text
stringlengths 66
377k
|
|---|
Epigenetic Features in Uterine Leiomyosarcoma and Endometrial Stromal Sarcomas: An Overview of the Literature
# Introduction
The body of the uterus is composed of a mucosa muscular interface derived from the Müllerian embryonic ducts and constituted of internal endometrium and external myometrium (MM) tissue layers [bib_ref] Endometrial-myometrial interface: Relationship to adenomyosis and changes in pregnancy, Uduwela [/bib_ref] [bib_ref] Epidemiological and genetic clues for molecular mechanisms involved in uterine leiomyoma development..., Commandeur [/bib_ref] [bib_ref] Non-coding rnas in uterine development, function and disease, Nothnick [/bib_ref] [bib_ref] Malignant tumors of the uterine corpus: Molecular background of their origin, Brany [/bib_ref]. The internal endometrium is composed of luminal epithelium, glandular epithelium, and endometrial stroma whereas the MM consists mainly of smooth muscle cells [bib_ref] Epidemiological and genetic clues for molecular mechanisms involved in uterine leiomyoma development..., Commandeur [/bib_ref] [bib_ref] Non-coding rnas in uterine development, function and disease, Nothnick [/bib_ref] [bib_ref] Malignant tumors of the uterine corpus: Molecular background of their origin, Brany [/bib_ref]. Cellular and molecular alterations in the endometrial stroma and smooth muscle cell layers can lead to uterine sarcoma (US) development [bib_ref] Malignant tumors of the uterine corpus: Molecular background of their origin, Brany [/bib_ref] [bib_ref] Tissue-engineered multi-cellular models of the uterine wall, Kuperman [/bib_ref] [bib_ref] Clinical and molecular features of uterine sarcomas, De Almeida [/bib_ref].
US accounts for 3-9% of all uterine malignancies and shows high rates of recurrence and metastasis [bib_ref] Systemic treatment in adult uterine sarcomas, Desar [/bib_ref] [bib_ref] Immunohistochemical studies on uterine carcinosarcoma, leiomyosarcoma, and endometrial stromal sarcoma: Expression and..., Koivisto-Korander [/bib_ref] , occupying the second place among all gynecological tumors [bib_ref] Systemic treatment in adult uterine sarcomas, Desar [/bib_ref]. The American Cancer Society (ACS) registered a total of 66,570 new cases of uterine tumors with about 12,940 related deaths in 2021 [bib_ref] Cancer statistics, 2021, Siegel [/bib_ref] and estimates 65,950 new cases for 2022 [fig_ref] Figure 1: The estimated incidence of gynecological tumors for 2022 according to the ACS [/fig_ref].
"Pure" sarcomas are composed exclusively of mesenchymal cells and include the leiomyosarcomas (LMS) and endometrial stromal sarcomas (ESS), which are morphologically classified mainly based on the tumor cells phenotype [bib_ref] Uterine mesenchymal tumors: Update on classification, staging, and molecular features, Parra-Herran [/bib_ref]. LMS arises from the smooth muscle compartment, while ESS arises from the stroma supporting the endometrial glands [bib_ref] Immunohistochemical studies on uterine carcinosarcoma, leiomyosarcoma, and endometrial stromal sarcoma: Expression and..., Koivisto-Korander [/bib_ref]. LMS and ESS are the most frequent uterine mesenchymal tumors in adult age [bib_ref] Molecular biomarkers for uterine leiomyosarcoma and endometrial stromal sarcoma, Tsuyoshi [/bib_ref]. For LMS and ESS, the disease stage is the single most important prognostic factor [bib_ref] Uterine sarcomas, Mbatani [/bib_ref]. The International Federation of Gynecology and Obstetrics (FIGO) classification and staging system has been specifically designed for these tumors [bib_ref] Uterine sarcomas: A review, D'angelo [/bib_ref]. In 2018, the ACS published the latest revision of the definitions and clinical staging of LMS and ESS [fig_ref] Table 1: Staging of LMS and ESS [/fig_ref] , based on the FIGO system and the American Joint Committee on Cancer (AJCC) TNM staging system [bib_ref] Uterine sarcomas, Mbatani [/bib_ref] [bib_ref] Uterine leiomyosarcoma: A review of the literature and update on management options, Roberts [/bib_ref].
It is well known that several molecular events may lead to tumor development. Among these, epigenetic mechanisms such as DNA methylation, post-translational modifications (PTMs), and non-coding RNA (ncRNA) regulation (e.g., microRNAs) can significantly affect the expression of relevant genes, leading to dramatic cell changes [bib_ref] Epigenetic and genetic landscape of uterine leiomyomas: A current view over a..., Laganà [/bib_ref] [bib_ref] Epigenetic regulation in human cancer: The potential role of epi-drug in cancer..., Lu [/bib_ref] [bib_ref] 5-Hydroxymethylcytosine promotes proliferation of human uterine leiomyoma: A biological link to a..., Navarro [/bib_ref]. Epigenetic alterations are characterized by reversibility and susceptibility to external factors and are the main regulatory events governing the development and progression of uterine sarcomas [bib_ref] Epigenetic and genetic landscape of uterine leiomyomas: A current view over a..., Laganà [/bib_ref] [bib_ref] Epigenetic regulation in human cancer: The potential role of epi-drug in cancer..., Lu [/bib_ref] [bib_ref] Let-7 miRNA's expression profile and its potential prognostic role in uterine leiomyosarcoma, De Almeida [/bib_ref]. Here, we reviewed and summarized the scientific and clinical reports from the past twelve years regarding epigenetic events and their role in the pathophysiology of the ESS and LMS. The most relevant articles written in English were meticulously reviewed and included in this review, and no restrictions for geographic location were applied. Articles without tumor type description or any identification as "uterine" were excluded.
## Lms etiology, prognosis, and treatment
LMS arises from the myometrium (MM) and often does not reach the endometrial cavity surface [bib_ref] Outcome and management of uterine leiomyosarcoma treated following surgery for presumed benign..., Tantitamit [/bib_ref]. Its incidence is 0.36 per 100,000 women a year, affecting mainly women of ≥40 years of age, and representing approximately 70% of all US [bib_ref] Uterine sarcoma-Current perspectives, Benson [/bib_ref] [bib_ref] Prognosis and treatment of uterine leiomyosarcoma: A national cancer database study, Seagle [/bib_ref] [bib_ref] Role of bevacizumab in uterine leiomyosarcoma, Bogani [/bib_ref] [bib_ref] Uterine leiomyosarcoma, Juhasz-Böss [/bib_ref] [bib_ref] Options for treating different soft tissue sarcoma subtypes, Ray-Coquard [/bib_ref]. LMS is a very heterogeneous tumor and represents the most common sarcoma of the uterine body [bib_ref] Uterine sarcomas, Mbatani [/bib_ref] [bib_ref] Uterine leiomyosarcoma: A review of the literature and update on management options, Roberts [/bib_ref] [bib_ref] Prognosis and treatment of uterine leiomyosarcoma: A national cancer database study, Seagle [/bib_ref] [bib_ref] Role of bevacizumab in uterine leiomyosarcoma, Bogani [/bib_ref]. Its pathogenesis is poorly understood, but several studies focus-ing on tumor clonality indicate that many of these tumors are de novo entities . Even though it is an extremely rare event , some authors defend the hypothesis that LMSs could arise from the malignant transformation of a pre-existing leiomyoma (LM) [bib_ref] Uterine sarcomas: A review, D'angelo [/bib_ref]. However, most of the patients do not exhibit predisposing factors such as prior radiation therapy to the pelvis (10-25%), tamoxifen use (1-2%), genetic syndromes (e.g., retinoblastoma and Li-Fraumeni syndrome), postmenopausal status, and ethnicity (African American) [41]. Tumor restricted to the uterus (less than 5 cm) (T1a). N0 M0
## Ib t1b
Tumor restricted to the uterus (more than 5 cm) (T1b). N0 M0 II T2
Tumor growing outside the uterus but is restricted to the pelvis (T2). N0 M0 IIIA T3a
Tumor growing in a single tissue located in the abdomen (T3a). N0 M0 IIIB T3b
Involvement of other extrauterine pelvic tissues, 2 or more sites (T3b). N0 M0 IIIC T1-T3
Tumor invades abdominal tissues (does not protrude from the abdomen) but does not grow into the bladder or rectum (T1 to T3). The cancer has spread to nearby lymph nodes (N1).
## N1 m0
IVA T4 Tumor spread to the rectum or urinary bladder (T4). It might or might not have spread to nearby lymph nodes (Any N). Any N M0 IVB Any T Tumor spread to distant sites (lungs, bones, or liver) (M1). It may or may not have grown into tissues in the pelvis and/or abdomen (any T) and it might or might not have spread to lymph nodes (Any N).
## Any n m1
Clinically, LMS is associated with a poor prognosis even when diagnosed in the early stages, consequently leading to a significant increase in uterine cancer-associated deaths [bib_ref] Molecular biomarkers for uterine leiomyosarcoma and endometrial stromal sarcoma, Tsuyoshi [/bib_ref] [bib_ref] Uterine leiomyosarcoma: A review of the literature and update on management options, Roberts [/bib_ref] [bib_ref] Role of bevacizumab in uterine leiomyosarcoma, Bogani [/bib_ref] [bib_ref] Options for treating different soft tissue sarcoma subtypes, Ray-Coquard [/bib_ref]. The recurrence rate of LMS reaches 53-75%, even at the initial stages of the disease, with locoregional or distant recurrence in the first two years after diagnosis [bib_ref] Systemic treatment in adult uterine sarcomas, Desar [/bib_ref] [bib_ref] Uterine sarcomas, Mbatani [/bib_ref] [bib_ref] Role of bevacizumab in uterine leiomyosarcoma, Bogani [/bib_ref] [bib_ref] Options for treating different soft tissue sarcoma subtypes, Ray-Coquard [/bib_ref]. The overall survival expectancy of LMS is 2.6 years, and the survival at 2, 5, and 10 years are approximately 57%, 24%, and 12%, respectively [bib_ref] Role of bevacizumab in uterine leiomyosarcoma, Bogani [/bib_ref] [bib_ref] Options for treating different soft tissue sarcoma subtypes, Ray-Coquard [/bib_ref] ]. The survival rates for patients with LMS decrease as the disease progresses; thus, for localized disease (i.e., restricted to the uterus) the estimated survival rates are 64%, for regional disease (afflicting nearby and adjacent tissues, i.e., lymph nodes) the survival rates are 36%, and for disseminated disease or metastatic disease (i.e., lungs and liver) the survival is 14% [51].
LMS-related symptoms are associated with vaginal bleeding in 56% of the cases, increased pelvic mass in 54% of the cases, and/or pelvic pain in 22% of the cases [bib_ref] Uterine sarcomas, Mbatani [/bib_ref] [bib_ref] Uterine sarcomas: A review, D'angelo [/bib_ref] [bib_ref] Uterine leiomyosarcoma: A review of the literature and update on management options, Roberts [/bib_ref]. Typically, 75% of the patients present a large tumor mass with an average diameter of 10 cm at the time of diagnosis [bib_ref] Uterine sarcomas, Mbatani [/bib_ref] [bib_ref] Uterine leiomyosarcoma: A review of the literature and update on management options, Roberts [/bib_ref]. Although LMSs occur primarily in postmenopausal women [52], both progesterone receptor (PR) and estrogen receptor (ER) are found to be expressed in 40% and 70% of the cases, respectively . A recent review has suggested that hormonal therapy applied to LMS expressing ER/PR is effective and presents favorable tolerance and reliability [bib_ref] Hormonal therapy in uterine sarcomas, Zang [/bib_ref].
There are preoperative methods that allow the differential diagnosis of benign and malignant uterine disease. Magnetic resonance imaging (MRI) remains the optimal imaging modality to characterize pelvic masses originating from the uterus, but distinguishing LMS from LM remains a challenge [bib_ref] Uterine leiomyosarcoma: A review of the literature and update on management options, Roberts [/bib_ref].
LMS histopathological analysis is characterized by the presence of spindle cells, with ruptured nuclei, perinuclear vacuolization, and eosinophilic cytoplasm arranged in intersecting fascicles within the analyzed sample. Meeting the Stanford criteria, LMS should be deemed intrinsically as a high-grade tumor [bib_ref] Mesenchymal tumors of the uterus, Oliva [/bib_ref]. Cell atypia can vary from moderate to severe, while nuclear atypia is always severe, large areas of tumor cell necrosis with variable mitotic index and atypical mitosis are often observed [bib_ref] Uterine sarcomas: A review, D'angelo [/bib_ref] [bib_ref] Uterine sarcoma-Current perspectives, Benson [/bib_ref] [bib_ref] Uterine leiomyosarcoma: A review article, Gockley [/bib_ref] [bib_ref] Systemic treatment of metastatic/recurrent uterine leiomyosarcoma: A changing paradigm, Arend [/bib_ref]. There are two uncommon subtypes of uterine LMS: myxoid and epithelioid LMS. These present mild or focal nuclear pleomorphism and lower mitotic degree, compared to typical LMS. Diagnostic mistakes between these types of LMS and other smooth muscle tumors are often common [bib_ref] Myxoid leiomyosarcoma of the uterus: A clinicopathologic analysis of 30 cases and..., Parra-Herran [/bib_ref] [bib_ref] Epithelioid smooth-muscle tumors of the uterus: A clinicopathologic study of 18 patients, Prayson [/bib_ref].
Immunohistochemical co-expression of Desmin, h-Caldesmon, smooth muscle actin (SMA), and HDAC8 can assist with LMS diagnosis [bib_ref] Differentiating soft tissue leiomyosarcoma and undifferentiated pleomorphic sarcoma: A miRNA analysis, Guled [/bib_ref] [bib_ref] Immunohistochemical and selected genetic reflex testing of all uterine leiomyosarcomas and STUMPs..., Ptáková [/bib_ref]. Several other potential biomarkers such as PDGFRA, WT1, GNRHR, BCL2, ESR, PGR, and LAMP2 have also been evaluated to distinguish LMS from other tumors, mainly from LM [bib_ref] Differentiating soft tissue leiomyosarcoma and undifferentiated pleomorphic sarcoma: A miRNA analysis, Guled [/bib_ref]. The cell proliferative index (determined by Ki-67 protein expression), the protein expression levels of the tumor suppressors p16 and p53, and the expression of several isoforms of the CD44 (hyaluronan receptor) are, however, routinely used [bib_ref] Immunohistochemical studies on uterine carcinosarcoma, leiomyosarcoma, and endometrial stromal sarcoma: Expression and..., Koivisto-Korander [/bib_ref]. Additionally, some patients show high amounts of lactate dehydrogenase (LDH) [bib_ref] Uterine mesenchymal tumors: Update on classification, staging, and molecular features, Parra-Herran [/bib_ref] and/or CA125 levels [bib_ref] Molecular biomarkers for uterine leiomyosarcoma and endometrial stromal sarcoma, Tsuyoshi [/bib_ref] , but these markers are quite unspecific [bib_ref] Uterine leiomyosarcoma: A review of the literature and update on management options, Roberts [/bib_ref].
Most recently, gene expression profile analysis has enabled the classification of LMS into two subtypes according to their molecular signature. The subtype I recapitulated the low-grade LMS and was enriched for LMOD1, SLMAP, MYLK, and MYH11, all of them smooth muscle-specific markers. Subtype II of LMS included tumors with worse prognosis and expressed genes associated with cell cycle, proliferation, and tumorigenesis (CDK6, BMP1, MAPK13, PDGFRL, and HOXA1) [bib_ref] Genetic and epigenetic features in uterine smooth muscle tumors: An update, Dos Anjos [/bib_ref] [bib_ref] Distinct molecular subtypes of uterine leiomyosarcoma respond differently to chemotherapy treatment, An [/bib_ref].
The gold standard treatment for LMS is still the tumor surgical excision. Total hysterectomy and bilateral salpingo-oophorectomy are recommended for early-stage tumors [bib_ref] Uterine leiomyosarcoma: A review of the literature and update on management options, Roberts [/bib_ref] [bib_ref] Mesenchymal tumors of the uterus, Oliva [/bib_ref] [bib_ref] Uterine leiomyosarcoma: A review of recent advances in molecular biology, clinical management..., Cui [/bib_ref]. Adjuvant chemo and radiotherapies are indicated to avoid recurrences, or for early-stage disease, but their effectiveness is still unclear [bib_ref] Uterine leiomyosarcoma: A review of the literature and update on management options, Roberts [/bib_ref] [bib_ref] Uterine leiomyosarcoma: 14-year two-center experience of 31 cases, Kim [/bib_ref] and they do not offer a significant advantage to improve overall survival [bib_ref] Let-7 miRNA's expression profile and its potential prognostic role in uterine leiomyosarcoma, De Almeida [/bib_ref] [bib_ref] Adjuvant management of operated uterine sarcomas: A single institution experience, Cordoba [/bib_ref] [bib_ref] Does adjuvant chemotherapy improve survival for women with early-stage uterine leiomyosarcoma?, Ricci [/bib_ref]. Recently, new chemotherapies, targeted therapies (pazopanib), and immunotherapies (nivolumab or pembrolizumab) seem to be promising new approaches to treat drugresistant LMS [41].
## Ess etiology, prognosis, and treatment
ESS is the second most common type of US [bib_ref] Molecular pathogenesis and prognostication of "low-grade" and 'high-grade' endometrial stromal sarcoma, Micci [/bib_ref] and arises from the uterine stroma. It is composed of endometrial stromal cells reminiscent of proliferative phase endometrium [bib_ref] Systemic treatment in adult uterine sarcomas, Desar [/bib_ref] [bib_ref] Molecular biomarkers for uterine leiomyosarcoma and endometrial stromal sarcoma, Tsuyoshi [/bib_ref] [bib_ref] Uterine sarcomas: A review, D'angelo [/bib_ref] [bib_ref] Endometrial stromal sarcoma: A review of the literature, Puliyath [/bib_ref]. It is predominantly intramural, showing both a myometrial invasion and myometrial lymphovascular space permeation [bib_ref] Systemic treatment in adult uterine sarcomas, Desar [/bib_ref]. ESS pathogenesis is unknown, but tamoxifen exposure and some medical conditions (e.g., polycystic ovary syndrome) may contribute to its development [bib_ref] Endometrial stromal sarcoma: A review of the literature, Puliyath [/bib_ref] [bib_ref] Endometrial stromal sarcoma presented as endometrial polyp: A rare case, Angelos [/bib_ref]. Although a rare tumor, representing less than 1% of all uterine tumors, ESS accounts for up to 25% of all uterine sarcomas [bib_ref] Mesenchymal tumors of the uterus, Oliva [/bib_ref]. Symptoms related to ESS development and progression include abnormal uterine bleeding (about 90% of patients), uterine enlargement (70% of cases), pelvic pain, and dysmenorrhea. In 25% of the cases, however, the patients can be asymptomatic [bib_ref] Endometrial stromal sarcoma: A review of the literature, Puliyath [/bib_ref] [bib_ref] Endometrial stromal sarcoma: A rare entity, Jabeen [/bib_ref].
The most recent World Health Organization (WHO) classification (2020) for ESS is based on both cytogenetic and molecular analyses (i.e., gene fusion or alterationswhere the tumors are divided into benign endometrial stromal nodules (ESN), low-grade endometrial stromal sarcoma (LG-ESS), high-grade endometrial stromal sarcoma (HG-ESS), and undifferentiated uterine sarcomas (UUS) [fig_ref] Table 2: Molecular features of endometrial stromal tumors [/fig_ref] [bib_ref] Uterine sarcomas: An updated overview part 2: Endometrial stromal tumor, Zappacosta [/bib_ref] [bib_ref] Uterine sarcomas, Prat [/bib_ref]. Morphologically, ESN is differentiable from LG-ESS only for the absence of lymphovascular invasion and myometrial infiltration. LG-ESS is usually positive for CD10, ER, and PR and can express actin, keratins, and calretinin [bib_ref] Endometrial stromal sarcomas and related neoplasms: New developments and diagnostic considerations, Hoang [/bib_ref] , which differ from the HG-ESS tumors carrying the YWAE-NUTM2 fusion that do not express these markers. HG-ESS shows high expression of Cyclin-D1, c-KIT, and BCOR, and when ZC3H7B-BCOR fusion is present, CD10 and variable staining for ER and PR are also observed. Finally, UUS exhibits myometrial invasion, severe nuclear pleomorphism, high mitotic activity and/or necrosis, and loss of differentiation. This tumor, however, does not show a specific immunohistochemical profile, instead showing a diffused and atypical staining for CD10 as well as heterogeneous patterns of ER and PR staining [bib_ref] Uterine sarcomas, Mbatani [/bib_ref] [bib_ref] Immunohistochemical and molecular characterization of endometrial stromal sarcomas, Subbaraya [/bib_ref].
HG-ESS and UUS can be difficult to diagnostically differentiate from LMS since the latter can mimic both ESS and UUS. In this case, the immunohistochemical markers and morphological features can be useful, but not accurate. Tumor location can also provide important information because LMS is exclusively related to MM, while UUS may also involve the endometrium. Furthermore, ESS is also diagnosed only post-surgery, but unlike LMS, they present an indolent course with relapses occurring up to 20 years after diagnosis [bib_ref] Endometrial stromal tumors: The new who classification, Conklin [/bib_ref].
Overall, patients with ESS have a better life expectancy than other sarcomas. Their five-year survival rates are higher than 80%. For disease stages I and II the five-year survival is approximately 90% whereas for stages III and IV (i.e., advanced disease) the survival rate for the same interval of time is significantly reduced, according to the FIGO stage system [bib_ref] Endometrial stromal sarcoma: A rare entity, Jabeen [/bib_ref].
Treatment protocols are defined based on the grade and stage of the tumor at the time of diagnosis. Total hysterectomy with bilateral salpingo-oophorectomy remains the standard treatment for ESS, and lymphadenectomy does not appear to improve overall survival rates [bib_ref] Recurrent endometrial stromal sarcoma: Treatment with a progestin and gonadotropin releasing hormone..., Dupont [/bib_ref]. Adjuvant radiotherapy and hormone therapy are not well-established therapeutic options yet, even though some studies have shown that hormonal agents can be an alternative to the management of LG-ESS [bib_ref] Systemic therapy for endometrial stromal sarcomas: Current treatment options, Serkies [/bib_ref]. In contrast, HG-ESS is generally detected in advanced stages with no effective adjuvant therapy available. In this case, immunotherapy with adoptive T cells transfer, targeting tumor fusion proteins, can be useful. Such an approach has been proved to be efficient in inhibiting tumor recurrences in other cancer types, thus inducing long-term memory cells and the persistent presence of these cells in the patient's blood.
[formula] JAZF1-SUZ12 1 [88] JAZF1-PHF1 [88] MEAF6-PHF1 [88] EPC1-PHF1 [89] MBTD1-EZHIP [89] JAZF1-BCORL1 [89] MAGED2-PLAG1 [90] MEAF6-SUZ12 [91] EPC2-PHF1 [92] BRD8-PHF1 [72] EPC1-BCOR [72] EPC1-SUZ12 [72] [/formula]
## Genetics and epigenetics mechanisms in lms and ess
Genetic changes are related to alterations in the DNA sequences, whereas epigenetic modifications involve specific regulatory events apart from DNA codification [bib_ref] Cancer epigenetics: Moving forward, Nebbioso [/bib_ref] [bib_ref] Cancer epigenetics comes of age, Jones [/bib_ref] [bib_ref] Advances in epigenetics link genetics to the environment and disease, Cavalli [/bib_ref] [bib_ref] Epigenetic research in cancer epidemiology: Trends, opportunities, and challenges, Verma [/bib_ref] [bib_ref] Epigenetic mechanisms and cancer an interface between the environment and the genome, Herceg [/bib_ref] [bib_ref] Epigenetics and precision oncology, Werner [/bib_ref] [bib_ref] Cancer epigenetics: An introduction, Kanwal [/bib_ref]. Epigenetic events play an important role in several normal cellular processes, including embryonic development, genetic imprinting, and X-chromosome inactivation. When altered, epigenetic mechanisms may lead to several diseases, including cancer initiation and progression [bib_ref] Epigenetic mechanisms and cancer an interface between the environment and the genome, Herceg [/bib_ref]. Epigenetic dysregulation affects gene functions by altering the gene expression mainly by (1) DNA methylation, (2) PTMs, and (3) RNA-mediated gene silencing by ncRNA (e.g., microRNA) [fig_ref] Figure 2: Graphical representation of epigenetic events potentially involved in the initiation and development... [/fig_ref] [bib_ref] Cancer epigenetics: Moving forward, Nebbioso [/bib_ref] [bib_ref] The emerging role of epigenetic therapeutics in immuno-oncology, Topper [/bib_ref]. The main clinical and scientific interest in epigenetic events resides in the fact that they are reversible mechanisms [bib_ref] Non-coding RNAs as regulators in epigenetics (Review), Wei [/bib_ref] [bib_ref] Alterations of histone modifications in cancer, Hosseini [/bib_ref].
## Dna methylation
DNA methylation is the most studied and understood epigenetic event described to date. Found in more than 70% of the human genome, DNA methylation is crucial for cellular differentiation and normal development [bib_ref] Epigenetic modifications of histones in cancer, Zhao [/bib_ref]. It consists of the addition of a methyl radical (CH3) to the 5-carbon on cytosine residues (5mC) in CpG dinucleotides [bib_ref] Cancer epigenetics comes of age, Jones [/bib_ref] [bib_ref] Advances in epigenetics link genetics to the environment and disease, Cavalli [/bib_ref] [bib_ref] Cancer epigenetics: An introduction, Kanwal [/bib_ref] [bib_ref] The emerging role of epigenetic therapeutics in immuno-oncology, Topper [/bib_ref] [bib_ref] Non-coding RNAs as regulators in epigenetics (Review), Wei [/bib_ref] [bib_ref] Alterations of histone modifications in cancer, Hosseini [/bib_ref] [bib_ref] Epigenetic modifications of histones in cancer, Zhao [/bib_ref] [bib_ref] DNA methylation machinery in the endometrium and endometrial cancer, Caplakova [/bib_ref] [bib_ref] Epigenetic regulation of gene expression in cancer: Techniques, resources and analysis, Kagohara [/bib_ref]. DNA methyltransferases (DNMTs)-enzymes responsible for DNA methylation-are known to act in cancer cells by either hypomethylation or hypermethylation of specific CpG regions in the DNA [bib_ref] Epigenetic regulation of gene expression in cancer: Techniques, resources and analysis, Kagohara [/bib_ref].
Global DNA hypomethylation or loss of methylation has been associated with genomic instability as well as aneuploidy, loss of imprinting, reactivation of transposable elements, and endogenous retrovirus (ERVs) [bib_ref] Cancer epigenetics: An introduction, Kanwal [/bib_ref] [bib_ref] DNA methylation machinery in the endometrium and endometrial cancer, Caplakova [/bib_ref] [bib_ref] Hypomethylation distinguishes genes of some human cancers from their normal counterparts, Feinberg [/bib_ref]. In cancer, hypomethylation is commonly followed by hypermethylation of localized CpG islands at the promoter and regulatory regions of target genes, which remain unmethylated in normal cells [bib_ref] The emerging role of epigenetic therapeutics in immuno-oncology, Topper [/bib_ref] [bib_ref] Hypomethylation distinguishes genes of some human cancers from their normal counterparts, Feinberg [/bib_ref] [bib_ref] Treatment with epigenetic agents profoundly inhibits tumor growth in leiomyosarcoma, Fischer [/bib_ref]. Hypermethylation of regulatory regions leads to transcriptional silencing by directly blocking the transcription factors binding to the promoter region, or by the binding of proteins with a high affinity for methylated DNA that compete with the transcription factors binding sites [fig_ref] Figure 3: Schematic representation of the DNA methylation process in LMS and ESS during... [/fig_ref] [bib_ref] Cancer epigenetics: An introduction, Kanwal [/bib_ref] [bib_ref] DNA methylation-how important in gene control?, Bird [/bib_ref].
DNMTs are commonly found overexpressed in tumors, constituting an attractive target for specific therapy. The FDA has approved "epidrugs" such as 5-azacitidine (5-Aza), 5-aza-2-deoxycytidine (DAC), and the second-generation of the demethylation agent guadecitabine [bib_ref] Treatment with epigenetic agents profoundly inhibits tumor growth in leiomyosarcoma, Fischer [/bib_ref] [bib_ref] DNA methyltransferase inhibitors and their emerging role in epigenetic therapy of cancer, Gnyszka [/bib_ref]. In LMS, [bib_ref] Treatment with epigenetic agents profoundly inhibits tumor growth in leiomyosarcoma, Fischer [/bib_ref] assessed the therapeutic potential of nucleoside analogs 5-Aza, DAC, and guadecitabine, using both in vitro and in vivo experiments. Their results show guadecitabine as a more effective inhibitor of both cell survival and colony formation in vitro. Additionally, animals who received this treatment showed a decrease in the tumor burden and increased survival [bib_ref] Treatment with epigenetic agents profoundly inhibits tumor growth in leiomyosarcoma, Fischer [/bib_ref]. Our group recently assessed the impact of DNMT inhibition on the Hedgehog (HH) signaling pathway with 5 -Aza-dc, in uterine LMS cells. We observed a reduction in the GLI1 mRNA, and SMO and GLI1 protein in response to the treatment. Moreover, GLI1 and GLI2 nuclear translocation were also decreased while nuclear translocation of GLI3 was increased. Our data showed that DNMT inhibitor, alone or in combination with pharmacological treatment, was able to block the HH pathway and showed a high inhibitory effect on the LMS malignant cells phenotype.
MGMT silencing due to its promoter hypermethylation has been commonly observed in several malignancies, including uterine sarcomas [bib_ref] Temozolomide in advanced and recurrent uterine leiomyosarcoma and correlation with O6-methylguanine DNA..., Ferriss [/bib_ref] [bib_ref] The promoter methylation and expression of the O6-methylguanine-DNA methyltransferase gene in uterine..., Bujko [/bib_ref]. The methylation of the MGMT promoter region, which contributes to genome instability and sensitizes the cells to alkylating agents (such as Temozolomide (TMZ)), has been correlated with improved prognosis and as a potential factor of response to TMZ-based therapy prediction [bib_ref] The promoter methylation and expression of the O6-methylguanine-DNA methyltransferase gene in uterine..., Bujko [/bib_ref]. Global DNA methylation studies have also found methylation patterns or signatures that have been associated with different cancer hallmarks [bib_ref] Role of DNA methylation in the resistance to therapy in solid tumors, Romero-Garcia [/bib_ref]. Braný et al. in 2019 observed differences in the methylation levels between MM and LM samples in the KLF4 and DLEC1 genes. Higher levels of methylation were found in LM compared to LMS cases, suggesting that methylation of KLF4 and DLEC1 are potential biomarkers to distinguish LM from LMS [bib_ref] Different methylation levels in the KLF4, ATF3 and DLEC1 genes in the..., Braný [/bib_ref]. [bib_ref] Epigenetic signatures differentiate uterine and soft tissue leiomyosarcoma, Hasan [/bib_ref] identified differentially methylated and differentially expressed genes associated with LMS. Among the 77 hypermethylated genes, chromatin-modifying enzymes, including KAT6A, KMT2A and EZH2, and chromatin/DNA binding proteins such as CTNNB1, PBX3, SATB1, MEIS and COMMD1-BMI1 were observed. The findings indicate the possible involvement of chromatin modulation in regulating the DNA methylation of these genes [bib_ref] Epigenetic signatures differentiate uterine and soft tissue leiomyosarcoma, Hasan [/bib_ref].
A higher DNA damage response and hypomethylation of estrogen receptor 1 (ESR1) target genes were both observed when comparing uterine to extra uterine LMS [bib_ref] Molecular biomarkers for uterine leiomyosarcoma and endometrial stromal sarcoma, Tsuyoshi [/bib_ref].
Gene silencing through methylation can occur as frequently as mutations or deletions, leading to aberrant silencing of tumor suppressor genes [bib_ref] DNA methylation and gene silencing in cancer, Baylin [/bib_ref]. In an LMS experimental model, the lack of BRCA1 function was associated with tumor initiation and development. This protein expression was next investigated in human samples, and a loss of 29% was associated with promoter methylation [bib_ref] A role for BRCA1 in uterine leiomyosarcoma, Xing [/bib_ref]. Methylation in the CDKN2A gene, using uterine LMS samples with a rhabdomyosarcomatous component, has also been described [bib_ref] Loss of CDKN2A promoter methylation coincides with the epigenetic transdifferentiation of uterine..., Roncati [/bib_ref]. The authors identified both methylated and unmethylated alleles, originating mainly from the smooth muscle component. Moreover, the loss of heterozygosity in the rhabdomyosarcomatous component has been described exclusively in the cells expressing p16 and p14 [bib_ref] Loss of CDKN2A promoter methylation coincides with the epigenetic transdifferentiation of uterine..., Roncati [/bib_ref].
Hierarchical clustering based on the hypermethylation of ALX1, CBLN1, CORIN, DUSP6, FOXP1, GATA2, IGLON5, NPTX2, NTRK2, STEAP4, PART1, and PRL allowed differentiation among several uterine tumors with up to 70% accuracy [bib_ref] DNA methylation-based profiling of uterine neoplasms: A novel tool to improve gynecologic..., Kommoss [/bib_ref] [bib_ref] Endometrial stromal sarcomas with BCOR-rearrangement harbor MDM2 amplifications, Kommoss [/bib_ref]. Clusters of distinct DNA methylation patterns have also been useful in distinguishing tumor types regardless of the number of CpG sites. Thus, LM and LMS were separated into two different clusters each, while ESS samples (LG-and HG-ESS) were grouped into two subtypes with specific profiles. The HG-ESS cluster included YWHAE and BCOR-rearranged tumors, distinct from LG-ESS and LMS [bib_ref] DNA methylation-based profiling of uterine neoplasms: A novel tool to improve gynecologic..., Kommoss [/bib_ref] [bib_ref] Endometrial stromal sarcomas with BCOR-rearrangement harbor MDM2 amplifications, Kommoss [/bib_ref]. Although LMS and HG-ESS are morphologically similar, the results show that the DNA methylation profile may be useful to discriminate against these closely related tumors [bib_ref] DNA methylation-based profiling of uterine neoplasms: A novel tool to improve gynecologic..., Kommoss [/bib_ref].
DNA methylation plays a key role in gene expression regulation, inducing functional changes in key genes that regulate endometrial homeostasis [bib_ref] DNA methylation machinery in the endometrium and endometrial cancer, Caplakova [/bib_ref].showed that the KLF4 promoter was hypermethylated in the ESS. They also found PCDHGC5 was highly methylated in the ESS samples when compared to endometrioid and endometrial serous carcinoma. sFRPs 1-5 are tumor suppressors that downregulate Wnt/β-catenin signaling [bib_ref] Cytogenetic and molecular aberrations in endometrial stromal tumors, Chiang [/bib_ref] [bib_ref] Coincident expression of β-catenin and cyclin D1 in endometrial stromal tumors and..., Kurihara [/bib_ref] [bib_ref] Molecular mechanisms of endometrial stromal sarcoma and undifferentiated endometrial sarcoma as premises..., Hrzenjak [/bib_ref]. Their consistent promoter hypermethylation and subsequent gene expression suppression were described in ESN, LG-ESS, and UUS.
Although the true role of DNA methylation patterns in the LMS and ESS initiation and progression is not completely understood, DNA methylation in normal endometrial stromal cells has been useful to identify signatures that indicate changes during decidualization or cell transformations [bib_ref] Genistein promotes DNA demethylation of the steroidogenic factor 1 (SF-1) promoter in..., Matsukura [/bib_ref] [bib_ref] Genome-wide DNA methylation analysis revealed stable DNA methylation status during decidualization in..., Maekawa [/bib_ref] [bib_ref] Transcription factor C/EBPβ induces genome-wide H3K27ac and upregulates gene expression during decidualization..., Tamura [/bib_ref] [bib_ref] Promoter hypermethylation of progesterone receptor isoform B (PR-B) in adenomyosis and its..., Nie [/bib_ref]. Further studies to determine the precise methylation profile in pure mesenchymal tumors are certainly necessary to enable the characterization and differentiation of these tumors as well as to establish new therapeutic options.
## Chromatin remodeling
The chromatin is composed of DNA molecules tightly coiled around proteins called histones. This structure condensation degree is directly associated with greater or lesser RNA synthesis, with greater condensation (higher chromatin closing) being the state of more transcriptional repression [bib_ref] Readers, writers, and erasers: Chromatin as the whiteboard of heart disease, Gillette [/bib_ref]. The basic unit of chromatin, called nucleosome, is constituted by two copies of each core histone (H2A, H2B, H3, and H4) enveloped by DNA molecules [bib_ref] Nucleosome and chromatin fiber dynamics, Luger [/bib_ref]. Chromatin regulation occurs through PTMs of core histones and can involve phosphorylation, acetylation, methylation, ubiquitination, SUMOylation, and GlcNAcylation [bib_ref] Nucleosome and chromatin fiber dynamics, Luger [/bib_ref].
Currently, the two most studied and best understood mechanisms of chromatin regulation are histones acetylation and methylation. Such events are regulated by very specialized proteins called writers, erasers, and readers-which respectively add, remove, or recognize these PTMs [bib_ref] Non-coding RNAs as regulators in epigenetics (Review), Wei [/bib_ref] [bib_ref] Alterations of histone modifications in cancer, Hosseini [/bib_ref] [bib_ref] Epigenetic modifications of histones in cancer, Zhao [/bib_ref] [bib_ref] Assessment of synergistic contribution of histone deacetylases in prognosis and therapeutic management..., Mastoraki [/bib_ref] [bib_ref] Histone acetylation and the role of histone deacetylases in normal cyclic endometrium, Gujral [/bib_ref] [bib_ref] Regulation of chromatin by histone modifications, Bannister [/bib_ref]. Included in the writers' group are histone acetyltransferases (HATs), DNA methyltransferases (DNMTs), histone lysine methyltransferases (HKMTs), and histone methyltransferases (HMTs). The erasers' group includes histone deacetylase (HDACs) and histone demethylases (HDMs), ten eleven translocations (TETs), and histone lysine demethylases (HKDMs). In the reader's group are methyl-CpG-binding domains (MBDs) and bromodomains [bib_ref] Epigenetic tools (the writers, the readers and the erasers) and their implications..., Biswas [/bib_ref].
Histone modifications affect the chromatin structure providing binding sites for several transcriptional factors. Its modification has a direct influence on gene expression, DNA replication and repair, chromatin compaction, and cell cycle control. Thus, loss of regulation of the histone modifications can lead to cancer pathogenesis and several developmental defects [fig_ref] Figure 4: Chromatin remodeling process in LMS and ESS [/fig_ref] [bib_ref] Epigenetic tools (the writers, the readers and the erasers) and their implications..., Biswas [/bib_ref].
The lysine residues of histones H3 and H4 are targeted for methylation by site-specific enzymes, culminating in activation or repression of the gene expression. [bib_ref] The diverse functions of histone lysine methylation, Martin [/bib_ref]. This molecular mechanism is uniquely able to originate three methylation levels: me1 (mono), me2 (di), and me3 (trimethylation). Lysine methylations may lead to both transcriptional activation and repression, depending on the lysine residue's location [bib_ref] Epigenetic tools (the writers, the readers and the erasers) and their implications..., Biswas [/bib_ref]. In endometrial stromal cells, the transition from a proliferative to a decidual phenotype occurs due to the loss of the EZH2-dependent methyltransferase activity, which is part of the chromatin remodeling process [bib_ref] Down-Regulation of the histone methyltransferase EZH2 contributes to the epigenetic programming of..., Grimaldi [/bib_ref]. The decidualization process in those cells downregulates EZH2, resulting in lower levels of H3K27me3 at the promoter region of PRL and IGFBP1 (two decidual marker genes). The H3K27me3 loss, associated with acetylation enrichment in the same lysine residue, indicates the transition from a transcriptionally repressive chromatin form to a permissive one [bib_ref] Down-Regulation of the histone methyltransferase EZH2 contributes to the epigenetic programming of..., Grimaldi [/bib_ref].
Little is known about the specific underlying mechanism of histone acetylation or methylation associated with the "pure" sarcoma pathobiology. A unique study available in the literature describes the fatty acid synthase (FASN)-enhanced expression inducing cell proliferation, migration, and invasion, in transfected cells of uterine LMS. It has been observed that FASN promotes H3K9me3 and H3K27ac by alteration in the HDAC, HDM, HMT, and HAT trimethylation activity. Thus, in the uterine LMS cells, the epigenome reprogramming by chromatin remodeling seems to induce a higher malignant phenotype [bib_ref] Fatty acid synthase reprograms the epigenome in uterine leiomyosarcomas, Guan [/bib_ref].
The polycomb group (PcG) proteins are well-characterized transcriptional repressors that are essential for the regulation of physiological processes in several organisms. PcG proteins are known to form two distinct complexes with defined enzymatic activities: polycomb repressive complex 1 (PRC1), a histone ubiquitin ligase related to chromatin compaction; and PRC2, an HMT that mediates both H3K27me3 and target genes repression [bib_ref] Bivalent promoter hypermethylation in cancer is linked to the H327me3/H3K4me3 ratio in..., Dunican [/bib_ref] [bib_ref] The roles of polycomb repressive complexes in mammalian development and cancer, Piunti [/bib_ref] [bib_ref] AEBP2 as a transcriptional activator and its role in cell migration, Kim [/bib_ref] [bib_ref] EZH2: Biology, disease, and structure-based drug discovery, Tan [/bib_ref]. In several cancer types, the expression and function of PcG proteins are often found dysregulated [bib_ref] The roles of polycomb repressive complexes in mammalian development and cancer, Piunti [/bib_ref] [bib_ref] CATACOMB: An endogenous inducible gene that antagonizes H3K27 methylation activity of Polycomb..., Piunti [/bib_ref] [bib_ref] Identification of a novel, recurrent MBTD1-CXorf67 fusion in low-grade endometrial stromal sarcoma, Dewaele [/bib_ref] [bib_ref] The core of the polycomb repressive complex is compositionally and functionally conserved..., Levine [/bib_ref] , and their targeted deletions generally induce lethal phenotypes [bib_ref] The core of the polycomb repressive complex is compositionally and functionally conserved..., Levine [/bib_ref].
PRC1 catalytic core has two related E3 ubiquitin ligases, the RING1 (RING1A) or RNF2 (RING1B) that catalyze ubiquitination and BMI1 (polycomb ring finger oncogene), and one of six PCGF orthologues. The latter constitutes a PRC1 variant containing BCOR and KDM2B [bib_ref] The variant polycomb repressor complex 1 component PCGF1 interacts with a pluripotency..., Oliviero [/bib_ref] [bib_ref] Pession, A. BCOR involvement in cancer, Astolfi [/bib_ref] [bib_ref] Fusion of the ZC3H7B and BCOR genes in endometrial stromal sarcomas carrying..., Panagopoulos [/bib_ref] [bib_ref] Aberrations of EZH2 in cancer, Chase [/bib_ref]. Translocations or chromosomal rearrangements, involv-ing fusion proteins have also been implicated in PRCs mechanisms. Fusions such as KDM2B-CREBBP [bib_ref] Cytogenetic and molecular profile of endometrial stromal sarcoma, Micci [/bib_ref] , ZC3H7-BCOR [bib_ref] Endometrial stromal sarcomas and related neoplasms: New developments and diagnostic considerations, Hoang [/bib_ref] [bib_ref] Pession, A. BCOR involvement in cancer, Astolfi [/bib_ref] [bib_ref] Fusion of the ZC3H7B and BCOR genes in endometrial stromal sarcomas carrying..., Panagopoulos [/bib_ref] [bib_ref] Gene of the month: BCOR, Aldera [/bib_ref] [bib_ref] A pan-cancer landscape analysis reveals a subset of endometrial stromal and pediatric..., Juckett [/bib_ref] , JAZF1-BCORL1 [bib_ref] Endometrial stromal sarcomas and related neoplasms: New developments and diagnostic considerations, Hoang [/bib_ref] [bib_ref] Low-grade endometrial stromal sarcoma with a novel MEAF6-SUZ12 fusion, Makise [/bib_ref] [bib_ref] A recurrent endometrial stromal sarcoma harbors the novel fusion JAZF1-BCORL1, Allen [/bib_ref] , EPC1-BCOR [bib_ref] Molecular pathogenesis and prognostication of "low-grade" and 'high-grade' endometrial stromal sarcoma, Micci [/bib_ref] , LPP-BCOR [bib_ref] Molecular pathogenesis and prognostication of "low-grade" and 'high-grade' endometrial stromal sarcoma, Micci [/bib_ref] and BCOR internal tandem duplications (BCOR-ITD) are frequently found in ESS, and have recently been also found in LMS samples [bib_ref] Immunohistochemical and molecular characterization of endometrial stromal sarcomas, Subbaraya [/bib_ref] [bib_ref] A pan-cancer landscape analysis reveals a subset of endometrial stromal and pediatric..., Juckett [/bib_ref] [bib_ref] The emerging role of H3K9me3 as a potential therapeutic target in acute..., Monaghan [/bib_ref] [bib_ref] Molecular classification of endometrial stromal sarcomas using RNA sequencing defines nosological and..., Brahmi [/bib_ref]. Additionally, gene fusion such as MBTD1-CXorf67 [bib_ref] Endometrial stromal sarcomas and related neoplasms: New developments and diagnostic considerations, Hoang [/bib_ref] [bib_ref] Low-grade endometrial stromal sarcoma with a novel MEAF6-SUZ12 fusion, Makise [/bib_ref] [bib_ref] CATACOMB: An endogenous inducible gene that antagonizes H3K27 methylation activity of Polycomb..., Piunti [/bib_ref] [bib_ref] Gene expression profiling of low-grade endometrial stromal sarcoma indicates fusion protein-mediated activation..., Przybyl [/bib_ref] , MBTD1-EZHIP [bib_ref] Novel fusion MAGED2-PLAG1 in endometrial stromal sarcoma, Mansor [/bib_ref] and MBTD1-PHF1 in ESS are found as products of PRC1-associated protein [bib_ref] Identification of a novel, recurrent MBTD1-CXorf67 fusion in low-grade endometrial stromal sarcoma, Dewaele [/bib_ref].
PRC2 has in its core subunits the EZH2 (or its homolog EZH1), EED, SUZ12, and RbAp46 (or 48). EZH1 and EZH2 are responsible for the H3K27me3 generation. EED binds to H3K27me3, enhancing the EZH2 catalytic activity, while SUZ12 is essential for the PRC2 activity, and bAp46/48 acts as a chaperone. H3K27 trimethylation leads to the suppression of several relevant genes, including tumor suppressors [bib_ref] Recent progress on small molecules targeting epigenetic complexes, Itoh [/bib_ref].
There are two well-characterized PcG proteins, the BMI1 and EZH2, which are required for the regulation of the PRC activity but are also known to display oncogenic functions in several cancers. BMI1 was the first identified PcG protein that was described as a protooncogene, and although there are no specific studies regarding the BMI1 role in uterine LMS [bib_ref] Aberrations of EZH2 in cancer, Chase [/bib_ref] [bib_ref] Epigenetic factors in cancer development: Polycomb group proteins, Piunti [/bib_ref]. The authors observed 91% of sensitivity and 100% of specificity for EZH2 positive staining in well-differentiated LMS, suggesting this expression is a specific marker for this tumor. Furthermore, the increased expression of EZH2 was inversely correlated with SUZ12 and EED expressions, leading to PRC2 suppression and H3K27me3 decrease [bib_ref] High expression of EZH2 as a marker for the differential diagnosis of..., Zhang [/bib_ref].
Chromosomal rearrangements in genes belonging to the PRC2 complex, or in proteins that interact with them, have previously been described in the ESS [bib_ref] Cytogenetic and molecular aberrations in endometrial stromal tumors, Chiang [/bib_ref]. JAZF1-SUZ12 (previously named JAZF1-JJAZ1) has been frequently reported as the most common feature of ESS [bib_ref] Cytogenetic and molecular aberrations in endometrial stromal tumors, Chiang [/bib_ref] [bib_ref] Gene expression profiling of low-grade endometrial stromal sarcoma indicates fusion protein-mediated activation..., Przybyl [/bib_ref] [bib_ref] Panagopoulos, I. Fusion of the genes BRD8 and PHF1 in endometrial stromal..., Micci [/bib_ref] [bib_ref] JAZF1/SUZ12 gene fusion in endometrial stromal sarcomas, Hrzenjak [/bib_ref] [bib_ref] Molecular cytogenetic analysis of JAZF1, PHF1, and YWHAE in endometrial stromal tumors:..., Hodge [/bib_ref]. Additionally, several other modifications such as LG-ESS harboring the EPC1-PHF1 fusion gene has decreased levels of H3K27me3 and a concomitant increase of H3K36me3 [bib_ref] Novel fusion of MYST/Esa1-associated factor 6 and PHF1 in endometrial stromal sarcoma, Panagopoulos [/bib_ref]. PHF1 acts in cell proliferation through the modulation of histone H3 methylation [bib_ref] Molecular cytogenetic analysis of JAZF1, PHF1, and YWHAE in endometrial stromal tumors:..., Hodge [/bib_ref].
EZH2 can interact with HDAC1 and HDAC2, through the EED protein, suggesting that the transcriptional repression by the PRC2 complex may be mediated by HDACs [bib_ref] Histone deacetylase inhibitors deplete enhancer of zeste 2 and associated polycomb repressive..., Fiskus [/bib_ref]. These enzymes act in the acetylation control of transcription factors [bib_ref] Different class IIa HDACs repressive complexes regulate specific epigenetic responses related to..., Giorgio [/bib_ref] , and their classification (Class I, IIa, IIb, III, and IV) is based on their activity, structural similarity, subcellular localization, and expression patterns [bib_ref] A short guide to histone deacetylases including recent progress on class II..., Park [/bib_ref]. In LMS patients, strong expression of HDACs 1, 2, 3, 4, 6, and 8 were associated with unfavorable prognosis, while HDACs 5, 7, or 9 weak expressions, together with p53 expression, were associated with favorable diseasefree survival (DFS). HDACs 5, 7, and 9 were associated with better survival outcomes, whereas HDAC5 expression was an independent predictor for DFS in epithelioid subtype tumors [bib_ref] SARC018-SPORE02: Phase II study of mocetinostat administered with gemcitabine for patients with..., Choy [/bib_ref]. In vitro analysis using SK-UT-1, SK-LMS-1, MES-SA, and DMR cell lines demonstrated that HDAC9 (Class IIa) transcription is under MEF2D direct control, and this axis sustains cell proliferation and survival through FAS repression [bib_ref] Histone deacetylase inhibitors deplete enhancer of zeste 2 and associated polycomb repressive..., Fiskus [/bib_ref].
In ESS, it has been observed that high expression of HDACs 1, 4, 6, 7, and 8 is associated with lower DFS [bib_ref] Molecular targets and emerging therapeutic options for uterine leiomyosarcoma, Miller [/bib_ref] whereas, in UUS, distant tumor recurrence was associated with a strong expression of HDAC6 [bib_ref] Assessment of synergistic contribution of histone deacetylases in prognosis and therapeutic management..., Mastoraki [/bib_ref] [bib_ref] Investigation of new therapeutic targets in undifferentiated endometrial sarcoma, Baek [/bib_ref].
The increased HDAC activity often observed in cancers justifies the number of current studies investigating HDAC inhibitors as novel therapeutic agents [bib_ref] Pharmacological histone deacetylation distinguishes transcriptional regulators, Rafehi [/bib_ref] [bib_ref] A phase I/II study targeting angiogenesis using bevacizumab combined with chemotherapy and..., Monga [/bib_ref]. These studies have shown promising results for metastatic LMS [bib_ref] Assessment of synergistic contribution of histone deacetylases in prognosis and therapeutic management..., Mastoraki [/bib_ref] [bib_ref] SARC018-SPORE02: Phase II study of mocetinostat administered with gemcitabine for patients with..., Choy [/bib_ref] [bib_ref] Molecular targets and emerging therapeutic options for uterine leiomyosarcoma, Miller [/bib_ref] [bib_ref] Uterine sarcoma part III-Targeted therapy: The taiwan association of gynecology (TAG) systematic..., Yen [/bib_ref]. In this context, mocetinostat acts by turning on tumor suppressor genes, restoring their normal function, and reducing tumor growth [bib_ref] Assessment of synergistic contribution of histone deacetylases in prognosis and therapeutic management..., Mastoraki [/bib_ref] [bib_ref] SARC018-SPORE02: Phase II study of mocetinostat administered with gemcitabine for patients with..., Choy [/bib_ref]. Its use as mono-or combinatorial therapy has been evaluated in metastatic extra-uterine and uterine LMS with resistance to gemcitabine and found to induce regression of tumors with acquired chemoresistance. Romidepsin, LBH589, belinostat, SAHA, and valproate (other HDAC inhibitors) have shown good results alone or in combination with decitabine [bib_ref] SARC018-SPORE02: Phase II study of mocetinostat administered with gemcitabine for patients with..., Choy [/bib_ref] [bib_ref] Effects of targeting endometrial stromal sarcoma cells via histone deacetylase and PI3K/AKT/mTOR..., Quan [/bib_ref].
Combinatorial therapy using SAHA, LY294002 (PI3K inhibitor), and rapamycin (mTOR inhibitor) were tested in ESS cell culture [bib_ref] Effects of targeting endometrial stromal sarcoma cells via histone deacetylase and PI3K/AKT/mTOR..., Quan [/bib_ref]. The results show that SAHA combined with either LY294002 or rapamycin, or both, reduce specifically phospho-p70S6K and 4E-BP1 levels, inhibiting the tumor cell proliferation [bib_ref] Effects of targeting endometrial stromal sarcoma cells via histone deacetylase and PI3K/AKT/mTOR..., Quan [/bib_ref] [bib_ref] Therapeutic applications of histone deacetylase inhibitors in sarcoma, Tang [/bib_ref]. A strong reduction of mTOR and phospho-mTOR levels has been reported after treatment with either SAHA or rapamycin, by targeting phospho-S6rp, in ESS cells [bib_ref] Autophagic and apoptotic effects of HDAC inhibitors on cancer cells, Rikiishi [/bib_ref]. [bib_ref] Epigenetic silencing of apoptosis-inducing gene expression can be efficiently overcome by combined..., Fröhlich [/bib_ref] showed the benefit of SAHA treatment associated with TRAIL/Apo-2L in two US cell lines [bib_ref] Epigenetic silencing of apoptosis-inducing gene expression can be efficiently overcome by combined..., Fröhlich [/bib_ref].
Another study assessed the effects of combined therapy with valproic acid (VPA, a weak histone deacetylase inhibitor), bevacizumab (mAb against VEGF), gemcitabine, and docetaxel, for extra-and intrauterine unresectable or metastatic soft tissue sarcomas [bib_ref] A phase I/II study targeting angiogenesis using bevacizumab combined with chemotherapy and..., Monga [/bib_ref]. This study found partial response in one case of carcinosarcoma, two extrauterine LMSs, two undifferentiated pleomorphic sarcomas, and one uterine LMS patient. This pharmacological combination was well tolerated and overall safe, showing that the combination of traditional medication and "epidrugs" may truly represent a new treatment strategy for sarcoma [bib_ref] A phase I/II study targeting angiogenesis using bevacizumab combined with chemotherapy and..., Monga [/bib_ref].
New therapeutic strategies to specifically treat US, such as regional hyperthermia, combined with chemotherapy, radiotherapy, and/or immunotherapy have emerged in the last few years. Pazopanib (a multitargeted tyrosine kinase inhibitor with antiangiogenic effects) combined with hyperthermia has demonstrated synergistic effects mainly for LMS growth inhibition, in vitro and in vivo [bib_ref] Uterine sarcoma part III-Targeted therapy: The taiwan association of gynecology (TAG) systematic..., Yen [/bib_ref]. This approach induces HAT1 downregulation by suppressing Clock which, in turn, is responsible for H3 and H4 acetylation [bib_ref] Synergistic effects of pazopanib and hyperthermia against uterine leiomyosarcoma growth mediated by..., Lin [/bib_ref].
Histone phosphorylation, which takes place predominantly but not exclusively on serine, threonine, and tyrosine residues at the histone tails [bib_ref] Regulation of chromatin by histone modifications, Bannister [/bib_ref] , has gained considerable attention, especially regarding the histone H3, due to its close association with mitotic chromosome condensation in mammalian cells [bib_ref] Histone H3 phosphorylation-A versatile chromatin modification for different occasions, Sawicka [/bib_ref]. A preliminary study evaluating the mitotic index, based on H3 phosphorylation in LMS, found Phospho Histone H3 (PHH3) positive staining to be a promising mitosis-specific marker for this tumor [bib_ref] Mitosis-specific marker phospho-histone h3 in the assessment of mitotic index in uterine..., Veras [/bib_ref].
Bromodomain-containing proteins (BRDs), as the "readers" of lysine acetylation are responsible for transducing regulatory signals carried by acetylated lysine residues into various biological phenotypes. BRDs can exert a wide variety of functions via multiple gene regulatory mechanisms [bib_ref] Functions of bromodomain-containing proteins and their roles in homeostasis and cancer, Fujisawa [/bib_ref] and the deregulation of BRDs is involved in many diseases, including cancer [bib_ref] The genetic alteration spectrum of the SWI/SNF complex: The oncogenic roles of..., Sima [/bib_ref] [bib_ref] Exploring the Value of BRD9 as a Biomarker, Therapeutic Target and Co-Target..., Barma [/bib_ref] [bib_ref] BRD9 defines a SWI/SNF sub-complex and constitutes a specific vulnerability in malignant..., Wang [/bib_ref]. BRD9 is a newly identified subunit of the noncanonical barrier-to-autointegration factor (ncBAF) complex and a member of the bromodomain family IV. Studies have demonstrated that BRD9 plays an oncogenic role in multiple cancer types, by regulating tumor cell growth. Furthermore, the connection of BRD9 with the PI3K pathway [bib_ref] PIK3CA Cooperates with KRAS to Promote MYC Activity and Tumorigenesis via the..., Bell [/bib_ref] , microRNAs [bib_ref] miR-140-3p functions as a tumor suppressor in squamous cell lung cancer by..., Huang [/bib_ref] , and STAT5 [bib_ref] Altucci, L. BRD9 binds cell type-specific chromatin regions regulating leukemic cell survival..., Gaudio [/bib_ref] is implicated in cancer progression. It has been shown that BRD9 is aberrantly overexpressed in uterine LMS tissues, compared to adjacent myometrium.. In addition, BRD9 expression was upregulated in uterine LMS cell lines compared to benign LM and myometrium cell lines. Notably, targeting the BRD9 with a specific chemical inhibitor (TP-472) can suppress the LMS cell growth, concomitantly sculpting the transcriptome of uterine LMS cells, altering the important pathways, reprogramming the oncogenic epigenome, and inducing the miRNA-mediated gene regulation. These studies reveal that BRD9 constitutes a specific vulnerability in malignant LMS and that targeting non-bromodomain and extra-terminal BRDs in uterine LMS may provide a promising and novel strategy for treating patients with this aggressive uterine cancer.
In summary, histone modifications are frequently found in US, thus representing potential targets for new therapeutic strategies development. Several studies have demon-strated that HDAC inhibitors could modulate several signaling pathways, activating, or inhibiting numerous cascades that lead to an antitumor response. Moreover, combination with chemo-or targeted-therapies is likely to strengthen the activity of HDAC inhibitors. However, it is still necessary to further elucidate how the histone modifications are regulated, as well as to understand the mechanism of action of their available inhibitors in LMS and ESS. This information will enable more efficient clinical trials, which could lead to an improvement in patients' response to treatment and overall survival [bib_ref] Assessment of synergistic contribution of histone deacetylases in prognosis and therapeutic management..., Mastoraki [/bib_ref].
## Non-coding rna (ncrnas)
ncRNA is known to regulate gene expression both at transcriptional and post-transcriptional levels. ncRNAs play an important role in epigenetic processes, including modulation of heterochromatin, histone modification, DNA methylation, and gene silencing [fig_ref] Figure 5: Graphical representation of the ncRNAs dysregulation in LMS and ESS [/fig_ref] [bib_ref] Cancer epigenetics: An introduction, Kanwal [/bib_ref]. These molecules can be divided into housekeeping and regulatory ncRNAs [bib_ref] Non-coding RNAs as regulators in epigenetics (Review), Wei [/bib_ref] [bib_ref] Long non-coding rnas as epigenetic regulators in cancer, Vafadar [/bib_ref]. Regulatory ncRNAs are classified according to their size in small non-coding RNAs (sncRNAs), with approximately 19-200 nucleotides (nt), and long non-coding RNAs (lncRNAs), with more than 200 nt [bib_ref] Epigenetic research in cancer epidemiology: Trends, opportunities, and challenges, Verma [/bib_ref] [bib_ref] Cancer epigenetics: An introduction, Kanwal [/bib_ref] [bib_ref] DNA methylation machinery in the endometrium and endometrial cancer, Caplakova [/bib_ref] [bib_ref] Interplay between long non-coding RNAs and epigenetic machinery: Emerging targets in cancer?, Hanly [/bib_ref] [bib_ref] Small non-coding RNA and cancer, Romano [/bib_ref]. The sncRNAs have a wide range of structural and functional roles in gene expression regulation, RNA splicing, and chromatin structure [bib_ref] Small noncoding rna expression in cancer, Guisier [/bib_ref] [bib_ref] Noncoding RNA genes, Eddy [/bib_ref]. sncRNAs includes four different categories: (1) small interfering RNA (siRNA), (2) microRNA (miRNA), (3) PIWI-interfering RNA (piRNA, with approximately 19-31 nt), and (4) small nucleolar RNA (snoRNA, with 60-300 nt) [bib_ref] Alterations of histone modifications in cancer, Hosseini [/bib_ref] [bib_ref] Long non-coding rnas as epigenetic regulators in cancer, Vafadar [/bib_ref]. miRNA and piRNA are probably the most studied sncRNAs categories to date, and their functions are well established in the literature [bib_ref] Small noncoding rna expression in cancer, Guisier [/bib_ref]. Due to miRNAs' broad roles, mainly at the post-transcriptional level, dysfunctions in their regulation have been associated with the development of several diseases, including cancer [bib_ref] Small non-coding RNA and cancer, Romano [/bib_ref] [bib_ref] Small noncoding rna expression in cancer, Guisier [/bib_ref] [bib_ref] MicroRNA expression profiles classify human cancers, Lu [/bib_ref]. To understand the complex biology of sarcomas, numerous correlative and functional studies aiming to integrate gene expression patterns and miRNAs have been carried out [bib_ref] MicroRNA-mediated gene regulations in human sarcomas, Subramanian [/bib_ref]. As mentioned above, the differential diagnosis of LMS is still a challenge, and studies focusing on new biomarkers to help distinguish uterine LMS from LM are extremely important [bib_ref] The new horizon of liquid biopsy in sarcoma: The potential utility of..., Wei [/bib_ref] [bib_ref] Liquid biopsy as a tool for differentiation of leiomyomas and sarcomas of..., Dvorská [/bib_ref] [bib_ref] Abstract 5242: miR-10b functions as a novel tumor suppressor in uterine leiomyosarcoma..., Anderson [/bib_ref] [bib_ref] MiRNAs 144-3p, 34a-5p, and 206 are a useful signature for distinguishing uterine..., Schiavon [/bib_ref]. [bib_ref] Serum microRNA profile enables preoperative diagnosis of uterine leiomyosarcoma, Yokoi [/bib_ref] demonstrate the feasibility of circulating serum miRNAs detection as a preoperative clinical assay to detect US. They identified two miRNA signatures (miR-1246 and miR-191-5p) in uterine LMS (95% confidence interval of 0.91-1.00) [bib_ref] Serum microRNA profile enables preoperative diagnosis of uterine leiomyosarcoma, Yokoi [/bib_ref]. [bib_ref] Liquid biopsy as a tool for differentiation of leiomyomas and sarcomas of..., Dvorská [/bib_ref] , and later, Wei et al. (2020) reviewed how liquid biopsies could increase the overall understanding of uterine LMS behavior and how its molecular profile could contribute to more accurate discrimination from LM [bib_ref] The new horizon of liquid biopsy in sarcoma: The potential utility of..., Wei [/bib_ref] [bib_ref] Liquid biopsy as a tool for differentiation of leiomyomas and sarcomas of..., Dvorská [/bib_ref].
Comparing LMS, LM, and MM, [bib_ref] Abstract 5242: miR-10b functions as a novel tumor suppressor in uterine leiomyosarcoma..., Anderson [/bib_ref] found 37 miRNAs differentially expressed in uterine LMS. The lack of miR-10b in LMS samples was critical for tumor growth and metastasis. Indeed, rescuing miR-10b expression in the cell lines resulted in prominent inhibition of cell proliferation, migration, and invasion, and increased apoptosis. Similarly, stable miR-10b expression significantly reduced the number and size of tumor implants in vivo by reducing cell proliferation and increasing apoptosis [bib_ref] Abstract 5242: miR-10b functions as a novel tumor suppressor in uterine leiomyosarcoma..., Anderson [/bib_ref].
Later, [bib_ref] MiRNAs 144-3p, 34a-5p, and 206 are a useful signature for distinguishing uterine..., Schiavon [/bib_ref] found that dysregulation of miR-148a-3p, 27b-3p, 124-3p, 183-5p, and 135b-5p expression was associated with tumor relapse, increased metastasis, and poor survival rates in uterine LMS patients [bib_ref] MiRNAs 144-3p, 34a-5p, and 206 are a useful signature for distinguishing uterine..., Schiavon [/bib_ref]. De Almeida et al. (2017) evaluated the miRNAs expression profile in cell lines of MM, LM, and LMS. Thirteen molecules presented differential expression profiles in LM and LMS, compared to normal tissue (MM). Additionally, the authors observed that miR-1-3p, miR-130b-3p, miR-140-5p, miR-202, miR-205, and miR-7-5p presented similar expression patterns between the cell lines and 16 patients' samples [bib_ref] Oncomirs expression profiling in uterine leiomyosarcoma cells, De Almeida [/bib_ref]. [bib_ref] Molecular analyses of 6 different types of uterine smooth muscle tumors: Emphasis..., Zhang [/bib_ref] demonstrated that miRNAs were significantly dysregulated among different types of uterine smooth muscle tumors (USMTs), including ordinary LM, mitotically active leiomyoma (MALM), cellular leiomyoma (CLM), atypical leiomyoma (ALM), uterine smooth muscle tumor of uncertain malignant potential (STUMP), and LMS samples. The miRNA expression profile showed that ALM and LMS shared similar signatures (including miR-34a-5p, miR-10b-5p, miR-21-5p, miR-490-3p, miR-26a-5p and miR-650). Unsupervised analysis divided the tumors into three clusters: LMS/ALM, LM/STUMP, and CLM/MM [bib_ref] Molecular analyses of 6 different types of uterine smooth muscle tumors: Emphasis..., Zhang [/bib_ref]. miR-200c was found to be significantly downregulated in LM, compared to MM [bib_ref] miR-200c is aberrantly expressed in leiomyomas in an ethnic-dependent manner and targets..., Chuang [/bib_ref] , acting directly in ZEB1/2, VEGFA, FBLNS, and TIMP2 regulation. Next, the authors observed a significant reduction of miR-200c in the SK-LMS-1 cells, compared to isolated LM cells, indicating this miRNA is an important marker for LM progression and malignance risk [bib_ref] Malignant tumors of the uterine corpus: Molecular background of their origin, Brany [/bib_ref] [bib_ref] The regulatory function of mir-200c on inflammatory and cell-cycle associated genes in..., Chuang [/bib_ref].
To date, the differential expression (i.e., up-or down-regulation) of several miRNAs has been directly correlated with US patients' prognosis. In 2018, Dos Anjos et al. analyzed the expression profile of 84 cancer-related miRNAs and associated their signatures with patients' clinical and pathological data. In LMS, specifically, the authors found an association between miRNA dysregulation and lower cancer-specific survival (CSS) and aggressive tumor phenotype. In ESS samples, alterations in miRNA regulation were related to both lower CSS and metastasis. [bib_ref] Let-7 repression leads to HMGA2 overexpression in uterine leiomyosarcoma, Shi [/bib_ref] found a significant inverse correlation between endogenous HMGA2 levels and let-7 expression in uterine LMS. Their study revealed that the ectopic expression of let-7a inhibits LMS proliferation by HMGA2 repression, suggesting that the let-7 loss of expression can represent a worse prognostic factor [bib_ref] Let-7 repression leads to HMGA2 overexpression in uterine leiomyosarcoma, Shi [/bib_ref]. [bib_ref] Profiling and functional analyses of microRNAs and their target gene products in..., Zavadil [/bib_ref] identified the way in which let-7s is responsible for the direct regulation of PPP1R12B, STARD13, TRIB1, BTG2, HMGA2, and ITGB3 genes (involved in the cell proliferation and extracellular matrix regulation) in LM samples. [bib_ref] Profiling and functional analyses of microRNAs and their target gene products in..., Zavadil [/bib_ref]. De Almeida et al. (2019), found that decreased expression of let-7 family members was directly correlated with worse prognosis, affecting both the overall survival (OS) and the DFS rates of the LMS patients [bib_ref] Let-7 miRNA's expression profile and its potential prognostic role in uterine leiomyosarcoma, De Almeida [/bib_ref].
Dysregulation of some miRNAs has also been correlated with acquired chemoresistance in uterine neoplasm. For instance, the loss of miR-34a expression and its release from LMS cells via exosomes contribute indirectly to the tumor doxorubicin chemoresistance. This mechanism seems to be mediated by MELK overexpression and the recruitment of M2 macrophages [bib_ref] Molecular analyses of 6 different types of uterine smooth muscle tumors: Emphasis..., Zhang [/bib_ref].
Although less studied than other ncRNAs, lncRNAs are known to interact with either DNA, RNA, or proteins, and play a significant regulatory function in several cellular processes [bib_ref] MicroRNA expression profiles classify human cancers, Lu [/bib_ref]. lncRNAs are responsible for regulating transcription on three different levels: pre-transcriptional (chromatin remodeling), transcriptional, and posttranscriptional [bib_ref] Long non-coding rnas as epigenetic regulators in cancer, Vafadar [/bib_ref] [bib_ref] The function of lncrnas as epigenetic regulators, Ayub [/bib_ref]. Some similarities can be found between lncRNAs and mRNAs, including size, transcription by RNA pol-II, 5 -capping, RNA splicing, and poly(A) tail (approximately 60% of all lncRNAs) [bib_ref] Non-coding RNAs in cancer: Platforms and strategies for investigating the genomic "dark..., Grillone [/bib_ref]. LncRNAs can be stratified into five categories: (1) intergenic (present between two protein-coding genes), [bib_ref] Epidemiological and genetic clues for molecular mechanisms involved in uterine leiomyoma development..., Commandeur [/bib_ref] intronic (between the introns of a protein-coding gene), (3) overlapping (a coding gene is located on the intron of a lncRNA), (4) antisense (the lncRNA is transcribed from the opposite strand of a proteincoding gene), and (5) processed lncRNAs (lacks an open reading frame ORF) [bib_ref] MicroRNA expression profiles classify human cancers, Lu [/bib_ref] [bib_ref] Non-coding RNAs in cancer diagnosis and therapy, Rasool [/bib_ref]. lncRNA can be expressed in distinguished cell regions and their functions are directly related to their sub-cellular location. However, these epigenetic regulators may suffer molecular alterations that affect their expression and, consequently, their physiological function. Accumulated evidence shows that several differentially expressed lncRNAs are related to cancer development, progression, and metastasis [bib_ref] Interplay between long non-coding RNAs and epigenetic machinery: Emerging targets in cancer?, Hanly [/bib_ref] [bib_ref] The function of lncrnas as epigenetic regulators, Ayub [/bib_ref].
Unfortunately, in uterine LMS and ESS, the molecular role of lncRNAs and their regulation remains unclear. Yet, [bib_ref] Integrated analysis of long noncoding RNAs and mRNAs reveals their potential roles..., Guo [/bib_ref] performed a microarray-based genomewide analysis of lncRNAs, including 35 LM and MM-matched samples. The authors showed, for the first time, the differential expression profile of the lncRNAs between these tissues. The expression pattern obtained was associated with the downregulation of the cytokine-cytokine receptor interaction pathway in large LM, and the upregulation of the fatty acid metabolism pathway in small LM. This study, although preliminary, sheds light on future studies that will attempt to elucidate the role of lncRNAs specifically in uterine mesenchymal tumors [bib_ref] Integrated analysis of long noncoding RNAs and mRNAs reveals their potential roles..., Guo [/bib_ref].
# Conclusions
Uterine pure sarcomas constitute the most frequently diagnosed group of malignant neoplasms in the uterine body. LMS and ESS are distinct tumors with a variety of features similar to other uterine neoplasms. The high heterogeneity, morphological and molecular variations pose challenges to subtypes differentiation and diagnosis. The origin of these tumors remains unclear, as well as the molecular mechanisms that drive their clinical and biological behavior. However, genetic, and epigenetic mechanisms have been shown to directly and indirectly influence the USMT malignant transformation, but the high complexity of this group of tumors still represents a barrier to diagnosis and disease management. In this review, we provided insights into the most recent studies regarding epigenetic events in LMS and ESS, and their potential as novel biomarkers or for developing new therapeutic modalities to treat these tumors. [CrossRef]
[fig] Figure 1: The estimated incidence of gynecological tumors for 2022 according to the ACS. [/fig]
[fig] Figure 2: Graphical representation of epigenetic events potentially involved in the initiation and development of the tumors, including uterine LMS and ESS. The biogenesis of miRNAs starts in the nucleus and ends in the cytoplasm. This process includes the participation of several enzymes and protein complexes that regulate the production of mature molecules capable of regulating gene expression, both through induction of mRNA degradation and translational repression. Likewise, dynamic alterations of histone modifications, including acetylation, methylation, and phosphorylation, modify gene expression, thus affecting DNA replication and repair, chromatin compaction, and cell cycle control. In addition, histone modification readers such as BRDs can recognize modified histones, therefore altering gene expression and responding to different signals. Dysregulation in the epigenetic machinery leads to malignant transformation of cells culminating in the development of cancer. Created with BioRender.com. [/fig]
[fig] Figure 3: Schematic representation of the DNA methylation process in LMS and ESS during cancer development and progression. Methyl groups are added to the DNA molecule and change its activity. Promoter hypermethylation has been shown to silence tumor suppressor genes in cancer cells, leading to either dysregulation of cell growth or inducing resistance to cancer therapies. Hypomethylation promotes genomic instability causing missegregation of chromosomes during cell division. Created with BioRender.com. [/fig]
[fig] Figure 4: Chromatin remodeling process in LMS and ESS. Chromatin remodeling is an important mechanism of gene expression regulation. In the histone acetylation induced by HATs, the condensed chromatin is transformed into a more relaxed structure (euchromatin) that is associated with greater levels of gene transcription, while histone hypoacetylation induced by HDAC activity is associated with more condensed chromatin (heterochromatin), inducing gene silencing. Altered expression and mutations of genes that encode HDACs have been linked to tumor development. Created with BioRender.com. [/fig]
[fig] Figure 5: Graphical representation of the ncRNAs dysregulation in LMS and ESS. ncRNAs have been identified as oncogenic drivers or tumor suppressors in cancer. LncRNAs often affect the expression of their target genes by interacting with miRNAs, which are the main post-transcriptional regulation factors. Some lncRNAs act like sponges, thereby preventing miRNAs from binding to their target mRNAs. As lncRNAs work as decoys for miRNAs, oncogene mRNA translation is allowed, starting the LMS and ESS carcinogenesis. Created with BioRender.com. [/fig]
[fig] Author: Contributions: B.C.d.A. and L.G.d.A. literature review, data collection and manuscript preparation; A.S.D., E.C.B., Q.Y. and A.A.-H. manuscript review, intellectual and technical support; K.C.C. idea conception, data collection and manuscript preparation, and review. All authors have read and agreed to the published version of the manuscript. [/fig]
[fig] Funding: This research was funded by Fundação De Amparo à Pesquisa do Estado Do São Paulo (FAPESP 2019/01109-2) and Coordenação de Aperfeiçoamento de Pessoal De Nível Superior-Brasil (CAPES). Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Not applicable. Acknowledgments: Not applicable. Conflicts of Interest: The authors declare no conflict of interest. 29. Meng, Y.; Yang, Y.; Zhang, Y.; Li, X. Construction and validation of nomograms for predicting the prognosis of uterine leiomyosarcoma: A population-based study. Med. Sci. Monit. 2020, 26, e922739-1. [CrossRef] 30. Mittal, K.R.; Chen, F.; Wei, J.J.; Rijhvani, K.; Kurvathi, R.; Streck, D.; Dermody, J.; Toruner, G.A. Molecular and immunohistochemical evidence for the origin of uterine leiomyosarcomas from associated leiomyoma and symplastic leiomyoma-like areas. Mod. Pathol. 2009, 22, 1303-1311. [CrossRef] [PubMed] 31. Yasutake, N.; Ohishi, Y.; Taguchi, K.; Hiraki, Y.; Oya, M.; Oshiro, Y.; Mine, M.; Iwasaki, T.; Yamamoto, H.; Kohashi, K.; et al.Insulin-like growth factor II messenger RNA-binding protein-3 is an independent prognostic factor in uterine leiomyosarcoma.Histopathology 2017, 12, 3218-3221. [CrossRef] 32. Patel, D.; Handorf, E.; von Mehren, M.; Martin, L.; Movva, S. Adjuvant chemotherapy in uterine leiomyosarcoma: Trends and factors impacting usage. Sarcoma 2019, 2019, 3561501. [CrossRef] [PubMed] 33. Mittal, K.; Joutovsky, A. Areas with benign morphologic and immunohistochemical features are associated with some uterine leiomyosarcomas. Gynecol. Oncol. 2007, 104, 362-365. [CrossRef] [PubMed] 34. Banas, T.; Pitynski, K.; Okon, K.; Czerw, A. DNA fragmentation factors 40 and 45 (DFF40/DFF45) and B-cell lymphoma 2 (Bcl-2) protein are underexpressed in uterine leiomyosarcomas and may predict survival. OncoTargets Ther. 2017, 10, 4579-4589. [CrossRef] 35. Williams, A.R.W. Uterine fibroids-What's new? F1000Research 2017, 6, 2109. [CrossRef] [PubMed] 36. Zhang, P.; Zhang, C.; Hao, J.; Sung, C.J.; Quddus, M.R.; Steinhoff, M.M.; Lawrence, W.D. Use of X-chromosome inactivation pattern to determine the clonal origins of uterine leiomyoma and leiomyosarcoma. Hum. Pathol. 2006, 37, 1350-1356. [CrossRef] [PubMed] 37. Oh, J.; Park, S.B.; Park, H.J.; Lee, E.S. Ultrasound features of uterine sarcomas. Ultrasound Q. 2019, 35, 376-384. [CrossRef] 38. Chen, L.; Yang, B. Immunohistochemical analysis of P16, P53, and Ki-67 expression in uterine smooth muscle tumors. Int. J. Gynecol. Pathol. 2008, 27, 326-332. [CrossRef] 39. Loizzi, V.; Cormio, G.; Nestola, D.; Falagario, M.; Surgo, A.; Camporeale, A.; Putignano, G.; Selvaggi, L. Prognostic factors and outcomes in 28 cases of uterine leiomyosarcoma. Oncology 2011, 81, 91-97. [CrossRef] 40. Devereaux, K.A.; Schoolmeester, J.K. Smooth muscle tumors of the female genital tract. Surg. Pathol. Clin. 2019, 12, 397-455. [CrossRef] 41. Byar, K.L.; Fredericks, T. Uterine leiomyosarcoma. J. Adv. Pract. Oncol. 2022, 13, 70-76. [CrossRef] 42. Hoang, H.L.T.; Ensor, K.; Rosen, G.; Leon Pachter, H.; Raccuia, J.S. Prognostic factors and survival in patients treated surgically for recurrent metastatic uterine leiomyosarcoma. Int. J. Surg. Oncol. 2014, 2014, 919323. [CrossRef] 43. Friedman, C.F.; Hensley, M.L. Options for adjuvant therapy for uterine leiomyosarcoma. Curr. Treat. Options Oncol. 2018, 19, 7. [CrossRef] 44. Stope, M.B.; Cernat, V.; Kaul, A.; Diesing, K. Functionality of the tumor suppressor microrna-1 in malignant tissue and cell line cells of uterine leiomyosarcoma. Anticancer Res. 2018, 1550, 1547-1550. [CrossRef] 45. Rizzo, A.; Pantaleo, M.A.; Saponara, M.; Nannini, M. Current status of the adjuvant therapy in uterine sarcoma: A literature review. World J. Clin. Cases 2019, 7, 1753. [CrossRef] [PubMed] 46. Li, Y.; Ren, H.; Wang, J. Outcome of adjuvant radiotherapy after total hysterectomy in patients with uterine leiomyosarcoma or carcinosarcoma: A SEER-based study. BMC Cancer 2019, 19, 697. [CrossRef] 47. Amant, F.; Lorusso, D.; Mustea, A.; Duffaud, F.; Pautier, P. Management strategies in advanced uterine leiomyosarcoma: Focus on trabectedin. Sarcoma 2015, 2015, 704124. [CrossRef] [PubMed] 48. Barlin, J.N.; Zhou, Q.C.; Leitao, M.M.; Bisogna, M.; Olvera, N.; Shih, K.K.; Jacobsen, A.; Schultz, N.; Tap, W.D.; Hensley, M.L.; et al. Molecular subtypes of uterine leiomyosarcoma and correlation with clinical outcome. Neoplasia 2015, 17, 183-189. [CrossRef] 49. Ben-Ami, E.; Barysauskas, C.M.; Solomon, S.; Tahlil, K.; Malley, R.; Hohos, M.; Polson, K.; Loucks, M.; Severgnini, M.; Patel, T.; et al. Immunotherapy with single agent nivolumab for advanced leiomyosarcoma of the uterus: Results of a phase 2 study. Cancer 2017, 123, 3285-3290. [CrossRef] [PubMed] 50. Elvin, J.A.; Gay, L.M.; Ort, R.; Shuluk, J.; Long, J.; Shelley, L.; Lee, R.; Chalmers, Z.R.; Frampton, G.M.; Ali, S.M.; et al. Clinical benefit in response to palbociclib treatment in refractory uterine leiomyosarcomas with a common CDKN2A alteration. Oncologist 2017, 22, 416-421. [CrossRef] 51. American Cancer Society Survival. Rates for Uterine Sarcoma. Available online: https://www.cancer.org/cancer/uterinesarcoma/detection-diagnosis-staging/survival-rates.html (accessed on 13 January 2022). 52. Kobayashi, H.; Uekuri, C.; Akasaka, J.; Ito, F.; Shigemitsu, A.; Koike, N.; Shigetomi, H. The biology of uterine sarcomas: A review and update. Mol. Clin. Oncol. 2013, 1, 599-609. [CrossRef] [PubMed] 53. Baek, M.-H.; Park, J.-Y.; Park, Y.; Kim, K.-R.; Kim, D.-Y.; Suh, D.-S.; Kim, J.-H.; Kim, Y.-M.; Kim, Y.-T.; Nam, J.-H. Androgen receptor as a prognostic biomarker and therapeutic target in uterine leiomyosarcoma. J. Gynecol. Oncol. 2018, 29, e30. [CrossRef] 54. Slomovitz, B.M.; Taub, M.C.; Huang, M.; Levenback, C.; Coleman, R.L. A randomized phase II study of letrozole vs. observation in patients with newly diagnosed uterine leiomyosarcoma (uLMS). Gynecol. Oncol. Rep. 2019, 27, 1. [CrossRef] 55. Chiang, S.; Oliva, E. Recent developments in uterine mesenchymal neoplasms. Histopathology 2013, 62, 124-137. [CrossRef] 56. Ducie, J.A.; Leitao, M.M., Jr. The role of adjuvant therapy in uterine leiomyosarcoma. Expert Rev. Anticancer Ther. 2016, 16, 45. [/fig]
[table] Table 1: Staging of LMS and ESS (FIGO 1 and AJCC 2 ). [/table]
[table] Table 2: Molecular features of endometrial stromal tumors (ESTs). [/table]
|
A low-cost smart system for electrophoresis-based nucleic acids detection at the visible spectrum
Nucleic acid detection by electrophoresis is still a quick and accessible technique for many diagnosis methods, primarily at research laboratories or at the point of care units. Standard protocols detect DNA/RNA molecules through specific bound chemical dyes using a UVtransilluminator or UV-photo documentation system. However, the acquisition costs and availability of these devices, mainly the ones with photography and internet connection capabilities, can be prohibitive, especially in developing countries public health units. Also, ultraviolet radiation is a common additional risk factor to professionals that use electrophoresisbased nucleic acid detection. With that in mind, this work describes the development of a lowcost DNA/RNA detection smart system capable of obtaining qualitative and semi-quantitative data from gel analysis. The proposed device explores the visible light absorption range of commonly used DNA/RNA dyes using readily available parts, and simple manufacturing processes, such as light-emitting diodes (LEDs) and 3D impression. By applying IoT techniques, our system covers a wide range of color spectrum in order to detect bands from various commercially used dyes, using Bluetooth communication and a smartphone for hardware control, image capturing, and sharing. The project also enables process scalability and has low manufacturing and maintenance costs. The use of LEDs at the visible spectrum can achieve very reproducible images, providing a high potential for rapid and point-of-care diagnostics as well as applications in several fields such as healthcare, agriculture, and aquaculture.OPEN ACCESSCitation: Cunha EN, de Souza MFB, Lanza DCF, Lima JPMS (2020) A low-cost smart system for electrophoresis-based nucleic acids detection at the visible spectrum. PLoS ONE 15(10): e0240536.
# Introduction
Nucleic acid detection by gel electrophoresis is a ubiquitous laboratory routine procedure used in several fields, like genetics and molecular biology, biochemistry, and forensic science, and an essential step in extraction, cloning, and PCR workflows. In the last decades, its increasing application allied with more accurate molecular diagnosis techniques has provided useful information on the general condition of patients, as well as contributing to the diagnosis and prognosis of various diseases. Despite its simplicity, equipment and analysis costs are still high, especially in low resource settings (LRS) or for point-of-care (POC) applications. The most used chemical dye for gel electrophoresis-based DNA/RNA detection is Ethidium Bromide (EtBr). This dye binds to DNA and, in the presence of ultraviolet (UV) light between wavelength 260 and 360 nm, fluoresces in orange-red range, in the length of 590 nm, being able to detect as little as 10 ng of DNA. However, the use of EtBr presents several disadvantages, among them, is the fact that it is highly mutagenic and carcinogenic, as well as its requirement for UV light exposure for detection. Non-toxic fluorescent dyes are good alternatives to EtBrnucleic acid staining, such as Methylene Blue, which has no mutagenic potential and no UV light need, although it is significantly less sensitiveand Sybr Green, which exhibits about the same sensitivity as the EtBr and can be visualized in the blue or UV light. More recently, there is also a line of more efficient and safe staining-ready gels, such as GelRed and GelGreen. However, they are expensive alternatives, and most laboratories, especially in developing countries, still use EtBr, despite its disadvantages.
The use of EtBr also popularized UV transilluminators for the visualization of nucleic acid gel electrophoresis. This device generates light at UV wavelengths to excite the fluorophore present in the gel, allowing the visualization of the molecule to which it is attached. However, these wavelengths are not the same for all DNA/RNA specific dyes. Therefore, most equipment available on the market is not specific for all applications in research and clinical laboratories. Although it is a conceptually simple instrument, transilluminators have a high cost, about U$ 500.00 to 2,000.00, even without adding photodocumentary capabilities, which is usually only justified by its application. Moreover, camera-coupled transilluminators usually need to be connected to a computer and positioned on a transilluminator or have an integrated chamber for the acquisition of images. The associated costs, coupled with its low portability, often make it infeasible to use in places with limited structures such as LRS laboratories or POC health units in developing countries.
With the advancement of technology and its availability, it is possible to develop systems that previously had a very high cost at very competitive prices in the market. Also, the growth of "do it yourself" mentality, the increasing accessibility to top-end technologies, and the Internet of Things (IoT), all have broken paradigms of connectivity and development of electronic products. The most significant search today in all areas is for integrated, portable, low energy, automatic, and low-cost smart devices. In the laboratory environment, simple technologies combined with excellent engineering had allowed a reduction of up to 20 times in the values of some equipment. At present, easily accessible materials can allow the development of thermal cyclers for PCR, electrophoresis apparatus, and nucleic acid direct detection systems. With that in mind, the aim of the present work was the development of a low-cost DNA/RNA detection smart system capable of obtaining qualitative and semi-quantitative data from gel electrophoresis analysis from different available dyes. With easily accessible material and technologies, as light-emitting diodes (LEDs), miniaturized computing, and 3D impression, and also using a smartphone and custom software, it was possible to obtain a capable device to substitute a conventional commercial UV-transilluminator in several applications.
# Materials and methods
The project of the device comprises two parts, software development and the system hardware per si, connected by Bluetooth protocols.
## Hardware
The system hardware consists of a microcontroller platform with wifi and Bluetooth connections. It also includes an adaptable controller to various dye types, selecting the wavelength emitted by the lighting system for the type of dye used in the experiment, summarized in "". Lighting intensity control is also implemented in hardware to optimize image capture, as shown in.
Lighting system. For the excitation of the dyes, the system uses fourteen 3 W RGB LED chips (3x1W, with six K1763 terminals), with a coupled heat sink, divided into three separate sets. Each set has unevenly powered red, blue, and green LEDs. This division in the electric
## Plos one
power does not happen in the same way (as will be described in the elaboration of the control circuit of the led) since the voltage of the red LED matrix is smaller. This uneven power division compensated the amount of lux emitted by the red LED and permitted an equal luminous power. The wavelength range of these LEDs is 620-630 nm, 520-530 nm, and 460-470 nm, for red, green, and blue, respectively. S1 File describes the LEDs full specification. Electronic circuit. For the design of the LED's control circuit, we used a pulse width modulation (PWM) system. To obtain control and optimize the lighting system according to the arrangement of the box, we used one PWM for each set of LEDs. The LED chip RGB consists of three parallel LEDs, mounted with a common anode, which need the same voltage (24 V) and current (600 mA) for each color matrix in each LED. The drive circuit has taken into account the switching speed for pulse modulation so that it is possible to obtain the full-color spectrum within the visible, and there is no excess energy dissipation by Joule effect, not compromising the durability of the card and components.
The microcontroller used was the ESP32 (Espressif), operating at 0.8 mA current at the port, well below its maximum capacity of 12 mA. The system's project also included touch screen buttons and signal operation LEDs mounted to an auxiliary circuitry. For controlling the LED cooling system, aluminum heat sinks were mounted directly on the LED board. The circuit contained correctly dimensioned components to avoid overload of the LED chips during use.describes the circuit project and shows the components already sized. 3D printing. To maintain a low cost and achieve a straightforward reproduction, we opted to use 3D printing for the development of the platform's on-site photographic documentation system.shows the final prototype. The calculation of the size and angles for the smartphone support depends on its camera's specifications and must guarantee that the entire illuminated area is covered. In the tested prototype, we used the following measures:
� The measures of the support.
■ Height 56 mm, width 120 mm, and depth 145 mm.
■ Minimum distance for image capture: 90 mm. White arrows indicate the filter lid, the status LEDs, the gel chamber, the power outlet, and the "SET" and "OK" buttons for manual operation of the system. S1 File includes simple calculations for the support's measures, for standard cameras available. The constructed prototype can accommodate gels in any range according to the calculations presented. The inclusion of the SET and OK buttons allows the system's operation as a simple transilluminator, without using a smartphone, though power and wavelength selections are software-managed only. S1 File describes the CAD drawing model for 3D printing.
## Software
The system includes two operating software. The ESP32 microcontroller software, written in C programming language, has the function of controlling the pulse width modulation system of the LED chips, the communication with the smartphone via Bluetooth low energy (BLE), and the touch buttons. The Android smartphone application is responsible for connecting to the ESP32 microcontroller via Bluetooth, as well as managing the type of dye used, its frequency (color) and intensity with the emitted light, and image capture. There is also an included function to execute an automated wavelength scanning process. At this step, the user has an option to automatically or manually validate the captured image. Another function implemented in smartphone software along with hardware is image capture optimization, to synchronize the camera frequency with the pulse cycle of the used LEDs. After the image validation process, the software sends data for storage or remote analysis. This application, developed in an online block programming environment at Thunkable1, also controls the included photography and sharing functions.
## Sample preparations and gel electrophoresis
The DNA used was the PCR product of a conserved fragment of 18S DNA from the shrimp (Penaeus vannamei) genome. This fragment has a similar pattern in almost all decapod with an 848 bp amplicon, corresponding to the sequence region 352-1200 of the 18S rRNA. The used primers were: 143F 5'-TGC-CTT-ATC-AGCTNT-CGA-TTG-TAG-3' and 145R 5'-TTC-AGN-TTT-GCA-ACC-ATA-CTT-CCC-3' (N represents G, A, T or C). PCR's reactions had a final volume of 75 μL containing: 0.25 μM from each primer, 2.5 mM MgCl 2 , 2.5 mM of each dNTP, 1 U GoTaq1 DNA Polymerase (Promega), and 6 μL of the template DNA. PCR products were quantified using the QuantiFluor One dsDNA system. The concentration of the PCR final product was 38 ng/μL. Eight samples from seven serial dilutions of the PCR final product (38.00, 19.00, 9.50, 4.75, 2.38, 1.19, 0.59, 0.30 ng/uL) were prepared and later observed in three distinct 1% agarose gels, using SYBR Green1, GelRed1, and ethidium bromide as chemical dyes. 6 μL of each diluted sample were applied to the gel's wells, totalizing the following DNA quantities: 228, 114, 57, 28.5, 14.28, 7.14, 3.54, and 1.8 ng.
## System calibration
In order to obtain the best viewing point for each of the dyes, we performed a calibration process using a tracking module implemented in the device's software, which scans the entire region of the visible spectrum with an automatic increment of 10 nm. After that, we compared the results with the spectroscopy provided in each dye manual to ensure the use of its highest excitement regions. For the calibration process, ten independently prepared gels were used for each type of dye, with a more thorough scan for the dye with higher excitation within the UV region. "" describes the power levels for each color LED and the RGB values used for each dye.
# Image analysis
During visualization within the system, all the gels were photographed with the coupled Xiaomi MI 9 SE smartphone (Android Version 10). For comparison reasons, the gels were also visualized and photographed using a 302 nm wavelength standard transilluminator. The resulting raw images (without software treatment) were cropped, then analyzed and compared in ImageJ software. From ImageJ densitometric analyses we calculated the Lower Limit of Detection (LLD) values, as described in.
# Results and discussion
## Assembly costs
We aimed to develop a complete device for DNA/RNA analysis using the electrophoresis technique, commonly used today in many research and health diagnostic laboratories. " " describes the cost of each part. The average assembly cost was US$ 53.76, without including smartphone cost, which is pronouncedly cheaper than every commercially available system
## Plos one
with similar functionalities. Rather than substituting a camera-coupled transilluminator entirely or for all its applications, this system is supposed to aid rapid and early electrophoresis-based diagnostics methods, especially in developing countries, either in POC units or at field or LRS conditions. Since there are different sizes of electrophoresis gels and their respective applications, the project presented here is easily modifiable to accommodate larger gels and can even be adapted to the construction of devices that use capillaries system for detection. Besides, by using a smartphone internet connection, results can be remotely sent to further analysis at reference health centers or laboratories. Regarding its low costs and uncomplicated manufacturer process, another potential application of this system is for science education at public schools, which usually have a low budget for scientific equipment.
## Software management and dye detection
Using RGB LEDs, we were able to get all colors within the visible spectrum by combining the primary colors, red, green, and blue. Therefore, we got a piece of equipment that can excite any commercially available dye, since it has a known visible-light emission wavelength range. As can be evidenced in, most DNA/RNA specific dyes have a low but sufficient excitement within the visible spectrum. For EtBr, a commonly used dye in molecular analysis laboratories, due to its low cost, we achieved excitation, emission, and detection without the use of UV light. Although EtBr has a higher excitement within the UVB (280-315 nm) region, conventional UV lamps used in transilluminators have a considerable displacement, not taking full advantage of this peak region, which is still significantly higher than the excitement region within the visible spectrum. However, our system can excite EtBr molecules at its peak visible absorption, around 480 nm, by using high-luminance LEDs, which obtain a better excitement optimization at this range and partially compensate for this energy difference. The developed smartphone associated software has a screen for selecting the dye used for detection. Afterward, it automatically sets LED colors to maximize dye-specific excitement. Also, since the tracking module scans the entire visible region, the adjustment of the system is possible to ensure the use of dyes' highest excitement regions. Therefore, calibration is essential for further optimization and also for the correct registration of wavelengths for each dye in the smartphone software. . Parts and materials used in the manufacture and assembly of the system, and respective costs (not including the smartphone price).
# Part/material
## Plos one
# Sensitivity analysis
To evaluate the visualization potential of the developed system, we used the same amplified DNA sample dyed with three distinct and commonly used dyes (EtBr, GelRed, and SYBR Green), comparing the results with the visualization using a standard UV-transilluminator. By using smartphone software communication, we selected the highest point of excitation from each dye within the spectrum for the RGB LED, as shown in. Therefore, the used excitation wavelengths were 490 nm (Cian), 500 nm (Green), and 530 nm (Green), for EtBr, GelRed, and SYBR Green, respectively. In the software, it is possible to monitor the device temperature and to make the fine-tune adjustments for dye selection. Initial screen (right). Preset for using ethidium bromide as a dye (left).
https://doi.org/10.1371/journal.pone.0240536.g005
## Plos one
As can be seen in, the visualization of EtBr stained bands is more definite and precise in the photos taken on the UV-transilluminator. As discussed before, this is due to the high adsorption potential of this dye within the UV region. By analyzing the plot of the grayscale difference of bands' densitometryfrom the proposed device, we perceived that it is not possible to view the samples at low concentrations. The lower limit of detection (LLD) values for the EtBr dye were 29.34 and 45.06 ng, using the UV-transilluminator and our LED system, respectively. Since reported EtBr sensitivity values in agarose gels range from 1.0-10.0 ng of DNA, the LLD of the proposed system is adequate for several EtBr applications.
Visually, we observed similar results for the GelRed staining Figs 6C, 6D and 7. Although this visual difference, the estimated LLD for GelRed from our system (47.58 ng) was lower than the calculated value from the UV-transilluminator detection (65.22 ng). This was probably due to the fact that GelRed excitation and detection are more uniform and proportional under visible light. However, both detection systems presented higher LLD values than the reported sensitivity values for GelRed dye. Despite the difficulty of visual
## Plos one
evaluation of the more diluted EtBr and GelRed-dyed bands in photos taken by using the proposed device, it does not invalidate its use for these dyes. Low concentrations are usually not recommended for qualitative or semi-quantitative analysis. Additionally, optimization of protocols and sample concentrations can compensate for this potential weakness, with the advantage of using a device that is safer than conventional UV-transilluminators, since it does not emit UV radiation.
Nevertheless, the device surpasses conventional UV-transilluminators for the use of SYBR green-based protocols. SYBR-stained DNA bands were more brilliant and pronounced in photos taken under the illumination of visible LEDs Figs 6E, 6F and 7. The system takes advantage of the high absorption potential of this dye within the visible region, as noticed by its significantly lower LLD values than the ones from the UV-transilluminator (42.36 ng x 101.58 ng, respectively). Under this chemical staining, it is indeed difficult to evidence more diluted samples under the transilluminator radiation. Although it also has its toxic effects, SYBRgreen and GelRed have surged as a safer alternative for EtBr nucleic acid staining, with the plus of having a more uncomplicated decontamination process.
The use of excitation and emission filters could reduce the background noise and enhance the proposed device's detection capabilities, as evidenced by. However, specific filters addition would increase the device's costs and the overall project complexity. We overcame this by using a red acrylic lid as a filter since all dyes have an emission above 500 nm. Without filter addition, our system can detect other chemical dyes, which remains an accessible, lowcost option for LRS in emerging countries, especially for qualitative molecular tests.
## Functionalities and features of the system
" " summarizes the main features and specifications of the proposed system compared to conventional commercial transilluminators. Our system is a low-cost example of a new generation of transilluminators that aims to apply the concepts of IoT in laboratory equipment.
## Plos one
The insertion of the technology has the clear objective of increasing security both in not using UV radiation but also avoiding the user's contact with the transilluminator, by the use of a smartphone as a control for dye adjustment and image capture and processing.
# Conclusion
The low-cost smart detection system here presented can substitute a conventional UV-transilluminator in several electrophoresis-based applications, allowing the detection of nucleic acid bands staining with three commercial dyes. Besides the low-cost, other advantages of our system are the maximization of dye-specific excitement under the visible light spectrum, portability, and connectivity. The performed analysis also demonstrated that, in some cases, the device could achieve better results than a standard UV-transilluminator for the detection of SYBR green-dyed DNA bands. These features, allied to its easily accomplishable construction, confirm the potential use of this device in low resource settings (LRS) or for point-of-care (POC) and educational applications.
Supporting information S1 File. Supplementary information. Document file describing the calculations for the support's measures for standard cameras available, the LED specifications, the CAD drawings for the system and linear regression plots from the densitometric analyses. (PDF) S1 Raw images. (PDF) |
International medical graduates (IMGs) needs assessment study: comparison between current IMG trainees and program directors
Background: International Medical Graduates (IMGs) training within the Canadian medical education system face unique difficulties. The purpose of this study was to explore the challenges IMGs encounter from the perspective of trainees and their Program Directors.Methods: Program Directors of residency programs and IMGs at the University of Toronto were anonymously surveyed and asked to rate (using a 5-point Likert scale; 1 = least important -5 = most important) the extent to which specific issues were challenging to IMGs and whether an orientation program (in the form of a horizontal curriculum) should be implemented for incoming IMGs prior to starting their residency.
# Background
Several studies note that International Medical Graduates (IMGs) experience additional problems relative to their Canadian counterparts such as loneliness, social isolation, concerns related to family members left behind in home countries, a decrease in social status with an accompanying diminishment of self-esteem, lack of financial resources and worries about visas/immigration issues [bib_ref] The educational needs of IMGs in psychiatric residencies, Kramer [/bib_ref] [bib_ref] Cultural sensitivity training among foreign medical graduates, Majumdar [/bib_ref]. Although some demographic commonalities exist among IMGs (such as generally being older, previously trained and/or practicing, and responsible for supporting a family), heterogeneity is also found due to variations in their life stages and perceptions of health and health care [bib_ref] Untangling the roots of some IMG's poor academic performance, Bates [/bib_ref] [bib_ref] Characteristics of IMGs who applied to the CaRMS 2002 match, Crutcher [/bib_ref]. Furthermore, the level of medical training they have received prior to entering residency varies, which may prove challenging to supervisors. To address the difficulties IMGs face, previous studies find that exposure to an orientation program prior to entering postgraduate training is beneficial [bib_ref] Intensive one-week orientation for foreign medical graduates entering an internal medicine residency..., Rosner [/bib_ref].
However, most orientation programs tend to concentrate on a specific skill, namely communication or cultural-sensitivity training. In a pilot study conducted at one Canadian medical school, IMGs participated in an 18-hour communication program, focusing on doctor-patient and doctor-colleague interaction during the first year of residency. Although the sample size was small, self-evaluations scores showed significant improvement after these sessions [bib_ref] Advanced medical communications: support for international residents, Goldszmidt [/bib_ref]. In another study, IMGs in one residency program were divided into two experimental groups to measure Emotional Resilience, Flexibility, Perceptual Acuity and Personal Autonomy. After cultural-sensitivity training, the experimental group (Group A) scored significantly higher in flexibility, emotional resilience and perceptual acuity than the control group. These IMGs were more self-confident, willing to learn from others, and able to correctly assess verbal and non-verbal cues [bib_ref] Cultural sensitivity training among foreign medical graduates, Majumdar [/bib_ref].
In light of this, a needs assessment study was conducted at the University of Toronto to explore the extent that these specific issues were challenging for IMGs in a residency program from the perspective of Program Directors and foreign-trained physicians themselves.
# Methods
## Survey instrument
Separate surveys were created for IMGs and Program Directors using iterative input from an expert panel of three faculty members. These individuals have worked with IMGs over the past four years with particular emphasis on mentoring, teaching and developing remediation plans.
Ten common issues were posed to both groups and covered the following three topics: Clinical Knowledge and Skills (3), Communication and Working Relations (4), and Macro Issues (3). These issues were identified by the expert panel as being the most pertinent and common across all subspecialties. For the purpose of this study, Macro Issues were defined as areas of concern affecting institutional, environmental or cultural experience and performance.
Currently, there is limited research exploring the impact of macro-level issues as challenges to the medical training of IMGs. Adapting to a new environment involves much more than learning how to communicate effectively. Essentially, the culture and principles, as well as the rules and procedures inherent in everyday operations of an organization, must be learned. Whether this translates to learning how to request a MRI or how to use software to track a patient's medical records, experiencing difficulty in adapting to a new environment may be mitigated by learning about these issues. Hence, under this premise, the survey included macro-level issues such as knowledge of the Canadian Healthcare System, knowledge of the hospital system (e.g. how to use pagers and computer software as well as adapting to hospital protocols) and hospital and pharmaceutical formularies.
Using a 5-point Likert scale, IMGs indicated to what extent each issue posed as a challenge to them during their residency. Program Directors were asked to what extent these issues were challenges to IMGs using the same scale. Further, both groups were asked whether an orientation program, in the form of a horizontal curriculum, was necessary for incoming IMGs. For this study, a horizontal curriculum is defined as a course of study with specific learning objectives for the purpose of obtaining a goal (i.e. providing a smoother transition of IMGs into residency programs at the University of Toronto).
## Recruitment
In the 2005 -2006 academic year, a total of 118 IMGs were enrolled in all residency programs at the University of Toronto, as identified through the Postgraduate Medical Education registration database. IMGs were defined as individuals (regardless of citizenship status and source of funding) who obtained their medical degree outside of Canada at an accredited medical school, and training in a residency program recognized by the Royal College of Physicians and Surgeons of Canada and/or the College of Family Physicians of Canada. With the University of Toronto being the largest medical school in Canada offering almost all specialty and subspecialty residency training programs, 73 residency Program Directors were identified as potential participants.
Due to the number of IMGs and the difficulty in recruiting (i.e. multiple training sites), email invitations were sent out which included a link to complete the survey on-line. To increase the response rate, residents were given the option to include their email address for a chance to win an iPOD.
Email invitations were sent out to all Program Directors with the survey attached, as well as instructions on how to complete the on-line questionnaire. In the event that the former was chosen, Program Directors were given the option to fax or email completed surveys. To encourage the participation of Program Directors, these multiple accommodating methods were offered. Three follow-up emails were sent to IMGs and two follow-up emails were sent to Program Directors.
Participation for both groups was voluntary and subjects were given the option to withdraw at any time. Ethics approval was obtained by the University of Toronto Research Ethics Board for this study.
# Analysis
Frequencies for all eleven close-ended questions were calculated, as well as the mean rating for each of the 10 Likert-scale questions.
# Results
The response rate for IMGs was 74% (87 out of 118) and the response rate for Program Directors was 62% (45 out of 73). Mean scores for all ten topics and corresponding standard deviations are presented in [fig_ref] Table 1: Mean scores of challenges faced by IMGs from the perspective of IMGs... [/fig_ref] Approximately three-quarters of all participants agreed that an orientation program specifically targeted for IMGs is needed. When broken down by group, 93% of all Pro-gram Directors surveyed believed that an orientation program was necessary. A smaller majority (63%) of IMGs believed such a program would be beneficial.
# Conclusions and discussion
Among the IMGs and Program Directors surveyed at our medical school, the majority agreed that an orientation program for all IMGs is required before starting a residency program. Such a program is currently provided for IMGs in the Faculty of Medicine and is reported to be insufficient (Childs and Herbert, Assessing IMG Performance at Ontario Medical Schools 2002-06 Final Report, December 2007). This finding is supported by previous work conducted in the United States. Mylonakis, Mega, and Schiffman [bib_ref] What do program directors in internal medicine think about IMGs? Results of..., Mylonakis [/bib_ref] found that 37% of US Internal Medicine Program Directors agreed that a pre-residency training program should be mandatory prior to IMGs entering training. Other studies by Levey [bib_ref] Internal medicine and the training of IMGs: A time for open discussion..., Levey [/bib_ref] , Kidd and Zulman [bib_ref] Educational support for overseas-trained doctors, Kidd [/bib_ref] and Kramer [bib_ref] The educational needs of IMGs in psychiatric residencies, Kramer [/bib_ref] found similar results.
## Mean scores for basic clinical skills differed between
IMGs and Program Directors and may be related to the variability of training IMGs possess prior to entering the Canadian medical education system. With regards to Communication and Working Relations, mean scores were relatively higher among Program Directors, which may indicate they are more concerned with the communication skills and interprofessionalism of IMGs, in comparison to IMGs themselves. This is consistent with previous literature [bib_ref] Communications skills, cultural challenges, and individual support: Challenges of IMGs in a..., Hall [/bib_ref] [bib_ref] Advanced medical communications: support for international residents, Goldszmidt [/bib_ref].
These findings are not exclusive to foreign-trained physicians. Yahes and Dunn [bib_ref] Enculturation of foreign nurse graduates: An integrated model, Yahes [/bib_ref] found that the biggest challenge experienced by foreign-trained nurses was a lack of communication skills. After completing two-hour sessions over the course of 12 weeks (dealing with group These concurrent surveys of IMG residents and their Program Directors at the University of Toronto illustrated that both cohorts feel that IMGs need better integration into their residency programs. Interestingly, the groups differed in the areas of concern with residents acknowledging their anxiety about a lack of knowledge and comfort in many macro areas, namely institutional and societal transition areas, whereas Program Directors were more concerned with areas of performance indicators in communication, collaboration and basic clinical skills. Each group responded based on their needs and anxieties, possibly indicating that they need to be educated about each other's perspectives.
IMGs are forming a major part of our training cohort, thus representing an increasing number of physicians entering into practice as a short-term solution to health human resource needs. A proliferation of IMG entries has not been accompanied by robust strategies to optimize their integration and bridge their transition into residency. In this study, IMGs have identified specific challenges as have their Program Directors. A comprehensive and integrated program is needed to facilitate the success of IMGs.
# Limitations
The project was designed as a needs assessment study to explore the extent to which specific issues were challenging to IMGs from the perspective of IMG trainees and Program Directors. As it stands, the survey instrument has only face validity.
Secondly, this study is exploratory in nature. As the survey only provided pre-selected issues, participants were restricted in ranking the choices presented. Follow-up studies using focus groups with IMGs may help to expand on their experiences or uncover specifics that cannot be captured using close-ended questions. The same qualitative methods should be used to explore the perceptions of Program Directors, not only for contrast, but also to identify other issues that may not be shared by the IMG cohort.
Conclusions of this study are specific to the perceptions of IMGs and Program Directors of medical residency programs at the Faculty of Medicine at the University of Toronto. It cannot be concluded that IMGs and Program Directors at other Canadian medical schools or elsewhere perceive these issues as challenges, or to the same extent as those who participated in this study.
[table] Table 1: Mean scores of challenges faced by IMGs from the perspective of IMGs and their program directors [/table]
|
Minimization of resource utilization data collected within cost-effectiveness analyses conducted alongside Canadian Cancer Trials Group phase III trials
Background: Cost-effectiveness analyses embedded within randomized trials allow for evaluation of value alongside conventional efficacy outcomes; however, collection of resource utilization data can require considerable trial resources. Methods: We re-analyzed the results from four phase III Canadian Cancer Trials Group trials that embedded costeffectiveness analyses to determine the impact of minimizing potential cost categories on the incremental costeffectiveness ratios. For each trial, we disaggregated total costs into component incremental cost categories and recalculated incremental cost-effectiveness ratios using (1) only the top 3 cost categories, (2) the top 5 cost categories, and (3) all cost components. Using individual trial-level data, confidence intervals for each incremental cost-effectiveness ratio simulation were generated by bootstrapping and descriptively presented with the original confidence intervals (and incremental cost-effectiveness ratios) from the publications. Results: Drug acquisition costs represented the highest incremental cost category in three trials, while hospitalization costs represented the other consistent cost driver and the top incremental cost category in the fourth trial. Recalculated incremental cost-effectiveness ratios based on fewer cost components (top 3 and top 5) did not differ meaningfully from the original published results. Based on conventional willingness-to-pay thresholds (US$50,000-US$100,000 per quality-adjusted life-year), none of the re-analyses would have changed the original perception of whether the experimental therapies were considered cost-effective. Conclusions: These results suggest that the collection of resource utilization data within cancer trials could be narrowed. Omission of certain cost categories that have minimal impact on incremental cost-effectiveness ratio, such as routine laboratory investigations, could reduce the costs and undue burden associated with the collection of data required for cancer trial cost-effectiveness analyses.
# Background
As the costs of cancer care continue to rise, clinical trials will increasingly need to demonstrate that new therapeutic options can offer both clinical benefit and value for money. Cost-effectiveness analyses embedded within randomized trials allow for formal evaluation of value alongside conventional efficacy determinations. Ideally, economic evaluations should capture all relevant costs in a large sample of patients. Ultimately, the perspective of the cost-effectiveness analysis (e.g. health system or societal) contributes to the determination of which resource utilization data should be collected alongside the usual clinical outcome elements.
However, at the level of the clinical trial execution, there is tension with respect to ideal study design to ensure comprehensive data collection without excessive use of trial resources (e.g. study coordinator time) or undue patient burden. Resource utilization data collected in the trials typically include parameters such as drug utilization, hospitalizations, emergency department and outpatient visits, diagnostics, laboratory, and personnel; subsequent cost calculations require that resource utilization parameters be multiplied by the unit cost of that parameter. Resources are collected within specifically designed case report forms and impart additional time and effort for participating patients and study personnel. Such burden may impact compliance with completion of required data elements and ultimately data validity. Following trial completion, central review processes (e.g. data validation, querying) are further required for the data incorporation into the trial database. The added costs and resources required for the completion of a prospective economic evaluation can themselves represent barriers to the conduct of such analyses.
Since trials are limited by the costs associated with patient accrual and with the practical collection of the resource utilization data, it would be important to determine the optimal data requirements to ensure judicious trial performance. The Canadian Cancer Trial Group is one of the only international cancer cooperative groups to maintain a Committee on Economic Analysis; its main aim is to build economic evaluations into phase III cancer trials. Re-analysis of previously completed cost-effectiveness analyses embedded into Canadian Cancer Trials Group trials could potentially clarify the optimal design elements and data requirements for future economic analyses. In particular, review of the costs and resources collected within these trials can determine which high-cost items have the largest impact on the economic sub-study results, including determination of an incremental costeffectiveness ratio (ICER). Moreover, a resource's influence on the overall costs might help determine which parameters should be collected within a trial while at the same time identifying parameters that have minimal impact and do not need to be collected. We hypothesize that the total cost of an intervention in a trial is often driven only by a few resource parameters, often drug acquisition or hospitalization costs, and that other resource parameters such as laboratory costs or diagnostic costs play a small role in the overall cost and therefore may be omitted from data collection. These data would inform the extent of information (and data collection) required to report a robust cost-effectiveness analysis as part of a prospective randomized cancer trial.
# Methods
We sought to re-analyze the results from phase III trials that embedded economic evaluations to determine the impact of minimizing potential cost categories collected on the ICERs reported.
The Committee on Economic Analysis at the Canadian Cancer Trials Group has embedded a series of high-quality economic evaluations into a series of Canadian-led randomized controlled trials for new cancer therapies (n = 4). The LY.12 study was a noninferiority trial that compared an outpatient salvage therapy (gemcitabine, dexamethasone, and cisplatin (GDP)) to an inpatient regimen (dexamethasone, cytarabine, and cisplatin (DHAP)) in patients with relapsed aggressive histology lymphoma (n = 619 overall trial population; n = 519 patients within the economic substudy). 1,2 The CO.17 trial evaluated the costeffectiveness of cetuximab relative to best supportive care in patients with metastatic colorectal cancer (n = 572 overall trial population; n = 557 economic evaluation). 3,4 BR.21 studied the cost-effectiveness of erlotinib versus best supportive care in patients with advanced non-small cell lung cancer who had failed cisplatin-based therapy (n = 731 patients in both the overall study population and economic evaluation).BR.10 studied the role of chemotherapy versus observation in patients with early-stage non-small cell lung cancer (n = 482 overall trial population; n = 172 economic evaluation).For each of the four trials, we disaggregated the total costs into component incremental cost categories (i.e. drug acquisition and administration costs, hospitalizations, outpatient visits, etc.), presented in tabular form (descriptive analysis). The total direct costs associated with the treatment strategies within the economic evaluations were recalculated in three ways: using only the top 3 cost categories, using only the top 5 cost categories, and then using all collected cost components. Using individual trial-level data, ICERs were subsequently recalculated in a stepwise manner. Confidence intervals for each of the ICER simulations were generated by bootstrapping and descriptively presented along with the original confidence intervals (and ICERs) from the publications. In the CO.17 trial, incremental cost-effectiveness (with life-years gained) and incremental cost-utility (with quality-adjusted life-years gained) analyses were completed in the overall trial population and in individuals with wild-type KRAS tumors. All ratios and confidence intervals were recalculated from this trial. In the LY.12 study, the experimental GDP arm was dominant, and no ICER was presented due to the non-inferiority trial design; as such, only the total direct cost recalculations were presented for this trial.
# Results
Disaggregated drug cost components from the four clinical trials are presented in. Drug acquisition and administration costs represented the highest incremental cost category in three trials (BR.10, BR.14, and CO.17) and the second highest category in the LY.12 study. Hospitalization costs represented the other consistent cost driver, representing the top incremental cost category in one trial (LY.12) and the second highest category in two studies (BR.10 and CO.17).
Several cost components consistently contributed less than 5% of the total incremental cost used to calculate study ICERs. For example, laboratory testing incremental costs for the CO.17 and BR.10 trials contributed only 1.2% and 4.9%, respectively, to the total incremental cost. Management of study drug toxicities (without requiring hospitalization) in BR.21 and CO.17 contributed only 0.5% and 4.8%, respectively, to the total incremental costs. Imaging studies in the CO.17 trial only contributed 0.9% to the total incremental cost.
ICERs were recalculated using top 3, top 5, and all cost components (original results) from the clinical trials, when available. Only the total cost differences were presented for LY.12. The re-analyses based on fewer cost components (top 3 and top 5 components) did not appear to differ meaningfully from the original published results in that the re-analyzed ICERs would not have changed the original trial conclusions of cost-effectiveness. Specifically, none of the reanalyses would have changed the original perception of whether the experimental treatment arms were considered cost-effective or not, based on conventional willingness-to-pay thresholds in the range of US$50,000 to US$100,000 per quality-adjusted life-year gained.
# Discussion
In individual-patient cost-effectiveness analyses conducted prospectively alongside cancer clinical trials, we ascertained that key cost drivers inform the calculation of ICERs, while other cost categories have minimal impact on the results. Within the sample of four Canadian Cancer Trials Group trials evaluated, drug costs and hospitalizations represented the top cost drivers of relevance. When re-calculating ICERs from the trials using only the top 3 and top 5 categories with the highest incremental costs, the resultant values did not meaningfully differ from the original published ICER result. In all of the re-calculations, there were no instances in our descriptive analyses in which the original conclusions regarding cost-effectiveness would have changed, based on currently accepted willingness-topay thresholds.
These results suggest that the collection of resource utilization data within cancer trials could be narrowed. Fewer cost categories or more targeted collection of data within cost categories might be considered in future clinical trials. We appreciate that this implies an a priori understanding of potential cost components or importance, which may not always be realized at the time of study design. Although our results suggest that drug acquisition costs and hospitalizations are consistently high-cost drivers, this may not invariably be the case. Moreover, certain costs might always warrant collection, if important to the trial outcomes or perspective of the analysis; for example, indirect costs including lost productivity and caregiver burden would likely be essential costs to collect in an economic evaluation that considers a societal perspective. Ultimately, cancer trialists and their health economic collaborators may be aware of cost categories that are likely to be of low impact based on their understanding of the clinical context to be studied. Collection of low-impact cost categories could be minimized on a trial-by-trial basis in future studies with embedded economic evaluations, ideally at the design phase of the cancer clinical trial. The advantages associated with simplifying data collection are potentially significant. A recent analysis of two Canadian Cancer Trials Group studies (HN.6 with 320 patients and BR.26 with 385 patients) with embedded economic evaluations revealed that a considerable number of data queries were required for the resource utilization data (Penny Bradbury, Canadian Cancer Trials Group, personal communication); HN.6 required 1603 queries (690 of which were manual), while BR.26 required 3865 trials (2809 of which were manual). Efforts to minimize the burden of data collection with more pragmatic trial designs could have demonstrable impact on both human and financial resources, including site investigators and researches, data managers, central co-operative group analysts, and overall trial budgets. A formal analysis of the value of attaining perfect information, to quantify the additional cost to collect items that may not appear to impact outcomes, will be a future research endeavor. The use of routinely collected administrative data may also represent a novel, low-burden alternative to establishing resource utilization, with preliminary work demonstrating an ability to collect complete and highly concordant data.Our analysis does have limitations. Our results were only applicable to studies that prospectively collected direct medical resources and costs from the health system perspective; our conclusions do not apply to the collection of indirect costs or caregiver burden, which may be substantial from a patient perspective. Our analysis only applied to a limited sample of previously completed trials that focused on drug interventions within Canada. This may represent a limitation in generalizing our results to all cancer clinical trials, and notably trials that might focus on radiation or surgicalbased interventions; however, we feel that the concept that certain cost categories contribute little to the final results of a trial-based economic analysis can be broadly applied to cancer clinical trials in general. This study serves to clarify the resource parameters required to calculate incremental costs borne when a new cancer therapy is considered. Key cost drivers including acquisition costs for novel cancer drugs and hospitalizations are identifiable and should be consistently collected. Other cost categories have a minimal impact on the final incremental cost calculations, and their omission does not change the value regarding cost-effectiveness. Limiting the collection of cost parameters that have minimal impact would ensure more efficient use of trial staff, improve allocation of trial resources, and ultimately reduce the burden on participating patients.
## Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
# Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by a 2018 Seed Grant awarded by the Canadian Center for Applied Research in Cancer Control.
## Orcid ids
Matthew C Cheung https://orcid.org/0000-0003-3193-5872 Annette Hay https://orcid.org/0000-0003-4581-3390 |
BioMethyl: an R package for biological interpretation of DNA methylation data
Motivation: The accumulation of publicly available DNA methylation datasets has resulted in the need for tools to interpret the specific cellular phenotypes in bulk tissue data. Current approaches use either single differentially methylated CpG sites or differentially methylated regions that map to genes. However, these approaches may introduce biases in downstream analyses of biological interpretation, because of the variability in gene length. There is a lack of approaches to interpret DNA methylation effectively. Therefore, we have developed computational models to provide biological interpretation of relevant gene sets using DNA methylation data in the context of The Cancer Genome Atlas. Results: We illustrate that Biological interpretation of DNA Methylation (BioMethyl) utilizes the complete DNA methylation data for a given cancer type to reflect corresponding gene expression profiles and performs pathway enrichment analyses, providing unique biological insight. Using breast cancer as an example, BioMethyl shows high consistency in the identification of enriched biological pathways from DNA methylation data compared to the results calculated from RNA sequencing data. We find that 12 out of 14 pathways identified by BioMethyl are shared with those by using RNA-seq data, with a Jaccard score 0.8 for estrogen receptor (ER) positive samples. For ER negative samples, three pathways are shared in the two enrichments with a slight lower similarity (Jaccard score ¼ 0.6). Using BioMethyl, we can successfully identify those hidden biological pathways in DNA methylation data when gene expression profile is lacking. Availability and implementation: BioMethyl R package is freely available in the GitHub repository (https://github.com/yuewangpanda/BioMethyl).
# Introduction
Epigenetic modification of DNA plays an important role in regulating gene activity and transcript levels without directly changing the gene sequence. DNA methylation is one of the most common epigenetic mechanisms and has been shown to impact multiple biological processes [bib_ref] Rett syndrome is caused by mutations in X-linked MECP2, encoding methyl-CpG-binding protein..., Amir [/bib_ref] [bib_ref] DNA methylation and epigenetics, Bender [/bib_ref] [bib_ref] Methylation matters, Costello [/bib_ref] [bib_ref] The power and the promise of DNA methylation markers, Laird [/bib_ref]. Consequently, aberrant DNA methylation has been associated with multiple human cancers including prostate cancer [bib_ref] DNA methylation changes in prostate cancer, Goering [/bib_ref] [bib_ref] Fluorescent methylation-specific polymerase chain reaction for DNA-based detection of prostate cancer in..., Goessl [/bib_ref] , breast cancer [bib_ref] Presence of tumor DNA in plasma of breast cancer patients: clinicopathological correlations, Silva [/bib_ref] [bib_ref] DNA methylation and breast cancer, Szyf [/bib_ref] and liver cancer [bib_ref] The altered DNA methylation pattern and its implications in liver cancer, De Zhu [/bib_ref] V C The Author(s) 2019. Published by Oxford University Press.
## 3635
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact [email protected] [bib_ref] Frequent p15 promoter methylation in tumor and peripheral blood from hepatocellular carcinoma..., Wong [/bib_ref]. Moreover, a strong relationship between DNA methylation and cancer development has been found, which has resulted in DNA methylation being used as a prognostic marker in many cancer types [bib_ref] DNA methylation biomarkers as diagnostic and prognostic tools in colorectal cancer, Gyparaki [/bib_ref] [bib_ref] DNA methylation profiling in the clinic: applications and challenges, Heyn [/bib_ref] [bib_ref] Hypermethylation of the CDKN2A gene in colorectal cancer is associated with shorter..., Maeda [/bib_ref] [bib_ref] Tumor p16M is a possible marker of advanced stage in non-small cell..., Ng [/bib_ref] [bib_ref] A prognostic DNA methylation signature for stage I non-small-cell lung cancer, Sandoval [/bib_ref]. Therefore, array-based or sequencing-based approaches have been developed to measure large-scale DNA methylation profiles [bib_ref] Principles and challenges of genomewide DNA methylation analysis, Laird [/bib_ref] [bib_ref] Advances in the profiling of DNA modifications: cytosine methylation and beyond, Plongthongkum [/bib_ref]. The Illumina methylation array platform provides an opportunity to generate genome-wide human DNA methylation profile, and the Illumina HumanMethylation450 BeadChip is one of the most commonly utilized platforms for investigating DNA methylation in a comprehensive manner. It contains 450 000 CpG probes that cover 99% of Refseq genes, encompassing promoters, gene bodies, UTRs and intergenic regions [bib_ref] Validation of a DNA methylation microarray for 450,000 CpG sites in the..., Sandoval [/bib_ref].
To interpret the underlying biology, differentially methylated CpG sites are routinely associated with particular phenotypes including cancer survival and biological pathway enrichment. Since gene ontology terms and pathways are generally represented as sets of genes, it is critical to accurately map DNA methylation sites to gene annotations. Mapping DNA methylation sites to genes usually involves either (1) identifying the genes with a single differentially methylated CpG site in their promoters [bib_ref] LRpath analysis reveals common pathways dysregulated via DNA methylation across cancer types, Kim [/bib_ref] [bib_ref] Association between DNA methylation in whole blood and measures of glucose metabolism:..., Kriebel [/bib_ref] [bib_ref] Integrated analysis of DNA methylation and gene expression reveals specific signaling pathways..., Li [/bib_ref] [bib_ref] DNA methylation signatures define molecular subtypes of diffuse large B-cell lymphoma, Shaknovich [/bib_ref] or (2) identifying genes cover differentially methylated regions (DMR) of the genome [bib_ref] DNA methylation array analysis identifies profiles of blood-derived DNA methylation associated with..., Marsit [/bib_ref] [bib_ref] DMRforPairs: identifying differentially methylated regions between unique samples using array based methylation..., Rijlaarsdam [/bib_ref] [bib_ref] IMA: an R package for high-throughput analysis of Illumina's 450K infinium methylation..., Wang [/bib_ref]. Additionally, several web-based tools can be used to analyze methylation data including Annotation-Modules [bib_ref] Annotation-Modules: a tool for finding significant combinations of multisource annotations for gene..., Hackenberg [/bib_ref] , EpiExplorer [bib_ref] EpiExplorer: live exploration and global analysis of large epigenomic datasets, Halachev [/bib_ref] , and Galaxy [bib_ref] Galaxy: a comprehensive approach for supporting accessible, reproducible, and transparent computational research..., Goecks [/bib_ref]. However, these methods have four main limitations. First, it is hard to capture the directionality of gene expression that results from DNA methylation. Generally, hypermethylation of the promoter causes repression [bib_ref] The role of DNA methylation in mammalian epigenetics, Jones [/bib_ref] , while hypermethylation in the gene body is correlated with activation [bib_ref] DNA methylation patterns associate with genetic and gene expression variation in HapMap..., Bell [/bib_ref] [bib_ref] The DNA methylation paradox, Jones [/bib_ref]. Therefore, it is difficult to predict the changes in gene expression based simply upon DNA methylation results. Second, it is difficult to precisely define the extent of gene promoter methylation due to variability in the size of canonical promoters and the presence of distal enhancers, which introduces biases into the association of methylated regions with gene models. Third, the longer length of a gene, the higher the probability that this gene could be selected due to the nearby differentially methylated CpG sites. Lastly, for the web-based studies, specialized tools are needed to reformat the methylation data to genomic region formats (i.e. BED and WIG), which increases the difficulty of usage. To our knowledge, tools utilizing complete largescale DNA methylation data to infer the directionality of gene expression and provide a framework for the interpretation of biological impact are lacking.
We have developed an R package that we have named Biological interpretation of DNA Methylation (BioMethyl), which identifies biologically meaningful trends from a complete DNA methylation profile by using all available CpG sites. By integrating DNA methylation and RNA sequencing (RNA-seq) profiles for 37 cancer types from The Cancer Genome Atlas (TCGA), we have developed linear regression models to analyze the relationship between any single gene's expression profile and its corresponding CpG sites methylation sites for each cancer type. We inferred the contribution of each single CpG site to the expression of a gene using the coefficient of linear regression calculated by the model. Therefore, BioMethyl captures the expression values for a gene, based on its association with CpG methylation sites. We compare our results to RNA-seq profiles and show that BioMethyl accurately estimates gene expression from DNA methylation data. We have integrated Gene Set Enrichment Analysis (GSEA) [bib_ref] Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles, Subramanian [/bib_ref] into BioMethyl to automatically identify enriched pathways. As a result, we show that the enriched pathways identified by BioMethyl were highly consistent with those identified by RNA-seq. The BioMethyl R package is freely available at GitHub accessing by https://github.com/yuewang panda/BioMethyl.
# Materials and methods
## Data collection
We downloaded RNA-seq and DNA methylation data for all 37 TCGA cancer types from Firehose (https://gdac.broadinstitute.org/, November, 2016). We fit models to the cancer types that contained more than 50 samples with both RNA-seq and DNA methylation profiles. We excluded ovarian cancer, lymphoid neoplasm diffuse large B-cell lymphoma and cholangiocarcinoma due to their low sample numbers (Supplementary [fig_ref] Table 1: Brief introduction of functions in BioMethyl R package [/fig_ref]. To train a non-cancer model, TCGA normal samples with paired RNA-seq and methylation profiles were collected (Supplementary . To validate the non-cancer model, DNA methylation data (GSE42861) [bib_ref] Epigenome-wide association data implicate DNA methylation as an intermediary of genetic risk..., Liu [/bib_ref] and gene expression data (GSE15573) [bib_ref] Transcriptome analysis describing new immunity and defense genes in peripheral blood mononuclear..., Teixeira [/bib_ref] were downloaded for rheumatoid arthritis (RA) samples.
## Development of biomethyl models
To train BioMethyl models for each cancer type, we first preprocessed the RNA-seq and DNA methylation data. For RNA-seq data, we log2-transformed the data and removed genes if expression values were zero in more than half of the samples. Then, a z-transformation was further applied to the RNA-seq profile across samples. For DNA methylation data, we removed CpG sites if methylation levels were missing values in more than half of sample size. Then, R package 'ENmix' [bib_ref] ENmix: a novel background correction method for Illumina HumanMethylation450 BeadChip, Xu [/bib_ref] was applied to methylation data to filter out outliers and to replace missing values using k nearest neighbors algorithm.
Then, we trained each BioMethyl model using linear regression to capture the association between gene expression and DNA methylation for each cancer types. For gene i , E ¼ {e 1 , e 2 , . . ., e n } is the gene expression across N samples, and M is the corresponding methylation matrix, containing all CpG sites associated to gene i , where cpg n, j is the beta value of j-th CpG in sample n. By calculating the correlations between beta values of each CpG site and E, we only selected the CpG sites whose beta values are correlated (Pearson correlation coefficient >j0.05j) with gene expression to build a model for gene i using the following function:
[formula] E ¼ a þ B Ã M; [/formula]
where B ¼ {b 1 , b 2 , . . ., b n } is the vector of coefficients estimated by linear regression model which explains the contribution on gene expression for each CpG site. We built cancer-specific models for all TCGA cancer types and recorded those models into BioMethyl. The non-cancer model was trained with the same way instead of using matched data of TCGA normal samples.
# Pathway enrichment analysis
GSEA and Fisher's exact test were applied to conduct pathway enrichment analyses in this study with the Molecular Signature Database (MSigDB) C2 dataset (c2.all.v5.2.symbols.gmt, 2016) [bib_ref] Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles, Subramanian [/bib_ref]. Statistical significance of enriched pathways was set to false discovery rate (FDR) < 0.01. For Fisher's exact test, FDR was calculated with Benjamini and Hochberg method. To automatically identify enriched pathways, GSEA R script method 'GSEA.1.0.R' was deployed in BioMethyl package to perform enrichment analysis with default settings [bib_ref] Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles, Subramanian [/bib_ref]. The C2 dataset was set as the default gene sets.
## Biomethyl validation
To validate our model, we applied 10-fold cross-validation to test the quality of the gene expression inferred by the linear regression model. Namely, for each validation, we used 9/10 samples as training dataset to train the model. Then, by integrating the DNA methylation data and trained model, we calculated a gene expression profile for the rest 1/10 samples. After 10-fold cross-validation, we merged those profiles to a gene expression profile containing all samples and compared it with the RNA-seq data to conduct downstream validations.
# Statistical analysis
To identify differentially methylated CpG sites and differentially expressed genes in two phenotypes (in our case is ERþ versus ERÀ), we applied the Student t test to calculate the t scores and corresponding P-values. The FDR was further calculated using the Benjamini-Hochberg multiple hypothesis testing correction method. By ranking t scores in a decreasing order, top-ranked CpG sites/ genes are differentially methylated/expressed in ERþ samples and bottom-ranked CpG sites/genes are differentially methylated/ expressed in ERÀ samples. For P-values in boxplots, ANOVA test was used for comparison in multiple groups and two-sample Wilcoxon test was applied to those two group comparisons.
# Results
## The limitations of previous approaches
Previous studies [bib_ref] LRpath analysis reveals common pathways dysregulated via DNA methylation across cancer types, Kim [/bib_ref] [bib_ref] Association between DNA methylation in whole blood and measures of glucose metabolism:..., Kriebel [/bib_ref] [bib_ref] Integrated analysis of DNA methylation and gene expression reveals specific signaling pathways..., Li [/bib_ref] [bib_ref] DNA methylation signatures define molecular subtypes of diffuse large B-cell lymphoma, Shaknovich [/bib_ref] have used gene promoter regions that contained differentially methylated CpG sites to perform pathway enrichment analyses. We tested the hypothesis that gene length might introduce biases in the downstream analyses. First, we examined the number of CpG sites associated with each gene according to the HumanMethylation450 platform mapping information. We found that 43.3% of genes (10 865 out of 25 094) map to less than 10 CpG sites, whereas more than 20% genes map to over 20 CpG sites [fig_ref] Figure 1: Simply translating CpGs to genes confuses downstream results because of gene length... [/fig_ref]. Moreover, only 7.4% of genes (1860 out of 25 094) map to one CpG site. Notably, 0.7% of genes (163 out of 25 094) map to more than 100 CpG sites. Broadly, genes with a longer length cover more CpG sites in the HumanMethylation450 platform [fig_ref] Figure 1: Simply translating CpGs to genes confuses downstream results because of gene length... [/fig_ref]. For example, PTPRN2, a receptor-type tyrosine-protein phosphatase N2 gene which occupies 1 048 741 bases, is associated with many diseases [bib_ref] Antibodies to the protein tyrosine phosphatases IAR and IA-2 are associated with..., Schmidli [/bib_ref] [bib_ref] DNA hypermethylation and DNA hypomethylation is present at different loci in chronic..., Smyth [/bib_ref] [bib_ref] Aberrant expression of proPTPRN2 in cancer cells confers resistance to apoptosis, Sorokin [/bib_ref] and covers 1288 CpG sites in the platform. Furthermore, by ranking genes based on the number of covered CpG sites in decreasing order, we found that the accumulation of CpG sites is derived by genes with longer length [fig_ref] Figure 1: Simply translating CpGs to genes confuses downstream results because of gene length... [/fig_ref]. In fact, the top 5000 genes in length collectively occupy more than 50% of measured CpG sites.
For further biological interpretation of study results, differentially methylated CpG sites are often translated to genes. Because these mappings are not linear, it is theoretically possible for a gene to be implicated in both up-and downregulated CpG groups in an analysis. To demonstrate this phenomenon, using TCGA breast cancer (BRCA) as an example, we identified the significantly (P < 0.01) differentially methylated CpG sites for estrogen receptor positive (ERþ) and estrogen receptor negative (ERÀ) samples which resulted in two distinct, non-overlapping lists of CpG sites. We then randomly selected 5%/10%/15%/20% of the sites from each of the groups, mapped these sites to genes, and calculated the Jaccard score between the groups of genes. This simulation was repeated 10 000 times. When using sites only in promoter region (within 1 kb of the transcription start site) of genes, we found that the average number of genes overlapping between the two lists increased as the number of random CpG sites were selected. We found the same trend when considering CpG sites that map to whole gene regions [fig_ref] Figure 1: Simply translating CpGs to genes confuses downstream results because of gene length... [/fig_ref]. Moreover, the number of overlapping genes was at least 4.2 times larger when using all CpG sites compared to using only promoter sites. These observations suggest that simply translating CpG sites to genes based on location may confuse the overall interpretation of results. This may be in part due to the effect of gene length because longer regions generally contain more CpG sites. Next, we identified the 500 genes that most frequently appeared in each simulation using whole region CpG sites and found that the average lengths of these genes are nearly identical between simulations but are much longer than the average length of all genes in the genome [fig_ref] Figure 1: Simply translating CpGs to genes confuses downstream results because of gene length... [/fig_ref] , ANOVA P < 2e-16). Furthermore, we found that 449 of the 500 genes are shared by these four different simulations, and those genes tend to cover more CpG sites compared to all genes in the genome [fig_ref] Figure 1: Simply translating CpGs to genes confuses downstream results because of gene length... [/fig_ref]. These 449 common genes were significantly involved in sensory system (olfactory transduction), signaling molecules and interaction (cytokine receptor interaction), genetic information processing (ubiquitin mediated proteolysis, spliceosome), signaling molecules and interaction (neuroactive ligand receptor interaction), neurodegenerative diseases (Huntington's disease and Alzheimer's disease) and nucleotide metabolism (purine metabolism and pyrimidine metabolism). Moreover, we found that genes involved in those significant pathways cover more CpG sites (P ¼ 2e-14) and have a longer gene length (P ¼ 4e-07) compared to all genes in the genome [fig_ref] Figure 1: Simply translating CpGs to genes confuses downstream results because of gene length... [/fig_ref]. Again, these results suggest that longer genes could be preferentially selected, and the length of genes might dramatically impact the biological interpretation of previously studies which use direct mapping approaches.
## Overview of our analyses
To overcome these limitations, we developed a simple and powerful method called BioMethyl to interpret DNA methylation data in a wide range of cancer types. We built a linear regression model for each gene to capture the association of its expression and corresponding CpG methylation sites by integrating RNA-seq and DNA methylation profiles for each TCGA cancer type. The coefficients of each model explain the contributions of CpG methylation sites associated with each gene's expression. To validate the model, we used 10-fold cross-validation to compare the biological outcome of the inferred gene expression profiles and the original RNA-seq profile [fig_ref] Figure 2: Workflow of our computational framework [/fig_ref]. We found a high concordance between those two profiles, both in terms differentially expressed genes and enriched pathways. Therefore, our method takes advantage of the complete profiles of DNA methylation and RNA-seq to build more accurate models for each cancer type. By assembling our framework and GSEA method, we developed a user-friendly R package, called BioMethyl, which is freely available at GitHub (https://github.com/yuewangpanda/ BioMethyl). When a user inputs a new DNA methylation dataset of interest for a given cancer type, BioMethyl utilizes the corresponding models to uncover relevant biological pathways hidden in the data [fig_ref] Figure 2: Workflow of our computational framework [/fig_ref].
## Validation of the biomethyl model
To validate BioMethyl, we first examined whether the linear regression model could accurately depict the gene expression profile through DNA methylation data. We trained the models with cancers in TCGA that had both RNA-seq and DNA methylation profiles using 10-fold cross-validation (see Section 2). Comparing the pseudo gene expression estimated by the model and the RNA-seq profile, we found that BioMethyl model could explain the relationship between RNA-seq and DNA methylation profiles of 94.1% (32 out of 34) cancers in TCGA with median Spearman correlation coefficients (SCCs) greater than 0.3 [fig_ref] Figure 1: Simply translating CpGs to genes confuses downstream results because of gene length... [/fig_ref].
We tested BioMethyl on the TCGA breast cancer (BRCA) dataset since it contained 785 tumors, the highest number of tumors in TCGA, with both RNA-seq and DNA methylation profiles. BioMethyl was able to recapitulate tumors' gene expression using their DNA methylation data alone with a median SCC 0.423 when compared to their RNA-seq profiles [fig_ref] Figure 3: Validation of BioMethyl in the context of breast cancer [/fig_ref]. Moreover, the majority (more than 88%) of genes have SCCs greater than 0.3 [fig_ref] Figure 2: Workflow of our computational framework [/fig_ref]. By dividing genes into low, intermediate and high variance groups based on the original RNA-seq dataset, we found that genes with high variance had significantly higher SSCs than the other two groups [fig_ref] Figure 3: Validation of BioMethyl in the context of breast cancer [/fig_ref] , ANOVA test P < 2e-16). To further test the reliability of our model, we compared the expression difference of each gene between ERþ and ERÀ patients for the estimated gene expression profile and RNA-seq profile. By calculating the correlation of t scores, the result showed that the estimated gene expression profile inferred from DNA methylation data is highly consistent with the RNA-seq data [fig_ref] Figure 3: Validation of BioMethyl in the context of breast cancer [/fig_ref] , SCC ¼ 0.88) which is only slightly lower than the comparison between TCGA microarray and RNA-seq profiles [fig_ref] Figure 4: Validation of BioMethyl using Fisher's exact test [/fig_ref] , SCC ¼ 0.94). These observations suggest that BioMethyl is able to accurately infer gene expression through DNA methylation data compared to RNA-seq data.
To further compare the similarity of biological findings identified by BioMethyl and RNA-seq analyses, we performed GSEA analysis [bib_ref] Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles, Subramanian [/bib_ref]. By dividing TCGA BRCA samples into ERþ and ERÀ groups, we identified pathways significantly enriched in those two groups of samples. When using RNA-seq data, there were 13 pathways significantly (FDR < 0.01) enriched in ERþ samples and four pathways significantly (FDR < 0.01) enriched in ERÀ samples [fig_ref] Figure 3: Validation of BioMethyl in the context of breast cancer [/fig_ref]. When using the gene expression profile inferred by BioMethyl, there were 14 pathways significantly (FDR < 0.01) enriched in ERþ samples and four pathways significantly (FDR < 0.01) enriched in ERÀ samples [fig_ref] Figure 3: Validation of BioMethyl in the context of breast cancer [/fig_ref]. We found that 12 pathways for ERþ samples are shared by those two methods (Jaccard score ¼ 0.8). For ERÀ samples, three pathways are shared in the two enrichments with a slight lower similarity (Jaccard score ¼ 0.6). We also investigated Jaccard scores using different FDR thresholds (from 0 to 0.25) for shared pathways in either ERþ or ERÀ samples . In our experience, setting a more conservative FDR threshold (<0.05) is important to achieve high similarity (Jaccard scores > 0.5) between the results of BioMethyl and RNA-seq.
## Validation of fisher's exact test enrichment results
Next, we compared the enriched pathways of genes covering differentially methylated CpG sites and the genes mostly differentially expressed in gene expression profiles. We identified the top 500 hypermethylated CpG sites (FDR < 1e-30) in ERþ samples and mapped those sites to genes, resulting in 270 genes. We identified 348 genes for ERÀ samples through the top 500 hypermethylated CpG sites (FDR < 4e-38) in ERÀ samples. Notably, there were 12 genes shared between these two gene sets [fig_ref] Figure 4: Validation of BioMethyl using Fisher's exact test [/fig_ref]. Next, we identified the top 270 and 348 differentially expressed genes in ERþ and ERÀ patients, respectively, using both the RNA-seq profile and the profile inferred from BioMethyl. We found very little overlap in the number of genes identified by CpG-mapped genes and the top differentially expressed genes identified by RNA-seq or BioMethyl for both ERþ [fig_ref] Figure 4: Validation of BioMethyl using Fisher's exact test [/fig_ref] and ERÀ [fig_ref] Figure 4: Validation of BioMethyl using Fisher's exact test [/fig_ref] samples. Then, we conducted Fisher's exact test to those six differentially expressed gene sets to identify enriched biological pathways, respectively. When using gene sets collected via differentially methylated CpG sites, there is only one pathway enriched in ERþ samples (compared to 23 identified by RNA-seq and 28 identified by BioMethyl) and no pathways enriched in ERÀ samples (compared to 143 identified by RNA-seq and 16 identified by BioMethyl) (data not shown). We found that 21 out of 28 significantly enriched pathways (FDR < 0.01) using gene set from BioMethyl profile are shared with those from RNA-seq in the context of ERþ samples [fig_ref] Figure 4: Validation of BioMethyl using Fisher's exact test [/fig_ref]. The common pathways are highly related to ERþ context including upregulated by ESR1, genes upregulated in luminal-B breast cancer and downregulated in invasive breast cancer . Similar, for ERÀ samples, 10 out of 16 pathways (FDR < 0.01) are shared in the comparison [fig_ref] Figure 4: Validation of BioMethyl using Fisher's exact test [/fig_ref] including downregulation by ESR1, genes upregulated in basal breast cancer and downregulated in luminal-B breast cancer . These observations suggest that using the gene expression values inferred by the BioMethyl method is able to identify common biological pathways using Fisher's exact test.
## Application of biomethyl to non-cancer diseases
To test whether the BioMethyl could be applied to non-cancer diseases as well, we selected RA, an autoimmune disease, as an example. By integrating the non-cancer model and the DNA methylation data, we estimated gene expression for samples in GSE42861. GSEA and Fisher's exact test were applied to both the estimated gene expression of GSE42861 and the real gene expression dataset GSE15573. When comparing the GSEA results, neither of them contains significant (FDR < 0.25) pathways. When comparing the results of Fisher's exact test, differentially expressed genes (FDR < 0.01) in our model were significantly enriched in immunerelated pathways such as T cell receptors, immune response and antigen response. This result is consistent with both the characteristics of RA [bib_ref] Understanding the dynamics: pathways involved in the pathogenesis of rheumatoid arthritis, Choy [/bib_ref] and the enrichment analysis reported from the RA gene expression (GSE15573) [bib_ref] Transcriptome analysis describing new immunity and defense genes in peripheral blood mononuclear..., Teixeira [/bib_ref]. This validation suggests that our non-cancer model could capture the common biological pathways for noncancer diseases, such as RA.
## Implementation of biomethyl package
For a user-friendly and straightforward usage, we compiled models for 37 cancer types into our BioMethyl R package including five functions [fig_ref] Table 1: Brief introduction of functions in BioMethyl R package [/fig_ref]. When a user inputs DNA methylation data, BioMethyl will produce the relevant biological pathways as final output. Here, we briefly introduce the procedure of BioMethyl package. BioMethyl preprocesses the data with filterMethyData() function and removes CpG sites that have missing values in more than half samples and imputes the rest missing values by integrating 'ENmix', an specialized R package for DNA methylation data [bib_ref] ENmix: a novel background correction method for Illumina HumanMethylation450 BeadChip, Xu [/bib_ref] , with default parameters. Next, calExpr() function is applied to the filtered methylation data to infer the gene expression profile for a given disease type. In this step, the inferred gene expression could be saved as a text file as an option for other customized applications. Then, differentially expressed genes (DEGs) are identified via calDEG() function in order to optimize gene expression for GSEA analysis. As well, the list of differentially expressed genes could be saved as a text file for gene set based pathway or GO term enrichment test. Lastly, using the inferred gene expression, BioMethyl integrates GSEA R code to perform pathway enrichment analysis using GSEA default settings. In this step, the parameter of cutoff for DEGs is a numeric vector in which the first element is the cutoff for t score (default is 0) and the second is for P-value (default is 0.01). Moreover, BioMethyl package has a friendly recommendation function so that it helps users select the best model for their DNA methylation data. By applying a centroid manner, referCancerType() function can suggest a suitable cancer type model having the best similarity with TCGA cancers when it is not clear. The BioMethyl package and demo code are freely available at GitHub (https://github.com/yuewangpanda/BioMethyl).
# Discussion
Since DNA methylation plays important roles in multiple biological processes, more and more efforts have been put on generating DNA methylation data. Attempts at investigating enriched pathways using DNA methylation profile has been an active area study. Previous studies used either single differentially methylated CpG sites or DMRs as an assumed proxy to identify the differentially expressed genes between samples. However, our results suggest that using the direct mapping method results in a pronounced overlapping of genes between opposing biological groups which could introduce bias to downstream analyses-pathway/genes associated with more CpG sites are more likely to be identified [fig_ref] Figure 1: Simply translating CpGs to genes confuses downstream results because of gene length... [/fig_ref]. Previous work has tried to correct this bias by modeling the probability of a gene to be selected by chance as a function of the number of CpG sites it associated with [bib_ref] Gene-set analysis is severely biased when applied to genome-wide methylation data, Geeleher [/bib_ref]. In this sense, all CpG sites associated with a gene are assumed to contribute equally to the transcriptional regulation of the gene. In our work, we developed the BioMethyl method to more reasonably map CpG sites to genes by assigning different weights to sites according to their relative contributions to gene expression. Our results showed that the enriched pathways determined by BioMethyl are highly consistent with those interpreted directly from RNA-seq.
Due to the internal relationship between DNA methylation and gene expression [bib_ref] DNA methylation and gene expression, Razin [/bib_ref] , several studies have developed computational methods to infer gene expression from DNA methylation data in the context of a certain cancer [bib_ref] Using epigenomics data to predict gene expression in lung cancer, Li [/bib_ref] [bib_ref] Modeling complex patterns of differential DNA methylation that associate with gene expression..., Schlosberg [/bib_ref]. BioMethyl applies linear regression models to capture the association between the expression of a gene and its CpG sites methylation levels for all TCGA cancer types. We found that the gene expression profile inferred from DNA methylation data is highly like RNA-seq profile [fig_ref] Figure 3: Validation of BioMethyl in the context of breast cancer [/fig_ref] and . Moreover, we found that using the inferred gene expression profile can classify ERþ from ERÀ breast cancer samples as well as using RNA-seq profile . These results suggest that BioMethyl captures the overall and true directions of gene expression via linear regression models. Therefore, the inferred gene expression profile could be applied to other downstream analyses (e.g. pathway enrichment using Fisher's exact test, identifying differentially expressed genes) when a study only has DNA methylation data available.
In this study, we took advantage of the TCGA cancer data to estimate the contribution of individual CpG site to gene expression. However, the weights of CpG sites might change in different tissues or under different physiological conditions. Therefore, lower performance might be expected when the models trained from the TCGA data are directly applied to DNA methylation data for other diseases. Nevertheless, the proposed framework can be used to re-calculate the weights of CpG sites under various contexts given matched gene expression and DNA methylation data, which has becoming more and more readily available in the future.
In summary, BioMethyl makes use of the whole DNA methylation profile and captures the association between gene expression and DNA methylation in a highly sensitivity way. In our models, we assigned coefficients to those CpG sites really associated with changes in gene expression. The models contained within BioMethyl span a large number of diverse cancers. Our freely available R package is easy to install and use. Moreover, our methods represent significant contributions to data interpretation when only DNA methylation is available. With the improvement of methylation platform (i.e. HumanMethylation850), BioMethyl hopefully could achieve a higher accuracy in terms of biological interpretation directly from DNA methylation data.
# Funding
[fig] Figure 1: Simply translating CpGs to genes confuses downstream results because of gene length diversity. (A) The mapping between number of genes and different CpG sites range (left y axis). The fraction above each bar shows the percentage of whole genome genes which located in this CpG range. Right y axis shows the relationship between the distribution of gene lengths in each CpG range. (B) The accumulation of CpG sites with gene length. (C) Barplot showing the average Jaccard scores from 10 000 times simulations. Error bars show standard deviation of Jaccard scores. Blue is only using promoter CpG sites and red is using whole CpG sites. (D) Boxplot showing the distribution of gene lengths for each simulation and all genes. ANOVA Pvalue is showed. (E) The distributions of covered CpG sites for common genes in simulations and all genes. (F) Boxplots for comparing gene length and covered CpG sites numbers between genes (subset of common genes) in significant pathways and all genome genes. Wilcoxon Rank Sum test P-values are showed [/fig]
[fig] Figure 2: Workflow of our computational framework. (A) Validation of models. Using a 10-fold cross-validation manner, BioMethyl trains models and calculates a new gene expression matrix for samples. To validate our models, we compare the inferred gene expression matrix with RNA-seq data in terms of gene difference and involved pathways. (B) Application of models. Using the complete DNA methylation and RNA-seq profiles, we build models for each cancer. Further by integrating GSEA analysis, we develop the R package BioMethyl to reveal the relevant pathways in a new DNA methylation data of interest [/fig]
[fig] Figure 3: Validation of BioMethyl in the context of breast cancer. (A) Density plot for SCC of genes by comparing gene expression inferred by BioMethyl and RNA-seq data. (B) Scatter plot of t scores (ERþ samples versus ERÀ samples) for genes between gene expression inferred by BioMethyl and RNA-seq data. Pathway enrichment results of GSEA are showed for (C) RNA-seq data and (D) gene expression inferred by BioMethyl by comparing ERþ to ERÀ samples. For pathways enriched in ERþ samples, Àlog10(FDR) are showed (red). The orange pathways are pathways shared by two results for ERþ samples. For pathways enriched in ERÀ samples, log10(FDR) are showed (green), in which green pathways are shared pathways [/fig]
[fig] Figure 4: Validation of BioMethyl using Fisher's exact test. Venn diagrams for (A) differentially expressed genes selected by hypermethylated CpG sites in ERþ and ERÀ samples; (B) differentially expressed genes in ERþ samples selected by hypermethylated CpG sites, RNA-seq and BioMethyl; (C) differentially expressed genes in ERÀ samples selected by hypermethylated CpG sites, RNA-seq and BioMethyl; (D) pathways enriched in ERþ samples between RNA-seq data and BioMethyl; (E) pathways enriched in ERÀ samples between RNA-seq data and BioMethyl [/fig]
[table] Table 1: Brief introduction of functions in BioMethyl R package [/table]
|
Effect of Topical Tranexamic Acid on Bleeding and Quality of Surgical Field during Functional Endoscopic Sinus Surgery in Patients with Chronic Rhinosinusitis: A Triple Blind Randomized Clinical Trial
July 7, 2010
30411July 7, 2010CONSORT CHECKLIST Table. CONSORT 2010 Checklist of Information to Include When Reporting a Randomized Trial a Section and Topic Item No. Checklist Item Reported on Page No. Title and abstract 1a Identification as a randomized trial in the title 1b Structured summary of trial design, methods, results, and conclusions (for specific guidance see CONSORT for abstracts) Print Form Title Abstract
Specific objectives or hypotheses Methods Trial design 3a Description of trial design (such as parallel, factorial) including allocation ratio 3b
Important changes to methods after trial commencement (such as eligibility criteria), with reasons Participants 4a Eligibility criteria for participants 4b
Settings and locations where the data were collected Interventions 5
The interventions for each group with sufficient details to allow replication, including how and when they were actually administered Outcomes 6a
Completely defined prespecified primary and secondary outcome measures, including how and when they were assessed 6b
Any changes to trial outcomes after the trial commenced, with reasons Sample size 7a How sample size was determined 7b
When applicable, explanation of any interim analyses and stopping guidelines Randomization Sequence generation 8a Method used to generate the random allocation sequence 8b
Type of randomization; details of any restriction (such as blocking and block size) Allocation concealment mechanism 9 M e c h a n i s mu s e dt oi m p l e m e n tt h er a n d o ma l l o c a t i o ns e q u e n c e( s u c ha ss e q u e n t i a l l yn u m b e r e d containers), describing any steps taken to conceal the sequence until interventions were assigned Implementation 10 Who generated the random allocation sequence, who enrolled participants, and who assigned participants to interventions Blinding 11a If done, who was blinded after assignment to interventions (for example, participants, care providers, those assessing outcomes) and how 11b
If relevant, description of the similarity of interventions Statistical methods 12a
Statistical methods used to compare groups for primary and secondary outcomes 12b
Methods for additional analyses, such as subgroup analyses and adjusted analyses Results Participant flow (a diagram is strongly recommended)
13a
For each group, the numbers of participants who were randomly assigned, received intended treatment, and were analyzed for the primary outcome 13b
For each group, losses and exclusions after randomization, together with reasons Recruitment 14a
Dates defining the periods of recruitment and follow-up 14b
Why the trial ended or was stopped Baseline data 15 A |
Enhanced recovery after surgery with intrathecal opioid in a patient of Gilbert's syndrome undergoing mitral valve replacement
Gilbert's syndrome, an inherited autosomal dominant disorder, is the most common cause of congenital unconjugated hyperbilirubinaemia. We report the anaesthetic management of a 46-year-old female with Gilbert's syndrome operated for mitral valve replacement (MVR), with a special focus on the role of intrathecal opioids.
# Introduction
Gilbert's syndrome, an inherited autosomal dominant disorder, is caused by the relative deficiency of glucuronyl transferase enzyme responsible for conjugation of bilirubin. It is the most common cause of congenital unconjugated hyperbilirubinaemia. Preoperative bilirubin level is a risk factor for mortality after cardiac surgery and stress due to fasting, anaesthesia, surgery, and cardiopulmonary bypass (CPB) cause exacerbation of hyperbilirubinaemia. [bib_ref] Gilbert's syndrome-clinical and pharmacological implications, Radu [/bib_ref] [bib_ref] Hyperbilirubinemia after cardiac surgery: Incidence, risk factors, and clinical significance, Wang [/bib_ref] Single preoperative administration of intrathecal opioids decreases pain and perioperative opioid requirements and hastens recovery in patients undergoing cardiac surgery. Written and informed consent has been obtained from patient to report this case.
## Case report
A 46-year-old female weighing 60 kg, a diagnosed case of Gilbert syndrome was posted for MVR. Her liver function profile [ [fig_ref] Table 1: Laboratory investigations [/fig_ref] ] revealed total bilirubin of 20.6 mg/dl with unconjugated bilirubin of 19.8 mg/dl. Preoperative echocardiography showed severe mitral stenosis, moderate tricuspid regurgitation (TR), moderate pulmonary hypertension with mild right ventricular (RV) dysfunction, and a left ventricular ejection fraction of 65%.
She was premedicated with oral. alprazolam 0.25 mg and fasted for 8 h but allowed clear water up to 2 hours before surgery. 5% dextrose infusion was started in the morning. Standard monitors including 5 lead electrocardiogram, invasive arterial blood pressure,
## Case report
Banashree Mandal, Srinath Damodaran, Harkant S Baryah 1 , Gayathri Warrier After systemic heparinisation with 4 mg/kg of injection heparin, cannulation was performed. Proper position of IVC cannula was confirmed by TOE. Normothermic (35°C-36°C) CPB was maintained with flow of 2.2-2.5 L/min/m2, and a mean arterial pressure of 70-80 mmHg. The mitral valve was replaced with 29 mm St. Jude's mechanical valve. Total ischaemia time and CPB time were 78 and 98 min, respectively. The patient was weaned from CPB with inotropic support of inj. milrinone 0.3 µg/kg/min and noradrenaline 0.05 µg/kg/min. Post CPB TOE showed mild right ventricular dysfunction and moderate tricuspid regurgitation, normal mitral prosthetic valve function, and hepatic blood flow was 175 ml/min. The patient was extubated after 4 h of ventilation and allowed orally 2 h after extubation. Postoperative pain was managed with inj. diclofenac 50 mg twice daily (IV) and intermitted boluses of inj. fentanyl with total opioid of 130 µg over next 48 h. Patient was discharged from intensive care unit after 5 days and hospital after 8 days.
# Discussion
Incidence of hyperbilirubinaemia following cardiac surgery is 10-40% and is associated with mortality as high as 25%. [bib_ref] Hyperbilirubinemia after cardiac surgery: Incidence, risk factors, and clinical significance, Wang [/bib_ref] Preoperative total bilirubin concentration, the number of valves replaced, and preoperative right atrial pressure are the most important risk factors for prediction of the postoperative hyperbilirubinaemia and mortality. [bib_ref] Hyperbilirubinemia after cardiac surgery: Incidence, risk factors, and clinical significance, Wang [/bib_ref] Our patient categorised to class B (modified Child-Pugh classification) was associated with mortality ranging from 18% to 80% while undergoing cardiac surgery. [bib_ref] Cardiac surgery in patients with end-stage liver disease, Diaz [/bib_ref] We took various measures to prevent further increase in bilirubin level perioperatively. Before CPB commencement, IVC cannula position was checked by TOE to prevent further liver damage during CPB. Hepatic blood flow measurement prior to CPB and after coming off CPB gives us objective idea about any compromise in hepatic perfusion during cardiac surgery. Its feasibility has been demonstrated in cardiac as well as noncardiac surgical patients. [bib_ref] Hepatic blood flow and right ventricular function during cardiac surgery assessed by..., Gfirdeback [/bib_ref] [bib_ref] The effects of intraabdominally insufflated carbon dioxide on hepatic blood flow during..., Meierhenrich [/bib_ref] Hyperbilirubinaemia can occur after CPB due to haemolysis caused by cardiotomy suction, hypothermia, the membrane oxygenator, and various other elements of CPB and nonpulsatile perfusion causing hepatic ischaemia. [bib_ref] Suction-induced hemolysis at various vacuum pressures: Implications for intraoperative blood salvage, Gregoretti [/bib_ref] [bib_ref] Mechanisms of hemolysis with mitral prosthetic regurgitation study using transesophageal echocardiography and..., Garcia [/bib_ref] [bib_ref] Pulsatile and non-pulsatile per-fusion: The continuing controversy, Hornick [/bib_ref] Similarly, blood transfusion increases bilirubin load. To prevent that, we followed CPB management goals, such as high pump flow, maintaining mean arterial pressure above 70 mmHg, normothermic bypass, and use of modified ultrafiltration. Modified ultrafiltration technique helps in maintaining hematocrit during CPB, which avoids blood transfusion further, limits hyperbilirubinaemia. Intrathecal administration of opioids significantly reduces intravenous opioids thus aids in fast tracking of patient undergoing cardiac surgery. [bib_ref] Intrathecal morphine for coronary artery bypass graft procedure and early extubation revisited, Chaney [/bib_ref] Although the onset of analgesia following intrathecal administration of fentanyl is <10 min, after morphine is >60 minutes. Duration of action of intrathecal fentanyl is 4 h, whereas morphine is >24 h. Thus, intrathecal administration of both opioids led to rapid onset of analgesia due to fentanyl and prolonged duration of action by morphine.
# Conclusion
Enhanced recovery after surgery is feasible with the use of intrathecal opioids in patient with Gilbert's syndrome undergoing cardiac surgery. Declaration of patient consent The authors certify that they have obtained all appropriate patient consent forms. In the form, the patient(s) has/have given his/her/their consent for his/her/their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed.
Financial support and sponsorship Nil.
## Conflicts of interest
There are no conflicts of interest.
[table] Table 1: Laboratory investigations [/table]
|
Sense and Learn: Recent Advances in Wearable Sensing and Machine Learning for Blood Glucose Monitoring and Trend-Detection
Diabetes mellitus is characterized by elevated blood glucose levels, however patients with diabetes may also develop hypoglycemia due to treatment. There is an increasing demand for non-invasive blood glucose monitoring and trends detection amongst people with diabetes and healthy individuals, especially athletes. Wearable devices and non-invasive sensors for blood glucose monitoring have witnessed considerable advances. This review is an update on recent contributions utilizing novel sensing technologies over the past five years which include electrocardiogram, electromagnetic, bioimpedance, photoplethysmography, and acceleration measures as well as bodily fluid glucose sensors to monitor glucose and trend detection. We also review methods that use machine learning algorithms to predict blood glucose trends, especially for high risk events such as hypoglycemia. Convolutional and recurrent neural networks, support vector machines, and decision trees are examples of such machine learning algorithms. Finally, we address the key limitations and challenges of these studies and provide recommendations for future work.
# Introduction
Blood glucose levels are closely regulated within a desirable range by several hormones released primarily by the pancreas. Diabetes mellitus is a group of metabolic diseases characterized by elevated glucose levels . The 2021 International Diabetes Federation (IDF) atlas has estimated that there are 537 million adults with diabetes and that number is expected to increase to 784 million by 2045. Type 1 diabetes affects between 5 and 10% of patients and is characterized by a lack of insulin production and a higher variability in blood sugars, which requires exogenous insulin and more regular monitoring of blood glucose (American Diabetes Association, 2020). Type 2 diabetes affects up to 90% of all patients diagnosed with diabetes and can be further subdivided into a distinct number of subtypes with varying degrees of insulin deficiency, insulin resistance, and propensity for developing complications [bib_ref] Subtypes of Type 2 Diabetes Determined from Clinical Parameters, Ahlqvist [/bib_ref] [bib_ref] Novel Subgroups of Adult-Onset Diabetes and Their Association with Outcomes: a Data-Driven..., Ahlqvist [/bib_ref].
Normal blood glucose concentrations should lie within a range of 4.0-5.5 mmol/L (72-99 mg/dl) after an 8 h fast and should be < 7.8 mmol/L (140 mg/dl) 2 h after eating [bib_ref] What Is a Normal Blood Glucose? Archives Dis, Güemes [/bib_ref]. Chronic hyperglycemia is associated with long-term microvascular and macrovascular complications [bib_ref] Insulin Therapy, Hyperglycemia, and Hypertension in Type 1 Diabetes Mellitus, De Boer [/bib_ref]. Hypoglycemia occurs due to excess insulin which can be exogenous or endogenous leading to glucose dropping below the normal range. Hypoglycemia occurs in both type 1 diabetes and type 2 diabetes [bib_ref] Hypoglycemia in Type 2 Diabetes, Zammitt [/bib_ref] [bib_ref] Hypoglycemia in Patients with Type 2 Diabetes Mellitus, Miller [/bib_ref]. There are clear American Diabetes Association (ADA) clinical practice recommendations to achieve optimal blood glucose targets using different therapies (American Diabetes Association, 2020a; American Diabetes Association, 2020b), but are inherently associated with an increased risk of hypoglycemia [bib_ref] The Comparative Effects of Intensive Glucose Lowering in Diabetes Patients Aged below..., Ohkuma [/bib_ref] , especially in patients who have reduced awareness of hypoglycemia [bib_ref] Interventions that Restore Awareness of Hypoglycemia in Adults with Type 1 Diabetes:..., Yeoh [/bib_ref].
The ADA categorizes hypoglycemia into: Level 1 with a bloodglucose value between 54 mg/dl and 70 mg/dl; level 2 with a blood-glucose value less than 54 mg/dl; and level 3 with severe hypoglycemia characterized by altered mental and/or physical status requiring external assistance . The symptoms of hypoglycemia are categorized as autonomic, with adrenergic (tremor, palpitations, tachycardia, and anxiety) and cholinergic (sweating, hunger, and paresthesia) manifestations. Neuroglycopenic symptoms may include dizziness, weakness, drowsiness, delirium, confusion, seizure, and coma. The blood-glucose threshold at which an individual patient will experience hypoglycemic symptoms depends on overall glycemic control as the threshold for hypoglycemia will be higher in patients with consistently elevated blood-glucose. Thus, self-monitoring of blood glucose is a key part of diabetes management.
The most widely used glucose monitoring systems rely on disposable test strips that read the blood glucose from a sample obtained using the finger prick approach. This method is associated with discomfort, especially when it is undertaken frequently in patients with type 1 diabetes [bib_ref] Finger Pricking and Pain: A Never Ending Story, Heinemann [/bib_ref]. Continuous glucose monitoring (CGM) devices measure interstitial glucose and have been shown to marginally improve overall glycemic control and reduce the incidence of hypoglycemia [bib_ref] Effect of Flash Glucose Monitoring in Adults with Type 1 Diabetes: a..., Nathanson [/bib_ref] , but suffer from several inherent limitations. There is a time lag between the interstitial glucose measured by CGM and actual blood glucose and there are inconsistencies in the readings of different CGM devices [bib_ref] Analysis of Time Lags and Other Sources of Error of the DexCom..., Kamath [/bib_ref] [bib_ref] A Comparative Effectiveness Analysis of Three Continuous Glucose Monitors, Damiano [/bib_ref]. Furthermore, most CGMs are relatively expensive, which limits the wider adoption of such technology among diabetic patients.
Wearables are lightweight devices capable of measuring different vital signs and modalities such as heart rate, temperature, respiration rate, activity level, and skin conductance [bib_ref] Wearable Devices for Precision Medicine and Health State Monitoring, Jeong [/bib_ref]. Wearable technologies have been proposed to improve the quality of life by collecting and sharing data between people and their carers [bib_ref] A Survey of Wearable Devices and Challenges, Seneviratne [/bib_ref]. Wearable devices have different functions and may include patches, belts, lenses, earphones, socks, glasses, watches, wrist bands, and bracelets [bib_ref] Wearables and the Medical Revolution, Dunn [/bib_ref]. Wearables have been recommended in the elderly to increase their independence and have been proposed for autism screening and therapy [bib_ref] The Usefulness and Actual Use of Wearable Devices Among the Elderly Population, Kekade [/bib_ref] [bib_ref] Sensing Technologies for Autism Spectrum Disorder Screening and Intervention, Cabibihan [/bib_ref] [bib_ref] Social Robots and Wearable Sensors for Mitigating Meltdowns in Autism -A Pilot..., Cabibihan [/bib_ref]. Recently, the readings from wearable technologies have been associated with blood glucose levels [bib_ref] Early Detection of Hypoglycemia in Type 1 Diabetes Using Heart Rate Variability..., Bekkink [/bib_ref] [bib_ref] Detection of Atrial Fibrillation Episodes Using a Wristband Device, Corino [/bib_ref] [bib_ref] Towards Wearable-Based Hypoglycemia Detection and Warning in Diabetes, Maritsch [/bib_ref]. Using such devices to indirectly monitor glucose represents a non-invasive, convenient, and inexpensive method over CGM. A typical glucose monitoring system consists of wearable devices and machine learning models assessed against FDA established evaluation metrics . surveyed different types of non-invasive blood glucose monitoring equipment and highlighted the potential role of artificial intelligence in solving some of the existing challenges [bib_ref] Pain-free Blood Glucose Monitoring Using Wearable Sensors: Recent Advancements and Future Prospects, Siddiqui [/bib_ref]. Non-invasive monitoring of glucose levels from breath samples and epidermal electrochemical glucose sensors hold promise [bib_ref] Recent Advancements and Future Prospects on E-Nose Sensors Technology and Machine Learning..., Lekha [/bib_ref] [bib_ref] Wearable Non-invasive Epidermal Glucose Sensors: A Review, Kim [/bib_ref]. Other noninvasive sensors that measure physiological parameters such as pulse and pulse pressure may help to improve glucose management among diabetic patients, especially during physical activity [bib_ref] Sensor Monitoring of Physical Activity to Improve Glucose Management in Diabetic Patients:..., Ding [/bib_ref]. A wide range of non-invasive techniques for blood glucose monitoring have been developed utilizing glucose monitoring in bodily fluids [bib_ref] Glucose Sensing for Diabetes Monitoring: Recent Developments, Bruen [/bib_ref].
Despite the existence of many studies related to non-invasive glucose monitoring, several questions remain unanswered:
1. What technologies can be used in blood glucose prediction and trends detection? 2. Which machine learning algorithms can be used? 3. To what extent are these technologies and techniques successful? 4. What are the limitations of the existing approaches?
To answer these questions, we have reviewed the advances over the past five years in non-invasive sensors, wearable technologies, and machine learning approaches to estimate blood glucose, especially to identify hypoglycemia. We have identified the major limitations and challenges faced by these studies and provided recommendations to overcome these constraints. We have reviewed studies that have developed non-invasive sensors and wearable devices as well as contributions that have employed existing commercially available devices. Studies without working prototypes were excluded. The considered search keywords included noninvasive or noninvasive, sensor, wearable, glucose, levels, trends, hypoglycemia, monitoring, machine learning, and prediction.
## Non-invasive sensors and wearables
This section presents the advances in wearables and sensors over the past five years. The technologies used have varied from direct measurement of glucose from bodily fluids such as saliva, sweat, and tears to the detection of physiological variables which change with blood glucose . Examples of such wearable devices and sensors are illustrated in . A summary of the surveyed studies is provided in [fig_ref] TABLE 1 |: A summary of the advances in non-invasive sensors and wearable technologies for... [/fig_ref].
## Photoplethysmography
Heart rate variability (HRV) reflects physiological changes in the human body and provides insights on cardiac health and the autonomic nervous system, which can be used as an indicator of current or impending diseases [bib_ref] Associations between Colonic Motor Patterns and Autonomic Nervous System Activity Assessed by..., Yuan [/bib_ref]. For example, it was considered as a predictor in rapid renal function deterioration [bib_ref] Heart Rate Variability as a Predictor of Rapid Renal Function Deterioration in..., Chou [/bib_ref] , sudden cardiac death [bib_ref] Heart Rate Variability as Predictive Factor for Sudden Cardiac Death, Sessa [/bib_ref] , and stroke or post-stroke complications [bib_ref] Heart Rate Variability as a Biomarker for Predicting Stroke, Post-stroke Complications and..., Lees [/bib_ref]. The analysis of HRV relies on methods in either timedomain or frequency domain. The root mean square of successive differences (RMSSD), the standard deviation of normal to normal interval (SDNN), and average heart rate (HR) are widely used time domain indices [bib_ref] Heart Rate Variability: a, Rajendra Acharya [/bib_ref]. High frequency (HF), low frequency (LF) and very low-frequency (VLF) are examples of frequency-domain parameters. HRV has been associated with the severity of hypoglycemia and fluctuation of interstitial glucose [bib_ref] Association between Severity of Hypoglycemia and Loss of Heart Rate Variability in..., Silva [/bib_ref] [bib_ref] Heart Rate Variability Is Associated with Interstitial Glucose Fluctuations in Type 2..., Klimontov [/bib_ref].
Estimating HRV parameters from wearables to predict hypoglycemia has been investigated in previous studies with changes in HRV occurring up to 90 min prior to hypoglycemia FIGURE 2 | A broad classification of the wearable and sensor technologies considered in glucose monitoring over the past five years.
FIGURE 3 | Examples of developed wearable devices and sensors that are aimed to aid in blood glucose management. (A) A wearable mouthguard biosensor that measures salivary glucose levels (Adapted with permission from. (B) Wearable sensor to analyze perspiration glucose (Adapted with permission from . (C) A biosensor that can measure tear glucose levels (Adapted with permission from [bib_ref] Clinical Evidence for Use of a Noninvasive Biosensor for Tear Glucose as..., Kownacka [/bib_ref]. (D) An electromagnetic wearable glove for continuous glucose monitoring (Adapted with permission from [bib_ref] Noninvasive, Wearable, and Tunable Electromagnetic Multisensing System for Continuous Glucose Monitoring, Mimicking..., Hanna [/bib_ref]. [bib_ref] Early Detection of Hypoglycemia in Type 1 Diabetes Using Heart Rate Variability..., Bekkink [/bib_ref]. Empatica E4 (Empatica Inc., US) is one example of a wearable used to estimate HRV parameters utilizing a photoplethysmography (PPG) sensor to measure the blood volume pulse. The PPG sensor in Empatica E4 has been shown to provide accurate arrhythmia classification [bib_ref] Detection of Atrial Fibrillation Episodes Using a Wristband Device, Corino [/bib_ref]. A previous study has demonstrated that Empatica E4 can be used to detect the occurrence of hypoglycemia [bib_ref] Towards Wearable-Based Hypoglycemia Detection and Warning in Diabetes, Maritsch [/bib_ref]. The study collected HRV time features (e.g., RMSSD) extracted from the recorded interbeat intervals from the wearable device and interstitial glucose readings using CGM (i.e., FreeStyle Libre) from a participant with T1DM. Using a machine learning model based on decision trees (i.e., gradient boosting), the developed model was able to predict hypoglycemia in the test set with an accuracy of 82.7%. The study had major limitations in terms of the features considered, physiological signals, and was only tested on one participant. However, the initial results demonstrated that wearables can indeed be used to predict hypoglycemic events using HRV parameters. A recent study demonstrated the application of a PPG device in the classification of blood glucose levels into normal and diabetic. With PPG signal and blood glucose data collected from 80 participants, the study reported an accuracy of 98% with a decision tree based classifier. Another recent study investigated blood glucose prediction based on PPG and physiological data acquired from 2,538 participants split into two groups with or without medication. Seventeen features were extracted from the collected data and were used in developing the prediction model. With quarterly measured HbA1c, the best model achieved an accuracy of 94.3% with an RMSE value of 12.4 mg/dl for the group without medication while the results for the group with medication were limited.
## Electrocardiogram
Electrical signals arise in the SA node in the right atrium, travel to the atrioventricular node in the interatrial septum and diverge through the left and right bundle of His to the Purkinje fibers terminating in the endocardium and ventricular epicardium causing ventricular contraction. Cardiac electrical activity can be monitored using an Electrocardiogram (ECG). A typical heartbeat is partitioned into three segments, namely PR segment, QRS complex, and ST segment and each segment can be used to identify underlying cardiac conduction defects such as short QT-interval or PR-interval [bib_ref] Prevalence, Predictors, and Prognostic Implications of PR Interval Prolongation in Patients with..., Nikolaidou [/bib_ref] [bib_ref] Early Repolarization and Risk of Arrhythmia Events in Long QT Syndrome, Hasegawa [/bib_ref] [bib_ref] Cardiac Autonomic Regulation and PR Interval Determination for Enhanced Atrial Fibrillation Risk..., Kališnik [/bib_ref].
The advancement in sensor miniaturization has enabled current wearables to measure changes in cardiac conduction and contraction with good accuracy. The Fitbit Sense smartwatch and Apple watch series 4 are examples of wrist wearables that can measure the heart rate and ECG signal to identify an irregular heart rhythm, e.g., atrial fibrillation. Chest sensors are another trend in wearables that can measure electrical signals from the heart. The Bioharness (3.0, Zephyr Technology, US) is a lightweight, portable chest wearable device that can provide live access to a variety of physiological parameters, including ECG and it has been tested to predict hypoglycemia. The HealthPatch (VitalConnect, San Jose, CA) is another example of an ECG wearable used to derive HRV to investigate hypoglycemia [bib_ref] Early Detection of Hypoglycemia in Type 1 Diabetes Using Heart Rate Variability..., Bekkink [/bib_ref]. Other promising wearable ECG devices that provide ease of monitoring include the Bittium Faros (Bittium Corporation, Finland) and Lifetouch (Isansys Lifecare Ltd., United Kingdom). Changes in blood glucose can alter cardiac repolarization and induce prolongation of the QT interval with an increased risk of cardiac arrhythmia [bib_ref] Cardiac Autonomic Regulation and PR Interval Determination for Enhanced Atrial Fibrillation Risk..., Kališnik [/bib_ref]. These changes in cardiac conduction prior to the development of arrythmia could be used to predict the occurrence of hypoglycemia [bib_ref] Non-invasive Hypoglycemia Monitoring System Using Extreme Learning Machine for Type 1 Diabetes, Ling [/bib_ref] [bib_ref] Precision Medicine and Artificial Intelligence: A Pilot Study on Deep Learning for..., Porumb [/bib_ref]. In a recent work, the relationship between ECG data and blood glucose in adults has been investigated [bib_ref] Modelling the Relationship between Continuously Measured Glucose and Electrocardiographic Data in Adults..., Charamba [/bib_ref]. The study collected the ECG, diary, CGM data for one week from seventeen patients with type 1 diabetes. The study identified a negative relationship between QTC and hypoglycemia.
A recent study utilized an ECG device (i.e., Bioharness) and a glucose monitoring system (i.e., FreeStyle Libre Flash) to develop personalized models to predict the occurrence of nocturnal hypoglycemic events [bib_ref] Precision Medicine and Artificial Intelligence: A Pilot Study on Deep Learning for..., Porumb [/bib_ref]. The study acquired the ECG data from healthy participants monitored for up to 14 days in relation to blood glucose levels below 4 mmol/L to define a hypoglycemia threshold. Using machine learning techniques, the study showed the feasibility of predicting nocturnal hypoglycemia from raw ECG signals. However, the participants in the study were all healthy, hence, low blood glucose instances were limited. Furthermore, the collected data were limited to nocturnal events, thus, ECG circadian changes during the day in relation to glucose concentration were not investigated. Another study acquired ECG and glucose data from 16 healthy adults and 5 with prediabetes to establish a machine learning model based on CNN to classify blood glucose values into three groups, namely, low (i.e., below 6.0 mmol/L), moderate (i.e., above 6.0 mmol/L and below 7.7 mmol/L), and high (i.e., above 7.7 mmol/L) [bib_ref] Non-invasive Monitoring of Three Glucose Ranges Based on Ecg by Using Dbscan-Cnn, Li [/bib_ref]. The results of the best trained model showed an accuracy of 81.69% in classifying the blood glucose values.
## Electromagnetic
Non-ionizing parts of the electromagnetic (EM) spectrum (e.g., visible light, radio, and ultraviolet) ranging from 0 to 3.0 PHz have been used for diagnostic and therapeutic medical applications [bib_ref] Emerging Medical Applications Based on Non-ionizing Electromagnetic Fields from 0 Hz to..., Mattsson [/bib_ref]. Non-invasive estimation of blood glucose has been investigated using electromagnetic waves and near infrared (NIR) waves based on the unique absorption spectrum of glucose [bib_ref] Noninvasive Electromagnetic Wave Sensing of Glucose, Zhang [/bib_ref] [bib_ref] Sensitivity Optimization of Antenna for Non-invasive Blood Glucose Monitoring, Jiang [/bib_ref] [bib_ref] Noninvasive Blood Glucose Detection Using Near Infrared Sensor, Hotmartua [/bib_ref] [bib_ref] Near-infrared LED Based Non-invasive Blood Glucose Sensor, Yadav [/bib_ref] [bib_ref] Non-invasive Prediction of Blood Glucose Trends during Hypoglycemia, Tronstad [/bib_ref]. Transmittance, which measures the scattered light after penetrating the tissue, and reflectance, which measures the reflected light from the skin surface, are two methods that rely on light to acquire information about a substance [bib_ref] Noninvasive Blood Glucose Detection Using Near Infrared Sensor, Hotmartua [/bib_ref]. By investigating the properties of both the reflected and transmitted waves, the level of glucose can be estimated [bib_ref] Near-infrared LED Based Non-invasive Blood Glucose Sensor, Yadav [/bib_ref].
Several devices have been developed that show that changes in EM correlate with glucose concentrations [bib_ref] Reflection and Transmission Measurements Using 60 Ghz Patch Antennas in the Presence..., Cano-Garcia [/bib_ref] [bib_ref] Non-invasive Monitoring of Glucose Level Changes Utilizing a Mm-Wave Radar System, Shaker [/bib_ref]. A microstrip patch antenna (1.50 mm × 1.50 mm) utilizing a millimetre band of the EM spectrum has been developed to sense glucose by using two facing antennas operating at 60 GHz to assess the variations in permittivity across the signal path [bib_ref] A Glucose Sensing System Based on Transmission Measurements at Millimetre Waves Using..., Saha [/bib_ref]. The device was capable of detecting changes in small glucose concentrations of 1.33 mmol/L in water-based samples and glucose spikes in humans and therefore has potential applications to detect hypoglycemia. However, the system was limited to experimental settings and was susceptible to noise (e.g., hand motion). Another study demonstrated the design of a compact antenna that operates in the frequency range of 24.1-29.3 GHz and has a bandwidth of 5.2 GHz [bib_ref] An Electromagnetic Band Gap-Based Complementary Split Ring Resonator Loaded Patch Antenna for..., Raj [/bib_ref]. The measured reflection coefficient of the antenna showed deviations due to changes in the electrical properties of plasma glucose, which could potentially be used to measure glucose concentration. Another study demonstrated the design of a circular two cell split ring resonator microwave sensor that displayed sensitivity to changes in glucose concentration in water [bib_ref] Hyper-sensitive Microwave Sensor Based on Split Ring Resonator (Srr) for Glucose Measurement..., Zidane [/bib_ref]. Their device displayed a glucose detection resolution of 1.665 mmol/L at a frequency of 1.9 GHz. Another study developed a microwave sensor with a relatively wide passband that correlated with blood glucose changes [bib_ref] Application of Broadband Microwave Near-Field Sensors for Glucose Monitoring in Biological Media, Zapasnoy [/bib_ref]. Glucose concentration between 0 and 25 mmol/L in sodium chloride solution could be detected with 1 mmol/L resolution in the frequency range of 1.4-1.7 GHz.
Recently, an innovative EM-based glove wearable device has been developed to monitor blood glucose levels [bib_ref] Noninvasive, Wearable, and Tunable Electromagnetic Multisensing System for Continuous Glucose Monitoring, Mimicking..., Hanna [/bib_ref]. The system consists of two flexible sensors (i.e., slot antenna and a reject filter) and operates in the frequency range of 500 MHz and 3 GHz, which reaches subcutaneous veins and arteries. The design of the multiband antenna was made to imitate the vascular anatomy of the hand which improved its sensitivity by concentrating the EM waves on the blood network allowing the monitoring of glucose over a wider frequency range. When the device was tested with glucose solutions at various concentrations, the reflection coefficients varied with the changes in glucose and achieved a high correlation (i.e., greater than 0.90). In in vivo experiments, the device showed high correlation (i.e., greater than 0.89) with hypoglycemia and hyperglycemia and reportedly high accuracy on the Clarke's error grid. Whilst it showed promise in estimating glucose levels in healthy participants, the device is still an experimental prototype that requires extra circuitry and was only tested in controlled conditions without physical activity. Subsequently, the same research team demonstrated high fidelity in serum from 21 participants and achieved 98% accuracy against reference glucose levels [bib_ref] A Vasculature Anatomy Inspired Flexible Slot Antenna for Continuous Non-invasive Glucose Monitoring, Hanna [/bib_ref].
## Bioimpedance
Bioimpedance measures the response of a biological medium (e.g., human body) to an electric current. The composition of the biological mediums affects the bioimpedance based on whether they act as insulators, dielectrics, or conductors (Naranjo-Hernández et al., 2019). Hence, bioimpedance measurements can be used to acquire information about body composition, such as fat, muscle, and water and thereby assess obesity ; sarcopenia in patients with renal disease [bib_ref] Sarcopenia and Cardiovascular Risk Indices in Patients with Chronic Kidney Disease on..., Lai [/bib_ref] , and gastrointestinal disease [bib_ref] A Portable Bioimpedance Measurement System Based on Red Pitaya for Monitoring and..., Ruiz-Vargas [/bib_ref]. Bioimpedance analysis has been suggested as a non-invasive method to screen for diabetes mellitus [bib_ref] Glucoseindependent Segmental Phase Angles from Multi-Frequency Bioimpedance Analysis to, Jun [/bib_ref].
Some studies have investigated the correlation between changes in glucose levels and bioimpedance [bib_ref] Early Detection of Diabetes Based on Skin Impedance Spectrogram and Heart Rate..., Das [/bib_ref] Frontiers in Bioengineering and Biotechnology | www.frontiersin.org May 2022 | Volume 10 | Article 876672 [bib_ref] A Noninvasive Method for Measurement of Blood Glucose Using Bio Impedance Technique, Jose [/bib_ref] [bib_ref] Impedance Spectroscopy of Aqueous Solution Samples of Different Glucose Concentrations for the..., Satish [/bib_ref]. A recent study showed an inverse relationship between glucose concentration and the bioimpedance difference in blood volume . Another study identified that a frequency band of below 40 kHz provided stable and reliable estimation for blood glucose based on bioimpedance [bib_ref] Measurement Frequency Evaluation for Bioimpedance-Based Blood-Glucose Estimation, Takamatsu [/bib_ref]. Another study demonstrated a wearable prototype system for non-invasive glucose monitoring based on bioimpedance measurement and showed that certain parameters of bioimpedance were sensitive to changes in blood glucose levels . Another study proposed a hybrid technique that combined bioimpedance with nearinfrared measurements to monitor glucose. Using machine learning (i.e., regression), the combination of the two measurements achieved better results based on Clarke's error grid (i.e., 90% points in region A) when compared to the reference blood glucose. However, the study was limited to one participant. In relation to blood trends, a recent study assessed noninvasive sensors that included bioimpedance measurements in detecting hypoglycemia among 20 patients with type 1 diabetes that underwent clamp procedures [bib_ref] Non-invasive Prediction of Blood Glucose Trends during Hypoglycemia, Tronstad [/bib_ref]. The study revealed that bioimpedance plays a correcting role in the prediction when paired with other sensors. Bioimpedance of the skin has also been used to detect nocturnal hypoglycemic events [bib_ref] Design and Validation of a Low-Cost Non-invasive Device to Detect Overnight Hypoglycemia, Lesko [/bib_ref]. Another study investigated the galvanic skin response (GSR) and correlated it with blood glucose levels [bib_ref] Non-invasive Blood Glucose Analysis Based on Galvanic Skin Response for Diabetic Patients, Snekhalatha [/bib_ref] , where a device was developed to acquire the GSR resistance and voltage data from 100 participants (i.e., 50 with diabetes). A negative correlation with GSR voltage and resistance was observed among diabetic patients. A recent study in 14 patients with type 2 diabetes collected bioimpedance data using a wearable ring over 2 weeks and used a gradient boosted model to estimate blood glucose levels and trends and showed excellent prediction performance with 99% of the values in zones A and B of the Parkes error grid [bib_ref] Evaluation of a Continuous Blood Glucose Monitor: A Novel and Noninvasive Wearable..., Sanai [/bib_ref].
## Sweat
Sweating is a normal physiological mechanism to regulate body temperature through evaporation, but it is also a key autonomic feature of hypoglycemia [bib_ref] Early Detection of Hypoglycemia Events Based on Biometric Sensors Prototyped on FPGAs, Escolar [/bib_ref]. Sweat consists mostly of water but also contains sodium, chloride, potassium, lactate, and urea [bib_ref] Physiology of Sweat Gland Function: The Roles of Sweating and Sweat Composition..., Baker [/bib_ref]. Chemical components in the sweat have been utilized as biomarkers of disease, e.g., sweat chloride in cystic fibrosis [bib_ref] Sweat Chloride as a Biomarker of CFTR Activity: Proof of Concept and..., Accurso [/bib_ref]. A study used different non-invasive sensors (sweat, temperature, and ECG) in patients with type 1 diabetes and showed that measurement of sweating in combination with the ECG signal predicted the development of hypoglycemia [bib_ref] Evaluation of Hypoglycaemia with Non-invasive Sensors in People with Type 1 Diabetes..., Elvebakk [/bib_ref]. Sweat also contain glucose at orders of magnitude lower in concentration (10-200 μM) compared to blood glucose [bib_ref] Development of Robust Calibration Models Using Support Vector Machines Frontiers in Bioengineering..., Bariya [/bib_ref] , but require careful considerations to avoid contamination when collected from the skin surface [bib_ref] Correlation between Sweat Glucose and Blood Glucose in Subjects with Diabetes, Moyer [/bib_ref].
Wearable sensors can take advantage of the non-invasive nature of using sweat as a predictor of the human health status [bib_ref] SLOCK (Sensor for Circadian Clock): Passive Sweat-Based Chronobiology Tracker, Upasham [/bib_ref]. There is a growing interest in developing sweat-based sensors and systems that are aimed to monitor health to help in the management of patients with diabetes [bib_ref] Noninvasive Diabetes Monitoring through Continuous Analysis of Sweat Using Flow-Through Glucose Biosensor, Karpova [/bib_ref]. One study developed a microfluidic device using a cotton thread and filter paper paired with a smartphone to sense sweat glucose [bib_ref] A Wearable, Cotton Thread/paper-Based Microfluidic Device Coupled with Smartphone for Sweat Glucose..., Xiao [/bib_ref]. The device showed a linear trend in the 50-250 μM range with a detection limit of 35 μM and the nature of the sensor construction enabled it to be flexible, easy to integrate, and to be produced at relatively low-cost. Another study developed a biosensor based on a graphene oxide nanostructured composite deposited with gold and platinum nano-particles to detect glucose in human sweat [bib_ref] A Wearable Electrochemical Glucose Sensor Based on Simple and Low-Cost Fabrication Supported..., Xuan [/bib_ref]. When tested with sweat samples, the device showed a short response time and high linearity. [bib_ref] Smartphone-addressable 3d-Printed Electrochemical Ring for Nonenzymatic Self-Monitoring of Glucose in Human Sweat, Katseli [/bib_ref] demonstrated that a 3D printed electrochemical sweat sensor shaped like a ring was capable of monitoring sweat glucose in the range of 12.5-400 μmol/L, which could be read by a smartphone. However, prototype testing was limited to one healthy volunteer. Similarly, [bib_ref] Touch-based Fingertip Blood-free Reliable Glucose Monitoring: Personalized Data Processing for Predicting Blood..., Sempionatto [/bib_ref] developed a touch-based sweat glucose sensor to estimate blood glucose. The electrochemical sensor consists of a sweat collecting layer, glucose biosensor, and a substrate that requires no sweat stimulation. The sensor achieved a high correlation (i.e., 0.95) with blood glucose and with all the points in the A and B regions on the Clarke error grid. Another study developed a tandem catalytic system for sweat glucose detection based on chemiluminescence with a high sensitivity and detection limit of 0.1 μM when compared to solutions containing different glucose concentrations [bib_ref] A Novel Luminescent "Nanochip" as a Tandem Catalytic System for Chemiluminescent Detection..., Gao [/bib_ref].
Smart wrist wearables are becoming an essential part of fitness and health monitoring. The convenience of wearing watches on the wrist only widened the adoption of such wearable devices. Several wrist wearable sensors have been developed to detect sweat glucose [bib_ref] Multifunctional Wearable System that Integrates Sweat-Based Sensing and Vital-Sign Monitoring to Estimate..., Hong [/bib_ref] [bib_ref] A Wearable Electrochemical Glucose Sensor Based on Simple and Low-Cost Fabrication Supported..., Xuan [/bib_ref] [bib_ref] Wearable Sweat Monitoring System with Integrated Micro-supercapacitors, Lu [/bib_ref]. One study demonstrated a nonenzymatic wearable sensor that allowed the analysis of sweat glucose . The sensor was made from a treated silver electrode coated with fluorocarbon-based materials. An integrated wristband containing the sensor provided continuous monitoring of sweat glucose and showed the results on a smartphone App. Their solution demonstrated the possibility of detecting glucose in the range of 30-1,100 μM. However, the developed wearable was tested with samples acquired from participants only and no correlations with blood glucose were made. Another research group developed a fully integrated device to provide continuous monitoring of sweat glucose . The device consists of flexible rechargeable batteries and photovoltaic cells that are used to power up the device (i.e., signal processing and display) using solar energy. Monitoring of sweat glucose is based on an electrochemical sensor connected to a controlling module. A small display is used to provide real time monitoring. The wearable displayed potential in detecting sweat glucose changes in the range of 50-200 μM during different activities (e.g., running vs. biking).
## Tears
The human eyes produce tears as a response to irritants, due to intense emotions, and to keep the ocular surface lubricated and protected. Tears are made of water, protein, lipids, and electrolytes [bib_ref] Tears and Their Secretion, Dartt [/bib_ref] , but also contain traces of glucose that correlate with blood glucose levels [bib_ref] Analysis of Tear Fluid by Ce/lif: a Noninvasive Approach for Glucose Monitoring, Chen [/bib_ref] [bib_ref] Noninvasive Diagnostic Devices for Diabetes through Measuring Tear Glucose, Zhang [/bib_ref]. The concentration of tear glucose is influenced by the method of collection. For example, a study found that onion-induced tear glucose concentration is up to 8 fold higher compared to one without stimulation [bib_ref] Analysis of Tear Glucose Concentration with Electrospray Ionization Mass Spectrometry, Taormina [/bib_ref]. This was attributed to the level of irritation in the onion-induced method that influenced the collected samples. Hence, a consistent method to collect the tears must be selected and careful consideration must be paid to the surrounding conditions [bib_ref] Evaluation of Commonly Used Tear Sampling Methods and Their Relevance in Subsequent..., Rentka [/bib_ref]. Despite the complicated nature of collecting tear samples, there is a growing interest in the development of sensors capable of tear glucose monitoring [bib_ref] A Non-enzymatic Glucose Sensor Enabled by Bioelectronic pH, Strakosas [/bib_ref].
A ratiometric fluorescent membrane capable of sensing tear glucose in the range of 0.1-10 mM has been developed [bib_ref] Development of a Ratiometric Fluorescent Glucose Sensor Using an Oxygen-Sensing Membrane Immobilized..., Duong [/bib_ref]. However, the testing conditions were limited to glucose solutions. Another study developed a low-cost, flexible, customizable, and disposable sensor strip based on engraved graphene to detect glucose in tears and saliva [bib_ref] Facile and Scalable Disposable Sensor Based on Laser Engraved Graphene for Electrochemical..., Tehrani [/bib_ref]. The sensor displayed a promising sensitivity and low detection limit (i.e., 250 nM) when tested in vitro. However, no testing with real samples of human tears or saliva has been undertaken. A low-cost and non-enzymatic glucose sensor based on an inkjet printed electrochemical sensor was developed in another study. The sensor was flexible and versatile in terms of fabrication, and demonstrated its ability in detecting glucose concentrations in human tears. However, tears were induced using onion and collected in glass capillaries. To overcome some of the limitations in tear glucose sensors, [bib_ref] Self-monitoring of Tear Glucose: the Development of a Tear Based Glucose Sensor..., Belle [/bib_ref] developed an integrated device that has a broad dynamic range, rapid analysis, low detection limit, works with small samples, and does not cause stress on the eye. The sensor could detect tear glucose in the range of 0.72 mg/dl to 111.6 mg/dl corresponding to 20 mg/dl to 600 mg/dl of blood glucose. However, the developed device was only tested in samples from an animal.
The development of wearable contact lenses capable of monitoring different physiological signs has gained considerable interest in recent years [bib_ref] Wearable Contact Lens Biosensors for Continuous Glucose Monitoring Using Smartphones, Elsherif [/bib_ref] [bib_ref] Wearable Smart Sensor Systems Integrated on Soft Contact Lenses for Wireless Ocular..., Kim [/bib_ref]. A contact lens comprised of three layers (silk fibroin, silver nanowires, and protected film) capable of sensing tear glucose in the range of 500 nM to 1 mM with a detection limit of 211 nM was developed by . Another contact lens was developed by [bib_ref] Wearable Smart Sensor Systems Integrated on Soft Contact Lenses for Wireless Ocular..., Kim [/bib_ref] which monitors tear glucose based on glucose oxidase linked to a graphene channel that can be read wirelessly by a coil. The in vivo experiments showed that the wearable sensor can detect tear glucose concentration when placed in a rabbit's eye. Another study developed a fully integrated soft contact lens that contains glucose sensors, wireless circuits, and a display . The developed smart contact lens is supposed to overcome some of the limitations of existing contact lens such as being brittle, blocking the vision, and requiring extra equipment to read the lens. The wearable was able to detect tear glucose when tested in vivo on a live rabbit.
An optical sensor embedded in a wearable contact lens was developed to provide continuous glucose monitoring [bib_ref] Wearable Contact Lens Biosensors for Continuous Glucose Monitoring Using Smartphones, Elsherif [/bib_ref]. The reading of the sensor was based on smartphone camera readouts that correlated the reflected power of the diffraction with glucose concentration. The developed sensor could detect glucose concentration less than 50 mM with a 12 nm mM −1 sensitivity. However, no in vivo experiments were reported. They subsequently developed a bifocal contact lens containing a hydrogel glucose sensor that could detect tear glucose within the 0-3.3 mM range in artificial tears. Another study developed and clinically tested a flexible tear glucose biosensor [bib_ref] Clinical Evidence for Use of a Noninvasive Biosensor for Tear Glucose as..., Kownacka [/bib_ref]. The coilshaped sensor is 1.3 mm in diameter and 15 mm long and consists of a flexible coil made of electrodes arranged in parallel and an antenna wire that can transfer the readings to an external device through telemetry. The sensor was designed to be placed under the lower eyelid and was reported to not cause any irritation or abnormalities in a sheep's eye. Clinical testing was conducted with six subjects who wore the sensors and CGMs (i.e., Abbott FreeStyle Libre) for 5 hours in a fasted state. On Clarke's error grid, 95% of the data points for the sensor were in the A and B regions with 70% in region A, which was comparable to the performance of the CGM. Geelhoed-Duijvestijn et al. (2021) evaluated the same biosensor in 24 patients with type 1 diabetes by simultaneously measuring blood and interstitial glucose levels every 15 min compared to continuous tear glucose and a neural network based regression model was used to convert tear glucose to blood glucose. The performance of the device was comparable to the CGM with a mean absolute relative difference in glucose of 16.7 mg/dl.
## Saliva
Saliva is a clear and slightly acidic secretion originating from the sublingual, submaxillary, parotid, and minor mucous glands and serves to lubricate and clean the oral tissues and assist in taste and digestion [bib_ref] A Review of Saliva: Normal Composition, Flow, and Function, Humphrey [/bib_ref] [bib_ref] Rhythms in Salivary Flow Rate and Composition, Dawes [/bib_ref] [bib_ref] Saliva Composition and Functions: a Comprehensive Review, De Almeida [/bib_ref]. Different concentrations of various electrolytes and minerals can be found in saliva including carbon dioxide, sodium, chloride, and potassium as well as traces of glucose [bib_ref] Salivary Secretion of Electrolytes, Schneyer [/bib_ref] [bib_ref] Excretion of Sodium, Potassium, Chloride and Carbon Dioxide in Human Parotid Saliva, Thaysen [/bib_ref] [bib_ref] Saliva Composition and Functions: a Comprehensive Review, De Almeida [/bib_ref]. The excretion and concentration of salivary glucose has been found to be higher in diabetic patients compared to control subjects and there is a significant correlation between the concentration of glucose in the saliva and blood in patients with diabetes [bib_ref] Salivary Glucose Concentration and Excretion in Normal and Diabetic Subjects, Jurysta [/bib_ref] [bib_ref] Salivary Glucose Concentrations in Patients with Diabetes Mellitus-Aa Minimally Invasive Technique for..., Amer [/bib_ref].
Interest in the use of saliva as a diagnostic fluid has grown considerably and several sensors have been developed [bib_ref] Clinical and Diagnostic Utility of Saliva as a Non-invasive Diagnostic Fluid: a..., Nunes [/bib_ref] [bib_ref] A Fully Inkjet-Printed Disposable Glucose Sensor on Paper, Bihar [/bib_ref]. A study fabricated a non-enzymatic electrochemical sensor to measure salivary glucose with a working range varying from 0.5 to 50 μg/ml and a detection limit of 1.9 μg/ml and showed a highly significant correlation (r = 0.96) with blood glucose measured using the finger prick method Frontiers in Bioengineering and Biotechnology | www.frontiersin.org May 2022 | Volume 10 | Article 876672 [bib_ref] A Nonenzymatic Electrochemical Glucose Sensor Based on Molecularly Imprinted Polymer and its..., Diouf [/bib_ref]. Another study fabricated a saliva glucose optical sensor and showed that the glucose concentration increased the absorbance of light when tested at a wavelength of 630 nm [bib_ref] A Lab-On-A-Chip-Based Non-invasive Optical Sensor for Measuring Glucose in Saliva, Jung [/bib_ref]. There was a good correlation between the glucose in blood and saliva. A disposable saliva glucose sensor based on dehydrogenase flavine-adenine dinucleotide was tested in nine healthy individuals and showed a detection range of 2.38-3.40 mg/dl corresponding to a blood glucose range between 90 and 143 mg/dl at a detection limit of 0.11 mg/dl [bib_ref] Toward the Development of a Glucose Dehydrogenase-Based Saliva Glucose Sensor without the..., Lin [/bib_ref]. A salivary glucose biosensor utilized a new hydrogel film to improve glucose detection sensitivity by 130% with good linearity for glucose concentration between 0 and 50 mg/L . Another study showed that enzymatic biosensors provided a linear relationship between electrical impedance and glucose concentration with the lowest detection limit being 14 μM [bib_ref] Design of a Bioelectronic Tongue for Glucose Monitoring Using Zinc Oxide Nanofibers..., Mercante [/bib_ref]. Another study developed bioconjugated nanoflowers which quickly (i.e., within 10 min) estimated salivary glucose concentration between 0.2-300 mg/dl [bib_ref] Bioconjugated Nanoflower for Estimation of Glucose from Saliva Using Nanozymes, Shende [/bib_ref]. There have been attempts to incorporate saliva glucose sensors into devices used daily to measure glucose levels. One study embedded a saliva glucose sensor into a smart toothbrush which integrated a bronze based sensor to provide non-enzymatic electrochemical measurement of salivary glucose . The sensor showed a linear range from 0 to 320 μM with a detection limit of 6.6 μM. To test the sensor, saliva samples were acquired from five participants before and after meals. The sensor readings for the saliva glucose reflected well with the changes in blood glucose values. Embedding a saliva glucose sensor with wireless communication capabilities into a mouthguard has also been considered [bib_ref] Mouthguard Biosensor with Telemetry System for Monitoring of Saliva Glucose: A Novel..., Arakawa [/bib_ref]. The same team embedded a biosensor based on cellulose acetate into a mouthguard and was able to detect glucose concentration wirelessly in the range of 1.75-10,000 μmol/L. However, the tests were limited to artificial saliva.
## Acceleration
Involuntary shaking part of the human body, such as the hand, is one manifestation of the symptoms that are associated with hypoglycemia [bib_ref] A Critical Review of the Literature on Fear of Hypoglycemia in Diabetes:..., Wang [/bib_ref]. The tremor that occurs during hypoglycemia is categorized under enhanced physiologic tremor and results from different mechanical and neuromuscular interactions. The enhanced physiologic tremor is usually more visible compared to normal tremors with a frequency that was estimated to be in the range of 5-14 Hz [bib_ref] Diagnosis and Treatment of Common Forms of Tremor, Puschmann [/bib_ref]. Smart wearables can detect tremors and identify different patterns with the help of machine learning techniques, but have not been widely assessed to predict hypoglycemia [bib_ref] Parkinson's Disease Tremor Detection in the Wild Using Wearable Accelerometers, San-Segundo [/bib_ref]. A lowcost wearable device based on an accelerometer mounted on the index finger showed that it can detect tremor in the range of 10-14 Hz, but was not validated in patients with diabetes in relation to hypoglycemia [bib_ref] A Wearable, Low-Cost Hand Tremor Sensor for Detecting Hypoglycemic Events in Diabetic..., Abbas [/bib_ref]. In a study of seven patients with type 1 and type 2 diabetes, tremors in the frequency range of 10-14 Hz were easily distinguishable under fatigue, but no evaluation was undertaken in relation to hypoglycemia [bib_ref] Features of Physiological Tremor in Diabetic Patients, Aljihmani [/bib_ref].
Acceleration can be used to provide information about human activities and wearables with an accelerometer have found increasing use in health care applications such as detecting falls among the elderly [bib_ref] A Comprehensive Analysis on Wearable Acceleration Sensors in Human Activity Recognition, Janidarmian [/bib_ref]. Given that severe hypoglycemia can markedly impair movement, information on physical activity may improve the prediction of glucose levels [bib_ref] Predicting Nocturnal Hypoglycemia with Measures of Physical Activity Intensity in Adolescent Athletes..., Jaggers [/bib_ref] [bib_ref] The Impact of Accelerometer and Heart Rate Data on Hypoglycemia Mitigation in..., Stenerson [/bib_ref]. Accelerometers have been used to plan physical activity levels in adults with type 1 diabetes [bib_ref] Lower Objectively Measured Physical Activity Is Linked with Perceived Risk of Hypoglycemia..., Keshawarz [/bib_ref]. In a study of ten adolescent athletes with type 1 diabetes, vigorous high intensity physical activity correlated with an increased risk of prolonged nocturnal hypoglycemia [bib_ref] Predicting Nocturnal Hypoglycemia with Measures of Physical Activity Intensity in Adolescent Athletes..., Jaggers [/bib_ref].
## Machine learning techniques
A typical machine learning algorithm uses data to build a predictive model that can map a set of inputs to a desired output. The core element in machine learning is collecting enough data to be used in the training and evaluation of the predictive model. Once enough data is collected, a portion of this data is typically selected to be used to train and optimize a model's parameters, and the remaining part is then used to evaluate the performance of the learned model. In glucose monitoring applications, wearable devices can be used to acquire physiological data as inputs while a CGM device is used to acquire the output or target values. Machine learning techniques have been considered for several glucose monitoring applications such as predicting current glucose levels, forecasting future values, and classifying ongoing trends.
This section presents the different machine learning techniques that have been used for blood glucose monitoring and trends detection [fig_ref] FIGURE 4 |: Machine learning algorithms utilized in glucose monitoring applications over the past five... [/fig_ref]. A summary of the surveyed studies is provided in [fig_ref] TABLE 2 |: A summary of machine learning based blood glucose monitoring contributions [/fig_ref].
## Artificial neural network
An ANN consists of interconnected layers of perceptrons that can learn patterns of data by adjusting numerical weights attached to each connection. Several popular ANN architectures have been considered for the purpose of blood glucose monitoring [bib_ref] A Deep Learning Approach to Diabetic Blood Glucose Prediction, Mhaskar [/bib_ref].
## Convolutional neural network
CNN is a type of ANN mainly used for processing and recognition of grid like data (e.g., images) and typically consists of three types of building blocks. The first two building blocks (i.e., convolution and pooling layers) perform feature extraction while the third (i.e., fully connected layer) matches the extracted features with outputs or classes. The stacking of these blocks constitutes the CNN architecture . To make a prediction, the first layer in CNN receives a vector of input values that can either be spatially related such as images, or short sequences of time series data Frontiers in Bioengineering and Biotechnology | www.frontiersin.org May 2022 | Volume 10 | Article 876672 such as multi-dimensional biometric data. These values are then passed to several layers that perform two operations; convolutions, and downsampling. In the convolution layers, specific features are extracted from nearby inputs by matching learned meaningful patterns with the sequence of data that are fed into the layer. The results of this operation are then forwarded to a next layer which chooses the patterns that were most apparent (i.e., max-pooling). This operation is repeatedly done depending on the number of layers used. Finally, the resulting patterns are fed into a fully connected ANN that produces a prediction based on the information presented on the last preceding layer. The number of patterns to match in each layer (i.e., filters) and number of stacked layers are hyper-parameters that are tuned based on the desired prediction performance and computational cost. CNN has been utilized in many applications, e.g., heart-beat classification [bib_ref] A Deep Convolutional Neural Network Model to Classify Heartbeats, Acharya [/bib_ref] , COVID-19 detection , and glaucoma detection [bib_ref] Glaucoma Detection Based on Deep Convolutional Neural Network, Chen [/bib_ref]. Models based on CNN have also been developed to predict and forecast blood glucose levels and trends [bib_ref] Automated Detection of Diabetes Using CNN and CNN-LSTM Network and Heart Rate..., Swapna [/bib_ref] [bib_ref] Deep Learning for Blood Glucose Prediction: CNN vs LSTM, El Idrissi [/bib_ref] [bib_ref] Interpretable Filter Based Convolutional Neural Network (If-cnn) for Glucose Prediction and Classification..., Kamalraj [/bib_ref]. A personalized CNN model employing a fine-tuning strategy improved the prediction horizon performance compared to standard CNN when evaluated using CGM data [bib_ref] A Personalized Blood Glucose Level Prediction Model with a Fine-Tuning Strategy: A..., Seo [/bib_ref]. Among six patients with type 1 diabetes, a dataset containing insulin dose, carbohydrate intake, and glucose levels for 8 weeks was used to train and benchmark a blood glucose forecasting model based on casual dilated CNN . Preprocessing (interpolation, extrapolation, and filtering) was performed to compensate for missing values and to clean the data (e.g., remove noise). The inputs to the neural network were the recorded CGM, insulin, carbohydrate intake, and time index mapped to 256 classes that represent a change of 1 mg/dl between each class. The results showed promise with an average root mean squared error (RMSE) of 21.72 mg/dl for a 30 min prediction horizon. A hybrid model consisting of CNN and gated recurrent unit neural networks has been proposed to reduce the error rate in predicting blood glucose levels [bib_ref] Predicting Continuous Blood Glucose Level Using Deep Learning, Shahid [/bib_ref]. Based on simulated data, the proposed model achieved an RMSE of 6.04 mg/dl for the 30 min prediction horizon and 8.12 mg/dl over 60 min.
A more recent study considered a combination of CNN with autoencoders to detect nocturnal hypoglycemia [bib_ref] Nocturnal Low Glucose Detection in Healthy Elderly from One-Lead ECG Using Convolutional..., Porumb [/bib_ref]. The study used a non-invasive wearable (i.e., ECG) and CGM to acquire data from 25 elderly subjects under controlled conditions for up to 36 h. The collected data were used to train and evaluate personalized deep learning models. The best trained model was able to distinguish low blood glucose trends with an accuracy of 90% in the test dataset. [bib_ref] Applying Deep Learning to Track Food Consumption and Human Activity for Nonintrusive..., Samir [/bib_ref] proposed another CNN based model to predict blood glucose trends by classifying food and human activity and classified hyperglycemia with an accuracy of 93.2% based on the D1NAMO dataset of diabetic patients using CGMs alongside food images and acceleration data [bib_ref] The Open D1namo Dataset: A Multi-Modal Dataset for Research on Non-invasive Type..., Dubosson [/bib_ref]. Another study also considered ECG wearable to classify blood glucose into three levels, namely, low, moderate, and high [bib_ref] Non-invasive Monitoring of Three Glucose Ranges Based on Ecg by Using Dbscan-Cnn, Li [/bib_ref]. The study used a finger pricking device (i.e., Accu-Chek) to acquire the blood glucose data from 21 healthy and prediabetic adult participants during fasting and oral glucose tests. A CNN-based model was able to classify low glucose (87.94%), moderate glucose (69.36%), and high glucose (86.39%) in the testing datasets.
## Recurrent neural network
RNN is a class of ANNs with feedback signals developed to learn sequential ordered data (i.e., time series) or time-varying patterns such as that found in speech. The prediction in RNN relies on previous information maintained internally. The hidden layers act like a memory that captures the information about a sequence. RNN models consist of interconnected layers of neurons just as in normal ANNs . The difference in RNN is their ability to take into account information from previous predictions. Specifically, each hidden layer also considers the outputs of hidden layers from preceding predictions. This allows the network to capture information from ordered sequences of data. The learning in a standard RNN structure might be limited and hindered due to the vanishing gradients problem, hence, a structure based on RNN (e.g., long-short term memory (LSTM)) is usually used to combat that [bib_ref] Long Short-Term Memory, Hochreiter [/bib_ref]. Several applications used RNN based models to recognize different sequences in human activities [bib_ref] Human Activity Recognition Using LSTM-RNN Deep Neural Network Architecture, Pienaar [/bib_ref] , emotion recognition based on videos [bib_ref] Video-based Emotion Recognition Using Cnn-Rnn and C3d Hybrid Networks, Fan [/bib_ref] , and the characterization of unwanted interactions with small robots [bib_ref] Influence of Reaction Time in the Emotional Response of a Companion Robot..., Alhaddad [/bib_ref]. RNN has also been considered for blood glucose prediction among patients with diabetes [bib_ref] Predicting Blood Glucose with an LSTM and Bi-LSTM Based Deep Neural Network, Sun [/bib_ref] [bib_ref] Using LSTMs to Learn Physiological Models of Blood Glucose Behavior, Mirshekarian [/bib_ref] [bib_ref] A Multi-Patient Data-Driven Approach to Blood Glucose Prediction, Aliberti [/bib_ref]. A recent study developed a model based on a RNN structure with the help of transfer learning to forecast future glucose levels . They believed that forecasting of future blood glucose will help to enhance the CGM and insulin pump systems by calculating the optimum insulin doses avoiding any adverse events. The study considered simulated and actual datasets containing information on meal intake, CGM readings, and insulin dosage to evaluate the developed model. The results showed a good 30 min forecasting performance of the developed model (RMSE = 18.9 mg/dl) compared to other tested algorithms. In another work, an LSTM-based algorithm was applied to 6months data on diet, glucose levels, and physical activity in 10 patients with type 2 diabetes to forecast daily glucose concentrations [bib_ref] Development of a Deep Learning Model for Dynamic Forecasting of Blood Glucose..., Faruqui [/bib_ref]. They were able to predict the next day glucose concentrations with 84.12% of the predicted values in zone A based on Clark error grid when compared to the true values of blood glucose. However, the study was limited to a small sample size and was affected by individual variations and data collection challenges. An inference system based on a smartphone to monitor blood glucose noninvasively was developed by. They collected data about insulin, drug dosage, food intake, sleep quality, and physical activities along with CGM. The authors evaluated their system on 112 subjects and the implementation of RNN achieved an accuracy of 82.14% in tracking blood glucose levels into four classes. A recent study also assessed the ability to estimate future (i.e., 30 min prediction horizon) blood glucose levels using an IoT device with CGM sensor, an application layer protocol, and prediction model on the cloud [bib_ref] Iot and Cloud Computing in Health-Care: A New Wearable Device and Cloud-Based..., Nasser [/bib_ref]. Based on a model consisting of RNNs and restricted boltzmann machines, the proposed system achieved an RMSE value of 15.59 mg/dl in data acquired from ten patients with type 1 diabetes. Another recent study proposed using Weibull Time To Event RNN (i.e., WTTE-RNNto predict future episodes of hypoglycemia and reported an RMSE of 12.56 mg/dl [bib_ref] A Wtte-Rnn Based Hypoglycemic Event Prediction, Eom [/bib_ref].
## Decision tree
A decision tree (DT) is a divide-and-conquer method that partitions data such that it becomes easy to classify. A typical FIGURE 5 | Illustrations of commonly considered artificial neural networks in glucose monitoring applications and an example of a study that considered a combined architecture. (A) Illustration of a CNN model. The first layer is the input layer which holds values of the input data that is followed by a convolution layer, which serves to extract meaningful patterns in partial regions of the input. Next, comes the pooling layer that will perform a downsampling operation that reduces the number of parameters. The extracted features are then passed to a fully connected layer to make the final prediction. (B) Illustration of RNN model. The RNN model consists of an input layer (X), hidden layers to model sequential information (from h 0 to h n where n is the number of hidden layers), and an output layer (O). The structure is connected with weights that link the input layer to the first hidden layer (W xh ), the hidden layers together (W hh ), and the last hidden layer to the output layer (W hy ). (C) An example of a study that considered the application of CNN and RNN to predict the occurrence of hypoglycemia based on ECG data (Adapted with permission from [bib_ref] Precision Medicine and Artificial Intelligence: A Pilot Study on Deep Learning for..., Porumb [/bib_ref]. The isolated heartbeats were combined into segments of 5 min each and each segment has been assigned a label (i.e., low or normal glucose level) based on the recorded CGM value.
Frontiers in Bioengineering and Biotechnology | www.frontiersin.org May 2022 | Volume 10 | Article 876672
DT model consists of nodes that split the data based on attributevalue combinations. Data are split repeatedly until a given criteria is satisfied (e.g., similarity of data in each partition). DT has been applied to extract and analyse information from large datasets (i.e., data mining) and in machine learning (e.g., classification and regression) [bib_ref] Overview of Use of Decision Tree Algorithms in Machine Learning, Navada [/bib_ref]. Classification trees are used when the target values are discrete while regression trees are used when the target values are continuous. Compared to other machine learning techniques, DT has the advantage of model interpretability. It provides an insight on the most influential data attributes related to the task at hand and helps to plan future experiments [bib_ref] An Introduction to Decision Tree Modeling, Myles [/bib_ref]. Different forms of DT (e.g., classification, regression, and forests) can be considered depending on the problem and the desired output [bib_ref] Overview of Use of Decision Tree Algorithms in Machine Learning, Navada [/bib_ref]. Decision trees were used as prediction models for risk factor interactions in diabetes and to identify subjects with impaired glucose metabolism [bib_ref] Decision Tree-Based Modelling for Identification of Potential Interactions between Type 2 Diabetes..., Ramezankhani [/bib_ref] [bib_ref] Decision Trees as a Simple-To-Use and Reliable Tool to Identify Individuals with..., Hische [/bib_ref]. Different DT models have been used to identify vital indicators in relation to blood glucose prediction [bib_ref] Predictive Analytics for Blood Glucose Concentration: an Empirical Study Using the Tree-Based..., Liu [/bib_ref]. One study developed a non-invasive system to detect blood glucose levels based on the conservation-of-energy method and physiological parameters . The study acquired data samples from 400 participants (i.e., healthy and diabetic patients) that were then used in the model development and evaluation. Using an algorithm that combines a DT and neural network, the proposed approach was able to provide a blood glucose prediction with an accuracy of 88.53% in classifying new blood glucose samples.
A recent study considered a wearable device to estimate blood glucose based on photoplethysmography (PPG) signals [bib_ref] Diabetes Care in Motion: Blood Glucose Estimation Using Wearable Devices, Tsai [/bib_ref]. The data were acquired from 9 patients with type 2 diabetes in a stable physical position (i.e., sitting). The blood glucose levels were acquired using a finger prick device (i.e., Accu-Chek). A machine learning algorithm based on decision tree (i.e., random forest) was considered to build personalized and generalized models that achieved 80% and 90%, respectively when tested with unseen data. Another study recently evaluated several ensemble machine learning models to provide a generalized blood glucose prediction . Simulated CGM data from 40 participants with type 1 diabetes were used to train and test the models. A combined model (i.e., DT and Adaboost) outperformed the other models and achieved an RMSE value of 2.204 mg/dl when evaluated against test samples. Another study developed a non-invasive platform to measure sweat glucose periodically and utilized a machine learning algorithm to generate sweat glucose readings from the discrete values [bib_ref] A Machine Learning-Based On-Demand Sweat Glucose Reporting Platform, Sankhala [/bib_ref]. A DT model was considered to provide sweat glucose readings based on the raw impedance signal, relative humidity, and temperature and the regression model achieved an RMSE value of 0.1 mg/dl in three participants.
DT based models were also utilized to predict blood glucose trends, such as hypoglycemia. One study proposed using a machine learning model based on decision trees (i.e., gradient boosting) to detect hypoglycemia using a wearable device [bib_ref] Towards Wearable-Based Hypoglycemia Detection and Warning in Diabetes, Maritsch [/bib_ref]. Using features acquired from the heart variability rate, the study developed a machine learning model based on the data acquired from one participant with type 1 diabetes. The results of the unseen samples demonstrated the possibility of detecting hypoglycemic events with 82.7% accuracy. [bib_ref] Prediction of Hypoglycemia during Aerobic Exercise in Adults with Type 1 Diabetes, Reddy [/bib_ref] investigated the possibility of predicting hypoglycemia in 43 adults with type 1 diabetes who were performing aerobic exercise. The extracted features included physical activity, heart rate, anthropometric data, energy expenditure estimate, glucose readings, and physical activity. The results of two developed models based on decision trees showed promising results in predicting hypoglycemia with an accuracy of 86.7%.
## Support vector machine
SVM is a supervised machine learning technique used in classification and regression. In classification problems, the SVM learns from the labeled training data how to best categorize data that belongs to one of two classes by finding the optimal hyperplane that separates them [bib_ref] Support Vector Machines, Hearst [/bib_ref]. The separation in SVM can be based on a linear, or non-linear combination of features depending on the complexity of the task at hand and feature dependencies. In case of a non-linear SVM, kernel functions are used to transform a problem to a linearly separable one by projecting the problem from a low-dimensional space to a high-dimensional one [bib_ref] SVM Kernel Functions for Classification, Patle [/bib_ref]. SVM is also used in linear and non-linear regression. The principle of SVM for regression is to find a flat function that satisfies a deviation criterion from the target outputs with less restriction to minimize the errors [bib_ref] A Tutorial on Support Vector Regression, Smola [/bib_ref]. In case of a non-linear regression, a similar technique to that used in classification is applied. SVM has been considered in different areas and in many different applications such as in cancer genomics [bib_ref] Applications of Support Vector Machine (Svm) Learning in Cancer Genomics, Huang [/bib_ref] , chemistry [bib_ref] Applications of Support Vector Machines in Chemistry, Ivanciuc [/bib_ref] , autism therapy [bib_ref] Detection of Challenging Behaviours of Children with Autism Using Wearable Sensors during..., Alban [/bib_ref] , and bioinformatics [bib_ref] Support Vector Machine Applications in, Byvatov [/bib_ref]. A custom-built optical sensor was used to investigate the relationship between wavelengths and glucose concentrations in aqueous solutions containing different glucose concentrations ranging from 40 to 250 mg/dl [bib_ref] Noninvasive Glucose Monitoring Using Optical Sensor and Machine Learning Techniques for Diabetes..., Shokrekhodaei [/bib_ref]. Intensity data based on the four optimal wavelengths (i.e., 485, 645, 860 and 940 nm) were considered in training a classifier and a regression model to predict either a discrete range (i.e., 21 classes) or a continuous value, respectively. A classifier based on SVM achieved the best results with an RMSE value of 12.5 mg/dl and 99.55% of the predictions in zones A and B on the Clarke error grid.
SVM has also been applied in glucose monitoring and longterm diabetes outcome prediction [bib_ref] Smartphone-based Personalized Blood Glucose Prediction, Li [/bib_ref] [bib_ref] Predicting Long-Term Type 2 Diabetes with Support Vector Machine Using Oral Glucose..., Abbas [/bib_ref]. A study surveying machine learning techniques for blood glucose prediction found that a regression model based on SVM performed best in the short term forecasting of blood glucose [bib_ref] Glycemic-aware Metrics and Oversampling Techniques for Predicting Blood Glucose Levels Using Machine..., Mayo [/bib_ref]. Based on near infrared (NIR) spectroscopy data in ten discrete artificial blood samples, SVM alone achieved an accuracy of 67.5% when evaluated with the testing dataset, but when paired with principle component analysis (PCA) it improved to 77.5% [bib_ref] Nir-spectroscopic Classification of Blood Glucose Level Using Machine Learning Approach, Habibullah [/bib_ref]. Another study tested several machine learning techniques that included SVM in detecting fasting blood glucose based on measuring the electrochemical properties of saliva [bib_ref] Non-invasive Detection of Fasting Blood Glucose Level via Electrochemical Measurement of Saliva, Malik [/bib_ref]. The fasting blood glucose was measured on venous plasma using an automatic biochemical analyzer and used as the target or true value. The electrochemical parameters of saliva (e.g., pH, conductivity, and sodium concentration) and fasting glucose were collected from 175 participants of whom half had diabetes. A SVM classifier trained with 70% of the data using the radial basis kernel function achieved the best results in distinguishing high vs. low fasting blood glucose levels in the remaining unseen testing data (i.e., accuracy and F1 score of 85%). Support vector regression has been used to predict future blood glucose levels using CGM in 12 patients and the best trained model achieved an RMSE of 12.95 mg/dl for a prediction horizon of 60 min .
A few studies have also considered using SVM to predict the occurrence of hypoglycemia among patients with type 1 diabetes. For example, one study used a non-invasive wearable device that measured air temperature, heart rate, and galvanic skin response to acquire data from one participant with type 1 diabetes for 2 months [bib_ref] Machine Learning Experiments with Noninvasive Sensors for Hypoglycemia Detection, Marling [/bib_ref]. The blood glucose data were acquired using a Dexcom CGM device and contained 34 hypoglycemic events, each lasting for 10 min or more. SVM with a linear kernel achieved the best performance, but the results were limited to one participant. Another study used a CGM device (FreeStyle Libre) in 10 participants with type 1 diabetes over 12 weeks and showed that SVM achieved a high sensitivity (78.75%) and specificity (82.15%) to predict nocturnal hypoglycemia [bib_ref] Prediction of Nocturnal Hypoglycemia in Adults with Type 1 Diabetes under Multiple..., Bertachi [/bib_ref].
## Autoregressive integrated moving average
ARIMA is a linear time series model used to predict or forecast future values based on past values. It is a function that includes differencing operators, and autoregressive and moving average terms [bib_ref] Utility of Big Data in Predicting Short-Term Blood Glucose Levels in Type..., Rodríguez-Rodríguez [/bib_ref]. ARIMA is considered as a generalized model of the autoregressive moving average (ARMA) as it incorporates a broad range of non-stationary series. ARIMA has been used to predict traffic noise pollution [bib_ref] Applications of Autoregressive Integrated Moving Average (Arima) Approach in Time-Series Prediction of..., Garg [/bib_ref] , web applications workload [bib_ref] Tasm: Technocrat Arima and Svr Model for Workload Prediction of Web Applications..., Singh [/bib_ref] , and traffic flow [bib_ref] Short-term Traffic Flow Prediction Using Seasonal Arima Model with Limited Input Data, Kumar [/bib_ref]. Models based on ARIMA have also been considered in the prediction of blood glucose levels [bib_ref] Utility of Big Data in Predicting Short-Term Blood Glucose Levels in Type..., Rodríguez-Rodríguez [/bib_ref]. A study used ARIMA to assist in predicting future blood glucose trend changes for hypoglycemia and hyperglycemia . Based on a combination of the ARIMA model and an adaptive algorthim, the study developed a prediction framework using continuous glucose monitoring (CGM) data from 100 patients with type 1 and type 2 diabetes. Their model provided early alarms with a 9.4% false rate at a sensitivity of 100%. Another recent study utilized CGM data to compare the 30 min prediction horizon performance of thirty linear and nonlinear algorithms [bib_ref] Forecasting of Glucose Levels and Hypoglycemic Events: Head-To-Head Comparison of Linear and..., Prendin [/bib_ref]. Individualized ARIMA was the best linear algorithm in terms of accuracy with an RMSE of 22.15 mg/dl and also in hypoglycemia detection with 64% precision and 82% recall.
# Discussion
The past five years has witnessed considerable advances in the development of sensors that measure different modalities which correlate with blood glucose. Glucose levels in tears, saliva, and sweat are related to the blood glucose levels and the advances in non-invasive wearable technology have not only allowed an estimation of blood glucose levels, but also the prediction of hypoglycemia. The developed wearables have targeted the heart, skin, eyes, and mouth using various technologies such as electromagnetic and bioimpedance. Some of these solutions have been augmented with machine learning techniques which have yielded promising outcomes, especially in hypoglycemia prediction. However, there remain considerable challenges before these devices can achieve FDA approval.
## Participants and testing conditions
A major limitation in the studies to date is the number of participants with the majority of studies being limited to a few participants. This limitation clearly affects their clinical use and hinders the generalizability of the findings. Furthermore, some studies did not even undertake trials in patients with diabetes, which limits the applicability of the proposed solutions to the target end-users. Many of the studies reported only the initial or exploratory results of the developed sensors or wearable prototypes. The majority of studies were conducted under controlled settings and were limited to subjects of a younger age. Future studies should recruit larger numbers of participants with a wider age range and especially patients with type 1 or type 2 diabetes.
## Bodily fluids glucose sensors
The glucose levels in these bodily fluids were reported to correlate with blood glucose concentrations, but the glucose concentrations in these bodily fluids are low compared to that found in the blood. This represents a major challenge for the development of sensors and technologies that rely on bodily fluids to estimate blood glucose levels and trends. Hence, extra considerations such as the enhancement of sensitivity and interference elimination become crucial in the development of such sensors [bib_ref] A Contact Lens with Embedded Sensor for Monitoring Tear Glucose Level, Yao [/bib_ref]. Another challenge is the way a body fluid sample is collected (e.g., stimulated or unstimulated) as it was found to influence the content of the acquired sample. For example, diagnostic biomarkers were found to be higher in unstimulated saliva compared to stimulated saliva [bib_ref] Current Developments in Salivary Diagnostics, Miller [/bib_ref]. A study has also shown that unstimulated saliva is highly accurate for predicting blood glucose [bib_ref] Unstimulated Parotid Saliva Is a Better Method for Blood Glucose Prediction, Cui [/bib_ref]. Hence, the collection techniques require standardization to avoid influencing the contents of the collected samples [bib_ref] Salivary Nicotine and Cotinine Concentrations in Unstimulated and Stimulated Saliva, Robson [/bib_ref] [bib_ref] Evaluation of Commonly Used Tear Sampling Methods and Their Relevance in Subsequent..., Rentka [/bib_ref]. Contamination in the collected samples is another challenge Frontiers in Bioengineering and Biotechnology | www.frontiersin.org that must be addressed as contaminated samples will generate false reading on tear glucose levels [bib_ref] Association between Tear and Blood Glucose Concentrations: Random Intercept Model Adjusted with..., Aihara [/bib_ref].
## Noise susceptibility and physiological effects
Some of the physiological sensors were found to be influenced by daily circadian rhythms in heart rate and sweating [bib_ref] Evaluation of Hypoglycaemia with Non-invasive Sensors in People with Type 1 Diabetes..., Elvebakk [/bib_ref]. Error can also arise due to nonlinear dynamics of physiological signs being measured, making sensors more susceptible to error and high noise levels [bib_ref] First Experiences with a Wearable Multisensor Device in a Noninvasive Continuous Glucose..., Zanon [/bib_ref] [bib_ref] Prediction of Adverse Glycemic Events from Continuous Glucose Monitoring Signal, Gadaleta [/bib_ref]. Some studies have reported the adverse impact of environmental conditions (e.g., pH, humidity, and temperature) [bib_ref] Electrochemical Non-enzymatic Glucose Sensors, Park [/bib_ref] [bib_ref] Wearable Contact Lens Biosensors for Continuous Glucose Monitoring Using Smartphones, Elsherif [/bib_ref]. The time delay of a sensor or wearable device must be minimized to capture the rapid changes in blood glucose [bib_ref] Improving Blood Glucose Level Predictability Using Machine Learning, Marcus [/bib_ref]. For patchlike sensors, the multidirectional stretchability of the sensor is an issue that needs careful considerations to ensure accuracy and stability under multi stretching cycles [bib_ref] Fully Stretchable Capillary Microfluidics-Integrated Nanoporous Gold Electrochemical Sensor for Wearable Continuous Glucose..., Bae [/bib_ref]. Furthermore, reproducibility of sensor characteristics can be influenced by the fabrication processes. Finally, the majority of devices remain in the development stage and require extra-large devices to be connected to read out the signal and provide the filtration needed [bib_ref] Noninvasive, Wearable, and Tunable Electromagnetic Multisensing System for Continuous Glucose Monitoring, Mimicking..., Hanna [/bib_ref]. Electronics and mechanical miniaturization in wearable sensors represents a major limitation that requires further technological development and optimization [bib_ref] Wearable Contact Lens Biosensors for Continuous Glucose Monitoring Using Smartphones, Elsherif [/bib_ref].
## Data acquisition and model generalizability
A key challenge in machine learning is acquiring enough comprehensive data to train and test a model that can then be generalized to a wider population. Many studies were limited in terms of the number of participants, duration of the study, and the incidence of clinically relevant severe hypoglycemia. Additionally, most studies established their methods and techniques in relation to CGM data that may be limited in certain scenarios. There is a need to include a larger population of patients, especially with diabetes, to account for the inter-individual differences and to establish better validation of the proposed solutions. To capture meaningful data, the duration of trials for data acquisition and the incidence of hypoglycemia need to be sufficient to avoid unbalanced or skewed data [bib_ref] Machine Learning Experiments with Noninvasive Sensors for Hypoglycemia Detection, Marling [/bib_ref]. In the case of imbalanced data, oversampling techniques can address this and improve the accuracy [bib_ref] Glycemic-aware Metrics and Oversampling Techniques for Predicting Blood Glucose Levels Using Machine..., Mayo [/bib_ref]. Acquiring data from the same participant for longer periods allows the machine learning algorithm to combat intraindividual differences and increases overall prediction performance [bib_ref] Predicting Hypoglycemia in Diabetic Patients Using Time-Sensitive Artificial Neural Networks, Eljil [/bib_ref]. The reliance on data annotated by the participants is an issue in acquiring accurate information as it depends solely on their commitment [bib_ref] Prediction of Nocturnal Hypoglycemia in Adults with Type 1 Diabetes under Multiple..., Bertachi [/bib_ref]. The experimental conditions should not be controlled to allow data acquisition of more realistic daily life settings. Future studies should also include a wider set of physiological parameters to investigate their individual or collective effect on estimating blood glucose trends using machine learning techniques. Additionally, data assimilation techniques could be considered in conjunction with machine learning techniques using data acquired from several sources and wearables [bib_ref] Personalized Glucose Forecasting for Type 2 Diabetes Using Data Assimilation, Albers [/bib_ref] [bib_ref] Mechanistic Machine Learning: How Data Assimilation Leverages Physiologic Knowledge Using Bayesian Inference..., Albers [/bib_ref].
## Model interpretability
Although machine learning algorithms have great predictive potential, the majority of these algorithms are black box models and lack the means to explain their predictions. The interpretability requirement of a machine learning model in health care is crucial [bib_ref] Interpretability in Healthcare: A Comparative Study of Local Machine Learning Interpretability Techniques, Elshawi [/bib_ref]. Clinicians need to make an informed decision based on a prediction but need to provide a proper explanation. Whilst wearable sensors are capable of estimating blood glucose trends based on physiological changes, such predictions without an underlying explanation may confuse the patient and render the wearable unreliable [bib_ref] Towards Wearable-Based Hypoglycemia Detection and Warning in Diabetes, Maritsch [/bib_ref]. Future studies must evaluate and incorporate interpretability techniques into their proposed solutions while making sure that the information acquired by the sensors are presented in a user-friendly interface [bib_ref] Interpretability in Healthcare: A Comparative Study of Local Machine Learning Interpretability Techniques, Elshawi [/bib_ref]. A support decision system along with the prediction was suggested to allow patients to provide feedback to evaluate the performance of a machine learning algorithm in real-life scenarios [bib_ref] Prediction of Nocturnal Hypoglycemia in Adults with Type 1 Diabetes under Multiple..., Bertachi [/bib_ref]. Another consideration of an algorithm is the computational cost and energy. A wearable device has limited resources to perform complex operations and to operate for a long duration. Hence, feature engineering and data reduction techniques are needed to reduce the computational cost and improve energy efficiency [bib_ref] Exploring the Computational Cost of Machine Learning at the Edge for Human-Centric..., Gómez-Carmona [/bib_ref] [bib_ref] Efficient Machine Learning for Big Data: A Review, Al-Jarrah [/bib_ref].
## Conclusions and future directions
This review highlights the considerable progress made over the past five years in the area of non-invasive blood glucose monitoring using wearable technologies and sensors alongside machine learning algorithms. The devices have varied modalities and adopted technologies with novel approaches utilizing machine learning techniques to provide meaningful interpretations of multiple physiological parameters. However, there remain considerable limitations and challenges that hinder FDA approval and more widespread adoption of such technologies in patients. We therefore recommend future studies to focus on the following areas:
1. The recruitment of a larger number of participants especially patients with diabetes to validate the proposed techniques for use in the clinical arena.
2. Validating glucose sensors by adequate collection without contamination of bodily fluids. 3. Miniaturization of electronics and sensors for practical deployment. 4. Consistent evaluation of algorithms in personalized vs.
generalized scheme, where a model is trained either on a target individual or a group of subjects. 5. The utilization of a wider set of physiological parameters for machine learning and data assimilation techniques while establishing effectiveness of each sensor's contribution. 6. Investigating means to reduce computational cost and energy in wearable devices. 7. The development of interpretable machine learning models.
# Author contributions
AYA wrote the first draft. HA, HG, AA-A, and KKS produced the figures and revised the manuscript. J-JC and RAM oversaw the work and revised the manuscript. All authors contributed to the work and approved the submitted version.
# Funding
The work is supported by an NPRP grant from the Qatar National Research Fund under the grant No. NPRP 11S-0110-180247. The statements made herein are solely the responsibility of the authors.
[fig] FIGURE 4 |: Machine learning algorithms utilized in glucose monitoring applications over the past five years.Frontiers in Bioengineering and Biotechnology | www.frontiersin.org May 2022 | Volume 10 | Article 876672 [/fig]
[table] TABLE 1 |: A summary of the advances in non-invasive sensors and wearable technologies for blood glucose monitoring. [/table]
[table] TABLE 2 |: A summary of machine learning based blood glucose monitoring contributions. [/table]
|
Improved patient-reported outcomes in patients with psoriatic arthritis treated with risankizumab: analysis of the Phase 3 trial KEEPsAKE 2
# Introduction
Psoriatic arthritis (PsA) is a heterogenous, chronic inflammatory disease that is frequently characterised by different disease domains, including psoriasis, axial and peripheral arthritis, enthesitis and dactylitis. Approximately 30% of patients with psoriasis will develop PsA. [bib_ref] Psoriatic arthritis, Ritchlin [/bib_ref] [bib_ref] The molecular pathophysiology of psoriatic Arthritis-The complex interplay between genetic predisposition, epigenetics..., Carvalho [/bib_ref] [bib_ref] Clinical features of psoriatic arthritis, Kishimoto [/bib_ref] Over time, PsA leads to significantly impaired physical functioning, resulting in reduced health-related quality of life (HRQoL) in these patients. Due to the significant physical limitations of PsA, the ability to perform daily living, work and social activities is often impaired. Indeed, overall disease activity is negatively associated with work presenteeism and productivity, and positively associated with absenteeism. Furthermore, duration of disease negatively impacts patients' ability to remain employed and, as such, 36% of working-age patients with PsA were reported to be unemployed. [bib_ref] Psoriatic arthritis, Ritchlin [/bib_ref] WHAT IS ALREADY KNOWN ON THIS TOPIC ⇒ Psoriatic arthritis (PsA) is a chronic inflammatory disease that frequently leads to impaired physical functioning, resulting in reduced health-related quality of life.
## What this study adds
⇒ This analysis determined the impact of treatment with risankizumab versus placebo on key patientreported outcomes in patients with PsA. ⇒ Risankizumab-treated versus placebo-treated patients reported greater improvements in pain, fatigue and health-related quality of life. ⇒ Risankizumab-treated versus placebo-treated patients also reported greater improvements in the ability to perform work.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE AND/OR POLICY ⇒ These results complement the achievement of primary clinical endpoints observed with treatment, with improvements perceived by patients that should be considered in treatment decision-making.
## Rmd open rmd open rmd open
The impact of PsA on physical limitations is also comparable to the impact of PsA on mental health and the prevalence of depression and anxiety is substantial in these patients. The OMERACT PsA Working Group recommends that clinical studies consider the 'life impact' that an emerging therapy may have, outside of the pathophysiological benefits. More specifically, they recommend obtaining patient-reported information regarding pain, physical functioning, perceived disease activity, fatigue and overall QoL. Patient-reported outcomes (PROs) are important tools for understanding a therapy's efficacy in the eyes of the patient. This is critical because perceived efficacy may significantly impact patients' willingness to receive treatment. Risankizumab (RZB) is an interleukin (IL-23) inhibitor currently under development for the treatment of adults diagnosed with active PsA. Phase 2 and Phase 3 studies in patients with PsA have shown that RZB treatment significantly improves patient pain, physical functioning (determined by Health Assessment Questionnaire-Disability index) and disease severity, as seen by reduced Psoriasis Area and Severity Index and American College of Rheumatology scores. [bib_ref] OP0307 Efficacy and safety of risankizumab, a selective il-23p19 inhibitor, in patients..., Mease [/bib_ref] [bib_ref] Efficacy and safety of risankizumab for active psoriatic arthritis: 24-week results from..., Östör [/bib_ref] This analysis aimed to characterise the impact of 24-week treatment with RZB versus placebo (PBO) on PROs in patients with active PsA who are enrolled in KEEPsAKE 2 to understand treatment benefits of RZB from the patient perspective.
## Patients and methods study design
KEEPsAKE 2 (NCT03671148) is a Phase 3, randomised, PBO-controlled, double-blind, multicentre study comparing the effects of RZB with PBO after 24 weeks of treatment in patients with active PsA who have an inadequate response or intolerance to one or two biologic disease-modifying antirheumatic drugs (DMARDs) (Bio-IR) and/or to ≥1 conventional synthetic DMARD (csDMARD-IR). 14 Patients were randomised 1:1 to receive RZB 150 mg or PBO by subcutaneous injection at weeks 0, 4 and 16.
Full details on study design and patient attrition were published previously. [bib_ref] Efficacy and safety of risankizumab for active psoriatic arthritis: 24-week results from..., Östör [/bib_ref]
## Patients
Patients eligible for the study were ≥18 years old with a clinical diagnosis of PsA and with symptom onset at least 6 months prior to study and fulfilment of the Classification Criteria for PsA (CASPAR) at screening. Patients had active disease as defined by ≥5 tender joints (based on 68 joint counts; TJC68) and ≥5 swollen joints (based on 66 joint counts; SJC66), and active plaque psoriasis with at least one psoriatic plaque ≥2 cm in diameter or presence of nail psoriasis at baseline. Patients were permitted to be on ≤2 of the following background csDMARDs: methotrexate (MTX), sulfasalazine, leflunomide, apremilast, hydroxychloroquine, bucillamine, iguratimod or ciclosporin A; csDMARD use had to remain stable. To be considered Bio-IR, patients must have demonstrated an inadequate response, defined as a lack of efficacy after ≥12 weeks therapy, or intolerance to one or two bDMARDs intended to treat PsA. To be considered csDMARD-IR, patients had to demonstrate an inadequate response or intolerance to a previous or current treatment with ≥1 of the following csDMARDs: MTX, sulfasalazine, leflunomide, apremilast, bucillamine, iguratimod or ciclosporin A. Patients in the csDMARD-IR only population must not have had any prior exposure to bDMARDs (ie, bDMARDnaïve) used to treat PsA to be eligible for inclusion; patients who were csDMARD-IR and Bio-IR entered the study in the Bio-IR population. Any patient with an active chronic infection, with prior exposure to IL-12 or IL-12/23 inhibitors, and who were pregnant, breastfeeding or planning to become pregnant, were excluded. Concomitant use of Janus kinase (JAK) inhibitors, biologic and biosimilar therapies, opiates (except low-potency opiates: tramadol, codeine or hydrocodone alone or in combination with acetaminophen), live vaccines and non-oral corticosteroids (except low-potency topicals) was not permitted during the study. A maximum oral dose of ≤10 mg/day of prednisone equivalent was permitted.
## Pro measures
This study assessed several PROs across several different domains of HRQoL in RZB-treated versus PBO-treated patients with PsA. PROs included the 36-item Short-Form Health Survey (SF-36), the Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-Fatigue), the EQ-5D-5 Level (EQ-5D-5L), Patient's Assessment of Pain by visual analogue scale (VAS), Patient's Global Assessment of disease activity (PtGA) and the PsA-specific Work Productivity and Activity Impairment (WPAI-PsA) questionnaire.
The SF-36 is a 36-item survey of patient health consisting of eight domains: physical functioning (PF), bodily pain (BP), role physical (RP), role emotional (RE), mental health (MH), social functioning (SF), vitality (VT) and general health perceptions (GH). There are also two composite scores included in the SF-36: the physical component score (PCS) and mental component score (MCS). All domains and component scores were determined on a 0-100 scale, with higher scores indicating a favourable health state. The FACIT-Fatigue is a 13-item questionnaire that evaluates fatigue and its impact on daily living activities and physical functioning. Scores are determined by a 5-point Likert scale, with a total score range of 0-52; higher scores denote more severe fatigue. [bib_ref] Functional assessment of chronic illness therapy-fatigue scale is valid in patients with..., Chandran [/bib_ref] The EQ-5D-5L is a general HRQoL questionnaire consisting of five dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/ depression. Patients rate these dimensions from 1 to 5 based on severity, and scores are indexed to ≤1 (the value of full health), with higher scores indicating higher health utility. [bib_ref] Development and preliminary testing of the new five-level version of EQ-5D (EQ-5D-5L), Herdman [/bib_ref] The EQ-5D-5L was also assessed by a horizontal 100 mm VAS ranging from 0 (worst health Psoriatic arthritis Psoriatic arthritis Psoriatic arthritis patient can imagine) to 100 (best health patient can imagine) in which patients report on their health status for that day. Patient's Assessment of Pain is measured on a VAS ranging from 0 to 100 mm. For the assessment, patients report their pain levels over the last 24 hours, with higher scores indicating more severe pain. [bib_ref] Measures of adult pain: visual analog scale for pain (vas pain), numeric..., Hawker [/bib_ref] PtGA was used to assess the patient's perception of overall funtionality in the context of disease activity using a VAS ranging from 0 to 100 mm, with higher scores denoting more severe impairment. The WPAI-PsA measure is
## Rmd open rmd open rmd open
a well-validated instrument used to assess impairment in work and daily activities. [bib_ref] The validity and reproducibility of a work productivity and activity impairment instrument, Reilly [/bib_ref] There are four domains to the WPAI-PsA: overall work impairment, activity impairment, absenteeism and presenteeism. Patients assess the impact of PsA on their abilities to perform work or regular activities over the past 7 days, and scores range from 0% to 100% impairment. Pain and PtGA were assessed at day 1 and weeks 4, 8, 12, 16 and 24. The SF-36, EQ-5D-5L, FACIT-Fatigue and WPAI were assessed at day 1 and weeks 12 and 24.
# Statistical analysis
Demographics and baseline characteristics endpoints were summarised using descriptive statistics. For binary endpoints, numbers and percentages were summarised, while mean and SD were reported for continuous endpoints. Efficacy analyses, including PROs, were assessed in the full analysis set population that was defined as all randomised patients who received ≥1 dose of study drug. For all PRO endpoints investigated in this manuscript, least square mean change from baseline and 95% CIs at week 24 were compared between RZB and PBO treatment groups, using mixed-effects repeated measures regression modelling controlled for stratifying variables of current use of csDMARD (0 vs ≥1), number of prior bDMARDs (0 vs ≥1) and extent of psoriasis (≥3% body surface area (BSA) or <3% BSA) at baseline. The mixedeffects repeated measures model used longitudinal data for up to 24 weeks and excluded data from patients after initiation of rescue therapy or concomitant treatments for PsA that could meaningfully impact the assessment. The SF-36 PCS and FACIT-Fatigue were ranked secondary endpoints and were controlled for multiplicity adjustment to ensure a strong control of family-wise type I error rate at significance level alpha=0.05 (two-sided). Analysis of all other measures was not multiplicity controlled, and the nominal p values are reported. Analyses were repeated by the following subgroups of the full analysis set population: csDMARD-IR only (bDMARD-naïve) patients versus bDMARD-IR, and body mass index (BMI) scores (BMI kg/m 2 : <25, ≥25 to <30, ≥30) at baseline.
## Ethics
The protocol, informed consent form(s), recruitment material, and all patient materials were approved by an independent ethics committee or institutional review board at all study sites (see online supplemental table S1 for full list). All patients provided written informed consent prior to enrolment. The clinical study was conducted in accordance with the current Declaration of Helsinki and is consistent with the International Conference on Harmonisation Good Clinical Practice and Good Epidemiology Practices, and all applicable local regulatory requirements. All patient data were anonymised and complied with patient confidentiality requirements.
# Results
## Baseline demographic and clinical characteristics
A total of 444 patients with PsA were randomised to receive RZB (n=224) or PBO (n=220), and baseline characteristics were balanced between groups (table 1). One patient was randomised but did not receive the study drug and thus was excluded from the full analysis set population. 14 As previously reported, 215 (96.0%) and 199 (90.5%) patients who received RZB or PBO, respectively, completed the week 24 study visit. [bib_ref] Efficacy and safety of risankizumab for active psoriatic arthritis: 24-week results from..., Östör [/bib_ref] The average patient age was 52.9±12.6 years, with a mean disease duration of 8.2 years, and 55% of patients
## Rmd open
RMD Open RMD Open
Fatigue, pain, general health status (EQ-5D-5L) and disease activity state Patients receiving RZB demonstrated notable improvements, as compared with PBO, in all PROs at week 24 [fig_ref] Table 3: Least squares mean change from baseline [/fig_ref]. Specifically, patients receiving RZB versus PBO reported a greater mean change from baseline in FACIT-Fatigue scores. This resulted in a significant difference (95% CI) between groups (2.2 (0.6 to 3.9), p=0.009). Notable differences between groups were also observed in EQ-5D-5L index and VAS scores. There was a 0.08 difference in EQ-5D-5L index scores between RZB-treated and PBO-treated groups (95% CI: 0.08 (0.04 to 0.11); nominal p<0.001). Similarly, there was a 5.9point (95% CI: 1.9 to 9.8) difference between RZB and PBO groups in EQ-5D-5L VAS scores (nominal p<0.01).
Patients receiving RZB reported greater improvement in pain (VAS) than those receiving PBO, resulting in a notable difference (95% CI) between groups (-8.1 (-12.8 to -3.5), nominal p<0.001). Similarly, greater improvements were also seen in PtGA (VAS) in RZBtreated versus PBO-treated patients, resulting in a -8.8point difference between groups (95% CI: -13.5 to -4.2; nominal p<0.001). shows the improvements in SF-36 domains and component scores in RZB-treated versus PBO-treated patients. RZB-treated patients had greater improvements, as determined by the differences between least squares mean changes from baseline, in seven of the eight SF-36 domains (PF, RP, BP, GH, VT, SF and MH; nominal p<0.05) and the multiplicity-adjusted SF-36 PCS score (p<0.001).
## Sf-36 domains and component scores
Visual depiction, via spydergram, of the differences between RZB-treated versus PBO-treated patients at week 24 in SF-36 domains is presented in online supplemental figure 1. The figure shows that all patients have substantial negative impacts on their HRQoL at baseline due to the nature of the disease and that RZB-treated patients report consistently higher scores at week 24 in comparison to PBO-treated patients.
## Work and activity impairment
After 24 weeks of treatment, patients receiving RZB reported greater improvements in three of the four WPAI domain scores than patients receiving PBO (figure 2). There was a -10% difference (95% CI: -17.5% to -2.4%, nominal p≤0.01) in overall work impairment between treatment groups. Similar differences (95% CI) between RZB and PBO groups were also shown for activity impairment (-7.9% (-12.7% to -3.2%), nominal p<0.001) and presenteeism (-9.0% (-15.4% to -2.6%), nominal p<0.01). There was no notable difference in change from baseline in absenteeism between groups (-4.6% (-11.0 to 1.8)).
## Mean change from baseline at 24 weeks by subgroup
Both csDMARD-IR (bDMARD-naïve) patients and bDMARD-IR treated with RZB versus PBO saw notable improvements in multiple PROs ; results were similar to those observed in the primary analysis. Differences between RZB-treated and PBO-treated patients at week 24 were numerically higher among bDMARD-IR patients for several PROs, including fatigue, EQ-5D-5L VAS, pain, PtGA, SF-36 PCS, overall work and activity Psoriatic arthritis Psoriatic arthritis Psoriatic arthritis impairment and presenteeism compared with patients with csDMARD-IR.
Additionally, RZB-treated versus PBO-treated patients had a trend towards greater improvements from baseline at week 24 regardless of BMI score at baseline [fig_ref] Table 5: Least squares mean change from baseline at 24 weeks among patients with... [/fig_ref].
# Discussion
It is recommended to assess the impact of PsA and treatment benefits from the patient's perspective regarding pain, physical functioning, perceived disease activity, fatigue and overall QoL. Results from KEEPsAKE 2, a study in patients with PsA who have inadequate response or intolerance to bDMARDs and/or csDMARDs, demonstrated that RZB greatly improved PROs at week 24 as compared with PBO across several HRQoL domains as recognised by the GRAPPA-OMERACT working group to measure the impact of disease on patients with PsA. [bib_ref] Defining outcome measures for psoriatic arthritis: a report from the GRAPPA-OMERACT Working..., Ogdie [/bib_ref] At baseline, patients had substantially reduced HRQoL as assessed by SF-36 domain scores. Patients who received RZB experienced notable improvements in fatigue and general health status as assessed by both EQ-5D-5L index and VAS scores. Patient-reported pain and overall perception of disease activity were also reduced at week 24 in RZB-treated versus PBO-treated patients. Patients treated with RZB reported notable improvements in seven of the eight SF-36 domains and SF-36 PCS scores. Improvements in WPAI overall work impairment, activity impairment and presenteeism were also reported. A majority of patients were on background csDMARD; however, there was still a major effect on PROs with the addition of RZB in comparison to PBO. Similar results were observed when patients were stratified by prior bDMARD exposure status and BMI scores at baseline. Differences between RZB-treated versus PBO-treated patients were higher among bDMARD-IR versus csDMARD-IR (bDMARDnaïve) patients, suggesting a further benefit of RZB treatment to those patients.
It is important to contextualise the findings of this study against other drug classes indicated for the treatment of PsA. A wide variety of therapies are approved Least squares mean change from baseline among patients who were csDMARD-IR (bDMARD-naïve) or bDMARD-IR at 24 weeks
## Rmd open rmd open rmd open
to treat PsA, such as csDMARDs (eg, MTX), bDMARDs (eg, tumour necrosis factor (TNF), IL-17, IL-12/23 and IL-23 inhibitors) and targeted synthetic DMARDs (eg, apremilast and JAK inhibitors). [bib_ref] Current treatments and recommendations, Vivekanantham [/bib_ref] Understanding the benefits of these treatments from a patient's perspective is critical, as perceived efficacy may significantly impact a patient's willingness to receive treatment. [bib_ref] Medication adherence in patients with rheumatoid arthritis: a critical appraisal of the..., Van Den Bemt [/bib_ref] [bib_ref] Comparison of healthrelated quality of life in rheumatoid arthritis, psoriatic arthritis and..., Strand [/bib_ref] Comparably, scores on SF-36 PCS and seven of eight domain scores were also greater in the RZB treatment group than the PBO group in this study after 24 weeks of treatment. In a study with the IL-17A inhibitor ixekizumab, patients reported significant improvement relative to PBO in the joint pain VAS, PtGA and EQ-5D-5L through week [bib_ref] Current treatments and recommendations, Vivekanantham [/bib_ref] [bib_ref] Ixekizumab improves patient-reported outcomes in patients with active psoriatic arthritis and inadequate..., Kavanaugh [/bib_ref] comparable to the results of this study for RZB. Achieving and maintaining improvements in work productivity is noteworthy because approximately onethird of respondents in a multinational survey 28 reported that they missed work because of PsA, and approximately one-third reported that their PsA impacted their ability to work full-time. In the current study, patients treated with RZB versus PBO reported greater reductions in impairments in work productivity, absenteeism and presenteeism. Patients treated with RZB versus PBO also reported greater reductions in daily activity impairment.
This study has strengths and limitations. Strengths included the blinded and randomised study design that allows for less biased reporting from each patient and mitigates biases due to differences between treatment groups. This study used multiple PROs to reflect and capture the multiple burdens experienced by patients with PsA, although the skin domain was not specifically assessed. A skin PRO such as the Dermatology Quality of Life Index was not included and should be considered Psoriatic arthritis Psoriatic arthritis Psoriatic arthritis for inclusion in future clinical trials assessing the impact of RZB on patients with PsA. At the time of the study development, GRAPPA-OMERACT had only provisionally included the Psoriatic Arthritis Impact of Disease questionnaire and therefore this PRO was also not assessed in the current trial despite wide clinical usage. Additional limitations included the collection of PRO data at fixed visits, sometimes weeks apart and with no day-to-day data. Prolonged recall of symptoms may introduce recall bias, which could affect patient perceptions of efficacy. In addition, most PROs presented here were not multiplicity controlled; thus, some significance values are nominal. Completing several PROs may result in response fatigue, thereby introducing bias. Results were limited to 24 weeks and should be confirmed long-term.
# Conclusion
In patients with active PsA who were Bio-IR and/or csDMARD-IR, RZB treatment resulted in greater improvements in physical functioning, fatigue, pain, HRQoL and ability to perform work compared with PBO. These results complement the reductions in disease severity observed with treatment, with improvements perceived by patients that need to be taken into consideration in treatment decisions.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
[fig] Figure 2: Least squares mean percent change from baseline in WPAI Scores at 24 weeks. *Reported only for patients who were employed. †Nominal p<0.001 for RZB versus PBO. ‡Nominal p≤0.01 for RZB versus PBO. LSM, least squares mean; PBO, placebo; RZB, risankizumab; WPAI, Work Productivity and Activity Impairment. [/fig]
[fig] *: Reported only for patients who were employed. bDMARD, biologic disease-modifying antirheumatic drug; BMI, body mass index; EQ-5D-5L, EuroQoL-5 Dimension-5 Level; FACIT-Fatigue, Functional Assessment of Chronic Illness Therapy-Fatigue; IR, inadequate responder; MCS, mental component summary; PCS, physical component summary; PRO, patient-reported outcome; PtGA, patient's global assessment of disease activity; SF-36, 36-Item Short-Form Health Survey; VAS, visual analogue scale. [/fig]
[table] Table 1: Baseline demographics and clinical characteristics [/table]
[table] Table 2: Baseline PRO Scores [/table]
[table] Table 3: Least squares mean change from baseline [/table]
[table] Table 5: Least squares mean change from baseline at 24 weeks among patients with varying BMI levels at baseline PRO, LS mean change (95% CI) [/table]
|
Monoterpenoids from Acanthopanax sessiliflorus Fruits
Three new acyclic monoterpenoids named (2E)-3,7-dimethylocta-2,6-dienoate-6-O--L-arabinopyranosyl-(16)--D-glucopyranoside (1), (3Z,6E)-3,7-dimethyl-3,6octadiene-1,2,8-triol(2)and (6E)-7-methyl-3-methylene-6-octene-1,2,8-triol (3) were isolated from Acanthopanax sessiliflorus fruits, along with three known monoterpenoid compounds. The structures of the new compounds were determined by means of extensive spectroscopic analysis (1D, 2D NMR and HRESIMS) and chemical methods.
arabinopyranosyl-(16)--D-glucopyranoside (1), (3Z,6E)-3,7-dimethyl-3,6-octadiene-1,2,8-triol [bib_ref] Antitumor and immunostimulating activities of Acanthopanax sessiliflorus fruits, Lee [/bib_ref] and (6E)-7-methyl-3-methylene-6-octene-1,2,8-triol [bib_ref] In vivo antithrombotic and antiplatelet activities of a quantified Acanthopanax sessiliflorus fruit..., Song [/bib_ref] , were isolated and identified along with the known compounds kenposide A (4) [bib_ref] Bioactive constituents of Chinese natural medicines. IV. Rhodiolae radix. (2).: On the..., Yoshikawa [/bib_ref] , sacranoside B (5) [bib_ref] Monoterpenoids from the fruit of Gardenia jasminoides, Yu [/bib_ref] and 1-O-[(S)-oleuropeyl]--Dglucopyranose [bib_ref] Pancreatic lipase-inhibiting triterpenoid saponins from fruits of Acanthopanax senticosus, Li [/bib_ref] [bib_ref] Non-volatile components of the essential oil secretory cavities of Eucalyptus leaves: Discovery..., Goodger [/bib_ref] , on the basis of spectral analysis, including MS, 1 H-NMR, [bib_ref] Bioactive constituents from Chinese natural medicines. XXVIII. Chemical structures of acyclic alcohol..., Nakamura [/bib_ref] C-NMR, DEPT, HMBC, HMQC and NOESY experiments [fig_ref] Figure 1: Structures [/fig_ref]. In the present report, we describe the structural elucidation of compounds 1~3.
# Results and discussion
Compound 1 was obtained as a colorless amorphous powder. . Acid hydrolysis of 1 liberated D-glucose and L-arabinose, which were identified by GC analysis using a hydrogen flame detector after treatment with L-cysteine methyl ester hydrochloride in pyridine [bib_ref] Triterpenes from Acanthopanax sessiliflorus fruits and their antiplatelet aggregation activities, Yang [/bib_ref] [bib_ref] Two new cyanogenic glucosides from the leaves of Hydrangea macrophylla, Yang [/bib_ref]. [bib_ref] Bioactive 3,4-seco-triterpenoids from the fruits of Acanthopanax sessiliflorus, Lee [/bib_ref] H-NMR and 13 C-NMR spectra of 1 and 5.10 (H-6), suggesting a -CH 2 CH 2 CHfragment. Two anomeric proton signals were assigned to two anomeric carbon signals at 95.2 (C-1') and 104.8 (C-1'') respectively, in the HMQC experiment. By comparing coupling constants and the chemical shifts of the sugar signals with those reported [bib_ref] Three new compounds from the leaves of Acanthopanax senticosus harms, Wang [/bib_ref] [bib_ref] Bioactive constituents from Chinese natural medicines. XXVIII. Chemical structures of acyclic alcohol..., Nakamura [/bib_ref] , the two sugars were deduced to be of -configuration for glucose and for arabinose. In the HMBC spectrum, long-range correlations between the methyl proton signals at 2.20 (m, 2H) 42.0 C-2, C-3, C-5, C-6, C-10 H-2 5
2.21 (m, 2H) 27.1 C-3, C-4, C-6, C-7 H-9 6
5.10 (dd, 1H, J = 6.6, 6.9) 124.1 C-4, C-5, C-8, C-9 H-8 7
133.6 8
1.68 (s, 3H) 25.9 C-6, C-7, C-9 H-6 9
1.62 (s, 3H) 17.8
[formula] C-6, C-7, C-8 H-5 10 2.18 (s, 3H) 19.3 C-2, C-3, C-4 1' 5.47 (d, 1H, J = 8.2) 95.2 C-1, C-5' 2' [/formula]
3.54 (overlapped) 77.9
## 3'
3.43 (dd, 1H, J = 8.9, 8.9) 77.7
## 4'
3.52 (overlapped) 73.9
## 5'
3.39 (m, 1H) 71.2
## 6'
3.72 (dd, 1H, J = 5.5, 11.0) 69. In the NOESY spectrum, NOE correlations were observed between the following proton pairs: suggesting the trans-configuration at the 6 position, Therefore, compound 3 was identified as (6E)-7-methyl-3-methylene-6-octene-1,2,8-triol.
## Experimental
## General
Open column chromatography was carried out using silica gel (200-300 mesh, Qingdao Marine Chemical Co., Qingdao, China) or octadecyl silica gel (ODS, 25-40 μm, Fuji, Tokyo, Japan) as stationary phase. TLC employed precoated silica gel plates (5-7 μm, Qingdao Marine). Preparative HPLC was carried out on a Waters 600 instrument equipped with a Waters RID-2414 detector. A Waters Sunfire prep C18 OBD (19 × 250 mm i.d.; Waters, Milford, MA, USA) column was used for preparative purpose. The IR spectra were recorded as KBr pellets on a Jasco 302-A spectrometer (Jasco, Tokyo, Japan). The UV spectra were recorded on a Shimadzu UV-2450 spectrophotometer (Shimadzu, Kyoto, Japan). Optical rotation was recorded on a Jasco P-2000 polarimeter. HRESIMS were measured on a FTMS-7 instrument (Bruker Daltonics, Karlsruhe, Germany). The 1 H, 13 C and 2D ( 1 H-1 H COSY, HMQC, HMBC, NOESY) NMR spectra were recorded on a JNM-ECA600 spectrometer (JEOL, Tokyo, Japan) using standard pulse sequence. Chemical shifts were reported in ppm (), and scalar coupling were reported in Hz. GC analyses were carried out using a Fuli 9790 instrument, DM-5 column (0.25 μm, 30 m × 0.25 mm, Dikma, China).
## Plant
The
## Extraction and isolation
The dried and powdered fruits (12 kg) of A. sessiliflorus (Rupr. et Maxim.) Seem. were extracted with 70% EtOH (3 × 32 L) under reflux (1 h). The combined extract was concentrated under vacuum yielding a residue (2.2 kg) which was dissolved in water, loaded on a D101 macroporous adsorption resin column and eluted successively with H 2 O, 30% EtOH, 60% EtOH and 95% EtOH.
The 30% EtOH fraction (100.0 g) was subjected to silica gel column chromatography with a stepwise gradient CHCl 3 -MeOH (10:1, 5:1, 3:1, 2:1, 1:1 v/v), and finally with MeOH alone, to give five fractions 1-6. Fraction 4 (9. [fig_ref] Table 1: showed the presence of three vinyl methyls [1 [/fig_ref]. [fig_ref] Table 2 1: H-and 13 C-NMR data for 2 and 3 [/fig_ref]. [fig_ref] Table 2 1: H-and 13 C-NMR data for 2 and 3 [/fig_ref].
## Acid hydrolysis of and determination of the absolute configuration of the monosaccharides
Compound 1 (3.03 mg) was hydrolyzed with 1 M HCl (1.0 mL) for 2 h at 85 °C. The reaction mixture was cooled and partitioned between CHCl 3 (2.0 mL) and H 2 O (2.0 mL). The aqueous layer was washed with CHCl 3 (3.0 mL × 3), neutralized with Ba(OH) 2 , filtered, and evaporated under reduced pressure. The residue was dissolved in pyridine (1.0 mL) and 0.1 M L-cysteine methyl ester hydrochloride in pyridine (2.0 mL) was added. The mixture was heated at 60 °C for 1 h. An equal volume of Ac 2 O was added with heating continued 1 h. The acetylated thiazolidine derivatives were analyzed by GC using a DM-5 Column (30 m × 0.25 mm, 0.25 μm). Temperatures of injector and detector were 280 °C for both. A temperature gradient system was used for the oven; starting at 160 °C and increasing up to 195 °C at a rate of 5 °C/min. Peaks of the hydrolysate were detected by comparison with retention time of authentic samples of D-glucose (10.08 min) and L-arabinose (6.55 min) after treatment with L-cysteine methyl ester hydrochloride in pyridine.
# Conclusions
In this paper, three new acyclic monoterpenoids, (2E)-3,7-dimethylocta-2,6-dienoate-6-O--Larabinopyranosyl-(16)--D-glucopyranoside (1), (3Z,6E)-3,7-dimethyl-3,6-octadiene-1,2,8-triol (2) and (6E)-7-methyl-3-methylene-6-octene-1,2,8-triol (3) were isolated from the EtOH extract of the dried fruits of Acanthopanax sessiliflorus together with three known monoterpenoids. To the best of our knowledge, this is the first scientific report of acyclic monoterpenoids from Acanthopanax plants.
[fig] Figure 1: Structures [/fig]
[fig] Figure 2: Key HMBC and 1 H-1 H COSY correlations of 1-3. [/fig]
[fig] 7: g) was subjected to reversed-phase silica gel column chromatography [200 g, MeOH-H 2 O 10:90→20:80→30:70→40:60→50:50, v/v)→MeOH] to afford six fractions [Fr. 4.1 3.0 g), Fr. 4.2 506 mg), Fr. 4.3 755 mg), Fr. 4.4 321 mg), Fr. 4.5 612 mg), Fr. 4.6 538 mg)]. Fr. 4.2 506 mg) was separated by HPLC [MeCN-H 2 O 7: 93, v/v)] to give 2 10.6 mg) and 3 11.2 mg). Fraction 4.3 755 mg) was separated by HPLC [MeOH-H 2 O 20:80, v/v)] to give 6 15.8 mg). Fraction 5 15.1 g) was subjected to reversed-phase silica gel column chromatography [300 g, MeOH-H 2 O 10:90→20:80→30:70→50:50→60:40, v/v)→MeOH] to afford ten fractions [Fr. 5.1 1.45 g), Fr. 5.2 1.61 g), Fr. 5.3 168 mg), Fr. 5.4 995 mg), Fr. 5.5 217 mg), Fr. 5.6 1.1 g), Fr. 5.7 121 mg), Fr. 5.8 674 mg), Fr. 5.9 2.0 g), Fr. 5-10 197 mg)]. Fraction 5.2 533 mg) was separated by HPLC [MeOH-H 2 O 35:65, v/v)] to give 1 20.1 mg). Fraction 5.3 197 mg) was separated by HPLC [MeOH-H 2 O 35:65, v/v)] to give 4 15.3 mg). Fraction 5.4 426 mg) was separated by HPLC [MeOH-H 2 O 40:60, v/v)] to give 5 8.5 mg). The known compounds kenposide A 2E)-3,7-Dimethylocta-2,6-dienoate-6-O--L-arabinopyranosyl-16)--D-glucopyranoside 1). [MeOH). HR-ESI-MS m/z 485.1995 [M+Na] + calc. C 21 H 34 O 11 Na, 485.1999); UV max MeOH) 221 nm; IR KBr) 3410, 1715, 1645, 1423, 1385, and 1073 cm −1 ; 1 H-and 13 C-NMR CD 3 OD) data, see [/fig]
[table] Table 1: showed the presence of three vinyl methyls [1.68 (s, 3H, H-8)], [1.62 (s, 3H, H-9)], [2.18 (s, 3H, H-10)], two trisubstituted alkenes [5.10 (dd, 1H, J = 6.6, 6.9 Hz, H-6), 5.74 (s, 1H, H-2)], and two anomeric protons [5.47 (d, 1H, J = 8.2 Hz, H-1')], [4.27 (d, 1H, J = 6.2 Hz, H-1'')]. 1 H-1 H correlations were observed between the following protons pairs: 2.20 (H-4) and 2.21 (H-5); 2.21 (H [/table]
[table] Table 2 1: H-and 13 C-NMR data for 2 and 3 (600 and 150 MHz, CD 3 OD, J in Hertz and in ppm). [/table]
|
A case report: paroxysmal nocturnal hemoglobinuria and systemic lupus erythematosus association
Paroxysmal nocturnal hemoglobinuria (PNH) is defined by acquired intravascular hemolytic anemia, thrombosis and bone marrow failure with pancytopenia. Systemic lupus erythematosus (SLE) also appears as an autoimmune disease. The coexistence of both is rarely reported. Here we report the case of a 30year-old female presenting with pancytopenia and diagnosed as SLE, who also had a PNH clone. Bone marrow biopsy did not support hypoplastic anemia. The patient was then followed up with the consideration of the existence of a PNH clone with SLE. She was treated by the rheumatology department and complete blood count improved under immunosuppressive treatment. The coexistence of CD59-CD55 deficiency with autoimmune diseases has been reported. It is an important example in terms of receiving clinical response with SLE-specific treatment.
Paroxysmal nocturnal hemoglobinuria (PNH) is a clonal hematopoietic stem cell disease manifested by chronic intravascular hemolysis findings, bone marrow failure and thrombosis. It has is an acquired defect of the glycosylphosphatidylinositol associated proteins CD55 and CD59 [bib_ref] Abnormalities in the expression of CD55 and CD59 surface molecules on peripheral..., Ruiz-Delgado [/bib_ref]. These protein deficiencies lead to complementmediated cell lysis and other clinical manifestations. Flow cytometry analysis of antibodies directed against CD55 and CD59 is the gold standard technique for the diagnosis [bib_ref] Detection of PNH cells by flow cytometry, using multiparameter analysis, Rocha [/bib_ref] [bib_ref] Diagnosis of paroxysmal nocturnal hemoglobinuria: recent advances, Manivannan [/bib_ref].
The disease manifests itself in two main clinical pictures: hemolytic and hypoplastic type. Chronic intravascular hemolysis findings are typically more prominent in the hemolytic type, whereas pancytopenia is typically more prominent in the hypoplastic type. Coombs-negative hemolytic anemia (laboratory findings of hemolysis) as well as presence of thrombosis, iron deficiency and/or cytopenia requires a PNH examination. PNH can be found alone (classical PNH) or accompanied by aplastic anemia and myelodysplastic syndrome (MDS) [bib_ref] ICCS/ESCCA consensus guidelines to detect GPI-deficient cells in paroxysmal nocturnal hemoglobinuria (PNH)..., Illingworth [/bib_ref]. PNH also has a clinical importance in terms of hematological malignancies with which it can be seen. Presence of PNH clone with MDS and aplastic anemia changes the treatment approach completely. Similarly, because of being a clonal stem cell disease, it also poses risk for the transformation of acute myeloid leukemia. This rate is 1% within publications [bib_ref] The role of complement regulatory proteins (CD55 and CD59) in the pathogenesis..., Ruiz-Argüelles [/bib_ref].
PNH clone presence with one of subtypes of MDS is not clearly known. In addition, the effect of PNH clone presence on prognosis or leukemic transformation is also controversial. In the series reported by Wang et al., 35% of patients with refractory anemia had a PNH clone and 20% of 5q-syndrome patients also developed a PNH clone during the follow-up period. However, deficiency of multiple glucose phosphate isomerase-linked (GPI-linked) proteins in granulocytes was not detected in patients with a diagnosis of sideroblastic anemia, refractory anemia with excessive blast or clear acute myeloid leukemia. In general, it can be stated that acute leukemia transformation is seen less frequently and the prognosis is better for MDS cases with a PNH clone; however, there is no difference at the point of treatment [bib_ref] Paroxysmal nocturnal hemoglobinuria and myelodysplastic syndromes: clonal expansion of PIG-A-mutant hematopoietic cells..., Young [/bib_ref]. MDS association with systemic lupus erythematosus (SLE) is extremely rare and is occasionally mentioned in some case reports. This case report shall present a case with PNH clone that was diagnosed with SLE and presented with bicytopenia.
## Case report
A 30-year-old female patient was examined in the rheumatology clinic for alopecia, bilateral pleural, pericardial effusion and nonerosive oligoarthritis, particularly involving the hand joints. Together with the laboratory results, it fully met the Systemic Lupus International Collaboration Criteria for SLE. She was referred to our hematology clinic on suspicion of aplastic anemia/MDS with a positive FLAER test. She had no complaints other than headache and weakness in her referral. Initial hemoglobin was 5 g/dl, leukocyte 800/mm 3 and platelet was 26,000/mm 3 . In the biochemical findings, LDH was 400 IU/l (normal range: 140-280 IU/l). Corrected reticulocyte count was 4.5%, haptoglobin was 0.3 (0.5-2.2 g/l), C3 level was 30 mg/dl (80-160 mg/dl) and C4 level was 5 mg/dl (10-40 mg/dl). The direct and indirect Coombs tests were negative and the antinuclear antibody titer 1:2560 homogeneous positive, anti-double stranded DNA result was positive. In her peripheral blood smear, it was possible to see reticulocytosis with any other pathological findings. Due to Coombs-negative hemolytic anemia, the PNH clone was rerun with FLAER test and was positive with erythrocytes 0.6%, 7.79% for monocytes and 11.25% for granulocytes. [fig_ref] Figure 1: FLAER Test [/fig_ref] reveals the presence of PNH clones: A in erythrocytes, B in granulocytes and C in monocytes. Bone marrow aspiration smears were normocellular with normoblastic erythroid hyperplasia, normal myeloid maturation and adequate megakaryopoesis. There was a mild increase in mature plasma cells, which was considered to be associated with SLE. There was no fibrosis or dysplastic changes in bone marrow specimen [fig_ref] Figure 2: Hematoxylin-eosin staining of the bone marrow specimen [/fig_ref].
She was treated in the rheumatology clinic with pulse 1 g/day for 2 days methylprednisolone treatment. After that, 20 mg of prednisone and cyclosporine 2 × 50 mg were added to her treatment. Since the need for transfusion completely disappeared by the end of the first month, treatment was continued with the same immunosuppressive therapy. In parallel with the publications that reported that CD59 expression decreased with SLE, the FLAER test was repeated and the PNH clone was confirmed. In addition to the stable immunosuppressive treatment, the patient was planned to have rituximab because of refractory oligoarthritis and the patient was taken to the hematology and rheumatology outpatient follow-up.
# Discussion
PNH presents with acquired intravascular hemolytic anemia, large vein thrombosis and pancytopenia clinics with moderate to severe bone marrow failure. In our case, there were no triad findings other than cytopenia. As mentioned, the pathology is based on CD55 and CD59 deficiency. However, CD55 and CD59 deficiency can also be seen in autoimmune thrombocytopenia, hemolytic anemia and systemic lupus erythematosus. It is possible to define them as diseases that create a 'PNH-like phenomenon' [bib_ref] Paroxysmal nocturnal hemoglobinuria in systemic lupus erythematosus: a case report, Nakamura [/bib_ref] [bib_ref] Recent advances in biological and clinical aspects of paroxysmal nocturnal hemoglobinuria, Luzzatto [/bib_ref].
Various pathogenetic mechanisms for cytopenia are defined in SLE patients. As an additional factor, the expression of CD55 and CD59 is also reduced in SLE. It has been observed that decreased expression of CD55 and CD59 has a facilitating role in autoimmune hemolytic anemias. The relationship between CD55 expression and antiphospholipid syndrome on erythrocyte surfaces from secondary autoimmune hemolytic anemia to SLE or primary autoimmune hemolytic anemia patients was investigated. In the presence of autoimmune hematolytic anemia, independent from SLE, CD55 and CD59 have been shown to be deficient. However, the relationship between antiphospholipid antibodies and this acquired defect has not yet been established [bib_ref] Relationship between bone marrow failure syndromes and the presence of glycophosphatidyl inositol-anchored..., Maciejewski [/bib_ref] [bib_ref] Development of paroxysmal nocturnal haemoglobinuria in systemic lupus erythematosus: an unusual cause..., Gupta [/bib_ref].
In SLE cases with autoimmune thrombocytopenia and lymphopenia, the role of CD55 and CD59 expression is another topic that has been studied. Lymphopenia in SLE has been associated with autoantibodies that target nuclear antigens. CD55 and CD59 in T and B lymphocytes decreased significantly in SLE patients with lymphopenia. Interestingly, on the contrary, in nonlymphopenic patients with SLE, normal expression was found in T and B cells. It has been suggested that deficiency of CD55 and CD59 expression may play a role in the pathophysiology of lymphopenia, possibly by increasing the complement-mediated cytolysis sensitivity of cells. Although there is no correlation between CD55 and CD59 surface density and the presence of any specific autoantibodies, there appears to be a relationship between high anti-double strand DNA, anti-Sm and antiribosomal P antibody titers and lymphopenia [bib_ref] The role of complement regulatory proteins (CD55 and CD59) in the pathogenesis..., Ruiz-Argüelles [/bib_ref].
In the presence of PNH clone associated with SLE, there is no data in the literature that clearly demonstrates the pathology of cytopenia. While SLE-related cytopenia can occur with autoantibody-based cell destruction, receptor blockage and apoptosis, the contribution to cytopenia can be mentioned in the presence of PNH clone. Considering SLE cases in which decreased expression of CD55 and CD59 is associated with hemolytic anemia and lymphopenia, it can be said that in our case, the presence of PNH clone contributed to complement-related cell destruction .
We present a case about this rare association. Instead of PNH specific therapy, the treatment of the underlying disease makes a clinically significant difference. Treatment was started by evaluating our patient's current clinical picture secondary to SLE, not the PNH clone. Therefore, determining the cause of the clinical picture becomes important in the association of rare diseases. In our case, corticosteroid was started first and as maintenance, it was planned to continue with cyclosporin. Although cytopenia seems to improve with the current treatment, rituximab was added to the treatment due to the continuing joint complaints of the patient and rheumatological follow-up continues under this plan. In the presence of resistant cytopenia, the addition of PNH-specific therapy may come to mind. In the treatment of PNH, the effect of corticosteroids decreasing hemolysis, even if not in every patient, can be observed. Corticosteroid therapy can be effective in increasing hemoglobin levels, especially during periods of refractory hemolysis. Side-effects in long-term treatment limit their use. With targeting complement C5 protein, eculizumab is a treatment option that has been shown to significantly reduce hemolysis and thus hemolysis-related side-effects in PNH cases. In this context, in the presence of resistant cytopenia, specific treatment for the PNH clone may be the subject of controversy in the future.
# Conclusion
The coexistence of CD59-CD55 deficiency with autoimmune hemolytic anemia, autoimmune thrombocytopenia or SLE have been reported. Therefore, the factor causing PNH clone in our case is associated with SLE and it is an important example for clinicians in terms of receiving a clinical response with SLE-specific treatment.
- Paroxysmal nocturnal hemoglobinuria (PNH) is a clonal hematopoietic stem cell disease manifested by chronic intravascular hemolysis findings, bone marrow failure and thrombosis. - PNH has a clinical importance in terms of hematological malignancies with which it can be seen. Because of being a clonal stem cell disease, it also poses risk for the transformation of acute myeloid leukemia. - Its association with systemic lupus erythematosus is extremely rare and is occasionally mentioned in some case reports. - However, CD55 and CD59 deficiency can also be seen in autoimmune thrombocytopenia, hemolytic anemia and systemic lupus erythematosus (SLE); it is possible to define them as diseases that create a 'PNH-like phenomenon'. - CD55 and CD59 in T and B lymphocytes are decreased significantly in SLE patients with lymphopenia.
- It has been suggested that deficiency of CD55 and CD59 expression may play a role in the pathophysiology of lymphopenia, possibly by increasing the complement-mediated cytolysis sensitivity of cells. - Considering SLE cases in which decreased expression of CD55 and CD59 is associated with hemolytic anemia and lymphopenia, it can be said that the presence of a PNH clone contributed to complement-related cell destruction. - In the treatment of PNH, the effect of corticosteroids decreasing hemolysis, even if not in every patient, can be observed. - Corticosteroid therapy can be effective in increasing hemoglobin levels, especially during periods of refractory hemolysis. - By targeting complement C5 protein, eculizumab is a treatment option that has been shown to significantly reduce hemolysis and thus hemolysis-related side-effects in PNH cases. In the presence of resistant cytopenia, specific treatment for the PNH clone may be the subject of controversy in the future.
[fig] Figure 1: FLAER Test. PNH clones; (A) in erythrocytes, (B) in granulocytes and (C) in monocytes. PNH: Paroxysmal nocturnal hemoglobinuria. [/fig]
[fig] Figure 2: Hematoxylin-eosin staining of the bone marrow specimen. [/fig]
|
Over-expression of the c-myc proto-oncogene in colorectal carcinoma.
Alterations in the c-myc proto-oncogene in colorectal cancer were studied at the level of RNA expression, gene amplification and rearrangements. One hundred cases of colorectal cancer, stratified by Dukes' stage were examined. The level of messenger RNA expression was measured in tumours and matched normal mucosa from the same patient. Between 5 and 400 fold over-expression was found in 66% of tumours. Neither the presence nor the level of over-expression correlated with tumour staging. A significant correlation (P<0.01) was found between over-expression of c-myc in tumours and the presence of synchronous adenomas elsewhere in the colon. In contrast to other tumours, no rearrangements of the gene were found on Southern analysis of colorectal cancers. Similarly, amplification of the gene was not found in the cancers examined.
Since the establishment of the National Cancer Registry in 1968, Singapore has seen a steady increase in the incidence of colorectal cancer, with a standardised rate of 19.9 and 15.7 per 100,000 for males and females for the period compared with rates of 31 and 26.3 for males and females for 1983-1987, giving an increase in incidence of 55% and 67% respectively. In terms of number of cases, there has been an average annual increase of 3.5% since 1968. Colorectal cancer was the sixth most common cancer at the start of the Cancer Registry; it ranks second today, and will be the most common cancer in Singapore by the end of this decade.
Being a small country with a dense population of 3 million, Singapore provides an ideal opportunity for studying the carcinogenic process of colorectal cancer, particularly, because it has three different races which manifest different risks. Among Singaporeans the incidence of colorectal cancer is highest in the Chinese population (which comprises 75% of the total). The Chinese also have the highest rate of increase. As most of the Chinese population originated from the Southern coastal provinces of China, any study here will provide an opportunity for comparing the aetiology of colorectal cancer for Chinese populations in China, California.
In western populations colorectal cancer is one of the most intensively studied malignancies, due to the availability of clearly defined stages between normal colonic mucosa and the fully malignant carcinoma, and which are now being correlated to specific gene changes [bib_ref] A genetic model for colorectal tumorigenesis, Fearon [/bib_ref] [bib_ref] Progressing towards a molecular description of colorectal cancer development, Fearon [/bib_ref]. A number of cellular protooncogenes have been examined in colorectal cancer, particularly the involvement of the cellular proto-oncogene c-Ki-ras [bib_ref] c-Ki-ras mutations in chronic ulcerative colitis and sporadic colon carcinoma, Burmer [/bib_ref] [bib_ref] Detection of high incidence of K-ras oncogenes during human tumorigenesis, Forrester [/bib_ref] [bib_ref] Genetic alterations during colorectal tumor development, Vogelstein [/bib_ref] [bib_ref] Mutations in the KRAS2 oncogene during progressive stages of human colon carcinoma, Burmer [/bib_ref] [bib_ref] ras genes, Barbacid [/bib_ref] [bib_ref] ras oncogenes in human cancer: A review, Bos [/bib_ref] [bib_ref] The biochemistry of ras p21, Grand [/bib_ref] , and the tumour suppressor gene p53 [bib_ref] p53 gene mutations occur in combination with 17p allelic deletions as late..., Baker [/bib_ref] [bib_ref] p53 mutations in colorectal cancer, Rodriguez [/bib_ref] [bib_ref] A genetic model for colorectal tumorigenesis, Fearon [/bib_ref] [bib_ref] Chromosome 17 deletions and p53 gene mutations in colorectal carcinomas, Baker [/bib_ref] [bib_ref] Mutations in the p53 gene occur in diverse human tumor types, Nigro [/bib_ref] [bib_ref] p53 mutations in human cancers, Hollstein [/bib_ref]. Less well studied in colorectal cancer is the proto-oncogene c-myc, althoughthis has been intensively studied in other malignancies such as Burkitts lymphoma [bib_ref] Translocation of the c-myc gene into the immunoglobulin heavy chain locus in..., Taub [/bib_ref] [bib_ref] Effect of somatic mutations within translocated c-myc genes in Burkitt's lymphoma, Rabbitts [/bib_ref] [bib_ref] Translocations among antibody genes in human cancer, Leder [/bib_ref] [bib_ref] Assignment of the human c-myc onc-gene to the region of chromosome 8..., Dalla Favera [/bib_ref] [bib_ref] Translocation and rearrangements of the c-myc onc-gene in human undifferentiated B-cell lymphomas, Dalla Favera [/bib_ref] [bib_ref] Translocation joins c-myc and immunoglobulin GI genes in a Burkitt lymphoma revealing..., Hamelyn [/bib_ref].
The role of c-myc in colorectal carcinomas is not well understood. Immunohistochemistry has shown that c-myc gene product in normal colonic tissue is located in the mid zone of the colonic crypts, which corresponds to the zone of maturation and differentiation of colonic epithelial cells [bib_ref] Detection of the c-myc oncogene product in colonic polyps and adenocarcinomas, Stewart [/bib_ref] [bib_ref] Distribution of cells expressing myc proteins in human colorectal epithelium, polyps and..., Melhem [/bib_ref]. localisation extends into the proliferative zone while in colorectal carcinoma c-myc staining can be found in the mature zone as well as the maturation and proliferative zones of colonic crypts [bib_ref] Detection of the c-myc oncogene product in colonic polyps and adenocarcinomas, Stewart [/bib_ref] [bib_ref] Distribution of cells expressing myc proteins in human colorectal epithelium, polyps and..., Melhem [/bib_ref]. Over-expression of the c-myc mRNA has been reported to occur in between 60%-80% [bib_ref] Expression of the myc gene family in different stages of human colorectal..., Finley [/bib_ref] [bib_ref] c-myc oncogene expression in colorectal cancer, Sikora [/bib_ref] [bib_ref] Evidence that c-myc expression defines two genetically distinct forms of colorectal adenocarcinoma, Rothberg [/bib_ref] [bib_ref] Expression of c-myc and other cell cycle dependent genes in human colon..., Calabretta [/bib_ref] [bib_ref] Deregulation of c-myc gene expression in human colon carcinoma is not accompanied..., Erisman [/bib_ref] [bib_ref] Expression of the c-myc gene in human gastrointestinal malignancies, Tsuboi [/bib_ref] [bib_ref] Expression of c-myc oncogene in colorectal polyps as a biological marker for..., Imaseki [/bib_ref] of colon carcinomas, although the number of samples in these studies is rather small, from a minimum of six tumours [bib_ref] Expression of c-myc and other cell cycle dependent genes in human colon..., Calabretta [/bib_ref] to a maximum of 38 [bib_ref] Evidence that c-myc expression defines two genetically distinct forms of colorectal adenocarcinoma, Rothberg [/bib_ref]. One study by [bib_ref] Evidence that c-myc expression defines two genetically distinct forms of colorectal adenocarcinoma, Rothberg [/bib_ref] reports a correlation between overexpression of c-myc and the location of the tumour, and although statistically significant, is based on a relatively small sample size of 38 tumours. However this result has not been supported by other workers [bib_ref] Expression of c-myc oncogene in colorectal polyps as a biological marker for..., Imaseki [/bib_ref] , but again this is based on a small sample size (11 tumours). As yet no correlation has been found between over-expression of the c-myc proto-oncogene and either patient survival or disease recurrence [bib_ref] Noncorrelation of the expression of the c-myc oncogene in colorectal carcinoma with..., Erisman [/bib_ref] , or metastatic potential [bib_ref] Expression of the c-myc gene in human gastrointestinal malignancies, Tsuboi [/bib_ref].
Amplification of the c-myc oncogene has been reported in fresh colonic tumours, although the incidence is low varying from 6%, (2/32; [bib_ref] Proto-oncogene abnormalities in colon cancers and adenomatous polyps, Meltzer [/bib_ref] , to 22% (2/9; [bib_ref] Oncogene alterations in primary human colon tumors, Alexander [/bib_ref]. A better correlation is found when only aggressive subtypes of colorectal tumours (such as mucinous or poorly differentiated tumours) were examined. In these cases slight amplification of the c-myc gene is found in approximately 50% of cases [bib_ref] Aggressive subtypes of human colorectal tumours frequently exhibit amplification of the c-myc..., Heerdt [/bib_ref]. None of these papers report any rearrangement of the c-myc gene in colorectal carcinomas.
The present study was undertaken with two main points in mind. Firstly we wished to examine the type of oncogenic changes occurring in an Asian population which is showing a rapid increase in incidence of colorectal cancer. Secondly we wished to determine, using a larger sample base, accurate correlations between c-myc and various clinical correlates such as Dukes' stage, age, sex, and tumour site.
# Materials and methods
## Tumour specimens
Samples used in this study were from patients admitted to the Department of Colorectal Surgery at Singapore General Hospital. No initial chemotherapy, radiotherapy or hormonal therapy was given prior to tumour excision. A portion of the surgically removed tumour was snap frozen in liquid nitrogen and stored at -80°C until required. The remainder of the tumour sample was sent for histopathological diagnosis. Control mucosa (sited at least 10 cm proximal to the site of the tumour) was also removed and similarly treated.
Isolation of RNA and Northern blotting RNA was extracted from tumour and mucosa samples by the method of Chomcznski and Sacchi [bib_ref] Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction, Chomcznski [/bib_ref] , followed by caesium chloride centrifugationand quantitated by UV spectrophotometry. Total RNA was fractionated through formaldehyde-agarose gels, transferred to a solid matrix (Hybond-N, Amersham, Arlington Heights, IL), and hybridised to 32P random primed labelled [bib_ref] A technique for radiolabelling DNA restriction endonuclease fragments to high specific activity, Feinberg [/bib_ref] [bib_ref] Amplification of the c-myc oncogene in human stomach cancers, Sugimurea [/bib_ref] cDNA probes. The following double-stranded probes were employed: c-myc cDNA (pGl-5'-c-myc; American Type Culture Collection, Rockville, MD); P-actin cDNA clone (Clontech Laboratories, Palo Alto, CA). After hybridisation filters were exposed to Fuji-RX medical X-ray film (Japan) for between 2-5 days. Signal was quantitated on a CS-9000 scanning densitometer (Shimadzu, Japan). Corrected myc signal in the tumour was compared to the corrected myc signal in the mucosa by comparison with the 13-actin control signal to obtain a number representing the level of overexpression in the tumour. Reproducibility was assessed by 10% of the samples being analysed on separate Northern filters. Reproducibility for all samples was found to be ± 30%, with the majority being ± 20%. The greatest variation was found in one sample from which RNA was extracted from two separate portions of the tumour and probably reflects differing amounts of stromal cell contamination.
Isolation of DNA, Southern hybridisation and DNA dot blot analysis DNA was isolated by standard methods [bib_ref] Preparation of DNA from eukaryotic cells, Davis [/bib_ref]. For dot blot analysis approximately 10 gg of genomic DNA was transferred to a Nylon filter (Hybond-N, Amersham), denatured by soaking in 1.5 M NaCl, 0.5 M NaOH and neutralised by soaking in 2 M NaCl, 0.5 M Tris-HCl pH 6.0. After drying, DNA was cross-linked by UV radiation and filter hybridised overnight to random prime [bib_ref] A technique for radiolabelling DNA restriction endonuclease fragments to high specific activity, Feinberg [/bib_ref] [bib_ref] Amplification of the c-myc oncogene in human stomach cancers, Sugimurea [/bib_ref] labelled cDNA probes. Probes used were c-myc exon 1 cDNA (pGl-5'-c-myc; American Type Culture Collection, Rockville, MD); carboxypeptidase H [bib_ref] Human carboxypeptidase E: isolation and characterisation of the cDNA clone, sequence conservation,..., Manser [/bib_ref] ; c-Ki-ras exon 1 [bib_ref] ras genes, Barbacid [/bib_ref] [bib_ref] Primerdirected enzymatic amplification of DNA with a thermostable DNA polymerase, Saiki [/bib_ref] [bib_ref] Specific synthesis of DNA in vitro via a polymerase catalysed chain reaction, Mullis [/bib_ref] [bib_ref] Enzymatic amplification of P-globin genomic sequences and restriction site analysis for diagnosis..., Saiki [/bib_ref] product (primers from Clontech Laboratories, Palo Alto, CA) and P-actin (Clontech Laboratories, Palo Alto, CA). After each exposure the filter was stripped by boiling in 0.1% SDS. For Southern analysis 10 tig of normal mucosa and tumour DNA was digested with EcoRl restriction endonuclease (New England Biolabs) and subjected to electrophoresis on a 0.8% agarose gel. After electrophoresis, DNA was denatured and neutralised as above and transferred to solid matrix (Hybond-N, Amersham) by overnight capillary action. The filter was hybridised overnight with c-myc cDNA probe (pGl-5'-c-myc; American type Culture Collection, Rockville, MD), labelled by the random prime method [bib_ref] A technique for radiolabelling DNA restriction endonuclease fragments to high specific activity, Feinberg [/bib_ref] [bib_ref] Amplification of the c-myc oncogene in human stomach cancers, Sugimurea [/bib_ref] with 32P, and autoradiography performed.
# Results
Over-expression of c-myc RNA The level of c-myc messenger RNA in 100 colorectal tumours, equally divided by Dukes' stage (25 Dukes' A, 25 Dukes' B, 25 Dukes' C and 25 Dukes' D) was measured and compared to levels found in normal mucosa from the same patient. The tumours came from 55 male and 45 female patients, average age 62 years (range 24-89 years). There were 92 Chinese patients and eight others. In each case, 10 fig of total RNA from both the tumour and from matched normal mucosa was fractionated through formaldehyde-agarose gels and hybridised initially with c-myc cDNA. To compensate for variations in the amount of RNA loaded in each lane a second hybridisation with ,-actin was undertaken. Levels of c-myc were then quantitated against levels of P-actin. In all 66 of the tumours were shown to over-express c-myc (see . Only tumours showing a greater than 3-fold increase in c-myc levels were considered to be over-expressing. Thirty-four per cent of the tumours showed no overexpression. Low levels of over-expression (3-10-fold increase in levels of c-myc RNA) was found in 20%, moderate levels of c-myc over-expression (11-30-fold increase) was found in 29%, and high levels of RNA over-expression (>30-fold increase) was found in 17% (see [fig_ref] Table I: Levels of c-myc expression in colorectal carcinomas [/fig_ref] of tumours. No correlation was found between the presence of c-myc overexpression and the stage of the tumour (see [fig_ref] Table I: Levels of c-myc expression in colorectal carcinomas [/fig_ref] ; the level of c-myc over-expression and Dukes' staging (see [fig_ref] Table I: Levels of c-myc expression in colorectal carcinomas [/fig_ref] ; the presence of c-myc over-expression and age [fig_ref] Table I: Levels of c-myc expression in colorectal carcinomas [/fig_ref].
Furthermore in contrast to other workers , no correlation was found between the site of the tumour, i.e. left or right side tumours (where left side tumours are those of the rectum, sigmoid colon, descending colon and splenic flexure and right side tumours are those of the caecum, hepatic flexure, ascending colon and transverse colon) and c-myc over-expression (see [fig_ref] Table I: Levels of c-myc expression in colorectal carcinomas [/fig_ref]. c-myc overexpression did however correlate with the presence of synchronous adenomas [fig_ref] Table I: Levels of c-myc expression in colorectal carcinomas [/fig_ref]. Of the 66 patients where over-expression was found in the tumour, 22 possessed synchronous polyps, while only two patients out of 34 not having c-myc over-expression also possessed synchronous polyps. Hence there is a significant correlation (P> 0.01, analysed by x2 test) between tumours over-expressing c-myc and the presence of synchronous polyps. Five patients had synchronous cancers and tumours from four out of the five expressing the tumour and 10 fig of DNA from matched normal mucosa of the same patient was digested with the restriction endonuclease Eco RI, the digest products were separated on a 0.8% agarose gel and then transferred to solid nylon matrix support prior to hybridisation with a c-myc cDNA probe. Results shown in show only the expected single band at 12.5 kb [bib_ref] Translocation of the c-myc gene into the immunoglobulin heavy chain locus in..., Taub [/bib_ref] [bib_ref] Translocation and rearrangements of the c-myc onc-gene in human undifferentiated B-cell lymphomas, Dalla Favera [/bib_ref] which is of equal intensity between the tumour sample and the normal mucosa. No amplification is therefore present. Furthermore no rearrangement of the c-myc gene was found at this level of resolution in any of the ten samples.
## Gene rearrangements
The Southern analysis in shows that of the ten samples analysed no detectable rearrangements were found. For this reason a larger number of samples was examined. A further 32 samples were digested with the restriction endonuclease EcoRl (in this case only DNA from the tumour was examined) and the DNA separated by agarose gel electrophoresis. The final sample composition was 11 Dukes' A tumours, 13 Dukes' B tumours, five Dukes' C tumours and 13 Dukes' D tumours. After Southern transfer, rearrangements of the c-myc proto-oncogene were analysed by hybridisation with a c-myc cDNA probe. Representative results are shown in . As can be seen at this level of resolution no gross rearrangement was found in any of the samples analysed.
Discussion patients had increased levels of c-myc in the main tumour analysed here. One of these patients with synchronous cancers also had synchronous polyps and expressed c-myc at moderate levels in the main tumour (9-fold over-expression).
## Gene amplification
A total of 50 colon carcinomas were examined for amplification of the c-myc cellular proto-oncogene by dot blot hybridisation. Of the 50 samples, 12 were Dukes' A, 17 were Dukes' stage B, eight were Dukes' stage C and 13 were Dukes' stage D. No tumour was found to contain amplified c-myc, as compared against hybridisation of P-actin, c-Ki-ras exon 1 and carboxypeptidase H [bib_ref] Human carboxypeptidase E: isolation and characterisation of the cDNA clone, sequence conservation,..., Manser [/bib_ref] , a gene known to be present as a single copy in the human genome (DRS, unpublished data), see for representative dot blots. In confirmation of the dot blot analysis, no amplification could be discerned in ten samples analysed by genomic Southern blotting. In this case, 10 ytg of DNA from
The c-myc proto-oncogene is the cellular homologue of the v-myc oncogene of avian myelocytomatosis virus [bib_ref] Isolation and characterisation of c-myc, a cellular homologue of the oncogene (v-myc)..., Venstrom [/bib_ref] and is a member of the myc family of oncogenes, which contains five other members besides c-myc; namely N-myc, Lmyc, Rmyc, Pmyc and Bmyc [bib_ref] myc family of cellular oncogenes, De [/bib_ref]. The c-myc proto-oncogene is present as a single copy gene in the normal human genome and has been localised to chromosome 8 (specifically at 8q24) [bib_ref] Translocation of the c-myc gene into the immunoglobulin heavy chain locus in..., Taub [/bib_ref] [bib_ref] Assignment of the human c-myc onc-gene to the region of chromosome 8..., Dalla Favera [/bib_ref] [bib_ref] Two human c-onc genes are located in the long arm of chromosome..., Neel [/bib_ref] , and consists of one non-coding exon and two coding exons separated by two introns [bib_ref] Translocation joins c-myc and immunoglobulin GI genes in a Burkitt lymphoma revealing..., Hamelyn [/bib_ref] [bib_ref] myc family of cellular oncogenes, De [/bib_ref]. We have examined the expression of c-myc in colorectal tumours. We have found that 66% of colorectal tumours show some degree of over-expression of c-myc RNA. The samples analysed were representative of the stages of colorectal tumour as determined by histopathologic analysis. The tumour samples analysed consisted of 25 Dukes' A tumours, [fig_ref] Table I: Levels of c-myc expression in colorectal carcinomas [/fig_ref] (although in agreement with other workers [bib_ref] Expression of the myc gene family in different stages of human colorectal..., Finley [/bib_ref] there is perhaps a slight reduction of the percentage of late state tumours over-expressing c-myc i.e. Dukes' stage C and D tumours, although as with Finley et al., this reduction is not statistically significant), or with the age or sex of the donor [fig_ref] Table I: Levels of c-myc expression in colorectal carcinomas [/fig_ref] and V respectively). The degree of overexpression ranges from 5-fold to in excess of 400-fold overexpression, but again this does not correlate with the stage of the tumour [fig_ref] Table I: Levels of c-myc expression in colorectal carcinomas [/fig_ref]. [bib_ref] Evidence that c-myc expression defines two genetically distinct forms of colorectal adenocarcinoma, Rothberg [/bib_ref] , found a significant correlation between the site of the tumour and the over-expression of c-myc in a study on 38 colorectal tumours. In their study they find that 81% of left side tumours (those of the rectum, splenic fixture, sigmoid colon and descending colon) overexpress c-myc, whereas only 36% of right side tumours (those of the caecum, hepatic flexure, ascending colon and transverse colon) show elevated levels of c-myc expression (or alternatively 85% of elevated c-myc expression is found in left side tumours, whereas only 15% of elevated expression is found in right side tumours). In our study, however, we find that an almost identical proportion of left and right side tumours over-express c-myc, i.e. 65% and 60% respectively, although it should be noted that 80% of the tumours in our sample cohort are left side tumours, whereas only 15% of the tumours are right side tumours (five tumours are not included in this analysis as the donors had multiple colorectal tumours). Nevertheless our sample size is more than 2.5 times the size of that used in the study by Rothberg et al.
A significant correlation was found in our study between c-myc being over-expressed in the tumour and the presence of synchronous polyps elsewhere in the colon [fig_ref] Table I: Levels of c-myc expression in colorectal carcinomas [/fig_ref]. Synchronous polyps were found in 24 of the patients. Twenty-two of these cases occurred in patients with c-myc over-expression in the tumour, whereas only two of the cases of synchronous polyps were found in those patients whose tumours did not over-express c-myc. This correlation is statistically significant (P<0.01). Furthermore the fact that 5% of the patients in this study showed synchronous cancers and that 80% of these patients showed c-myc over-expression in the main tumour, might indicate that c-myc changes are more widespread throughout the colon than has been found with other oncogenes and tumour suppressor genes such as c-Ki-ras and p53, whose changes tend to be localised to the site of the tumour. It is possible to speculate therefore that a wide spread colonic alteration could be genetically predetermined. This supposition can be partially supported by the wide range of basal levels of c-myc messenger RNA noted by ourselves (data not shown) and others [bib_ref] Expression of the myc gene family in different stages of human colorectal..., Finley [/bib_ref]. Amplification and rearrangement of the c-myc cellular proto-oncogene has been shown to be associated with several different malignancies [bib_ref] Amplification and expression of the c-myc oncogene in human lung cancer cell..., Little [/bib_ref] [bib_ref] High correlation between molecular alterations of the c-myc oncogene and carcinoma of..., Ocadiz [/bib_ref] [bib_ref] Homogeneously staining chromosomal regions contain amplified copies of an abundantly expressed cellular..., Alitalo [/bib_ref] [bib_ref] Amplification of the c-myc oncogene in one of five human breast carcinoma..., Kozbar [/bib_ref] [bib_ref] Amplification of endogenous myc-related sequences in a human myeloid leukaemia cell line, Collins [/bib_ref] [bib_ref] Genomic alterations involving the c-myc proto-oncogene locus during the evolution of a..., Mccarthy [/bib_ref] [bib_ref] onc gene amplification in promyelotic leukaemia cell line HL-60 and primary leukaemic..., Dalla Favera [/bib_ref] [bib_ref] Amplification of the EGF receptor and c-myc genes in human esophaegeal cancers, Lu [/bib_ref] [bib_ref] Structure and expression of the oncogene c-myc in fresh tumor material from..., Rothberg [/bib_ref] [bib_ref] Aggressive subtypes of human colorectal tumours frequently exhibit amplification of the c-myc..., Heerdt [/bib_ref]. However, we have examined 50 colorectal tumours of different stages and find no evidence of gene amplification. Furthermore, examination of some 42 tumours by genomic Southern analysis shows no evidence of gene rearrangements. The lack of amplification or rearrangement of the c-myc gene in colorectal carcinomas clearly indicate that a different mechanism of activation is occurring in these tumours as opposed to tumours of the lymphatic system such as Burkitt's lymphoma, and tumours derived from uterine cervix [bib_ref] High correlation between molecular alterations of the c-myc oncogene and carcinoma of..., Ocadiz [/bib_ref] , esophageal cancers [bib_ref] Amplification of the EGF receptor and c-myc genes in human esophaegeal cancers, Lu [/bib_ref] , hematopoietic malignancies [bib_ref] Structure and expression of the oncogene c-myc in fresh tumor material from..., Rothberg [/bib_ref] and stomach cancers , where amplification and/or rearrangement have been shown to be correlated with tumorigenesis in primary biopsy samples.
Hence, in colorectal cancers as activation of c-myc is not a result of either amplification or rearrangement then activation could result from either point mutations in the c-myc gene, either in the promoter region or within the first exon as has been shown for some other malignancies, or possibly by the activation or deactivation of a trans-activating factor. This latter possibility is supported by studies that show that c-myc over-expression is correlated with loss of chromosome 5 alleles [bib_ref] Evidence that the familial adenomatous polyposis gene is involved in a subset..., Erisman [/bib_ref]. Furthermore the introduction of chromosome 5 by microcell fusion into colon carcinoma cell lines leads to the suppression of c-myc deregulation [bib_ref] Suppression of deregulated c-MYC expression in human colon carcinoma cells by chromosome..., Rodriguez-Alfageme [/bib_ref] , possibly by the reintroduction of a functional APC gene, a gene that is implicated in the genesis of spontaneous colorectal cancers and highly implicated in the familial adenomatous polyposis syndrome, an inherited susceptibility to colon cancer and which is known to reside on chromosome 5 [bib_ref] Identification of deletion mutations and three new genes at the familial polyposis..., Joslyn [/bib_ref].
[fig] Figure 1, Figure 2: Representative Northern blot analysis of 21 carcinomas (T) and their corresponding normal mucosa (M) hybridised initially with c-myc cDNA and then P-actin and shown as a composite. The number above each pair (T,Representative dot blot hybridisation of DNA from 16 tumour samples. Each dot contains 10 #g of DNA and filter was hybridised sequentially with probes for c-myc, carboxypeptidase H, c-Ki-ras and P-actin. Numbers correspond to patients. 25 Dukes' B tumours, 25 Dukes' C and 25 Dukes' D tumours. The degree of over-expression does not correlate with either stage of the tumour as is shown in [/fig]
[fig] Figure 3, Figure 4: Representative Southern hybridisation analysis of c-myc in seven carcinomas (T) and compared with their corresponding normal mucosa (M). Each lane consists of 10 ,ug of DNA digested with EcoRl. The 12.5 Kb c-myc band is indicated. Band size is estimated against . DNA digested with HindIII (Lane MA) and sized as shown. The number above each pair of lanes corresponds to the patient number. Representative Southern analysis of c-myc gene rearrangement in 20 carcinoma samples. Numbers above lanes correspond to patient numbers. Each lane contains approximately 10 jig of DNA digested with EcoRl. Indicated is the 12.5 Kb c-myc product as compared with A DNA digested with HindIII (Lane MA, size of bands shown). [/fig]
[table] Table I: Levels of c-myc expression in colorectal carcinomas [/table]
[table] Table VI: Over-expression of c-myc stratified by tumour location [/table]
|
In Vitro Models of Ovarian Cancer: Bridging the Gap between Pathophysiology and Mechanistic Models
Ovarian cancer (OC) is a disease of major concern with a survival rate of about 40% at five years. This is attributed to the lack of visible and reliable symptoms during the onset of the disease, which leads over 80% of patients to be diagnosed at advanced stages. This implies that metastatic activity has advanced to the peritoneal cavity. It is associated with both genetic and phenotypic heterogeneity, which considerably increase the risks of relapse and reduce the survival rate. To understand ovarian cancer pathophysiology and strengthen the ability for drug screening, further development of relevant in vitro models that recapitulate the complexity of OC microenvironment and dynamics of OC cell population is required. In this line, the recent advances of tridimensional (3D) cell culture and microfluidics have allowed the development of highly innovative models that could bridge the gap between pathophysiology and mechanistic models for clinical research. This review first describes the pathophysiology of OC before detailing the engineering strategies developed to recapitulate those main biological features.Biomolecules 2023, 13, 103 2 of 28 fluid, known to contribute to patient morbidity and mortality by facilitating metastasis and contributing to chemoresistance, has become a hallmark of OC[6]. It is thus of particular importance to integrate fluids in in vitro models, both in terms of composition and shear stress. After a survey of 3D cultures and spheroid models, we focus on the integration of mechanotransduction in in vitro models as generated by the shear stress from surrounding ascites. In particular, we discuss the ability of microfluidics to model the peritoneal cavity with the associated mechanical constraints, enabling control of the shear stress to quantitatively monitor the cell genetic and phenotypic response. The ability to reproduce OC plasticity and dissemination by modulating both biochemical and mechanical cues makes these advanced in vitro models highly promising tools that open new perspectives in biomedical and clinical research.The Pathophysiology of Ovarian Cancers2.1. The OC Environment 2.1.1. The Peritoneal Cavity and the Accumulation of Ascites Epithelial ovarian cancer (EOC) represents more than 90% of diagnosed OC cases, classified into five subtypes: high-grade and low-grade serous carcinoma (HGSOC and LGSOC, respectively), mucinous carcinoma, endometrioid carcinoma, and clear cell carcinoma[1,7]. Because HGSOC is the most predominant form of EOC, it is the focus of this review. Although the biological origin of the disease is not clear, it is suspected to originate from fallopian tumors that rapidly metastasize within the ovaries [8-10], or from incorrect wound repair after ovulation and inflammation[11]. The ovaries are covered by the germinal epithelium of Waldeyer, which is composed of one layer of squamous-to-cuboidal epithelial cells. This epithelium delimitates the borders of the ovaries, within the peritoneal cavity, from the same embryonic origin as the peritoneal mesothelium delimitating the coelomic cavity (Figure 1).First, we describe the pathophysiology of OCs with emphasis on the heterogeneities from genetic to phenotypic scales. Then, we focus on the interactions of cells with their microenvironment, including the scaffold and the circulating microenvironment. The latter plays a fundamental role in OC because of the build-up of fluid in the peritoneal cavity, the ascites, which contains a variety of cellular and acellular components. This fluid, known to contribute to patient morbidity and mortality by facilitating metastasis and contributing to chemoresistance, has become a hallmark of OC[6]. It is thus of particular importance to integrate fluids in in vitro models, both in terms of composition and shear stress. After a survey of 3D cultures and spheroid models, we focus on the integration of mechanotransduction in in vitro models as generated by the shear stress from surrounding ascites. In particular, we discuss the ability of microfluidics to model the peritoneal cavity with the associated mechanical constraints, enabling control of the shear stress to quantitatively monitor the cell genetic and phenotypic response. The ability to reproduce OC plasticity and dissemination by modulating both biochemical and mechanical cues makes these advanced in vitro models highly promising tools that open new perspectives in biomedical and clinical research.The Pathophysiology of Ovarian CancersThe OC EnvironmentThe Peritoneal Cavity and the Accumulation of AscitesEpithelial ovarian cancer (EOC) represents more than 90% of diagnosed OC cases, classified into five subtypes: high-grade and low-grade serous carcinoma (HGSOC and LGSOC, respectively), mucinous carcinoma, endometrioid carcinoma, and clear cell carcinoma[1,7]. Because HGSOC is the most predominant form of EOC, it is the focus of this review. Although the biological origin of the disease is not clear, it is suspected to originate from fallopian tumors that rapidly metastasize within the ovaries [8-10], or from incorrect wound repair after ovulation and inflammation[11]. The ovaries are covered by the germinal epithelium of Waldeyer, which is composed of one layer of squamous-tocuboidal epithelial cells. This epithelium delimitates the borders of the ovaries, within the peritoneal cavity, from the same embryonic origin as the peritoneal mesothelium delimitating the coelomic cavity (Figure 1).
# Introduction
Ovarian cancer (OC) is associated with difficult detection, poor prognosis, and high probability of relapse after surgery or chemotherapy, due to late-stage diagnosis [bib_ref] Advances in Ovarian Cancer Therapy, Cortez [/bib_ref] [bib_ref] Treatment of Epithelial Ovarian Cancer, Kuroki [/bib_ref]. This raises several challenges: (i) understanding the biological origin of the disease, (ii) identifying the relevant markers of tumor development, (iii) mapping the phenotypic heterogeneities within primary tumor and metastases, and (iv) controlling the interplay between cells and their 3D microenvironment. Developing biologically relevant in vitro models ahead of preclinical models is a milestone to develop new strategies to tackle these clinical challenges [bib_ref] Preclinical Models of Ovarian Cancer: Pathogenesis, Problems, and Implications for Prevention BACKGROUND-Ovarian..., Karnezis [/bib_ref]. Promising studies have started to emerge and OCs have greatly benefited from recent advances in various fields such as single-cell analysis, 3D culture, and microfluidics. Those innovative models must integrate multiple parameters to recapitulate the progression of the pathological context, which include the multiple cellular populations as well as both the mechanical and biochemical cues. In this review, we detail these parameters by associating the advanced technologies developed for their reconstruction in vitro.
First, we describe the pathophysiology of OCs with emphasis on the heterogeneities from genetic to phenotypic scales. Then, we focus on the interactions of cells with their microenvironment, including the scaffold and the circulating microenvironment. The latter plays a fundamental role in OC because of the build-up of fluid in the peritoneal cavity, the ascites, which contains a variety of cellular and acellular components. This . Scheme of the peritoneal cavity, which contains the ovaries and is surrounded by the parietal and visceral leaflets [bib_ref] The Untapped Potential of Ascites in Ovarian Cancer Research and Treatment, Ford [/bib_ref]. The abnormal increase in ascites volume lines with enrichment in pro-inflammatory cytokines, circulating nucleic acids, and various cell types, as well as proteins from the extracellular matrix. Reproduced with permission of Springer Nature.
The peritoneal cavity contains a thin film of peritoneal fluid composed of water, electrolytes, and other substances derived from the interstitial fluid of adjacent tissues that ensures the harmless movements of the surrounding organs. In a pathological context, an inflammatory liquid-called ascites-accumulates in the cavity due to decreased lymphatic drainage, lymphatic obstruction, and increased vascular permeability [bib_ref] Pathogenesis of Malignant Ascites Formation: Initiating Events That Lead to Fluid Accumulation, Nagy [/bib_ref]. The early presence of malignant ascites has been shown to generally match with poor prognosis, high chemoresistance, and aggressive metastasis [bib_ref] Ascites as a Predictor of Ovarian Malignancy, Shen-Gunther [/bib_ref] [bib_ref] Meeting the Challenge of Ascites in Ovarian Cancer: New Avenues for Therapy..., Kipps [/bib_ref]. It is composed of cancer and non-cancer cells, cell-free DNA, and numerous signaling molecules, as well as extracellular matrix (ECM) proteins and proteases. Additionally, exosomes play a key role in promoting the metastatic niche in this environment [bib_ref] Exosomes Promote Pre-Metastatic Niche Formation in Ovarian Cancer, Feng [/bib_ref].
## The cellular environment
Ovarian primary tumor tissue gathers diverse types of cells, where only a subgroup is defined as malignant. EOC cells are first players, together with other cell types including cancer-associated fibroblasts (CAFs). Cancer cells produce TGF-β, which stimulates interactions with CAFs. CAFs are stromal cells, which contribute to ECM synthesis and remodeling from physiological to pathological environments. This ability of CAFs enables a pro-tumoral activity favoring the metastatic niche [bib_ref] A Tense Situation: Forcing Tumour Progression, Butcher [/bib_ref]. The combination of CAFs ECM remodeling and pro-inflammatory molecule secretion (cytokines, chemokines, growth factors) is responsible for diverse effects from matrix mechanical properties on growth, invasion, and angiogenesis during tumorigenesis [bib_ref] Use of a Novel 3D Culture Model to Elucidate the Role of..., Kenny [/bib_ref] [bib_ref] Cancer-Associated Fibroblasts Are Activated in Incipient Neoplasia to Orchestrate Tumor-Promoting Inflammation in..., Erez [/bib_ref].
Immune cells also play major roles in the tumor microenvironment. In particular, T-cells and tumor-associated macrophages (TAMs) are crucial actors in tumor development. They play a strong immuno-modulating role by producing a large number of cytokines, chemokines, and growth factors. They can hijack the immune response to preserve the tumor [bib_ref] Tumor Derived UBR5 Promotes Ovarian Cancer Growth and Metastasis through Inducing Immunosuppressive..., Song [/bib_ref] [bib_ref] M2-like Macrophages Dictate Clinically Relevant Immunosuppression in Metastatic Ovarian Cancer, Hensler [/bib_ref] , promote tumor development [bib_ref] Ovarian Cancer Stem Cells and Macrophages Reciprocally Interact through the WNT Pathway..., Raghavan [/bib_ref] , and even increase metastasis [bib_ref] Omental Macrophages Secrete Chemokine Ligands That Promote Ovarian Cancer Colonization of the..., Krishnan [/bib_ref] [bib_ref] Alternatively-Activated Macrophages Upregulate Mesothelial Expression of p-Selectin to Enhance Adhesion of Ovarian..., Carroll [/bib_ref]. In this review, we do not describe in detail the role of immune cells and rather focus on the impact of cytokine secretions in the surrounding fluid in ovarian cancer. Finally, adipocytes have also been shown to play a role in the tumor environment, acting in metabolite transfer and supporting tumor growth [bib_ref] Adipocytes Promote Ovarian Cancer Metastasis and Provide Energy for Rapid Tumor Growth, Nieman [/bib_ref] [bib_ref] Adipose Tissue and Adipocytes Supports Tumorigenesis and Metastasis # NIH Public Access..., Nieman [/bib_ref].
## Heterogeneities in oc cells: from genetics to signaling pathways
In about 10% of cases, OC has a genetic cause. Intensive work has been carried out over the past ten years to screen for genetic mismatches in cancer, including in OC [bib_ref] Major Oncogenes and Tumor Suppressor Genes Involved in Epithelial Ovarian Cancer (Review), Aunoble [/bib_ref] [bib_ref] A Single-Cell Landscape of High-Grade Serous Ovarian Cancer, Izar [/bib_ref]. One of the extensively studied tumor suppressor genes-TP53-is known to be associated with HGSOC in 96% of cases, and mutations in BRCA1/2 increase the risk of developing ovarian cancer by 10 to 25 times. BRCA1/2 play multiple and unique roles in homologous recombination repair. More than 200 different mutations of the BRCA1 gene and 80 of the BRCA2 gene have been listed, 80% of which lead to a non-functional, absent, or truncated protein. Other prevalent genetic mutations implicated in HGSOCs include BRCA1 (12%), BRCA2 (11%), KRAS (6%), NF1 (4%), CDK12 (3%), PIK3CA (2%), and BRAF (2%) [bib_ref] The COSMIC (Catalogue of Somatic Mutations in Cancer) Database and Website, Bamford [/bib_ref]. Although some of these mutations are rare, they act as important drivers of HGSOCs. Beyond their role in disease initiation or progression, more than 25 oncogenes have been detected to promote drug resistance [bib_ref] Oncogenes Associated with Drug Resistance in Ovarian Cancer, Liu [/bib_ref]. It is also important to mention the impact of epigenetics in the initiation of OC.
Most of these mutations directly affect the signaling pathways by modulating the expression of cytokines, including growth factors, and of their receptors through which the signal is transduced [bib_ref] Fluid Shear Stress-Induced IL-8/CXCR Signaling in Human Ovarian Cancer Cells, Sun [/bib_ref] [bib_ref] CXCR7 Signaling Induced Epithelial-Mesenchymal Transition by AKT and ERK Pathways in Epithelial..., Yu [/bib_ref]. Among these, TGF-β is involved in cell proliferation, apoptosis, adhesion, invasion, and angiogenesis. As such, the dysregulation of this pathway results in OC plasticity and leads to OC progression [bib_ref] Roles of Other TGFβ Superfamily Members in Cancer, Wakefield [/bib_ref] [bib_ref] TGFβ Signaling Networks in Ovarian Cancer Progression and Plasticity, Kumari [/bib_ref].
## Phenotypic heterogeneities and the epithelial-to-mesenchymal transition 2.3.1. the epithelial-to-mesenchymal transition
The epithelial-to-mesenchymal transition (EMT) is a morphogenetic process of cellular plasticity that epithelial cells undergo during embryonic development and during several pathophysiological processes such as wound healing, fibrosis, and cancer. During this transition, epithelial cells lose some of their characteristics while gaining mesenchymal features. Among many, epithelial cells present altered cell-matrix interactions due to an associated switch in the population of integrins at the surface of transitioning cells. Similarly, cadherins are also affected on the cell surface, with the up-and downregulation of the N-and E-cadherin expressions, respectively [bib_ref] Tumor 1 (WT1) Promotes Ovarian Cancer Progression by Regulating E-Cadherin and ERK1/2..., Han [/bib_ref] [bib_ref] MiR-9 Accelerates Epithelial-Mesenchymal Transition of Ovarian Cancer Cells via Inhibiting e-Cadherin, Sui [/bib_ref] [bib_ref] Effects of UPF1 Expression on EMT Process by Targeting E-Cadherin, N-Cadherin, Vimentin..., Li [/bib_ref] , and the reduction in Ep-CAM expression. Additionally, E-cadherin downregulation combined with occludin can lead to the weakening of tight junctions. This has been proposed to end with the detachment of cells from the ovarian epithelium, involving a role in the dissemination of multicellular aggregates. During EMT, the internal cell structure is also remodeled [bib_ref] The Morphological and Molecular Features of the Epithelial-to-Mesenchymal Transition, Moreno-Bueno [/bib_ref] [bib_ref] Cell Stiffness Is a Biomarker of the Metastatic Potential of Ovarian Cancer..., Xu [/bib_ref]. This is illustrated by the change in the expression of α-smooth muscle actin and the nuclear localization of β-catenin. Another illustration of the thorough reorganization of the cytoskeleton is the upregulation of vimentin, switching from a low and perinuclear distribution to a high cortical distribution [bib_ref] Circulating Tumour Cells Escape from EpCAM-Based Detection Due to Epithelial-to-Mesenchymal Transition, Gorges [/bib_ref] [bib_ref] Vimentin Contributes to Epithelial-Mesenchymal Transition Cancer Cell Mechanics by Mediating Cytoskeletal Organization..., Liu [/bib_ref]. This transitioning process is not binary, but rather includes a continuum associated with great phenotypic evolutions, well-illustrated by the term "Epithelial-Mesenchymal Plasticity" (EMP) [bib_ref] Guidelines and Definitions for Research on Epithelial-Mesenchymal Transition, Yang [/bib_ref]. This makes cells capable of escaping anoikis, a mechanism of cellular death induced by the loss of anchorage.
Ovarian cells are very EMT-active and display an important cellular heterogeneity, as observed within the Waldeyer epithelium after ovulation [bib_ref] Ascites-Induced Shift along Epithelial-Mesenchymal Spectrum in Ovarian Cancer Cells: Enhancement of Their..., Carduner [/bib_ref] [bib_ref] Epithelial-mesenchymal Transition Spectrum Quantification and Its Efficacy in Deciphering Survival and Drug..., Tan [/bib_ref]. Indeed, cells from the outer layers of the ovary dedifferentiate and proliferate to create new stroma and epithelium, which fill the damaged area. In a pathological context, this transition has been shown to be partial because of the lack of the classical transcription factors that are typically observed in wounded areas [bib_ref] Single-Cell EMT-Related Transcriptional Analysis Revealed Intra-Cluster Heterogeneity of Tumor Cell Clusters in..., Kan [/bib_ref]. EOC cells use this innate ability when exposed to promoters such as CAF-secreted TGF-β to detach from the ovaries or fallopian tubes and adopt a floating regime within the ascites [fig_ref] Figure 2: Figure 2 [/fig_ref] [bib_ref] TGFβ Signaling Networks in Ovarian Cancer Progression and Plasticity, Kumari [/bib_ref] [bib_ref] The Morphological and Molecular Features of the Epithelial-to-Mesenchymal Transition, Moreno-Bueno [/bib_ref] [bib_ref] The Role of Epithelial-to-Mesenchymal Plasticity in Ovarian Cancer Progression and Therapy Resistance, Loret [/bib_ref]. Interestingly, the response of cells to TGF-β strongly depends on the cellular subtype and EOC grade [bib_ref] TGF-Beta Induces Serous Borderline Ovarian Tumor Cell Invasion by Activating EMT but..., Cheng [/bib_ref]. Finally, several articles suggest that this rapid transition can be reversed (the mesenchymal-to-epithelial transition-MET) with cells acquiring epithelial features with strong cell-cell interactions directly on the ovaries [7,51].
## The role of cancer stem cells in emt
Cancer stem cells (CSCs) are prone to metastatic behaviors and capable of selfrenewal [bib_ref] Cell Plasticity and Heterogeneity in Cancer, Marjanovic [/bib_ref] [bib_ref] Biomechanical Profile of Cancer Stem-like/Tumor Initiating Cells Derived from a Progressive Ovarian..., Babahosseini [/bib_ref]. It is not known whether this population originates from EOC cells in the course of EMT, or if CSCs appear stochastically with a mesenchymal phenotype to drive tumor growth. Some studies indicate that CSCs possess epithelial characteristics, which supports the first hypothesis [bib_ref] Transformation of Epithelial Ovarian Cancer Stemlike Cells into Mesenchymal Lineage via EMT..., Jiang [/bib_ref] [bib_ref] Mukhopadhyay, A. Aggressive Serous Epithelial Ovarian Cancer Is Potentially Propagated by EpCAM..., Akhter [/bib_ref]. Moreover, these cells are shown to have improved capacities for spheroid formation and anoikis resistance [bib_ref] The FZD7-TWIST1 Axis Is Responsible for Anoikis Resistance and Tumorigenesis in Ovarian..., Tan [/bib_ref] , but they still lack invasive properties to be considered as the sole actors of the metastatic process [bib_ref] Functional Isolation of Tumor-Initiating Cells Using Microfluidic-Based Migration Identifies Phosphatidylserine Decarboxylase as..., Chen [/bib_ref]. Additionally, their phenotype is different from both primary and metastatic constructs [bib_ref] Heterotypic CAF-Tumor Spheroids Promote Early Peritoneal Metastatis of Ovarian Cancer, Gao [/bib_ref]. This suggests that they are the result of several steps of back-and-forth EMT transitions.
The ascitic population is known to house an important proportion of CSCs as compared to tumor tissues [bib_ref] Transformation of Epithelial Ovarian Cancer Stemlike Cells into Mesenchymal Lineage via EMT..., Jiang [/bib_ref]. In this pathological fluid, they are identified as key actors, being involved in the survival of floating EOC cells, either as single cells or in aggregates [bib_ref] Identification and Characterization of Ovarian Cancer-Initiating Cells from Primary Human Tumors, Zhang [/bib_ref] [bib_ref] Metastasis Model of Cancer Stem Cell-Derived Tumors, Mansour [/bib_ref] [bib_ref] Sialyl Lewisx-P-Selectin Cascade Mediates Tumor-Mesothelial Adhesion in Ascitic Fluid Shear Flow, Li [/bib_ref] [bib_ref] Multicellular Detachment Generates Metastatic Spheroids during Intra-Abdominal Dissemination in Epithelial Ovarian Cancer, Al Habyan [/bib_ref]. CSCs act as leader cells that drive the evolution of a whole group, although this population is composed of heterogeneous EOC cells [bib_ref] Keratin-14 (KRT14) Positive Leader Cells Mediate Mesothelial Clearance and Invasion by Ovarian..., Bilandzic [/bib_ref]. Biomolecules 2023, 13, x 5 of 30 [fig_ref] Figure 2: Figure 2 [/fig_ref]. EMT events in ovarian cancer hematogenous metastasis. In Step 1, lesions with characteristic alterations in TP53 develop into cancers in the fallopian tube and the ovaries. OC cells detach and shed into the peritoneal fluid for transcoelomic spread or enter the blood vessels leading to hematogenous metastasis. In Step 2, OC cells in the ascites show high heterogeneity along the EMT continuum, forming anoikis-resistant cell aggregates. Ascites facilitates cell aggregate adhesions to the peritoneal membrane. Such adhesions in Step 3 can undergo MET (reverse EMT), enabling the cells to establish and grow at secondary sites. At the peritoneal interface, cancer cells invade peritoneal mesothelial cells facilitated by integrins and TGF-β, developing secondary tumors and metastasis [bib_ref] TGFβ Signaling Networks in Ovarian Cancer Progression and Plasticity, Kumari [/bib_ref]. Reproduced with permission of Springer Nature.
## The role of cancer stem cells in emt
Cancer stem cells (CSCs) are prone to metastatic behaviors and capable of self-renewal [bib_ref] Cell Plasticity and Heterogeneity in Cancer, Marjanovic [/bib_ref] [bib_ref] Biomechanical Profile of Cancer Stem-like/Tumor Initiating Cells Derived from a Progressive Ovarian..., Babahosseini [/bib_ref]. It is not known whether this population originates from EOC cells in the course of EMT, or if CSCs appear stochastically with a mesenchymal phenotype to drive tumor growth. Some studies indicate that CSCs possess epithelial characteristics, which supports the first hypothesis [bib_ref] Transformation of Epithelial Ovarian Cancer Stemlike Cells into Mesenchymal Lineage via EMT..., Jiang [/bib_ref] [bib_ref] Mukhopadhyay, A. Aggressive Serous Epithelial Ovarian Cancer Is Potentially Propagated by EpCAM..., Akhter [/bib_ref]. Moreover, these cells are shown to have improved capacities for spheroid formation and anoikis resistance [bib_ref] The FZD7-TWIST1 Axis Is Responsible for Anoikis Resistance and Tumorigenesis in Ovarian..., Tan [/bib_ref] , but they still lack invasive properties to be considered as the sole actors of the metastatic process [bib_ref] Functional Isolation of Tumor-Initiating Cells Using Microfluidic-Based Migration Identifies Phosphatidylserine Decarboxylase as..., Chen [/bib_ref]. Additionally, their phenotype is different from both primary and metastatic constructs [bib_ref] Heterotypic CAF-Tumor Spheroids Promote Early Peritoneal Metastatis of Ovarian Cancer, Gao [/bib_ref]. This suggests that they are the result of several steps of back-and-forth EMT transitions.
The ascitic population is known to house an important proportion of CSCs as compared to tumor tissues [bib_ref] Transformation of Epithelial Ovarian Cancer Stemlike Cells into Mesenchymal Lineage via EMT..., Jiang [/bib_ref]. In this pathological fluid, they are identified as key actors, being involved in the survival of floating EOC cells, either as single cells or in aggregates EMT events in ovarian cancer hematogenous metastasis. In Step 1, lesions with characteristic alterations in TP53 develop into cancers in the fallopian tube and the ovaries. OC cells detach and shed into the peritoneal fluid for transcoelomic spread or enter the blood vessels leading to hematogenous metastasis. In Step 2, OC cells in the ascites show high heterogeneity along the EMT continuum, forming anoikis-resistant cell aggregates. Ascites facilitates cell aggregate adhesions to the peritoneal membrane. Such adhesions in Step 3 can undergo MET (reverse EMT), enabling the cells to establish and grow at secondary sites. At the peritoneal interface, cancer cells invade peritoneal mesothelial cells facilitated by integrins and TGF-β, developing secondary tumors and metastasis [bib_ref] TGFβ Signaling Networks in Ovarian Cancer Progression and Plasticity, Kumari [/bib_ref]. Reproduced with permission of Springer Nature.
EMT is thus intrinsically correlated to cellular aggregation, where a balance between retaining epithelial characteristics and developing mesenchymal traits is needed to ensure the survival of a multicellular aggregate and its metastatic outcomes [bib_ref] Cell Plasticity and Metastasis, Chaffer [/bib_ref]. The non-aggregative population of EOC cells is genetically heterogeneous and cohabits with other cell types within the peritoneal cavity [bib_ref] A Single-Cell Landscape of High-Grade Serous Ovarian Cancer, Izar [/bib_ref]. Because of their ability to undergo EMT and to detach from their anchorage, EOC cells adopt a new phenotype that allows them to survive without tight cell-cell junctions. This switch in their cadherin expression also tends to promote their aggregation, as cells that achieve partial EMT often express a variety of cadherins that enable temporary cell-cell junctions [bib_ref] Single-Cell EMT-Related Transcriptional Analysis Revealed Intra-Cluster Heterogeneity of Tumor Cell Clusters in..., Kan [/bib_ref]. Consequently, these cells are likely to form multicellular spheroid constructs. Spheroids are formed by self-assembly of cells through different types of intercellular interactions, including cadherin-and integrin-mediated interactions, as well as involving the vitronectin/αv integrin adhesive system. Spheroids contribute to protect cells from the stressful microenvironment, accounting for the resistance to anoikis of cells within spheroids [bib_ref] Cell Cycle Arrest or Survival Signaling through Av Integrins, Activation of PKC..., Carduner [/bib_ref]. In vivo tumor spheroids are also characterized by the presence within these multicellular clusters of non-cancerous and non-ovarian cells. Matte et al. described the presence of mesothelial cells in the center of the spheroids (in vivo and in vitro), which makes the spheroids more compact [bib_ref] Mesothelial Cells Interact with Tumor Cells for the Formation of Ovarian Cancer..., Matte [/bib_ref]. CAF-tumor spheroids have also been recently reported to promote early peritoneal metastasis of OC [bib_ref] Sialyl Lewisx-P-Selectin Cascade Mediates Tumor-Mesothelial Adhesion in Ascitic Fluid Shear Flow, Li [/bib_ref]. Interestingly, a study from Al Habyan et al. showed that most of the spheroids are likely to come from collective detachment from the ovaries, with little aggregation or proliferation within the ascites [bib_ref] Multicellular Detachment Generates Metastatic Spheroids during Intra-Abdominal Dissemination in Epithelial Ovarian Cancer, Al Habyan [/bib_ref]. Besides, numerous articles have shown that spheroids are more prone to resist chemotherapies, such as carboplatin and paclitaxel [bib_ref] Personalized Medicine-Based Approach to Model Patterns of Chemoresistance and Tumor Recurrence Using..., Raghavan [/bib_ref] [bib_ref] Ovarian Cancer Multicellular Spheroids, Platelet Releasate Promotes Growth, Expansion of Aldh+ and..., Casagrande [/bib_ref] , with a mechanism that remains unknown. Such organization has hence been suspected of modifying gene expression and increasing aggressiveness [bib_ref] Single-Cell EMT-Related Transcriptional Analysis Revealed Intra-Cluster Heterogeneity of Tumor Cell Clusters in..., Kan [/bib_ref] [bib_ref] Ovarian Cancer Spheroid Cells with Stem Cell-like Properties Contribute to Tumor Generation,..., Liao [/bib_ref]. However, some argue that spheroids favor the development of CSCs, and that only the latter would be chemoresistant, giving the colony the ability to rapidly regenerate after a conventional drug treatment [bib_ref] Ovarian Cancer Spheroid Cells with Stem Cell-like Properties Contribute to Tumor Generation,..., Liao [/bib_ref]. Still, spheroids seem to be the preferential mode of survival of floating EOC cells, as well as the starting point for metastatic activity.
## The mechanical microenvironment of oc: from scaffold to circulating environment
The mechanical microenvironment regulates ovarian cancer cell morphology, migration, and spheroid disaggregation [bib_ref] The Mechanical Microenvironment Regulates Ovarian Cancer Cell Morphology, Migration, and Spheroid Disaggregation, Mckenzie [/bib_ref]. Indeed, in addition to the interplay between the different signaling pathways, there are interactions with the scaffold and the circulating environments, namely the extracellular matrix (ECM) and ascites. Let us first consider the role of ECM in EOC.
## The extracellular matrix and its mechanosensing
The ECM is a key player in regulating cell migration, differentiation, and proliferation, and is a determinant for the growth and progression of solid tumors. In the ovary, it is made of a variety of molecules including the collagen superfamily, glycoproteins, proteoglycans, and hyaluronan [bib_ref] Extracellular Matrix of Ovarian Tumors, Ricciardelli [/bib_ref]. While the most abundant proteins of the ovarian interstitial matrix include fibrillary collagens (I and III), the basement membrane underlying the ovarian surface epithelium is composed of a dense network of collagen IV and laminin. The ECM is constantly remodeled both in normal and tumor development. Important changes in terms of composition, topology, and stiffness have been reported ex vivo in normal and malignant human ovarian biopsies [bib_ref] Basement Membrane Structures in Tumors of the Ovary, Stenbäck [/bib_ref] [bib_ref] Molecular Events Associated with Dysplastic Morphologic Transformation and Initiation of Ovarian Tumorigenicity, Yang [/bib_ref]. Basement membrane remodeling in pre-malignant ovarian surface epithelium has been shown to mostly impact collagen IV and laminin with transient loss [bib_ref] Loss of Surface and Cyst Epithelial Basement Membranes and Preneoplastic Morphologic Changes..., Roland [/bib_ref] [bib_ref] Alterations of the Extracellular Matrix in Ovarian Cancer Studied by Second Harmonic..., Nadiarnykh [/bib_ref] [bib_ref] Endogenous Optical Biomarkers of Ovarian Cancer Evaluated with Multiphoton Microscopy, Kirkpatrick [/bib_ref]. Note that fibronectin and vitronectin also play a key role in ovarian cancer, notably as circulating components of the ascitic malignant fluid.
Apart from its biochemical composition, the ECM is also key in terms of biophysical properties. This is commonly known as mechanosensing. Mechanosensing and the impact of the ECM on cell behavior have motivated an increasing number of works during the past few years [bib_ref] Mechanobiology of Tumor Invasion: Engineering Meets Oncology, Carey [/bib_ref] [bib_ref] Effects of Extracellular Matrix Viscoelasticity on Cellular Behaviour, Chaudhuri [/bib_ref]. In particular, mechanosensing has been shown to influence many cell processes including tumor activity of EOC. For example, Zhou and colleagues have shown that growing SKOV-3 cells in hydrogel of different stiffness modulates the formation of spheroids, the presence of CSC characteristics, and their resistance to doxorubicin [bib_ref] Enrichment of Ovarian Cancer Stem Cells by PEG Cross-Linked PMVE-Co-MA Hydrogel with..., Zhou [/bib_ref]. They have highlighted a significant stiffness-dependent increase in the expression of several markers such as ALDH1, which is supposed to be involved both in chemoresistance and in differentiation of EOC cells, as well as of CD117, a receptor tyrosine kinase that allows cells to acquire stem-like properties such as self-renewal, and also CD133, which is known to play a role in EMP.
Furthermore, the invasion of EOC cells has been shown to be dependent on the substrate, with a softer substrate promoting more aggressive and faster invasion according to Dikovsky et al. [bib_ref] Defining the Role of Matrix Compliance and Proteolysis in Three-Dimensional Cell Spreading..., Dikovsky [/bib_ref] or McGrail et al. [bib_ref] The Malignancy of Metastatic Ovarian Cancer Cells Is Increased on Soft Matrices..., Mcgrail [/bib_ref] , although still controversial [bib_ref] The Effects of Macroporosity and Stiffness of Poly[(Methyl Vinyl Ether)-: Alt -(Maleic..., Guo [/bib_ref]. McGrail and colleagues explained that SKOV-3 ovarian cancer cells display a more malignant phenotype on polyacrylamide soft substrates (ca. 3 kPa) undergoing an EMT-characteristic morphological elongation. Single-cell motility analysis revealed large increases in migration on soft substrates, as quantified by the calculated cell migration coefficient. In addition, they have used traction force microscopy to quantify the force exerted by cells on the underlying substrate. These experiments have shown that when cultured on soft matrices, OC cells exerted more force than when cultured on stiffer polyacrylamide substrates (ca. 35 kPa,, which is attributed to an increased metastatic phenotype. Moreover, the increased intensity and polarization of phosphorylated myosin light chain (pMLC) indicates that cells were more capable of polarizing these forces on the soft substrates, a crucial step for effective cell migration. These findings show that the mechanical environment is key to determine cancer progression and highlight how crucial it is to identify key parameters to be reproduced for reconstructing a biologically relevant in vitro environment. This has been highlighted by Pearce and co-workers in their profiling of HGSOC metastases [bib_ref] Deconstruction of a Metastatic Tumor Microenvironment Reveals a Common Matrix Response in..., Pearce [/bib_ref]. They integrated gene expression, matrisome proteomics, ECM organization, biomechanical properties, and cytokine and chemokine levels all from the same sample. By doing so, they attempted to predict the extent of disease, while also revealing the dynamic nature of matrisome remodeling during tumor development. Further studies are probably needed to better identify the range of mechanical forces involved and to always increase the level of complexity by integrating the interaction with the topology of the environment, the biomolecular composition, without neglecting the circulating environment. These findings show that the mechanical environment is key to determine cancer progression and highlight how crucial it is to identify key parameters to be reproduced for reconstructing a biologically relevant in vitro environment. This has been highlighted by Pearce and co-workers in their profiling of HGSOC metastases [bib_ref] Deconstruction of a Metastatic Tumor Microenvironment Reveals a Common Matrix Response in..., Pearce [/bib_ref]. They integrated gene expression, matrisome proteomics, ECM organization, biomechanical properties, and cytokine and chemokine levels all from the same sample. By doing so, they attempted to predict the extent of disease, while also revealing the dynamic nature of matrisome remodeling during tumor development. Further studies are probably needed to better identify the range of mechanical forces involved and to always increase the level of complexity by integrating the interaction with the topology of the environment, the biomolecular composition, without neglecting the circulating environment.
## The importance of fluids in oc: soluble factors and shear stress
Ascites acts as a regulator of cell-ECM interactions, exposing cells to a high concentration of ECM proteins. In particular, integrin α5β1 and αvβ3 ligands are believed to be a preferential path of adhesion for EOC cells to the ascites fibronectin and vitronectin, respectively [bib_ref] Targeting Constitutively Activated B1 Integrins Inhibits Prostate Cancer Metastasis, Lee [/bib_ref] [bib_ref] Ovarian Cancer Ascites-Derived Vitronectin and Fibronectin: Combined Purification, Molecular Features and Effects..., Carduner [/bib_ref]. We and others have previously shown that mesothelial secretions, including vitronectin, promote the invasion of OC cells [bib_ref] Ovarian Cancer Ascites-Derived Vitronectin and Fibronectin: Combined Purification, Molecular Features and Effects..., Carduner [/bib_ref] [bib_ref] Malignant Ascites Determine the Transmesothelial Invasion of Ovarian Cancer Cells, Mikuła-Pietrasik [/bib_ref] [bib_ref] Ascites from Ovarian Cancer Patients Stimulates MUC16 Mucin Expression and Secretion in..., Matte [/bib_ref] [bib_ref] Mesothelial Vitronectin Stimulates Migration of Ovarian Cancer Cells, Heyman [/bib_ref]. However, these findings are still controversial [bib_ref] Patient-Derived and Artificial Ascites Have Minor Effects on MeT-5A Mesothelial Cells and..., Estermann [/bib_ref] , which can partly be attributed to the diversity of cell lines and the heterogeneity in patient ascites. This highlights the fact that further studies have to be conducted on the interactions between invasive cells and proteins in the ascites.
Biophysical inputs are also at play in the circulating environment. Indeed, Bascetin et al. have shown that the macromolecular crowding (MMC) of the ascites microenvironment impacts OC cell phenotype [bib_ref] A Biomimetic Model of 3D Fluid Extracellular Macromolecular Crowding Microenvironment Fine-Tunes Ovarian..., Bascetin [/bib_ref]. In their recent work, they used two inert crowders, Ficoll 400 kDa and Dextran 250 kDa, to mimic the MMC and the estimated total protein concentration in OC ascites. They notably looked at the effect of extracellular MMC on actin organization of two different OC cell lines. Without MMC, SKOV3 and IGROV1 cells were well spread, with peripheral cortical actin and/or central stress fibers. In contrast, in the presence of crowders, cell spreading significantly decreased with much reduced cortical actin and stress fibers. In this line, ascites viscosity was previously shown to be a marker of the degree of cancer malignancy and correlated with metastasis speed and with the concentration of floating cells [bib_ref] The Role of Ascitic Fluid Viscosity in the Differential Diagnosis of Ascites, Gokturk [/bib_ref]. This illustrates the importance of ascites MMC and viscosity when designing a relevant in vitro model of the peritoneal cavity. Both can indeed decrease the metastatic potential, as crowding can inhibit spheroid formation and viscosity can reduce the invasion rate due to limited diffusion.
Biomolecules 2023, 13, x 9 of 30 microenvironment impacts OC cell phenotype [bib_ref] A Biomimetic Model of 3D Fluid Extracellular Macromolecular Crowding Microenvironment Fine-Tunes Ovarian..., Bascetin [/bib_ref]. In their recent work, they used two inert crowders, Ficoll 400 kDa and Dextran 250 kDa, to mimic the MMC and the estimated total protein concentration in OC ascites. They notably looked at the effect of extracellular MMC on actin organization of two different OC cell lines. Without MMC, SKOV3 and IGROV1 cells were well spread, with peripheral cortical actin and/or central stress fibers. In contrast, in the presence of crowders, cell spreading significantly decreased with much reduced cortical actin and stress fibers. In this line, ascites viscosity was previously shown to be a marker of the degree of cancer malignancy and correlated with metastasis speed and with the concentration of floating cells [bib_ref] The Role of Ascitic Fluid Viscosity in the Differential Diagnosis of Ascites, Gokturk [/bib_ref]. This illustrates the importance of ascites MMC and viscosity when designing a relevant in vitro model of the peritoneal cavity. Both can indeed decrease the metastatic potential, as crowding can inhibit spheroid formation and viscosity can reduce the invasion rate due to limited diffusion. Specifically, ovarian cells are exposed to a continuous fluid shear stress imposed by diaphragm and organ movements, as well as ascites build-up in advanced OC stages. This makes shear stress dependent on the increase in ascites volume in the peritoneal cavity of each individual patient. An interesting study by Klymenko and colleagues evidenced that Specifically, ovarian cells are exposed to a continuous fluid shear stress imposed by diaphragm and organ movements, as well as ascites build-up in advanced OC stages. This makes shear stress dependent on the increase in ascites volume in the peritoneal cavity of each individual patient. An interesting study by Klymenko and colleagues evidenced that the increase in ascites volume triggers a dramatic increase in the intraperitoneal pressure that EOC cells are subjected to, which has few effects on their proliferation, but modifies their metastatic abilities, notably their cadherins expression [bib_ref] Modeling the Effect of Ascites-Induced Compression on Ovarian Cancer Multicellular Aggregates, Klymenko [/bib_ref]. This also points out the impact of shear stress as a potential inducer of tumor cell cycle modification or arrest. Chang et al. found that shear stress of 12 dyne/cm 2 led to a G2/M arrest in four lines of cancer cells, as opposed to their control in static conditions [bib_ref] Tumor Cell Cycle Arrest Induced by Shear Stress: Roles of Integrins and..., Chang [/bib_ref]. In a systematic study, they characterized the mechanisms by which shear stress regulates cell cycle in tumor cells, proposing specific roles to integrins and Smad proteins. They hypothesized that the in vivo behavior of these cells could result in the preferential invasion of areas with smaller, laminar shear stress. However, a precise magnitude of shear stress within the peritoneal cavity is still difficult to obtain. Most of the articles now agree on the upper barrier [bib_ref] Fluid Shear Stress Impacts Ovarian Cancer Cell Viability, Subcellular Organization, and Promotes..., Hyler [/bib_ref] [bib_ref] A Dynamic Culture Method to Produce Ovarian Cancer Spheroids under Physiologically-Relevant Shear, Masiello [/bib_ref] [bib_ref] Stemness and Chemoresistance in Epithelial Ovarian Carcinoma Cells under Shear Stress, Ip [/bib_ref] , which could be placed at some units to 10 dyne/cm 2 .
Modeling and integrating key parameters of the cell microenvironment is a crucial step to foster progress in biological and biomedical research, and particularly here for investigating cell EMP, migration, and metastasis progression. In the last part of this review, we will discuss how the different characteristics of the scaffold, which are all known to affect cells, must be controlled and reproduced using biomaterials approach for designing in vitro models. In addition to these crucial aspects, we will also discuss the need for integrating co-culture, 3D models, and the fluid microenvironment.
## Reconstructing oc in its multidimensional environment
New tools are constantly being developed to always improve the biological relevance of in vitro models. This comprises 3D culture setups and co-cultures, which represent a great improvement as compared to traditional 2D culture models. Recent smart culture setups also include hydrogel seeding, multilayered constructs, droplet culture, and spheroid growth. Additionally, microfluidics and additive fabrication have also come across to enrich the battery of possibilities, easing the engineering and vascularization of 3D culture models.
All these tools and strategies should be used in a complementary fashion to reach the highest possible integration level. This is illustrated in [fig_ref] Figure 5: Scheme of the main steps of the OC metastasis process [/fig_ref] , which highlights the necessity to increase the integration of the in vitro model, starting from ECM models, 3D cellular models, and recapitulating the circulating environment, in terms of composition, dynamics, and shear stress. The tools developed in the past decade to tackle these issues are discussed in the following sections.
## Co-culture, which cells?
Important research efforts have been dedicated to ovarian carcinomas, either to provide a reliable model on which screening assays could be conducted for drug development [bib_ref] Quantitative High Throughput Screening Using a Primary Human Three-Dimensional Organotypic Culture Predicts..., Kenny [/bib_ref] , or to build up an integrative model to mimic and better understand OC physiology [bib_ref] Vascular Cell Adhesion Molecule-1 Is a Regulator of Ovarian Cancer Peritoneal Metastasis, Slack-Davis [/bib_ref]. As discussed before, several cell types are involved in OC such as CAFs, mesothelial and EOC cells. EOC cells are first players, widely commercially available for their use in biomedical research, and with significant differences in behavior and expression [bib_ref] Ascites-Induced Shift along Epithelial-Mesenchymal Spectrum in Ovarian Cancer Cells: Enhancement of Their..., Carduner [/bib_ref]. Among those, SKOV3 cells have the advantage of exhibiting an innate plasticity, which makes it a relevant model cell line for EOC [bib_ref] Epithelial-mesenchymal Transition Spectrum Quantification and Its Efficacy in Deciphering Survival and Drug..., Tan [/bib_ref]. One step further, EOC cells' coculture is the first challenge to meet for building in vitro models. In this line, several groups have reported the growth of EOC cells onto a layer of mesothelial cells to assess the influence of mesothelial secretions [bib_ref] Mesothelial Vitronectin Stimulates Migration of Ovarian Cancer Cells, Heyman [/bib_ref] [bib_ref] Mesothelial Cells Induce the Motility of Human Ovarian Carcinoma Cells, Rieppi [/bib_ref] [bib_ref] Ovarian Carcinoma Ascites Spheroids Adhere to Extracellular Matrix Components and Mesothelial Cell..., Burleson [/bib_ref]. ture models.
All these tools and strategies should be used in a complementary fashion to reach the highest possible integration level. This is illustrated in [fig_ref] Figure 5: Scheme of the main steps of the OC metastasis process [/fig_ref] , which highlights the necessity to increase the integration of the in vitro model, starting from ECM models, 3D cellular models, and recapitulating the circulating environment, in terms of composition, dynamics, and shear stress. The tools developed in the past decade to tackle these issues are discussed in the following sections.
## Co-culture, which cells?
Important research efforts have been dedicated to ovarian carcinomas, either to provide a reliable model on which screening assays could be conducted for drug development [bib_ref] Quantitative High Throughput Screening Using a Primary Human Three-Dimensional Organotypic Culture Predicts..., Kenny [/bib_ref] , or to build up an integrative model to mimic and better understand OC Besides mesothelial cells, the influence of adipocytes on EOC metastasis has also been a focus of interest. Indeed, adipocytes have been shown to increase proliferation and chemoresistance of EOC cells, and may constitute a preferential invasion site for floating aggregates [bib_ref] Adipocytes Promote Ovarian Cancer Metastasis and Provide Energy for Rapid Tumor Growth, Nieman [/bib_ref] [bib_ref] The Malignancy of Metastatic Ovarian Cancer Cells Is Increased on Soft Matrices..., Mcgrail [/bib_ref] [bib_ref] Activation of SphK1 by Adipocytes Mediates Epithelial Ovarian Cancer Cell Proliferation, Dai [/bib_ref] [bib_ref] Adipocytes Induce the Resistance of Ovarian Cancer to Carboplatin through ANGPTL4, Zhou [/bib_ref]. On this basis, multi-cellular culture models were developed by Pearce and Balkwill. In their study on the impact of platelets on extracellular matrix production and tissue invasion, they used a 3D model where fibroblasts were plated on top of a gel loaded with adipocytes, followed by a layer of mesothelial cells. This multicellular construct was cultured for 24 h. Then, a tetra-culture construct was obtained by the addition of HGSOC cells. Ultimately, a further level of integration was reached by the addition of fresh isolated platelets.-D reproduces representative confocal images obtained after 7 days. These images show that ECM molecules such as fibronectin (FN1) and versican (VCAN) were present to a higher level on the pentaculture construct. In addition, quantification of EpCAM deposition demonstrates the presence of a higher number of malignant cells in the penta-culture. This was further confirmed by flow cytometry analysis of EpCAM-positive cells. This indicates that platelets stimulate the production of ECM molecules and malignant cell invasion, further associated with poor prognosis.
Immune cells should also be included in co-culture setups to investigate the role of tumor-associated macrophages (TAMs) in tumor growth. Indeed, such works have shown that OC cells and macrophages interact via cell-cell contacts within spheroids, contributing to the switch towards malignant EOC phenotypes [bib_ref] Ovarian Cancer Stem Cells and Macrophages Reciprocally Interact through the WNT Pathway..., Raghavan [/bib_ref]. More generally, this provides a platform to study the deregulation of the immune balance in a pathological environment [bib_ref] 1-Methyl-Tryptophan Attenuates Regulatory T Cells Differentiation Due to the Inhibition of Estrogen-IDO1-MRC2..., Wei [/bib_ref]. Finally, pioneering works were also reported by Lengyel, Kenny, and co-workers at Chicago University with the design of organotypic models of the peritoneal cavity [bib_ref] Major Oncogenes and Tumor Suppressor Genes Involved in Epithelial Ovarian Cancer (Review), Aunoble [/bib_ref] [bib_ref] Modeling the Early Steps of Ovarian Cancer Dissemination in an Organotypic Culture..., Peters [/bib_ref] [bib_ref] Throughput Screening Model of the Tumor Microenvironment for Ovarian Cancer Cell Growth, Lal-Nag [/bib_ref] [bib_ref] Organotypic 3D Models of the Ovarian Cancer Tumor Microenvironment, Watters [/bib_ref]. These are composed of primary human omental fibroblasts mixed with ECM components (fibronectin and type I collagen) and covered with a confluent layer of mesothelial cells, on top of which EOC cells were seeded. These works have allowed a great advance in the description of the impact, and hence in the design, of the properties of the matrix in a cellular model with increased biological relevance. As such, these works are further discussed in Section 3.3. taculture construct. In addition, quantification of EpCAM deposition demonstrates the presence of a higher number of malignant cells in the penta-culture. This was further confirmed by flow cytometry analysis of EpCAM-positive cells. This indicates that platelets stimulate the production of ECM molecules and malignant cell invasion, further associated with poor prognosis. . Data are expressed as mean ± SD (* p < 0.05 and ** p < 0.01; unpaired t-test for (C-E) and paired t-test for (F)). Adapted with permission from.
Co-culture models are hence of high relevance to better reproduce the pathophysiological context, and to better understand the interactions between several cell types, with an increasing consideration for 3D cellular constructs [bib_ref] Comparative Analysis of Cell-Cell Contact Abundance in Ovarian Carcinoma Cells Cultured in..., Kutova [/bib_ref].
## Spheroid models
Many researchers have reported that cell behavior, including OC cells, and response to drugs are different in 2D monolayers and 3D spheroid models [bib_ref] Comparative Analysis of Cell-Cell Contact Abundance in Ovarian Carcinoma Cells Cultured in..., Kutova [/bib_ref] [bib_ref] Carboplatin Sensitivity in Epithelial Ovarian Cancer Cell Lines: The Impact of Model..., Patra [/bib_ref]. This highlights the challenges of choosing the appropriate pre-clinical models for drug testing. As a result, most of the current researches in the field elaborate protocols to grow cells in 3D constructs [bib_ref] Dangles-Marie, V. Spherical Cancer Models in Tumor Biology, Weiswald [/bib_ref]. These techniques involve the seeding of cells on non-adherent substrates such as bovine serum albumin (BSA) [bib_ref] 3-Dimensional Cell Culture for on-Chip Differentiation of Stem Cells in Embryoid Body, Kim [/bib_ref] [bib_ref] On-Chip Anticancer Drug Test of Regular Tumor Spheroids Formed in Microwells by..., Kim [/bib_ref] , polyHEMA [bib_ref] A Dynamic Culture Method to Produce Ovarian Cancer Spheroids under Physiologically-Relevant Shear, Masiello [/bib_ref] [bib_ref] High-Throughput Cancer Cell Sphere Formation for Characterizing the Efficacy of Photo Dynamic..., Chen [/bib_ref] , agarose [bib_ref] Cell Cycle Arrest or Survival Signaling through Av Integrins, Activation of PKC..., Carduner [/bib_ref] [bib_ref] CD44 Regulates Formation of Spheroids and Controls Organ-Specific Metastatic Colonization in Epithelial..., Suarez [/bib_ref] [bib_ref] Fabrication of Micro-Cages and Caged Tumor Spheroids for Microfluidic Chip-Based Assays, He [/bib_ref] , or Pluronic F127 [bib_ref] Configurable 2D and 3D Spheroid Tissue Cultures on Bioengineered Surfaces with Acquisition..., Kuo [/bib_ref] or the use of EOC cell lines known to spontaneously aggregate in given conditions [bib_ref] Ascites-Induced Shift along Epithelial-Mesenchymal Spectrum in Ovarian Cancer Cells: Enhancement of Their..., Carduner [/bib_ref] [bib_ref] Human Platelet Lysates-Based Hydrogels: A Novel Personalized 3D Platform for Spheroid Invasion..., Monteiro [/bib_ref]. Alternatively, spheroids can be produced by using hydrogels [bib_ref] Enrichment of Ovarian Cancer Stem Cells by PEG Cross-Linked PMVE-Co-MA Hydrogel with..., Zhou [/bib_ref] [bib_ref] The Malignancy of Metastatic Ovarian Cancer Cells Is Increased on Soft Matrices..., Mcgrail [/bib_ref] [bib_ref] Bioengineered 3D Platform to Explore Cell-ECM Interactions and Drug Resistance of Epithelial..., Loessner [/bib_ref] [bib_ref] Functionalization, Preparation and Use of Cell-Laden Gelatin Methacryloyl-Based Hydrogels as Modular Tissue..., Loessner [/bib_ref] [bib_ref] Three-Dimensional Culture and Clinical Drug Responses of a Highly Metastatic Human Ovarian..., Song [/bib_ref] [bib_ref] Effect of RGD Content in Poly(Ethylene Glycol)-Crosslinked Poly(Methyl Vinyl Ether-Alt-Maleic Acid) Hydrogels..., Zhou [/bib_ref] , hanging drop culture [bib_ref] Hanging Drop Cell Culture Protocol for Generation of 3D Spheroids, Foty [/bib_ref] [bib_ref] Empirical Chemosensitivity Testing in a Spheroid Model of Ovarian Cancer Using a..., Das [/bib_ref] , or rotating wall culture [bib_ref] Fluid Shear Stress Impacts Ovarian Cancer Cell Viability, Subcellular Organization, and Promotes..., Hyler [/bib_ref] [bib_ref] Short Exposure to Millimolar Concentrations of Ethanol Induces Apoptotic Cell Death in..., Castaneda [/bib_ref] [bib_ref] An Air Bubble-Isolating Rotating Wall Vessel Bioreactor for Improved Spheroid/Organoid Formation, Phelan [/bib_ref] that have all proven their efficiency in the past ten years for the generation of reproducible and stable organoids [bib_ref] Initial Formation of IGROV1 Ovarian Cancer Multicellular Aggregates Involves Vitronectin, Kellouche [/bib_ref] [bib_ref] Evaluation of the Potential of a New Ribavirin Analog Impairing the Dissemination..., Wambecke [/bib_ref]. Alternatively, we have recently reported the development of microfabricated supports to engineer human OC spheroids. We showed that playing with the dimension of the support allows tuning the spheroid size in a controlled and reproducible manner. Single cell approaches have revealed the stemness potential of spheroid cell populations. Those findings are relevant for cancer cell phenotypic heterogeneity, as well as for drug resistance [bib_ref] Single cell-derived spheroids capture the self-renewing subpopulations of metastatic ovarian cancer, Velletri [/bib_ref]. Alessandri and co-workers have investigated the impact of confinement on the internal cellular organization of spheroids by encapsulating cells in an aqueous core enclosed by a hydrogel shell [bib_ref] Cellular Capsules as a Tool for Multicellular Spheroid Production and for Investigating..., Alessandri [/bib_ref]. DAPI staining shows that the nuclei are smaller in confined spheroids [fig_ref] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement [/fig_ref] , leading to a cell density twice as large as in the freely growing spheroids [fig_ref] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement [/fig_ref]. They also examined cell proliferation by staining with KI-67, showing that proliferative phenotypes are homogeneously distributed throughout free spheroids [fig_ref] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement [/fig_ref] , contrary to confined spheroids, where cell division mostly occurs at the periphery [fig_ref] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement [/fig_ref]. Finally, fibronectin was found to be fibrillar and homogeneously distributed in free spheroids, whereas it was restricted to the periphery in confined spheroids [fig_ref] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement [/fig_ref]. One step further, and by performing invasion assays in a collagen matrix, they have reported that peripheral cells readily escape pre-confined spheroids, while cell-cell cohesion is maintained for freely growing spheroids. This suggests that mechanical cues from the surrounding microenvironment may trigger cell invasion from a growing tumor [bib_ref] Cellular Capsules as a Tool for Multicellular Spheroid Production and for Investigating..., Alessandri [/bib_ref]. Single cell approaches have revealed the stemness potential of spheroid cell populations. Those findings are relevant for cancer cell phenotypic heterogeneity, as well as for drug resistance [bib_ref] Single cell-derived spheroids capture the self-renewing subpopulations of metastatic ovarian cancer, Velletri [/bib_ref]. Alessandri and co-workers have investigated the impact of confinement on the internal cellular organization of spheroids by encapsulating cells in an aqueous core enclosed by a hydrogel shell [bib_ref] Cellular Capsules as a Tool for Multicellular Spheroid Production and for Investigating..., Alessandri [/bib_ref]. DAPI staining shows that the nuclei are smaller in confined spheroids [fig_ref] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement [/fig_ref] , leading to a cell density twice as large as in the freely growing spheroids [fig_ref] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement [/fig_ref]. They also examined cell proliferation by staining with KI-67, showing that proliferative phenotypes are homogeneously distributed throughout free spheroids [fig_ref] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement [/fig_ref] , contrary to confined spheroids, where cell division mostly occurs at the periphery [fig_ref] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement [/fig_ref]. Finally, fibronectin was found to be fibrillar and homogeneously distributed in free spheroids, whereas it was restricted to the periphery in confined spheroids [fig_ref] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement [/fig_ref]. One step further, and by performing invasion assays in a collagen matrix, they have reported that peripheral cells readily escape pre-confined spheroids, while cell-cell cohesion is maintained for freely growing spheroids. This suggests that mechanical cues from the surrounding microenvironment may trigger cell invasion from a growing tumor [bib_ref] Cellular Capsules as a Tool for Multicellular Spheroid Production and for Investigating..., Alessandri [/bib_ref].
Finally, Zhou et al. observed a progressive decrease in the growth of their multicellular aggregates when cultured within hydrogel fragments, which can be explained by the limited oxygen and nutrient diffusion in a crowded environment [bib_ref] Enrichment of Ovarian Cancer Stem Cells by PEG Cross-Linked PMVE-Co-MA Hydrogel with..., Zhou [/bib_ref]. These results are an additional illustration of the impact of the matrix and importance of cell-matrix interactions in driving cell behaviors, including cell growth and invasion. This then requires to consider the scaffolding environment of cells to be reconstructed. Finally, Zhou et al. observed a progressive decrease in the growth of their multicellular aggregates when cultured within hydrogel fragments, which can be explained by the limited oxygen and nutrient diffusion in a crowded environment [bib_ref] Enrichment of Ovarian Cancer Stem Cells by PEG Cross-Linked PMVE-Co-MA Hydrogel with..., Zhou [/bib_ref]. These results are an additional illustration of the impact of the matrix and importance of cell-matrix interactions in driving cell behaviors, including cell growth and invasion. This then requires to consider the scaffolding environment of cells to be reconstructed.
## Ecm and scaffold
Among different features, the biochemical composition, topology, and mechanical properties of the ECM scaffold are crucial and must be scrupulously monitored and reproduced. This requires advanced techniques of biofabrication. Polymer science and biomaterials research have greatly improved our ability to engineer biological matrices. Among these, electrospinning, which is an electrically assisted extrusion method, is a promising cost-effective technique to mimic the fibrillar aspect of the ECM, which has been shown to influence cell behavior [bib_ref] Nano-on-Micro Fibrous Extracellular Matrices for Scalable Expansion of Human ES/IPS Cells, Liu [/bib_ref] [bib_ref] A 3D Fibrous Scaffold Inducing Tumoroids: A Platform for Anticancer Drug Development, Girard [/bib_ref]. Moghadas et al. used electrospinning to design layers of ECM fibers that can host cells for 3D culture and proliferation [bib_ref] Fabrication and Characterization of Low-Cost, Bead-Free, Durable and Hydrophobic Electrospun Membrane for..., Moghadas [/bib_ref]. In another work, a thick electrospun matrix could further be integrated within a microfluidic chip to study cell invasion throughout the ECM [bib_ref] A Novel Method to Understand Tumor Cell Invasion: Integrating Extracellular Matrix Mimicking..., Eslami Amirabadi [/bib_ref]. Alternatively, bioprinting methods enable a great spatial control on cells and ECM patterns and allow making complex structures with layers of mixed compounds and/or different cell types [bib_ref] A Three-Dimensional in Vitro Ovarian Cancer Coculture Model Using a High-Throughput Cell..., Xu [/bib_ref]. Recent works have also shown that the materials used and the way they are printed has an influence on cell behavior [bib_ref] Placental Basement Membrane Proteins Are Required for Effective Cytotrophoblast Invasion in a..., Kuo [/bib_ref]. This confirms that ECM is a key player in the development and migration of normal and tumor cells [bib_ref] Bioengineered 3D Platform to Explore Cell-ECM Interactions and Drug Resistance of Epithelial..., Loessner [/bib_ref]. This is further amplified by the findings of Zhou et al., who have shown that the molecular nature of the cell-matrix bindings also determines cellular spreading [bib_ref] Effect of RGD Content in Poly(Ethylene Glycol)-Crosslinked Poly(Methyl Vinyl Ether-Alt-Maleic Acid) Hydrogels..., Zhou [/bib_ref].
Cell-matrix interactions involve biomolecular recognition with specific ECM proteins. This has been thoroughly investigated by Kenny and co-workers, who have studied the impact of different ECM proteins on the adhesion and invasion of OC cells [bib_ref] Use of a Novel 3D Culture Model to Elucidate the Role of..., Kenny [/bib_ref]. From
## Ecm and scaffold
Among different features, the biochemical composition, topology, and mechanical properties of the ECM scaffold are crucial and must be scrupulously monitored and reproduced. This requires advanced techniques of biofabrication. Polymer science and biomaterials research have greatly improved our ability to engineer biological matrices. Among these, electrospinning, which is an electrically assisted extrusion method, is a promising cost-effective technique to mimic the fibrillar aspect of the ECM, which has been shown to influence cell behavior [bib_ref] Nano-on-Micro Fibrous Extracellular Matrices for Scalable Expansion of Human ES/IPS Cells, Liu [/bib_ref] [bib_ref] A 3D Fibrous Scaffold Inducing Tumoroids: A Platform for Anticancer Drug Development, Girard [/bib_ref]. Moghadas et al. used electrospinning to design layers of ECM fibers that can host cells for 3D culture and proliferation [bib_ref] Fabrication and Characterization of Low-Cost, Bead-Free, Durable and Hydrophobic Electrospun Membrane for..., Moghadas [/bib_ref]. In another work, a thick electrospun matrix could further be integrated within a microfluidic chip to study cell invasion throughout the ECM [bib_ref] A Novel Method to Understand Tumor Cell Invasion: Integrating Extracellular Matrix Mimicking..., Eslami Amirabadi [/bib_ref]. Alternatively, bioprinting methods enable a great spatial control on cells and ECM patterns and allow making complex structures with layers of mixed compounds and/or different cell types [bib_ref] A Three-Dimensional in Vitro Ovarian Cancer Coculture Model Using a High-Throughput Cell..., Xu [/bib_ref]. Recent works have also shown that the materials used and the way they are printed has an influence on cell behavior [bib_ref] Placental Basement Membrane Proteins Are Required for Effective Cytotrophoblast Invasion in a..., Kuo [/bib_ref]. This confirms that ECM is a key player in the development and migration of normal and tumor cells [bib_ref] Bioengineered 3D Platform to Explore Cell-ECM Interactions and Drug Resistance of Epithelial..., Loessner [/bib_ref]. This is further amplified by the findings of Zhou et al., who have shown that the molecular nature of the cell-matrix bindings also determines cellular spreading [bib_ref] Effect of RGD Content in Poly(Ethylene Glycol)-Crosslinked Poly(Methyl Vinyl Ether-Alt-Maleic Acid) Hydrogels..., Zhou [/bib_ref].
Cell-matrix interactions involve biomolecular recognition with specific ECM proteins. This has been thoroughly investigated by Kenny and co-workers, who have studied the impact of different ECM proteins on the adhesion and invasion of OC cells [bib_ref] Use of a Novel 3D Culture Model to Elucidate the Role of..., Kenny [/bib_ref]. From these assays, they reported that SKOV3 cells preferentially adhered to and invaded collagen I, followed by binding to collagen IV, fibronectin, vitronectin, and laminin 10 and 1. these assays, they reported that SKOV3 cells preferentially adhered to and invaded collagen I, followed by binding to collagen IV, fibronectin, vitronectin, and laminin 10 and 1. Important research efforts have been dedicated to the investigation of the dynamic behavior of EOC cells on biomimetic substrates to understand how tumors spread throughout the peritoneal cavity. Invasion assays have been proposed that consist of a monolayer of mesothelial cells cultured over a decellularized human tissue or a biocompatible substrate (mostly hydrogel) [bib_ref] Human Platelet Lysates-Based Hydrogels: A Novel Personalized 3D Platform for Spheroid Invasion..., Monteiro [/bib_ref] [bib_ref] Understanding the Effect of Mechanical Forces on Ovarian Cancer Progression, Martinez [/bib_ref]. This bi-layered barrier is then used to isolate EOC cells from a chamber filled with FBS to create a chemotactic gradient. By doing so, these teams were able to scrutinize the ability of EOC cells to invade the mesothelial layers by weakening and breaking cell-cell bindings [bib_ref] Keratin-14 (KRT14) Positive Leader Cells Mediate Mesothelial Clearance and Invasion by Ovarian..., Bilandzic [/bib_ref]. One step further, other works reported that the ascites fosters this invasion process thanks to several proteins [bib_ref] Ovarian Cancer Spheroids Use Myosin-Generated Force to Clear the Mesothelium, Iwanicki [/bib_ref]. In parallel, advances in hydrogel synthesis have allowed research groups to build 3D substrates, which migrating cells modify with secreted proteins (including matrix metalloproteases and lysyl oxidases) [bib_ref] Collagen I but Not Matrigel Matrices Provide an MMP-Dependent Barrier to Ovarian..., Sodek [/bib_ref]. Such hydrogel systems also provide models to study contraction effects within the 3D network in the course of cell migration [bib_ref] Effects of Migrating Cell-Induced Matrix Reorganization on 3D Cancer Cell Migration, Sun [/bib_ref].
Finally, the impact of the mechanical properties [bib_ref] Decoupling the Effects of Stiffness and Fiber Density on Cellular Behaviors via..., Berger [/bib_ref] and more recently, the impact of the topology have been thoroughly studied in different cancer types [bib_ref] A Novel Method to Understand Tumor Cell Invasion: Integrating Extracellular Matrix Mimicking..., Eslami Amirabadi [/bib_ref] [bib_ref] Three-Dimensional Collagen Topology Shapes Cell Morphology, beyond Stiffness, Chen [/bib_ref]. In particular, cells have been shown to modulate their morphology and invade the matrix with a speed that depends on its topology: by using the same matrix with two different fiber sizes, Eslami Amirabadi et al. observed that cells formed larger protrusions onto matrices with smaller fibers, a phenomenon attributed to the maximization of the cell-matrix interface [bib_ref] A Novel Method to Understand Tumor Cell Invasion: Integrating Extracellular Matrix Mimicking..., Eslami Amirabadi [/bib_ref]. In parallel, Guzman et al. found that the viscoelastic properties of the matrix, as well as its pore size (given that it is not sub-nuclear in order not to impair cell migration), had little influence on the invasive distance that OC cells travel [bib_ref] The Effect of Fibrillar Matrix Architecture on Tumor Cell Invasion of Physically..., Guzman [/bib_ref]. They propose that the invasion is mostly determined by the polarization of OC cells along the preferential orientation of the fibers in the matrix. Important research efforts have been dedicated to the investigation of the dynamic behavior of EOC cells on biomimetic substrates to understand how tumors spread throughout the peritoneal cavity. Invasion assays have been proposed that consist of a monolayer of mesothelial cells cultured over a decellularized human tissue or a biocompatible substrate (mostly hydrogel) [bib_ref] Human Platelet Lysates-Based Hydrogels: A Novel Personalized 3D Platform for Spheroid Invasion..., Monteiro [/bib_ref] [bib_ref] Understanding the Effect of Mechanical Forces on Ovarian Cancer Progression, Martinez [/bib_ref]. This bi-layered barrier is then used to isolate EOC cells from a chamber filled with FBS to create a chemotactic gradient. By doing so, these teams were able to scrutinize the ability of EOC cells to invade the mesothelial layers by weakening and breaking cell-cell bindings [bib_ref] Keratin-14 (KRT14) Positive Leader Cells Mediate Mesothelial Clearance and Invasion by Ovarian..., Bilandzic [/bib_ref]. One step further, other works reported that the ascites fosters this invasion process thanks to several proteins [bib_ref] Ovarian Cancer Spheroids Use Myosin-Generated Force to Clear the Mesothelium, Iwanicki [/bib_ref]. In parallel, advances in hydrogel synthesis have allowed research groups to build 3D substrates, which migrating cells modify with secreted proteins (including matrix metalloproteases and lysyl oxidases) [bib_ref] Collagen I but Not Matrigel Matrices Provide an MMP-Dependent Barrier to Ovarian..., Sodek [/bib_ref]. Such hydrogel systems also provide models to study contraction effects within the 3D network in the course of cell migration [bib_ref] Effects of Migrating Cell-Induced Matrix Reorganization on 3D Cancer Cell Migration, Sun [/bib_ref].
Finally, the impact of the mechanical properties [bib_ref] Decoupling the Effects of Stiffness and Fiber Density on Cellular Behaviors via..., Berger [/bib_ref] and more recently, the impact of the topology have been thoroughly studied in different cancer types [bib_ref] A Novel Method to Understand Tumor Cell Invasion: Integrating Extracellular Matrix Mimicking..., Eslami Amirabadi [/bib_ref] [bib_ref] Three-Dimensional Collagen Topology Shapes Cell Morphology, beyond Stiffness, Chen [/bib_ref]. In particular, cells have been shown to modulate their morphology and invade the matrix with a speed that depends on its topology: by using the same matrix with two different fiber sizes, Eslami Amirabadi et al. observed that cells formed larger protrusions onto matrices with smaller fibers, a phenomenon attributed to the maximization of the cellmatrix interface [bib_ref] A Novel Method to Understand Tumor Cell Invasion: Integrating Extracellular Matrix Mimicking..., Eslami Amirabadi [/bib_ref]. In parallel, Guzman et al. found that the viscoelastic properties of the matrix, as well as its pore size (given that it is not sub-nuclear in order not to impair cell migration), had little influence on the invasive distance that OC cells travel [bib_ref] The Effect of Fibrillar Matrix Architecture on Tumor Cell Invasion of Physically..., Guzman [/bib_ref]. They propose that the invasion is mostly determined by the polarization of OC cells along the preferential orientation of the fibers in the matrix.
Investigating the impact of the matrix on tumor growth and progression is still an important challenge that should now be considered with the integration of flow. This adds a dynamic control over the circulating environment and reproduces shear stress at play in the peritoneal cavity.
## Contribution of microfluidics: consideration of shear forces in mechanotransduction
## Examples of microfluidic setups
Although microfluidics has been more traditionally used as a tool to increase the throughput of anticancer drug assays [bib_ref] Empirical Chemosensitivity Testing in a Spheroid Model of Ovarian Cancer Using a..., Das [/bib_ref] , it is now becoming widely used to grow and differentiate spheroids or embryoid bodies [bib_ref] 3-Dimensional Cell Culture for on-Chip Differentiation of Stem Cells in Embryoid Body, Kim [/bib_ref] [bib_ref] Development of a Three-Dimensional Microfluidic System for Long-Term Tumor Spheroid Culture, Ziółkowska [/bib_ref] , as well as for building in vitro cancer models with flow control for mimicking body flow and vasculature [bib_ref] Metastasis in Context: Modeling the Tumor Microenvironment with Cancer-on-a-Chip Approaches, Sleeboom [/bib_ref]. In particular, microfluidics has been considered as a candidate to reproduce the complex interplays within the peritoneal cavity during OC. Since first reported chips with a unidirectional perfusion and separated inlet and outlet reservoirs for co-culturing [bib_ref] A New Tool for Probing of Cell-Cell Communication: Human Embryonic Germ Cells..., Song [/bib_ref] , different setups have been reported for investigating OC [fig_ref] Figure 10: Schemes representing different microfluidic setups proposed in the literature [/fig_ref]. Li and co-workers reported the engineering of a 3D ovarian cancer-mesothelium microfluidic platform, where OC spheroids were co-cultured with primary human peritoneal mesothelial cells. In this chip, mesothelial cells were plated on fibronectin. Non-adherent OC spheroids were seeded into microfluidic channels with continuous flow medium perfused by a syringe pump [fig_ref] Figure 10: Schemes representing different microfluidic setups proposed in the literature [/fig_ref] [bib_ref] Modeling Ovarian Cancer Multicellular Spheroid Behavior in a Dynamic 3D Peritoneal Microdevice, Li [/bib_ref]. This setup allowed them to study the impact of shear stress on spheroids, as well as the interplays between fibronectin, mesothelial cells, and EOC spheroids, as expected in the peritoneal cavity during metastasis. In another work, Rizvi et al. designed a linear microfluidic chip to study the impact of flow on the attachment and growth of OC cells [bib_ref] Flow Induces Epithelial-Mesenchymal Transition, Cellular Heterogeneity and Biomarker Modulation in 3D Ovarian..., Rizvi [/bib_ref]. In their setup, tumor cells entered the channels through gas permeable silicone tubing and flowed over stromal beds of growth factor reduced (GFR) Matrigel. A portion of the cells are effluxed from the chip via the outlet tubing [fig_ref] Figure 10: Schemes representing different microfluidic setups proposed in the literature [/fig_ref]. Those that adhered to the Matrigel beds were cultured under continuous flow (shear stress of 0.5 to 3 dyne/cm 2 ) for seven days. In this work, they reported the flow-induced, transcriptionally regulated decrease in E-cadherin protein expression and the simultaneous increase in vimentin, indicating an increasing metastatic potential.
Investigating the impact of the matrix on tumor growth and progression is still an important challenge that should now be considered with the integration of flow. This adds a dynamic control over the circulating environment and reproduces shear stress at play in the peritoneal cavity.
## Contribution of microfluidics: consideration of shear forces in mechanotransduction
## Examples of microfluidic setups
Although microfluidics has been more traditionally used as a tool to increase the throughput of anticancer drug assays [bib_ref] Empirical Chemosensitivity Testing in a Spheroid Model of Ovarian Cancer Using a..., Das [/bib_ref] , it is now becoming widely used to grow and differentiate spheroids or embryoid bodies [bib_ref] 3-Dimensional Cell Culture for on-Chip Differentiation of Stem Cells in Embryoid Body, Kim [/bib_ref] [bib_ref] Development of a Three-Dimensional Microfluidic System for Long-Term Tumor Spheroid Culture, Ziółkowska [/bib_ref] , as well as for building in vitro cancer models with flow control for mimicking body flow and vasculature [bib_ref] Metastasis in Context: Modeling the Tumor Microenvironment with Cancer-on-a-Chip Approaches, Sleeboom [/bib_ref]. In particular, microfluidics has been considered as a candidate to reproduce the complex interplays within the peritoneal cavity during OC. Since first reported chips with a unidirectional perfusion and separated inlet and outlet reservoirs for co-culturing [bib_ref] A New Tool for Probing of Cell-Cell Communication: Human Embryonic Germ Cells..., Song [/bib_ref] , different setups have been reported for investigating OC [fig_ref] Figure 10: Schemes representing different microfluidic setups proposed in the literature [/fig_ref]. Li and co-workers reported the engineering of a 3D ovarian cancer-mesothelium microfluidic platform, where OC spheroids were co-cultured with primary human peritoneal mesothelial cells. In this chip, mesothelial cells were plated on fibronectin. Non-adherent OC spheroids were seeded into microfluidic channels with continuous flow medium perfused by a syringe pump [fig_ref] Figure 10: Schemes representing different microfluidic setups proposed in the literature [/fig_ref] [bib_ref] Modeling Ovarian Cancer Multicellular Spheroid Behavior in a Dynamic 3D Peritoneal Microdevice, Li [/bib_ref]. This setup allowed them to study the impact of shear stress on spheroids, as well as the interplays between fibronectin, mesothelial cells, and EOC spheroids, as expected in the peritoneal cavity during metastasis. In another work, Rizvi et al. designed a linear microfluidic chip to study the impact of flow on the attachment and growth of OC cells [bib_ref] Flow Induces Epithelial-Mesenchymal Transition, Cellular Heterogeneity and Biomarker Modulation in 3D Ovarian..., Rizvi [/bib_ref]. In their setup, tumor cells entered the channels through gas permeable silicone tubing and flowed over stromal beds of growth factor reduced (GFR) Matrigel. A portion of the cells are effluxed from the chip via the outlet tubing [fig_ref] Figure 10: Schemes representing different microfluidic setups proposed in the literature [/fig_ref]. Those that adhered to the Matrigel beds were cultured under continuous flow (shear stress of 0.5 to 3 dyne/cm²) for seven days. In this work, they reported the flow-induced, transcriptionally regulated decrease in E-cadherin protein expression and the simultaneous increase in vimentin, indicating an increasing metastatic potential. [bib_ref] Fluid-Flow Induced Wall Shear Stress and Epithelial Ovarian Cancer Peritoneal Spreading, Avraham-Chakim [/bib_ref]. The chip could be disassembled to install the membrane in custom-designed wells and the cells in the testing flow chamber, and then re-assembled for biological testing [fig_ref] Figure 10: Schemes representing different microfluidic setups proposed in the literature [/fig_ref]. Schemes representing different microfluidic setups proposed in the literature. (A) 3D ovarian cancer-mesothelium microfluidic platform. Adapted with permission from [bib_ref] Modeling Ovarian Cancer Multicellular Spheroid Behavior in a Dynamic 3D Peritoneal Microdevice, Li [/bib_ref]. (B) Matrigelbased linear microfluidic chip for investigating the impact of flow on 3D tumor growth. Reproduced with permission from the National Academy of Sciences, USA [bib_ref] Flow Induces Epithelial-Mesenchymal Transition, Cellular Heterogeneity and Biomarker Modulation in 3D Ovarian..., Rizvi [/bib_ref]. (C) Scheme of the experimental system and for application of wall shear stress on cultured cells. Reproduced with permission from [bib_ref] Fluid-Flow Induced Wall Shear Stress and Epithelial Ovarian Cancer Peritoneal Spreading, Avraham-Chakim [/bib_ref].
As a last example, Avraham-Chakim et al. developed a flow chamber for direct application of fluid flow induced wall shear stress (WSS) on a monolayer of OVCAR-3 OC cells cultured on denuded amniotic membranes (mainly collagen and fibronectin) [bib_ref] Fluid-Flow Induced Wall Shear Stress and Epithelial Ovarian Cancer Peritoneal Spreading, Avraham-Chakim [/bib_ref]. The chip could be disassembled to install the membrane in custom-designed wells and the cells in the testing flow chamber, and then re-assembled for biological testing [fig_ref] Figure 10: Schemes representing different microfluidic setups proposed in the literature [/fig_ref]. The flow chamber was designed to hold three wells hosting cells for multiple experiments. The pump could generate a steady flow with a uniform field of shear forces on top of the cells (0.5, 1, and 1.5 dyne/cm 2 ). The space under the well-bottoms inside the flow chamber was filled with static culture medium in contact with the bottom plane of the membrane. From this work, they have suggested that WSS has a significant impact on the mechanical regulation of EOC spreading in the peritoneal cavity by acting on cytoskeleton reorganization (cell elongation, stress fibers formation, and microtubules generation).
## Contribution of microfluidics in understanding the progression of ovarian cancer
While the pattern and magnitude of fluid motion in the peritoneal cavity remains difficult to map precisely, the shear stress on cells is estimated to be in the 0-10 dyne/cm 2 range [bib_ref] Fluid Shear Stress Impacts Ovarian Cancer Cell Viability, Subcellular Organization, and Promotes..., Hyler [/bib_ref]. This flow-induced mechanotransduction triggers rapid signaling events, which impact cytoskeleton organization and further drive cell proliferation, adhesion, and invasion.
## How microfluidic shear stress impact ovarian cancer
In their work dating back to 2010, van der Meer and co-workers have shown how microfluidics can provide a mechanistic insight into cytoskeleton remodeling when cells-endothelial cells in this case-experience directional shear stress [bib_ref] Analyzing Shear Stress-Induced Alignment of Actin Filaments in Endothelial Cells with a..., Van Der Meer [/bib_ref]. In the context of ovarian cancer, Hyler et al. proposed more recently to use swirling and rotating fluid circulation to better mimic peritoneal fluid motion [bib_ref] Fluid Shear Stress Impacts Ovarian Cancer Cell Viability, Subcellular Organization, and Promotes..., Hyler [/bib_ref]. In addition, they have exposed cells to repetitive 96 h periods of fluid shear stress on disseminated and adherent OC cells. This setup is supposed to mimic the physiological exfoliation of surface cells in the peritoneal cavity, which could re-adhere at a different location. They used mouse ovarian cancer epithelial cells (MOSE) from benign (MOSE-E), slow (MOSE-L), and fast (MOSE-LTICν)-developing cancer together with human SKOV-3 OC cells. After exposure to fluid shear stress, they observed that actin protrusions were increased in all tumorigenic cells. In parallel, they investigated the shear-induced changes in cell adhesion by determining the number and length of vinculin-containing focal adhesions. Overall, cells exhibit a higher number of focal adhesions under shear stress, which was particularly significant for the benign MOSE-E cells. However, the focal adhesion length was found to slightly increase in tumor cells experiencing flow, contrary to the benign MOSE-E cells. This is in line with the work of Avraham-Chakim et al., which has evidenced shear stress-induced cytoskeleton reorganization of OVCAR-3 OC cells cultured in flow chambers coated with denuded amniotic membranes [fig_ref] Figure 10: Schemes representing different microfluidic setups proposed in the literature [/fig_ref] [bib_ref] Fluid-Flow Induced Wall Shear Stress and Epithelial Ovarian Cancer Peritoneal Spreading, Avraham-Chakim [/bib_ref].
In their work, Hyler et al. also showed that cells exposed to fluid shear stress exhibited a high increase in CREST-positive micronuclei, which signed for chromosome mis-segregation during mitosis. Finally, they have reported that tumorigenic OC cell lines responded to shear stress by detaching and forming spheroids [bib_ref] Fluid Shear Stress Impacts Ovarian Cancer Cell Viability, Subcellular Organization, and Promotes..., Hyler [/bib_ref]. The formation and stemness of OC spheroid have also been investigated in another microfluidic chip by the group of Wong [bib_ref] Stemness and Chemoresistance in Epithelial Ovarian Carcinoma Cells under Shear Stress, Ip [/bib_ref]. Their device provides a continuous well-defined flow rate (0.002 and 0.02 dyne/cm 2 ), with a synthetic polymer coating (poly-HEMA) preventing cell attachment and matrix deposition [fig_ref] Figure 2: Figure 2 [/fig_ref]. This aims at keeping tumor spheroids in suspension, as observed in patient ascites. From this setup, they have reported that spheroids under flow expressed stem cell markers (Oct-4, c-Kit, ABCG2, and P-gp), contrary to what was observed under static conditions. Furthermore, they detected an enrichment in CD117 + /CD44 + cells in spheroids exposed to shear stress, together with enhanced self-renewal potential, differentiation ability, and increased tumor-initiating capability. Finally, they identified a mechanosensitive miRNA-miR-199a-3p-that showed a marked decrease under physiologic shear stress [bib_ref] Stemness and Chemoresistance in Epithelial Ovarian Carcinoma Cells under Shear Stress, Ip [/bib_ref]. These findings stimulated further works on this signaling pathway using the same polymer-coated microfluidic device. Working with a shear stress of 0.02 dyne/cm 2 , they identified c-Met as a shear stress-responsive receptor tyrosine kinase [bib_ref] Flow Induces Epithelial-Mesenchymal Transition, Cellular Heterogeneity and Biomarker Modulation in 3D Ovarian..., Rizvi [/bib_ref]. This provides a mechanistic insight into the downstream regulation of miR-199a-3p and the consecutive impact on drug resistance. These results confirm previous reports from Rizvi et al., who described the increased biomarker expression and tumor morphology consistent with increased EMT, and that is attributed to hydrodynamic forces [fig_ref] Figure 10: Schemes representing different microfluidic setups proposed in the literature [/fig_ref] [bib_ref] Flow Induces Epithelial-Mesenchymal Transition, Cellular Heterogeneity and Biomarker Modulation in 3D Ovarian..., Rizvi [/bib_ref]. These findings indicate that fluid shear stress induces a motile and aggressive tumor phenotype, which is driven in part by a post-translational upregulation of epidermal growth factor receptor (EGFR) expression and activation, in turn associated with the worst prognosis in ovarian cancer. In their work, Hyler et al. also showed that cells exposed to fluid shear stress exhibited a high increase in CREST-positive micronuclei, which signed for chromosome missegregation during mitosis. Finally, they have reported that tumorigenic OC cell lines responded to shear stress by detaching and forming spheroids [bib_ref] Fluid Shear Stress Impacts Ovarian Cancer Cell Viability, Subcellular Organization, and Promotes..., Hyler [/bib_ref]. The formation and stemness of OC spheroid have also been investigated in another microfluidic chip by the group of Wong [bib_ref] Stemness and Chemoresistance in Epithelial Ovarian Carcinoma Cells under Shear Stress, Ip [/bib_ref]. Their device provides a continuous well-defined flow rate (0.002 and 0.02 dyne/cm²), with a synthetic polymer coating (poly-HEMA) preventing cell attachment and matrix deposition [fig_ref] Figure 2: Figure 2 [/fig_ref]. This aims at keeping tumor spheroids in suspension, as observed in patient ascites. From this setup, they have reported that spheroids under flow expressed stem cell markers (Oct-4, c-Kit, ABCG2, and P-gp), contrary to what was observed under static conditions. Furthermore, they detected an enrichment in CD117 + /CD44 + cells in spheroids exposed to shear stress, together with enhanced self-renewal potential, differentiation ability, and increased tumor-initiating capability. Finally, they identified a mechanosensitive miRNA-miR-199a-3p-that showed a marked decrease under physiologic shear stress [bib_ref] Stemness and Chemoresistance in Epithelial Ovarian Carcinoma Cells under Shear Stress, Ip [/bib_ref]. These findings stimulated further works on this signaling pathway using the same polymer-coated microfluidic device. Working with a shear stress of 0.02 dyne/cm², they identified c-Met as a shear stress-responsive receptor tyrosine kinase [bib_ref] Flow Induces Epithelial-Mesenchymal Transition, Cellular Heterogeneity and Biomarker Modulation in 3D Ovarian..., Rizvi [/bib_ref]. This provides a mechanistic insight into the downstream regulation of miR-199a-3p and the consecutive impact on drug resistance. These results confirm previous reports from Rizvi et al., who described the increased biomarker expression and tumor morphology consistent with increased EMT, and that is attributed to While this illustrates how microfluidics enable addressing the interplay between biophysical and biochemical cues in the ascitic microenvironment, it also highlights the necessity to develop advanced devices for therapeutic development.
## Shear stress in chemoresistance
Wong and his research group have devoted research efforts to understanding the impact of fluid shear stress on chemoresistance in ovarian cancer. Their device provides a continuous well-defined flow rate (0.002 and 0.02 dyne/cm 2 ), with a synthetic polymer coating (poly-HEMA) preventing cell attachment and matrix deposition [fig_ref] Figure 2: Figure 2 [/fig_ref] [bib_ref] Stemness and Chemoresistance in Epithelial Ovarian Carcinoma Cells under Shear Stress, Ip [/bib_ref]. Spheroids were grown from SKOV3 cells with a mean diameter of 104.6 ± 1.67 µm. These spheroids were treated with two antitumoral drugs cisplatin and paclitaxel, in the presence or absence of shear stress. The results are analyzed with an Annexin V/PI staining to detect viable, necrotic, early and late apoptotic cells (see [fig_ref] Figure 2: Figure 2 [/fig_ref]. In static conditions, cells in OC spheroids rapidly underwent apoptosis upon cisplatin and paclitaxel treatment (upper right panels, [fig_ref] Figure 2: Figure 2 [/fig_ref]. In contrast, under flow, cells in tumor spheroids showed significantly greater chemoresistance in the presence of cisplatin and paclitaxel, with 65% to 70% of cells in spheroids remaining viable (lower left panels, [fig_ref] Figure 2: Figure 2 [/fig_ref].
Deciphering the mechanisms of chemoresistance remains an open challenge. Still, the PI3K/Akt signaling is of particular relevance for chemoresistance. In recent works, Wong's group again deepened this understanding of the inverse correlation of miR-199a-3p expression with enhanced drug resistance in chemoresistant OC cell lines. In particular, they proposed that the shear stress-dependent downregulation of miR-199a-3p expression may activate PI3K/Akt signaling. These results should provide an additional key for correlating stemness and chemoresistance in OC, and hopefully represent a viable target for therapeutic development.
impact of fluid shear stress on chemoresistance in ovarian cancer. Their device provides a continuous well-defined flow rate (0.002 and 0.02 dyne/cm²), with a synthetic polymer coating (poly-HEMA) preventing cell attachment and matrix deposition [fig_ref] Figure 2: Figure 2 [/fig_ref] [bib_ref] Stemness and Chemoresistance in Epithelial Ovarian Carcinoma Cells under Shear Stress, Ip [/bib_ref]. Spheroids were grown from SKOV3 cells with a mean diameter of 104.6 ± 1.67 µ m. These spheroids were treated with two antitumoral drugs cisplatin and paclitaxel, in the presence or absence of shear stress. The results are analyzed with an Annexin V/PI staining to detect viable, necrotic, early and late apoptotic cells (see [fig_ref] Figure 2: Figure 2 [/fig_ref]. In static conditions, cells in OC spheroids rapidly underwent apoptosis upon cisplatin and paclitaxel treatment (upper right panels, [fig_ref] Figure 2: Figure 2 [/fig_ref]. In contrast, under flow, cells in tumor spheroids showed significantly greater chemoresistance in the presence of cisplatin and paclitaxel, with 65% to 70% of cells in spheroids remaining viable (lower left panels, [fig_ref] Figure 2: Figure 2 [/fig_ref]. Deciphering the mechanisms of chemoresistance remains an open challenge. Still, the PI3K/Akt signaling is of particular relevance for chemoresistance. In recent works, Wong's group again deepened this understanding of the inverse correlation of miR-199a-3p expression with enhanced drug resistance in chemoresistant OC cell lines. In
## Vascularized microfluidic models
Microfluidics has been applied to the world of biology and medicine to build models where cells are confronted with the flow conditions of living systems. This concerns the shear rate and the resulting mechanotransduction pathways. It is also about the supply of nutrients and oxygen to cells. In other words, it is about mimicking the vascular system and its essential role in the optimal functioning of organs.
In this context, the group of Jain built an organ-on-a-chip model of ovarian cancer dedicated to the investigation of the cross-talk between vessels, platelets, and ovarian cancer cells. In a first setup, called the ovarian cancer-on-chip (OvCa-Chip), they superimposed two PDMS fluidic chambers [bib_ref] OvCa-Chip microsystem recreates vascular endothelium-mediated platelet extravasation in ovarian cancer, Saha [/bib_ref]. The top microchannel is seeded with human ovarian A2780 tumor cells and mimics the peritoneal cavity. The bottom channel is lined with human primary endothelial cells that form continuous monolayers and cover all four sides of the microchannel, creating a blood-perfused vessel. One step further, they have incorporated a collagen-based ECM adjacent to the tumor cell chamber, ending with a so-called ovarian tumor microenvironment organ-on-chip (OTME-Chip). Freshly derived platelets from human blood were perfused through the bottom microchannel for 3 days and cell invasion dynamics were monitored with respect to platelet extravasation. Using gene-edited tumors and RNA sequencing, they investigated the impact of the interactions between glycoprotein VI (GPVI) and galectin-3 in mediating plateletpromoted tumor metastasis. To this end, they compared the OTME-Chip seeded with OC cells with a chip involving galectin-3 knocked out ovarian cancer cells (KO-OTME-Chip). Further comparison was performed with a Control-Chip, where no platelet was perfused.
Rapid ECM invasion was seen in the OTME-Chip compared with the KO version, indicating a role of GPVI and galectin-3 in the platelet-promoted metastasis. The role of platelets in promoting ovarian cancer metastasis was further confirmed by the even lower invasion observed in the platelet-free Control-Chip. Finally, they perfused the GPVI inhibitor Revacept, which impaired metastatic potential, illustrating the possibility of using the OTME-Chip for therapeutic exploration. From a technological point of view, this work shows how important it is to associate vascularization to validate in vitro models. channel for 3 days and cell invasion dynamics were monitored with respect to platelet extravasation. Using gene-edited tumors and RNA sequencing, they investigated the impact of the interactions between glycoprotein VI (GPVI) and galectin-3 in mediating platelet-promoted tumor metastasis. To this end, they compared the OTME-Chip seeded with OC cells with a chip involving galectin-3 knocked out ovarian cancer cells (KO-OTME-Chip). Further comparison was performed with a Control-Chip, where no platelet was perfused. Rapid ECM invasion was seen in the OTME-Chip compared with the KO version, indicating a role of GPVI and galectin-3 in the platelet-promoted metastasis. The role of platelets in promoting ovarian cancer metastasis was further confirmed by the even lower invasion observed in the platelet-free Control-Chip. Finally, they perfused the GPVI inhibitor Revacept, which impaired metastatic potential, illustrating the possibility of using the OTME-Chip for therapeutic exploration. From a technological point of view, this work shows how important it is to associate vascularization to validate in vitro models. From vascularized in vitro chip models, a further step in integrating multicellular constructs has been recently achieved by Ibrahim and co-workers [bib_ref] Omentum-on-a-chip: A multicellular, vascularized microfluidic model of the human peritoneum for the..., Ibrahim [/bib_ref]. In their work, they co-cultured tumor cells (TCs) with mesothelial and endothelial cells (MCs and ECs), together with adipocytes (ADs). The 3D aspect of the model is ensured by the use of a fibrin hydrogel seeded with ADs providing TCs with a rich stroma. This fibrin gel can be vascularized by ECs and topped with a monolayer of MCs. The combination of the four different types of cell in a 3D model enable investigating the impact of the permeability of the mesothelial monolayer, as well as the impact of vascular permeability in promoting intraperitoneal metastases.
In a first step, they investigated the ability of cells to secrete ECM proteins in the chip-integrated 3D model. To this end, they checked the expression of collagen VI and fibronectin, which are highly expressed in the ECM of the omentum and peritoneum. Then, TCs were seeded on the mesothelial layer with varying cell densities. By doing so, they have shown that a critical cell density is required for tumor growth. In addition, tumor growth was further enhanced by stromal ADs and ECs present in the peritoneal omentum. As a result, and beyond vascularization, this work shows that multicellular 3D models are essential to elucidate tumor-stromal cell interactions during intraperitoneal metastasis of ovarian cancer. ectin, which are highly expressed in the ECM of the omentum and peritoneum. Then, TCs were seeded on the mesothelial layer with varying cell densities. By doing so, they have shown that a critical cell density is required for tumor growth. In addition, tumor growth was further enhanced by stromal ADs and ECs present in the peritoneal omentum. As a result, and beyond vascularization, this work shows that multicellular 3D models are essential to elucidate tumor-stromal cell interactions during intraperitoneal metastasis of ovarian cancer.
# Conclusion and perspectives
# Conclusions and perspectives
Ovarian cancer as a living system is multifaceted and multiparametric, notably combining biochemical and biophysical cues. It is also the crossroads of many pathways and the seat of a heterogeneity-including cellular-which remains difficult to describe at present. This makes it a huge challenge to tackle. A crucial direction to take to meet this challenge is to work on the design of biologically relevant models. What parameters should in vitro models integrate to be relevant for bringing new findings in a pathophysiological context? It is a whole scientific field that brings together researchers with different expertise, working together for a common objective. This line has seen the development of biological and biomedical engineering to fabricate model systems (3D cell culture models, spheroid engineering, scaffold-induced mechanotransduction) and associated physical and metric instruments to describe cellular behaviors. More recently, microfluidics has been introduced to the world of cell biology through high throughput and drug screening. Moreover, microfluidics has for years brought new perspectives in the study of cancer thanks to the integration of flow, which is now recognized as a crucial actor in the biological context, both in terms of circulating factors and shear-induced mechanotransduction. Research effort on ovarian cancers, among others, has strongly emphasized the need to develop devices based on flow for the development of therapeutics.
Microfluidics enable fine tuning of parameters to model the influence of specific signals from the microenvironment in cellular heterogeneity. We believe that the combination of quantitative tools for monitoring and measuring cellular parameters with microfluidics constitutes the Rosetta stone for deciphering the complex interactions of the microenvironment and the cells involved in the pathophysiological processes of ovarian cancer. A challenge remains in the ability to control and switch those interactions not only externally with microfluidics, but from the cell perspective. Controlling cell interaction with the microenvironment with microfluidic-like speed and precision is a key challenge to achieve this level of decipherment.
Microfluidics has already contributed to the identification of several general mechanisms that govern processes during cancer progression, including tumor growth, metastasis, and chemoresistance. Questions that have been discussed in this review still remain to be mastered. In particular, the possibility of increasing the integration of different cell types, and the proper measurement of each in a so-called co-culture system should always be improved and deepened. Other aspects beyond the scope of this review are worth mentioning here in the perspective of ever better recapitulating living systems. To name a few, one can emphasize the need to reproduce and control the dynamic remodeling of the extracellular matrix during tumor progression. Moreover, the communication with distant organs must also be modeled. This must be based on the possibility of vascularizing the in vitro models. This is another active area of research in microfluidics and organ-on-chip, whose maturity should make it possible, in the near future, to make a leap forward in understanding the mechanisms of tumor progression, in particular by repositioning ovarian cancer in a whole organism.
[fig] Figure 2: Figure 2. EMT events in ovarian cancer hematogenous metastasis. In Step 1, lesions with characteristic alterations in TP53 develop into cancers in the fallopian tube and the ovaries. OC cells detach and shed into the peritoneal fluid for transcoelomic spread or enter the blood vessels leading to hematogenous metastasis. In Step 2, OC cells in the ascites show high heterogeneity along the EMT continuum, forming anoikis-resistant cell aggregates. Ascites facilitates cell aggregate adhesions to the peritoneal membrane. Such adhesions in Step 3 can undergo MET (reverse EMT), enabling the cells to establish and grow at secondary sites. At the peritoneal interface, cancer cells invade peritoneal mesothelial cells facilitated by integrins and TGF-β, developing secondary tumors and metastasis [35]. Reproduced with permission of Springer Nature. [/fig]
[fig] Figure 3: (A) Cell motility tracking on soft and hard matrices and analysis revealing a significantly higher cell migration coefficient on soft substrates. (B) Heat maps of traction stresses (Pascals) overlaid with black arrows showing cell-induced matrix displacements. The cell center of mass is shown by the circle, and the triangle shows the force-weighted center of mass. Scale bars: 10 µ m. (C) Staining for pMLC reveals a corresponding increase in pMLC intensity on soft substrates. Data are shown as the mean ± s.e.m.; *** p < 0.001. Reproduced with permission from [81]. [/fig]
[fig] 2. 4 2: The Importance of Fluids in OC: Soluble Factors and Shear Stress [/fig]
[fig] Figure 4: (A) Adherent SKOV3 and IGROV1 cells were starved overnight in serum-free medium, and cultured 6 h in their culture media supplemented with 0 (-MMC) or 75 mg/mL of Ficoll 400 kDa (Fc) or Dextran 250 kDa (Dx). Cells were stained for actin (green) and DNA (blue). Representative images of two independent experiments in duplicate reveal the reorganization of the actin cytoskeleton in crowded environments. Reproduced with permission from Elsevier[92]. (B) Schematic representation of the signaling pathways regulating cell cycle and differentiation in tumor cells in response to shear stress. ↑ up and ↓ downregulation by shear; dotted double-arrow line represents the interaction pathway that has not been defined. Copyright (2008) National Academy of Sciences, USA[93]. [/fig]
[fig] Figure 5: Scheme of the main steps of the OC metastasis process: EMP, stemness, and shear stress. Under braces are the different in vitro models, which aim at recapitulating the corresponding pathophysiological context. Made with BioRender. [/fig]
[fig] Figure 6: (A) Scheme of the multi-cellular setup. (B) Immunofluorescence staining of EpCAM-positive G164 cells, VCAN and FN1 in tetra-and penta-cultures. (C,D) Quantification of FN1 and VCAN deposition in the tetra-and penta-culture (at least 2 gels per condition, n = 3). (E) Quantification of EpCAM deposition. (F) Flow cytometry analysis of EpCAM-positive cells (n = 3). Data are [/fig]
[fig] Figure 7: (A) Photograph and (B,C) SEM images of the bare patch. (D,E) Optical microscopy images of spheroids. (F) Immunofluorescence 3D reconstruction of a SKOV-3 spheroid with epithelial markers EpCAM (in green) and E-cadherin (in red), and nuclei (DAPI in blue). Reproduced with permission from[131]. [/fig]
[fig] Figure 8: Imaging of the internal cellular organization of growing spheroids under elastic confinement. Confocal images of free (A,B) and confined (D,E) spheroids after cryosection and immunolabeling for DAPI (blue), KI67 (magenta), and fibronectin (red). Quantification of cell nuclei (blue), proliferating cells (purple), and dead cells (gray) radial densities for (C) free and (F) confined spheroids (scale bars 100 μm). Copyright (2013) National Academy of Sciences, USA [133]. [/fig]
[fig] Figure 9: (A) Adhesion and (B) invasion assays of SKOV3ip.1 or IOSE fluorescently labeled cells on different ECM compositions. Reproduced with permission from[17]. [/fig]
[fig] Figure 10: Schemes representing different microfluidic setups proposed in the literature. (A) 3D ovarian cancer-mesothelium microfluidic platform. Adapted with permission from [153]. (B) Matrigel-based linear microfluidic chip for investigating the impact of flow on 3D tumor growth. Reproduced with permission from the National Academy of Sciences, USA [154]. (C) Scheme of the experimental system and for application of wall shear stress on cultured cells. Reproduced with permission from [155]. As a last example, Avraham-Chakim et al. developed a flow chamber for direct application of fluid flow induced wall shear stress (WSS) on a monolayer of OVCAR-3 OC cells cultured on denuded amniotic membranes (mainly collagen and fibronectin) [/fig]
[fig] Figure 11: (A) Immunofluorescence images showing the reorganization of the actin (green) cytoskeleton in adherent MOSE-E, MOSE-L, SKOV-3, and MOSE-LTICv cells after exposure to fluid shear stress. (B) Quantitation of vinculin-positive focal adhesion number and size in control and after exposure to shear stress (t-test, ** p < 0.01, *** p < 0.001). Adapted with permission from[94]. [/fig]
[fig] Figure 12: (A) Scheme of the experimental setup and side view of a poly-HEMA-coated (non-adherent) microfluidic channel under perfusion. (B) The chemosensitivity against cisplatin and paclitaxel of OC spheroids under static and 0.02 dyne/cm² shear stress was analyzed by Annexin V/PI staining.Adapted with permission from Springer Nature[96]. [/fig]
[fig] Figure 13: (A) Scheme of the microdevice containing two PDMS compartments separated by a thin, porous membrane mimicking the tumor-vascular interface. The right part shows a crosssectional view of the chip. (B) Cross-section of 3D confocal imaging of OTME-Chip showing cancer cells (yellow), endothelial cells (red), and platelets (cyan) at 0 and 72 h after platelet extravasation.(C) Fluorescence microscopy images showing cancer cell (green) invasion (marked by arrows) into hydrogel ECM due to extravasated platelets (yellow). (D) Bar graph of the quantification of ECM invasion at 48 h (Dunnett test, * p < 0.05, ** p < 0.01). Adapted with permission from Science Advances[159]. [/fig]
[fig] Figure 14: (A) Scheme of the 3D model and (B) PDMS mold with patterned channels fabricated using soft lithography. The central region (green) contains cells and a fibrin hydrogel. The side channels and reservoirs (purple) as well as the top channel and reservoir (orange) were filled with cell culture medium. (C) Scheme and confocal image of the vascularized model, in which ECs express GFP (green), nuclei are stained with DAPI (blue), and AD lipid droplets are stained with LipidTox (red). (D) Immunohistochemistry analysis of ECM.(E) 2D projected confocal z-stack of SKOV3 TCs (red) on the MC membrane (blue), but not invading the ECM of vascularized ECs (green) and ADs (not stained) on day 14, 7 days after high-density TC seeding. Adapted with permission from Elsevier[160]. [/fig]
[fig] Author: Contributions: All authors have contributed to the redaction of this review. All authors have read and agreed to the published version of the manuscript. Funding: The authors thank the China Scholarship Council for the PhD grant of C.C., and the Oversea Study Program of Guangzhou Elite Project for the PhD grant of Z.W. We thank the CNRS through the MITI interdisciplinary programs and CY Initiative of Excellence for financial support. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Not applicable. [/fig]
|
A modified surgical technique for Descemet's stripping automated endothelial keratoplasty (DSAEK) in altered or abnormal anatomy
A B S T R A C TPurpose: This paper describes a modified technique for the Descemet's Stripping Automated Endothelial Keratopasty (DSAEK) surgery in eyes with abnormal or altered anatomy. Certain anatomic abnormalities increase the level of surgical complexity, and lead to increased risk of donor lenticule detachment. These challenges include aniridia, abnormal iris, aphakia and hypotony from previous vitrectomy. Observations: The Sheets glide was trimmed to 4mm in width and inserted into the clear-corneal wound. The "needle-push"/Fichman glide technique was used to insert the DSAEK donor graft. The modification of the technique involves maintaining the Sheets glide in the eye for the duration of the intraoperative air bubble. After confirming desired orientation of the graft, air was injected into the anterior chamber, creating a large air bubble between the graft and the Sheets glide. After 10 minutes, the Sheets glide was removed, and the main wound was closed. Conclusions and Importance: This technique overcomes the challenges of posterior air bubble migration and posterior dislocation of the donor lenticule in eyes with altered anatomy.
# Introduction
By 2014, Descemet stripping endothelial keratoplasty (DSAEK) became the most common (50%) type of corneal transplantation technique in the United States. [bib_ref] Keratoplasty in the United States: a 10-year review from, Park [/bib_ref] As increased number of patients with endothelial dysfunction undergo DSAEK, more eyes with altered anatomy are encountered during surgery. During DSAEK surgery, an air bubble is injected into the anterior chamber to allow for DSAEK lenticule graft adhesion. Anatomic abnormalities such as aniridia, abnormal iris, aphakia and hypotony from previous history of vitrectomy, can present intra-operative challenges, since adherence of the graft lenticule depends on the ability to maintain an air bubble in the anterior chamber. [bib_ref] Factors affecting DSAEK graft lenticle adhesion: an in vitro experimental study, Vaddavalli [/bib_ref] This paper describes a modified technique, which involves using the presence of the Sheets glide in the anterior chamber to eliminate the challenges posed by anatomic abnormalities.
## Surgical technique
During Descemet's Stripping Automated Endothlial Keratopasty (DSAEK) surgery the Sheets glide, trimmed to 4mm, was inserted through the clear corneal wound made via the 4.1mm keratome blade. Viscoelastic was placed on the glide to protect the endothelial side of the DSAEK graft. The "needle-push"/Fichman glide technique 1,2 was used to facilitate the insertion of the donor graft, with the use of a bent 27-gauge. The glide was deliberately maintained in the eye. After confirming the correct orientation and centration of the graft, air was introduced into the anterior chamber to create a large air bubble between the graft and the glide. After 10 minutes, the glide was removed, and the main wound was closed with interrupted 10-0 nylon sutures (Video 1).
Supplementary video related to this article can be found at https:// doi.org/10.1016/j.ajoc.2019.100497.
## Case reports
## Case 1
DSAEK surgery was recommended in this woman with Fuchs' corneal dystrophy, who presented with corneal decompensation following cataract surgery. Intra-operatively, she was found to have floppy iris syndrome.
The DSAEK procedure was performed using the technique described above. The Sheets glide acted as a physical barrier that could "hold" down the floppy iris. Through a paracentesis incision, the anterior chamber was filled with air with the Sheets glide in place. We recommend a "tight" fit of the trimmed Sheet glide through the main T wound-the Sheets glide trimmed to 4mm, was placed in the 4.1mm wound. The tight fit ensured minimal to no escape of air through the wound during this step. The Sheets glide prevented the air bubble from escaping posteriorly behind the iris and maintained chamber stability . This case demonstrated that this technique worked well for floppy iris, as the Sheets glide provided a physical barrier to prevent iris prolapse through the wound.
At 1 month following the procedure, vision had improved from Counting Fingers (CF) vision pre-operatively, to 20/70. At 8 months following the procedure, vision improved to 20/40 and the cornea remained clear.
## Case 2
This is a woman who presented with a failed DSAEK graft. She had a history of advanced glaucoma, with prior glaucoma drainage implantation and cataract surgery. She had an enlarged pupil, iris defects, and inferior iridocorneal touch to the DSAEK donor graft, with peripheral anterior synechiae. A repeat DSAEK surgery with synechialysis was recommended. The DSAEK surgery was performed in the abovedescribed method. In this case, the glide acted as a "scaffold." The enlarged pupil as well as iris defects pose a challenge during surgery, especially during the introduction of air into the anterior chamber, as air can easily migrate posteriorly behind the pupil. The glide allowed for air bubble to be easily introduced in the anterior chamber by preventing posterior migration of the air.
At 3 months, vision improved from 20/800 prior to the procedure, to 20/50, and the cornea was clear.
## Case 3
This patient had undergone pars plana vitrectomy and retinal detachment repair surgery in 2006. Then, he underwent cataract surgery that was complicated by posterior capsular rupture and an anterior chamber intraocular lens (ACIOL) was placed. He had bullous keratopathy, an enlarged pupil, as well as chronic cystoid macular edema at the time of presentation. Lens explantation and DSAEK surgery were recommended.
The ACIOL was explanted via a superior scleral tunnel wound, which was closed immediately. The DSAEK surgery proceeded in the afore-mentioned method. Again, the Sheets glide acted as a "scaffold." In this case, the enlarged pupil in a vitrectomized, aphakic patient posed a challenge in maintaining air in the anterior chamber. In addition, there was also the risk of graft dislocation into the posterior segment. The glide helped to prevent posterior graft dislocation and posterior migration of the air bubble, while maintaining air in a wellformed anterior chamber.
At 6 months following the procedure, vision improved to 20/40 with an aphakic contact lens, with near-resolution of macular edema.
# Discussion
While Descemet's Stripping Automated Endothelial Keratopasty (DSEK) surgery has improved the outcomes of patients needing corneal transplantations, it has its own complications, the most common of which being donor graft detachment 5,6, . [bib_ref] Complications after endothelial keratoplasty: three years of experience, Villarrubia [/bib_ref] Previously, higher intra-operative and immediate post-operative intra-ocular pressure has been shown to facilitate DSEK donor graft lenticule adhesion to the recipient. [bib_ref] Factors affecting DSAEK graft lenticle adhesion: an in vitro experimental study, Vaddavalli [/bib_ref] Conversely, a complete air-fluid exchange has been shown to decrease graft lenticule adhesion. [bib_ref] Factors affecting DSAEK graft lenticle adhesion: an in vitro experimental study, Vaddavalli [/bib_ref] Thus, it becomes imperative that an air bubble is successfully introduced and maintained in the anterior chamber in order to allow for successful graft attachment.
Understandably, abnormalities of the anterior segments would pose surgical challenges during this step of the surgery when air needs to be maintained in the anterior chamber. These abnormalities include aniridia, enlarged pupil, intraoperative floppy iris syndrome, aphakia and hypotony from previous history of vitrectomy. [bib_ref] Modified surgical technique for improving donor adherence in DSAEK in the aphakic..., Titiyal [/bib_ref] All these factors could lead to posterior migration of the air and inability to maintain air in the chamber. In addition, eyes that are aphakic, or have undergone previous vitrectomy and complicated intraocular lens placements, are at risk for graft dislocation into the posterior segment. [bib_ref] Dislocation of the donor graft to the posterior segment in descemet stripping..., Afshari [/bib_ref] Graft dislocation into the posterior segment is a rare but serious complication, which could lead to permanent vision loss. [bib_ref] Dislocation of the donor graft to the posterior segment in descemet stripping..., Afshari [/bib_ref] This modified technique uses the Sheets glide as a "scaffold" to maintain air in the anterior chamber while preventing posterior migration of air, as well as posterior dislocation of the graft. The sizing of the wound is critical as the 4mm Sheets glide should fit tightly into the wound. We recommend using a 4.1mm keratotome blade for the clear corneal incision. In all three cases, there was no issue with air escaping via the wound, nor were there issues with wound gape/wound-related complications. In addition, the presence of the Sheets glide provided a physical barrier to hold the iris down, as seen in Case 1 with floppy-iris, thus preventing iris prolapse.
Since the Sheets glide only covers the central 4mm of the iris-plane, large peripheral iridotomies or iris defects could theoretically allow unwanted posterior air migration. However, as seen in the description of Case 2 with prior glaucoma surgery and large iris defects, and Case 3 with aphakia, this technique worked well in terms of introducing and maintaining an air bubble. We recommend a controlled introduction of a single large air-bubble with the tip of cannula positioned near the pupillary axis while aimed anteriorly towards the graft, to allow for successful completion of this technique .
One point of consideration, is that at the end of the 10 minutes of complete air-fill, care must be taken to remove the Sheets glide in a gentle, controlled manner to prevent loss of air during this step. As The Sheets glide is inserted into the anterior chamber through the main clear-corneal incision. B) The DSAEK lenticule is inserted via the "needle-push" technique. C) An air bubble is injected through the paracentesis wound, while the Sheets glide is intentionally left in the eye. D) The Sheets glide is removed carefully from the eye. E) Two interrupted 10-0 nylon sutures are placed at the wound. F) The air bubble is reduced to approximately 2 mm inside the margins of the donor graft.
shown in and in the video, most of the air did in fact remain in the anterior chamber as seen by the near-complete air-fill (parts D and E in after the Sheets glide removal. In all three cases, after closure of the wound, the air-bubble was deliberately reduced via the paracentesis wound to 2mm inside the margins of the donor graft (part F in as would be normally done in a DSAEK procedure.
Previously, authors have shown that continuous fluid infusion in the vitrectomized posterior segment is helpful in aphakic, vitrectomized eyes and enhances graft adherence. [bib_ref] Modified surgical technique for improving donor adherence in DSAEK in the aphakic..., Titiyal [/bib_ref] Others have described suturing the graft lenticule in eyes with altered anatomy. [bib_ref] Endothelial keratoplasty technique for aniridic aphakic eyes, Price [/bib_ref] [bib_ref] AC Suture technique for Descemet stripping and endothelial keratoplasty, Macsai [/bib_ref] [bib_ref] Transcorneal suture fixation of posterior lamellar grafts in eyes with minimal or..., Patel [/bib_ref] This paper describes a modified technique using the Sheets glide to enable air bubble maintenance in the anterior chamber, allowing successful graft adherence in challenging eyes.
## Patient consent
Written consent to publish this case has not been obtained. This report does not contain any personal identifying information.
## Authorship
The author attests that she meets the current ICMJE criteria for Authorship.
# Disclosure of funding
None.
# Funding
No funding or grant support.
## Conflicts of interest
The following author E.K., has no financial disclosures.
AcknowledgementsNone. |
Resin Diterpenes from Austrocedrus chilensis
Seventeen diterpenes belonging to the labdane, abietane and isopimarane skeleton classes were isolated from the resin of the Chilean gymnosperm Austrocedrus chilensis and identified by spectroscopic and spectrometric methods. The diterpene 12-oxo-labda-8(17),13E-dien-19 oic acid is reported for the first time as a natural product and 14 diterpenes are reported for the first time for the species.
# Introduction
The gymnosperm tree Austrocedrus chilensis (D. Don) Florin et Boutelje (Syn.: Austrocedrus chilensis (D.Don) Pic. Ser. et Bizz) (Cupressaceae) is known as "ciprés de cordillera" and grows in sandy soils in the Eastern Andean slopes up to 2,000 m over sea level. The plant leaves has been used as a sudorific and the powdered fruits to treat diarrhea. Little information exists on other medicinal uses of the plant, but the wood is highly appreciated. The monoterpene carvacrol and β-thujaplicin (4-isopropyl tropolone = hinokitiol) and the flavonoid taxifolin were isolated from the wood of A. chilensis. An ethanol extract from the aerial parts of the tree, including leaves, twigs and stems, afforded the lignan desoxypodophyllotoxin and the diterpenes 8,20-dihydroxy-8(11),13-abietadien-12one and pisiferol [bib_ref] The distribution of lignanoids in the order coniferae, Castro [/bib_ref] [bib_ref] Plant anticancer agents. XIII. Constituents of Austrocedrus chilensis, Cairnes [/bib_ref]. Flavonoids were identified from a 70% ethanol extract of leaves and branchlets OPEN ACCESS of A. chilensis [bib_ref] Biflavones of the subfamily Callitroideae, Gadek [/bib_ref] , while sugiol and another unidentified diterpene were isolated from a tree bark methanol extract [bib_ref] Extractable compounds of native trees. Chemical and biological study I: Bark of..., Flores [/bib_ref].
However, scarce information is available on the composition of its resin. Cox et al. [bib_ref] Oxygenated di-and tricyclic diterpenoids of shouthern hemisphere conifers, Cox [/bib_ref] , working on a comparison of gymnosperm resins using hyphenated techniques, described the identification of two labdane/clerodane C-19 acids, sandaracopimaric acid, ferruginol, 2,3-dehydroferruginol, 6,7-dehydroferruginol and 7-oxoferruginol from the external resin of A. chilensis.
The investigation of gymnosperm resin composition has been carried out using chromatographic means for the isolation and further identification of the constituents by spectroscopic and spectrometric methods [bib_ref] Labdane acids from polar extracts of Juniperus thurifera, Feliciano [/bib_ref] [bib_ref] Further diterpene acids from Juniperus thurifera, Feliciano [/bib_ref] [bib_ref] Diterpene composition of Cupressus dupreziana and Cupressus sempervirens, Piovetti [/bib_ref]. New techniques used to characterize the resin constituents include gas chromatography coupled to mass spectrometry (GC-MS) [bib_ref] Sesqui-and diterpenoid biomarkers preserved in Taxodiumrich Oligocene oxbow lake clays, Weisselster basin, Otto [/bib_ref] [bib_ref] Chemosystematics and diagenesis of terpenoids in fossil conifer species and sediment from..., Otto [/bib_ref] [bib_ref] Resin compounds from the seed cones of three fossil conifer species from..., Otto [/bib_ref] [bib_ref] Characterization of archaeological frankincense by gas chromatography-mass spectrometry, Mathe [/bib_ref] , and proton magnetic resonance spectroscopy [bib_ref] Distinctions among conifer exudates by proton magnetic resonance spectroscopy, Lambert [/bib_ref].
Following our studies on native South American plants, we have now examined the composition of resin exudates from a mature population of Austrocedrus chilensis trees to establish suitable conditions for metabolic profiling.
# Results and discussion
Seventeen diterpenes with isopimarane, labdane and abietane skeletons were isolated from the resin of Austrocedrus chilensis, and identified by spectroscopic and spectrometric methods. Fourteen of these diterpenes are reported for the species for the first time. The compounds were identified by spectroscopic means and by comparison of their spectral data with literature values. The 13 C-NMR data of compounds 4, 6, 12-14a is shown in [fig_ref] Table 1 13: C-NMR data of the compounds 4, 6, 12, 13 and 14a [/fig_ref]. The structures of the compounds identified in the resin are shown in [fig_ref] Figure 1: Structure of the compounds isolated and/or identified from the resin of Austrocedrus... [/fig_ref]. From the compounds identified in the present work, only sandaracopimaric acid (compound 5), ferruginol (compound 15), and 7-oxoferruginol (sugiol, compound 17) were previously reported [bib_ref] Oxygenated di-and tricyclic diterpenoids of shouthern hemisphere conifers, Cox [/bib_ref] in a resin sample of this species collected in southern Chile. On the other hand, the previously reported diterpenes 2,3-dehydroferruginol and 6,7-dehydroferruginol [bib_ref] Oxygenated di-and tricyclic diterpenoids of shouthern hemisphere conifers, Cox [/bib_ref] were not identified in our samples. The differences can be explained due to different plant populations and collection time. [bib_ref] Labdanes from Cryptomeria japonica, Su [/bib_ref]. In a new article, additional labdanes, abietanes and pimaranes were identified, including junicedric acid, 13-epi-cupressic acid methyl ester, copalol, 13-oxo-14,15-dinorlabd-8(17)-en-19 oic acid methyl ester, trans-communic acid, cis-communic acid, 19-acetoxyferruginol, sugiol methyl ether, 6α-hydroxydemethyl-cryptojaponol, 5,6-dehydrosugiol methyl ether, cupresol, nejukol, isopimarinol and isopimaric acid [bib_ref] Diterpenoids from leaves of Cryptomeria japonica, Su [/bib_ref].
The compound 14 was isolated as the corresponding methyl ester and is described as a new natural product. The structure of compound 14 follows from the HR-MS indicating a molecular formula C 20 H 30 O 3 and C 21 H 32 O 3 after methylation (compound 14a), accounting for six degrees of unsaturation (i.e., three double bonds, two rings and one carbonyl function). The IR spectrum of the methyl ester shows an ester function and α,β-unsaturated ketone at 1,724 and 1,672 cm −1 , respectively. In the 1 H-NMR spectrum, an olefinic side chain proton at δ 6.77 (H-14) coupled with two allyl methyl groups at δ 1.84 (H-15) and 1.74 ppm (H-16), indicates an α,β-unsaturated ketone system in the side chain of the diterpene. The E-configuration of the side chain double bond follows from the chemical shift of the olefinic proton and the C-15 methyl group and is in agreement with the data reported for related diterpenes isolated from the liverwort Scapania undulata [bib_ref] Scapaundulins A and B, two novel dimeric labdane diterpenoids and related compounds..., Yoshida [/bib_ref].
The pair of dd at δ 2.94 (17.5, 10.9) and δ 2.50 (17.5, 3) ppm (H-11) places the ketone at C-12. An exo methylene (δ 4.67, s, H-17 and δ 4.23, s, H-17′), a methoxy group at δ 3.60 and two quaternary methyl groups appears as singlets at δ 1.18 (H-18) and 0.54 ppm (H-20), respectively, indicating that the diterpene has a labdane skeleton. The 1 H-NMR data of compound 14 are summarized in the Experimental section. The 13 C-NMR spectrum is in agreement with the proposed structure. Related compounds with a hydroxy function at C-15 were reported from Chloranthus henryi including henrilabdane A and the 13-oxo derivative henrilabdane C [bib_ref] Bis-sesquiterpenes and diterpenes from Chloranthus henryi, Li [/bib_ref] and 12,13RS-dihydroxycommunic acid from Platycladus orientalis [bib_ref] Diterpenoids from the pericarp of Platycladus orientalis, Wang [/bib_ref]. GC trace of the Austrocedrus chilensis resin (as methyl esters). Compounds were identified by comparison with standards isolated from the resin and comparison of the mass fragmentation patterns with literature. Compounds: 1: 18-hydroxy isopimar-15ene; 4: Isopimara-8(9),15-dien-19-ol; 5a: Sandaracopimaric acid methyl ester; 9: Torulosal; 10a: Torulosic acid methyl ester; 11: 8(17),12,14-labdatriene; 12a: E-communic acid methyl ester; 13a: Z-communic acid methyl ester; 14a: 12-oxo-labda-8(17),13E-dien-19 oic acid methyl ester; 15: Ferruginol; 16a: Dehydroabietic acid methyl ester. For compounds 6, 9 and 10, M + -water was detected instead of the molecular ion. The resin sample A was obtained from a female tree while the sample B was taken from a male tree from the same population. Both samples were collected in November 2010. For compound structures please see [fig_ref] Figure 1: Structure of the compounds isolated and/or identified from the resin of Austrocedrus... [/fig_ref].
The compounds isolated and identified from the A. chilensis resin are constituents occurring in several gymnosperms and in the Cupressaceae family. The C-13 epimer of compound 3 was previously described in a study of cationic rearrangements and cyclizations of diterpenes [bib_ref] Cationic rearrangements and cyclizations of diterpenoid olefins, Hall [/bib_ref] as well as from Phoma betae [bib_ref] Diversity of diterpene hydrocarbons in fungus Phoma betae, Oikawa [/bib_ref]. Compound 4 was identified as 8,15-isopimaradien-19-ol and shows clear differences with the ent-isopimaranes isolated from Calceolaria polifolia [bib_ref] Diterpenes from Calceolaria polifolia, Chamy [/bib_ref] and C. latifolia [bib_ref] Diterpenes from Calceolaria latifolia, Garbarino [/bib_ref]. Sandaracopimaric acid (5) as well as several abietanes were isolated from the endemic species Taxus mairei occurring in Taiwan [bib_ref] Diterpenes from Taxus mairei, Yang [/bib_ref] , Juniperus chinensis [bib_ref] Diterpenes from the bark of Juniperus chinensis, Fang [/bib_ref] and other gymnosperms. 8(17),14-Labdadien-13-ol (6) was previously reported from Cupressus sempervirens, C. torulosa, Tetraclinis articulate, Vitex rotundifolia, Cupressus dupreziana and C. sempervirens [bib_ref] Diterpene composition of Cupressus dupreziana and Cupressus sempervirens, Piovetti [/bib_ref]. The 13 C-NMR data of compound 6 agree with that reported for epimanool derivatives [bib_ref] Labdanes from Cryptomeria japonica, Su [/bib_ref]. Compounds 6-10 were described as constituents of Cryptomeria japonica leaves [bib_ref] Diterpenoids from leaves of Cryptomeria japonica, Su [/bib_ref]. Torulosal (9) has also been previously isolated from Cupressus torulosa, Araucaria cooki, Tetraclinis articulate and Dacrydium biforme. The communic acid isomers (compounds 12 and 13) has been reported from several Cupressaceae, including Juniperus chinensis [bib_ref] Diterpenes from the bark of Juniperus chinensis, Fang [/bib_ref] and ferruginol [bib_ref] Characterization of archaeological frankincense by gas chromatography-mass spectrometry, Mathe [/bib_ref] has been found in several Podocarpaceae and gymnosperms [bib_ref] Phenolic diterpenoids of Podocarpus ferrugineus and other Podocarps, Cambie [/bib_ref]. Dehydroabietic acid 16 is a common constituent in resins from gymnosperms and 7-oxoferruginol (sugiol, compound 17) was identified [bib_ref] Oxygenated di-and tricyclic diterpenoids of shouthern hemisphere conifers, Cox [/bib_ref] from a Chilean resin sample of A. chilensis.
Overall, the diterpene composition of the South American Cupressaceae Austrocedrus chilensis shares similarities with Juniperus and Cryptomeria species, including the occurrence of E-and Z-communic acid, sandaracopimaric acid, ferruginol and several labdane and pimarane diterpenes [bib_ref] Diterpenes from the bark of Juniperus chinensis, Fang [/bib_ref] [bib_ref] Labdanes from Cryptomeria japonica, Su [/bib_ref]. Labdane diterpenes have been reported from other Cupressaceae including Platycladus orientalis [bib_ref] Terpenoids from the seed of Platycladus orientalis, Inoue [/bib_ref] , while Juniperus brevifolia afforded sandaracopimaranes and abietanes [bib_ref] Diterpene constituents of leaves from Juniperus brevifolia, Seca [/bib_ref]. Juniperus phoenicea and J. thurifera var. africana yielded several abietane and pimarane derivatives [bib_ref] Oxygenated diterpenes and other constituents from Moroccan Juniperus phoenicea and Juniperus thurifera..., Barrero [/bib_ref]. Labdanes and pimarane acids have been isolated from Juniperus thurifera [bib_ref] Labdane acids from polar extracts of Juniperus thurifera, Feliciano [/bib_ref] [bib_ref] Further diterpene acids from Juniperus thurifera, Feliciano [/bib_ref].
Several of the diterpenes occurring in the A. chilensis resin have been found to display relevant biological activities in different in vitro as well as in in vivo systems. Communic acid has been reported as a selective COX-2 enzyme inhibitor from Curcuma mangga [bib_ref] Labdane diterpenes in Curcuma mangga rhizomes inhibit lipid peroxidation, cyclooxygenase enzymes and..., Liu [/bib_ref] and antimycobacterial compound in Juniperus communis [bib_ref] Antimycobacterial terpenoids from Juniperus communis L. (Cuppressaceae), Gordien [/bib_ref]. Ferruginol present gastroprotective and cytotoxic activity [bib_ref] Gastroprotective and cytotoxic effect of semisynthetic ferruginol derivatives, Areche [/bib_ref] [bib_ref] Potencial antitumor promoting diterpenoids from the stem bark of Thuja standishii, Iwamoto [/bib_ref] , antiproliferative [bib_ref] Ferruginol suppresses survival signaling pathways in androgen-independent human prostate cancer cells, Bispo De [/bib_ref] , anti-staphylococcal [bib_ref] Antibacterials and modulators of bacterial resistance from the immature cones of Chamaecyparis..., Smith [/bib_ref] and antiplasmodial effect [bib_ref] Synthesis of totarol amino alcohol derivatives and their antiplasmodial activity and cytotoxicity, Clarkson [/bib_ref] , among others. The related diterpene 7-oxoferruginol was associated with the anti-inflammatory and hepatoprotective effect of Cryptomeria japonica [bib_ref] Hepatoprotective phytocompounds from Cryptomeria japonica are potent modulators of inflammatory mediators, Shyur [/bib_ref]. Dehydroabietic acid and its derivatives present gastroprotective effect in animal models of induced gastric ulcers [bib_ref] Gastroprotective and cytotoxic effect of dehydroabietic acid derivatives, Sepúlveda [/bib_ref]. The abietane diterpenes 20-hydroxyferruginol and 6α-hydroxysugiol from the cones of Sequoia sempervirens, have been shown to present antitumour effect in oncogenic H-ras transformed cells and also against the following human tumor cell lines: colon (SW620 and HCT116), breast (MDA-MB-231) and lung (NCI-H23 and A 549) [bib_ref] Anti-tumor abietane diterpenes from the cones of Sequoia sempervirens, Son [/bib_ref]. Labdane diterpenes from Thuja standishii stem bark including 6α-hydroxysugiol, 12S-hydroxylabda-8(17), 13(16),14-trien-19-oic acid, 12-methoxyabieta-8,11,13-trien-11-ol and 15-oxolabda-8(17),11Z,13Etrien-19 oic acid displayed inhibitory effects as potential cancer chemopreventive agents [bib_ref] Anti-tumor promoting diterpenes from the stem bark of Thuja standishii (Cupressaceae), Iwamoto [/bib_ref].
While the composition of resin samples from a female and a male tree showed the same main compounds, the relative ratio of the diterpenes was different. This fact can be related to several factors, including gender, individual, seasonal, or response to pathogens, among others. However, a much larger number of samples should be analyzed to disclose the significance of the present findings. The establishment of a chromatographic method for the fast and reliable identification of the resin constituents opens new possibilities for the comparative study of populations of this tree as well as the response of A. chilensis to environmental stress and microorganisms [bib_ref] Hehl-Lange, S. Plant health and global change-Some implications for landscape management, Pautasso [/bib_ref] [bib_ref] Pathogenicity of Phytophthora austrocedrae on Austrocedrus chilensis and its relation with mal..., Greslebin [/bib_ref] [bib_ref] Phytophthora austrocedrae sp. nov., a new species associated with Austrocedrus chilensis mortality..., Greslebin [/bib_ref] [bib_ref] Stand development patterns as a consequence of the mortality in Austrocedrus chilensis..., Amoroso [/bib_ref] , including the pathogenic fungus Phytophthora austrocedrae.
## Experimental
## General
Optical rotations were obtained for solutions in CHCl 3 (concentrations expressed in g/100 mL) on a Jasco DIP 370 polarimeter (Jasco Analytical Instruments, Easton, MD, USA). IR spectra were recorded on a Nicolet Nexus FT-IR instrument (Thermo Electron Corporation, Waltham, MA, USA). All NMR experiments were performed on a Bruker Avance 400 NMR spectrometer (Bruker BioSpin GmbH, Rheinstetten, Germany) equipped with a 5 mm inverse detection z-gradient probe. The 1 H and 13 C spectra (at 400 and 100 MHz, respectively) were measured at room temperature (22-23 °C) using CDCl 3 as solvent. Chemical shifts are given on the scale and were referenced to residual CHCl 3 at 7.25 ppm for 1 H spectra and to the solvent at 77.00 ppm for 13 C spectra. One-dimensional 1 H and 13 C-NMR spectra were acquired under standard conditions. The pulse programs of the COSY, gHSQC, gHMBC experiments were taken from the Bruker software library. Homonuclear twodimensional spectra (COSY) and inverse proton-detected heteronuclear two-dimensional spectra (gHSQC) were acquired in the phase-sensitive mode and gHMBC spectra were acquired in the absolute value mode. The data for the gHSQC spectra were collected in a 1024 × 256 matrix with a spectral width of [bib_ref] Plant anticancer agents. XIII. Constituents of Austrocedrus chilensis, Cairnes [/bib_ref] Compounds were characterized by electron-ionization (EI) mass spectra. Retention time (Rt, min) and MS fragmentation patterns of the known compounds were compared with literature.
# Gc-ms analysis
# Plant material
The resin of Austrocedrus chilensis was collected from a mature tree population growing at Las Trancas, VIII Region, Chile (36°54′03′′S, 31°32′47′′W). Voucher herbarium specimens have been deposited at the Herbario de la Universidad de Talca and were identified by Patricio Peñailillo. The voucher specimens correspond to the individuals from those the resin was obtained.
## Isolation of the resin constituents
The crude resin (362 g) was extracted with a 1:1 dichloromethane (DCM)-ethyl acetate (EtOAc) mixture. Some 110.58 g of solubles were obtained. The resin was adsorbed on silica gel and submitted to flash chromatography with a petroleum ether (PE)/ethyl acetate (EtOAc) gradient (PE/EtOAc 95:5, 90:10, 80:20, 70:30, 60:40, 50:50 0:100). The volumen of each fraction was 1 L. After TLC analysis (SiO 2 , PE/DCM 1:1 and PE/EtOAc 9:1), fractions with similar patterns were pooled as follows: 1 (280 mg), 2 (20 mg), 3 (7 g), 4 (16 g), 5 (17.5 g), 6 (12 g), 7 (11 g) and 8 (15 g) (total: 78.8 g).
Fractions 1 and 2 containing mainly non-polar compounds were submitted to GC-MS. The GC-MS chromatogram of both fractions showed a main compound, identified as 3 according to the MS fragmentation pattern (NIST database) and interpretation of the spectra, with 4 as a second product. The GC-MS profile of fractions 3 and 4 presented a main constituent identified as 8(17), 12,14-labdatriene (11), the alcohol 1 and the compound 3. Fractions 5, 6 and 7 differed only in the relative proportion of constituents. Part of the fraction 6 (12 g) was permeated on a Sephadex LH-20 column (column length 125 cm, Sephadex content 60 cm, i.d. 5 cm) with a PE/DCM/MeOH 1:1:1 mixture in two batches (2 × 6 g each). Some 32 fractions of 20 mL each were collected and pooled together according to the TLC patterns (SiO 2 , PE/EtOAc 8:2). Fractions 1-10 did not contained compounds of interest and were discarded. The pool of fractions 11-14 (1.67 g) was chromatographed on a silica gel column (190 g silica gel, 70 cm length, 4 cm i.d.) with a PE/EtOAc 8:2 mixture. Some 44 fractions of 25 mL each were collected and grouped after TLC analysis as follows: 1-2 (discarded), 3-5 (319 mg) with manool (6) as the main constituent and the acetate 2 as a minor compound, 6-7 (42 mg manool 6), 8-9 (32 mg isopimara-8(9),15-dien-19-ol, 4), 10-21 (350 mg) and 22-43 (638 mg, complex mixture). The fraction pool 10-21 afforded after preparative TLC on silica gel (toluene/EtOAc 9:1) 64 mg of a mixture of 18-acetoxymanool (8) and torulosal [bib_ref] Labdane acids from polar extracts of Juniperus thurifera, Feliciano [/bib_ref].
The fraction pool 15-19 (1.92 g) was chromatographed on a RP-8 silica gel column with MeOH/H 2 O 9:1 as the mobile phase in three batches. Some 90 fractions of 5 mL each were collected and pooled after TLC to afford a mixture of E-communic acid (main compound) and Z-communic acid (compounds 12 and 13) in a 3:1 ratio (350 mg) and sandaracopimaric acid (90 mg, 5). A representative sample of the fractions 1-32 (136 mg) was treated with diazomethane. After preparative TLC (silica gel, PE/EtOAc/acetone 9:2:1) 51 mg of torulosic acid methyl ester (10a) and 33 mg of a mixture of torulosic acid methyl ester (10a) and the alcohol 7 were obtained. After methylation, preparative TLC of fraction pool 20-23 (toluene/DCM/Et 2 O 4.5:4.5:1) yielded 12.3 mg of compound 14a and 13 mg of a mixture of the compounds 6 and 7 in a 4:1 ratio.
# Conclusions
The composition of A. chilensis resin was established by spectroscopic and spectrometric means. A GC-MS method was developed for the fast identification of the diterpene constituents in the resin, setting the conditions for new studies on the physiological response of the tree under different environmental stimuli.
[fig] Figure 1: Structure of the compounds isolated and/or identified from the resin of Austrocedrus chilensis. [/fig]
[fig] Figure 2: Figure 2. GC trace of the Austrocedrus chilensis resin (as methyl esters). Compounds were identified by comparison with standards isolated from the resin and comparison of the mass fragmentation patterns with literature. Compounds: 1: 18-hydroxy isopimar-15ene; 4: Isopimara-8(9),15-dien-19-ol; 5a: Sandaracopimaric acid methyl ester; 9: Torulosal; 10a: Torulosic acid methyl ester; 11: 8(17),12,14-labdatriene; 12a: E-communic acid methyl ester; 13a: Z-communic acid methyl ester; 14a: 12-oxo-labda-8(17),13E-dien-19 oic acid methyl ester; 15: Ferruginol; 16a: Dehydroabietic acid methyl ester. For compounds 6, 9 and 10, M + -water was detected instead of the molecular ion. The resin sample A was obtained from a female tree while the sample B was taken from a male tree from the same population. Both samples were collected in November 2010. For compound structures please see Figure 1. [/fig]
[fig] Figure 3: Flow chart summarizing the isolation of constituents from Austrocedrus chilensis resin. Some 4.06 g of fraction 8 were permeated on a Sephadex LH-20 column (125 cm length, 5 cm i.d., 60 cm Sephadex filling), with a PE:DCM:MeOH 1:1:1 mixture. Some 34 fractions of 20 mL each were collected and pooled according to the TLC patterns to afford after additional CC on silica gel, 10 mg of the acetate 2 with traces of the corresponding aldehyde, 2.5 mg manool (6) and 15.6 mg of the acetate 8 with traces of torulosal (9), torulosic acid (10, 362 mg) and traces of 7-oxoferruginol (sugiol, 17). The isolation of the compounds is summarized inFigure 3. [/fig]
[fig] 3. 5: Compound Characterization 18-Hydroxyisopimar-15-ene (1). GC-MS: Rt 14.98 min; C 20 H 34 O (290); m/z 290 (11), 275 (34), 257 (40), 179 (20), 137 (41), 109 (65), 95 (66), 81 (79), 69 (84), 55 (100). [/fig]
[fig] 18 -: Acetoxyisopimar-15-ene (2). [α] D 20 : +82.7 (c 0.10, CHCl 3 ); IR v max (film) 2928, 2864, 1736, 1461, 1378, 1242. GC-MS: Rt 16.33 min; C 22 H 34 O 2 (330); m/z 330 (9), 315 (7), 288 (6), 270 (9), 255 (70), 187 (30), 105 (47), 91 (53), 43 (100). Isopimara-8(9),15-diene (3). GC-MS: Rt 14.13 min; C 20 H 32 (272); m/z 272 (11), 257 (34), 137 (53), 107 (53), 95 (89), 81 (100), 69 (91), 55 (89). [/fig]
[table] Table 1 13: C-NMR data of the compounds 4, 6, 12, 13 and 14a (100 MHz, CDCl 3 , δ-values). [/table]
|
Candidate Gene Association Studies of Anthracycline-induced Cardiotoxicity: A Systematic Review and Meta-analysis
SupplementaryTable 3: Genetic polymorphisms, association with ACT and HWE status. This table summarised the findings on the association between each of the 147 SNPs in 84 genes and anthracycline-induced cardiotoxicity. (Words) Search StrategiesOVID Medline, Cochrane Central Register of Controlled Studies, EMBASE and AMED Text word search(1) anthracycline*.mp.; (2) doxorubixin*.mp.; (3) daunorubicin*.mp. (4) epirubicin*.mp. (5) idarubicin*.mp. (6) 1 or 2 or 3 or 4 or 5; (7) cardiotoxicity*.mp.; (8) cardiomyopathy*mp.; (9) heart*.mp.; (10) failure*.mp.; (11) 9 and 10; (12) arrhythmia*.mp.; (13) 7 or 8 or 11 or 12; (14) genetic*.mp.; (15) polymorphism*.mp.; (16) pharmacogenomics*.mp.; (17) variant*.mp.
## Gene
Nucleotide change |
The Muscularity-Oriented Eating Test, Drive for Muscularity Scale, and Muscle Dysmorphic Disorder Inventory among Chinese Men: Confirmatory Factor Analyses
Citation: He, J.; Murray, S.; Compte, E.J.; Song, J.; Nagata, J.M. The Muscularity-Oriented Eating Test, Drive for Muscularity Scale, and Muscle Dysmorphic Disorder Inventory among Chinese Men: Confirmatory Factor Analyses. Int. J.Abstract: Research on eating disorders (EDs) and body image disturbances has focused mostly on females from Western countries, and little is known about EDs in male populations in China, which is partially due to the lack of validated assessment measures. The current work aims to translate the Muscularity-Oriented Eating Test (MOET), Drive for Muscularity Scale (DMS) and Muscle Dysmorphic Disorder Inventory (MDDI) into Chinese and examine their psychometric properties. The factor structures, reliability and validity of the translated scales were examined with two samples: male university students (n = 295, M age = 18.92 years) and general adult men (n = 406, M age = 28.53 years). With confirmatory factor analyses, the original factor structures are replicated for the MOET, DMS and MDDI. The results also support the adequate internal consistency for both samples. Strong evidence of convergent and incremental validity for the three measures is also found in both samples. Overall, the three measures prove to be good instruments for use among Chinese male university students and general adult men.MOET contains 15 items that are rated on a 5-point Likert-type scale from 0 ("never true") to 4 ("always true"). In the work of Murray et al.[6], the MOET showed a unidimensional structure, high internal consistency (omega = 0.92~0.93), high test-retest correlation (r = 0.75) and construct validity, as indicated by large correlations between the MOET scores and theoretically closely related measures (e.g., drive for muscularity) but small correlations between the MOET scores and theoretically weak or unrelated measures (e.g., socially desirable response). Currently, the MOET has been translated into Spanish, and the Spanish version replicated the unidimensional structure and showed good psychometric properties in a sample of Argentinian college men[10].To date, research on EDs or body image disturbances among men is mainly conducted in Western societies among adolescents or young adults of Caucasian ethnicity [2]. However, there has been a rising trend of ED prevalence in Asian countries, including China[11,12], and more attention has been called to EDs in Chinese men[13]. Thus, it is imperative and significant to conduct more research on EDs in Chinese male populations for exploring whether there are unique clinical manifestations, risk factors and health consequences of EDs in Chinese men.Furthermore, as body image is a key contributing factor of EDs[14], it is also important to assess male-specific body image disturbances to explore the etiology of EDs among men. However, different from women and girls, the major body image concern among men and boys is drive for muscularity[15][16][17], which refers to an individual's desire to achieve an ideal muscular body[15]. Previous studies have shown that drive for muscularity is a significant risk factor for EDs among men[3,18,19]. Furthermore, drive for muscularity is also a significant risk factor or characteristic of muscle dysmorphia (MD)[20][21][22]. Specifically, as a subtype of body dysmorphic disorder, muscle dysmorphia (MD) represents the pathological extreme pursuit of muscularity and is characterized by a pervasive belief or fear around insufficient muscularity[23,24]. Muscle dysmorphia is mostly researched among men, and it leads to considerable functional impairment[25].However, to the best our knowledge, there have been no such measures available in Chinese for assessing male-specific body image disturbances or muscle dysmorphia. To bridge this gap, another two widely used measures assessing male-specific body image disturbances were also chosen to be translated and validated in the current study: the Drive for Muscularity Scale (DMS)[15]and the Muscle Dysmorphic Disorder Inventory (MDDI)[26].Specifically, the DMS is a 15-item self-report questionnaire rated on a 6-point Likerttype scale with responses from 1 ("always") to 6 ("never"), which has two subscales: the muscle-oriented body image subscale (MBI, or the muscularity attitudes subscale; 7 items) and the muscle-oriented behavior subscale (MB, or the muscularity behavior subscale; 8 items)[15]. All items are reverse-coded, in which the total score (or mean score) of the 15 items then represents an individual's overall drive for muscularity. The higher the total score, the stronger the attitudes and behaviors of drive for muscularity. Currently, the DMS has been translated into a number of different languages and shown adequate reliability and validity, including Romanian [27], Spanish[28,29], French [30], Turkish [31], Portuguese [32] and Japanese [33].The MDDI [26] contains 13 items that comprise three subscales: drive for size (DFS, 5 items), appearance intolerance (AI, 4 items) and functional impairment (FI, 4 items). The MDDI is rated on a 5-point Likert-type scale with responses from 1 ("never") to 5 ("always"). A total score can be derived by the sum of all items or all subscales. Higher total scores indicate more symptoms of muscle dysmorphia. The MDDI has also been translated into a number of different languages and demonstrated good psychometric properties in samples from different countries, including Portuguese [34], Italian[35], German [25], Turkish[36]and Spanish[37].Overall, the current study aimed to translate and validate three measures (i.e., the MOET, DMS and MDDI) to facilitate future research on EDs and body image disturbances among Chinese men. We hypothesized that (1) the MOET would have a unidimensional
# Introduction
Eating disorders (EDs) are severe mental illnesses characterized by disturbances in eating behaviors and body image. Traditionally, research on EDs and their key risk factors-body image disturbances-was mainly conducted among women. However, there has been ample evidence showing that EDs and body image disturbances increasingly occur in men, and the presentations of EDs and body image disturbances among men are often distinct to those observed in women. Specifically, according to the previous literature, EDs and body image disturbances among women are typically thinnessoriented (i.e., drive for thinness), while EDs and body image disturbances among men are typically muscularity-oriented (i.e., drive for muscularity).
The majority of the existing measures that are widely used for assessing ED symptomatology and body image disturbances are thinness-focused, such as the Eating Attitude Test, the Eating Disorder Examinationand the Eating Disorder Inventory. Thus, the measures are not sensitive enough to capture the core features of disordered eating or body image disturbances among men.
To fill this gap, Murray et al.developed and validated the Muscularity-Oriented Eating Test (MOET) with a sample of 511 undergraduate men in the US. Specifically, the factor structure, the DMS would have a two-factor structure, and the MDDI would have a three-factor structure; (2) the three measures would have adequate reliably as indicated by internal consistency; (3) the total scores of the three measures would be closely related to each other; and (4) the total scores of the three measures would be positively related to traditional eating disorder symptomology, psychological impairment, psychological distress and thinness-oriented body dissatisfaction.
# Materials and methods
## Participants and procedure
The protocol of data collection was approved by the Institutional Review Board of the Chinese University of Hong Kong, Shenzhen. We used two male samples for the current study: university students and male adults from the general population.
## University students
For the sample of male university students, the survey was conducted in a paper-andpencil format in a comprehensive university in Hunan province. Psychological teachers introduced the information about the project (i.e., the study design) to 1223 first-and second-year undergraduate students during class time and invited them to participate. Two attention check questions were used to ensure the quality of the survey. Finally, 1059 students met our inclusion criterion (i.e., ≥18 years old) and provided written informed consent. Of those, 812 (36.3% males; 295) passed the two attention check questions and were included in the study. As the current study is about male university students' body image and eating behaviors, only the sample of 295 males was used in the current study, and they had a mean (SD) age of 18.92 (1.04) years (range: 18~23 years). The mean (SD) of body mass index (BMI) based on self-reported height and weight
## Male adults from the general population
To recruit general male adults, we used an online Chinese survey platform, Credamo, which is similar to Qualtrics Online Sample. Credamo recruits participants from all over China. In the online survey, participants provided informed consent by selecting the "agree to participate" option at the end of the first page. If any participant did not give his informed consent, the survey would end immediately. Moreover, the participants were also informed that they could stop at any time during the survey by closing the survey page. Two attention check questions were used to ensure the quality of the survey, and surveys from the participants who failed to pass the two attention check questions were removed automatically by Credamo. By setting the criteria of the target sample (i.e., ≥18 years old), 532 male adults filled out our survey, and 406 of them were included by passing the two attention check questions. The included participants' ages ranged from 18 to 53 years old, with a mean (SD) of 28.53 (5.56) years. Their BMIs ranged from 15.87 to 32.96 kg/m 2 , with a mean (SD) of 22.28 (2.80) kg/m 2 .
According to the guidelines to estimate the sample size for confirmatory factor analysis, for a power of 0.80, a root mean square error of approximation value of 0.05 and an alpha level of 0.05, a minimum sample size of 174 participants was required for the MOET, while it was 175 for the DMS and 215 for the MDDI. Thus, the sample sizes of 295 and 406 were adequate for the purposes of the current work.
## Translation
The Chinese versions of the MOET, DMS and MDDI were obtained based on the procedures recommended in the previous literature for cross-cultural translation of scales. Specifically, the English versions of the three measures were first translated to Chinese (Mandarin) by a Ph.D. student and a master's student independently, and both students were proficient in English. Then, the two translations for each of the three measures were compared, and the differences were discussed within the research team, leading to a synthesized translation for each of the three measures. Back-translation was conducted independently by two new bilingual master's students who had no prior knowledge of the original English measures. Afterward, the forward-and back-translations were evaluated by the research team. The research team, as an expert review committee, consisted of the four translators and three experts in eating behaviors and body image. In addition, we successfully involved the developers of the MOET and MDDI into the translation process by requesting them to evaluate the equivalence of the back-translations and the original English versions. As the research team did not identify translation issues, no modifications to the translations were made. Next, the preliminary Chinese translations were administered to 30 male university students (M age = 19.80 years, SD = 1.03) for evaluating the understandability of the items. They were asked to evaluate the understandability of the items on a 5-point Likert scale from 1 ("I do not understand at all") to 5 ("I completely understand, and I have no doubts"). The results showed that the understandability of the MOET ranged from 4.03 to 4.77, the understandability of the MDDI ranged from 3.50 to 4.53, and the understandability of the DMS ranged from 4.20 to 4.67. Based on Campos et al., means of understandability ≥3 were considered adequate understandability, representing degrees of understanding of approximately 80%. As such, the preliminary Chinese translations (i.e., the C-MOET, C-DMS and C-MDDI) were used in the subsequent studies.
## Measures
In addition to the three measures (i.e., the MOTE, DMS and MDDI) being validated in the current study, the following measures were used for examining the construct validity of the three measures.
## Psychosocial impairment
The Clinical Impairment Assessment Questionnaire (CIA 3.0)was used to assess psychosocial impairment secondary to eating disorder symptomatology. The CIA 3.0 contains 16 items which are rated on a 4-point Likert-type scale from 0 ("not at all") to 3 ("always"). The summation of the 16 items is used as the overall indicator of psychological impairment, with higher total scores indicating greater psychological impairment. Previous studies showed that the CIA 3.0 had good psychometric properties. In the current study, the Chinese version of the CIA 3.0 (C-CIA 3.0) was used. The CIA 3.0 was translated into Chinese with similar procedures to those described in the translation section, and the only difference was that the preliminary C-CIA 3.0 was administrated to adolescent boys, adolescent girls, undergraduate men and undergraduate women for evaluating the meanings, clarity and understandability of the items. Cronbach's α of the CIA 3.0 was 0.94 and 0.90 for the samples of university students and general adults, respectively.
## Psychological distress
The 6-item Kessler Psychological Distress Scale (K6)was used to assess psychological distress during the past month. The items are rated on a 5-point Likert scale from 1 ("none of the time") to 5 ("all of the time"). Higher total scores indicate greater psychological distress. The Chinese version of the K6 was used in the current study, and it had shown good psychometric properties in Chinese populations. In the sample of university students, Cronbach's α of the K6 was 0.88, and in the sample of general adults, Cronbach's α of the K6 was 0.82.
## Thinness-oriented eating disorder symptomatology
The 12-item self-reporting Eating Disorder Examination Questionnaire (EDE-QS)was used to examine traditional eating disorder symptomatology (i.e., thinness-oriented disordered eating). The response scale ranges from 0 to 3. The summed scores of the 12 items are the total score. Higher total scores indicate a higher level of traditional eating disorder symptomatology. Recent findings suggest that the Chinese version of the EDE-QS showed strong reliability and validity. In the current study, the Cronbach's α of the EDE-QS was 0.92 and 0.88 for the samples of university students and general adults, respectively.
## Thinness-oriented body dissatisfaction
The 9-item body dissatisfaction subscale of the Eating Disorder Inventory (EDI-BD)was used to measure thinness-oriented body dissatisfaction. The items are rated on a 6-point Likert-type scale from 1 ("never") to 6 ("always"). Higher total scores represent a higher level of body dissatisfaction. Adequate reliability and validity of the EDI-1 were reported for Chinese male university students. Cronbach's α of the EDI-BD was 0.90 and 0.92 and for the university sample and the general adult sample, respectively.
# Statistical analysis
All data analyses in the current work were carried out via the R 4.0.0 with the lavaan package version 0.6-7and psych package version 2.0.9. Specifically, confirmatory factor analysis (CFA) was used to confirm the one-factor structure of the MOET, two-factor structure of the DMSand three-factor structure of the MDDI. To estimate the CFA models, the diagonally weighted least squares (WLSMV) estimator was adopted, as it was recommended for Likert-type rating scales. To evaluate the model fit of CFA, according to Hu and Bentlerand Shi et al., the following fit indicators were reported: the comparative fit index (CFI; close to or >0.95 indicates a good fit, but 0.90 indicates an acceptable fit), Tucker-Lewis index (TLI; close to or >0.95 indicates a good fit, but 0.90 indicates an acceptable fit) and the standardized root mean square residual (SRMR; close to or <0.06 indicates a good fit, but 0.10 indicates a mediocre fit). It should be noted that we decided not to report the root mean square error of approximation (RMSEA) because it is unreliable and inferior to SRMR when used in CFA with ordinal responses (e.g., Likert-type responses). We also used the multi-group CFA method to assess measurement invariance by sample type (university students vs. general adults). According to Cheung and Rensvold, CFI < 0.01 and SRMR < 0.030 indicate invariance between two nested models (e.g., configural model vs. metric model and metric model vs. scalar model).
To examine the reliability, we calculated the McDonald's omega values for the MOET, DMS subscales and MDDI subscales. Furthermore, we evaluated the concurrent validity of the MOET, DMS and MDDI by exploring the correlations between the scores of the three scales being validated and the theoretically correlated measures. Based on previous research (e.g.,, we expected significant and positive associations among the MOET scores, DMS scores, MDDI scores, CIA scores, K6 scores, EDE-QS scores and EDI-BD scores. According to Cohen, correlations of 0.1, 0.3, and 0.5 were considered small, medium and large, respectively.
Finally, incremental validity was explored with hierarchical multiple regression analyses to determine whether the scores of the three measures being validated explained the unique variances in the key criterion variables after controlling for confounding variables. Specifically, since both muscularity-oriented disordered eating and muscle dysmorphic disorder symptoms, as measured by the MOET and MDDI, respectively, were already "outcome or criterion" variables, we cared about whether the MOET scores and the MDDI scores could uniquely explain the variances in psychosocial impairment and psychological distress when the confounding variables (e.g., age, BMI, traditional eating disorder symptomatology and traditional thinness-oriented body dissatisfaction) were controlled. Regarding the DMS scores, the criterion variables selected were muscularity-oriented disordered eating and muscle dysmorphic disorder symptoms, and the cofounding variables were age, BMI, traditional eating disorder symptomatology and traditional thinness-oriented body dissatisfaction.
# Results
## Confirmatory factor analysis and internal consistency
## Muscularity-oriented eating test
For the university sample, the results of CFA showed that the one-factor structure had an adequate model fit, with χ 2 = 281.68 (df = 90, p < 0.001), CFI = 0.98, TLI = 0.98 and SRMR = 0.05. For the general adult sample, the one-factor structure also had an adequate model fit, with χ 2 = 345.74 (df = 90, p < 0.001), CFI = 0.95, TLI = 0.94 and SRMR = 0.06. The standardized factor loadings are shown in Within the university sample, the two-factor structure had an acceptable model fit, with χ 2 = 502.08 (df = 89, p < 0.001), CFI = 0.96, TLI = 0.95 and SRMR = 0.11. As for the general adult sample, the two-factor structure had a marginally acceptable model fit, with χ 2 = 714.71 (df = 89, p < 0.001), CFI = 0.91, TLI = 0.89 and SRMR = 0.09. Since certain items of the DMS share similar wording, the modification indices (M.I.) were checked. Based on the M.I. values, the covariances among three items (i.e., Items 13, 14, and 15) were freely estimated one by one, as these items had similar wording (i.e., I think that my arms (for Item 13), chest (for Item 14) or legs (for Item 15) are not muscular enough). The modified model showed an adequate model fit, with χ 2 = 493.63 (df = 86, p < 0.001), CFI = 0.94, TLI = 0.93 and SRMR = 0.08. The standardized factor loadings are described in
## Muscle dysmorphic disorder inventory
In both the university sample and general adult sample, the three-factor structure of the MDDI had an acceptable model fit, with χ 2 = 317.95 (df = 62, p < 0.001), CFI = 0.95, TLI = 0.94 and SRMR = 0.09 for the university sample and χ 2 = 271.13 (df = 62, p < 0.001), CFI = 0.94, TLI = 0.92 and SRMR = 0.07 for the general adult sample. The standardized factor loadings of the model were described in
## Invariance and mean difference tests
## Muscularity-oriented eating test
To use the WLSMV estimator for the invariance tests, the missing patterns of the response categories should be the same across different samples. However, there was no response category missing in the general adult sample, but there was a missing response category of 4 ("always true") in Item 8 ("I have pre-cooked several meals in advance to ensure that I do not deviate from my diet plan.") for the university sample. Thus, the invariance and mean difference tests were not conducted for the MOET.
## Drive for muscularity scale
As shown in, all indices suggested that configural, metric and scalar invariance were supported across the samples for the DMS. Latent mean difference tests showed that the general adults had significantly higher latent means for MBI (Cohen's d = 1.02; p < 0.001) and MB (Cohen's d = 1.32; p < 0.001) than those of the university students.
## Muscle dysmorphic disorder inventory
The results inshow that configural, metric and scalar invariance were also supported across the samples for the MDDI. Latent mean difference tests showed that the general adults had significantly higher latent means of DS (Cohen's d = 1.47; p < 0.001), AI (Cohen's d = 0.69; p < 0.001) and FI (Cohen's d = 1.37; p < 0.001) than those of the university students.
## Concurrent validity
## Muscularity-oriented eating test
As shown in, for the university sample, there were medium to large positive correlations with the theoretically closely related constructs, namely the drive for muscularity, muscle dysmorphic disorder symptoms, psychological impairment and traditional eating disorder symptomatology. For the general adult sample, the MOET scores had large positive correlations with all the theoretically closely related constructs. For both samples, the MOET scores also had a medium correlation with psychological distress and a small correlation with thinness-oriented body dissatisfaction, but the MOET scores were not significantly related to BMI or age.
## Drive for muscularity scale
As shown in, for both samples, the DMS scores had significant and large or close-to-large correlations with muscle-oriented disordered eating, muscle dysmorphic disorder symptoms, psychological impairment and traditional eating disorder symptomatology. Moreover, for both samples, the DMS scores had a small but significant positive correlation with thinness-oriented body dissatisfaction, and the DMS scores were not significantly related to BMI.
## Muscle dysmorphic disorder inventory
As seen in, for both samples, the MDDI scores showed significant and large or close-to-large correlations with muscle-oriented disordered eating, the drive for muscularity, psychological impairment, traditional eating disorder symptomatology and psychological distress. In addition, for both samples, the MDDI scores were not significantly related to BMI or age.presents the hierarchical regression models with the MOET, DMS and MDDI scores entered in the second step. Specifically, for both samples, the inclusion of the MOET and MDDI scores significantly increased the variance explained in psychological impairment, and the inclusion of the DMS scores significantly increased the variance explained in muscularity-oriented disordered eating and muscle dysmorphic disorder symptoms.
## Incremental validity
# Discussion
The need to further study male EDs and body image disturbances among Chinese men has been made clear. However, due to the lack of validated instruments assessing eating and body image disturbances among Chinese men, limited research about these topics is possible at this time in China. Thus, the current study aimed to translate three measures (i.e., the MOET, DMS and MDDI) that specifically assess eating and body image disturbances among men and examine the psychometric properties of the Chinese translations of the three measures. The results showed that all measures presented good psychometric properties with both a sample of male university students and a sample of general male adults.
Specifically, in line with our hypotheses, the respective one-factor structure, twofactor structureand three-factor structureof the MOET, DMS and MDDI were successfully confirmed via CFA in the current study. This indicates that the structures of the scales are stable across different cultures, especially considering that the factor structures of the DMS and MDDI have been successfully replicated in a number of samples from different countries. Thus, future researchers may consider conducting cross-cultural measurement invariance tests (e.g.,for the three measures to explore whether these measures can be used for cross-cultural comparisons in disordered eating and body image disturbances among men, which can greatly forward our understanding of the cultural differences in these areas.
The McDonald's omega values for the MOET, DMS subscales and MDDI subscales were acceptable in the current two samples, suggesting that the scores of the three measures had adequate internal consistency reliability, which is in line with previous validation studies (e.g.,. In the present work, as hypothesized, the scores of the three measures were found to be significantly and positively related to each other. These large inter-correlational findings provide evidence for the convergent validity of the three measures, since all three constructs being measured (i.e., muscularity-oriented disordered eating, drive for muscularity and muscle dysmorphic disorder symptoms) are focused on muscularity.
We also found that the total scores of the three measures were closely related to traditional eating disorder symptomology. These findings are also consistent with the previous literature. As muscularity concerns in men include not only muscularity but also leanness, it is not surprising to find large correlations between the scores of the three measures and traditional eating disorder symptomology, which is thinnessoriented. Furthermore, the close relationships between the scores of the three measures and psychosocial impairment secondary to eating disorder symptomology suggest the potential negative effects from muscularity-oriented disordered eating and body image disturbances on men's quality of life. Together, with the medium correlations between the total scores of the three measures and psychological distress, these correlational findings warrant interventions for EDs and body image disturbances among Chinese men.
The results also showed that thinness-oriented body dissatisfaction, as measured by the EDI-BD, had small to medium associations with the scores of the three measures. These findings are not surprising, since the EDI-BD focuses on body weight and shape dissatisfaction due to body fat, while the three measures focus on muscularity. Moreover, the scores of the three measures contributed a significant amount of unique variance to each criterion variable, indicating the good incremental validity of the three measures.
Finally, we found that the subscale scores of the DMS and MDDI were significantly higher in general adult men than university students, indicating that Chinese general adult men may have more muscularity-oriented body image disturbances. Except for the sampling differences in recruiting the two samples (paper and pencil in university students vs. online in general adults), the large differences in professional status may also help explain the score differences in the DMS and MDDI. Specifically, different from Chinese university students who generally live in university dormitories and rely on the monthly financial support from their parents to live (called "monthly living fees"), the majority of the general adults in the current study were employed and had stable income. Thus, general adults should have more time and money to get involved in activities to strengthen their muscles.
The current work is not free from limitations. First, because the sample sizes were only enough for us to run CFAs, we did not conduct exploratory factor analysis by splitting the samples as recommended in validating body image related measures. Thus, the current study only confirmed the original factor structures shown in the previous literature, and future studies with larger sample sizes are needed to explore whether there are different factor structures in the three measures in the Chinese context. However, it should also be noted that as the original factor structures of the DMS and MDDI had been replicated in a wide variety of populations, and a recent validation of MOET also replicated the original single-factor structure, we chose to run CFAs only for confirming the original factor structures. Second, our study was limited to Chinese adult men. Thus, the findings cannot be generalized to Chinese women, adolescent boys or clinical populations (i.e., patients with eating disorders or muscle dysmorphia). Third, the invariance tests by sample type were not conducted, as certain important factors, such as sexual orientation, were not included because we did not collect such information. However, future studies should be conducted to check whether these measures can be used to make group comparisons. Fourth, the test-retest reliability of the measures was not assessed. Thus, the test-retest reliability of the three measures remains unknown. Finally, as a common issue in survey research, socially desirable respondingmight have affected the scores obtained in the scales, so future studies may test whether socially desirable responding is an important issue by including an instrument to assess socially desirable response tendencies (e.g.,. Future researchers are highly encouraged to conduct research by considering the limitations of the current work.
# Conclusions
In summary, the three measures (i.e., the MOET, DMS and MDDI) showed good psychometric properties in two samples of Chinese adult men. To date, there has been limited research related to muscularity-oriented disordered eating and body image disturbances in China, and the translation and validation of the three measures can serve as the first step to further understanding these areas of research within the Chinese context. |
A novel k-mer mixture logistic regression for methylation susceptibility modeling of CpG dinucleotides in human gene promoters
Background: DNA methylation is essential for normal development and differentiation and plays a crucial role in the development of nearly all types of cancer. Aberrant DNA methylation patterns, including genome-wide hypomethylation and region-specific hypermethylation, are frequently observed and contribute to the malignant phenotype. A number of studies have recently identified distinct features of genomic sequences that can be used for modeling specific DNA sequences that may be susceptible to aberrant CpG methylation in both cancer and normal cells. Although it is now possible, using next generation sequencing technologies, to assess human methylomes at base resolution, no reports currently exist on modeling cell type-specific DNA methylation susceptibility. Thus, we conducted a comprehensive modeling study of cell type-specific DNA methylation susceptibility at three different resolutions: CpG dinucleotides, CpG segments, and individual gene promoter regions. Results: Using a k-mer mixture logistic regression model, we effectively modeled DNA methylation susceptibility across five different cell types. Further, at the segment level, we achieved up to 0.75 in AUC prediction accuracy in a 10-fold cross validation study using a mixture of k-mers.Conclusions:The significance of these results is three fold: 1) this is the first report to indicate that CpG methylation susceptible "segments" exist; 2) our model demonstrates the significance of certain k-mers for the mixture model, potentially highlighting DNA sequence features (k-mers) of differentially methylated, promoter CpG island sequences across different tissue types; 3) as only 3 or 4 bp patterns had previously been used for modeling DNA methylation susceptibility, ours is the first demonstration that 6-mer modeling can be performed without loss of accuracy.
# Background
DNA methylation is the chemical modification of DNA bases, mostly on cytosines that precede a guanosine in the DNA sequence, i.e., the CpG dinucleotides. This epigenetic modification involves the addition of a methyl group to the number 5 carbon of the cytosine pyrimidine ring. DNA methylation is essential for cellular growth, development and differentiation [bib_ref] DNA methylation patterns and epigenetic memory, Bird [/bib_ref] , playing a fundamental role in the activation of genes at the transcriptional level. In cancer cells, aberrant DNA methylation patterns, such as genome-wide hypomethylation and region-specific hypermethylation, are frequently observed [bib_ref] Cancer-epigenetics comes of age, Jones [/bib_ref]. CpG islands, short CpG-rich regions of DNA often located around gene promoters and normally protected from DNA methylation, become hypermethylated in cancer, contributing to transcriptional silencing [bib_ref] The cancer epigenome-components and functional correlates, Ting [/bib_ref] [bib_ref] Gene silencing in cancer in association with promoter hypermethylation, Herman [/bib_ref]. As CpG island methylation patterns have been shown to differ across cancer types, recent studies have revealed that some CpG islands are "methylation sensitive", while others are "resistant" to DNA methylation [bib_ref] Aberrant CpGisland methylation has non-random and tumour-type-specific patterns, Costello [/bib_ref]. Recent technological breakthroughs allow, for the first time, the capability to measure human methylomes at base resolution [bib_ref] Principles and challenges of genome-wide DNA methylation analysis, Laird [/bib_ref] , providing unprecedented opportunities for understanding the phenomenon of methylation susceptibility.
## Previous work
Several recent studies have attempted to predict CpG island methylation patterns in normal and cancer cells. DNA pattern recognition and supervised learning techniques were used by Feltus et al to discriminate methylation-prone (MP) and methylation-resistant (MR) CpG islands based on seven DNA sequence patterns [bib_ref] Predicting aberrant CpG island methylation, Feltus [/bib_ref]. McCabe et al then developed a classifier (PatMAn) based on the frequencies of those seven patterns in cancer [bib_ref] PatMaN: rapid alignment of short sequences to large databases, Prüfer [/bib_ref] , followed by "SUPER-PatMAn" for predicting methylation susceptible CpG islands using both local sequence context and transacting factors such SUZ12 [bib_ref] A multifactorial signature of DNA sequence and polycomb binding predicts aberrant CpG..., Mccabe [/bib_ref]. In addition, Feltus et al used motifs related to 28 MP and MR CpG islands to predict DNA methylation susceptibility [bib_ref] DNA motifs associated with aberrant CpG island methylation, Feltus [/bib_ref] , and Keshet et al showed evidence of instructive mechanisms in cancer cells, finding common sequence motifs in the regions of promoters whose genes show tumor-specific "methylation susceptibility" [bib_ref] Evidence for an instructive mechanism of de novo methylation in cancer cells, Keshet [/bib_ref]. A prediction method for finding a minority class in an imbalanced data setting (which is the case for DNA methylation data), called "cluster_boost", was recently developed by Goh et al and used to identify novel hypermethylated genes in cancer [bib_ref] Genomic sweeping for hypermethylated genes, Goh [/bib_ref]. Fang et al developed "MethCGI" to predict the methylation status of CpG islands using a support vector machine and both local sequence context and transcription factor binding sites [bib_ref] Predicting methylation status of CpG islands in the human brain, Fang [/bib_ref]. Finally, a prediction method using DNA sequence features of various types, including sequence, repeats, predicted structure, CpG islands, and genes, was developed by Bock et al to predict binding sites, conservation, and single nucleotide polymorphisms [bib_ref] CpG island methylation in human lymphocytes is highly correlated with DNA sequence,..., Bock [/bib_ref].
While the focus of the above studies was on CpG island methylation susceptibility, recent experiments have convincingly demonstrated that methylation levels of CpG sites, i.e. genomic location of CpG dinucleotides, within a CpG island can be highly variable. For example, Handa et al found that certain sequence features flanking CpG sites were associated with high-and low-methylation CpG sites in an in vitro DNMT1 overexpression model [bib_ref] Profound flanking sequence preference of Dnmt3a and Dnmt3b mammalian DNA methyltransferases shape..., Handa [/bib_ref]. Moreover, at single base pair resolution, Zhang et al demonstrated that DNA methylation levels frequently differ within a CpG island [bib_ref] DNA methylation analysis of chromosome 21 gene promoters at single base pair..., Zhang [/bib_ref]. To investigate the role of DNA methylation during development in human embryonic stem cells Brunner et al developed Methylseq, which assays DNA methylation at more than 90,000 regions throughout the genome [bib_ref] Distinct DNA methylation patterns characterize differentiated human embryonic stem cells and developing..., Brunner [/bib_ref]. Using bisulfite sequencing data, Lister et al determined the first genome-wide, single-base-resolution maps of methylated cytosines in mammalian genomes (human embryonic stem cells (ESC) and fetal broblasts) [bib_ref] Human DNA methylomes at base resolution show widespread epigenomic differences, Lister [/bib_ref]. By using "ultradeep" sequencing data from Taylor et al [bib_ref] Ultradeep bisulfite sequencing analysis of DNA methylation patterns in multiple gene promoters..., Taylor [/bib_ref] , we demonstrated that CpG flanking sequences can be used to model methylation susceptible CpG sites [bib_ref] Predicting DNA methylation susceptibility using CpG flanking sequences, Kim [/bib_ref]. Finally, Previti et al analyzed tissue-specific CpG island methylation status, in terms of profiles created by probabilistically combining two sources of independent clusters (clusters from methylation data in 12 tissues and clusters from CGIs attributes) to demonstrate the predictive power of their method with a decision tree classifier [bib_ref] Profile analysis and prediction of tissuespecific CpG island methylation classes, Previti [/bib_ref]. Those investigators categorized profiles into four classes: constitutive unmethylated, constitutive methylated, unmethylated in sperm, and differentially methylated [bib_ref] Profile analysis and prediction of tissuespecific CpG island methylation classes, Previti [/bib_ref].
## Motivation
Previous CpG island methylation susceptibility prediction studies have not considered cell type-specific methylation status. Considering variations in DNA methylation level even in the same genomic regions of different types of cells, we asked the question: can cell type-specific DNA methylation susceptibility be modeled? The significance of exploring this question is based on evidence supporting the strong association of genomic sequence features with DNA methylation status. Furthermore, recent studies strongly indicate the existence of methylation sensitive/ resistant CpG islands in different cancer types [bib_ref] Aberrant CpGisland methylation has non-random and tumour-type-specific patterns, Costello [/bib_ref]. In this paper, we performed a comprehensive DNA methylation susceptibility modeling study in five different cell lines at three different levels: CpG sites, entire promoter regions, and short DNA segments. We focused on DNA methylation in the context of CpG dinucleotides in adult cells (we are aware of a recent study [bib_ref] Human DNA methylomes at base resolution show widespread epigenomic differences, Lister [/bib_ref] reporting non-CpG methylation in ESC).
# Methods
The problem: methylation susceptible dna segment modeling problem The need for segment modeling Bisulfite sequencing data clearly demonstrates that methylation levels, even within a single gene promoter, can be highly variable. Furthermore, a figure in Additional file 1 shows highly variable methylation of the same promoter sequence in five different cell lines, i.e. cell type-specific DNA methylation susceptibility (bisulfite sequencing data obtained from [bib_ref] DNA methylation analysis of chromosome 21 gene promoters at single base pair..., Zhang [/bib_ref].
## Definition of the problem
The following notations were used to formally define the problem. A small set of pre-selected k-mers x = {x i }, where a k-mer is fixed number of DNA base pairs. Labels t = {t j } on data are assigned as +/-depending on methylation level p j of each sample.
For each cell type, a k-mer mixture logistic regression model (equation 1) was built using a small set of preselected patterns, i.e. k-mers. To select the best logistic model, predicted methylation at a CpG site (based on the logistic model under consideration) was compared with actual CpG methylation obtained from the bisulfite sequencing data. To make the comparison, we calibrated the predicted methylation level between 0 and 1 (below).
[formula] y = 1 1 + e −f (x) [/formula]
(1)
where f (x) = i β i x i and b i 's are parameters to be learned for the machine learning predictor. The k-mer mixture modeling problem Our goal was to test whether methylation susceptibility can be modeled by a logistic regression model using a small set of kmers. Although using k-mers for DNA methylation modeling is not entirely new, to our knowledge, only short k-mers (3 or 4 bp in length) were used in previous studies [bib_ref] CpG island methylation in human lymphocytes is highly correlated with DNA sequence,..., Bock [/bib_ref]. As short k-mers can occur in almost every DNA sequence, modeling using 3 or 4 bp relies on k-mer frequency.
1. First, we attempted to use longer k-mers (up to 6 bp) to utilize those that only occur in methylation susceptible sequences (vs. frequency for short k-mers, described above). 2. Our goal of determining whether machine learning predictors can be built by using k-mers required that we address two important issues: over-fitting and generalizability of prediction beyond the test data. The over-fitting problem was addressed by selecting a small number of k-mers from the training data set (using a larger number of k-mers can easily over-fit the training data). The cross validation technique was used to test the generalizability of prediction power. We selected k-mers and built machine predictors by using only the training data set. We then assessed the predictor on the test data set not used for either selecting k-mer features or building predictors.
## Two k-mer feature selection methods
We used a selected set of k-mers for DNA methylation susceptibility modeling in the different cell types. The research question explored in this paper is the feasibility of modeling methylation susceptible segments given a set of k-mers. As selection of the "best set" of k-mers for modeling was not explored (a solution to the combined problem was too difficult), we used two standard pattern selection methods for a two-class data set.
1. Feature selection with t-test: A popular t-test method was used to select k-mers because of its simplicity and applicability for all modeling approaches.
For each attribute a, occurrences of a were counted in positive samples and negative samples. Then, the P-value of a was measured by t-test. A fixed number of patterns was selected from a list of k-mers ordered by P-value. Alternatively, patterns with a P-value below a threshold were selected. 2. Feature selection with the random forest technique: The RF algorithm [bib_ref] NIPS: Feature selection challenge, Breiman [/bib_ref] can be used for feature selection. The usefulness of the RF-based feature selection method was clearly demonstrated by Yi-Wei Chen and Chih-Jen Lin at the NIPS 2003 feature selection challenge . We used an extended version of the RF-based feature selection method. Multiple rounds of the RF-based feature selection were performed using a balanced data set of methylation-susceptible and non-susceptible sequences. We performed k times of RF runs, where each RF run used n random trees; only top N attributes with z-scores > 0 were collected. After k RF runs, a subset of attributes, which had appeared p% times, were selected. The values were set k = 30, n = 100, N = 100, and p = 90 for the k-mer feature selection.
In both methods, we extracted a set of patterns in the balanced data set. First, centered at each CpG site, we extracted a flanking sequence of length l, where we set l = 100. A label of the CpG site was given as +/-depending on methylation level. Then, we balanced the data with even number of +/-classes. A set of all k-mers obtained in sliding windows on each sequence were used for k-mer feature selection.
## Modeling methylation levels of dna segments
Definition A boundary variable B i at a genomic sequence position is an indicator variable that is defined where two adjacent CpG sites have different labels. The value 1 of B i denotes that the genomic position is a boundary and the value 0 denotes that the position is not a boundary. A DNA segment S is defined by two boundary variables B a and B z where B a = 1 and B z = 1 and for all a <i <z, B i = 0. [fig_ref] Figure 1: Illustration of the initial segment definition [/fig_ref] illustrates how boundary variables are used to define 10 segments. We call a set of DNA segments defined by the boundary variables a configuration. Labeling data Given a segment S i , the methylation probability p i of a segment was defined as a ratio of the number of CpG sites with the + label to the number of CpG sites in the segment. Then, the label t i of S i was assigned + if p i is greater than 0.5. Otherwise, a labelwas assigned to t i . Attributes for modeling K-mer occurrences in segments in the training data set were used as attributes. A small subset of k-mers features x was selected from all k-mers using the feature selection methods.
Modeling A single logistic regression model was used to model all DNA segments for each cell line, using attributes x and labels t.
## Segment-level modeling challenges: exponential search space
Although the methylation status of a DNA segment is defined by an aggregation of the methylation status of individual promoter regions (as we did for the whole promoter region-modeling approach), how to define methylation susceptible DNA segments is currently unknown. For example, consider a DNA segment with five CpG sites {s1, s2, s3, s4, s5} in a short DNA segment and assume that three sites, s1, s2, s4 are methylation susceptible and the other two sites s3, s5 are resistance methylation. By definition, the DNA segment is methylation susceptible, as the majority of sites (three) are methylation susceptible. However, if we divide the segment into two sub-segments, {s1, s2} and {s3, s4, s5}, there will be a segment that is susceptible to methylation and one that is resistant. To determine which of the two segment definitions can be better modeled for methylation susceptibility, enumeration of all possible definitions of segment configurations and for each definition of segment is required. We thus computed a "best fit" logistic model for methylation data in a cell line. The complexity of this problem can be discussed in terms of the well-known "counting the number of parenthesization" problem, because a parenthesis can define a segment of CpG sites. The number of parenthesis P (n) for n CpG sites is P (1) = 1; P(n) = n−1 i=1 P(i)P(n − i)for n ≥ 2. Given the complexity, an optimal solution using an exhaustive search algorithm is unlikely to be found (known to be Ω (2 n ). Thus, we developed a heuristic algorithm that used a random segment merging starting from the finest definition of segments.
## A random binary segment merging algorithm
A Naïve approach to segment modeling simply enumerates all possible segment configurations. Every combination of segment boundaries is considered, while changing the setting of values for boundary indicator variable B i {0, 1}. Then, an error function for each segment set definition is computed. However, this requires the enumeration of a 2 m possible segment configurations, where m is the number of B i . To compute the optimal k-mer logistic regression model, segment boundaries must first be identified; however, as these are unknown, we started with an initial presumption of the methylation susceptible and resistant segments. We then used an iterative improvement procedure in search of both the segment definition and the best fitting logistic regression model. The major steps of the segment modeling algorithm are as follows:
1. Initialization of a configuration: Define a boundary variable B i = 1 at every genomic position where labels (+ or -) of two adjacent CpG sites around the position are different. Define a segment as a DNA region between two boundary variables set to 1. By taking this approach, we start with a configuration of smallest possible segments. By merging segments in many different ways and re-calculating the logistic regression model, the algorithm attempts to find the best segment configuration. This is how INITIALCONFIGURATION() is implemented in the HillClimbingConfigurationSearch in Algorithm 1.
## 2.
Computing a logistic regression model: Given a k-mer occurrence and a segment configuration, compute a logistic regression model by [bib_ref] DNA methylation patterns and epigenetic memory, Bird [/bib_ref]. This is how COMPUTEMODEL() is implemented in the HillClimbingConfigurationSearch in Algorithm 1.
## 3.
Computing an error of a segment configuration: Errors in the segment set S are measured by [bib_ref] Cancer-epigenetics comes of age, Jones [/bib_ref].
[formula] O(S) = |S| i=1 w i (ŷ i − t i ) 2(2) [/formula]
where |S| is the total number of segments,ŷ i is the predicted methylation level of the segment i, t i is the actual methylation level of the segment i, and w i is the weight of each segment. A segment weight is defined as w i = |S|/ |S i |, where |S| is the average count of CpG sites in all segments and |S i | is the count of CpG in a segment. A weight of each segment w i is given as an inverse proportion to average segment size. In this way, large segments are penalized less, and vice versa. This is how COMPUTEERROR() is implemented in the Hill-ClimbingConfigurationSearch in Algorithm 1.
## The random binary segment merging algorithm
Given the current segment configuration {B i }, a segment is randomly chosen using a distribution of errors measured by a weighted square error. For a segment B j , the weighted square error is defined by e j = β j (ŷ j − t j ) [bib_ref] Cancer-epigenetics comes of age, Jones [/bib_ref] where the weight of the segment β j = |S i | /|S|,ŷj is the predicted methylation level of the segment j, and t j is the actual methylation level of the segment j. A segment is chosen by random sampling using a segment error vector <e 1 , . . . , e n > where n is the number of segments in the current segment configuration. The random sampling using a segment error vector <e 1 , . . . , e n > guides choosing a segment with a higher prediction error, but also ensure a random sampling. Note that segments that are already considered for merging are excluded for the next round of sampling (see the use of visit[] in the HillClimbingConfigurationSearch in Algorithm 1).
Once a segment B j is chosen, it is tentatively merged with segment B j+1 next to B j . Then a logistic regression model is re-calculated. The two segment merging is accepted only if the merging of two segments reduces the weighted squared error (equation 2). Otherwise, the original segment configuration is retained, rejecting the merging. A segment B j considered for merging is marked so that the segment will not be repeatedly chosen for the next step. This sampling and marking a segment is repeated until all segments in the current configuration are considered for merging.
Input
# Results
## Data set
We used data from Zhang et al [bib_ref] DNA methylation analysis of chromosome 21 gene promoters at single base pair..., Zhang [/bib_ref] for DNA methylation patterns in chromosome 21 (297 amplicons from 190 gene promoters using bisulfite conversion, subcloning and sequencing DNA as the major experimental methods). The bisulfite sequencing data were collected in five cell types: viz. human peripheral blood (primarily leukocytes), fibroblast, the human embryonic kidney cell line HEK293, the human hepatocellular liver carcinoma cell line HepG2 and fibroblast cells derived from a patient with Down syndrome (trisom 21). Methylation patterns differed widely and specific to each cell types.
## Experimental setup
The 10-fold cross validation (described above) was used to compare the performances of three modeling approaches. For each round of 10-fold validation, one of the 10 subsets was set aside for testing, and the k-mer features were selected only from the training set, ensuring that the test data would have no influence on the k-mer feature selection. Also, regression coefficients were computed in only training stage. We measured the area under the ROC curve (AUC) score for performance comparison.
## Effectiveness of the segment modeling approach
We extensively tested the effectiveness of the segment modeling algorithm using 4-mer, 5-mer, and 6-mer patterns. For each of the experiments, the AUC score was measured from 10-fold cross validation for the initial segment definition vs. the final segment definition. The RFbased algorithm with 100 trees was used for k-mer feature selection. For each k-mer selection procedure, 30 random experiments were performed, and k-mers with zscore > 0 that appeared in at least 90% of experiments were selected as k-mer features. Using the set of k-mers, the optimal logistic regression model was computed.
## 10-fold cross validation experiments
The performance comparison between the initial segments and the final segments in the test set is shown in [fig_ref] Figure 2: Effectiveness of segment modeling in 10-fold cross-validation experiments [/fig_ref]. Bars between adjacent dotted lines show the improvement in the between prediction results of two models with the initial segment setting and the final segment setting in terms of the AUC scores. We measured the performance improvement using 4-mer, 5-mer, and 6-mer features. For each cell type, the segment modeling algorithm identified significantly improved segment definitions. Five panels in each plot correspond to tissue types: (A) Fibroblast, (B) HEK293, (C) HepG2, (D) Leukocytes, and (E) Trisom 21. Our algorithm achieved approximately 10% improvement in most cell types, illustrating the effectiveness of the segment modeling algorithm.
## Search behavior
The search behavior of the segment modeling algorithm is shown in . In this experiment, we used the whole data set to show the algorithmic convergence of our approach. The learning error (Equation 2) was reduced at each iteration of segment merging and model re-calculation. Our random segment sampling algorithm converged for all 15 cases of 5 different cell lines with 4-, 5-, and 6-mers.
## Discussion on the predictive power of the model
The predictive power of the model measured by 10-fold cross validation is encouraging. For 6-mers, the predictive accuracy was 0.69 for Fibroblast, 0.70 for HEK293, 0.54 for HepG2, 0.73 for Leukocytes, and 0.65 for Trisom 21. These prediction accuracies using 6-mer cannot be achieved in random data sets where the expected prediction accuracy is 0.5. Variations in the prediction accuracy for the five cell types, especially for HepG2, may be due to the cell type specific characteristics. On the other hand, the data obtained from [bib_ref] DNA methylation analysis of chromosome 21 gene promoters at single base pair..., Zhang [/bib_ref] was of a low coverage. Amplicons covered less than 0.2% of entire Chromosome 21.
Thus variations in the prediction accuracy may due to the low coverage of the data used. We were not able to further verify why the prediction accuracy varied. In fitting the whole data set, as opposed to 10 fold cross validation, the final model was able to accurately predict methylation susceptibility. The search behavior of the segment modeling algorithm using the whole data set. Pairwise plots showing reduced learning error (2) at each iteration of segment merging and model recalculation. The columns for the pairwise plots are k-mers; rows are cell lines. In each plot, the X-axis denotes the number of iterations and the weighted squared prediction error is denoted on the Y-axis. The HillClimbing search algorithm effectively reduced the error between prediction and observation. In fitting the whole data set, as opposed to 10 fold cross validation, the final model predicted methylation susceptibility in the different cell types.
## Effect of the number of k-mers used for prediction
The three modeling approaches were compared in terms of AUC obtained by 10-fold cross-validation technique. We conducted comprehensive modeling of cell-type specific DNA methylation susceptibility at three different resolutions: individual CpG sites, CpG segments, and promoter regions in terms of AUC obtained by the 10fold cross validation technique. The methods for modeling at individual CpG sites and at promoter regions are described in Additional file 2. To measure the effect of the number of k-mer patterns used for modeling, 10-fold cross-validations were performed with a varying number of k-mer patterns from 10 to 100 (with an increase of 10 k-mers). P-values from t-tests were used to select the kmers. The experimental results are illustrated in . Only the segment modeling approach was effective for all 4-, 5-, and 6-mer experiments. Interestingly, the number of k-mers used for modeling had little impact on the prediction result, demonstrating that the prediction accuracy did not derive from the over-fitting the data and indicating that the selection of a small number of k-mers can effectively model methylation susceptibility without a loss of prediction power. Moreover, when a longer k-mer was used (up to 6-mer), the prediction accuracy did not decrease. This finding is highly encouraging because on average, a 6-mer is unlikely to occur by chance in a short (274 bp) DNA segment. Thus, a set of 6-mers can be used to model DNA methylation susceptibility.
# Conclusion
We conducted a comprehensive modeling study for cell-type specific DNA methylation susceptibility. By performing extensive computational experiments of data from five distinct cell types, we show that DNA methylation susceptibility can be accurately modeled at the segment level, achieving up to 0.75 in AUC prediction accuracy in a 10-fold cross validation study. The two-step iterative segment modeling algorithm successfully identified optimal segments that can be modeled as a logistic regression model using a set of k-mers. Our model further shows the significance of certain kmers for the mixture model, which can potentially highlight DNA sequence features (k-mers) of differentially methylated promoter CpG island sequences in different cells and tissues, including malignancies. As only used 4 bp patterns were used in previous modeling studies of DNA methylation susceptibility, this is the first report to show that k-mer modeling can be performed using up to 6-mer without the loss of modeling accuracy. Effect of the number of k-mers used for three modeling approaches. The performance of three modeling approaches was measured from 10-fold cross-validation. Each bar is the AUC value of the experiment. X-axis is the number of most significant variables (p-value in t-test) used in each experiment. Consistently in 4-mer to 6-mer and regardless of number of patterns, segment modeling outperformed other modeling approaches. More importantly, from the experiments using variable numbers of k-mers from 10 to 100, we have shown that the selection of k-mers does not have a big impact on the model performances and the higher accuracies of the segment modeling approach, compared to the promoter and site-specific modeling approaches, is likely due to the effectiveness of the segment model.
[fig] Figure 1: Illustration of the initial segment definition. Because all boundary variables are set to 1, 10 initial segments are defined. Later, the segment modeling algorithm considers alternative segment definition by changing the boundary variable values. Figure was modified from[16,25]. [/fig]
[fig] Figure 2: Effectiveness of segment modeling in 10-fold cross-validation experiments. Bars between adjacent dotted lines show improvement in the between prediction results of two models with the initial segment setting and the final segment setting in terms of AUC scores. We measured the performance improvement using 4-mer, 5-mer, and 6-mer features. For each cell type, the segment modeling algorithm identified significantly improved segment definitions. Five panels in each plot corresponds to tissue types: (A) Fibroblast, (B) HEK293, (C) HepG2, (D) Leukocytes, and (E) Trisom 21. [/fig]
|
IgY antibodies in human nutrition for disease prevention
Oral administration of preformed specific antibodies is an attractive approach against infections of the digestive system in humans and animals in times of increasing antibiotic resistances. Previous studies showed a positive effect of egg yolk IgY antibodies on bacterial intoxications in animals and humans. Immunization of chickens with specific antigens offers the possibility to create various forms of antibodies. Research shows that orally applied IgY's isolated from egg yolks can passively cure or prevent diseases of the digestive system. The use of these alternative therapeutic drugs provides further advantages: (1) The production of IgY's is a non-invasive alternative to current methods; (2) The keeping of chickens is inexpensive; (3) The animals are easy to handle; (4) It avoids repetitive bleeding of laboratory animals; (5) It is also very cost effective regarding the high IgY concentration within the egg yolk. Novel targets of these antigen specific antibodies are Helicobacter pylori and also molecules involved in signaling pathways in gastric cancer. Furthermore, also dental caries causing bacteria like Streptococcus mutans or opportunistic Pseudomonas aeruginosa in cystic fibrosis patients are possible targets. Therefore, IgY's included in food for human consumption may be able to prevent or cure human diseases.
## The avian immune system
To protect the host against invading microorganisms and exogenous antigens, the avian species have also developed an immune system similar to mammals. The avian immune system consists of primary and secondary lymphoid organs. Thymus, located in the neck and Bursa Fabricius, located adjacent to the cloaca, are primary lymphoid organs. Secondary organs are spleen, caecal tonsils, Harderian gland, bone marrow, lymph nodes, and various lymphoid tissues e.g. lymphoid tissues associated with mucosal surfaces (MALT); including bronchialassociated lymphoid tissues (BALT), gut associated lymphoid tissues (GALT) and conjunctival associated lymphoid tissues (CALT). The thymus is considered to be the primary lymphoid organ for T-cell differentiation and the antibody-synthesizing B-cells are formed in the Bursa of Fabricius, an organ which is unique in birds [bib_ref] The bursa of fabricius and antibody production, Glick [/bib_ref] [bib_ref] The structure and function of the avian immune system, Sharma [/bib_ref]. The plasma cell proliferation and memory B-cells are situated within the spleen [bib_ref] Chicken antibodies, Carlander [/bib_ref]. The avian immune system basically consists of two types: first the innate, non-specific system and second the acquired, specific system. The latter is mainly characterized by specificity and memory. This system can be distinguished into the cellular and non-cellular immune response, once again.
The cellular response can be defined as those cells that react with a high specificity to their specific antigen but not including the cells involved in antibody production. T-lymphocytes, as part of the cellular branch, recognize processed antigens on antigen-presenting macrophages. Among other tasks, T-cells are able to enhance or suppress the activity of B-cells, macrophages and T-helper cells. Further T-cells can directly destroy foreign organisms.
The non-cellular, also referred to as the humoral response, includes plasma proteins, e.g. immunoglobulins that passively circulate in blood or lymph, and antibody producing B-lymphocytes. These cells are formed in the embryonic liver, yolk sack and bone marrow. After maturation in the Bursa Fabricius, B-cells move to the blood, spleen, caecal tonsils, bone-marrow, Harderian gland, and thymus. T-helper cells are able to activate B-cells into plasmocytes. These plasmocytes are then able to secrete antibodies (immunoglobulins) which are highly specific for binding antigens. After re-exposing a chicken to the same antigen a memory effect occurs, involving an increased antibody production at a faster rate [bib_ref] Hen egg yolk antibodies (IgY), production and use for passive immunization against..., Chalghoumi [/bib_ref].
## Immunoglobulin classes in birds in comparison to mammals
Three classes of immunoglobulins in birds are known. First described is the mammalian IgG which is analogous to IgY and is mainly present in serum and egg yolk [bib_ref] Phylogen of immunoglobulin structure and function. 3. Immunoglobulins of the chicken, Leslie [/bib_ref]. In previous studies, IgY has been called IgG because of its similar function and concentration in the serum. Nowadays, this terminology has been found to be incorrect, due to clear differences in molecular structure. Furthermore, recent studies describe the evolutionary relationship between avian IgY's and human IgE [bib_ref] The crystal structure of an avian IgY-Fc fragment reveals conservation with both..., Taylor [/bib_ref]. In addition, IgY's are also found in reptiles, amphibian and lungfish. The comparion of different Immunoglobulin classes are shown in [fig_ref] Table 1: Basic comparison of the adaptive immune system between avian and mammals [/fig_ref].
Research has shown that other immunoglobulins including chicken IgA and IgM have similar molecular weight, structure and electrophoretic mobility compared to mammalian IgA and IgM.
IgY antibodies make about 75 % of all immunoglobulins. The serum concentrations of IgY, IgA, and IgM have been reported to be 5.0, 1.25, and 0.61 mg . ml −1 , respectively [bib_ref] Studies on the secretory immunologic system of fowl III. Serum and secretory..., Leslie [/bib_ref].
In spite of functional homology between avian IgY and mammalian IgG, there are differences in molecular weight, structure and biochemical functions. They are composed of two identical heavy and light chains bound together with disulphide bonds. Furthermore, they own a variable antigen-binding-site and a constant highly conserved region. IgY's are distinguished from IgG's with major heavy chains and therefore a higher molecular weight. Moreover, avian IgY has a shorter and thus less flexible hinge region than IgG [bib_ref] IgY: clues to the origins of modern antibodies, Warr [/bib_ref]. It has also been suggested that IgY's have more hydrophobic molecules than IgG antibodies and also have a lower isoelectric point [bib_ref] A comparative study between the adsorption of IgY and IgG on latex..., Dávalos-Pantoja [/bib_ref].
IgY's neither activate the complement system like IgG's [bib_ref] Peroral immunotheraphy with yolk antibodies for the prevention and treatment of enteric..., Carlander [/bib_ref] nor interact with rheumatoid factors in Immunoassays [bib_ref] A comparative study between the adsorption of IgY and IgG on latex..., Dávalos-Pantoja [/bib_ref]. The structural differences between IgY's and IgG's are shown in . Furthermore, IgY's do not interfere with protein A or C. This may not simplify the purification but there are several methods for IgY extraction from the egg yolk.
The antibody diversity in chicken is distinguished from mammals and is based on gene conversion and somatic hyperconversion.
## Production of igy's starts with the immunization of chickens
IgY's are an inexpensive and an alternative form of polyclonal antibodies. For immunization the hens were injected with specific antigens intramuscularly for several intervals. Antibodies are transferred from hen to the offspring via the latent stage of the egg. The immune-incompetent chick is able to resist various infectious diseases due to the innate immune system given from the hen. The transfer of IgY antibodies from maternal serum to the egg yolk is analogous to the cross-placental transmission in mammals [bib_ref] Immunoglobulin classes in the hen's egg: their segregation in yolk and white, Rose [/bib_ref]. During the last days of the embryonic development phase, IgY is transported across the yolk sac membrane into the embryonic bloodstream [bib_ref] Quantitation of maternal-fetal IgG transport in the chicken, Kowalczyk [/bib_ref]. Recent studies show that the transfer of IgY from serum to egg yolk is a receptormediated process which allows a selective transfer of antibodies from the maternal serum [bib_ref] Deposition of genetically engineered human antibodies into the egg yolk of hens, Mohammed [/bib_ref] [bib_ref] Sequences in antibody molecules important for receptor-mediated transport into the chicken egg..., Morrison [/bib_ref]. Research supported that a specific sequence (His-Glu-Ala-Leu: HEAL) in the FC-region and an intact hinge region are required for transport. Changes in this amino acid sequence inhibit IgY transport into the egg yolk. Roughly 4-6 days after inoculation, IgY's can be detected in the egg yolk [bib_ref] Antibody production and transfer to egg yolk in chickens, Patterson [/bib_ref] [bib_ref] Comparison of antibody production to human interleukin-6 (IL-6) by sheep and chickens, Woolley [/bib_ref].
The antigen dose significantly influences the immune response. Through renewed immunization the concentration of antibodies can be strongly increased in the egg yolk. This process avoids bleeding of animals, stress and permits the harvest of large amounts of antibodies. Furthermore, long-lasting high titre of IgY can be detected in egg yolk [bib_ref] Efficient production of chicken egg yolk antibodies against a conserved mammalian protein, Gassmann [/bib_ref].
Schade et al. [bib_ref] Chicken egg yolk antibodies (IgY-technology): a review of progress in production and..., Schade [/bib_ref] published a review about all IgY extraction and purification methods. The most frequent used processes are with help of polyethyleneglycol [bib_ref] IgY technology: extraction of chicken antibodies from egg yolk by polyethylene glycol..., Pauly [/bib_ref] , ammonium or sodium sulfate [bib_ref] IgY antibodies protect against human Rotavirus induced diarrhea in the neonatal gnotobiotic..., Vega [/bib_ref] [bib_ref] Comparison of four purification methods for the production of immunoglobulins from eggs..., Akita [/bib_ref]. After purification IgY's show a high stability over a few months to a few years under specified conditions [bib_ref] Chicken egg yolk antibodies (IgY-technology): a review of progress in production and..., Schade [/bib_ref].
Gene-specific antibodies make the complicated multistep process for specific antigen synthesis redundant [bib_ref] The use of gene-specific IgY antibodies for drug target discovery, Zhang [/bib_ref].
## Stability of avian igy's
For food fortification and the intestinal treatment to cure or prevent diseases, it is necessary, to improve the heat and pH resistance of IgY's. Several studies have been conducted to evaluate the stability of these antibodies. The The adaptive immune system of avian and mammals is based on immunoglobulins. Birds produce three types of immunoglobulins (IgM, IgY and IgA), and the mammals five (IgM, IgD, IgG, IgE and IgA). In both cases the acquired immunity includes T-cell receptors (TCRs), polymorphic MHC class I and II molecules, primary and secondary lymphoid organs, rearrangement of the recombination-activating gene (RAG) and antibody class switch activity of IgY may be reduced by gastric conditions, particularly due to a low pH value [bib_ref] Anti-E. coli lmmunoglobulin Y isolated from egg yolk of immunized chickens as..., Shimizu [/bib_ref]. Chicken antibodies are quite stable against digestive enzymes trypsin and chymotrypsin. Although there is a high loss of activity through Pepsin under low pH conditions in a short time [bib_ref] Anti-E. coli lmmunoglobulin Y isolated from egg yolk of immunized chickens as..., Shimizu [/bib_ref] [bib_ref] Use of egg yolk-derived immunoglobulin as an alternative to antibiotic treatment for..., Shin [/bib_ref] , IgY is stable at pH 4-9 and up to 65°C in aqueous conditions. This resembles IgG, which is stable at pH 3-10 and up to 70°C [bib_ref] Oral passive immunization effect of anti-human rotavirus IgY and its behavior against..., Hatta [/bib_ref] [bib_ref] Molecular stability of chicken and rabbit immunoglobulin G, Shimizu [/bib_ref]. However, the resistance of IgY to low pH conditions increases if high salt concentrations or stabilizing reagents e.g. sorbitol are present. Xylitol does not have such an effect on heat resistance [bib_ref] Acid stability of antihelicobacter pyroli IgY in aqueous polyol solution, Lee [/bib_ref]. Furthermore, the addition of sucrose increases the resistance against low pH ranges, heat and pressure [bib_ref] Egg Yolk Antibody (Ig Y) stability in aqueous solution with high sugar..., Shimizu [/bib_ref]. The egg yolk may be able to stabilize IgY under low pH conditions and higher temperatures as well [bib_ref] Studies on the stability of chicken IgY in different sugars, complex carbohydrates..., Jaradat [/bib_ref]. Several studies also describe protein modifications and coatingmethods, respectively, and their positive effect against inactivation through digestion, heat or acidic conditions. Heat resistance of human IgG antibodies can be increased with help of Polyethyleneglycol-modifications [bib_ref] Physicochemical and biological properties of poly(ethylene glycol)-coupled immunoglobuling G, Suzuki [/bib_ref]. Encapsulation on IgY's in liposomes has been detected to stabilize antibodies against peptic hydrolysis under acidic conditions [bib_ref] Encapsulation of chicken egg yolk Immunoglobulin G (IgY) by liposomes, Shimizu [/bib_ref]. Furthermore, the protective effect of microencapsulation with chitosan-alginate on IgY during gastric passage in vivo has been evaluated [bib_ref] Chitosan-alginate microcapsules for oral delivery of egg yolk immunoglobulin (IgY): in vivo..., Li [/bib_ref]. IgY was reported to be stable for an extended durability for 14 weeks except at temperatures over 50°C [bib_ref] Studies on the stability of chicken IgY in different sugars, complex carbohydrates..., Jaradat [/bib_ref] [bib_ref] Eggs: conveniently packaged antibodies. Methods for purification of yolk IgG, Jensenius [/bib_ref]. Gujral et al. analyzed the IgY antibody stability during 78 weeks of storage at room temperature. In this case a combination with mannitol stabilized the IgY during the extended period [bib_ref] In-vitro and in-vivo binding activity of chicken egg yolk immunoglobulin Y (IgY)..., Gujral [/bib_ref].
## Advantages for using igy's
The production of IgY's is a non-invasive alternative to current methods. The keeping of chickens is inexpensive and the animals are easy to handle. It avoids repetitive bleeding and pain of laboratory animals. Furthermore, it is also very effective. The IgY-titre in the egg yolk of immunized chickens remains very high for a long period of time [bib_ref] Efficient production of chicken egg yolk antibodies against a conserved mammalian protein, Gassmann [/bib_ref]. One egg yolk contains more antibodies compared to the average isolated from the blood of immunized rabbits [bib_ref] Generation and application of chicken egg-yolk antibodies, Tini [/bib_ref] [bib_ref] Advances in the value of eggs and egg components for human health, Kovacs-Nolan [/bib_ref]. IgY's are able to prevent or cure human diseases as described previously. IgY's are not able to pass the gastric barrier [bib_ref] Acid stability of antihelicobacter pyroli IgY in aqueous polyol solution, Lee [/bib_ref]. This offers various possibilities for passive immunization in the digestive system against pathogenic viruses and bacteria. But there are further application possibilities, e.g. in diagnostic tests or protein purification processes. IgY's are less acid-and heat resistant than rabbit IgG [bib_ref] Molecular stability of chicken and rabbit immunoglobulin G, Shimizu [/bib_ref] [bib_ref] In vitro studies of chicken egg yolk antibody (IgY) against Salmonella enteritidis..., Lee [/bib_ref]. In 2002 Lee has shown that it is possible to stabilize the pH resistance with the help of sorbitol [bib_ref] Acid stability of antihelicobacter pyroli IgY in aqueous polyol solution, Lee [/bib_ref]. A further problem as well is the molecular resistance against proteolytic cleavage during the stomach passage. In 1993, Hatta and colleagues published, that IgY activity was reduced within a short time by pepsin, trypsin and chymotrypsin [bib_ref] Oral passive immunization effect of anti-human rotavirus IgY and its behavior against..., Hatta [/bib_ref].
An advantage is that IgY's neither activate the complement system nor bind to protein A and G, rheumatoid factors or cell surface [bib_ref] IgY: clues to the origins of modern antibodies, Warr [/bib_ref] [bib_ref] Generation and application of chicken egg-yolk antibodies, Tini [/bib_ref]. The phylogenetic distance between mammals and avian causes a high immune response of birds to mammalian antigens. This increases Structure of avian IgY versus mammalian IgG. Both molecules contain two heavy and two light chains, which consist of a variable domain (VH and VL) and four constant domains (CH1, CH2, CH3 and CH4), respectively. IgG has a longer hinge region, which makes it more flexible than avian IgY the binding specificity of IgY which is shown through immune diagnostics procedures like immunohistochemistry, ELISAs or immunofluorescence.
## Igy antibodies and bacteria
In 1984 Marshall and Warren firstly described Helicobacter pylori-previously named Campylobacter pylori-and its association with active chronic gastritis. They characterized the bacterium as a spiral, gram-negative organism with sheathed flagella. The organism is found in the human stomach and induces inflammation of the gastric mucosa [bib_ref] Unidentified curved bacilli in the stomach of patients with gastritis and peptic..., Marshall [/bib_ref]. It is known that H. pylori causes chronic gastritis, peptic ulcers, gastric mucosa associated lymphoid tissue (MALT) lymphoma and gastric cancer [bib_ref] Management of helicobacter pylori infection-the Maastricht IV/ Florence consensus report, Malfertheiner [/bib_ref] [bib_ref] Pathogenesis of helicobacter pylori infection, Kusters [/bib_ref]. Presently 70 % of all people in the world are infected with H. pylori [bib_ref] What is gastritis? What is gastropathy? How is it classified?, Kayaçetin [/bib_ref]. But only a small number develop pathological symptoms. Previous studies showed that the eradication of H. pylori inhibits the reoccurrence of new gastric ulcers, cures several MAL-Tomas and prevents peptic ulcers and gastric cancer [bib_ref] Management of helicobacter pylori infection-the Maastricht IV/ Florence consensus report, Malfertheiner [/bib_ref] [bib_ref] Effect of Helicobacter pylori eradication on subsequent development of cancer after endoscopic..., Uemura [/bib_ref].
Nowadays the infection with the bacterium can be detected fast and often successfully medicated with specific antibiotic regimes which Wu et al. described in the Review of evidence-based therapies against H. pylori. But novel tools are necessary because there is a critical increase of antibiotic resistances.
Previous studies showed a positive effect of specific IgY's against bacterial intoxications. The use of an oral therapy with IgY's offers the possibility of passive immunization. There are several strategies of using IgY's in host protection: a) inhibition of bacterial enzyme activity, b) toxin neutralization, c) blocking cell adhesion of microorganisms. The most effective way for passive immunization is using IgY produced as a response to previously selected characteristic bacterial antigens.
First trials for application of IgY's in neonatal piglets were already conducted in 1998. Threatening enteric infections were treated with specific IgY antibodies against E. coli K88+ [bib_ref] Passive protective effect of egg-yolk antibodies against enterotoxigenic Escherichia coli K88+ infection..., Marquardt [/bib_ref]. Later these therapeutic options were also validated with other E. coli, Salmonella, Campylobacter and Rotavirus strains [bib_ref] In vitro studies of chicken egg yolk antibody (IgY) against Salmonella enteritidis..., Lee [/bib_ref] [bib_ref] Growth inhibition of escherichia coli 987P by neutralizing IgY antibodies, Sunwoo [/bib_ref] [bib_ref] Immune responses in chickens against lipopolysaccharide of escherichia coli and salmonella typhimurium, Sunwoo [/bib_ref] [bib_ref] Randomized, placebo-controlled, clinical trial of hyperimmunized chicken egg yolk immunoglobulin in children..., Sarker [/bib_ref] [bib_ref] Oral administration of antibodies as prophylaxis and therapy in Campylobacter jejuni-infected chickens, Tsubokura [/bib_ref]. Enteric infections caused by Salmonella are another problem in chicken meat. Chalghoumi and colleagues published a summarized review about the need of alternative methods to prevent enteric infections in 2010. IgY's may be a useful and attractive tool to passively immunize chickens and lower the infection rate in humans after consumption [bib_ref] Hen egg yolk antibodies (IgY), production and use for passive immunization against..., Chalghoumi [/bib_ref].
LeClaire et al. delivered chickens with staphylococcal enterotoxin B (SEB) from Staphylococcus aureus. After immunization they purified the egg yolk and treated rhesus monkeys with the extract. At the same time the monkeys were exposed to a lethal aerosolized dose of SEB. All animals which were treated with specific anti-SEB-IgY survived [bib_ref] Protection against bacterial superantigen staphylococcal enterotoxin B by passive vaccination, Leclaire [/bib_ref].
The actual transmission of H. pylori has never been clarified. The H. pylori urease-enzyme may be mainly responsible for colonization of human gastroduodenal mucosa. This enzyme allows the survival in the acidic gastric milieu. Urease creates a neutral microenvironment around the organism. It catalyzes the cleavage of urea to ammonia and carbon dioxide. This immediately increases the ph value and allows the implantation in the mucosa. The organism burrows into the mucosa to reach the epithelial cells where a less acidic environment is present [bib_ref] Host-bacterial interactions in helicobacter pylori infection, Amieva [/bib_ref]. Developed IgY's against the whole-cell lysate of a H. pylori strain inhibit growth and urease activity of the organism in vitro [bib_ref] Use of egg yolk-derived immunoglobulin as an alternative to antibiotic treatment for..., Shin [/bib_ref]. In 2004 Horie and colleagues demonstrated the efficiency of specific antiurease-IgY in designed drinking yogurt against H. pylori infections. They showed that the oral intake of highly specialized antibodies suppress the infection in humans. Due to urease inhibition the adhesion to the mucosa was also inhibited [bib_ref] Suppressive effect of functional drinking yogurt containing specific egg yolk immunoglobulin on..., Horie [/bib_ref]. In 2009 Attallah and his group developed a reliable model for H. pylori infection in mice. Most cases of gastritis with inflammatory changes were detected. They developed IgY-antibodies against the cell lysate of a human pathogenic H. pylori strain in chicken and treated diseased mice with it. Results demonstrated that the passive immunized mice showed a significant lower degree of infection and gastritis than untreated animals [bib_ref] Efficacy of passive immunization with IgY antibodies to a 58-kDa H. pylori..., Attallah [/bib_ref]. Oral administration of preformed specific antibodies is an attractive approach against infections of the digestive system in humans and animals [bib_ref] IgY: clues to the origins of modern antibodies, Warr [/bib_ref] [bib_ref] Antibodies to rotaviruses in chickens' eggs: a potential source of antiviral immunoglobulins..., Yolken [/bib_ref].
## Monoclonal igy-antibodies in gastric cancer therapy
H. pylori is a significant risk factor for the genesis of gastric cancer. As mentioned above, the eradication inhibits reoccurrence of gastric cancer. Changes in cellular signaling cascades lead to cancer development. Cells proliferate and grow uncontrolled quickly. The reasons for this are somatic mutations leading to permanently activated receptors and over-expression of receptors or ligands. Novel therapies addressed these targets. Monoclonal antibodies which are able to bind epitopes of the extracellular domain of growth factor receptors were able to prevent the receptor-ligand interaction and receptor dimerization. There are possibilities influencing the signal molecules, enzymes and intracellular receptors. The antibodies used at present are chimeric, humanized biological therapeutics [bib_ref] Targeted therapies in gastroesophageal cancer, Kasper [/bib_ref]. It is conceivable that IgY's administered orally cause no allergic reaction and can be used as a preventative therapy for gastric cancer caused by H. pylori or as a general prevention against inflammation caused by this bacteria. Additionally, because of its high stability it can be used as a food supplement, which increases the access of this therapeutic tool worldwide [bib_ref] A novel solution for prevention and treatment of alimentary tract diseases, Rahman [/bib_ref].
## Novel therapies against dental caries and periodontitis
Dental caries are a result of oral bacteria like Streptococcus mutans or Streptococcus sobrinus. To inhibit dental breakdown, a sugarless diet or use of fluoride is recommended. In addition to antibiotics or other antibodies, IgY's can be applied against caries causing bacteria [bib_ref] Immunization against dental caries, Koga [/bib_ref]. This was first demonstrated positively in rats [bib_ref] Protection of rats against dental caries by passive immunization with hen-egg-yolk antibody..., Otake [/bib_ref] and then in humans as well [bib_ref] Passive immunization against dental plaque formation in humans: effect of a mouth..., Hatta [/bib_ref].
In addition to caries, periodontitis also plays an important role in dental diseases. Porphyromonas gingivalis may be the main cause for periodontitis [bib_ref] Virulence factors of porphyromonas gingivalis, Holt [/bib_ref]. In 2007 it was demonstrated that IgY's against a membrane-protein of P. gingivalis inhibits the development of a pathogen biofilm on the teeth surface and probably also following periodontal diseases [bib_ref] Egg yolk-derived immunoglobulin (IgY) against Porphyromonas gingivalis 40-kDa outer membrane protein inhibits..., Hamajima [/bib_ref].
## New target for egg yolk antibodies: pseudomonas aeruginosa
Cystic fibrosis (CF), an autosomal recessive hereditary disease, affects mainly the lungs of patients among other symptoms. Causes are mutations in the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR) protein [bib_ref] Identification of the cystic fibrosis gene: chromosome walking and jumping, Rommens [/bib_ref]. This leads to a reduced transport of chloride from exocrine cells and causes the development of viscous mucus in the bronchia. Chronicle lung infections by the opportunistic bacteria Pseudomonas aeruginosa (PA) are the main causes of morbidity and mortality in CF patients. In 2003, Kollberg and colleagues published that rinsing with a solution of specific anti-PA IgY in the evening may prevent the initial adhesion of the bacteria to the mucosal surface of the oropharynx in CF patients [bib_ref] Oral administration of specific yolk antibodies (IgY) may prevent Pseudomonas aeruginosa infections..., Kollberg [/bib_ref]. At the end of their study, all participants in the IgY treated group were still without chronicle PA infections. It was demonstrated that IgY could be used as a possible prophylaxis treatment in CF patients and thus reduces the necessity of frequent applications of antibiotics [bib_ref] Oral administration of specific yolk antibodies (IgY) may prevent Pseudomonas aeruginosa infections..., Kollberg [/bib_ref].
IgY's as a new treatment approach for Celiac disease Celiac disease (CD) is an autoimmune disease triggered by the ingestion of gluten-containing grains in susceptible individuals. Here, wheat gluten and similar alcohol-soluble proteins are the causes responsible for the development of intestinal damage. The disease is characterized by a loss of absorptive villi and hyperplasia of the crypts. A gluten free diet is currently the only accepted treatment for celiac disease [bib_ref] Submit your next manuscript to BioMed Central and take full advantage of:..., Chand [/bib_ref]. In 2012 Gujral and colleagues evaluated specific IgY antibodies against gliadin. As previously stated, purified IgY antibodies are extremely sensitive to gastric conditions and are rapidly inactivated by Pepsin under low pH ranges. The development of appropriate sugar protectants enables neutralization and the adsorption of gliadin in both in vitro and in vivo [bib_ref] In-vitro and in-vivo binding activity of chicken egg yolk immunoglobulin Y (IgY)..., Gujral [/bib_ref]. This could mean a revolution in the celiac disease treatment regime and harbors new hope for all affected persons.
# Conclusion
Using IgY in disease prevention and outlook Today IgY technology is a fast spreading field in life sciences. It may not be only a future vision anymore. IgY's in nose spray, cosmetics, body lotions, functional food e.g. yogurt, powder supplement may be able to prevent or cure human diseases. The use of IgY antibodies offers the possibility for reducing antibiotics in the treatment of bacterial infections of the digestive system. In addition, egg yolk antibodies provide a new approach for attending Candida albicans and intestinal parasites as well. Furthermore, IgY treatment promises passive immunization in newborns and immune compromised patients. Patients during chemotherapy do not produce sufficient amounts of antibodies in response to vaccines. In this case passive immunization has provided new opportunities and an alternative to current treatment strategies. It may also be a new approach in therapy for chronic inflammatory bowel diseases e. g. crohn's disease or ulcerative colitis. The local therapy with IgY antibodies in the digestive system could replace the current systemic treatment regime. Furthermore, IgY's promise a new field for all basic prophylactic treatment strategies.
There is a high potential for IgY antibodies in the treatment of a variety of diseases and an even more prospective future.
Competing interests TD is the President and CEO of IgY Immune Technologies and Life Sciences Inc. CO is a consultant to IgY Immune Technologies and Life Sciences Inc.
Authors' contributions SM, AS, JZ, TD and CO: Contributed to the original plan for the paper, wrote sections of the paper and contributed to draft revisions. All authors read and approved the final manuscript.
[fig] Fraunhofer: Institute for Cell Therapy and Immunology (IZI), Perlickstraße 1, 04103 Leipzig, Germany. 2 IgY Immune Technologies and Life Sciences Inc., Thunder Bay, Canada. 3 Project Centre for Biomedical Engineering and Advanced Manufacturing, McMaster University, Hamilton, Canada. [/fig]
[table] Table 1: Basic comparison of the adaptive immune system between avian and mammals [/table]
|
Patterns and correlates of mis-implementation in state chronic disease public health practice in the United States
Background: Much of the disease burden in the United States is preventable through application of existing knowledge. State-level public health practitioners are in ideal positions to affect programs and policies related to chronic disease, but the extent to which mis-implementation occurring with these programs is largely unknown. Mis-implementation refers to ending effective programs and policies prematurely or continuing ineffective ones.Methods: A 2018 comprehensive survey assessing the extent of mis-implementation and multi-level influences on mis-implementation was reported by state health departments (SHDs). Questions were developed from previous literature. Surveys were emailed to randomly selected SHD employees across the Unites States. Spearman's correlation and multinomial logistic regression were used to assess factors in mis-implementation. Results: Half (50.7%) of respondents were chronic disease program managers or unit directors. Forty nine percent reported that programs their SHD oversees sometimes, often or always continued ineffective programs. Over 50% also reported that their SHD sometimes or often ended effective programs. The data suggest the strongest correlates and predictors of mis-implementation were at the organizational level. For example, the number of organizational layers impeded decision-making was significant for both continuing ineffective programs (OR=4.70; 95% CI=2.20, 10.04) and ending effective programs (OR=3.23; 95% CI=1.61, 7.40). Conclusion: The data suggest that changing certain agency practices may help in minimizing the occurrence of mis-implementation. Further research should focus on adding context to these issues and helping agencies engage in appropriate decision-making. Greater attention to mis-implementation should lead to greater use of effective interventions and more efficient expenditure of resources, ultimately to improve health outcomes.
burden is preventable [bib_ref] Applying what we know to accelerate cancer prevention, Colditz [/bib_ref] , gaps in delivery of EBPs hinders effective public health practice to improve health.
The mechanisms behind mis-implementation are an important area of inquiry for public health practitioners and researchers [bib_ref] Toward optimal implementation of cancer prevention and control programs in public health:..., Padek [/bib_ref] [bib_ref] A cross-country study of mis-implementation in public health practice, Furtado [/bib_ref] [bib_ref] Understanding mis-implementation in public health practice, Brownson [/bib_ref]. The term mis-implementation refers to the inappropriate termination of evidencebased programs or the inappropriate continuation of non-evidence based programs [bib_ref] Toward optimal implementation of cancer prevention and control programs in public health:..., Padek [/bib_ref]. An example of inappropriate termination of an evidence-based policy in the United States is notable with the rollback of Bush and Obama era healthy school lunch standards, which were relaxed despite evidence they increased school-aged children's consumption of healthy foods [bib_ref] Effect of school wellness policies and the healthy, hunger-free kids act on..., Mansfield [/bib_ref]. Alternately, an example of inappropriate continuation of non-evidence-based programs is the continued use of health fairs for community screenings, interventions and education. While they may help increase visibility of services to subsets of the community, there is limited evidence that they increase screening follow-up, enhance sustained health knowledge, or improve health outcomes [bib_ref] Understanding the impact of colorectal cancer education: a randomized trial of health..., Briant [/bib_ref] [bib_ref] Screening in health fairs. A critical review of benefits, risks, and costs, Berwick [/bib_ref]. Previous work in this area suggests that between 58 and 62% of public health programs are evidence-based [bib_ref] Improving the public health workforce: evaluation of a training course to enhance..., Dreisinger [/bib_ref] [bib_ref] Training the Workforce in Evidence-Based Public Health: An Evaluation of Impact Among..., Gibbert [/bib_ref]. However, only 37% of chronic disease prevention staff in state health departments reported programs are often or always discontinued when they should continue [bib_ref] Understanding mis-implementation in public health practice, Brownson [/bib_ref]. These studies set a baseline for mis-implementation context but did not further explore the contributing factors to these decision-making processes and did not assess the degree to which misimplementation is occurring in chronic disease public health practices.
Exploring the evidence-based decision-making (EBDM) and related literature suggests that a mix of individual, organizational, agency and policy related factors are at play in organizational decision-making, including whether or not to begin or continue implementing programs, and program outcomes [bib_ref] Evidence-based public health: a fundamental concept for public health practice, Brownson [/bib_ref] [bib_ref] Understanding mis-implementation in public health practice, Brownson [/bib_ref]. EBDM which is an approach to decision-making that combines the appropriate research evidence, practitioner expertise, and the characteristics, needs, and preferences of the community, can have a significant impact on health-related outcomes [bib_ref] Evidence-based public health: a fundamental concept for public health practice, Brownson [/bib_ref] [bib_ref] Building capacity for evidence-based public health: reconciling the pulls of practice and..., Brownson [/bib_ref]. Specifically, leadership support in applying EBDM frameworks can enhance an organization's capacity for improved public health practices [bib_ref] Evidence-based public health: a fundamental concept for public health practice, Brownson [/bib_ref] [bib_ref] Fostering moreeffective public health by identifying administrative evidence-based practices: a review of..., Brownson [/bib_ref] [bib_ref] The relationship between local public health agency administrative variables and county health..., Harris [/bib_ref]. Concurrently, contextual factors, cost burden and characteristics of early adopters are also factors in mis-implementation outcomes [bib_ref] Training the Workforce in Evidence-Based Public Health: An Evaluation of Impact Among..., Gibbert [/bib_ref] [bib_ref] Unpacking the complexities of de-implementing inappropriate health interventions, Norton [/bib_ref]. These concepts support the factors that inform our original mis-implementation framework [bib_ref] Toward optimal implementation of cancer prevention and control programs in public health:..., Padek [/bib_ref]. In a cross-sectional U.S. study of local health departments, higher perceived organizational supports for EBDM were associated with lower perceived frequency of inappropriate continuation [bib_ref] Perspectives on program misimplementation among U.S. local public health departments, Allen [/bib_ref]. In crosscountry comparisons of mis-implementation involving Australia, Brazil, China, and the United States, leadership support and political contexts were common factors in whether chronic disease programs continued or ended inappropriately across four countries [bib_ref] A cross-country study of mis-implementation in public health practice, Furtado [/bib_ref].
State health departments (SHDs) are a significant driver of public health programs within the United States. Most federal funds for chronic disease prevention are directed through state health departments, and they provide resources and guidance to local level implementation of public health programs. This dynamic of the SHD as the pass-through organization means that their organizational dynamics are key in the successful outcomes of these programs. A common delivery structure in the U.S. is for local public health agencies and community-based organizations to design how they will implement topic-specific programs as they respond to SHD requests for proposals. Another delivery structure is contractual relationships generated by the SHDs in which local agencies choose from a menu of programmatic approaches to chronic disease prevention provided by the SHD. And while an estimated $1.1 billion dollars flow through state public health chronic disease and cancer prevention programs annually, a majority of these funds focus on secondary prevention (e.g., cancer screening), leaving a scarcity for primary prevention resources. With this scarcity in prevention funding, it is essential that every dollar being directed towards chronic disease programs have maximum impact.
This study seeks to: 1) assess the extent to which misimplementation of chronic disease programs is occurring in state health departments, and 2) identify the most important factors associated with mis-implementation among programs overseen by SHDs [bib_ref] Toward optimal implementation of cancer prevention and control programs in public health:..., Padek [/bib_ref] [bib_ref] The association between evidence-based decision making and accreditation of state health departments, Erwin [/bib_ref].
# Methods
This study is a cross-sectional assessment of decisionmaking practices within state health departments. We surveyed current SHD employees across the U.S. to gather quantitative data to identify the perceived frequency and correlates of mis-implementation within SHD chronic disease units. Human subjects approval was obtained from the Washington University in St. Louis Institutional Review Board (#201812062).
## Survey development
To develop a survey informed by the literature and addressing knowledge and survey gaps, we undertook an extensive literature review. Survey development was also guided by the study team's previously described conceptual framework to ensure measures included EBDM skills, organizational capacity for EBDM, and external influences such as funding and policy climate [bib_ref] Toward optimal implementation of cancer prevention and control programs in public health:..., Padek [/bib_ref].
A literature review of several databases (i.e., PubMed, SCOPUS, Web of Science) was conducted to search for existing survey instruments regarding organizational decision-making. Identified measures were summarized according to setting, audience, psychometric properties, and survey question themes. From our review of 63 surveys, we ended up selecting items from 23 measures to examine in relation to our conceptual framework . Questions pertaining to political influence and mis-implementation decision-making were iteratively developed and refined as there was little published literature available at the time to inform these questions. Drafts for questions in each domain (individual skills, organizational/agency capacity, mis-implementation decision-making, external influences) were updated, and underwent three separate reviews by the study team and a group of practitioner experts to develop a final draft of the study instrument. Since the survey had not been previous validated, the final draft survey underwent cognitive response testing with 11 former SHD chronic disease directors. Reliability test-retest of the revised draft with 39 current SHD chronic disease unit staff found consistency in scores and only minor changes to the survey were needed.
## Measures
Survey measures addressed the following topics: participant demographic characteristics, EBDM skills, perceived frequency of mis-implementation, reasons for mis-implementation, perceived organizational supports for EBDM, and external influences. External influences included perceived governor office and state legislative support for evidence-based interventions (EBIs), and perceived importance of multi-sector partnering. Exact item wording is provided in the national survey located in Additional file 1. Survey questions for EBDM skills, organizational supports, and external influences consisted of 5-point Likert scale responses. Response options ranged from either "Strongly Disagree to Strongly Agree" or "Not at all" to "Very great extent".
Perceived frequency of mis-implementation was assessed with two questions: "How often do effective programs, overseen by your work unit, end when they should have continued"; and "How often do ineffective programs, overseen by your work unit, continue when they should have ended." The response options were: never, rarely, sometimes, often, and always. These variables will subsequently be referred to as inappropriate termination and inappropriate continuation, respectively.
## Participants
Participants for the survey were recruited from the National Association of Chronic Disease Directors (NACD D) membership list. The NACDD membership lists consists of SHD employees working in their respective chronic disease units. Participants were randomly selected from the membership roll after individuals from territories and non-qualifying positions (administrative support & financial personnel) were removed. Emails were sent out in June 2018 inviting a randomly selected sample of 1239 members to participate in a Qualtrics online survey. Participants were offered the option of a $20 Amazon gift card or to have us make a donation to a public health charity of their choosing. A follow-up email was sent two weeks after the initial email with a reminder phone call a week later. Non-respondents could have received up to three reminder emails and two reminder voicemails or a single phone conversation to address questions. There was no ability to directly compare non-respondents with respondents given the lack of key characteristics (e.g., role in the agency, years working in the agency) in our initial list for sampling. The online survey closed at the end of August 2018.
## Data cleaning and analysis
Respondents who answered any of the questions beyond demographic questions were included in the sample. State-level variables, such as population size, funding from the Centers for Disease Control and Prevention (CDC) (the major funding source for state chronic disease control), and state governance type, were added to the data set from other publically available datasets such as the CDC grant funding profile, Association of State and Territorial Health Officials (ASTHO) State Profiles 1 and Public Health Accreditation Board data [bib_ref] The association between evidence-based decision making and accreditation of state health departments, Erwin [/bib_ref]. Dichotomized versions of Likert scale variables were created given the limited distribution of responses across the original scale and to facilitate interpretation. Responses that included Agree or Strongly Agree were coded as 1 while all other remaining responses were coded as 0.
Descriptive statistics were calculated for all variables in SPSS version 26. To assess associations, a Spearman's correlation was calculated between each nondichotomized mis-implementation variables and the individual demographic characteristics, individual skills, organizational capacity for EBIs and external factors. Multinomial logistic regression was then used to assess how variables were predictive of mis-implementation outcomes. The dependent variables (inappropriate termination & inappropriate continuation) were recategorized to 1) often/always 2) sometimes and 3) never/rarely (reference category). Multinomial regression was used as the assumption of proportional odds was violated with an ordinal regression. The independent variables were dichotomized (as described above). Two separate models were fit: the first assessing inappropriate termination among programs overseen by SHDs and the second assessing inappropriate continuation among programs overseen by SHDs. We decided two separate models were appropriate as inappropriate termination and inappropriate continuation are two different phenomena within the overall mis-implementation concept. An initial model for each of the two dependent variables was run for each domain with all their respective variables included. All variables shown to be significant in these first runs of the model were then added to a final version of each model (inappropriate termination and inappropriate continuation).
# Results
## Demographic characteristics
The final response rate was 48.3% (n=643). There were respondents from every state, but the number of responses per state was not proportional to state population size. In the interest of confidentiality, responses were grouped by ASTHO defined regions, and there was a relatively even distribution of participants across regions [fig_ref] Table 1: Participant characteristics of U [/fig_ref]. Half (50.7%) of the respondents were chronic disease program managers and on average had been in their position for over six years. Most respondents worked across multiple health areas with cancer as the most represented program area. Thirty-five percent of respondents had a master's or higher degree related to public health.
## Mis-implementation patterns
When asked "How often do effective programs, overseen by your work unit, end when they should have continued," 50.7% of respondents indicated sometimes, often or always [fig_ref] Table 2: Reported frequency and reasons for mis-implementation from a survey of U [/fig_ref]. Respondents were asked to choose the top three reasons for effective programs ending (but not in a ranked order). The most common responses were: funding priorities changed/funding ended (87.6%); support from leaders in your agency changed (38.9%); support from policy makers changed (34.2%) and program was not sustainable (30.2%) [fig_ref] Table 2: Reported frequency and reasons for mis-implementation from a survey of U [/fig_ref].
Regarding inappropriate continuation, when asked "How often do ineffective programs, overseen by your work unit, continue when they should have ended," 48.5% of respondents indicated sometimes, often or always. Respondents were also asked to choose the top three common reasons for ineffective programs continuing (not in ranked order). The most commons responses were: funder priorities to maintain program (43.4%); policy makers' request or requirements to continue (42.9%); agency leadership requests to continue (37.9%); and standard is to maintain status quo (36.5%) [fig_ref] Table 2: Reported frequency and reasons for mis-implementation from a survey of U [/fig_ref].
## Mis-implementation correlates
The number of years a participant had been working in their current position (r= − 0.11), years they had been working at their agency (r= − 0.09) and years they had been working in public health (r= − 0.10) were shown to have small negative significant correlations with inappropriate continuation [fig_ref] Table 3: Practitioner, Organization, and External Correlates of Mis-Implementation in U [/fig_ref] , meaning more years of experience were associated with lower likelihood of inappropriate continuation. None of the individual skills were shown to have a statistically significant association with either inappropriate termination or inappropriate continuation. All of the organizational capacity variables were shown to have a small negative significant association with both mis-implementation variables, meaning higher perceived organizational capacity was associated with lower perceived frequency of mis-implementation [fig_ref] Table 3: Practitioner, Organization, and External Correlates of Mis-Implementation in U [/fig_ref]. External variables related to lawmakers' priorities and support were shown to have small negative significant, associations with both the inappropriate termination and inappropriate continuation variable.
In the final model for inappropriate termination (
# Discussion
A set of organization/agency capacity factors demonstrated more consistent association with misimplementation outcomes than individual skills of staff. These factors demonstrated an inverse relationship with mis-implementation outcomes (e.g., as agency capacity increased, the association with mis-implementation rates decreased). These findings are consistent with our earlier study among US local health departments, which found organizational supports for EBDM were associated with lower perceived frequency of inappropriate continuation [bib_ref] Perspectives on program misimplementation among U.S. local public health departments, Allen [/bib_ref]. This suggests agency culture and capacity are significant protective factors against mis-implementation in multiple public health organizations rather than the skills of individual staff. Importantly, the agency-level variable reporting that the number of layers of authority impedes decision-making about programs continuation or ending was found to be strongly associated with both inappropriate termination and continuation. This suggests that highly vertical organizations may be more vulnerable to ineffective decision-making around program continuation or ending. Given that state health departments vary widely in their organizational structures, further work is needed to understand how a large number of layers may affect decision making that leads to more frequent use of evidence-based decision making in public health practice [bib_ref] Building capacity for evidence-based public health: reconciling the pulls of practice and..., Brownson [/bib_ref]. Outside of funding, the primary correlates for inappropriate termination or continuation were changing support from leaders and policymakers. We saw more variability in the reasons for inappropriate continuation versus termination. Inappropriate termination was heavily skewed towards funding priorities changing or ending, which is to be expected given the predominance of state public health programs based on time-limited grant funding. The top four reasons for inappropriate continuation were more spread out across multiple domains. This variability in reasons could demonstrate that the processes that result in an ineffective program continuing may tend to involved multiple domains, but this also allows for more opportunity for modifiability.
The two factors most strongly negatively correlated with inappropriate continuation related to leaders' Again, this suggests agency culture and leadership are strong drivers of mis-implementation outcomes but more specifically how leadership can be related to the importance of EBI use and flexibility in program implementation and adaption. Our findings are largely consistent with the literature in EBDM and have several implications for public health practice. Reviews found organizational climate, leadership support, staff commitment and skills, adequate staffing and low staff turnover, organizational resources, and partnerships affect EBI sustainability [bib_ref] The sustainability of new programs and innovations: a review of the empirical..., Wiltsey Stirman [/bib_ref] [bib_ref] Sustained implementation of evidence-based programs in disadvantaged communities: a conceptual framework of..., Hodge [/bib_ref] [bib_ref] The sustainability of evidence-based interventions and practices in public health and health..., Shelton [/bib_ref]. Policy, in the form of legislation and regulation, are also associated with sustainment of programs in community, clinical and social service settings [bib_ref] The sustainability of evidence-based interventions and practices in public health and health..., Shelton [/bib_ref]. Engaging community leaders and other policy makers throughout programmatic decision-making can increase likelihood of program sustainment [bib_ref] Sustained implementation of evidence-based programs in disadvantaged communities: a conceptual framework of..., Hodge [/bib_ref]. While de-implementation of ineffective clinical tests or services has been studied, there is sparse parallel literature on de-implementation of ineffective public health programs [bib_ref] Toward optimal implementation of cancer prevention and control programs in public health:..., Padek [/bib_ref] [bib_ref] Studying de-implementation in health: an analysis of funded research grants, Norton [/bib_ref] [bib_ref] Interventions aimed at reducing use of low-value health services: a systematic review, Colla [/bib_ref]. As our study illustrates, effective public health practice is not solely based on the effectiveness of the programs themselves but also the capacity of the organizations deliver them. Capacity is multi-faceted, and understanding an organization's culture and hierarchy could reveal more about successful public health program implementation.
# Limitations
Our response rates across states was varied enough that we were not able to study state-level correlates in detail. In the absence of other organizational and administrative data, this study relied on self-report surveys of individual and perceived organizational characteristics. While we asked respondents their level of involvement in decisionmaking, they were not always in the position to be privy to the reasons about decision-making or they joined the agency after a decision about a program had concluded.
Compared with previous pilot work, perceived frequency of mis-implementation in SHD was higher in this study (36.5% vs 50.7% for inappropriate termination and 24.7% vs 48.5% for inappropriate continuation), although some of this difference may be attributable in part to updates to the mis-implementation survey item definitions and changes in the approach to categorization of responses [bib_ref] A cross-country study of mis-implementation in public health practice, Furtado [/bib_ref] [bib_ref] Understanding mis-implementation in public health practice, Brownson [/bib_ref] [bib_ref] Perspectives on program misimplementation among U.S. local public health departments, Allen [/bib_ref]. In earlier studies, the recoded dichotomized mis-implementation variables only included the often/always response. After examining the distribution of the mis-implementation variables responses, we thought it was important to include the "sometimes" response in categorizing misimplementation because "sometimes" still captured the phenomena occurring and that excluding it could potentially leave out nuances in the data.
## Future directions
These results provide a first look at factors that may be related to the phenomena of mis-implementation in public health practice. Later phases of this study include eight case studies highlighting lessons learned around mis-implementation and agent-based modeling to identify the dynamic interactions between the individual, [bib_ref] Toward optimal implementation of cancer prevention and control programs in public health:..., Padek [/bib_ref]. The results of these qualitative case studies will be available in future publications. These models should provide decision-making tools to better facilitate evidence-based decision making. There is also a need to explore mis-implementation in other public health settings. While our study focuses on SHDs, local health departments and non-profit settings are significant implementers of public health programs. There is also sparse information on how mis-implementation may vary across public health program areas (e.g., chronic disease, infectious disease, maternal and child health). Additional comparisons of organizational structures across state health departments could also explore the context underlying the "flattening" variable we found as an important correlate of misimplementation.
# Conclusion
While our understanding of mis-implementation in public health practice is in an early stage, our findings Boldface indicates statistical significance (p< 0.05) The survey items for independent variables were asked on a 5-point scale (strongly disagree to strongly agree) and were dichotomized for model specification into strongly agree/agree (1) and neither disagree or agree/disagree/strongly disagree (0). Model fit statistics for model 1 were X 2 (12)= 83.88, p< 0.001 and for model 2 were X 2 (8)= 103.11, p< 0.001. OR: odds ratio, CI: confidence interval provide practitioners and applied researchers some actionable findings. For example, based on our study and related literature [bib_ref] Fostering moreeffective public health by identifying administrative evidence-based practices: a review of..., Brownson [/bib_ref] [bib_ref] Health sector reform and public sector health worker motivation: a conceptual framework, Franco [/bib_ref] [bib_ref] Larger, smaller, and flatter: the evolution of the modern health care organization, Mcconnell [/bib_ref]. it appears that efficiency and effectiveness may be gained via flattening of public health agencies along with an organizational culture that supports EBDM. Given the emergence of evidence that chronic diseases are a significant moderating factor in outcome of timely disease concerns i.e. COVID-19 and cancer risk [bib_ref] Cancer risk associated with chronic diseases and disease markers: prospective cohort study, Tu [/bib_ref] , suggestions like these could help maximize dollars spent on public health programs ensuring that appropriate evidence-based programs are contributing to improved health outcomes and benefiting the communities they serve.
## Supplementary information
The online version contains supplementary material available at https://doi. org/10.1186/s12889-020-10101-z.
Additional file 1. Mis-Implementation National Survey.
# Acknowledgments
This work was previously presented at the 12th Annual Conference on the Science of Dissemination and Implementation in Washington D.C. in December 2019. The abstract from that presentation was published in proceedings from that conference at Implementation Science [bib_ref] Predictors of mis-implementation of chronic disease control programs in state health departments, Padek [/bib_ref]. We would like to acknowledge Melissa Franco for help with initial survey development, testing and data collection and Rebekah Jacob for consultation regarding the data analysis. We'd like to acknowledge other members of our research team who have provided feedback and input about the survey development, data collection process and data analyses: Sarah Moreland-Russell, Ross Hammond, Paul Erwin, Joe Ornstein, and Matt Kasman. We would like to acknowledge our stakeholder advisory board which consisted of former state health department employees who provided feedback throughout this process. The National Association of Chronic Disease Directors have also provided consultation during this project.
# Authors' contributions
The authors contributions are as follows: MMP coordinated the survey development, data collection, conducted initial data analysis and led the writing of the manuscript. SM, PA, & DL contributed to survey development, provided feedback on data analysis and edited and reviewed the final manuscript. EWR assisted with data collection, conducted preliminary data analysis and provided input in the final manuscript. ET provided feedback on data analysis and provided input on final manuscript. RCB is the principal investigator of this study and contributed to survey development, data analysis feedback and reviewed the manuscript. All authors read and approved the final manuscript.
# Funding
This project is funded by the National Cancer Institute of the National Institutes of Health (R01CA214530). Additional support for this project came from National Cancer Institute (P50CA24431, T32CA190194), the Centers for Disease Control and Prevention (U48DP006395). The findings and conclusions in this paper are those of the authors and do not necessarily represent the official positions of the National Institutes of Health or the Centers for Disease Control and Prevention. The funders did not have any influence on the design of the study, data collection, data analysis, interpretation of the data or in the writing of the manuscript.
## Availability of data and materials
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.
Ethics approval and consent to participate Human subjects' approval was obtained from the Washington University in St. Louis Institutional Review Board (#201812062). This study received exempt status per the U. S Health and Human Services, Office for Human Research Protections guidelines 45 CFR 46.101(b) [bib_ref] Evidence-based decision making in public health, Brownson [/bib_ref]. The first page of the online survey required participants to click the "I Consent" button in order to progress forward and participate in the study. Participants were sent an Exempt IRB Information Sheet with their email invitation that noted participation in the survey was indicative of their consent.
## Consent for publication
Not applicable.
[fig] 4: My work unit chooses EBPs because they work in similar populations to those we servePlease indicate the extent to which you agree with each statement b Leadership in my work unit has developed a plan to implement Leadership in my work unit preserves through the ups and downs of implementing EBIs 3.60 (1.9) −0.2** −0.25** Leadership in my work unit supports employee's efforts to use EBIs 3.89 (1.03) −0.22** −0.26** Leadership in my work unit reacts to critical issues regarding the implementation of EBIs by openly and effectively addressing the problem. 3.31 (0.93) −0.28** −0.30** flexibility were: work unit's leaders are competent at managing change (r=− 0.30); and leadership reacts to critical issues regarding the implementation of EBIs (r= − 0.30). The two factors most strongly correlated with inappropriate termination were: work unit leadership react to critical issues regarding the implementation of EBIs; work unit leadership encourages planning for sustainability of programs (r= − 0.28, − 0.27 respectively). [/fig]
[table] Table 1: Participant characteristics of U.S. SHD employees in chronic disease prevention units, 2018 survey (N=643) [/table]
[table] Table 3: Practitioner, Organization, and External Correlates of Mis-Implementation in U.S. SHD chronic disease units, 2018 (N=613)To what extent do you agree with the statements regarding your agency and work unit a [/table]
|
Synovial fluid leukocyte apoptosis is inhibited in patients with very early rheumatoid arthritis
# Introduction
Inhibition of T-cell apoptosis in the synovium of patients with established rheumatoid arthritis (RA) was first described in 1995 [bib_ref] Apoptosis in rheumatoid arthritis synovium, Firestein [/bib_ref]. Subsequent work contrasted the virtually complete inhibition of T-cell apoptosis in RA with high levels of T-cell apoptosis in gout [bib_ref] Inhibition of T cell apoptosis in the rheumatoid synovium, Salmon [/bib_ref]. The phenotype of rheumatoid synovial T cells (Bcl-X L high , Bcl-2 low ) demonstrated that their survival was maintained by stromal mechanisms rather than by the common γ-chain cytokines, and IFN-β was identified as a key fibroblastderived survival signal [bib_ref] Interferon-beta mediates stromal cell rescue of T cells from apoptosis, Pilling [/bib_ref]. In addition to their effects on T cells, both IFN-β and synovial fluid from RA patients delay neutrophil apoptosis [bib_ref] Cytokine-mediated inhibition of apoptosis in non-transformed T cells and neutrophils can bedissociated..., Scheel-Toellner [/bib_ref] [bib_ref] Synovial fluid from patients with rheumatoid arthritis inhibits neutrophil apoptosis: role of..., Ottonello [/bib_ref]. These observations led to the concept that inhibition of leukocyte apoptosis, mediated by an expanded fibroblast network in the rheumatoid joint, was an important mechanism that maintains the leukocyte infiltrate in RA and perpetuates disease [bib_ref] Fibroblasts regulate the switch from acute resolving to chronic persistent inflammation, Buckley [/bib_ref].
We recently showed that patients with very early RA, within the first 3 months of symptom onset, have a synovial fluid cytokine profile that is distinct from those of patients with other forms of very early synovitis and of patients with established RA [bib_ref] Early rheumatoid arthritis is characterized by a distinct and transient synovial fluid..., Raza [/bib_ref]. The synovial fluid of patients with very early RA is char-DMARD = disease-modifying antirheumatic drug; G-CSF = granulocyte colony-stimulating factor; GM-CSF = granulocyte-macrophage colony-stimulating factor; IFN = interferon; IL = interleukin; IQR = interquartile range; RA = rheumatoid arthritis; SDF-1α = stromal cell-derived factor-1α. acterized by elevated levels of cytokines that are survival factors for T cells (IL-2, IL-4 and IL-15) and neutrophils (granulocyte-macrophage colony-stimulating factor [GM-CSF] and granulocyte colony-stimulating factor [G-CSF]). We therefore sought to determine whether synovial fluid neutrophil and lymphocyte apoptosis was inhibited in patients with very early RA compared with patients with other very early inflammatory arthritides. We found that patients with very early RA had significantly lower levels of neutrophil apoptosis than did patients who developed non-RA persistent arthritis and those with a resolving disease course. Similarly, lymphocyte apoptosis was absent in patients with early RA, whereas it was seen in patients with other early arthritides. The inhibition of synovial fluid leukocyte apoptosis in the earliest clinically apparent phase of RA distinguishes this from other early arthritides.
# Materials and methods
## Patients
Patients were recruited through the rapid access clinic for early inflammatory arthritis at City Hospital, Birmingham, UK. Ethical permission was obtained and all patients gave written informed consent. All patients had one or more swollen joints and a symptom duration of 3 months or less. Patients with evidence of previous inflammatory joint disease were excluded. No patient had commenced a disease-modifying antirheumatic drug (DMARD) before initial assessment. Joints were aspirated under either palpation or ultrasound guidance. Patients were included in the study if adequate synovial fluid was obtained by palpation or ultrasound-guided aspiration/lavage at initial assessment using a method described previously [bib_ref] Ultrasound guidance allows accurate needle placement and aspiration from small joints in..., Raza [/bib_ref]. Patients were subsequently assessed at 1, 2, 3, 6, 12 and 18 months. If joint effusions were present at follow-up assessments, and if consent to a further arthrocentesis was obtained, then these effusions were aspirated. Patients were assigned to their final diagnostic groups at 18 months. Patients were classified as having RA in accordance with the 1987 American Rheumatism Association criteria [bib_ref] The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid..., Arnett [/bib_ref] , allowing criteria to be satisfied cumulatively. Although the 1987 American Rheumatism Association criteria have no exclusions, we excluded from the RA category patients with alternative rheumatological diagnoses explaining their inflammatory arthritis. Patients were diagnosed with reactive arthritis, psoriatic arthritis, and a number of miscellaneous conditions according to established criteria.
## Synovial fluid cytospin preparation and assessment
When synovial fluid was directly aspirated from the joint, the number of leukocytes per millilitre of synovial fluid was counted using a haemocytometer; a cell count measurement was not performed when the sample was obtained by lavage. Synovial fluid cytospins were made within 30 minutes of joint aspiration or lavage [bib_ref] Ultrasound guidance allows accurate needle placement and aspiration from small joints in..., Raza [/bib_ref]. The short sample processing time minimized the potential for artefactual results due to apoptosis ex vivo. Slides were air dried and stained with Diff-Quik (Dade Behring AG, Düdingen, Switzerland). Cytospins were assessed by an observer who was blinded to clinical details of the patients. Leukocytes were identified on the basis of morphology. Apoptotic cells were identified on the basis of a condensed or fragmented nucleus. Apoptotic neutrophils were distinguished from apoptotic lymphocytes on the basis of acidophilic or basophilic cytoplasmic staining [fig_ref] Figure 1: Apoptotic lymphocytes and neutrophils in synovial fluid cytospin preparations from patients with... [/fig_ref]. Up to 500 cells on each slide were counted where possible. A minimum of 200 cells were counted or the sample was disregarded because of lack of precision.
# Statistical analysis
Categorical variables were compared using the χ 2 test. Numerical variables were compared between patient groups using the Mann-Whitney test. Spearman's test was used to assess the correlation between levels of neutrophil and lymphocyte apoptosis.
# Results
## Patients
Synovial fluid was obtained at presentation from 81 patients with very early inflammatory arthritis. Characteristics of these patients at presentation are shown in [fig_ref] Table 1: Baseline characteristics of patients with very early inflammatory arthritis [/fig_ref]. Seventeen patients developed RA. Of the 17 patients who developed a non-RA persistent inflammatory arthritis, the final diagnoses were unclassified inflammatory arthritis (n = 9), psoriatic arthritis (n = 3), arthritis related to connective tissue disease (n = 3), ulcerative colitis related arthritis (n = 1) and arthritis related to Behçet's disease (n = 1). Of the 47 patients whose disease resolved, the final diagnoses were unclassified inflammatory arthritis (n = 25), gout (n = 9), pseudogout (n = 3), reactive arthritis (n = 7), sarcoidosis (n = 1), ulcerative colitis related arthritis (n = 1) and psoriatic arthritis (n = 1). One of the patients labelled as having an unclassified resolving inflammatory arthritis fulfilled classification criteria for RA at presentation. This patient received an intramuscular injection of steroid after 10 weeks of symptoms, following which the patient remained in remission off therapy throughout follow-up. Patients whose disease progressed to RA were significantly older than patients who developed non-RA persistent synovitis or patients whose disease resolved. Patients whose disease resolved had a significantly shorter duration of symptoms at presentation compared with patients in the other two groups.
The following joints were aspirated at clinical presentation: knee (n = 60), ankle (n = 11), metacarpophalyngeal joint (n = 4), proximal interphalyngeal joint (n = 2), shoulder joint (n = 1), elbow joint (n = 1), wrist (n = 1) and talonavicular joint (n = 1). The knee was aspirated in 10 (59%) of the patients who developed RA, 15 (83%) of the patients who developed a non-RA persistent arthritis and 36 (76%) of the patients with resolving disease. There was no significant difference between the groups in terms of the numbers of patients who had the knee or another joint aspirated (χ 2 test; P = 0.22).
## Synovial fluid leukocyte apoptosis at clinical presentation
Levels of synovial fluid neutrophil apoptosis were available for 71 patients at their initial visit. For the remaining 10 patients the number of neutrophils was too low to quantify the percentage of cells that were apoptotic. Sixteen of the 71 patients developed RA, 15 developed non-RA persistent synovitis and 40 had resolving disease. There was a significant difference in the percentage of apoptotic synovial fluid neutrophils between patients who developed RA and patients in the other two groups (RA versus non-RA persistent synovitis, P = 0.02; RA versus resolving synovitis, P = 0.003; [fig_ref] Figure 2: Synovial fluid neutrophil and lymphocyte apoptosis in patients with very early inflammatory... [/fig_ref]. The median numbers of neutrophils per millilitre of synovial fluid in the initial samples of patients in the three groups were as follows: 2.5 × 10 6 (interquartile range [IQR] 0.2-6.0 × 10 6 ) for patients with RA; 7.1 × 10 6 (IQR 3.1-22.6 × 10 6 ) for patients with non-RA persistent synovitis; 3.9 × 10 6 (IQR 0.04-11.4 × 10 6 ) for patients with resolving synovitis. Patients with RA had a lower absolute number of apoptotic neutrophils per millilitre in their initial synovial fluid samples (0.02 × 10 6 [IQR 0.003-0.09 × 10 6 ]) than did patients with non-RA persistent synovitis (0.14 × 10 6 [IQR 0.02-0.99 × 10 6 ]; P = 0.046) or patients with resolving synovitis (0.14 × 10 6 [IQR 0.02-0.54 × 10 6 ]; P = 0.023).
Levels of synovial fluid lymphocyte apoptosis were available from synovial fluid cytospin preparations of 75 patients at their initial visit. For the remaining six patients the number of lymphocytes in the synovial fluid sample was too low to quantify the percentage of cells that were apoptotic. Sixteen of these patients developed RA, 15 developed non-RA persistent synovitis and 44 had resolving disease. There was a trend toward a lower level of lymphocyte apoptosis in the initial samples of patients with very early synovitis that developed into RA compared with initial samples of patients whose disease resolved, but this did not achieve statistical significance [fig_ref] Figure 2: Synovial fluid neutrophil and lymphocyte apoptosis in patients with very early inflammatory... [/fig_ref]. The median numbers of lymphocytes per millilitre of synovial fluid in the initial samples of patients in the three groups were as follows: 0.9 × 10 6 (IQR 0.1-1.4 × 10 6 ) for patients with RA; 1.1 × 10 6 (IQR 0.8-5.0 × 10 6 ) for patients with non-RA-persistent synovitis; and 0.5 × 10 6 (IQR 0.1-1.6 × 10 6 ) for patients with resolving synovitis. There was no significant difference between the absolute numbers of apoptotic lymphocytes per millilitre in the initial synovial fluid samples from patients in the three groups. Only an occasional apoptotic macrophage was seen, and there were no differences between groups in terms of apoptotic macrophages.
Longitudinal assessment of synovial fluid leukocyte apoptosis Several patients underwent repeat joint aspiration during their follow-up. Results for synovial neutrophil apoptosis from follow-up joint aspirations were available for 16 synovial fluid samples from 10 RA patients, 25 samples from 11 of the patients with non-RA persistent synovitis, and 17 samples from 11 of the patients with resolving synovitis. The number of weeks after symptom onset at which these follow-up samples were collected is shown in [fig_ref] Table 2: Details of patients with very early inflammatory arthritis from whom follow-up synovial... [/fig_ref]. More neutrophil apoptosis was seen in a follow-up sample than at initial presentation in two of the 10 RA patients, six of the 11 patients with non-RA persistent synovitis and four of the 11 patients with resolving synovitis. Prior therapy with a DMARD or parenteral steroid was not associated with enhanced levels of synovial neutrophil apoptosis at subsequent follow up [fig_ref] Table 2: Details of patients with very early inflammatory arthritis from whom follow-up synovial... [/fig_ref]. The maximum percentage of synovial neutrophil apoptosis observed for each patient is shown in [fig_ref] Figure 2: Synovial fluid neutrophil and lymphocyte apoptosis in patients with very early inflammatory... [/fig_ref]. The maximum level of synovial fluid neutrophil apoptosis was significantly lower in patients with RA than in patients with non-RA persistent synovitis (P = 0.0004) or in patients with resolving synovitis (P = 0.002).
Results for synovial lymphocyte apoptosis from follow-up joint aspirations were available for 12 synovial fluid samples from seven RA patients, 26 samples from 11 of the patients with non-RA persistent synovitis, and 21 samples from 13 of the patients with resolving synovitis. The number of weeks after symptom onset at which these follow-up samples were collected is shown in [fig_ref] Table 2: Details of patients with very early inflammatory arthritis from whom follow-up synovial... [/fig_ref]. More lymphocyte apoptosis was seen in a follow-up sample than at initial presentation in two of the seven RA patients, four of the 11 patients with non-RA persistent synovitis, and seven of the 13 patients with resolving synovitis. Prior therapy with a DMARD or parenteral steroid was not associated with enhanced levels of synovial lymphocyte apoptosis at subsequent follow-up [fig_ref] Table 2: Details of patients with very early inflammatory arthritis from whom follow-up synovial... [/fig_ref]. The maximum percentage of synovial lymphocyte apoptosis observed for each patient is shown in [fig_ref] Figure 2: Synovial fluid neutrophil and lymphocyte apoptosis in patients with very early inflammatory... [/fig_ref]. There was a trend toward patients with RA having less synovial lymphocyte apoptosis than patients with non-RA persistent synovitis (P = 0.09) or patients with resolving synovitis (P = 0.054).
Levels of synovial fluid neutrophil and lymphocyte apoptosis in initial and subsequent samples in patients who developed either RA or non-RA persistent inflammatory arthritis are shown in [fig_ref] Figure 2: Synovial fluid neutrophil and lymphocyte apoptosis in patients with very early inflammatory... [/fig_ref] and 2f. The highest levels of leukocyte apoptosis in patients who developed non-RA persistent inflammatory arthritis were seen within the first 20 weeks of symptom onset. Only one patient who developed RA had significant levels of neutrophil apoptosis (12%) and lymphocyte apoptosis (1.5%) in a synovial fluid sample obtained after 10 weeks of symptoms. Lymphocyte and neutrophil apoptosis were virtually absent from all other synovial fluid samples from RA patients, irrespective of disease duration.
There was a statistically significant correlation between the levels of neutrophil and lymphocyte apoptosis in the synovial fluid samples (data not shown; Spearman r = 0.26; P = 0.004).
# Discussion
The early phase of RA, within 3 months of symptom onset, was characterized by very low levels of apoptotic leukocytes within the synovial compartment. During this phase of disease, patients with very early inflammatory arthritis who eventually developed RA had a significantly lower level of synovial fluid neutrophil apoptosis than did patients whose disease resolved. In addition, synovial fluid lymphocyte apoptosis was rarely observed in very early RA. There was a trend toward a higher level of lymphocyte apoptosis in patients with resolving disease compared with those who had very early RA. The inhibition of synovial fluid leukocyte apoptosis that characterizes established RA is thus also apparent in the earliest clinically apparent phase of disease. It is likely that this process contributes to the accumulation of leukocytes in the early rheumatoid lesion and is important in the development of the microenvironment that characterizes established RA. Follow-up assessments over the first 18 months of disease revealed that leukocyte apoptosis was inhibited at all time points in patients who developed RA, despite the fact that a proportion of patients had been treated, before the follow-up arthrocentesis, with antirheumatic drugs that have been reported to induce leukocyte apoptosis [bib_ref] Low dose methotrexate induces apoptosis with reactive oxygen species involvement in T..., Herman [/bib_ref] [bib_ref] Molecular mechanisms of sulfasalazine-induced T-cell apoptosis, Liptay [/bib_ref].
The mechanisms underlying the inhibition of leukocyte apoptosis in patients with very early RA remain undefined. In addition to IFN-β, other mechanisms for the rescue of leukocytes from apoptosis may operate in established RA. Exposure of CD4 + T cells to stromal cell-derived factor-1α(SDF-1α (produced by synovial fibroblasts)) renders T cells less susceptible to apoptosis induced by anti-CD3 stimulation [bib_ref] Stromal cell-derived factor-1-CXC chemokine receptor 4 interactions play a central role in..., Nanki [/bib_ref] and to cytokine deprivation induced apoptosis [bib_ref] Diverse transcriptionalresponse of CD4 + T cells to stromal cell-derived factor (SDF)-1:cell..., Suzuki [/bib_ref]. In Crohn's disease and experimental colitis, a role has been suggested for IL-6 trans signaling in mediating the inhibition of apoptosis in lamina propria T cells [bib_ref] Blockade of interleukin 6 trans signaling suppresses T-cell resistance against apoptosis in..., Atreya [/bib_ref]. The contributions of such mechanisms to the type 1 interferon mediated T-cell rescue that operates in the rheumatoid joint is, at present, unclear. However, it is likely that the high levels of common γ-chain cytokines (IL-2, IL-4 and IL-15) and of G-CSF and GM-CSF that we recently reported in the very early rheumatoid lesion [bib_ref] Early rheumatoid arthritis is characterized by a distinct and transient synovial fluid..., Raza [/bib_ref] play a role in lymphocyte and neutrophil survival in this phase of disease. Although these factors are not present in the joints of patients who progress to non-RA persistent disease, synovial leukocyte apoptosis was partially inhibited in these patients. Therefore, other factors are likely to contribute to leukocyte survival in the very early phase of disease that eventually persists. Macrophagederived IFN-α is a potent survival factor for T cells and neu-trophils [bib_ref] Interferon-beta mediates stromal cell rescue of T cells from apoptosis, Pilling [/bib_ref] [bib_ref] Inhibition of neutrophil apoptosis by type 1 IFN depends on cross-talk between..., Wang [/bib_ref]. Synovial macrophages may thus contribute to the inhibition of leukocyte apoptosis that is seen in early arthritides that progress to persistence.
The low level of leukocyte apoptosis in the initial samples of some patients whose disease eventually resolved was intriguing. A study of lymphocyte apoptosis during the course of a delayed-type hypersensitivity response [bib_ref] The role of apoptosis in the resolution of T cell-mediated cutaneous inflammation, Orteu [/bib_ref] showed that levels of lymphocyte apoptosis change as the lesion develops, being absent during the generation of the response and present during its resolution. In patients with self-limiting inflammatory lesions, the level of lymphocyte apoptosis thus appears to depend on the stage of the process at which the lesion is sampled. Our results suggest a transient inhibition of lymphocyte apoptosis at early stages of the disease process in some patients with a resolving inflammatory arthritis. Potential mechanisms for this include the transient presence of soluble survival factors such as IFN-α. In this study patients who developed RA were older than patients with other early arthritides. It has previously been reported that neutrophils from older individuals are more susceptible to spontaneous apoptosis than are neutrophils from younger persons, and that rescue from spontaneous and Fasinduced apoptosis by cytokines such as GM-CSF and G-CSF is less effective in neutrophils from older individuals than in those from younger ones [bib_ref] Changes in apoptosis of human polymorphonuclear granulocytes with aging, Fulop [/bib_ref] [bib_ref] Spontaneous and Fas-induced apoptotic cell death in aged neutrophils, Tortorella [/bib_ref]. In addition, the rate of spontaneous lymphocyte apoptosis is enhanced in elderly individuals compared with younger individuals [bib_ref] Age-related changes of apoptotic cell death in human lymphocytes, Schindowski [/bib_ref] [bib_ref] Excessive apoptosis of mature T lymphocytes is a characteristic feature of human..., Phelouzat [/bib_ref]. Consequently, it is unlikely that the reduced leukocyte apoptosis observed in RA patients is a feature of their greater age compared with other early arthritis patients.
# Conclusion
Synovial fluid leukocyte apoptosis is inhibited during the earliest clinically apparent phase of RA. This contrasts with other early arthritides, in which significantly higher levels of neutrophil apoptosis are seen in the early lesion. The inhibition of leukocyte apoptosis in the joints of patients with RA, within the first 12 weeks of symptoms, and the presence of apoptosis in the joints of patients whose disease resolves suggest that therapies that induce leukocyte apoptosis may be useful within the first few weeks of symptoms in patients at high risk for development of RA.
KR participated in the design of the study, recruited and followed-up the early arthritis patients, analyzed and interpreted the data, and drafted the manuscript. DST participated in the design of the study and the writing of the manuscript. CYL participated in assessing patients and in performing ultrasoundguided joint aspirations. DP participated in the design of the study and the writing of the manuscript. SJC, VT and FF contributed to the analysis and interpretation of the data. KK participated in assessing patients with early synovitis. JML participated in the design of the study and interpretation of data. CG participated in the design of the study and interpretation of data. CB participated in the design of the study and interpretation of data. MS participated in the design of the study and interpretation of data, and was involved in drafting the manuscript. All authors have read and approved the final manuscript. Number of patients in whom more apoptosis was seen in a follow-up sample than in the initial sample
[fig] Figure 1: Apoptotic lymphocytes and neutrophils in synovial fluid cytospin preparations from patients with very early inflammatory arthritis Apoptotic lymphocytes and neutrophils in synovial fluid cytospin preparations from patients with very early inflammatory arthritis. (a) An apoptotic lymphocyte (dashed arrow). (b) An apoptotic lymphocyte (dashed arrow) and apoptotic neutrophil (solid arrow). (c) Four apoptotic neutrophils (solid arrows). [/fig]
[fig] Figure 2: Synovial fluid neutrophil and lymphocyte apoptosis in patients with very early inflammatory arthritis Synovial fluid neutrophil and lymphocyte apoptosis in patients with very early inflammatory arthritis. (a) Synovial fluid neutrophil apoptosis at clinical presentation in patients with very early inflammatory arthritis divided according to outcome. (b) Synovial fluid lymphocyte apoptosis at clinical presentation in patients with very early inflammatory arthritis divided according to outcome. (c) Maximum synovial fluid neutrophil apoptosis observed in each patient with very early inflammatory arthritis; patients divided according to outcome. (d) Maximum synovial fluid lymphocyte apoptosis in each patient with very early inflammatory arthritis; patients divided according to outcome. (e) Synovial fluid neutrophil apoptosis over time in all samples obtained from patients with very early inflammatory arthritis that eventually persisted divided according to outcome (open circles = non-RA persistent synovitis; closed circles = RA). (f) Synovial fluid lymphocyte apoptosis over time in all samples obtained from patients with very early inflammatory arthritis that eventually persisted divided according to outcome (open circles = non-RA persistent synovitis; closed circles = RA). Comparisons were made using the Mann-Whitney test. RA, rheumatoid arthritis. [/fig]
[table] Table 1: Baseline characteristics of patients with very early inflammatory arthritis [/table]
[table] Table 2: Details of patients with very early inflammatory arthritis from whom follow-up synovial fluid samples were obtained in which neutrophil or lymphocyte apoptosis could be quantified [/table]
|
Oral antibiotics for neonatal infections: a systematic review and meta-analysis
Background: Worldwide many neonates suffer from bacterial infections. Adequate treatment is important but is associated with prolonged hospitalization for intravenous administration. In older children, oral switch therapy has been proven effective and safe for several indications and is now standard care.Objectives: To evaluate the currently available evidence on pharmacokinetics, safety and efficacy of oral antibiotics and oral switch therapy in neonates (0-28 days old).Methods:We performed systematic searches in Medline, Embase.com, Cochrane, Google Scholar and Web of Science. Studies were eligible if they described the use of oral antibiotics in neonates (0-28 days old), including antibiotic switch studies and pharmacological studies.Results: Thirty-one studies met the inclusion criteria. Compared with parenteral administration, oral antibiotics generally reach their maximum concentration later and have a lower bioavailability, but in the majority of cases adequate serum levels for bacterial killing are reached. Furthermore, studies on efficacy of oral antibiotics showed equal relapse rates (OR 0.95; 95% CI 0.79-1.16; I 2 0%) or mortality (OR 1.11; 95% CI 0.72-1.72; I 2 0%). Moreover, a reduction in hospital stay was observed.Conclusions: Oral antibiotics administered to neonates are absorbed and result in adequate serum levels, judged by MICs of relevant pathogens, over time. Efficacy studies are promising but robust evidence is lacking, most importantly because in many cases clinical efficacy and safety are not properly addressed. Early oral antibiotic switch therapy in neonates could be beneficial for both families and healthcare systems. There is a need for additional well-designed trials in different settings.
# Introduction
Infections remain a main cause of morbidity and mortality among newborns.Early-onset sepsis, defined as a proven bacterial infection in the first 72 h of life, has an overall incidence of $1/1000 live births, with a higher incidence in premature and/or very-low-birthweight infants.Forty-five percent of all childhood mortality under 5 years occurs in the neonatal period, of which 22% is due to neonatal infections, including pneumonia.Early diagnosis remains challenging due to non-specificity of both clinical symptoms and laboratory findings.When bacterial infection is probable or proven, parenteral antibiotics are usually prescribed for at least 7 days.Occasionally, when intravenous (iv) access problems occur, or when hospital referral is not possible, as in low-and-middle-income countries (LMICs), newborns are treated with oral antibiotics. In high-income countries (HICs), the full course is generally completed iv.
Intravenous therapy and thus prolonged hospitalization interferes with parent-child bonding and is associated with other hospital-related risks and substantial costs.In older children, oral switch therapy, defined as a switch to oral antibiotics within a treatment course once the patient is clinically well, has been proven to be effective and safe for a variety of indications and is now part of standard practice.The adequacy of antibiotic treatment depends on its specific pharmacological mode of action. Efficacy of penicillins and cephalosporins, both commonly used drugs in neonatology, depends on T .MIC . For vancomycin, efficacy depends on AUC/MIC and for aminoglycosides it depends on C max . The MIC is pathogen specific and cut-off values vary by antibiotic.To our knowledge, no systematic review evaluating the use of oral antibiotics in neonates has been performed. Together with the uncertainties regarding oral absorption in the first weeks of life, the lack of evidence may be a possible reason why oral switch therapy is not yet standard care in neonates. The aim of this systematic review is therefore to evaluate the currently available evidence on safety and efficacy of iv-to-oral switch therapy in neonates, and to evaluate whether, following oral antibiotic administration, adequate serum concentrations are attainable in neonates (0-28 days).
# Methods
## Search strategy and study selection
We performed a systematic review in accordance with the Preferred Reported Items for Systematic Reviews and Meta-analysis (PRISMA),searching Medline, Embase.com, Cochrane Central, Google Scholar and Web of Science on 22 February 2019. The PRISMA statement and full search strategies can be found in the Supplementary data (available at JAC Online). Titles and abstracts were screened and the full text of potential articles was reviewed independently by two reviewers (F. M. K. and G. A. T.-S.). Disagreements were resolved by discussion or through consultation with a third investigator (R. F. K.). Congress abstracts, reference lists and reviews were screened for additional studies. Eligible studies were limited to those performed in humans. Since we expected the amount of evidence to be small, we did not apply any restriction regarding year of publication or language. We included randomized controlled trials (RCTs), intervention studies and retrospective studies describing the use of oral antibiotics including oral switch therapy and pharmacological studies in newborns 0-28 days of age.
The protocol was registered in PROSPERO (protocol number CRD42017070854).
## Data extraction
Three authors (F. M. K., G. A. T.-S. and K. A.) independently extracted the data following a predefined extraction form (see Supplementary data). We did not contact authors for additional information.
## Quality assessment
Quality assessment was performed independently by two authors (F. M. K. and either K. A. or G. A. T.-S.) using the Cochrane Risk of Bias Tool for RCTs 12 and the Newcastle-Ottawa Quality Assessment Scale (NOS) for nonrandomized trials.Since a tool for quality assessment of pharmacological papers is currently lacking, we used the ClinPK statement, a descriptive tool without a grading system, to assess quality of pharmacokinetics papers .
# Data analysis
When possible, data were pooled to assess efficacy of oral treatment. We calculated pooled ORs with 95% CI using Review Manager V5.3. Heterogeneity was assessed using Q statistics and I 2 values and interpreted following the thresholds of the Cochrane Handbook for Systematic Reviews of Interventions.A fixed-effects model was applied when heterogeneity was low (I 2 ,40%), otherwise a random-effects model was used. We performed a sensitivity analysis based on indication for antibiotic treatment. In addition, a subgroup analysis was performed with respect to the clinical indication and antibiotic regimen.
# Results
From a total of 4559 studies, we reviewed the full text of 102 potential articles.shows the selection process. Additionally, five articles were selected through screening of reference lists, leading to 31 selected publications for this review. The characteristics of included studies are described in .
## Quality assessment
Risk of bias in seven out of nine RCTs was low; in the remaining two it was unclear.In all studies, blinding of patients and personnel was considered unethical [e.g. repeated intramuscular (im) placebo administration] and therefore not performed. However, the independent outcome assessors were blinded for treatment allocation. Seven RCTs were registered in a public trial register.The quality of the six observational papers was acceptable . With regard to the pharmacological studies, with focus on pharmacokinetics, overall, quality seems adequate taking into account available methods of analysis at that time. However, in some cases crucial information was missing, such as gestational age (GA) or postnatal age (PNA), or the exact methods used. The complete assessment is included in .
## Study population
As expected, the study population was quite heterogeneous, including both term and preterm infants of different postnatal ages. Four studies were performed in healthy newborns, admitted for a non-infectious indication.The remaining 27 studies included subjects with a clinical condition requiring antibiotics, ranging from prophylactic use to culture-proven infection. Two studies evaluated oral switch therapy in neonates with cultureproven sepsis.Thirteen studies were performed in LMICs. In these trials, antibiotic therapy indication was defined solely on clinical symptoms.Absorption of oral antibiotics Pharmacokinetic analysis and interpretation In 10 papers serum levels were determined using the agar plate diffusion method; the remaining and more recently published papers used HPLC. Most studies provided descriptive data on absorption, mainly C max without further pharmacokinetic estimates (e.g. V and CL). Three papers provided AUC estimates.Regarding interpretation, six papers reported MIC cut-off valueswith only one study reporting a T .MIC .Extracted pharmacokinetic data and administered doses are described in .
## Penicillin
Penicillin, a narrow-spectrum b-lactam antibiotic, was the first oral antibiotic studied in neonates.A weight-equivalent dose was administered orally or im to small groups of healthy subjects of different age (preterm and term newborns, infants or children). This resulted in a lower C max following oral compared with im administration in all age groups. Moreover, a higher AUC following oral administration was reported in newborns compared with older children.
## Systematic review
## Jac
## Ampicillin/amoxicillin
Absorption of oral ampicillin and amoxicillin, both broad-spectrum b-lactam antibiotics, was evaluated in several studies in newborns (GA 28-40 weeks; PNA 0-6 days).Following im injection T max was 30 min, whereas this was on average 4 h for oral therapy. Compared with adults, C max was higher and was reached later in neonates, with even higher levels found in preterm newborns. A small switch study evaluated the bioavailability of ampicillin and amoxicillin, reporting lower plasma concentrations following oral administration compared with equivalent im doses (AUC oral/im, ampicillin 59%, range 22%-94%; amoxicillin 75%).A randomized study in neonates suspected of a bacterial infection compared oral with iv amoxicillin. Initial serum levels were higher in the iv group but comparable concentrations were reached 2 h after oral administration.Most recently a population pharmacokinetic study has been performed among 44 neonates receiving parenteral gentamicin combined with oral amoxicillin. 32 Sampling 2-3 and 6-8 h after administration showed concentrations exceeding the susceptibility breakpoint for amoxicillin against Streptococcus pneumoniae (MIC 2.0 mg/L) strains at both timepoints, meaning that T .MIC is .50% for a 12 h dosing interval.
## Flucloxacillin/nafcillin
Levels of flucloxacillin and nafcillin, both narrow-spectrum blactam antibiotics, have been reported following single-dose administration and combined with other antibiotics to newborns (28-42 weeks GA; 0-6 days PNA). Both drugs appear to be absorbed faster than other penicillins, with a T max of 2 h for both following oral administration.The corrected bioavailability of oral flucloxacillin (corrected for a change in terminal half-life) was reported to be 47.7%, which is almost equivalent to that in adults.Chloramphenicol Chloramphenicol, a broad-spectrum antibiotic, is not generally used in neonatal care due to substantial side effects (e.g. grey baby syndrome).Plasma levels following identical oral and iv dose administration have been evaluated, showing a lower steady-state concentration following oral treatment (oral 13.3 mg/L; iv 25.7 mg/L).Similar results were found in a multicentre study, with only half of term infants reaching therapeutic levels (recommended range in study 10-25 mg/L) following oral administration (25-50 mg/kg/day q12h or q24h depending on PNA).
## Efficacy of oral antibiotics amoxicillin
Amoxicillin is the most studied oral antibiotic in neonates with a probable or proven bacterial infection. Its efficacy depends on the T .MIC . In preterm and term newborns (PNA 1-8 days) with a Systematic review probable bacterial infection, no relapse was reported after oral treatment (80-150 mg/kg/day q12h). Moreover, no side effects occurred and all measured serum concentrations were reported to be above the MICs of targeted pathogens.In a clinical study on Escherichia coli urinary tract infection (UTI), four neonates showed no re-infections in the next 2 years following a 14 day oral treatment of 120 mg/kg/day (in an era with low E. coli amoxicillin resistance).In an RCT including 21 neonates with suspected infection, 11 switched to oral amoxicillin (120 mg/kg q8h) after 48 h of iv therapy (ampicillin/netilmicin). The control group switched to amoxicillin iv. All patients included in the study had negative blood cultures and tolerated oral feeding well without any vomiting. Concentrations remained above the MIC for E. coli for all but three patients (n"2 iv, n"1 oral).Dose optimization through increasing the dosing frequency was suggested and subsequently evaluated in a second study. Ten infants switched to oral amoxicillin (100 mg/kg/day q6h). All plasma concentrations were above the MIC for E. coli without substantial side effects or re-infections.An uncontrolled iv-to-oral switch trial was performed in 222 term neonates with probable or proven group B-streptococcal (GBS) sepsis. Subjects switched to oral amoxicillin (300 mg/kg/day q6h) after 48 h of iv amoxicillin (100 mg/kg per day). All infants had to be asymptomatic and enterally fed at the moment of switch. Because of high serum concentrations, the dose was reduced (to 200 mg/kg/day q6h) in the remaining 158 patients. Serum levels were all above the MIC for GBS. Moreover, therapy was well tolerated without any side effects or reinfections and a reduction of 5 days in hospital admission was seen.Amoxicillin/clavulanic acid A retrospective study evaluated the clinical course and treatment of 172 newborns with a UTI. An increase in use of oral instead of iv therapy was seen over the years. In total, 119 patients switched to oral amoxicillin/clavulanic acid (dose not reported) as continuation therapy. None of the orally treated newborns experienced a relapse in the 6 months after treatment.In another study, oral amoxicillin/clavulanic acid (80 mg/kg/day q12h) was administered successfully to neonates at risk of infection without any reinfections or treatment failure in the first month after treatment completion.Cefalexin A study from Pakistan described the outcome of oral management in neonates with clinical omphalitis. Omphalitis was categorized based on severity; cases without sepsis were treated with cefalexin suspension (50 mg/kg/day q8h) with a success rate of 99.5%, showing that outpatient treatment of clinically well neonates with omphalitis using oral therapy is feasible.Cefpodoxime Switching therapy from iv to oral was performed in 36 term neonates with a probable or proven bacterial infection. After 72 h of iv treatment (ampicillin/sulbactam ! amikacin), patients who were asymptomatic switched to oral cefpodoxime (10 mg/kg/day), a third-generation cephalosporin. Seventy-two matched controls continued on iv therapy. Outcomes were comparable for the two groups, with identical inflammatory parameters in the first week of treatment and no mortality after 1 month. Admission duration was significantly lower and breastfeeding rate was significantly higher among neonates with an oral switch.Flucloxacillin In a small switch study, performed in 1987, neonates at risk of sepsis switched to oral flucloxacillin combined with oral amoxicillin after severe bacterial infection had been ruled out. Plasma concentrations following oral administration were all above the MIC cutoffs for Staphylococcus aureus.
## Linezolid
In a retrospective study, five preterm infants (GA 28+3.5 weeks), treated for late-onset sepsis, who experienced renal failure, switched from iv vancomycin (30 mg/kg/day) to oral linezolid (30 mg/kg/day q8h). C max for all patients but one was above the measured MIC for the causative pathogen.
## Larger efficacy studies including trials in lmic settings
Since there is a need for good outpatient-based management in LMICs, several large trials have taken place evaluating regimens including oral antibiotics. In a controlled trial in .80 villages in India, health workers in the intervention villages were trained in providing neonatal care.When clinical sepsis was suspected but admission refused, neonates received home-based treatment including oral co-trimoxazole. Sepsis-related mortality decreased from 16.6% to 6.9% compared with the period before introduction. Subsequently, several large RCTs comparing home-based antibiotic regimens have been published. The evaluated regimens are described in.
Three regimens were compared in 434 Pakistani children 0-59 days old (72% were 28 days old). Higher treatment failure rates were seen among patients treated with oral co-trimoxazole plus gentamicin compared with other regimens.In a Nepalese study, oral co-trimoxazole was administered in combination with im gentamicin to 67 newborns with a possible bacterial infection.The authors reported a 100% completion rate of oral therapy without any treatment failure. An Ethiopian trial evaluated the implementation of im gentamicin and oral amoxicillin.When infection was suspected, pre-referral medication was given and the patient was referred to the hospital. If referral was not possible, the intervention group continued with home-based treatment; the control group did not receive further treatment. Results seem promising, with a decline in mortality from 17.9 deaths per 1000 live births at baseline to 9.4 per 1000 in the intervention group. In the comparison group, mortality rates declined to a lesser extent, from 14.4 to 11.2 per 1000. However, mortality rates were not significantly lower in the intervention group compared with the control (P " 0.33).
Three RCTs, with a total of 8834 subjects, compared regimens including oral amoxicillin with standard im regimens (penicillin/ gentamicin) in newborns at risk of severe infection. The first trial, in Bangladesh, compared three regimens, including an oral switch regimen, among 2490 children (10% aged 0-6 days)The second trial, in the Democratic Republic of the Congo, Kenya and Nigeria (AFRINEST study) included 3564 infants 0-59 days old (30% 0-6 days old)comparing four regimens including one oral switch to Systematic review JAC amoxicillin. The third study included 2453 infants (44% 0-6 days of age) evaluating similar regimens.Heterogeneity between studies was low. Primary outcome was treatment failure within 8 days, defined as death, clinical deterioration, hospital admission or treatment-related serious adverse events. The combined OR for the orally treated group was 0.95 (95% CI 0.79-1.16; I 2 0%) Mortality within 2 weeks after enrolment was comparable in both groups, with an OR of 1.11 (95% CI 0.72-1.72; I 2 0%). Forest plots are shown in .
Finally, two trials evaluated the use of oral amoxicillin in neonates with tachypnoea as a single symptom of possible infection. The first, in which oral treatment was compared with placebo in 849 infants (78% 0-28 days old; dropout: n"121), showed a higher mortality in the placebo group compared with the treatment group, underlining the potential benefits of antibiotic treatment in infants with fast breathing alone.A second trial, including 2333 neonates (38% 0-6 days old), showed equivalence of oral amoxicillin compared with an im regimen in newborns with fast breathing, with comparable treatment failure rates [22% (im regimen) versus 19% (oral regimen)] and mortality rates (,1% in both groups).
# Discussion
In this systematic review, we collected the currently available evidence on oral antibiotics in neonates. While oral administration is not commonly considered at present in neonates, several pharmacological and efficacy studies have been performed with different types of antibiotics.
In general, adequate serum levels according to the MICs of relevant pathogens can be achieved after oral administration in neonates. Inter-individual variation is observed, which has also been reported following iv administration and should therefore not be used as an argument for discarding oral therapy. C max is reached later after oral administration compared with other routes. Thus, as in older patients, initial therapy should consist of iv antibiotics to quickly reach target concentrations, but can subsequently be switched to oral therapy once the neonate is clinically well.
The efficacy studies showed equal relapse rates and good toleration of oral therapy compared with iv therapy without reporting an increase in side effects. Moreover, in two studies oral administration led to a shorter stay in hospital and more exclusively breast-fed infants. In LMICs, mortality rates have decreased through the introduction of home-based therapy when referral is not possible and simplified antibiotic regimens with an oral switch have shown efficacy similar to that of standard im therapy.
The strength of this review is the fact that we provide a complete overview of all retrieved studies on oral antibiotic use in neonates. Although this provides a great historical overview of an idea that has existed since the 1950s, the heterogeneity of the studies found makes pooling and generalizability to current clinical practice difficult. In an attempt to translate findings to contemporary practice, limitations will be discussed in the light of study design and setting, ethics, techniques used and analysis.
First, study groups were small and without randomization, except for a few large RCTs, introducing a possible selection bias with exclusion of the sicker newborns. In most studies, clinical efficacy, bacterial re-infection or treatment failure is used as the primary outcome. Given the fact that the bacterial re-infection rate is low, a much larger study sample is needed to show non-inferiority or efficacy of oral treatment.Moreover, the clinical indication for antibiotic treatment and infection severity is unclear in a number of studies; therefore data cannot be translated to current practice.
The included studies were performed in both preterm and term infants, sometimes without providing the GA or PNA of the subjects. Drug clearance differs between preterm and term infants and improves with increasing postnatal age, thereby influencing plasma concentrations.Finally it must be stressed that a great(i) ceftriaxone (50 mg/kg/day) im (7 days) benzylpenicillin im ! gentamicin im (7 days) (ii) oral co-trimoxazole (5 mg/kg q8h) ! gentamicin im (7 days) Baqui et al.*(i) oral amoxicillin (50 mg/kg q12h) ! gentamicin im (7 days) benzylpenicillin im ! gentamicin im (7 days)* (ii) benzylpenicillin ! gentamicin im (2 days) followed by oral amoxicillin (5 days)* Tshefu et al.*(i) oral amoxicillin (50 mg/kg q12h) ! gentamicin im (7 days) benzylpenicillin im ! gentamicin im (7 days)* (ii) benzylpenicillin ! gentamicin im (2 days) followed by oral amoxicillin (5 days) (iii) gentamicin im ! oral amoxicillin (2 days) followed by oral amoxicillin (50 mg/kg q12h) (5 days)* Tshefu et al.oral amoxicillin (50 mg/kg q12h) benzylpenicillin im ! gentamicin im (7 days) Mir et al.*(i) gentamicin im ! oral amoxicillin (50 mg/kg q12h) (7 days) benzylpenicillin im ! gentamicin im (7 days)* (ii) procaine benzylpenicillin im ! gentamicin (2 days) followed by oral amoxicillin (5 days)* Degefie Hailegebriel et al.oral amoxicillin (40 mg/kg q8h) ! gentamicin im no treatment Tikmani et al.oral amoxicillin (50 mg/kg q12h) (7 days) placebo *Included in the meta-analysis.
Systematic review variety of antibiotic regimens have been used, including single-dose administration, and sometimes without mentioning the administered dose. Some of the therapies and regimens are rarely prescribed nowadays, partly due to increased concerns regarding antibiotic resistance and the availability of alternatives with fewer side effects. In LMICs, simplified regimens including oral antibiotics are already recommended by the WHO when referral is not possible.Unfortunately, the setting differs greatly from HICs, with refusal of hospital admission still being common and accepted, especially in remote areas. In addition to the differences in setting, the majority of patients are solely diagnosed on clinical symptoms since diagnostic tools are often lacking, possibly leading to an overestimation of the actual number of bacterial infections. Furthermore, the intensity of surveillance due to the execution of the study combined with exclusion of the sicker neonates may have biased mortality rate numbers.
Regarding the pharmacokinetic analysis, ethics requirements of studies have changed and the same holds true for the administration of antibiotics to healthy newborns. With regard to blood sampling, it is no longer considered ethical to collect large volumes or many samples in neonates. Advanced population pharmacokinetic approaches should be applied in further research, using a reduced number of samples per newborn.Further, improved knowledge and better techniques have led to novel antibiotic assays, replacing agar plate dilution methods. Advanced analysis programs are available in order to develop pharmacokinetic models, used for prediction of exposure and drug response, following different dosage regimens in a target population. Those models take into account covariates such as gestational and postnatal age or disease characteristics that possibly influence the pharmacokinetics and dynamics of a drug. Notably, none of the included papers reported covariates in their analysis.
Finally, for the interpretation of results and thus the evaluation of efficacy, the pharmacological mode of action of the specific antibiotic should be considered. The effect of b-lactam antibiotics depends on T .MIC , whereas for aminoglycosides it depends on the C max /MIC ratio. Although six papers do refer to MIC, only one reports T .MIC . Comparison of C max with a single MIC value in case of b-lactam antibiotics has no clinical relevance and cannot be used as a relevant surrogate marker for therapy efficacy. Moreover, MIC levels have increased in recent years, due to an increase in bacterial resistance. In 1992, Giustardi and Coppolareported an amoxicillin MIC of 5 mg/L for E. coli, whereas now an MIC of .8 mg/L is advised to properly treat an E. coli infection. Given these limitations, the currently published studies cannot be used as conclusive evidence to safely change our current guidelines on management of neonatal bacterial infection. However, our findings do give the impression that such studies may be undertaken safely.
## Conclusion and future directions
Early switch to oral antibiotics after a short course of iv antibiotics could be promising in term neonates with a (probable) bacterial infection. This claim is partly supported by the available evidence retrieved in this systematic review. Unfortunately, the lack of large well-designed studies in a high-income setting, evaluating the efficacy of oral antibiotics, together with the uncertainties regarding pharmacokinetics has obstructed further implementation. Future research should focus on the clinical efficacy of oral therapy and the safety of iv-to-oral antibiotic switch therapy in neonates using different types of antibiotics, taking into account the mode of action of the specific antibiotic. These studies should include pharmacokinetic analyses when . (a) Forest plot comparing treatment failure of reference treatment (penicillin/gentamicin im for 7 days) with switch regimen (penicillin/gentamicin im for 2 days followed by oral amoxicillin for 5 days). The regimens used are further described in. (b) Forest plot comparing mortality of reference treatment (penicillin/gentamicin im for 7 days) with switch regimen (penicillin/gentamicin im for 2 days followed by oral amoxicillin for 5 days). The regimens used are further described in. PEN, penicillin; GEN, gentamicin.
Systematic review JAC possible, to properly evaluate currently used dosing regimens.
Once iv-to-oral switch therapy is proven to be safe and effective in neonates, its implementation may have a strong effect on health-cost reduction and quality of life. |
Distinguishing anthropogenic and natural contributions to coproduction of national crop yields globally
Crop production is a crucial ecosystem service that requires a combination of natural and anthropogenic contributions to high and stable yields, which is a coproduction process. We analysed this coproduction based on nationally aggregated data for 15 major crops for 67 countries and the European Union with data for four time steps (2000, 2006, 2010, 2014). We found strong increases in fertilizer use, net capital stock and manure use intensity for lower-middle-income countries and stagnation or decrease of these for high-income countries. We used a multiple linear regression model predicting yield to distinguish the effect of anthropogenic contributions (crop-specific fertilizer use intensity, net capital stock intensity, manure use intensity) and natural contributions (crop-specific agricultural suitability, including soil characteristics, topography and climate). We found that in particular fertilizer use intensity, manure use intensity and agricultural suitability explained variation in yields to a considerable degree (R 2 = 0.62).
Yields of food, feed and energy crops have rapidly increased over the past decades in many areas [bib_ref] Crop intensification, land use, and on-farm energy-use efficiency during the worldwide spread..., Pellegrini [/bib_ref] [bib_ref] Recent patterns of crop yield growth and stagnation, Ray [/bib_ref]. Crop production is a crucial ecosystem service that is provided as a result of a combination of natural and anthropogenic contributions that establish a coproduction process [bib_ref] Chapter six-disentangling the pathways and effects of ecosystem service co-production, Palomo [/bib_ref] [bib_ref] Co-producing ecosystem services for adapting to climate change, Lavorel [/bib_ref]. Crop yield is, despite being regularly used as an ecosystem service indicator 5,6 not a well-suited measure for natural contributions to human well-being, as crop yield is only partly related to what agricultural ecosystems contribute and as it strongly depends on human management [bib_ref] Stocks and flows of natural and human-derived capital in ecosystem services, Jones [/bib_ref] [bib_ref] Ecosystem services must tackle anthropized ecosystems and ecological engineering, Barot [/bib_ref]. For instance, in a case study on accounting for natural contributions for six arable crops, Remme et al. [bib_ref] Monetary accounting of ecosystem services: a test case for Limburg province, the..., Remme [/bib_ref] found that the resource rent, as a proxy for natural contributions, only accounts for 12% of the total revenue of six arable crops, while operating, labour and capital costs account for the rest. On the one hand, crop production depends on a range of natural contributions such as soil formation and nutrient cycling leading to soil fertility [bib_ref] Soil organic carbon, soil formation and soil fertility, Gaiser [/bib_ref]. Other important ecosystem services that contribute to crop production include animal pollination and biological pest control [bib_ref] A global synthesis reveals biodiversity-mediated benefits for crop production, Dainese [/bib_ref] [bib_ref] Ecological intensification: harnessing ecosystem services for food security, Bommarco [/bib_ref]. Abiotic natural aspects that influence agricultural suitability include topography and climate [bib_ref] Global agricultural land resources-a high resolution suitability evaluation and its perspectives until..., Zabel [/bib_ref].
On the other hand, the importance of anthropogenic contributions for producing crops has been highlighted frequently [bib_ref] Crop intensification, land use, and on-farm energy-use efficiency during the worldwide spread..., Pellegrini [/bib_ref]. One key aspect of anthropogenic contributions is an increasing intensity of land use through use of fertilizers, pesticides, fuel and machinery [bib_ref] Crop intensification, land use, and on-farm energy-use efficiency during the worldwide spread..., Pellegrini [/bib_ref] [bib_ref] Energy intensity of agriculture and food systems, Pelletier [/bib_ref]. The framework of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES [bib_ref] The IPBES conceptual framework-connecting nature and people, Díaz [/bib_ref] prominently features coproduction as the interplay of anthropogenic and natural contributions in creating ecosystem services. Little is known, however, about how to distinguish anthropogenic and natural contributions to yield of food, feed and energy crops, how relatively important these are and how they interact [bib_ref] Chapter six-disentangling the pathways and effects of ecosystem service co-production, Palomo [/bib_ref] [bib_ref] Ecological intensification: harnessing ecosystem services for food security, Bommarco [/bib_ref] [bib_ref] Research frontiers in ecosystem service science, Bennett [/bib_ref].
Land use intensity trajectories, i.e. the development pathways of land use intensity over time, might characterize the change in the way agricultural products are coproduced. Such trajectories have been shown to differ across countries and regions with different development stages [bib_ref] Crop intensification, land use, and on-farm energy-use efficiency during the worldwide spread..., Pellegrini [/bib_ref] [bib_ref] Energy and the food system, Woods [/bib_ref] and generally indicate agricultural transitions from smallholder, subsistence farming to intensive agriculture [bib_ref] Global consequences of land use, Foley [/bib_ref]. According to trajectory theories, yields are generally expected to increase over time (while increases slow down [bib_ref] Synchronized peak-rate years of global resources use, Seppelt [/bib_ref] , and conventional intensification is expected to lead to a lower energy return on investment, i.e. the amount of energy gained from yield (caloric value of crops) as a proportion of energy invested in anthropogenic input (machinery, fuels, fertilizer) [bib_ref] Middle-range theories of land system change, Meyfroidt [/bib_ref]. These overall changes in agricultural systems question to what extent natural contributions can be replaced by anthropogenic contributions when producing crops [bib_ref] Are ecosystem services replaceable by technology?, Fitter [/bib_ref] www.nature.com/scientificreports/ substitute (replacing the ecosystem service nutrient cycling through chemical fertilizers) or enhance (making nutrient cycling and primary productivity more available through technology) ecosystem services in agricultural production. Substitutability of natural contributions (or capital) is a key question in sustainability research [bib_ref] Is natural capital really substitutable?, Cohen [/bib_ref] [bib_ref] A framework for the practical application of the concepts of critical natural..., Ekins [/bib_ref]. Substitutability of natural through man-made capital would be in line with the weak sustainability position (which could imply some types of capitals could decrease) while no substitutability would be in line with strong sustainability assumption (no substitution implies the need to conserve all types of capital). The aims of this study are twofold. First, we identified trajectories of anthropogenic contributions to coproduction of agricultural crops. For this purpose we quantified yield and three anthropogenic contributions of 15 crops over the period 2000-2014 on a national scale: crop-specific fertilizer use, manure applied to soils, and net capital stock in agriculture for three country income groups. Second, we modelled national yield globally with a set of explanatory variables to disentangle the effect of different anthropogenic contributions (fertilizer use intensity, net capital stock intensity, manure use intensity) and natural contributions (agricultural suitability, including soil characteristics, topography and climate 14 ) over the same time period. We hypothesised both natural contributions and anthropogenic contributions to have positive effects on yields, but also expected interaction effects indicating substitution between natural and anthropogenic contributions, i.e. the higher the application of fertilizer or manure the lower the effect of agricultural suitability on yield or, the higher agricultural suitability the lower the effect of the application of fertilizer/manure. We hypothesised that the interaction between net capital stock in agriculture and agricultural suitability could show either an enhancement effect, i.e. the higher the amount of investment in agriculture, the higher the effect of natural contributions to yield due to a more effective use of natural resources, or a substitution effect as net capital stock might positively correlate with the other anthropogenic contributions and be indirect drivers of yield.
# Results
Trajectories of anthropogenic contributions. We identified trajectories for three anthropogenic contributions vs. yield of the 15 crops aggregated over the four time steps between 2000 and 2014 for three different income groups [fig_ref] Figure 1: Trajectories of three anthropogenic coproduction factors for crop production in three Worldbank... [/fig_ref]. Yield increased most in lower-middle-income countries (aggregated over all countries: 36.2% in 2014 compared to 2000, mean across countries: 58.0%, range across countries: − 0.03-252.7%), followed by upper-middle (aggregated: 29.4%, mean: 30.4%, range: 20.8-44.4%) and high-income countries (aggregated: 23.6%, mean: 25.5%, range: 7.3-44.5%).
Fertilizer use intensity increased most in lower-middle-income countries over the considered time period (aggregated: 74.9%, mean: 57.3%, range: 6.6-159.8%), followed by upper-middle-income countries Net capital stock in agriculture increased most in lower-middle-income countries (aggregated: 162.5%, mean: 120.2%, range: 11.8-204.3%), followed by upper-middle-income countries (aggregated: 146.2%, mean: 112.0%, range: 5.0-384.6%), and high-income countries (aggregated: 32.4%, mean: 56.5%, range: 4.2-151.2%).
Agricultural suitability and fertilizer use intensity for the year 2014 are plotted against yield in [fig_ref] Figure 2: Bivariate choropleth maps of selected coproduction factors plotted against yield for the... [/fig_ref]. These maps highlight how relative levels of each factor align with levels of yield per country or region. For instance, the European Union as a high yield region has medium levels of agricultural suitability and fertilizer use intensity [fig_ref] Figure 2: Bivariate choropleth maps of selected coproduction factors plotted against yield for the... [/fig_ref]. On the other hand, China, as a high yield country, has high levels of agricultural suitability and fertilizer use intensity. India and the Philippines are the two countries that have relatively high agricultural suitability but relatively low yields. Pakistan, for instance, has high fertilizer use intensity, but relatively low yields.
## Model of yield as a function of anthropogenic and natural contributions. standardized coeffi-
cients of the multiple regression model are reported in [fig_ref] Table 1: Model results [/fig_ref]. The explanatory power of the model was R 2 = 0.62 (adjusted R 2 , n = 64 observations). As expected, agricultural suitability is positively associated with the yields of the 15 crops (p < 0.01). Our results also confirm that fertilizer use intensity is positively associated with yield (p < 0.001). Among all logged and standardised variables, this variable had the strongest positive effect on yield. The association of manure use intensity also had a positive association with yield (p < 0.01). Together, coefficients of these variables show the importance of anthropogenic contributions to the coproduction of crop yield. We found no significant main effects of net capital stock intensity on yield.
We found no interaction effects (at significance level p < 0.05) of agricultural suitability and fertilizer use intensity, contrary to our expectation that an increase in one of these variables could decrease the effect of the other on yield. However, the interaction effect between agricultural suitability and fertilizer use intensity, albeit above common significance levels, was present and negative, which could point to the fact that an increase in one coproduction factor leads to a decrease of the other in its effect on yield.
# Discussion
We found evidence for increasing efficiency in the use of fertilizer and manure as coproduction factors. While yield increased over the time period , the use of nitrogen fertilizer and manure stagnated. This was most pronounced for high-income countries, many of which are known to have increased their fertilizer use efficiency [bib_ref] 50 year trends in nitrogen use efficiency of world cropping systems: the..., Lassaletta [/bib_ref] [bib_ref] Drivers of changes in agricultural intensity in Europe, Levers [/bib_ref] and their ratio of outputs to inputs in general (total factor productivity [bib_ref] Leveraging total factor productivity growth for sustainable and resilient farming, Coomes [/bib_ref] [bib_ref] RD spillovers, and productivity growth in world agriculture, Fuglie [/bib_ref]. Continued energy input to agriculture (largely driven by nitrogen fertilizer, machinery and fuels) has been described for Asia, Latin America, while energy input decreased in Europe and stagnated North America 1 . This is in line with the development we observe for net capital stock in agriculture, with stagnation in high-income countries and a continued increase in lower-and upper-middle countries.
Our findings for coproduction of 15 crops on a national level confirm the positive association of nitrogen fertilizer with yield, which is in line with earlier findings, for instance within the European Union 26 and globally [bib_ref] Closing yield gaps through nutrient and water management, Mueller [/bib_ref]. We found a positive association of agricultural suitability with yield across all countries. This confirms the importance of natural contributions (here: soil fertility, topography and climate) in the coproduction of crops. Soil fertility and quality, as a considerable part of suitability, has been found to have a positive correlation with yield in the European Union 26 . German et al. [bib_ref] Relationships among multiple aspects of agriculture's environmental impact and productivity: a meta-analysis..., German [/bib_ref] have also found strong correlations between different aspects of soil quality and yield across different agricultural systems in their systematic review. However, another systematic review on pairwise associations of ecosystem services by Lee and Lautenbach 31 found no clear direction of association (i.e. both positive and negative associations) between soil formation and composition and food production, and even a negative association between soil formation and composition and biomass production (including fodder). Note, however, that the composite indicator for suitability which we used here also contains climatic factors (temperature, precipitation), which have been found to have differing effects on yields, depending on crop type and world region [bib_ref] Closing yield gaps through nutrient and water management, Mueller [/bib_ref] [bib_ref] Climate trends and global crop production since 1980, Lobell [/bib_ref]. In addition, it contains irrigated areas.
The literature on coproduction has, so far, largely remained conceptual and is hardly operationalised (e.g. [bib_ref] Stocks and flows of natural and human-derived capital in ecosystem services, Jones [/bib_ref] [bib_ref] Ecosystem services must tackle anthropized ecosystems and ecological engineering, Barot [/bib_ref] , which might be due to the lack of data allowing to distinguish these processes. Data availability on agricultural resources is relatively high and hence an ideal showcase for an understanding coproduction. Here we have operationalized the analysis of coproduction of crops, and included anthropogenic contributions at different stages in the provision of ecosystem services 4 , namely the management of agricultural landscapes (through the independent variables net capital stock, manure and fertilizer use intensity) and the mobilization or harvest of crops (net capital stock intensity). Our study of coproduction of agricultural crops is related to studying (the effects of) input intensity (i.e., amount of different forms of anthropogenic capital per area [bib_ref] A conceptual framework for analysing and measuring land-use intensity, Erb [/bib_ref]. Our analysis, in addition, includes natural contributions as inputs and also studies the interactions between different contributions. We found that modelling the final ecosystem service (crop production) as the dependent variable of different types of contributions as independent variables is a well-suited way to study coproduction that could be applied to other services.
While we did not find evidence that the effect of natural contributions (soil fertility, topography and climate) on yield decreases when more fertilizer is applied at common significance levels, we found a negative interaction effect between fertilizer use intensity and agricultural suitability. Future studies could test whether this interaction of natural and anthropogenic contributions in producing crops occurs at smaller scales. Moreover, research could look in more detail into studying the effect of agricultural intensification also need to consider aspects of human wellbeing (such as food security) as an outcome [bib_ref] Putting meaning back into "sustainable intensification, Loos [/bib_ref]. While we did not find evidence for substitutability of agricultural suitability we found positive effects of agricultural suitability as an independent variable. In line with ecological or sustainable intensification measures, soil fertility to enhance yield could, for instance, be promoted through conservation tillage, crop rotation and planting of cover crops [bib_ref] Ecological intensification: bridging the gap between science and practice, Kleijn [/bib_ref].
Despite the use of aggregated data, we were able to identify the effect of natural and anthropogenic contributions on the coproduction of crops at the national level with global coverage. We argue that aggregation here is a "requisite simplicity" 36 that allows studying the complex problem of coproduction of crops across systems. In our analysis, we used aggregated country data, which neglects potential within-country spatial variation. This was necessary as some data is only collected at the national level in an aggregated form (production, fertilizer use intensity), while other data would allow for a more fine-grained, spatially explicit analysis (agricultural suitability). With more data available in the future, one could study spatial-temporal correlations between observations of a country over time. The lack of spatial information on many land-use intensity factors in particular at the global level is a well-known challenge [bib_ref] Challenges and opportunities in mapping land use intensity globally, Kuemmerle [/bib_ref]. The same lack of data holds for other potentially relevant natural contributions (e.g. animal pollination, biological pest control). While the models explained a considerable proportion of the variation of yield, we lack information on other relevant anthropogenic contributions like the application of pesticides, or fuels for machinery, or agricultural knowledge 3 . Furthermore, we merged all crops, which limits the ability to detect potential variations between crops. The dataset for agricultural suitability we used, however, accounts for long-term crop-specific growing-conditions [bib_ref] Global agricultural land resources-a high resolution suitability evaluation and its perspectives until..., Zabel [/bib_ref]. We also used anthropogenic contributions that characterize agriculture in general at the national level, which might actually differ for the 15 crops as compared to all crops. Agricultural suitability contains irrigated areas and hence also anthropogenic contributions.
Our study paves the way for addressing further research questions. For instance, the effect of other important contributions to agricultural production, like pollination 38 or biological pest control 39 could be studied at large scales. These interactions would be of particular interest as agriculture both harms biodiversity contributing to these services and benefits from it (biodiversity-production mutualism [bib_ref] Deciphering the biodiversity-production mutualism in the global food security debate, Seppelt [/bib_ref]. Moreover, different levels of anthropogenic and natural coproduction factors that potentially replace each other in producing comparable amounts of crops also need to be studied in terms of associated benefits and costs including the environmental effects of applying fertilizer 41 or the negative effect of intensification on biodiversity [bib_ref] Conventional land-use intensification reduces species richness and increases production: a global metaanalysis, Beckmann [/bib_ref]. This also includes the socioeconomic effects of enhancing different levels of natural and anthropogenic contributions across farming systems on food security and profitability [bib_ref] Farming approaches for greater biodiversity, livelihoods, and food security, Garibaldi [/bib_ref]
# Conclusion
Yields of food, feed and energy crops depend on both natural and anthropogenic contributions. We have analysed coproduction of the yields of 15 major food, feed and energy crops globally on a national level. We found that crop-specific fertilizer use, manure use intensity and agricultural suitability for these crops explained variation in yields to a considerable degree. Our results contribute to a better understanding of how natural and anthropogenic contributions to agricultural yields interact. We found no significant interaction effect between agricultural suitability (including soil fertility, topography and climate) and fertilizer use intensity. The significant association of agricultural suitability with yield over all countries implies the importance of agricultural suitability, of which soil fertility could be enhanced through measures of ecological or sustainable intensification.
# Methods
We collected data for indicators for coproduction and yield of 15 crops for 67 countries and the European Union (including the United Kingdom) across the globe for four time steps . We excluded Croatia because of the lack of data for net capital stock intensity, and Belarus for the year 2000 because of inconsistently low reported values of fertilizer use. Choice of crops, countries and time steps was restricted by data availability of different indicators (see below). Data for all variables and time steps was available for 20 countries and the European Union, i.e. 47 countries in total (see Supplementary Information, [fig_ref] Table 1: Model results [/fig_ref]. Agricultural suitability: We used an aggregated suitability index, available for 15 major crops: barley, cassava, groundnut, maize, millet, oil palm, potato, rapeseed/canola, rice, rye, sorghum, soybean, sugarcane, sunflower, wheat (including summer and winter wheat) [bib_ref] Global agricultural land resources-a high resolution suitability evaluation and its perspectives until..., Zabel [/bib_ref]. The index comprises crop-specific conditions for soil structure and fertility, climate and topography based on a fuzzy logic approach (for further details see [bib_ref] Global agricultural land resources-a high resolution suitability evaluation and its perspectives until..., Zabel [/bib_ref]. It serves as an indicator for natural contributions to crop production, including the ecosystem services that contribute to soil fertility. Note that it also contains irrigated areas [bib_ref] Global agricultural land resources-a high resolution suitability evaluation and its perspectives until..., Zabel [/bib_ref] , and hence indirectly includes anthropogenic contributions. The index contains average values for the period 1981-2010. In order to account for the spatial distribution of the 15 crops we used data on harvested area by Monfreda et al. [bib_ref] Farming the planet: 2. Geographic distribution of crop areas, yields, physiological types,..., Monfreda [/bib_ref] to create a harvested area-weighed agricultural suitability index. The agricultural suitability dataset was resampled to the grid cell size of the Monfreda et al. [bib_ref] Farming the planet: 2. Geographic distribution of crop areas, yields, physiological types,..., Monfreda [/bib_ref] dataset (5 min by 5 min) using the bilinear resampling method in ArcMap 10.7. For each grid cell we created a weight by dividing the total area harvested of the 15 crops by the respective sum across the country and then multiplied this weight with the agricultural suitability data. The resulting grid cell values were then summed by country.
Fertilizer use intensity: We collected data on nitrogen fertilizer use by crops from the International Fertilizer Association IFA. Crop-specific fertilizer application data was available for all 15 crops and for four time steps around 2000, 2006/07, 2010/11, 2014/15 (years slightly varying due to data collection and reporting method). This dataset determined the choice for the four time steps. For in total 67 countries and the European Union data was available for at least one time step (see [fig_ref] Figure 3: Global map of the harvested area weighted agricultural suitability index on cropland... [/fig_ref] and [fig_ref] Figure 1: Trajectories of three anthropogenic coproduction factors for crop production in three Worldbank... [/fig_ref]. According to the dataset these countries applied a large proportion of the global artificial nitrogen fertilizer, namely 92.4% www.nature.com/scientificreports/ Net capital stock intensity: We took annual data of net capital stock (i.e. fixed capital, accounting for depreciation, used in agriculture, including machinery, buildings and equipment) from FAOSTAT for the selected countries and years . This data was used as a composite measure for anthropogenic capital used as an input to produce crops. As this data is not crop-specific we made the assumption that the intensity of capital stock in agriculture per area harvested as an average for all crops also holds for the 15 selected crops. That is, we divided net capital stock in agriculture by the total area harvested for all crops to get an average intensity measure. Values were transformed to 2014 US$ values using inflation factors.
Manure use intensity: We took annual data on manure applied to soils from FAOSTAT for the selected countries and years . As this data is not crop-specific we made the assumption that manure applied to soils is used proportionally to area harvested. That is, we multiplied the value for each country by the fraction of area harvested of the selected 15 crops of the total area harvested per year and then divided this value by area harvested for the 15 crops to get an intensity value.
Yield: Yield was calculated as production per area harvested. Data on annual crop production and the respective area harvested was taken from FAOSTAT for the selected countries, crops and years . All crop production data was transformed to caloric values from tonnes to kJ based on FAO. We then summed the values of the 15 crops for each country or region and year.
To visualise trajectories of three anthropogenic coproduction factors against yield we summed the values for production, fertilizer use, manure use and net capital stock in agriculture, each divided by the sum of total area harvested for countries belonging to the same income class according to the Worldbank ("lower-middle", "upper-middle" and "high" income)and for which data was available for each of the four time steps. This was the case for seven of 24 lower-middle income countries, seven of 19 upper-middle income countries, and seven of 17 high income countries (including the EU as one region) considered in this analysis (see [fig_ref] Table 1: Model results [/fig_ref]. We reported changes relative to the values for the year 2000 (100%). In addition we calculated the mean values and the minimum and maximum values of all countries belonging to the same income class.
We used the statistical software package R 3.6.1 55 run via RStudio 1.2.1335 56 for data analysis. We used linear regression model predicts yield as a function of anthropogenic and natural coproduction factors. We also included interaction effects with the independent variable agricultural suitability to test for enhancement or substitution effects between natural and anthropogenic coproduction factors. Hypothesised main and interaction effects are found in [fig_ref] Table 2: Explanatory variables and hypotheses [/fig_ref]. For the linear regression model all variables were available for each country for one, two, three or four years. We took the mean of the variables fertilizer use intensity, net capital stock intensity, manure use intensity and years over the available years per country to reduce the risk of pseudoreplication of observations of one country. The linear model was: Yield = f(agricultural suitability + fertilizer use intensity + net capital stock intensity + manure use intensity + agricultural suitability:fertilizer use intensity + agricultural suitability:net capital stock intensity + agricultural suitability:manure use intensity).
We checked outliers with the help of the function olsrr: ols_plot_cooksd_bar in R and identified four countries that we then removed from the dataset due to high Cook's distances 57 : Azerbaijan, Egypt, Kuwait, and Malaysia. We tested the model for collinearity. None of the explanatory variables were highly correlated (Pearson's r between fertilizer use intensity and net capital stock intensity was p = 0.705, and both were kept in the model).
To equalize spread and reduce leverage, we log-transformed all variables. We scaled variables (mean-centred and divided by their standard deviations) to make regression coefficients comparable. We used the number of observations per country as weights in the linear model. Code for data preparation and analysis as well as datasets used are available (see data availability statement).
## Variable
Hypothesised effect on yield
## Main effects
Agricultural suitability (index on climate, soil, topography, i.e. natural coproduction factors) + A higher agricultural suitability leads to higher yields Fertilizer use intensity (anthropogenic coproduction factor) + Higher fertilizer use intensity leads to higher yields Manure use intensity (anthropogenic coproduction factor) + A higher manure use intensity leads to higher yields Net capital stock intensity (anthropogenic coproduction factor) in 2014 US$ + A higher proportion of net capital stock/area leads to higher yields due to more industrialised agriculture
## Interaction effects
Agricultural suitability:fertilizer use intensity − A substitution effect, i.e. the higher fertilizer use intensity the lower the effect of suitability on yield Agricultural suitability:manure use intensity − A substitution effect, i.e. the higher the amount of manure applied to soils the lower the effect of suitability on yield Agricultural suitability:net capital stock intensity + / − An enhancement effect, i.e. the higher the amount of net capital stock, the more effective a country can make use of agricultural suitability for achieving higher yields (e.g., using machinery for less suitable soils); or a substitution effect
[fig] Figure 1: Trajectories of three anthropogenic coproduction factors for crop production in three Worldbank income classes ("lower-middle", "upper-middle", "high" income): fertilizer use intensity, manure use intensity and net capital stock intensity. Values are relative to the values in the year 2000 (100%). First arrow: 2000 to 2006, second arrow: 2006 to 2010, third arrow: 2010 to 2014. Yield increases over the years across income groups come along with stagnating or decreasing anthropogenic contributions in high income countries, and strong increase in net capital stock intensity in lower-middle and upper-middle income countries. Figure was [/fig]
[fig] Figure 2: Bivariate choropleth maps of selected coproduction factors plotted against yield for the year 2014. Values were split into three tertiles; low, medium, high. (a) Agricultural suitability, (b) Fertilizer use intensity. Countries in white have no data for this year. Figure was created in R (version 3.6.1, https:// www.r-proje ct. org/) based on 58 . [/fig]
[fig] Figure 3: Global map of the harvested area weighted agricultural suitability index on cropland of the 15 selected crops (based on 14,44 ). Countries included in the analysis are marked white, countries excluded (no reported crop-specific nitrogen use) are dashed. Figure was created in ArcMap (version 10.7, https:// deskt op. arcgis. com/ en/ arcmap/) based on 14,44,58 . Scientific Reports | (2021) 11:10821 | https://doi.org/10.1038/s41598-021-90340-1 [/fig]
[table] Table 1: Model results: coefficients of the independent variables. Fertilizer use intensity and agricultural suitability are positively associated with yield. a Significance levels: p < 0.1; *p < 0.05; **p < 0.01, ***p < 0.001. [/table]
[table] Table 2: Explanatory variables and hypotheses. [/table]
|
A Population-Based 16-Year Study on the Risk Factors of Surgical Site Infection in Patients after Bone Grafting
Bone grafting is a commonly used orthopedic surgical procedure that will provide bone formation in bone defects or regions of defective bone healing. A major complication following bone grafting is a postoperative recipient graft site infection that is associated with substantial mortality and increased use of medical resources. The purpose of the study was to identify the risk factors associated with infection after bone-grafting surgery.Data from 1,303,347 patients listed in the Taiwan National Health Insurance Research Database (NHIRD) and admitted to hospitals from 1997 through 2012 who underwent primary bone grafting (mean age: 46.57 years old; mean length of hospital stay: 8.04 days) were analyzed. The incidence of infection by age, hospital stay, gender, income, chronic disease (tuberculosis [TB]; diabetes mellitus [DM]; acquired immunodeficiency syndrome [AIDS]), fracture complications (nonunion; delayed union fracture), types of graft and hospital was evaluated.Three percent of the patients developed a postoperative recipient graft site infection. Multivariable analysis revealed that patients were more likely to develop a post bone-grafting surgery infection if they were older, had a longer hospital stay, were male, had a lower income, or had comorbid TB, DM, or AIDS. Patients were more likely to develop an infection if they had a nonunion, an alloplast graft, or treated in a local clinic.Our findings should provide a clinically relevant reference for surgeons who perform bone grafting. Patients should be informed of the potential risks.(Medicine 94(47):e2034)Abbreviations: AIDS = acquired immunodeficiency syndrome, AOR = adjusted odds ratio, CI = confidence interval, COR = crude odds ratio, DM = diabetes mellitus, ICD-9-CM = International à n, number of cases admitted for bone graft surgery. y Mean AE standard deviation. z The number in parentheses indicated the percentage of the number of either infected or noninfected one over the total number of bone-grafting case in each category. § Two graft types used: autograft and alloplast.
# Introduction
B one grafting is common in neurosurgery for spinal fusion and in orthopedic reconstruction surgery for revision hip replacement, repairing skeletal defects after a trauma, and for support when removing a tumor. Three types of bone graft are used: autograft (autologous graft), allograft, and alloplastic graft. An autograft is usually harvested from the patient's own iliac crest. An allograft is taken from a donor or cadaver. [bib_ref] Biology of bone grafts, Stevenson [/bib_ref] [bib_ref] Biology of allografting, Garbuz [/bib_ref] An alloplastic graft is a synthetic substitute, usually created from ceramics (calcium phosphate, calcium sulfate, or bioactive glass), polymer (derivative from acrylate), or growth factors (the family of bone morphogenetic proteins). [bib_ref] Bone graft and bone graft substitutes: a review of current technology and..., Damien [/bib_ref] Intraoperative complications might occur and affect subsequent tissue healing, motor function recovery, and the surgical outcome. Thus, bone grafting poses a significant health risk to the patient.
Approximately 500,000 surgical site infections caused by bacteria and viruses occur every year in the United States. [bib_ref] Health and economic impact of surgical site infections diagnosed after hospital discharge, Perencevich [/bib_ref] Postoperative transmission of an infection can be devastating to the graft and the patient: it may lead to serious illness and even death.Patients with a preexisting illness are at a higher risk for an infection than are those who are otherwise healthy. Delayed union and nonunion fracture are two of the most common complications after orthopedic surgery, and both require additional reconstruction using bone grafts. [bib_ref] Bone grafts and bone substitutes for opening-wedge osteotomies of the knee: a..., Lash [/bib_ref] [bib_ref] Intercalary allograft reconstructions following resection of primary bone tumors: a nationwide multicenter..., Bus [/bib_ref] Moreover, after bone surgery, patients with an immunocompromised disease, like diabetes mellitus (DM), have a greater risk of infection. [bib_ref] Postoperative infection rates in foot and ankle surgery: a comparison of patients..., Wukich [/bib_ref] [bib_ref] Factors associated with infection following anterior cruciate ligament reconstruction, Brophy [/bib_ref] Two other increasing chronic infectious diseases in Asia, tuberculosis (TB) and acquired immunodeficiency syndrome (AIDS), may compromise the immunity of patients and relate to the incidence of bone infection. Treating deep infection requires long-term antibiotic use and repetitive irrigation and debridement. Alternatively, a skin graft or flap might be required to improve the outcome after infection. For most patients, removing the graft might be the only way to fix the problem, because the fresh tissue may be the source or a nidus for infection recurrence. Nevertheless, there is no prior study that examines the risk factors of infection after bone-grafting surgery.
We examined the risk factors of infection in patients that underwent bone grafting in Taiwan. Data were collected from the Taiwan National Health Insurance Research Database (NHIRD) (http://nhird.nhri.org.tw/en/Background.html) and analyzed for all patients who were admitted to hospitals from 1997 through 2012 for primary bone-grafting treatment. The incidence of recipient graft site infection in patients who underwent bone grafting, and the risk factors associated with postoperative infection, were evaluated. This is the first large national database study of the factors associated with bonegrafting surgery.
# Methods
## Data source
Approximately 98% of Taiwan's population (ca. 23 million) participate in the National Health Insurance program (http://www.nhi.gov.tw/English/ Index.aspx?menu¼13&menu_id¼486&WD_ID¼486).
The Taiwan NHIRD, which uses the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis and procedure codes, 10 covers all inpatient medical claims data for enrolled residents. The established categories of expenditure were provided in inpatient expenditure by admission (DD files) (http://nhird.nhri.org.tw/date_ 01.html). The personal information of all patients was encrypted with a double scrambling protocol for research purposes to protect patient privacy. This study was approved by the Institutional Review Board of Jianan Psychiatric Center, and the protocol was evaluated by the National Health Research Institutes (NHRI), which consented to this planned analysis of the NHIRD (Agreement Number: NHIRD-103-161).
## Study design
A population-based retrospective cohort study was conducted. We analyzed 1,303,347 NHIRD-registered cases of bone grafting between 1997 and 2012. The participant demographic characteristics, including age, gender, hospital stay, and lowincome households, were identified at the index date. The risk factors included in this study were those that have been found to be associated with or possibly related to the development of bone infection. A diagnosis of bone infection was defined on the basis of deep infection after surgery (excluding superficial infection) as prescribed by the National Health Insurance system (ICD 9 codes 996.4, 996.66, 996.67, 996.69, and 730.10-730.19; [fig_ref] TABLE 1: Disease Definition and Diagnostic Codes [/fig_ref]. Data regarding deep infection-related surgeries (ICD 9 codes 77.1-77.9, 79.3, and 86.22) were collected. The preexisting comorbidities include DM, TB, and AIDS. They were identified from data of the NHIRD based mainly on ICD-9-CM diagnostic codes [fig_ref] TABLE 1: Disease Definition and Diagnostic Codes [/fig_ref]. The diagnosis in this study were judged and determined by related specialists and physicians according to the standard criteria. The minimum living expense standard was defined as 60% of the average monthly disposable income for each region. Low-income households were defined as those with a monthly average per-member gross income of less than the monthly minimum living expense standard in the household's residence region. The ICD-9-CM procedure code used for alloplasts and allografts were the same (79.3); thus, grafts were grouped into three categories: alloplast/allograft, autograft, and combined (autograft plus alloplast/allograft). Union was defined based on combined clinical and radiographic criteria. The criterion for a nonunion fracture was an unhealed broken bone 9-month postfracture that had shown no healing progression between the affected ends of the bone for 3 months after any specific treatment. [bib_ref] Enhancement of fracture-healing, Einhorn [/bib_ref] A fracture that failed to heal within the normal time for that fracture type and location was defined as a delayed union fracture. According to the chapter II, laws of medical care act, Ministry of Health and Welfare in Taiwan (http://law.moj.gov.tw/Eng/LawClass/LawContent.aspx?PCODE ¼L0020021), medical care institutions with wards for the inpatient care of patients shall be referred to as hospitals; those with only out-patient facilities shall be referred to as clinics. Clinics referred may set up no more than 9 beds for observation. In Taiwan, local clinics are set up with different specificities. Therefore, the data regarding to patients with bone infection can be collected from those local clinics specialized in orthopedics, general trauma, rehabilitation, infectious disease, etc.
Statistical Analysis SPSS 18.0 for Windows (SPSS Inc., Chicago, Illinois) was used for all statistical analyses. Descriptive statistics were analyzed for all variables using frequency, percentage, mean, and standard deviation (SD). Bivariate analyses were used to determine appropriate variables to include in the multivariate models. Only variables that were significantly associated with bone infection (P < 0.05) were entered simultaneously into multivariate logistic regression models. The effects of outcome variables were measured using the adjusted odds ratio (AOR) with a 95% confidence interval (CI). The statistical analysis has been examined by the epidemiologist, Dr. C.Y. Li, who is also one of the authors of this article.
# Results
We enrolled 1,303,347 cases (mean age: 46.57 AE 22.16 years old; mean length of hospital stay: 8.04 AE 7.73 days; male: 58.4%; low income: 1.3%) [fig_ref] TABLE 2: Characteristics of Patients Who Underwent a Bone Graft [/fig_ref]. The infection rate after bone graft surgery was 3.05%. The infection rate of male and [fig_ref] TABLE 2: Characteristics of Patients Who Underwent a Bone Graft [/fig_ref]. All variables but ''Previous Bone Graft Surgery'' were significantly correlated with surgical site infection after bone grafting [fig_ref] TABLE 3: Bivariate Associations Between Variables and Bone Infection [/fig_ref]. Univariate and multivariate analyses both showed that the likelihood of postoperative infection was significantly positively dependent upon the patient-related factors of age, gender, income, and comorbid diseases [fig_ref] TABLE 4: Factors Associated with Infection After a Bone Graft [/fig_ref]. Patients more easily developed surgical site infections if they were older, were male, had a lower income, and had comorbid diseases. Patients with comorbid AIDS had the highest odds of infection (AOR ¼ 6.945, 95% CI ¼ 2.405-20.054, P < 0.001). Univariate and multivariate analyses also showed that the likelihood of postoperative infection was significantly dependent upon the surgery-related factors of hospital stay length and fracture, graft, and hospital type. Patients with a history of nonunion have relatively high odds of developing a surgical site infection (AOR ¼ 7.005, 95% CI ¼ 6.774-7.245, P < 0.001); our data also indicated that patients who underwent an autograft were the least likely to develop a postoperative infection. The AOR of patients who had their bone grafts done in medical centers was 0.928 compared with those who had them done in local clinics (95% CI ¼ 0.904-0.953, P < 0.001). The AOR of patients who had their bone grafts done in regional hospitals was 0.764 compared with those who had them done in local clinics (95% CI ¼ 0.743-0.785, P < 0.001). The AOR of patients who had their bone grafts done in district hospitals was 0.544 compared with those who had them done in local clinics (95% CI ¼ 0.336-0.881, P ¼ 0.013). The relative risk factor of surgical site infection was the highest for bone grafts done in local clinics, followed by those done in medical centers, regional hospitals, and district hospitals. To our surprise, the odds of infection in the medical centers were the second highest.
# Discussion
In this study, we evaluated various risk factors of surgical site infection in patients who undergo bone grafting. Among the 1,303,347 cases that had undergone a bone-grafting operation, 39,816 cases developed a postoperative infection. The infection rate was 3.05%, lower than the 5.4% allograft infection rate in 128 patients reported by Delloye et al. [bib_ref] Local complications of massive bone allografts: an appraisal of their prevalence in..., Delloye [/bib_ref] Our infection rate was also lower than the 7.8% rate in 140 patients after cranioplasty. [bib_ref] Analysis of the factors influencing bone graft infection after cranioplasty, Lee [/bib_ref] Our infection rate might be lower because our patients were recruited from the general population and because we did not restrict enrollment based on the surgery site or the graft materials used.
Surprisingly, except for local clinics, the smallest units where the AOR was highest, the second highest one was in the medical centers, the largest units, followed by the regional hospitals and district hospitals. The commonsense and intuitive order should have been local, district, and regional hospitals, and then medical centers, that is, the smallest rate of infection should have been at the largest unit. This paradoxical finding might be explained by a national study 14 of nosocomial infection, which found that the infection rate was highest in medical centers (0.42%) followed by regional hospitals (0.34%). It may also explain why staying 1 extra day in a hospitals increased by 1.054 times the odds of developing a surgical site infection. Patients given an autograft were less likely to develop an infection than those who were given a combined graft or an alloplast/allograft. Similar results were reported in a study on anterior cruciate ligament reconstruction: 9 bone-tendon-bone autografts were associated with a lower risk of postoperative infection than were allografts and combined grafts. The principles involved in successful bone grafts include osteoconduction (guiding the reparative growth of the natural bone), Adjusted for age, gender, low income, diabetes mellitus, tuberculosis, AIDS, hospital stay, nonunion, delayed union, type of graft and type of hospital; COR ¼ crude odds ratio; AOR ¼ adjusted odds ratio; CI ¼ confidence interval for AOR. Ã Crude odds ratio is the ratio that is not stratified. y Adjusted odds ratios are the odds of a dichotomous event being adjusted for or controlling for other possible contributions from other variables in the model. osteoinduction (encouraging undifferentiated cells to become active osteoblasts), and osteogenesis (living bone cells in the graft material contribute to bone remodeling). [bib_ref] Is autograft the gold standard in achieving radiographic fusion in one-level anterior..., Samartzis [/bib_ref] Osteogenesis occurs only when using autografts; therefore, using an autograft is preferable. However, an autograft might lead to complications such as pain, infection, scarring, and blood loss, as well as a variable quantity of bone to harvest from each patient. The alternative is an allograft, but that lacks osteoactive capacity and risks carrying infectious agents or leading to an immune rejection. Other approaches have used alloplasts, which are various types of organic and synthetic substitutes for bone, and they have focused on improving the efficacy of bone grafts or other scaffolds by incorporating bone progenitor cells and growth factors to stimulate cell proliferation. Based on our study, in addition to using an autograft alone, the combined use of an alloplast with an autograft might help reduce infection. We collected data without exclusion of prior infection history. Although the bone grafting was performed when no sign of infection or the infection was totally under control in surgical sites, an exclusion of these patients with prior infection would underestimate the infection rate. The debridement of wound infection would not be considered to use bone grafts. However, in rare cases, when repetitively deep infection occurred, a vascular bone graft might be applied from the donor site of patients for bone grafting. In that case, this type of bone graft belonged to autograft that was the least likely to develop a postoperative infection compared to the other two groups in our study. Nevertheless, preexisting complications can increase the risk of infection; for example, a nonunion increases it 7.005 times and a delayed union increases it 2.781 times after bone grafting. Both of these remain major complications after skeletal trauma. Careful attention is required when doing a bone graft, particularly in patients with these histories.
Our study indicated that patients 1 year older were 1.009 times more likely to develop a postoperative infection than patients 1 year younger. Moreover, patients with comorbid chronic or infectious diseases were more likely to develop an infection. DM is a recognized important risk factor for a variety of intracellular bacterial infections. Oxidative stress in the tissue of patients with DM reduces the phagocytic and antibacterial activity of neutrophils and macrophages and provides an intracellular niche for the pathogen to replicate. [bib_ref] Immunological mechanisms contributing to the double burden of diabetes and intracellular bacterial..., Hodgson [/bib_ref] Immunosuppression makes it easier for patients to become infected. It should not, therefore, be surprising that patients with comorbid AIDS are 6.945 times more likely to become infected after bonegrafting surgery. Notably, patients with comorbid TB, another infectious disease with a rapidly rising incidence worldwide, are 1.780 times more likely to become infected than those without TB. CD4 þ T cells are protective against Mycobacterium tuberculosis: CD4 þ T cell depletion is responsible for the increased susceptibility to develop active M tuberculosis in HIV-infected patients. [bib_ref] Interaction between HIV and Mycobacterium tuberculosis: HIV-1-induced CD4 T-cell depletion and the..., Geldmacher [/bib_ref] Although the role of CD8 þ T cells in TB is less clear than the role of CD4 þ T cells, they are generally considered to contribute to optimal immunity and protection. [bib_ref] Induction and regulation of CD8þ cytolytic T cells in human tuberculosis and..., Brighenti [/bib_ref] These results suggested that compromised immunity because of aging and comorbid disease are determining risk factors of bone-graft infection. Compromised immunity contributes to the double burden of chronic disease and surgical site infection. How other noninfectious comorbid diseases were involved in bone-graft infection required for further investigation. We also found that low-income earners were 1.634 times more likely compared with higher income earners to develop an infection. Consistent with this risk factor for other bone surgery, 19,20 economic stress might physically and psychologically make patients less aware of the state of their individual health.
The National Health Insurance Bureau has established a uniform system to control the quality of medical services, and the quality of the coded data in the present study was reliable. [bib_ref] Estimated prevalence of orthopaedic fractures in Taiwan: a cross-sectional study based on..., Yang [/bib_ref] [bib_ref] Epidemiological survey of orthopedic joint dislocations based on nationwide insurance data in..., Yang [/bib_ref] Compared with self-reported surveys, the methodology used in this study reduced the patients' recall bias and thereby increased its accuracy. Nevertheless, the NHIRD provides only insurancerelated data; information such as education level and marital status are not provided. Second, the limited data acquired made it difficult to explain the different infection rate between males and females. Suzuki et al. identified that several risk factors were associated with infection after total knee arthroplasty surgery, including previous history of open reduction internal fixation, remnants of prior internal fixation material, and male gender. [bib_ref] Previous fracture surgery is a major risk factor of infection after total..., Suzuki [/bib_ref] Male gender was significantly correlated with postoperative infection. This is probably why the mortality rate was higher for male patients orthopedic and trauma admissions. [bib_ref] Mortality among orthopaedic and traumatology admissions: a ten year review, Orimolade [/bib_ref] Compliance with the doctor's prescribed medication regimen may be different between men and women.
In conclusion, we surveyed the epidemiology of surgical site infection in patients who had undergone bone grafting in Taiwan between 1997 and 2012. Our findings revealed the risk factors of a postoperative bone graft infection. They consists of the patient-related factors like age, income, gender, and chronic diseases, and the surgery-related factors like the length of hospital stay, disease complications, type of graft used, and type of hospital in which the surgery was done. As the frequency of using different types of bone grafts for orthopedic reconstructive surgery increase, our findings should provide an understanding of the risk factors associated with infection and should be a valuable reference for orthopedic surgeons.
[table] TABLE 1: Disease Definition and Diagnostic Codes [/table]
[table] TABLE 2: Characteristics of Patients Who Underwent a Bone Graft [/table]
[table] TABLE 3: Bivariate Associations Between Variables and Bone Infection [/table]
[table] TABLE 4: Factors Associated with Infection After a Bone Graft [/table]
|
Development and preliminary validation of the Adolescent Psoriasis Quality of Life instrument: a disease‐specific measure of quality of life in adolescents with psoriasis
Age-appropriate and psoriasis-specific instruments may be more sensitive for HRQoL issues experienced by this unique group.
## Summary
Background No age-appropriate and disease-specific instrument currently exists to measure health-related quality of life in adolescents with psoriasis (patients aged 12-17 years). Objectives To develop and provide preliminary validation of the Adolescent Psoriasis Quality of Life instrument. Methods Qualitative interviews with adolescents with psoriasis, parents of adolescents with psoriasis, and healthcare professionals informed the development of an initial item pool for the instrument, which was subsequently refined through cognitive interviews. Finally, data from an independent sample of adolescents with psoriasis (n = 50) were used for item reduction, scale construction and initial validation, using a combination of techniques from classical test theory and Rasch modelling. Results Rich qualitative data concerning health-related quality of life in adolescents with psoriasis (from 18 adolescents, 14 parents and four healthcare professionals), combined with cognitive interview testing (n = 12), resulted in a 41-item draft version. Item reduction led to the final version, a 17-item instrument consisting of two subscales showing good fit to their respective Rasch models: psychosocial impact and the impact of physical symptoms and treatment (five items). All a priori stated hypotheses regarding construct validity were supported. Both subscales and the total scale showed acceptable test-retest reliabilities (intraclass correlations 0Á97, 0Á89 and 0Á96) and internal consistencies (Cronbach's a 0Á94, 0Á81 and 0Á95). Conclusions The preliminary form of the Adolescent Psoriasis Quality of Life instrument shows promising psychometric properties. It can be used in daily clinical practice and research to support a patient-centred approach and inform treatment planning.
What's already known about this topic?
- Health-related quality of life (HRQoL) instruments should be targeted towards narrowly defined age groups, as life contexts of children, adolescents and adults may differ substantially.
- Dermatology-specific instruments have been used to measure HRQoL in adolescents with psoriasis, but it is not known whether these instruments accurately capture all relevant HRQoL aspects in adolescent psoriasis.
What does this study add?
- The Adolescent Psoriasis Quality of Life instrument represents the first age-appropriate and disease-specific instrument for measuring HRQoL in adolescentsyears old) with psoriasis.
- It is intended for use in daily clinical practice to support dermatologists and other healthcare professionals in providing optimal care for adolescents with psoriasis.
Health-related quality of life (HRQoL) reflects patients' physical, psychological and social function and wellbeing as it relates to medical diseases and their treatments. 1 Its content may vary across the lifespan as life contexts of children, adolescents and adults differ substantially, 2,3 and evidence shows that the better targeted an HRQoL instrument is towards a certain patient group, the higher its sensitivity in detecting relevant issues.Psoriasis treatment aims at minimizing the extent of the disease and reducing its impact on HRQoL.The sensitivity of HRQoL instruments used in daily clinical practice will likely affect how patients are met in the clinic and how treatments are initiated and monitored. While psoriasis-specific instruments have been developed for adults,no such measure exists for adolescents, except for one scalp-specific instrument designed for children (6-17 years old).Existing studies on HRQoL in adolescent psoriasis have therefore been restricted to using general dermatology-related or generic measures.The most commonly used measure of HRQoL in adolescent psoriasis is the Children's Dermatology Life Quality Index (CDLQI), intended for children 4-16 years old.Previous research in skin disease has highlighted adolescent-specific HRQoL issues distinct from those of both adults and children.The validity of using the same instrument in both 4-year-old patients and 16-year-old patients can thus be questioned.
The adult parallel to the CDLQI is the Dermatology Life Quality Index (DLQI), which targets patients 16 years and older.Confusion exists over whether to use the DLQI or the CDLQI in 16year-old patients. Further complications arise, as several studies used the CDLQI in 17-year-old patients,and another study compared scores for the CDLQI and the DLQI in patients aged 16 and 17 years, revealing a difference in total score between the two instruments.The authors concluded that an HRQoL measure specifically designed for adolescents was necessary.To address this issue, the Teenagers' Quality of Life (T-QoL) instrument, which targets adolescents (patients aged 12-19 years) with skin disease, was recently developed. 14 However, very few adolescents with psoriasis contributed information during the development of this instrument; only two of 32 patients during the concept elicitation phase, and eight of 203 during the psychometric testing phase, had psoriasis.In comparison, more than half of the patients (n = 105) in the psychometric testing phase had acne. This discrepancy suggests that the T-QoL instrument may be less sensitive to HRQoL issues experienced by patients with psoriasis than to HRQoL issues experienced by patients with other skin diseases, particularly acne.
This assumption is supported by our previous qualitative study aimed at developing a conceptual model of HRQoL in adolescents (patients aged 12-17 years old) with psoriasis.Although many themes identified in our interviews are covered in existing HRQoL measures, several important issues are not, such as psoriasis-related worry, what implications having fluctuating physical symptoms has for daily life and planning, and important treatment-related concerns.
To address this knowledge gap, which is relevant to both research and clinical practice, we used a combination of qualitative and quantitative methods to develop and establish the preliminary validity of the Adolescent Psoriasis Quality of Life instrument (APso-QoL), which is the first disease-specific instrument designed to measure HRQoL in adolescent psoriasis (psoriasis in patients aged 12-17 years) via two scales: one that assesses psychosocial impact (APso-PI), and one assesses impact of physical symptoms and treatment (APso-PST).
Being both age appropriate and psoriasis specific, the APso-QoL is intended to support patient-clinician communication about HRQoL-related issues of importance for each individual patient, and aid treatment planning.
# Patients and methods
The development of the APso-QoL was based on recommendations from regulatory agencies and experts within patientreported outcomes.It involved five main steps , which were approved by the Danish National Committees on Biomedical Research Ethics under the exempt status (ID 15007705) and followed Danish Data Protection Agency guidelines (ID 2015-57-0002).
Qualitative data analysis was supported by NVivo software (versions 11 and 12; QSR International, Melbourne, Australia). Statistical analyses were conducted with SPSS (version 25; IBM, Armonk, NY, U.S.A.) and WINSTEPS (version 4Á4Á0; Winsteps.com, Beaverton, OR, U.S.A).
## Patients and recruitment
Eligible participants were patients who were aged 12-17 years, had been diagnosed with psoriasis and were able to understand and read Danish. In order for the study to achieve maximum variation in participant demographic and clinical characteristics, participants were recruited from several sites and sources, including two dermatology hospital clinics, the Danish Psoriasis Association (via the summer school, e-mails to members, and social media), and the Psoriasis in Adolescents cohort.Recruitment sources and demographic details of each sample of patients in the study are shown in.
For each participating adolescent with psoriasis, one parent/caregiver was asked to rate his/her child's HRQoL (proxy report). Healthy controls were recruited from the Danish Psoriasis Association (siblings, children of parents with psoriasis) and the Danish National Birth Cohort. 21
## Phase 1: concept elicitation interviews
HRQoL-related experiences of adolescents with psoriasis were explored using grounded theory data collection methods and inductive thematic analysis.This phase has been described elsewhere.Phase 2: development of the draft questionnaire Preliminary items were generated based on the concept elicitation interviews. To maximize respondent understanding, items and instructions were constructed using simple, age-appropriate language based on the adolescents' own wordings. Existing HRQoL questionnaires used in adult psoriasis or paediatric dermatology were screened to detect additional issues not identified from interviews.
## Phase 3: cognitive interviews
To evaluate and refine the initial draft version, we asked the participants to complete it in front of an interviewer. To explore their understanding of each item, we asked participants to rephrase the items. Participants were also asked about the instructions and response options, as well as their ability and willingness to answer each item. All interviews were conducted by H.R. in a quiet meeting room at a hospital or university building or in the participants' home.
Interviews were conducted in blocks of three to five interviews and transcribed verbatim. After a thorough analysis of transcripts from each block (conducted by H.R.), consensus on revisions was based on group discussion (H.R., R.Z. and J.J.L.). If adolescents repeatedly used different words to describe existing items, we revised the item accordingly to improve understanding and their relevance. Selected items could be presented in a subsequent block of interviews along with one or more alternative items intended to be semantically identical, and participants were asked to comment on their potential similarities and/or differences. If the items were evaluated as identical, participants were asked which item they would prefer and why, and the most appropriate item was retained.
Following cognitive interviews, an expert panel (three dermatologists, a dermatology nurse and a psychologist) evaluated the revised version regarding the format, content, clarity and relevance of the questions and response categories. The experts were also invited to suggest additional items.
We also conducted cognitive interviews on our Danish translation of the disease severity indicator of the self-assessed Simplified Psoriasis Index (saSPI-s; n = 4), which is the only existing disease severity measure validated for paediatric psoriasis and only exists in Dutch. 25
## Phases 4 and 5: item reduction, scale development, initial validation and test-retest assessment
Based on the qualitative phases (phases 1 and 3), we hypothesized that HRQoL in adolescent psoriasis would be best captured using a three-dimensional framework focusing on (i) psychosocial impact, (ii) impact of physical symptoms and (iii) treatment-related impact, and the fact that it might be possible to operationalize the latter two as a single subscale. An electronic questionnaire package consisting of the initial version of the APso-QoL, a background survey and comparator instruments were completed by an independent cohort (sample 3; see. Using these data, we aimed to refine our instrument by reducing the number of items. We relied on the following decision rules. Firstly, we inspected Spearman's rank inter-item correlations. Items with low correlations (< 0Á20) with most other items were suspected of measuring a different construct than hypothesized and removed. Next, item pairs with strong correlations (> 0Á80) were taken to indicate that one item was redundant and could be removed. Likewise, items with several moderate-to-strong inter-item correlations were removed as it was felt the content of the items might be contained in other items. The decision to retain or discard any item was determined by discussion (by H.R. and R.Z.) of the qualitative content and importance of each individual item, and a review of cognitive interview transcripts.
[formula] Mean AE SD 14Á6 AE 1Á5 1 4 Á8 AE 1Á4 1 5 Á7 AE 1Á3 1 5 Á7 AE 1Á2 Median (IQR) 15Á0 (2Á0) 15Á0 (2Á0) 16Á0 (2Á0) 16Á0 (2Á0) Range 12-17 12-17 12-17 13-17 Age at psoriasis onset (years) 0-5 2 - 7 5 6-8 8 - 8 3 9-11 6 - 22 8 12-14 2 - 12 7 15-17 0 - 1 0 Disease severity PASI Mean AE SD 3Á2 AE 2Á8 - - - Median (IQR) 2Á4 (2Á8) - - - saSPI-s Mean AE SD - - 1Á5 AE 2Á2 2 Á5 AE 3Á3 Median (IQR) - - 0Á75 (1Á0) 1Á50 (3Á0) Global question Mean AE SD - - 4Á1 AE 2Á4 3 Á3 AE 2Á0 Median (IQR) - - 4Á0 (4Á0) 3Á0 (3Á0) Current [/formula]
Secondly, we examined the response distribution of items and excluded items displaying floor or ceiling effects, as such items cannot be used to distinguish between different levels of the measured construct. Floor and ceiling effects were defined as ≥ 80% of adolescents endorsing one of the two extreme scores.
Thirdly, for each hypothesized subscale, a one-parameter Andrich Rating Scale Rasch model was fitted to items to evaluate unidimensionality and individual item fit. Items with mean-square fit values between 0Á6 and 1Á4 were considered acceptable.Items with values outside this range were removed in a stepwise manner. Furthermore, we inspected the expected score item characteristic curves of the individual items to evaluate whether item information was appropriately captured by the model.
Fourthly, evidence of local response dependence was evaluated by calculating the Q 3,* index.Items with high residual correlations were evaluated and removed if redundancy was considered the cause of the local response dependence, as violations might lead to inflated estimates of reliability and problems with construct validity.
Rasch analyses were also used to evaluate how well the instrument distinguished between different levels of HRQoL impairment. A low person separation index (< 2, and person reliability < 0Á8) implies that the instrument may not be sensitive enough to distinguish among several levels.Although less than the recommended values, a person separation index of 1Á5 and a reliability coefficient of 0Á7 were considered the minimum values required to divide the patients in the sample into two distinct strata.In addition, the item separation index was used to verify the item hierarchy. Low item separation (< 3, and item reliability < 0Á9) was taken to imply that the number of patients in the sample was not large enough to confirm the item difficulty hierarchy (i.e. construct validity) of the instrument.
## Construct validity
Three comparator instruments were used to evaluate the construct validity of the instrument. The CDLQI is a 10-item dermatology-specific instrument, 10 with a total score ranging from 0 to 30, and is the most commonly used measure of HRQoL in paediatric psoriasis. Furthermore, we used validated Danish versions of the Pediatric Quality of Life Inventory (PedsQL) and the World Health Organization (Five) Wellbeing Index (WHO-5) to assess generic HRQoL and mental wellbeing, respectively.We hypothesized that the APso-QoL would correlate most strongly with the CDLQI, less so with the PedsQL and least with the WHO-5.
As indicators of disease severity, we used the saSPI-s and a global question. The global question asked adolescents to evaluate the overall severity of their psoriasis during the last 4 weeks on a 10-point visual analogue scale. Based on previous research in adult psoriasis,we hypothesized that there would be a small-to-medium correlation between the APso-QoL and self-assessed disease severity.
Agreement between adolescent self-reports and parent proxy ratings was assessed with the intraclass correlation coefficient (ICC; two-way mixed effects, absolute agreement, single measure).An ICC of ≤ 0Á41 was considered to indicate poor-tofair agreement, 0Á41-0Á60 was considered to indicate moderate agreement, 0Á61-0Á80 was considered to indicate good agreement, and 0Á81-1Á00 was considered to indicate excellent agreement. The proxy version was constructed purely for validation purposes and was created as a parallel version of the APso-QoL in which words such as 'I' and 'my psoriasis' were replaced with 'my child' and 'my child's psoriasis', respectively. Based on the results of previous studies,we hypothesized that there would be moderate agreement between self-and proxy reports, with higher agreement for the more easily observable APso-PST than for the APso-PI.
Discriminative validity assesses the ability of an instrument to distinguish between groups differing according to a certain factor. We used a Mann-Whitney U-test to evaluate whether the APso-QoL would be able to distinguish between adolescents with psoriasis and those without. In the nonpsoriasis version of the questionnaire, the word 'psoriasis' was replaced with 'my skin'.
## Reliability
Internal consistencies of the final instrument were assessed with Cronbach's a coefficient. Values above 0Á70 were considered acceptable.Test-retest reliabilities were estimated for each subscale, and the total score was based on the ICC for agreement. Values > 0Á7 are generally considered acceptable,and correlations ≥ 0Á85 indicate suitability for use in clinical trials.
# Results
## Concept elicitation interviews (phase 1)
Analysis of 36 interviews (18 adolescents with psoriasis, 14 parents of adolescents with psoriasis, and four health professionals) resulted in a conceptual framework consisting of six main themes (physical symptoms, feeling different, psoriasisrelated worry about the future, increased attention, attempts to conceal skin, and treatment-related frustration and worry). For further details, see Development of the draft questionnaire (phase 2)
A preliminary 48-item instrument was constructed. The recall period was set to 1 month. All items were answered on a five-point categorical response scale ranging from 'not at all' to 'very much'.
## Cognitive interviews (phase 3)
Three blocks of cognitive interviews conducted between March and June, 2017, were necessary to achieve saturation, that is, so that all adolescents found items, response options and instructions to be straightforward, easy to understand and relevant. Interviews lasted 31-103 min (median 55 min). The characteristics of the 12 adolescent patients who participated are described in.
During this phase, one new item was added, seven items underwent minor language revisions, four items were merged into a single item, and five items were discarded. For example, one item ('I have to be especially careful to prevent my psoriasis from bleeding') was omitted, as it was raised only briefly by a single participant during concept elicitation, and none of the adolescents participating in the cognitive interviews found it relevant. Overall, this phase resulted in a reduction from 48 items to 41 items. In addition, the recall period was changed to 'within the last four weeks', as participants interpreted 'within the last month' in various ways. Finally, four professionals with expertise in paediatric psoriasis and a psychologist reviewed and approved the 41-item questionnaire.
## Item reduction, scale development and initial validation (phases 4 and 5)
Data used for item reduction, scale development and initial validation were collected between January 2018 and February 2019.shows the characteristics of the patients in the sample (n = 50). Inter-item correlations confirmed the hypothesized framework and showed that HRQoL could be reflected in two preliminary scales: (i) psychosocial impact (APso-PI; 33 items) and (ii) impact of physical symptoms and treatment (APso-PST; seven items). One item ('I have quarrels with my parents regarding my psoriasis or my psoriasis treatment') was discarded, owing to the fact that it had low correlations with the remaining items.
Several examples of item groups with moderate-to-high inter-item correlations were identified. For example, the three items 'I think it is embarrassing to have psoriasis', 'I worry about what other people might think about me when seeing my psoriasis', and 'I feel uncomfortable when people are looking at my psoriasis' had high inter-item correlations (> 0Á80), as well as several moderate-to-high correlations with other items. Omitting the second and third of these resulted in the most optimal solution, as the first item showed fewer moderate-to-high inter-item correlations with the other items. This step resulted in nine items from the APso-PI being discarded. Four additional items from the APso-PI were discarded due to floor effects. No items from the APso-PST were omitted during these steps.
Based on independent Rasch analyses, seven items from the APso-PI (out of 20 items) and two items from the APso-PST (out of seven items) were discarded due to misfit. Calculation of the Q 3,* revealed strong evidence of local dependence for the APso-PI (Q 3,* = 0Á63). For the APso-PST subscale, there was less evidence (Q 3,* = 0Á25). After inspection of item content and location, this led to the removal of one item in the APso-PI. The APso-PST subscale was retained. No items displayed unexpected distributions according to their expected score item characteristic curves.
Following item reduction, the final version of the APso-QoL consisted of 17 items across two subscales: APso-PI (12 items) and APso-PST (five items; Appendix S1, Figs S1 and S2; see Supporting Information). Both subscales showed adequate fit to the model (mean-square range 0Á62-1Á36 and 0Á61-1Á25, respectively) and acceptable item separation (4Á00 and item reliability 0Á94, and 3Á65 and item reliability 0Á95). There was some evidence of local response dependence for the APso-PI [three item pairs (numbers 2 and 10, 15 and 17, and 9 and 6) had Q 3,* values just above 0Á3; results not shown].
While the value of the person separation index was acceptable for the APso-PI (2Á53, with person reliability 0Á86), the results for the APso-PST (1Á55, with person reliability 0Á71) suggest that adding more items may result in better discrimination. Item difficulty ranged from À1Á38 to 1Á54 for the APso-PI and from À0Á99 to 1Á44 for the APso-PST (Figs S1 and S2; see Supporting Information). Scores on the APso-PI ranged from zero to 48, and those on the APso-PST ranged from zero to 20, with higher scores representing greater impairment.
## Construct validity
The APso-QoL total score was most strongly correlated with the CLDQI (r = 0Á87), less so with the PedsQL (r = 0Á62) and least with the WHO-5 (r = 0Á50). The APso-QoL correlated moderately with the saSPI (r = 0Á61) and the global question (r = 0Á56). Furthermore, a Mann-Whitney U-test showed that the APso-QoL (item 15 was omitted from this analysis, as no healthy control parallel version could be created) was able to distinguish adolescents with psoriasis (n = 50, mean age 15Á7, 60% female, median APso-QoL total score 14Á5, interquartile range 18Á3) from those without psoriasis (n = 38, mean age 16Á1, 61% female, median APso-QoL total score 4Á0, interquartile range 9Á0; P < 0Á001 for all).
Parents and adolescents (dyad n = 19; all parents were mothers) showed moderate agreement on both subscales (APso-PI ICC = 0Á49 and APso-PST ICC = 0Á57) as well as for the total score (total ICC = 0Á49).
## Reliability
Internal consistency was high for both the APso-QoL total score (a = 0Á95) and the subscales APso-PI (a = 0Á94) and APso-PST (a = 0Á81).
Test-retest coefficients of the total score (ICC = 0Á96), APso-PI (ICC = 0Á97), and APso-PST (ICC= 0Á89) indicated acceptable reliability and stability over time.
# Discussion
This study presents the development and initial validation of the APso-QoL, the first age-and psoriasis-specific instrument for measuring HRQoL in adolescents (patients aged 12-17 years) with psoriasis. Using state-of-the-art methods,our initial validation study found the APso-QoL to be psychometrically sound and to have good preliminary support for construct validity. Furthermore, the test-retest reliability of our instrument was good. Due to their nonuniform nature, the two subscales should ideally be scored separately. Yet, it has been suggested that using a combined summary score may be acceptable for instruments of similar structure, such as the T-QoL and the Skindex-Teen,and thus may be justified for the APso-QoL as well.
The 17-item APso-QoL includes a comprehensive set of items directly reflecting views and experiences of adolescents with psoriasis in ways not previously captured in a single HRQoL instrument. For example, one of the most prevalent HRQoL issues mentioned by adolescents in our qualitative phases was receiving questions and comments from others regarding their psoriasis.In general, adolescents found that most comments and questions arose from ignorance rather than harmful intentions, and only very few adolescents in our samples reported being teased or bullied. This issue is not covered in the skin-specific T-QoL instrument. While CDLQI item 8 includes experiencing troubles from 'people asking questions', the instrument does not cover other important issues. For example, the CDLQI employs a narrow perspective on social relations by focusing on existing friendships only. An important qualitative finding from our research was that, while most adolescents felt at ease showing their psoriasis to friends and family, many reported having great difficulties displaying their psoriasis to people from outside their closest social network and that this negatively affected their ability to form new friendships and feel at ease in larger groups of people.For a scale to be appropriately targeted towards specific patient groups, responses should show adequate variability in a range that is appropriate to its intended use.For this reason, we excluded items where ≥ 80% of adolescents endorsed the lowest response category. In comparison, six out of 10 CDLQI items violated this criterion and hence had poor discriminative value in our sample. This finding may reflect differences in recruitment strategies used in the validation studies. While the CDLQI recruited patients from a specialized paediatric dermatology setting, we aimed at including a broader section of patients. For example, 28% of the participants in our study were not in active treatment.
When the clinical disease severity (PASI and saSPI-s) and HRQoL data for our samples are compared with published data, the differences are striking. To illustrate, a recent systematic review and meta-analysis based on 17 studies of 1185 paediatric patients with psoriasis, who had mainly been recruited from dermatology clinics, revealed a mean CDLQI score of 7Á7 (with higher scores indicating greater impairment of HRQoL) and a mean PedsQL score of 74Á5 (with higher scores indicating better HRQoL).In comparison, the median scores in our validation sample were 2 for CDLQI and 81 for PedsQL. As our sample is less affected, our instrument may be more sensitive for detecting subtle HRQoL issues experienced by adolescents in the general population, making it relevant to individuals seen not only in specialized dermatology hospitals but also in primary care, where most paediatric patients are managed. Future studies could provide more information about the comprehensiveness of the APso-QoL in more severely affected patients.
Some limitations should be mentioned. Firstly, while we aimed at recruiting more patients than we did for the validation, data collection proved to be more challenging than expected, and time limits forced us to terminate data collection before reaching our goal. The difficulty of recruiting adolescent patients with psoriasis is reflected in previous studies on paediatric psoriasis, with a systematic review revealing that only eight of the 17 available studies were based on samples of more than 50 patients.Furthermore, in validation studies of CDLQI, Skindex-Teen and T-QoL, patients with psoriasis only accounted for 10Á7% or less of the total number of patients with skin disease.The small sample size in our validation study influenced our analytical choices, for example, we did not evaluate response category functioning and differential item functioning, 38,39 which may limit the generalizability of our results. Although the Rasch model used for validation purposes has been shown to produce relatively robust item calibrations even with small samples, 40 this view has also been challenged,and it is possible that validation studies with larger samples could show evidence of item misfit or local response dependence. The present 17-item APso-QoL should therefore be considered preliminary.
Yet, Rasch analyses, in combination with other techniques, allowed us to reduce our item pool into a 17-item instrument that confirmed our hypotheses regarding dimensionality and showed promising psychometric properties. Due to the robust methods used during the qualitative phases that ensured strong face and content validity,we feel confident that the APso-QoL in its present form will be highly informative when used in clinical practice. To illustrate, scores on the different items might alert clinicians regarding issues of importance for each individual patient, facilitating patient-physician communication about issues that otherwise could be left undetected. The APso-QoL can also be used as a screening instrument to identify patients who might need referral to additional services such as a psychologist or peer support groups.
Although the ages of the participants during qualitative phases were evenly distributed, most participants in our validation study were older adolescents, and only a few 12-and 13-year-old patients participated. Further validation of the APso-QoL in independent samples is thus essential, especially in younger adolescents.
For the APso-QoL to be used in clinical trials, future studies should evaluate additional psychometric properties. These include differential item functioning, response category functioning, responsiveness, interpretability and administrative burden. Future studies should also test the validity of using a combined APso-QoL summary score. As the APso-QoL was developed and preliminarily validated in Danish-speaking patients, future studies should also examine its cross-cultural validity.
We were unable to confirm the diagnosis of psoriasis in the subgroup of participants recruited from the Danish Psoriasis Association. However, participants in a psoriasis-specific patient organization are likely to have the disease, and all adolescents confirmed that they had received the diagnosis from a physician. Furthermore, self-selection bias may have limited the representativeness of our samples, especially in sample 3, as many of these adolescents were invited to participate via email and social media.
In conclusion, our results indicate the promising psychometric properties of the APso-QoL, the first age-appropriate and disease-specific HRQoL instrument for use in adolescents (patients aged 12-17 years old) with psoriasis. Further testing in independent samples of larger size is needed, and additional attributes of the APso-QoL should be examined. However, in both clinical and nonclinical settings, the APso-QoL could be a valuable and sensitive tool for evaluating HRQoL impairment in adolescents with psoriasis. |
Unique Symmetry-Breaking Phenomenon during the Self-assembly of Macroions Elucidated by Simulation
Various soluble hydrophilic macroions can self-assemble into hollow, spherical, monolayered supramolecular "blackberry"-type structures, despite their like-charged nature. However, how the 3-D symmetrical macroions prefer to form 2-D monolayers in bulk solution, especially for the highly symmetrical "Keplerate" polyoxometalates and functionalized C 60 macroions has been a mystery. Through molecular dynamics simulations, using a model specifically designed for macroions in solution, the mechanism of this intriguing symmetry-breaking process is found to be related to the apparently asymmetric charge distribution on the surface of macroions in the equatorial belt area (the area which can be effectively involved in the counterion-mediated attraction). As a result, the electric field lines around macroions during the self-assembly process clearly show that the symmetry-breaking happens at the dimer level effectively defining the plane of the self-assembly. These findings are expected to contribute to our fundamental knowledge of complex solution systems that are found in many fields from materials science to biological phenomena.Published: xx xx xxxx OPEN www.nature.com/scientificreports/
Hydrophilic macroions possess fascinating solution behaviors. Such large ions cannot be described by either Debye-Hückel theory 1 for simple ions (due to their large sizes) or the DLVO theory 2,3 for colloids because they still form thermodynamically stable solutions and the van der Waals (VDW) forces are very weak. Recent studies indicate that the like-charged macroions can strongly attract with each other when carrying moderate amount of charges, leading to the reversible formation of thermodynamically stable, hollow, spherical, and monolayered "blackberry"-type structures in polar solvents [bib_ref] A Complete Macroion−"Blackberry" Assembly− Macroion Transition with Continuously Adjustable Assembly Sizes in..., Kistler [/bib_ref] [bib_ref] Deprotonations and charges of well-defined {Mo72Fe30} nanoacids simply stepwise tuned by pH..., Liu [/bib_ref] [bib_ref] Counterion Distribution around Hydrophilic Molecular Macroanions: The Source of the Attractive Force..., Pigga [/bib_ref] [bib_ref] Solution behaviors and self-assembly of polyoxometalates as models of macroions and amphiphilic..., Yin [/bib_ref]. Vairous macroions (1-6-nm size) are found to do so, such as inorganic metal-oxide molecular clusters [bib_ref] Introduction: Polyoxometalatess Multicomponent Molecular Vehicles To Probe Fundamental Issues and Practical Problems, Hill [/bib_ref] [bib_ref] Polyoxometalate clusters, nanostructures and materials: from self assembly to designer materials and..., Long [/bib_ref] [bib_ref] Supramolecular structures of polyoxomolybdate-based giant molecules in aqueous solution, Liu [/bib_ref] [bib_ref] Self-assembly in aqueous solution of wheel-shaped Mo 154 oxide clusters into vesicles, Liu [/bib_ref] [bib_ref] An unusually slow self-assembly of inorganic ions in dilute aqueous solution, Liu [/bib_ref] , polyhedral oligomeric silsesquioxane [bib_ref] Self-Assembly of Subnanometer-Scaled Polyhedral Oligomeric Silsesquioxane (POSS) Macroions in Dilute Solution, Zhou [/bib_ref] , functionalized fullerenes [bib_ref] Charge-regulated spontaneous, reversible self-assembly of the carboxylic acid-functionalized hydrophilic fullerene macroanions in..., Yin [/bib_ref] , dendrimers [bib_ref] Molecular structure encodes nanoscale assemblies: understanding driving forces in electrostatic selfassembly, Willerich [/bib_ref] [bib_ref] Supramolecular assembly of poly (propyleneimine) dendrimers driven by simple monovalent counterions, Eghtesadi [/bib_ref] , metal-organic nanocages [bib_ref] Self-assembly of ten molecules into nanometre-sized organic host frameworks, Fujita [/bib_ref] [bib_ref] Self-Assembly of Discrete Cyclic Nanostructures Mediated by Transition Metals, Leininger [/bib_ref] [bib_ref] Spontaneous Self-Assembly of Metal− Organic Cationic Nanocages to Form Monodisperse Hollow Vesicles..., Li [/bib_ref] [bib_ref] Viral-Capsid-Type Vesicle-Like Structures Assembled from M12L24 Metal-Organic Hybrid Nanocages, Li [/bib_ref] , bio-macromolecules and small nanoparticles [bib_ref] Self-assembly of polyoxometalate macroanion-capped Pd0 nanoparticles in aqueous solution, Zhang [/bib_ref] [bib_ref] Self-assembly of nanoparticles into biomimetic capsid-like nanoshells, Yang [/bib_ref] [fig_ref] Figure 1: Coarse-graining of various macroions that form blackberry structures [/fig_ref]. The blackberry structure formation has been confirmed to be due to the counterion-mediated attraction [bib_ref] A Complete Macroion−"Blackberry" Assembly− Macroion Transition with Continuously Adjustable Assembly Sizes in..., Kistler [/bib_ref] [bib_ref] Deprotonations and charges of well-defined {Mo72Fe30} nanoacids simply stepwise tuned by pH..., Liu [/bib_ref] [bib_ref] Counterion Distribution around Hydrophilic Molecular Macroanions: The Source of the Attractive Force..., Pigga [/bib_ref] [bib_ref] Solution behaviors and self-assembly of polyoxometalates as models of macroions and amphiphilic..., Yin [/bib_ref]. When the macroions have inhomogeneous surface charge distribution, tubular shaped assemblies have been observed [bib_ref] The self-assembly of a macroion with anisotropic surface charge density distribution, Haso [/bib_ref].
Simulations have provided several important information in understanding the diffusion of macroions in solution and the distribution of water molecules and counterions around them [bib_ref] Elucidating the origin of the attractive force among hydrophilic macroions, Liu [/bib_ref] [bib_ref] Crystal structure and dynamics of 12-heteropoly compounds as investigated by molecular dynamics, Tsujimichi [/bib_ref] [bib_ref] Polyoxometalates in solution: molecular dynamics simulations on the α-PW12O403-Keggin anion in aqueous..., Lopez [/bib_ref] [bib_ref] Keggin polyoxoanions in aqueous solution: Ion pairing and its effect on dynamic..., Leroy [/bib_ref].
Our recent simulation study agrees with the hypothesis that the attraction between the like-charged macroions comes mainly from electrostatic interactions mediated by their counterions [bib_ref] Elucidating the origin of the attractive force among hydrophilic macroions, Liu [/bib_ref]. The VDW interaction is also contributing to the attractive forces, but we found the magnitude of this interaction to be about two orders of magnitude smaller than the electrostatic interaction for 2.5-nm-size spherical macroions [bib_ref] Elucidating the origin of the attractive force among hydrophilic macroions, Liu [/bib_ref]. Further investigations have discovered the chaotic nature of the electrostatic forces among macroions and their counterions, which dramatically decelerates the formation of self-assembled structures and makes the process more statistically dependent. The effect of macroionic charge density was also studied, which showed how the interactions between macroions as well as the dynamics of both macroions and counterions are dependent on the charge density [bib_ref] Elucidating the origin of the attractive force among hydrophilic macroions, Liu [/bib_ref] [bib_ref] Crystal structure and dynamics of 12-heteropoly compounds as investigated by molecular dynamics, Tsujimichi [/bib_ref] [bib_ref] Polyoxometalates in solution: molecular dynamics simulations on the α-PW12O403-Keggin anion in aqueous..., Lopez [/bib_ref] [bib_ref] Keggin polyoxoanions in aqueous solution: Ion pairing and its effect on dynamic..., Leroy [/bib_ref].
The most intriguing remaining question is why the macroions assemble into hollow, spherical structures? Many types of macroions, such as the Keplerates 30,31 , C 60 and some nanocages, are structurally isotropic, which is different from the structurally-anisotropic surfactants. To form the hollow, spherical blackberry structure, the macroions need to have stronger intermolecular attraction along certain directions in a homogeneous bulk solution. That means, a symmetry-breaking process should take place, but how that happens is still a major mystery. In addition, the effect of macroionic size on their self-assembly behavior is also unclear.
# Results and discussion
The effect of macroionic size on their self-assembly behavior has not been explored through simulation so far, to the best of our knowledge, probably due to the huge computational cost when trying to simulate macroions by all-atom molecular dynamics approaches. To overcome this difficulty, our coarse-grained model that has been used in a previous study 25 is expanded to macroions with different sizes. The original CG model was designed to study the source of attraction between macroions in solution and the general self-assembly behaviors of various types of macroions, thus each macroion is represented by a hollow spherical structure which has both charged and uncharged beads on the surface, to mimic the structure of the well-studied 2.5-nm-size spherical "Keplerate" metal-oxide molecular cluster {Mo 72 Fe 30 } 30,31 . The counterions and solvents used with this model are also coarse-grained (the details are described in supporting information). Because of the simplicity and flexibility of this CG model, it can be easily expanded to macroions with different sizes and charge distributions, and simulations of larger macroions (up to 10 nm) using these types of CG models are accessible.
In order to study the effect of the macroionic size on their self-assembly behaviors, CG models of macroions with four different sizes from 2.5-10 nm were created [fig_ref] Figure 1: Coarse-graining of various macroions that form blackberry structures [/fig_ref] , and four systems were built accordingly. All species were initially randomly distributed in solution, followed by equilibration for >200 ns. The visualization of simulation results is shown in [fig_ref] Figure 2: Size effect on the self-assembly behavior of macroionic solutions [/fig_ref] -c. Regardless of the size difference, each of the four types of macroions forms into one large aggregate, indicating that either electrostatic or VDW forces drives the self-assembly process.
Further investigation was performed by varying the charge density (0-0.16 C/cm 2 ) on each type of macroion, in order to understand the role of the electrostatic interaction in the assembly process. Interestingly, for macroions >7.5 nm in size with low charge densities, a great tendency to aggregate was found. The macroion-macroion radial distribution function (RDF) was then calculated for each system (an example is shown in [fig_ref] Figure 2: Size effect on the self-assembly behavior of macroionic solutions [/fig_ref] with detailed discussion). The number of nearest neighbors is then calculated by integrating the area under the first two peaks observed in the RDF calculations, leading to a full-spectrum comparison of the charge density and size effect on the self-assembly of macroions [fig_ref] Figure 2: Size effect on the self-assembly behavior of macroionic solutions [/fig_ref]. Several intriguing features are observed from this comparison: first, as the size of the uncharged macromolecules increases, larger assemblies tend to form, indicating that the strength of the VDW interactions between macroions correlate with their sizes; secondly, increasing the charge density of macroions always has a positive effect on the tendency of forming assemblies, implying that in all cases the electrostatic forces are attractive. However, as the macroionic size increases, the electrostatic attraction between macroions diminishes, and may even become repulsive as predicted by the DLVO theory of large colloids. This observation supports the experimental results that the electrostatic interaction is responsible for the blackberry structure formation of macroions, and is dominant when the macroionic size is less than ~10 nm. VDW attractions start to dominate and cause electrostatic forces to become less attractive and even repulsive as the macroions grow larger, leading to typical colloidal behavior such as precipitation from the solution.
The most interesting yet unsolved mystery is the hollow, spherical, single-layered morphology of the blackberry structure. An attempt has been made to observe blackberry structure formation using our CG model for macroions in which their size, surface charge density, and charge distribution can be easily altered to simulate a full range of VDW/electrostatic interaction between them. Although it is difficult to simulate the formation of a whole blackberry structure, because it may contain thousands of single macroions; it is possible to simulate the early stage of the formation since the macroions should initially self-assemble into a 2D monolayer. However, simulation of a large system containing 50 macroions (each with 2.5-nm-diameter and 20-charges), 1,000 counterions, and 5 million solvent molecules didn't show any sign of 2D monolayer formation, only a large 3D irregular aggregate was formed. This result is reasonable because there should not be directional preference for the assembly since the macroions were assigned random charge distribution on their surface. Then the key question is: how do macroions initially self-assemble into a monolayer structure? Since many macroions have rigid structures, in order to break the isotropic symmetry and form a monolayer, the positions of the charged sites on the macroions may be reconfigured depending on the solution environment. Thus we hypothesize that the charge distribution on macroions is the key for the formation of blackberry structure.
To verify our hypothesis, various types of macroions covering a big range of charge densities and charge distributions were created. [fig_ref] Figure 3: 2D monolayer structures formed by macroions with equatorial charge distributions and the... [/fig_ref] -h show eight types of representative macroions which have moderate charge densities and different charge distributions. Interestingly, among macroions with different charge distributions, some did self-assemble into nicely packed 2D monolayers, when the charges are distributed near the "equator" of the macroions [fig_ref] Figure 3: 2D monolayer structures formed by macroions with equatorial charge distributions and the... [/fig_ref] that they can form 2D monolayers. The others all ended up forming 3D aggregates. Careful examination of the monolayers [fig_ref] Figure 3: 2D monolayer structures formed by macroions with equatorial charge distributions and the... [/fig_ref] reveals that the macroions seem to have a well-defined hexagonal packing (also confirmed by the RDF characterizations, [fig_ref] Figure 4: Self-assembly of macroions with a quasi-isotropic [/fig_ref] , which was also observed in {Mo 154 } solution although the packing was only short-ranged [bib_ref] Self-assembly in aqueous solution of wheel-shaped Mo 154 oxide clusters into vesicles, Liu [/bib_ref]. Furthermore, the monolayers in solution seem to be not rigid most of the time, instead they demonstrate fluctuating surfaces. Some defects are also observed [fig_ref] Figure 3: 2D monolayer structures formed by macroions with equatorial charge distributions and the... [/fig_ref].
Since the charge distribution is crucial for the 2D monolayer formation, in reality would the charges on macroionic surface redistribute during the self-assembly process? If they do, could the reason be that the new states with redistributed charges are more energetically favorable? Several batches of simulations were then performed. In each batch, different macroionic solution systems were built with same conditions except the charge distributions on the macroionic surface [fig_ref] Figure 5: Electric field characterization during the self-assembly process of macroions with a quasi-isotropic... [/fig_ref]. All comparisons show that the systems with macroions having charges distributed close to their "equators" always have the lowest total energy after forming stable assembled structures. These results indicate that if the charges on the macroions are movable, they may eventually redistribute closer to the macroion's equatorial area to lower the system energy, resulting in a 2D monolayer structure that slowly forms over time.
Furthermore, the dynamic process of 2D monolayer formation was evaluated. -f show a typical monolayer formation from macroions carrying 10 charges on the "equator". The self-assembly process is very slow as can be seen in the figure, fully consistent with experimental observations [bib_ref] An unusually slow self-assembly of inorganic ions in dilute aqueous solution, Liu [/bib_ref]. In addition, different methods were developed to quantify this process. [fig_ref] Figure 3: 2D monolayer structures formed by macroions with equatorial charge distributions and the... [/fig_ref] shows how the average size and number of the oligomers formed by macroions change with time. A clear sigmoidal curvature is observed, again consistent with our previous observations [bib_ref] Lag periods during the self-assembly of {Mo72Fe30} macroions: connection to the virus..., Zhang [/bib_ref]. Moreover, the calculated number of nearest neighbors [fig_ref] Figure 3: 2D monolayer structures formed by macroions with equatorial charge distributions and the... [/fig_ref] also displays a similar sigmoidal feature. Accordingly, an interesting picture is revealed: the slow initial induction comes from the difficult oligomer formation, possibly due to the short-ranged and chaotic nature of the electrostatic attraction 25 mediated by counterions. As larger oligomers slowly form, the increase in the attractive electrostatic forces accelerates the assembling process; eventually the dramatic drop in local concentration of free macroions and oligomers results in very slow completion of large monolayers.
A very critical question is why the structurally isotropic macroions such as Keplerates, with apparently isotropic charge distributions, also self-assemble into 2D monolayers prior to the blackberry structure formation. To better understand this, a model was designed: 30 negative charges were symmetrically assigned on the 30 vertices of an icosidodecahedron on the surface of a macroion (as shown in , identical to the locations of possible charge sites on {Mo 72 Fe 30 } surface. Surprisingly, such macroions also forms a 2D monolayer [fig_ref] Figure 4: Self-assembly of macroions with a quasi-isotropic [/fig_ref]. Although a minor defect is observed, the macroions pack in hexagonal closest-packed structure, like the ones with charges distributed close to their equators. Compared with other macroions having purely isotropic charge distributions that we have studied before, such as the ones with charges distributed on the vertices of a cube [fig_ref] Figure 3: 2D monolayer structures formed by macroions with equatorial charge distributions and the... [/fig_ref] or the ones with all surface beads charged 25 , this result seems particularly intriguing since all the others form 3D aggregations instead of 2D monolayers.
To better understand the reason, we carefully examined the symmetry of charge distribution on the icosidodecahedron. [fig_ref] Figure 4: Self-assembly of macroions with a quasi-isotropic [/fig_ref] ,f show the top and side views of this structure, when it sits on one of its pentagons on the surface. No matter how one rotates the sphere, the side view of the vertices shows an anisotropic charge distribution: the density of the vertices on the "belt" surrounding the sphere seems to be relatively higher. Since the average distance between the surface of macroions after forming a monolayer is ~0.6 nm [fig_ref] Figure 4: Self-assembly of macroions with a quasi-isotropic [/fig_ref] and experimentally confirmed) and assuming the counterions need to stay in the region between two macroions where the distance between the surfaces of the macroions is ~1.0 nm in order to effectively mediate the attraction (see , we can then define a belt area which has a width of ~1.4 nm on the surface of macroions. When placing a macroion [fig_ref] Figure 4: Self-assembly of macroions with a quasi-isotropic [/fig_ref] , the charge density inside this belt area is about 57% higher than the rest of the surface area. This explains the mechanism of the monolayer formation: the higher charge density around the belt area breaks the symmetry and attracts more counterions to this area and becomes the plane for a 2D self-assembly, similar to the case of equatorially charged macroions like {Mo 154 }. However, when placing the macroion on one of its triangles, the charge density inside this belt area is about 47% lower than the rest of the surface area, which explains why most of the macroions in the monolayer align themselves along their pentagons instead of the triangles. In general, macroions with even very slight anisotropy in their charge distribution may cause a "belt" area with higher charge density to form around them, which enable them to form a 2D monolayer structure. This means, in our hypothesis regarding the relationship between the charge distribution and the ability to form blackberry structures, the charges on the macroions need not move very far, even a slight tendency of redistributing the charges closer to the equator may lead to the formation of 2D monolayer structures.
Two approaches of expanding the size of the monolayer structure were designed to test if the monolayer structure was formed out of coincidence. The first one is to introduce the same type of macroions one at a time with corresponding counterions and solvent molecules into the solution which contains the monolayer. As expected, the single macroions slowly merged with the monolayer on the edge, leading to an expansion of the packing [fig_ref] Figure 4: Self-assembly of macroions with a quasi-isotropic [/fig_ref]. Another approach is to replicate the small monolayer four times and mix them in a large solution system. Again the monolayers quickly merged with each other along their edges and formed one large monolayer [fig_ref] Figure 4: Self-assembly of macroions with a quasi-isotropic [/fig_ref]. Comparing the two tests, the merging of the monolayer and single macroions is much slower but leads to a fine packing structure, while the merging of monolayers is faster due to the stronger electrostatic interactions, but the assembled structure is less perfect (some defects can be observed in [fig_ref] Figure 4: Self-assembly of macroions with a quasi-isotropic [/fig_ref]. These observations also correlate with the sigmoidal curvature in the self-assembly process, and maybe related to the size difference of blackberry structures formed by different types of macroions.
To better understand why the monolayers grow along the edges instead of stacking on top of each other or crossing each other's planes, and how the single macroions gradually merge with monolayers, an approach of characterizing the electric field surrounding the charged species was developed to visualize the electrostatic interaction among them, since it has been proved that the electrostatic force is the driving force for the self-assembly [bib_ref] Elucidating the origin of the attractive force among hydrophilic macroions, Liu [/bib_ref]. A test of two single macroions using this method reveals how counterions mediate the attraction . Once a dimer is formed, a resultant electric field that is attractive to other macroions is observed around the "belt" area of the dimer that dictates the plane of monolayer growth direction. Further investigation illustrates how the single macroions merge with a monolayer [fig_ref] Figure 5: Electric field characterization during the self-assembly process of macroions with a quasi-isotropic... [/fig_ref] -d, a more detailed process is shown in , and how two monolayers merge with each other [fig_ref] Figure 5: Electric field characterization during the self-assembly process of macroions with a quasi-isotropic... [/fig_ref]. The self-assembly process becomes clear now. A resultant electric field that is attractive to other macroions is clearly observed around the plane of the monolayer [fig_ref] Figure 5: Electric field characterization during the self-assembly process of macroions with a quasi-isotropic... [/fig_ref]. This electric field is responsible for the monolayer to grow larger by attracting other macroions and we hypothesize that this will finally result in the formation of the blackberry structure.
In conclusion, we have performed large-scale CG molecular dynamics simulations to address the fundamental question that how the symmetry-breaking process is achieved during the blackberry structure formation in dilute solution. We believe our general approach of understanding the process of self-assembly of charged molecules in solution will open a new direction in the study of the self-assembly and the nature of interactions in the broadly defined macroionic solutions, which cover a variety of fields from materials science to biological phenomena.
# Methods
In order to study the general self-assembly behaviors of various hydrophilic macroions, a versatile coarse-grained (CG) model that represents macroions of varying charge density and size was developed in a previous work [bib_ref] Elucidating the origin of the attractive force among hydrophilic macroions, Liu [/bib_ref] and applied in this work. The design of the CG model is based on the molecular structure of typical macroions such as polyoxometalate molecules. One macroion is represented by one hollow sphere with two different types of beads on the surface. The surface beads are either uncharged or charged in order to represent the van der Waals and electrostatic interactions among macroions, counterions and solvent molecules in the solution. The size and charge values of each surface bead, the size of the macroion, and the number of charged beads and their distribution on the surface can all be tuned to represent a specific type of macroion. The surface beads of a macroion are designed to move as one rigid body, which benefits the efficiency of the molecular dynamics simulations since the intra-molecular interactions within each macroion are not considered. This design is based on the assumption that the shape, size and composition of each macroion will not change in the process of self-assembly in solution, which means that there is no obvious relative movement of atom groups on the surface of each macroion.
The Lennard-Jones (LJ) 12-6 potential energy function was used to describe the van der Waals interactions between different kinds of species in the solution:
[formula] ε σ σ = − ≤ > U r r r r r r r ( ) 4 , [/formula]
## 0,
LJ c c Here U LJ (r) is the VDW interaction between pairs of beads separated by a distance of r; ε is the energy term and r c is the cut-off distance for LJ potential. The CG force field parameters for solvent were taken from the model of water in MARTINI force field. In this CG model of water, one bead is equivalent to four water molecules. The CG beads on the surface of macroions also have the same size (5 Å) and van der Waals interaction parameters as the solvent beads to account for their hydrophilic characteristic, so are the counterions. The CG force field parameters of the solvent can be tuned to represent a good or bad solvent for the macroions.
In this work all coarse-grained molecular simulations were performed by LAMMPS package, and the force field parameters used in all CG simulations are unitless quantities (LJ-style units). Without losing generality, LAMMPS has set the fundamental quantities, such as mass, sigma, epsilon and the Boltzmann constant to 1, and the specified masses, distances and energies in simulations are multiples of these fundamental values. The formulas that correlate the reduced (unitless) quantities to the same quantities with real units is provided in the LAMMPS manual [bib_ref] Fast parallel algorithms for short-range molecular dynamics, Plimpton [/bib_ref]. Therefore one can use the mass, sigma and epsilon values for a specific material and convert the simulation results from a unitless LJ-type simulation into physical quantities. The coarse-grained simulation studies performed in this work all used reduced units, while when interpreting the force field parameters in the discussions, equivalent real units are normally used in order to depict a more realistic picture. in the SI shows detailed conversion between the reduced LJ-style quantities and real quantities. The conversions have been carefully performed according to the LAMMPS manual [bib_ref] Fast parallel algorithms for short-range molecular dynamics, Plimpton [/bib_ref]. In the CG model of macroionic solutions the ε of all pair interactions between all kinds of species is set to 4.5 kJ/mol, and the σ is set to 5 Å in order to obtain a good solvent environment. The cut-off distance r c is set to 12.5 Å for all LJ interactions. Furthermore, the interactions between the charged beads on the surface of the macroions and the corresponding counterions in the solution were described by the Coulomb pair-potential: Here U Coul is the Coulomb potential for a pair of beads separated by a distance of r, q α and q β are the charges on each bead, respectively; and k q = 1/4πε 0 , where ε 0 is the permittivity of vacuum. Long range Coulombic interactions were calculated using the particle-particle/particle-mesh (PPPM) Ewald algorithm. In the current model, each charged bead on the surface of macroions has one negative charge, and accordingly each counterion has one positive charge, while the solvent beads are not charged. All the beads on each macroion move as one rigid body, thus the optional RIGID package in LAMMPS was installed and utilized for this purpose. The time step used in these coarse-grained MD simulations is 0.005 in reduced CG unit, equivalent to 10 fs. Each simulation was started with distributing the macroions, counterions and solvent molecules randomly in the system. The code written for distributing different types of particles in the initial simulation box ensures that no solvent or counterion particle is in the cavity of a macroion. All simulations was performed under isothermal-isobaric (NPT) ensemble in the beginning at a temperature of 1.0 and a pressure of 0.1 in reduced unit, and once the density of the system reaches a stable value, new simulations were continued from the restart configurations with isothermal (NVT) ensemble to enhance the computational efficiency. Due to the slow dynamics of macroions, most of the simulations took more than 100 ns to reach a state where the assembled structures are stable.
## Statistical information.
To ensure the various morphologies generated from the MD simulations are in the equilibrium state, same simulations have been repeated several times from different initial distributions, and all tests show that the assembled structures are similar in equilibrium state.
## Data availability
All relevant data are available from the authors upon reasonable request.
[fig] Figure 1: Coarse-graining of various macroions that form blackberry structures. (a) Examples of different kinds of macroions, including inorganic metal-oxide molecular clusters (1, 2) 8-13 , metal-organic nanocages (3) 18-21 , functionalized fullerenes (4) 15 , cyclodextrins (5) 14 and dendrimers (6) 17 . (b) A typical blackberry structure self-assembled from metal-oxide molecular clusters (a1), which is a monolayer hollow sphere. (c) A coarse-grained model designed for general spherical macroions. In this model, the cyan beads have only VDW interactions while the yellow beads have both VDW and electrostatic interactions. Scientific REPORTS | (2018) 8:13076 | DOI:10.1038/s41598-018-31533-z [/fig]
[fig] Figure 2: Size effect on the self-assembly behavior of macroionic solutions. (a-c) The final assembled states of macroions with a size of 2.5, 5.0 and 7.5 nm, respectively. The uncharged beads on macroions are cyan, while the charged ones are yellow. The counterions are red. Explicit solvent molecules are hidden for clarity. (d) The number of nearest neighbors for macroions with size ranging from 2.5 to 10 nm and charge density ranging from 0 to 0.16 C/m 2 . [/fig]
[fig] Figure 3: 2D monolayer structures formed by macroions with equatorial charge distributions and the time evolution. (a) Final assembly of macroions with an equatorial charge distribution (10 charges). (b) Final assembly of macroions with 20 charges distributed on the equator and two "tropics". Solvents are hidden for clarity. (c)The size and number of assembled clusters plotted as a function of time. (d) The average number of nearest neighbors of macroions plotted with time. The solid lines are drawn to guide the eyes. Scientific REPORTS | (2018) 8:13076 | DOI:10.1038/s41598-018-31533-z on one of its pentagons as shown in Fig. 4f (see also the top view in [/fig]
[fig] Figure 4: Self-assembly of macroions with a quasi-isotropic (icosidodecahedron shaped) charge distribution. (a) Self-assembled structure of 10 macroions. (b) After adding four macroions one by one into a. (c) 2D monolayer merged from four small monolayers as shown in a. (d) The shape of an icosidodecahedron. (e) The top view of this polyhedron when sitting on one of its pentagons on the surface. (f) The side view when sitting on one of the pentagons. (g) The side view when sitting on one of the triangles. Scientific REPORTS | (2018) 8:13076 | DOI:10.1038/s41598-018-31533-z [/fig]
[fig] Figure 5: Electric field characterization during the self-assembly process of macroions with a quasi-isotropic charge distribution. (a-d) The evolution of the electric field in the system where a single macroion merges with a monolayer (as shown in Fig. 4a). (e,f) The initial and end states of two monolayers merging with each other. The coloring method is the same as in Fig. S8. Scientific REPORTS | (2018) 8:13076 | DOI:10.1038/s41598-018-31533-z [/fig]
|
Direct Oral Anticoagulant Drugs: On the Treatment of Cancer-Related Venous Thromboembolism and their Potential Anti-Neoplastic Effect
Cancer patients develop a hypercoagulable state with a four-to seven-fold higher thromboembolic risk compared to non-cancer patients. Thromboembolic events can precede the diagnosis of cancer, but they more often occur at diagnosis or during treatment. After malignancy itself, they represent the second cause of death. Low molecular weight heparins are the backbone of the treatment of cancer-associated thromboembolism. This treatment paradigm is possibly changing, as direct oral anticoagulants (DOACs) may prove to be an alternative therapeutic option. The currently available DOACs were approved during the first and second decades of the 21st century for various clinical indications. Three molecules (apixaban, edoxaban and rivaroxaban) are targeting the activated factor X and one (dabigatran) is directed against the activated factor II, thrombin. The major trials analyzed the effect of these agents in the general population, with only a small proportion of cancer patients. Two published and several ongoing studies are specifically investigating the use of DOACs in cancer-associated thromboembolism. This article will review the current available literature on the use of DOACs in cancer patients. Furthermore, we will discuss published data suggesting potential anti-cancer actions exerted by non-anticoagulant effects of DOACs. As soon as more prospective data becomes available, DOACs are likely to be considered as a potential new therapeutic option in the armamentarium for patients suffering of cancer-associated thromboembolism.
# Introduction
Thromboembolism represents a common complication of cancer. The risk of developing a venous thromboembolic event (VTE) associated with malignancy is estimated to be four to seven-fold higher compared to the normal population [bib_ref] Venous thromboembolism and cancer: Risks and outcomes, Lee [/bib_ref] [bib_ref] Risk of venous thromboembolism in patients with cancer: A systematic review and..., Horsted [/bib_ref]. Between 20% and 30% of first thromboembolic events (TEs) are associated with malignancy [bib_ref] Epidemiology of cancer-associated venous thrombosis, Timp [/bib_ref]. Due to increasing aging of the general population, a much more rapid diagnosis of malignancies and a prolonged survival of patients, the incidence of thromboembolic events among cancer patients is increasing.
A VTE is the second cause of death in addition to the malignant disease itself [bib_ref] Thromboembolism is a leading cause of death in cancer patients receiving outpatient..., Khorana [/bib_ref]. Cancer patients who develop a TE have a shorter survival compared to those without [bib_ref] Prognosis of cancers associated with venous thromboembolism, Sorensen [/bib_ref] [bib_ref] Prognostic relevance of an asymptomatic venous thromboembolism in patients with cancer, Dentali [/bib_ref] [bib_ref] Body height and risk of venous thromboembolism: The Tromso Study, Braekkan [/bib_ref]. Moreover, treatments of the underlying neoplastic disease can be delayed, hospitalisation stay prolonged and, eventually, health-care costs are raised. All these characteristics highlight the importance of a well-managed treatment. Such a treatment should be simple (improving patient's compliance), cheap, and safe-with the smallest possible rate of recurrences and/or bleedings.
## Treatment of cancer-associated venous thromboembolism: current guidelines
Many guidelines for the management of the cancer-associated thromboembolism (CAT) are published and regularly updated. The most recent updates date from 2016 by the American College of Chest Physicians (ACCP), 2015 by the American Society of Clinical Oncology (ASCO) [bib_ref] Venous thromboembolism prophylaxis and treatment in patients with cancer: American society of..., Lyman [/bib_ref] , 2015 by the Haemostasis and Thrombosis Task Force of The British Committee for Standards in Haematology [bib_ref] Guideline on aspects of cancer-related venous thrombosis, Watson [/bib_ref] , 2016 by the International Initiative on Thrombosis and Cancer (ITAC-CME) [bib_ref] International clinical practice guidelines including guidance for direct oral anticoagulants in the..., Farge [/bib_ref] and 2011 by the European Society of Medical Oncology (ESMO) [bib_ref] National Comprehensive Cancer Network. Clinical Practice Guidelines in Oncology. Cancer-Associated Venous Thromboembolic..., Mandala [/bib_ref]. Based on these guidelines, LMWHs are still the treatment of choice for TEs in patients with a malignancy. LMWHs are the preferred agent for both the initial and for the long-term treatment, which should last at least three months.
Currently, DOACs are not considered for the acute and long-term treatment of CAT in most guidelines. However, several guidelines recognize the potential of these drugs. For instance, the 2016 updated consensus recommendations from the International Initiative on Thrombosis and Cancer (ITAC-CME) consider DOACs for patients with stable disease not receiving systemic anticancer therapy [bib_ref] International clinical practice guidelines including guidance for direct oral anticoagulants in the..., Farge [/bib_ref]. The 2016 published ACCP guidelines state that there is no preference between DOACs and VKAs when LMWHs are not used. Since 2016, even if based on weak clinical evidence, DOACs are considered as a good alternative to LMWHs, particularly in the presence of problems (e.g., costs, painful injections) which could lead to poor long-term compliance. The recently published update 2018 of the clinical guidelines by the National Comprehensive Cancer Network (NCCN) consider edoxaban for the treatment of patients diagnosed with CAT with a level 1 of evidence . Moreover, recommendations edited in June 2018 by the Scientific and Standardization Committee (SSC) from the International Society on Thrombosis and Haemostasis (ISTH) suggest the use of specific DOACs for cancer patients with an acute diagnosis of VTE. The recommendation considers the use of edoxaban or rivaroxaban as these two molecules were compared with LMWHs in randomized clinical trials. DOACs should only be administered in patients presenting with a low risk of bleeding, without any drug-drug interactions with systemic cancer treatment and after an in-depth discussion about the risks and benefits with the patients [bib_ref] Role of direct oral anticoagulants in the treatment of cancer-associated venous thromboembolism:..., Khorana [/bib_ref].
## Low-molecular-weight heparins (lmwhs)
LMWHs were developed in the late seventies and early eighties of the last century. They are obtained from the chemical or enzymatic depolymerization of heparin so that the molecules reach a weight of 4000-5000 Da. The anticoagulant effect is due to the activation of antithrombin, which promotes the inactivation of factor Xa (and to a lesser extent of thrombin). The major advantage of LMWHs was represented by the subcutaneous administration compared to intravenous administration of unfractionated heparin (UFH) with the same antithrombotic efficacy [bib_ref] Lower mortality in cancer patients treated with low-molecular-weight versus standard heparin, Green [/bib_ref]. Their introduction in daily clinical practice started in the second part of the nineties after the publication of four randomized clinical trials between 1996 and 1997 confirming safety and efficacy to treat TEs compared to UFH [bib_ref] A comparison of low-molecular-weight heparin administered primarily at home with unfractionated heparin..., Levine [/bib_ref] [bib_ref] Treatment of venous thrombosis with intravenous unfractionated heparin administered in the hospital..., Koopman [/bib_ref] [bib_ref] Low-molecular-weight heparin in the treatment of patients with venous thromboembolism, Columbus [/bib_ref] [bib_ref] A comparison of low-molecular-weight heparin with unfractionated heparin for acute pulmonary embolism...., Simonneau [/bib_ref].
UFH followed by VKAs were the standard of care before the introduction of LMWHs in patients with malignancy. Prandoni and colleagues showed an approximately double bleeding risk and a threefold thrombotic recurrence risk when patients with malignancy were treated with VKAs compared to non-cancer patients [bib_ref] Recurrent venous thromboembolism and bleeding complications during anticoagulant treatment in patients with..., Prandoni [/bib_ref].
Five randomized trials compared VKAs to LMWHs for the treatment of cancer associated VTE: the CLOT trial [bib_ref] Low-molecular-weight heparin versus a coumarin for the prevention of recurrent venous thromboembolism..., Lee [/bib_ref] , the LITE trial [bib_ref] Long-term low-molecular-weight heparin versus usual care in proximal-vein thrombosis patients with cancer, Hull [/bib_ref] , the CANTHANOX trial [bib_ref] Comparison of low-molecular-weight heparin and warfarin for the secondary prevention of venous..., Meyer [/bib_ref] , the ONCENOX trial [bib_ref] Oncenox Investigators. Secondary prevention of venous thromboembolic events in patients with active..., Deitcher [/bib_ref] and the CATCH trial [bib_ref] Warfarin for Treatment of Acute Venous Thromboembolism in Patients With Active Cancer:..., Lee [/bib_ref]. The first randomized clinical study (CANTHANOX trial) included 146 patients comparing warfarin versus enoxaparin. It showed that more patients treated with warfarin (21.1%) versus patients treated with LMWHs (10.5%) developed the complications defined as the primary combined endpoints at three months, namely recurrent VTE (4% versus 2.8%) or major bleeding (16% versus 7%). Of note, in this trial there were more bleedings than recurrent VTE. Due to slow recruitment, the study was interrupted early. The consequent lack of power of the study did not reveal a significant difference in the two groups [bib_ref] Comparison of low-molecular-weight heparin and warfarin for the secondary prevention of venous..., Meyer [/bib_ref].
The most important study which defined LMWHs as the standard of care was the randomized CLOT trial that included 672 patients with cancer being diagnosed six months prior to inclusion (squamous and basal cell carcinoma excluded) comparing dalteparin versus warfarin. LMWHs were superior to warfarin permitting the reduction of the risk of recurrent VTE (9% versus 17% at six months) without an increasing in bleeding events or in mortality [bib_ref] Low-molecular-weight heparin versus a coumarin for the prevention of recurrent venous thromboembolism..., Lee [/bib_ref].
The other LMWHs enoxaparin (ONCENOX trials including 102 patients [bib_ref] Oncenox Investigators. Secondary prevention of venous thromboembolic events in patients with active..., Deitcher [/bib_ref] , and the CANTHANOX trials including 146 patients [bib_ref] Comparison of low-molecular-weight heparin and warfarin for the secondary prevention of venous..., Meyer [/bib_ref] and tinzaparin (LITE trials including 200 patients [bib_ref] Long-term low-molecular-weight heparin versus usual care in proximal-vein thrombosis patients with cancer, Hull [/bib_ref] , and CATCH trial including 900 patients [bib_ref] Warfarin for Treatment of Acute Venous Thromboembolism in Patients With Active Cancer:..., Lee [/bib_ref] , were compared to warfarin, as well. Of these trials, only the results of LITE study were able to show a statistically significant reduction on recurrent VTE with similar rates of bleeding and mortality at 12 months. In fact, even if the much bigger CATCH study showed a reduction of all VTE recurrence at six months, the difference was not statistically significant. However, the results of the CATCH study showed a significant reduction of recurrent proximal DVT under LMWHs. Therefore, the CATCH study is seen as confirming current guidelines.
A recent meta-analysis of 15 randomized controlled trials published in 2018 by Hakoum and colleagues confirmed the efficacy of LMWHs as the treatment of choice for the initial treatment of venous thromboembolism in patients with cancer [bib_ref] Anticoagulation for the initial treatment of venous thromboembolism in people with cancer, Hakoum [/bib_ref]. However, LMWHs are far from being the ideal treatment of CAT taking into consideration the peculiar features of these patients. LMWH treatment requires daily injections, the dose must be weight-adjusted and withhold in case of severe thrombocytopenia. Moreover, management of patients whose renal function is inferior to 30 mL/min is difficult and LMWH treatment can be complicated by heparin-induced thrombocytopenia (HIT).
## General aspects of doacs and concerns in cancer patients
The currently available DOACs were approved during the first and second decades of the 21st century. Three molecules (apixaban, edoxaban and rivaroxaban) are targeting the activated factor X and one (dabigatran) is directed against the activated factor II, thrombin. They are employed to prevent and to treat TEs (deep vein thrombosis (DVT) or acute pulmonary embolism (PE)) or to prevent stroke and a systemic embolization in cases of non-valvular atrial fibrillation. Hemodynamically unstable patients, pregnant women, patients with extensive DVT and patients with severe renal impairment should not be treated with these agents. The most important advantages are represented by the absence of regular monitoring, the oral administration and a lower bleeding risk.
Their efficacy and safety in treating and preventing VTE were proved by randomized non-inferiority designed trials that compared the newer molecule with the standard treatment (LMWHs followed by VKAs). We briefly resume the most important aspects of these trials as they provided the first basic information on the utilization of DOACs in the subgroup of cancer patients.
Rivaroxaban showed a non-inferiority profile compared to VKAs with a similar rate of relapse of VTE and of bleeding in a total of 8281 patients included in two trials (3449 patients in the EINSTEIN-DVT trial and 4832 in the EINSTEIN-PE trial) [bib_ref] Oral rivaroxaban for symptomatic venous thromboembolism, Investigators [/bib_ref] [bib_ref] Oral rivaroxaban for the treatment of symptomatic pulmonary embolism, Investigators [/bib_ref].
Apixaban was non-inferior compared to the standard of treatment in the rate of recurrence and in mortality, and was superior in the rates of bleeding in a total of 5395 patients enrolled in the AMPLIFY trial [bib_ref] Oral apixaban for the treatment of acute venous thromboembolism, Agnelli [/bib_ref].
Edoxaban was able to demonstrate similar results than warfarin in a total of 4118 patients enrolled in the trial of the Hokusai-VTE investigators [bib_ref] Edoxaban versus warfarin for the treatment of symptomatic venous thromboembolism, Hokusai [/bib_ref]. Dabigatran was non-inferior in the capacity of preventing recurrence of VTE compared to the standard of treatment and in mortality as well in bleeding events. This was demonstrated in the acute VTE trials (RECOVER I) which included 2539 patients [bib_ref] RE-COVER Study Group. Dabigatran versus warfarin in the treatment of acute venous..., Schulman [/bib_ref] , in the 2589 enrolled patients of the acute VTE trials (RECOVER II) [bib_ref] Treatment of acute venous thromboembolism with dabigatran or warfarin and pooled analysis, Schulman [/bib_ref] , and in the pooled analysis of RE-COVER I and II [bib_ref] Treatment of acute pulmonary embolism with dabigatran versus warfarin. A pooled analysis..., Goldhaber [/bib_ref].
It is important to consider that these drugs have some limitations and some concerns arise, especially in the population of cancer patients. The management of bleeding in such patients is still an issue, not only because antidotes are limited. Patients with cancer often experience a modification of thrombocyte count mainly, but not only, due to treatments. LMWHs are usually given at full dose when platelet counts are >50 × 10 9 /L. With the exception of the AMPLIFY trial, which specified a platelet count of >100 × 10 9 /L as inclusion criterion, no precise cutoff was defined in the other studies. A platelet count >50 × 10 9 /L is generally considered adequate for therapeutic anticoagulation. Evidence of utilization of DOACs with lower platelets count is lacking.
Drug interactions with DOACs are of cardinal importance and clinicians are to be aware of this problem. The metabolism of DOACs relies on P-glycoprotein transport and on CYP3A4 pathway. Drugs that modify these metabolisms should be avoided. Some examples of agents inducing these pathways are dexamethasone, vinblastine, doxorubicin and paclitaxel. Drugs inhibiting these pathways are cyclosporine, tamoxifen, and tyrosine kinase inhibitors [bib_ref] Treatment of cancer-associated thrombosis, Lee [/bib_ref] [bib_ref] Management of challenging cases of patients with cancer-associated thrombosis including recurrent thrombosis..., Carrier [/bib_ref].
Renal and hepatic function influence the choice of the anticoagulant treatment. With the exception of apixaban-AMPLIFY trial excluded patients whose creatinine clearance was <25 mL/min-a minimal creatinine clearance of 30 mL/min is recommended to consider a treatment with DOACs. The same is true for the treatment with LMWHs. Even if not precisely defined-values of ASAT/ALAT >2 times the upper limit of normal or a bilirubin >1.5 times the upper limit of normal were considered as exclusion criteria in the main trials-a significant modification of liver function represents a contra-indication to DOACs.
Oral intake can be a concern for cancer patients. Rivaroxaban has to be administered with food and to ensure an adequate bioavailability has to be administered to the stomach via PEG tube, when required [bib_ref] Effect of major gastrointestinal tract surgery on the absorption and efficacy of..., Hakeam [/bib_ref]. Apixaban can be employed independently of food intake, as well as via PEG. Edoxaban and dabigatran are to be administered as an intact capsule, the first one because of limited data and the second because of an increased bioavailability when the capsule is removed. Both are not recommended for utilization via PEG.
Finally, patients affected by cancer often suffer from gastro-intestinal alterations (i.e., lack of appetite, nausea, vomiting, diarrhea, mucositis). Therefore, not only the simple swallowing of a pill can be problematic, but these patients are also at an increased risk of gastro-intestinal bleeding and at a higher risk of a modified absorption or clearance (e.g., diarrhea which results in a diminished bioavailability) of DOACs [bib_ref] Management of challenging cases of patients with cancer-associated thrombosis including recurrent thrombosis..., Carrier [/bib_ref] [bib_ref] How I treat with anticoagulants in 2012: New and old anticoagulants, and..., Schulman [/bib_ref]. A recently published review by Barr and Epps on the managing of DOACs nicely resumes the correct strategies to minimize errors by prescribing these drugs [bib_ref] Direct oral anticoagulants: A review of common medication errors, Barr [/bib_ref].
## Evidence of utilization of doacs in cancer patients
In comparison to the data on treatment of VTE with LMWHs in cancer patients, evidence on the use of DOACs in this patient's group is limited.
# Subgroup analysis
The only available data until 2017 were obtained from subgroups analysis of cancer patients of the main phase III randomized controlled trials comparing DOACs with VKAs in acute VTE. Altogether, these data were retrieved from about 1500 patients enrolled in such studies as having a known malignancy at baseline, which represent 6% of the patients treated with DOACs [bib_ref] Management of venous thromboembolism in cancer patients and the role of the..., Wharin [/bib_ref] [bib_ref] Direct oral anticoagulants compared with vitamin K antagonists for acute venous thromboembolism:..., Van Es [/bib_ref]. Consequently, because of the small sample size and the exploratory nature of the subgroups analysis, these data did not permit any reliable conclusions but offered a primary evaluation of the efficacy and safety.
In the subgroup analysis, 655 patients were randomly assigned to receive rivaroxaban versus VKAs (EINSTEIN-DVT and EINSTEIN-PE trials) [bib_ref] Oral rivaroxaban for symptomatic venous thromboembolism, Investigators [/bib_ref] [bib_ref] Oral rivaroxaban for the treatment of symptomatic pulmonary embolism, Investigators [/bib_ref] [bib_ref] Oral rivaroxaban versus enoxaparin with vitamin K antagonist for the treatment of..., Prins [/bib_ref] , 771 patients assigned to receive edoxaban versus VKAs (Hokusai-VTE) [bib_ref] Edoxaban versus warfarin for the treatment of symptomatic venous thromboembolism, Hokusai [/bib_ref] [bib_ref] Edoxaban for venous thromboembolism in patients with cancer: Results from a non-inferiority..., Raskob [/bib_ref] , 335 patients assigned to receive dabigatran versus VKAs [bib_ref] RE-COVER Study Group. Dabigatran versus warfarin in the treatment of acute venous..., Schulman [/bib_ref] [bib_ref] Extended use of dabigatran, warfarin, or placebo in venous thromboembolism, Schulman [/bib_ref] [bib_ref] Treatment with dabigatran or warfarin in patients with venous thromboembolism and cancer, Schulman [/bib_ref] , and 159 patients assigned to receive apixaban versus VKAs [bib_ref] Oral apixaban for the treatment of acute venous thromboembolism, Agnelli [/bib_ref] [bib_ref] Oral apixaban for the treatment of venous thromboembolism in cancer patients: Results..., Agnelli [/bib_ref]. We focus our attention on the subgroup of patients who presented an active cancer at enrolment or whose cancer was diagnosed during the study, in an attempt to consider a patient population as homogeneous as possible, excluding those patients with a previous history of cancer.
## Rivaroxaban
In the pooled rivaroxaban subgroup analysis [bib_ref] Oral rivaroxaban versus enoxaparin with vitamin K antagonist for the treatment of..., Prins [/bib_ref] , a total of 655 patients (8% of the 8281 included in EINSTEIN-DVT and EINSTEIN-PE trials) presented any active cancer. Of these, 462 (6%) had an active cancer at study inclusion and 193 (2%) were diagnosed during the study. Patients with a history of cancer (defined as not meeting the definition characteristic of these first two subgroups) were 469 (6%) and 7157 patients (86%) were never diagnosed with cancer.
Diagnosis of cancer or treatment within six months before inclusion, recurrent or metastatic cancer, including basal-cell or squamous-cell carcinoma of the skin, were the criteria defining active cancer at study inclusion. A new diagnosis of cancer after the randomization or a recurrence during the study defined patients with active cancer during the study. In the first subgroup (active cancer at study inclusion), rates of VTE recurrence was 2% in patients on rivaroxaban and 4% in patients on VKAs. Major bleeding occurred in the same proportion as VTE recurrence (2% under rivaroxaban and 4% on VKAs). In the second subgroup (cancer diagnosed during the study), rate of VTE recurrence were 10% in patients treated with rivaroxaban and 12% in patients treated with VKAs. Major bleeding occurred in 3% on rivaroxaban and in 7% on VKAs. The data suggested that rivaroxaban could be considered as an alternative of VKAs in patients with CAT. In all three subgroups, in fact, the conclusions for efficacy and safety did not significantly differ in patients receiving rivaroxaban or VKAs.
## Edoxaban
In the edoxaban subgroup analysis [bib_ref] Edoxaban for venous thromboembolism in patients with cancer: Results from a non-inferiority..., Raskob [/bib_ref] , 771 patients (9.4%) of the 8240 patients in the modified intention-to-treat and safety analysis of the Hokusai-VTE trial were defined as having had a history of cancer at study inclusion. Unlike rivaroxaban, patients were treated with an initial parenteral anticoagulation either with unfractionated heparin (UFH) or LMWHs for ≥5 days.
Patients were defined as having active cancer based on the clinical judgement of the local investigator. Furthermore, a review of all patients with a history of cancer were performed by an independent physician without knowledge of treatment group and patients. The data was collected in a post-hoc analysis. In this case, the definition of cancer was more accurate and defined precisely as either a presence of solid measurable cancer except non-melanoma skin cancer or the presence of a not in remission haematological malignancy. In the analysis of the patients with any history of cancer (n = 771), rates of VTE recurrence were 4% in patients on edoxaban and 7% in patients on VKAs. Rates of major bleeding were similar (3% of patients who received edoxaban and 3% of patients on warfarin). Clinically relevant bleeding (major and non-major) occurred in 12% in the edoxaban group and in 19% in the warfarin group.
In the pre-specified analysis of patients with active cancer (n = 208), rate of VTE recurrence were the same as in the groups with any history of cancer (4% for edoxaban and 7% for VKAs). Clinically relevant bleeding occurred at a higher rate in both arms, 18% for edoxaban and 25% for VKAs, respectively. In the group of patients diagnosed with cancer during the study (n = 175), rate of VTE recurrence were 17% (edoxaban) and 20% (VKAs). Similarly to rivaroxaban, these data suggested that edoxaban could be considered as an alternative of VKAs in patients with CAT as well.
## Dabigatran
In the dabigatran subgroup analysis 335 patients were included. This corresponds to 6.6% of the 5107 patients of the entire study population (RECOVER I/II trials) [bib_ref] Treatment with dabigatran or warfarin in patients with venous thromboembolism and cancer, Schulman [/bib_ref]. Active cancer at inclusion (n = 221) was defined as diagnosed within 5 years, except basal-or squamous-cell carcinoma of the skin, with any treatment within this period or any recurrent/metastatic disease. The remaining participants (n = 114) were patients with a diagnosis of cancer made during the study. Like edoxaban, an initial parenteral treatment with UFH, LMWHs or fondaparinux were conducted for ≥5 days.
In the subgroup of patients with active cancer, rate of VTE recurrence was 3.5% for dabigatran and 4.7% for VKAs. Major bleeding occurred in 3.8% on dabigatran and in 3.0% on VKAs. In the subgroup of patients diagnosed during the study, the rate of VTE recurrence was 8.5% for dabigatran and 13% for VKAs. The risk of major bleeding in this subgroup was 3.7% for dabigatran and 7.7% for VKAs. In the latter group, the rate of recurrence was higher than in the group of patients with active cancer at inclusion (hazard ratio (HR) 2.6, 95% confidence interval (CI) 1.1-6.2). In terms of efficacy and safety, there was no difference in patients receiving dabigatran or VKAs.
## Apixaban
Of the 5395 patients included in the AMPLIFY trial [bib_ref] Oral apixaban for the treatment of acute venous thromboembolism, Agnelli [/bib_ref] , 169 patients (3.1%) had an active cancer, whereas 365 patients (6.8%) had a history of cancer in the subgroup analysis [bib_ref] Oral apixaban for the treatment of venous thromboembolism in cancer patients: Results..., Agnelli [/bib_ref]. Active cancer was defined as diagnosed or treated within the previous 6 months of inclusion. Patients with a history of cancer were defined as those with a diagnosis >6 months previous enrollment without receiving any treatment. It is to be noted that 25 patients diagnosed with cancer during the study were included in the cohort of the remaining 4861 patients (90.1%) without any active cancer nor a history of cancer. These cancers were identified based on the analysis of adverse event reports.
In the subgroup of patients with active cancer at study inclusion, the primary efficacy outcome (recurrent symptomatic VTE or VTE-related death) occurred in 3.7% for apixaban and 6.4% for VKAs. Major bleeding occurred in this subgroup in 2.3% under apixaban and 5.0% under VKAs. These findings suggest that apixaban is effective as VKAs in preventing VTE-recurrence without any significative difference in the safety profile.
# Meta-analysis
The data of at least six (semi-)systematic reviews and meta-analysis and of all the phase 3 data published between 2014 and 2015 illustrate that DOACs present a similar efficacy and safety profile as VKAs [bib_ref] Direct oral anticoagulants compared with vitamin K antagonists for acute venous thromboembolism:..., Van Es [/bib_ref] [bib_ref] Meta-analysis of the efficacy and safety of new oral anticoagulants in patients..., Van Der Hulle [/bib_ref] [bib_ref] Efficacy and safety of anticoagulant therapy for the treatment of acute cancer-associated..., Carrier [/bib_ref] [bib_ref] Non-vitamin K antagonist oral anticoagulants and the treatment of venous thromboembolism in..., Larsen [/bib_ref] [bib_ref] Treatment of venous thromboembolism in patients with cancer: A network meta-analysis comparing..., Posch [/bib_ref]. Altogether, the findings of DOACs issued from the four subgroups analysis and from their meta-analysis, showed at least a similar trend in the capacity of preventing recurrent VTE with a similar safety profile as VKAs in the treatment of CAT [bib_ref] Management of venous thromboembolism in cancer patients and the role of the..., Wharin [/bib_ref] [bib_ref] Direct oral anticoagulants compared with vitamin K antagonists for acute venous thromboembolism:..., Van Es [/bib_ref] [bib_ref] Treatment of venous thromboembolism in patients with cancer: A network meta-analysis comparing..., Posch [/bib_ref].
However, the subgroup analysis had many important limitations. First, the comparators of DOACs were VKAs, which are less effective in preventing recurrent VTE than LMWHs. Second, in the phase 3 clinical trials only 6% of patients were suffering from active cancer [bib_ref] Direct oral anticoagulants compared with vitamin K antagonists for acute venous thromboembolism:..., Van Es [/bib_ref]. Third, there is also a paucity of data of the type of cancer, the staging of disease and the oncologic treatment in the different subgroup analysis. Forth, the state of active cancer at baseline was not well defined and differed from the trials evaluating LMWHs for CAT [bib_ref] Treatment of venous thromboembolism in patients with cancer: A network meta-analysis comparing..., Posch [/bib_ref]. Therefore, patients with aggressive cancer (with end-organ dysfunction or reduced life expectancy) were excluded limiting the representability of patients at higher risk for cancer-associated thrombosis (roughly, 30% were under chemotherapy and only 15% to 30% were metastatic). For these reasons, based on the data coming from highly selected cancer population (subgroup analysis and their meta-analysis), it is not possible to draw firm conclusions on the safe utilisation of DOACs in cancer patients.
## Observational studies
Several observational studies describe the use of DOACs in cancer patients [bib_ref] Oral rivaroxaban versus subcutaneous low molecular weight heparin treatment for venous thromboembolism..., Seo [/bib_ref] [bib_ref] Rivaroxaban shows promise as effective therapy for cancer patients with venous thromboembolic..., Theberge [/bib_ref] [bib_ref] Comparative effectiveness and safety of direct oral anticoagulants (DOACs) versus conventional anticoagulation..., Ross [/bib_ref] [bib_ref] Safe and effective use of rivaroxaban for treatment of cancer-associated venous thromboembolic..., Mantha [/bib_ref] [bib_ref] Effectiveness and safety of anticoagulants for the treatment of venous thromboembolism in..., Streiff [/bib_ref] [bib_ref] Healthcare resource use in XALIA: A subgroup analysis of a non-interventional study..., Mantovani [/bib_ref] and are summarized in a recently published systematic review [bib_ref] Direct oral anticoagulant (DOAC) versus low-molecular-weight heparin (LMWH) for treatment of cancer..., Li [/bib_ref]. The most frequently used treatments in these studies were enoxaparin and rivaroxaban. The duration of the utilization of DOACs was longer than that of LMWHs, possibly due to patient preferences for oral drugs or to lower costs of DOACs.
With the exception of one study [bib_ref] Oral rivaroxaban versus subcutaneous low molecular weight heparin treatment for venous thromboembolism..., Seo [/bib_ref] , the review reports lower rates of recurrent VTE in patients treated with DOACs in comparison to patients treated with LMWHs. In terms of safety (major bleedings and clinically relevant non-major bleedings, CRNMB), the results of the observational studies were heterogeneous. The main limitations of the observational studies are indications and patient selections bias by clinicians.
## Concluded prospective studies
So far, as of December 2018, only two prospective, multicentre, open-label, randomized trials comparing the efficacy and safety of DOACs versus LMWHs in cancer patients with CAT have been fully published (Hokusai VTE Cancer, Select-D), and one has been presented in abstract form (ADAM VTE, [fig_ref] Table 1: Summary of the ongoing or in 2018 concluded clinical trials [/fig_ref].
## Edoxaban: the hokusai vte cancer clinical trial
The Hokusai VTE cancer clinical trial, published in February 2018, compared a DOAC, edoxaban (at least 5 days of LMWHs followed by 60 mg daily), versus dalteparin (200 IU/kg daily for one month followed by a reduction at 150 IU/kg daily) [bib_ref] Edoxaban for the Treatment of Cancer-Associated Venous Thromboembolism, Raskob [/bib_ref]. Patients were treated for at least 6 months and up to 12 months.
In this non-inferiority trial, a total of 1050 patients, from 114 centers in 13 countries, were included from July 2015 to December 2016. With the exclusion of basal-or squamous-cell skin cancer, patients had to be diagnosed in the previous two years or had to present with an active cancer. Defined as having active cancer at diagnosis were patients with any treatment for malignancy in the previous six months, or a cancer not in complete remission (hematologic neoplasm), or with recurrent or metastatic disease. Patients were randomized to be treated with either edoxaban or dalteparin in a 1:1 ratio.
The primary outcome was a composite of recurrent VTE or major bleeding in the first 12 months after randomization. Major bleeding was defined according to the criteria of the International Society on Thrombosis and Haemostasis (ISTH) [bib_ref] Subcommittee on Control of Anticoagulation of the Scientific; Standardization Committee of the..., Schulman [/bib_ref]. The primary outcome occurred in 12.8% of patients in the edoxaban group and in 13.5% of patients in the LMWHs group (HR 0.97, 95% CI 0.70-1.36), confirming the non-inferiority of edoxaban. Rate of recurrent VTE was 7.9% in the edoxaban arm and 11.3% in the dalteparin arm (HR 0.71, 95% CI 0.48-1.06), again without a statistically significant difference. In contrast, the rate of major bleeding was significant higher, 6.9%, for the edoxaban arm and 4.0% for the dalteparin arm (HR 1.77, 95% CI 1.03-3.04). However, the higher rate of bleeding was mainly due to an upper gastrointestinal origin who occurred in patients with gastrointestinal cancer at enrollment. The rate of severe major bleedings (considered as a clinical emergency) was the same in the two groups (2.3%). Intracranial bleeding occurred in two patients treated with edoxaban and in four patients treated with dalteparin. Two patients died of major bleeding in the dalteparin group. There were no fatal VTE in both arms. Even if the rate of CRNMB, defined according to the criteria of the ISTH [bib_ref] Subcommittee on Control of Anticoagulation of the Scientific; Standardization Committee of the..., Schulman [/bib_ref] , was higher under edoxaban (14.6%) in comparison to dalteparin (11.1%), this was not statistically significant. There were no differences in both groups in terms of overall survival. In summary, edoxaban was non-inferior to dalteparin in the composite outcome of recurrent venous thromboembolism or major bleeding.
## Rivaroxaban: the select-d trial
Data on the efficacy and safety on the use of rivaroxaban in cancer patients was published in July 2018 in the SELECT-D trial, a pilot trial conducted in the United Kingdom. Recruited between September 2013 and December 2016, the study reported 406 patients with active cancer at inclusion who presented not only symptomatic or incidental PE or symptomatic lower-extremity proximal DVT, but all locations of thrombotic were considered. Patients were randomly assigned to receive either rivaroxaban (15 mg twice daily for 21 days followed by 20 mg daily) or dalteparin (200 IU/kg daily for the first 30 days, followed by 150 IU/kg) for a total of six months. Active cancer was defined in the same way as in the Hokusai VTE cancer study. Both arms were well balanced (58% of patients with metastatic disease in both arms) and all locations of thrombotic events were considered (not only DVT or PE).
The main objective of this study was to assess VTE recurrence and safety (rates of major bleedings and CRNMB) in the first six months of treatment. The second objective of the study, which was to assess the treatment duration beyond six months, remained unanswered because of low recruitment at the second planned randomization (lower assignment due to high mortality rate and clinician's decision).
The cumulative VTE recurrence rate was 11% in patients receiving dalteparin and 4% in patients receiving rivaroxaban (HR 0.43, 95% CI 0.19-0.99) at six months. Concerning safety, the cumulative major bleeds rate at six months was 6% in the rivaroxaban arm and 4% in the dalteparin arm (HR 1.83, 95% CI 0.68-4.96). There were no central nervous system (CNS) bleeds and most of major bleeding was of gastro-intestinal origin. In particular, patients with esophageal or gastroesophageal cancer had more bleeds under rivaroxaban than with dalteparin (four of 11 (36%) versus one of 19 (11%)). The cumulative rate of CRNMB at six months was 13% for rivaroxaban compared to 4% for dalteparin (HR 3.76, 95% CI 1. . Most of CRNMB had a gastrointestinal or urologic origin. One fatal PE and one fatal major bleed occurred in each arm. No statistically significant differences of overall survival were noted in either the two arms at six months (70% for dalteparin and 75% for rivaroxaban). To conclude, the recurrence rate of VTE was lower under rivaroxaban, at the cost, however, of an increased rate of CRNMB.
## Meta-analysis of the prospective studies
A meta-analysis of these two prospective trials confirmed the conclusions of both studies. At 6 months, patients with cancer treated with DOACs had a lower recurrence rate of VTE (42/725) compared to patients treated with LMWHs (64/727) (risk ratio [RR] 0.65, 95% CI 0.42-1.01). On the other hand, patients under DOACs presented a higher rate of major bleeding (40/725) in comparison to patients under LMWHs (23/727) (RR 1.74, 95% CI 1.05-2.88). Moreover, patients treated with DOACs showed a higher CRNMB rate (RR 2.31, 95% CI 0.85-6.28) [bib_ref] Direct oral anticoagulant (DOAC) versus low-molecular-weight heparin (LMWH) for treatment of cancer..., Li [/bib_ref]. At least three systematic reviews and network meta-analysis including all clinical trials, namely comprising the two prospective studies, were published in 2018 [bib_ref] Anticoagulation for the Treatment of Cancer-Associated Thrombosis: A Systematic Review and Network..., Sobieraj [/bib_ref] [bib_ref] Efficacy and safety of anticoagulant agents in patients with venous thromboembolism and..., Vedovati [/bib_ref] [bib_ref] Direct oral anticoagulants for the treatment of venous thromboembolism in patients with..., Yami [/bib_ref]. According to these results DOACs were superior in decreasing VTE recurrence, but an increased risk of major bleeding or CRNMB could not be excluded. [fig_ref] Table 1: Summary of the ongoing or in 2018 concluded clinical trials [/fig_ref] summarizes ongoing trials and two in 2018 completed studies, on the utilization of DOACSs in cancer patients diagnosed with VTE, based on a search of the Website https://clinicaltrials. gov . Most of the studies are conducted with rivaroxaban, only two with apixaban and dabigatran respectively. Evaluating the role of rivaroxaban in the prevention and treatment of CAT is the goal of an international collaboration-the Cancer Associated thrombosis expLoring soLutions for patients through Treatment and prevention with rivaroxaban, abbreviated CALLISTO program. The studies CASTA-DIVA and COSIMO belong to this program. A trial that will possibly reflect the real-life situation, comprises all four DOACs in comparison to LMWHs.
## Ongoing or in 2018 concluded clinical trials
## Anticancer effect
## Anticancer effect of heparins
A potential anti-tumoral activity of unfractionated heparin has been suspected for more than 60 years [bib_ref] Two cases of neoplasms not justified for classical therapy, treated with intravenous..., Albert-Weil [/bib_ref]. The overall survival was beneficially influenced by UFH according to the data, published in 1998, obtained from some retrospective analyses of cancer patients included in randomized clinical trials comparing thromboprophylaxis with UFH versus no prophylaxis [bib_ref] Heparin and cancer, Zacharski [/bib_ref]. Two studies published one year later gave contradictory results. The first one, published by Lebeau and colleagues, found a statistically significant amelioration of survival in patients diagnosed with small cell carcinoma of the lung (SCLC) who received UFH (therapeutically dosed) in combination with chemotherapy versus without UFH in a prospective randomized multicentre trial [bib_ref] Subcutaneous heparin treatment increases survival in small cell lung cancer, Lebeau [/bib_ref]. The second one, a systematic review of clinical studies edited by Smorenburg and colleagues, found no solid evidence of a survival advantage in patients with cancer without VTE who were treated with UFH versus no treatment or placebo [bib_ref] The effects of unfractionated heparin on survival in patients with malignancy-A systematic..., Smorenburg [/bib_ref]. The same year, a meta-analysis evaluating the efficacy of LMWHs versus UFH in cancer patients with VTE, showed a lower mortality in patients initially treated with LMWHs [bib_ref] Do heparins do more than just treat thrombosis? The influence of heparins..., Hettiarachchi [/bib_ref]. Yet seven years earlier, in 1992, Prandoni and colleagues [bib_ref] Comparison of subcutaneous low-molecular-weight heparin with intravenous standard heparin in proximal deep-vein..., Prandoni [/bib_ref] , as well as Green and colleagues [bib_ref] Lower mortality in cancer patients treated with low-molecular-weight versus standard heparin, Green [/bib_ref] , provided evidence of a survival advantage of LMWHs compared to UFH in cancer A potential survival advantage of anticoagulation in cancer patients was more recently reviewed [bib_ref] The effect of low-molecular-weight heparin on cancer survival. A systematic review and..., Lazo-Langner [/bib_ref] [bib_ref] A meta-analysis and systematic review of the efficacy and safety of anticoagulants..., Kuderer [/bib_ref] [bib_ref] Parenteral anticoagulation in patients with cancer who have no therapeutic or prophylactic..., Akl [/bib_ref]. The positive effect on survival relies not only on the anticoagulation action but also on a putative direct antineoplastic effect, in particular of LMWHs. For instance, a post-hoc analysis of the CLOT trial showed a mortality decrease at twelve months in patients without metastatic disease treated with dalteparin (20% versus 36%, HR 0.50, 95% CI 0.27-0.95) [bib_ref] Randomized comparison of low molecular weight heparin and coumarin derivatives on the..., Lee [/bib_ref].
Many data, frequently very interesting and convincing, obtained from experimental models demonstrated a potential anticancer effect of heparins through multiple mechanisms [bib_ref] The complex effects of heparins on cancer progression and metastasis in experimental..., Smorenburg [/bib_ref] [bib_ref] Anti-cancer properties of low-molecular-weight heparin: Preclinical evidence, Mousa [/bib_ref] [bib_ref] Role of endogenous thrombin in tumor implantation, seeding, and spontaneous metastasis, Hu [/bib_ref]. Several biological mechanisms, which could explain the direct antitumoral activity of LMWHs were elucidated. They comprise: 1. inhibition of angiogenesis (by interfering with TF-FVIIa activation and with thrombin), 2. inhibition of P-and L-selectin-mediated cell adhesion, 3. inhibition of tumour invasion (by lowering heparanase activity), 4. inhibition of thrombin and fibrin production, 5. modulation of the immune system, 6. interference with tumor cell glycosaminoglycans, 7. induction of apoptosis, 8. some evidence of potential interference with cancer cell proliferation [bib_ref] Biological and clinical aspects of anticancer effects of antithrombotics, Falanga [/bib_ref] [bib_ref] The effects of low molecular weight heparins on venous thromboembolism and survival..., Lee [/bib_ref] [bib_ref] Low molecular weight heparin suppresses receptor for advanced glycation end products-mediated expression..., Takeuchi [/bib_ref] [bib_ref] In vitro study of low molecular weight heparin effect on cell growth..., Balzarotti [/bib_ref].
From a clinical point of view, alongside studies showing the efficacy of LMWHs for the treatment of CAT [bib_ref] Low-molecular-weight heparin versus a coumarin for the prevention of recurrent venous thromboembolism..., Lee [/bib_ref] , analyses evaluating the effect of LMWHs in a cancer population without VTE have also been conducted [bib_ref] Low molecular weight heparin, therapy with dalteparin, and survival in advanced cancer:..., Kakkar [/bib_ref] [bib_ref] A randomized clinical trial of combination chemotherapy with and without low-molecular-weight heparin..., Altinbas [/bib_ref]. The first two studies which showed a statistically significant improvement of overall survival in cancer patients treated with LMWHs were published in 2004 by Altinbas and colleagues [bib_ref] A randomized clinical trial of combination chemotherapy with and without low-molecular-weight heparin..., Altinbas [/bib_ref] , and by Kakkar and colleagues (FAMOUS study) [bib_ref] Low molecular weight heparin, therapy with dalteparin, and survival in advanced cancer:..., Kakkar [/bib_ref]. Even if in the latter study the survival advantage was not statistically significant, a post hoc analysis in the subgroups of patients with a better prognosis at enrolment who received LMWHs had a statistically significant improvement in survival. In the following years at least nine other studies appeared showing conflicting results (i.e., MALT, PROTECHT, SAVE-ONCO, PRODIGE, TOPIC-I and -II, INPACT and others) [bib_ref] The effect of low molecular weight heparin on survival in patients with..., Klerk [/bib_ref] [bib_ref] Low-molecular-weight heparin in patients with advanced cancer: A phase 3 clinical trial, Sideras [/bib_ref] [bib_ref] Nadroparin for the prevention of thromboembolic events in ambulatory patients with metastatic..., Agnelli [/bib_ref] [bib_ref] PRODIGE: A randomized placebo-controlled trial of dalteparin low-molecular-weight heparin thromboprophylaxis in patients..., Perry [/bib_ref] [bib_ref] Randomized trial of the effect of the low molecular weight heparin nadroparin..., Van Doormaal [/bib_ref] [bib_ref] Semuloparin for thromboprophylaxis in patients receiving chemotherapy for cancer, Agnelli [/bib_ref] [bib_ref] Low-molecular-weight heparin versus placebo for the prevention of venous thromboembolism in metastatic..., Haas [/bib_ref] [bib_ref] Gemcitabine versus gemcitabine plus dalteparin thromboprophylaxis in pancreatic cancer, Maraveyas [/bib_ref] [bib_ref] Adjuvant therapy with bemiparin in patients with limited-stage small cell lung cancer:..., Lecumberri [/bib_ref]. Finally, the results of two randomised multicentre phase III controlled trials addressing the question of a survival advantage of LMWHs on a single cancer type-lung-were published in 2016 and 2018. The first one, the FRAGMATIC trial, did not show an improvement in overall survival [bib_ref] Randomized Phase III Trial of Standard Therapy Plus Low Molecular Weight Heparin..., Macbeth [/bib_ref]. The very recently published (October 2018) study, the TILT trial was not able to demonstrate a beneficial effect of overall survival or cancer recurrence even in early-stage lung cancer [bib_ref] Anti-tumour effect of low molecular weight heparin in localised lung cancer: A..., Meyer [/bib_ref]. All in all, while the results of some clinical trials and meta-analysis recognized a statistically significant effect of LMWHs on survival [bib_ref] The effect of low-molecular-weight heparin on cancer survival. A systematic review and..., Lazo-Langner [/bib_ref] [bib_ref] A meta-analysis and systematic review of the efficacy and safety of anticoagulants..., Kuderer [/bib_ref] [bib_ref] Low molecular weight heparin, therapy with dalteparin, and survival in advanced cancer:..., Kakkar [/bib_ref] [bib_ref] A randomized clinical trial of combination chemotherapy with and without low-molecular-weight heparin..., Altinbas [/bib_ref] [bib_ref] The effect of low molecular weight heparin on survival in patients with..., Klerk [/bib_ref] [bib_ref] Adjuvant therapy with bemiparin in patients with limited-stage small cell lung cancer:..., Lecumberri [/bib_ref] , others found opposite conclusions [bib_ref] Low-molecular-weight heparin in patients with advanced cancer: A phase 3 clinical trial, Sideras [/bib_ref] [bib_ref] Nadroparin for the prevention of thromboembolic events in ambulatory patients with metastatic..., Agnelli [/bib_ref] [bib_ref] PRODIGE: A randomized placebo-controlled trial of dalteparin low-molecular-weight heparin thromboprophylaxis in patients..., Perry [/bib_ref] [bib_ref] Randomized trial of the effect of the low molecular weight heparin nadroparin..., Van Doormaal [/bib_ref] [bib_ref] Semuloparin for thromboprophylaxis in patients receiving chemotherapy for cancer, Agnelli [/bib_ref] [bib_ref] Low-molecular-weight heparin versus placebo for the prevention of venous thromboembolism in metastatic..., Haas [/bib_ref] [bib_ref] Gemcitabine versus gemcitabine plus dalteparin thromboprophylaxis in pancreatic cancer, Maraveyas [/bib_ref] [bib_ref] The effect of low molecular weight heparin on survival in cancer patients:..., Sanford [/bib_ref] [bib_ref] The efficacy and safety of low-molecular-weight heparin use for cancer treatment: A..., Che [/bib_ref]. Many reasons can explain this divergence. Some of them are the heterogeneity of studies design, of type and stage of included cancers, of used chemotherapy regimens and the type, dosing and duration of the treatment of LMWHs.
As a conclusion, despite the clinical evidence and the available experimental data, the two most recent meta-analyses did not observe a straightforward advantage of LMWHs. The first one, a 2014 meta-analysis of nine randomized trials, did not show a survival benefit of anticoagulation with LMWHs (odds ratio (OR) for one-year mortality 0.87, 95% CI 0.70-1.08) [bib_ref] The effect of low molecular weight heparin on survival in cancer patients:..., Sanford [/bib_ref]. The second one, the Cochrane review published the same year, including 15 randomized trials comparing heparin with no intervention or placebo, showed a borderline effect on survival, a decrease in VTE incidence and a rise in minor bleeding [bib_ref] Parenteral anticoagulation in ambulatory patients with cancer, Akl [/bib_ref]. Accordingly, the use of anticoagulation to exert an antineoplastic effect is not supported by evidence coming from these meta-analyses and Cochrane reviews. According to different international guidelines, anticoagulation in absence of VTE is not an option in the attempt to improve survival in cancer patients.
## Doacs beyond the anticoagulation: a potential antineoplastic effect?
In analogy to research looking for a potential antineoplastic role of LMWHs [bib_ref] Anti-cancer properties of low-molecular-weight heparin: Preclinical evidence, Mousa [/bib_ref] [bib_ref] Antimetastatic activities of heparins and modified heparins, Borsig [/bib_ref] , some investigations indicate a potential effect of DOACs in this setting as well. The effects of DOACs beyond their principal role of inhibiting specific steps in the coagulation cascade eventually determining the anticoagulation state were summarized in 2014 [bib_ref] The new direct oral anticoagulants in special indications: Rationale and preliminary data..., Alberio [/bib_ref].
Among many other mechanisms, cancer relies on angiogenesis to develop. Blocking this process, malignancy development could be limited. Angiogenesis is one of the effects of the tight interplay between inflammation and hemostasis [bib_ref] The interactions between inflammation and coagulation, Esmon [/bib_ref] [bib_ref] The coagulant response in sepsis and inflammation, Levi [/bib_ref] , whose molecular link is constituted by protease-activated receptors (PARs).
Inflammation induces a procoagulant state in different ways (e.g., by reducing natural anticoagulants, by activating platelets, by releasing microparticles, by elevating the expression of the tissue factor (TF), and through NETosis) [bib_ref] Neutrophil extracellular traps kill bacteria, Brinkmann [/bib_ref] [bib_ref] NETosis in Cancer-Platelet-Neutrophil Crosstalk Promotes Tumor-Associated Pathology, Olsson [/bib_ref]. Conversely, the inflammation process is influenced by the coagulation cascade. PARs are transmembrane G-protein-coupled receptors that were discovered at the beginning of the nineties of the last century [bib_ref] Van Obberghen-Schilling, E. cDNA cloning and expression of a hamster alpha-thrombin receptor..., Rasmussen [/bib_ref] [bib_ref] Molecular cloning of a functional thrombin receptor reveals a novel proteolytic mechanism..., Vu [/bib_ref] [bib_ref] Protease-activated receptors in hemostasis, thrombosis and vascular biology, Coughlin [/bib_ref]. Fours PARs (PAR 1-4) have been identified and are expressed on cancer cells, on the cell surface of hematopoietic cells (leucocytes and platelets, but not on erythrocytes) and on several other cells (endothelial cells, vascular smooth muscle cells, fibroblasts and dendritic cells). Platelets express only PAR-1 and PAR-4, the first receptor having high affinity and the second having low affinity to the activator. The activation of the PAR-2 receptors induces prolonged secondary signaling. Thrombin acts as a protease on the extracellular domain of the N-terminus of the receptor. The so formed new N-terminal domain binds to a specific region of the extracellular part of the receptor inducing transmembrane signaling through G-proteins leading eventually to cell proliferation, activation and adhesion [bib_ref] Molecular cloning of a functional thrombin receptor reveals a novel proteolytic mechanism..., Vu [/bib_ref].
Thrombin influence many steps of cancer progression and seems to be a key player in this setting [bib_ref] Cancer cells in transit: The vascular interactions of tumor cells, Konstantopoulos [/bib_ref] [bib_ref] Thrombin's central role in angiogenesis and pathophysiological processes, Tsopanoglou [/bib_ref]. Coagulation protease, such as factor Xa and thrombin, by cleaving PARs on cell membranes trigger intracellular signaling events [bib_ref] Factor Xa: At the crossroads between coagulation and signaling in physiology and..., Borensztajn [/bib_ref] [bib_ref] Thrombin signalling and protease-activated receptors, Coughlin [/bib_ref]. Thrombin can activate all types of PARs with the exception of PAR-2. Factor Xa, alone or bound to the complex TF-Factor VIIa, can activate PAR-1 and PAR-2. Thrombin and Factor Xa, through the interaction with PARs, eventually favor angiogenesis, inflammation and tissue fibrosis, thus stimulating tumor progression [fig_ref] Figure 1: Schematic representation of the interaction between hemostasis and inflammation promoting tumor progression [/fig_ref] [bib_ref] Protease-activated receptors (PARs)-Biology and role in cancer invasion and metastasis, Wojtukiewicz [/bib_ref]. Wojtukiewicz and colleagues summarized some of the scientific evidence pointing out the key role of thrombin in the tumor progression in 2017 [bib_ref] Antiplatelet agents for cancer treatment: A real perspective or just an echo..., Wojtukiewicz [/bib_ref].
Data on animal models suggest an anti-angiogenic effect of rivaroxaban at least as powerful as LMWHs [bib_ref] Investigation of the antiangiogenic behaviors of rivaroxaban and low molecular weight heparins, Yavuz [/bib_ref]. Dabigatran is able to block free as well clot-bound thrombin. The latter favors angiogenesis and tissue repair at sites of injury [bib_ref] Targeting factor Xa and thrombin: Impact on coagulation and beyond, Esmon [/bib_ref]. In similar way, evidence from animal models or studies on cell lines on the inhibition of thrombin showed a diminution of tumor progression [bib_ref] Inhibition of the platelet-aggregating activity of two human adenocarcinomas of the colon..., Pearlstein [/bib_ref] [bib_ref] Oral thrombostatin FM19 inhibits prostate cancer, Nieman [/bib_ref] [bib_ref] Use of dabigatran etexilate to reduce breast cancer progression, Defeo [/bib_ref] [bib_ref] Dabigatran antagonizes growth, cell-cycle progression, migration, and endothelial tube formation induced by..., Vianello [/bib_ref] [bib_ref] Thrombin inhibition and cyclophosphamide synergistically block tumor progression and metastasis, Alexander [/bib_ref].
Thus, by inhibiting factor Xa and thrombin, it is theoretically conceivable to exert other effects than merely the anticoagulating one. By interfering with these processes, DOACs could exert an anti-neoplastic action. Whether this theoretic effect can traduce in clinical practice, needs to be investigated. Studies evaluating the risk/benefit ratio of DOACs for primary prevention VTE in intermediate-to-high-risk cancer patients, such as the recently published AVERT trial [bib_ref] Apixaban to prevent venous thromboembolism in patients with cancer, Carrier [/bib_ref] , could possibly contribute to this knowledge.
adhesion [bib_ref] Molecular cloning of a functional thrombin receptor reveals a novel proteolytic mechanism..., Vu [/bib_ref].
Thrombin influence many steps of cancer progression and seems to be a key player in this setting [bib_ref] Cancer cells in transit: The vascular interactions of tumor cells, Konstantopoulos [/bib_ref] [bib_ref] Thrombin's central role in angiogenesis and pathophysiological processes, Tsopanoglou [/bib_ref]. Coagulation protease, such as factor Xa and thrombin, by cleaving PARs on cell membranes trigger intracellular signaling events [bib_ref] Factor Xa: At the crossroads between coagulation and signaling in physiology and..., Borensztajn [/bib_ref] [bib_ref] Thrombin signalling and protease-activated receptors, Coughlin [/bib_ref]. Thrombin can activate all types of PARs with the exception of PAR-2. Factor Xa, alone or bound to the complex TF-Factor VIIa, can activate PAR-1 and PAR-2. Thrombin and Factor Xa, through the interaction with PARs, eventually favor angiogenesis, inflammation and tissue fibrosis, thus stimulating tumor progression [fig_ref] Figure 1: Schematic representation of the interaction between hemostasis and inflammation promoting tumor progression [/fig_ref] [bib_ref] Protease-activated receptors (PARs)-Biology and role in cancer invasion and metastasis, Wojtukiewicz [/bib_ref]. Wojtukiewicz and colleagues summarized some of the scientific evidence pointing out the key role of thrombin in the tumor progression in 2017 [bib_ref] Antiplatelet agents for cancer treatment: A real perspective or just an echo..., Wojtukiewicz [/bib_ref].
# Conclusions
The recommended treatment for cancer-associated venous thromboembolism is based on LMWHs. Increasing clinical evidence places DOACs in a position to represent a real treatment option for CAT. However, their possible benefits have to be carefully balanced with the disadvantages of DOACs in the special setting of CAT.
The main advantages include an ease way of administration, a reduction in treatment costs, an absence of monitoring with the effect of reducing VTE recurrence. Inconveniences with the treatment of DOACs consist of an increased risk of bleeding (e.g., patients with gastro-intestinal or urologic malignancy, chemotherapy-induced thrombocytopenia) and a considerable interaction with other drugs. The latter is particularly sensitive to a rapidly changing landscape of cancer treatment. Reduced renal function and extreme body weights are limiting factors for the use of both LMWHs and DOACs. In these special circumstances, monitoring plasma trough-concentrations of the chosen anticoagulant is a prudent way to warrant safety. The preferences and the values of the patient are also to be included in the decision-making process of defining the type of anticoagulation.
Considering the different mechanism of actions (inhibition of thrombin or factor Xa) and pharmacokinetics aspects (clearance through P-glycoprotein and/or cytochrome P450), a uniformly effect of the class of DOACs cannot be assumed. Randomized clinical trials evaluating efficacy and safety for each DOACs versus LMWHs are therefore indispensable. Results of such studies are eagerly awaited in the next few years. A direct anti-neoplastic effect of DOACs is an interesting field of research that needs to be deepened before drawing conclusions on clinical efficacy.
In summary, the treatment of CAT must be personalized and adapted to the specific clinical condition of the patient. Based on the currently available data for edoxaban and rivaroxaban, DOACs represent an additional therapeutic option for patients with CAT and a low risk of bleeding.
Author Contributions: Conceptualization, F.G. and L.A.; writing-original draft preparation, F.G.; writing-review and editing, F.G., L.A.; supervision, L.A.
Funding: This research received no external funding.
## Conflicts of interest:
The authors declare no conflict of interest.
[fig] Figure 1: Schematic representation of the interaction between hemostasis and inflammation promoting tumor progression. PARs (protease-activated receptors) are expressed on several cells. Thrombin activates PAR-1, PAR-3 and PAR-4. Factor Xa, alone or bound to the tissue factor (TF)-activated factor VII (VIIa) complex, activates PAR-1 and PAR-2. In red: inhibition of either thrombin or factor Xa could potentially inhibit cancer progression. [/fig]
[table] Table 1: Summary of the ongoing or in 2018 concluded clinical trials. [/table]
|
Electronic patient self-Reporting of Adverse-events: Patient Information and aDvice (eRAPID): a randomised controlled trial in systemic cancer treatment
on behalf of the eRAPID systemic treatment work group Abstract Background: eRAPID (electronic patient self-Reporting of Adverse-events: Patient Information and aDvice) is an internet based system for patients to self-report symptoms and side effects (adverse events or AE) of cancer treatments. eRAPID allows AE reporting from home and patient reported data is accessible via Electronic Patient Records (EPR) for use in routine care. The system can generate alerts to clinical teams for severe AE and provides patient advice on managing mild AEs. The overall aims of eRAPID are to improve the safe delivery of cancer treatments, enhance patient care and standardise AE documentation.Methods:The trial is a prospective randomised two-arm parallel group design study with repeated measures and mixed methods. Participants (adult patients with breast cancer on neo-adjuvant or adjuvant chemotherapy, colorectal and gynaecological cancer receiving chemotherapy) are randomised to receive the eRAPID intervention or usual care over 18 weeks of treatment. Participants in the intervention arm receive training in using the eRAPID system to provide routine weekly adverse event reports from home. Hospital staff can access eRAPID reports via the EPR and use the information during consultations or phone calls with patients. Prior to commencing the full trial an internal pilot phase was conducted (N = 87 participants) to assess recruitment procedures, consent and attrition rates, the integrity of the intervention information technology and establish procedures for collecting outcome data. The overall target sample for the trial is N = 504. The primary outcome of the trial is quality of life (FACT-G) with secondary outcomes including health economics (costs to patients and the NHS), process of care (e.g. contacts with the hospital, number of admissions, clinic appointments and changes to treatment/medications) and patient self-efficacy. Outcome data is collected at baseline, 6, 12, 18 weeks and 12 months. The intervention is also being evaluated via end of study interviews with patient participants and clinical staff.
(Continued from previous page) Discussion: The pilot phase was completed in February 2016 and recruitment and attrition rates met criteria for continuing to the full trial. Recruitment recommenced in May 2016 and is planned to continue until December 2017. Overall findings will determine the value of the eRAPID intervention for supporting the care of patients receiving systemic cancer treatment.
Trial registration: Current Controlled Trials ISRCTN88520246. Registered 11 September 2014.
Keywords: Cancer, Adverse events, Patient reported outcome measures (PROMs), Patient reported outcomes (PROs), Electronic patient records, Electronic health records, Internet, Intervention, Self-management, Chemotherapy Background Systemic drug treatments for cancer (chemotherapy, hormonotherapy, biological therapy, targeted agents) are associated with significant adverse events (AEs). An AE is an untoward symptom or disease associated with (but not necessarily causally related to) a medical treatment or intervention AEs may lead to changes in drug dosage, cessation of treatment and can significantly compromise patients' quality of life. Severe AEs can escalate to hospitalisation for potentially life-threatening toxicities: 18% of cancer patients present to emergency services within 14 days of a scheduled hospital visit for symptom management (infection, fever, nausea/vomiting, pain, breathlessness) [bib_ref] Demographics, clinical presentations and outcomes of cancer patients admitted to the emergency..., Bozdemir [/bib_ref] [bib_ref] Analysis of reasons for admission to the emergency department for cancer patients, De Luigi [/bib_ref] [bib_ref] Emergency department waiting times for patients with cancer with febrile neutropenia: a..., Nirenberg [/bib_ref] [bib_ref] Cancer pain as the presenting problem in emergency departments: incidence and related..., Tsai [/bib_ref]. Patients with breast, gastrointestinal, colorectal cancers and those with metastatic disease are amongst those most likely to have emergency admissions [bib_ref] Cancer pain as the presenting problem in emergency departments: incidence and related..., Tsai [/bib_ref] [bib_ref] An audit of acute oncology services: patient experiences of admission procedures and..., Warrington [/bib_ref].
Many patients however, delay seeking care especially out of hours [bib_ref] Emergency department waiting times for patients with cancer with febrile neutropenia: a..., Nirenberg [/bib_ref] [bib_ref] An audit of acute oncology services: patient experiences of admission procedures and..., Warrington [/bib_ref]. This concurs with the findings of a UK enquiry into patient outcome and death (National Confidential Enquiry into Patient Outcome and Death, NCEPOD) which found that of patients dying within 30 days of systemic cancer therapy, 17% delayed seeking advice for over 24 h. AEs are documented consistently by physicians in clinical trials however in routine care recording of AEs by clinicians and reporting by patients is variable and often omitted. It has been recognised for some time that a structured AEs reporting system would be useful to facilitate correct documentation and grading of AE severity to support tailored management. Consequently, the National Cancer Institute (NCI) in the US have developed the Common Terminology Criteria for Adverse Events (CTCAE v 4.0)as a reporting and severity grading system for cancer clinical trials. These have recently been adapted for patients to self-report (NCI-PRO CTCAE) [bib_ref] Validation study of the patient-reported outcomes version of the common terminology criteria..., Dueck [/bib_ref] and these items have concordance with nurse evaluated AE [bib_ref] Patient versus clinician symptom reporting using the National Cancer Institute common terminology..., Basch [/bib_ref] and similar items created for self-report correlate with quality of life measures [bib_ref] Adverse symptom event reporting by patients vs clinicians: relationships with clinical outcomes, Basch [/bib_ref]. The need for routine monitoring of cancer treatment AE is at odds with a health care system relying increasingly on patient self-management and home based care. In order to bridge the gap in service provision to detect, identify and manage AE in cancer patients the Electronic patient self-Reporting of Adverse-events: Patient Information and aDvice (eRAPID): system was developed [bib_ref] Towards safer delivery and monitoring of cancer treatments. Electronic patient self-reporting of..., Ziegler [/bib_ref].
## Patient reported outcome measures (proms)
PROMs have been used in clinical practice to support care of individual patients, recent reviews suggest they improve symptom/function monitoring, physician patient communication and decision making [bib_ref] Provision of feedback on perceived health status to health care professionals: a..., Espallargues [/bib_ref] [bib_ref] The effectiveness of the use of patient-based measures of health in routine..., Greenhalgh [/bib_ref] [bib_ref] Patient participation in the consultation process: a structured review of intervention strategies, Haywood [/bib_ref] [bib_ref] Impact of patient-reported outcome measures on routine practice: a structured review, Marshall [/bib_ref] [bib_ref] Impact of patient-reported outcomes in oncology: a longitudinal analysis of patient-physician communication, Takeuchi [/bib_ref] [bib_ref] Patient reported outcome measures: a model-based classification system for research and clinical..., Valderas [/bib_ref] , can save time during clinic visits and improve the accuracy of symptom reporting [bib_ref] Electronic patient-reported outcome systems in oncology clinical practice, Bennett [/bib_ref]. In the UK the 2008 Darzi reportrecommended that collection of PROMs data should be an essential component of health care evaluationand the Department of Health (DOH) subsequently produced guidelines to aid their implementation . Following this, use of PROMs in the health service is most advanced in England (particularly for performance comparisons) [bib_ref] Patient reported outcome measures could help transform healthcare, Black [/bib_ref]. Two recently published reports by the Independent Cancer Taskforce and NHS England have continued to highlight the need to put PROMs at the centre of strategies to improve patient centred cancer care and quality of life.
## Electronic and mobile reporting technology
Electronic reporting of patient reported outcome measures (PROMs) has proven extremely acceptable to patients in the clinic setting [bib_ref] Measuring quality of life in routine oncology practice improves communication and patient..., Velikova [/bib_ref] [bib_ref] Computer-based quality of life questionnaires may contribute to doctor-patient interactions in oncology, Velikova [/bib_ref] [bib_ref] Patients report improvements in continuity of care when quality of life assessments..., Velikova [/bib_ref]. Examples of successful implementation of electronic symptom reporting in oncology clinical practice include PatientViewpoint [bib_ref] Patient viewpoint: a website for patient-reported outcomes assessment, Snyder [/bib_ref] , the symptom tracking and reporting system (STAR) system for patients to report chemotherapy AE [bib_ref] Patient online self-reporting of toxicity symptoms during chemotherapy, Basch [/bib_ref] and the Tell Us™ [bib_ref] Tell us: a webbased tool for improving communication among patients, families, and..., Dy [/bib_ref] system for advanced cancer patients in hospices undergoing palliative care (all in the U.S.). In Austria the Computer-based Health Evaluation System (CHES) software [bib_ref] Evaluation of electronic patient-reported outcome assessment with cancer patients in the hospital..., Wintner [/bib_ref] has been developed and an interactive online system (ISAAC) is in use in Canada. In the UK the ASyMS mobile phone system is currently being evaluated [bib_ref] Patients' perceptions and experiences of using a mobile phone-based advanced symptom management..., Mccann [/bib_ref]. Electronic patient reported outcome systems have proven very acceptable even for patients coping with extreme symptom burden and reduced quality of life; indeed a mean monthly PROM completion rate of 83% at 34 weeks has been achieved with patients receiving cancer treatment [bib_ref] Feasibility of long-term patient self-reporting of toxicities from home via the internet..., Judson [/bib_ref].
## Erapid development work
The eRAPID research programme was designed to develop and evaluate an online system to support the collection and clinical integration of patients' symptom/AE reports during cancer treatment. It utilises a web-based questionnaire builder system called QTool. QTool Version 1 was originally used in a large prospective study of cancer survivors, recruiting 636 patients in 12 months, 81% of whom completed web-based questionnaires at baseline [bib_ref] Effect of a self-management program on patients with chronic disease, Lorig [/bib_ref] (www.epocs.leeds.ac.uk), confirming the feasibility of web-based patient-reporting and QTool acceptability. Between 2010 and 2013 the eRAPID developmental work was conducted (funded by an National Institute of Health Research grant: Programme Development Grant scheme RP-DG-1209-10,031), which focused on:
1) Developing the electronic platform to allow QTool data to be securely linked to the electronic patient records used by Leeds Teaching Hospitals (see [fig_ref] Figure 1: eRAPID system overview [/fig_ref]. 2) Selection, adaption and evaluation of items for patients to report symptoms and AE resulting in the development of patient-reported AE (PRAE) items based on CTCAE grades [bib_ref] Asking the right questions to get the right answers: using cognitive interviews..., Holch [/bib_ref]. staff interviews, a local audit of care pathways/acute triage processes, mapping the existing chemotherapy pathways for the detection and management of AE and an assessment of patient experience of acute admissions and prospective patient interviews and diaries during chemotherapy to record AEs and costs to patents and services. The latter aimed to develop a questionnaire for health economic analysis [bib_ref] An audit of acute oncology services: patient experiences of admission procedures and..., Warrington [/bib_ref].
This developmental work led to the:
Successful mapping of current systemic treatment pathway, establishing where eRAPID is best placed Identification of staff requiring training to deliver eRAPID Adaptation of a health economic questionnaire for cancer patients receiving treatment
## The erapid intervention
An overview of the eRAPID intervention is described in 1, 2a and b. [fig_ref] Figure 1: eRAPID system overview [/fig_ref] represents the technical components and their integration to support reporting of AEs immediately available in the EPR. The architecture protects patient confidentiality providing security whilst allowing immediate linkage to individual patient records to support care. The intervention consists of the following components:
Patients can log in to QTool (using a unique username and password) to access the eRAPID symptom questionnaire anywhere with internet access (including home or hospital).
For mild/moderate problems information about self-managing these issues are provided via brief instructions in QTool along with hyperlinks to more detailed advice on the eRAPID patient website . Where severe symptoms are reported patients are advised to contact the hospital. The patient reported data is immediately available for staff to view in the individuals' electronic patient records in Leeds Teaching Hospitals NHS Trust (Patient Pathway Manager, PPM). See . Alerts for severe symptom reports are sent directly to staff via email. Clinicians can then log into PPM and view the patients' symptom reports and take appropriate action where needed.
Prior to the start of the current trial the eRAPID system underwent usability testing with N = 14 breast cancer patients receiving adjuvant or neo-adjuvant chemotherapy and relevant staff.
## Hypotheses
We hypothesise the eRAPID intervention has the potential to bring benefit to patients, staff and the NHS in the following ways:
Benefits for patients ○ Earlier symptom detection and improved selfmanagement, timely admissions ○ Improved supportive medication use ○ Appropriate hospital, GP, community contacts ○ Better outcomes (improved symptom control, functioning and quality of life) Benefits for staff ○ Reduce the number of hospital, GP, community contacts ○ Save time spent on enquiring and recording AEs ○ Focus attention during clinical contacts on most important or severe AEs ○ Support decision making in routine care Benefits to the NHS ○ eRAPID provides a cost-effective approach to support patient self-management and reduce hospital and GP contacts
## Study design
This study is a single centre 1:1 allocation prospective randomised two-arm parallel group trial design with repeated measures and mixed methods.
## Patient sample
The study sample includes patients with gynaecological or colorectal cancer requiring chemotherapy, or breast cancer undertaking either neo-adjuvant or adjuvant following systemic treatment pathways at St. James's Institute of Oncology, Leeds, UK.
# Methods
Participants are randomised to either the intervention arm (eRAPID plus usual care) or the control arm (usual care). See [fig_ref] Figure 3: Trial flow diagram [/fig_ref] for the trial flow diagram. Participants are on the study for an 18 week period from the start of chemotherapy. A subset of participants (where feasible within the funding timeframe) will also be assessed at a 12 month time point to examine any potential longer term impact of the intervention on quality of life and clinical processes.
## Usual care
Includes an initial consultation with an oncologist to decide whether to commence systemic treatment. Patients are provided with verbal and written information on treatment benefits and expected AEs, and are given instructions on how to contact the hospital. They have a nurse assessment before starting their treatment. During treatment patients are routinely assessed in clinics for AE and to prescribe their next cycle of treatment by an oncologist, Clinical Nurse Specialist (CNS) or staff grade doctor. Depending on AE experienced by the patient, treatment doses can be reduced, and/or supportive medications changed (e.g. anti-sickness drugs, anti-diarrhoea drugs). When at home if patient has a serious AE they are asked to contact the hospital and the nurse dealing with the patient phone call uses an Acute Triage Form to record reasons for the call, document the AE and gives advice.
## Erapid intervention
In addition to usual care, participants randomised to the eRAPID intervention arm will receive training on using the system and will be asked to complete the eRAPID symptom report routinely from home at least weekly and a b a Screenshots of eRAPID intervention (Patient login and symptom reports). b Screenshots of eRAPID intervention-Clinician view of symptom reports in electronic patient record (EPR) when they experience symptoms over 18 weeks during treatment. Clinicians are given access to patients' selfreported AEs via the electronic patient record system (PPM) and asked to utilise the information when seeing patients in clinic or providing telephone advice.
## Aims and study objectives
To evaluate the potential benefits of eRAPID for patients and staff, the intervention and usual care arm will be compared on the following areas through the collection of appropriate clinical information, patient reported outcomes and interview data:
1. Assessment of hypothesised benefits to patients with mild or moderate AE: a) Number of hospital, GP and community contacts during the study b) Improved patient reported outcomes c) Improved symptom detection and supportive medication use 2. Assessment of hypothesised benefits to patients with severe AE: a) Improved detection and treatment of AEs and admissions (e.g. number of clinician alerts generated from eRAPID, number of admissions and hospital contacts) b) Levels of morbidity (percentage of planned chemotherapy received, changes to treatment plans (dose reductions, dose delays/interruptions)).
## Assessment of hypothesised benefits to clinicians:
Staff will be interviewed about their views of the value of eRAPID in saving time currently spent enquiring and recording patients' AE and supporting treatment decision-making. In addition oncologists will complete a feedback form at routine review appointments after seeing eRAPID intervention participants to assess how/if patient reports are used. 4. Monitor patient safety, assessed by monitoring acute admissions, cumulative deaths and cause of death.
The FACT-G Physical Wellbeing Score [bib_ref] The functional assessment of cancer therapy scale: development and validation of the..., Cella [/bib_ref] (measured at 18 weeks) is the primary outcome. The main secondary outcome is cost effectiveness assessed via use of health care services (including hospital admissions, telephone contacts and consultations, medication and personal expenses). In addition participant records will be linked to costs held within the local pilot database of the National Patient-Level Information and Costing System (PLICS) scheme. This provides a cost for hospital based accident and emergency department visits, outpatient attendances and inpatient stays.
# Ethical approval
The study was approved by the National Research Ethics Service (now part of the Health Research Authority) Yorkshire & The Humber Leeds East Committee in September 2014 (Reference 14/YH/1066). Local approvals from the Leeds Teaching Hospitals NHS Trust Research and Innovation Department were also obtained.
## The rct has two phases
I. An internal pilot phase to assess the feasibility and acceptability of the intervention and allow for minor modifications before further large scale recruitment was conducted. If no meaningful changes are made to the intervention the study would progress to the main trial and patients recruited during the pilot phase will be included in the analysis. II. The full trial phase will continue to recruit the target sample (at most N = 504 participants, see sample size calculation below) using the best recruitment and retention methods established in the internal pilot.
## Internal pilot phase
Prior to starting the full trial an internal pilot phase was conducted with the aim of assessing recruitment and attrition rates, refining the intervention, testing the integrity of information technology (IT) systems and to establish procedures and methods for collecting outcome measure data. We aimed to achieve (i) recruitment levels of >10 patients per month), (ii) 60% to consent to randomisation, and (iii) <30% attrition. The pilot sample size was set at 30 participants perarm [bib_ref] Understanding distress and distressing experiences in patients living with multiple myeloma: an..., Potrata [/bib_ref] allowing for 30% overall attrition, the overall target was a minimum of 42 patients per-arm (N = 84). Recruitment took place between January-September 2015.134 patients were approached, 87 consented, 22 declined and 25 were excluded after further screening (no Internet access or not continuing on to chemotherapy). The consent rate when including those patients excluded post-screening was 65% (87 consented/134 approached). However the "true" consent rate excluding the 25 patients was 80% (134 approached -25 ineligible). Fortyfour participants were allocated to the Intervention arm and 43 to Usual Care. Only 13 participants (15%) withdrew. No significant problems with the IT systems underpinning the eRAPID online intervention were encountered and the research team was able to develop robust methods of gathering information on clinical process data (e.g. hospital contacts, changes to treatment). Based on participant feedback some refinements were made to patient "use of resources forms" to aid comprehension of questions and ease of completion. The overall recruitment and attrition targets were met and the Trial Steering Committee (TSC) recommended progression to the main trial. The study procedures described below reflect the protocol for the main trial approved by Yorkshire & The Humber Leeds East Research Ethics Committee in December 2016, protocol version number 1.5.
## Patient eligibility inclusion criteria
Adult patients (aged 18 years or over) attending St James' Institute of Oncology, Leeds with breast cancer undertaking either neo-adjuvant or adjuvant systemic treatment pathways, gynaecological or colorectal cancer requiring chemotherapy Prescribed at least 3 months of planned chemotherapy cycles at the time of study consent Able and willing to give informed consent Able to read and understand English Access to the internet at home
## Exclusion criteria
Patients are excluded from participation if they are:
Taking part in other clinical trials involving the completion of extensive patient reported outcome or quality of life measures or have previously participated in an eRAPID trial Exhibiting overt psychopathology/cognitive dysfunction
## Recruitment processes
## Identification of eligible patients
Patients are recruited from outpatient clinics and day case wards at St James' Institute of Oncology clinics.
Eligible patients are identified by screening of the clinic, in-patient or day-case lists by the most appropriate clinical staff. Prior to study commencement, consultants responsible for the care of patients within each eligible tumour group are contacted via email and sent an introduction to the study and permission is requested for the research team to approach their patients.
## Approaching patients
An appropriate member of the clinical team seeks permission from eligible patients for the researcher to speak to them about the study. After introduction from clinical staff, eligible patients are approached by a member of the research team who explain the study and provide the information sheet. Patients are given as much time as they need to read the information and ask questions and should they wish to participate they are consented at the visit. Where patients prefer more time to consider participation, they can take the information home and discuss the study again with the researcher at their next visit.
When patients are happy to participate they are asked to provide written informed consent. The participant is then randomised to either the intervention or control arm. Participants who are randomised to the intervention arm receive training in using the eRAPID system.
## Randomisation
After trial eligibility has been confirmed and consent given, randomisation is performed via the University of Leeds Clinical Trials Research Unit (CTRU) telephone system. Participants are randomised with 1:1 allocation to intervention and control groups. Patients are stratified by cancer site (breast, gynaecological or colorectal), gender and previous chemotherapy (gynaecological cancer patients only) in variable random permuted blocks of 4, 6 or 8, see [fig_ref] Figure 4: Stratification factors used in randomising patients in the eRAPID RCT [/fig_ref].
## Erapid intervention: participant and staff training participant training
Researchers provide a short demonstration on how to use the eRAPID system and provide patients with a unique user name and password to access the system, on an eRAPID 'postcard'. Participants are given a user manual to take home providing a step-by-step guide on how to log in and use the eRAPID system. Participants are asked to complete the remote eRAPID Adverse Events (AEs) questionnaire weekly (from home or during clinic visits) and at any time when they experience any side-effects/symptoms during the duration of their treatment. The questionnaire consists of 12-15 items depending on the disease group assessing the severity of common symptoms such as: nausea, vomiting, pain, fatigue, diarrhoea, constipation, sore mouth/tongue, temperature, chills, performance status, fatigue, sleep, and appetite. Participants can also provide details about additional problems at the end of the standard questions. A weekly text message or email reminder are sent to the participants as a prompt to complete the eRAPID AE questionnaire.
## Staff training
Prior to study commencement the appropriate staff received training on eRAPID. The aims of training are to support staff in understanding:
1. How patients use and interact with eRAPID and the content of self-reported AE questionnaire/website 2. Accessing patients' eRAPID self-report data in the electronic patient records 3. Interpreting patient self-reported AE scores and methods of incorporating the data into clinical encounters with patients. Including information on how the symptom scores relate to mild, moderate and severe problems and how the cut-offs or alerts for severe symptoms have been developed
During one-to-one/small group interactive sessions eRAPID is demonstrated by the research team, giving staff an opportunity to see the patient interface. Staff are shown the practicalities of locating the data within the electronic patient records. Manuals are provided outlining the key steps in all the processes covered in the session. Training highlights that the self-report information should be seen as a supplementary resource for staff to use in conjunction with routine practices for clinical decisions.
## Outcome measures
The following measures and data are being collected to enable comparison between the usual care and eRAPID intervention arms. An overview of the outcome data and time points are outlined in [fig_ref] Table 1: eRAPID RCT in systemic cancer treatment [/fig_ref]. Secondary outcomes-Self-efficacy Self-Efficacy for Managing Chronic Disease [bib_ref] Effect of a self-management program on patients with chronic disease, Lorig [/bib_ref] 6-Item scale covering several domains common across chronic diseases (symptom control, role function, emotional functioning and communicating with physicians)
Items rated from 1-(not at all confident) to 10 (totally confident)
- How confident are you that you can keep physical discomfort or pain of your disease from interfering with the things you want to do?
## Baseline and 18 weeks
The score for the scale is calculated from the mean of the six items.
- How confident are you that you can do things other than just taking medication to reduce how much you illness affects your everyday life?
Cancer Behaviour Inventory-Brief (CBI-B) [bib_ref] Assessing self-efficacy for coping with cancer: development and psychometric analysis of the..., Heitzmann [/bib_ref] A measure of self-efficacy for coping with cancer. 14 items (adapted from full 33 item measure)
Items are rated on a 9-point scale ranging from 1 ("not all confident") to 9 ("totally confident") Responses are combined to provide a single score of between 0 and 100 with higher scores representing higher levels of patient activation.
- I know what treatments are available for my health problems.
Scores can be classified into one of four groups, known as 'levels of activation'.
Secondary outcomes-eRAPID/IT system performance System Usability Scale (SUS)10 item instrument to assess views of usability of an IT systems.
Each statement rated from 1 strongly disagree to 5 strongly agree.
- I think that I would like to use this system frequently 18 weeks
- I thought there was too much inconsistency in this system
Responses are calculated into a total score ranging from 0 to 100 with higher scores representing better system usability.
- I felt very confident using the system eRAPID end of study questionnaire 15 statements/free text boxes to assess participant views of using eRAPID and suggestions for improvements Statements rated on 3-5 response option scales (e.g. very easy-very difficult) and free text boxes for comments.
- How easy or difficult was it to learn how to use the eRAPID system?
## Weeks
- How did you feel about the amount of time it took to complete the symptom questions?
- To what extent do you feel that the symptom questionnaire was useful for the doctors and nurses you saw during your treatment?
- Have you got any suggestions about how the eRAPID system could be improved?
## Patient outcome measures
Functional assessment in cancer therapy scale-General (FACT-G) [bib_ref] The functional assessment of cancer therapy scale: development and validation of the..., Cella [/bib_ref] The FACT-G is a cancer specific measure widely used in clinical trials. It has four subscales: physical wellbeing, social or family wellbeing, emotional wellbeing, and functional wellbeing. Question responses range from 0 to 4. Higher scores on the questionnaire indicate better quality of life.
## Eq-5d-5 l [38]
The EQ-5D is a standardised instrument for use as a measure of health outcome developed by the EuroQol Group. The instrument assesses five dimensions: mobility; self-care; usual activities; pain/discomfort and anxiety/depression. Each dimension has five response levels (ranging from no problems to extreme problems). The instrument also includes a scale to rate health from 0 (worst health you can imagine) to 100 (best health you can imagine).
## Use of resources
Resource use is assessed using patient forms (detailing non-hospital contacts e.g. appointments with GPs/community services, counsellors, local support services), as well as medication use and costs incurred as a consequence of cancer diagnosis/treatment. This form is based on those developed by Hulme for a recently completed trial assessing treatment for chemotherapy-related nausea/vomiting (https://njl-admin.nihr.ac.uk/document/download/2002381).
## Eortc-qlq-c30 [39]
The EORTC QLQ-C30 is a 30-item questionnaire consisting of five functional scales (physical, emotional, cognitive, social, role), three symptom scales (fatigue, - 3 month prior to 12 month follow-up assessment Clinical process-Information from general practice
- GP recorded problems/concurrent illnesses - Data extracted from medical notes for 18 week study period - Prescribed medications and reasons for prescription (where available) - 3 month period prior to 12 month follow-up assessment for subset of participants IT/System functioning - Researcher maintained log of IT issues (e.g. server downtime, contacts with study participants reporting IT problems or issues logging into eRAPID) and how these were resolved Throughout trial pain, nausea/vomiting), a global health related quality of life scale, and six single items (anorexia, insomnia, dyspnoea, diarrhoea, constipation, financial difficulties).
Questions are rated on a 4 or 7 point response scale and overall scale scores are calculated from 0 to 100 with higher scores indicating better quality of life or functioning. Symptoms scales are scored so that higher scores indicate worse symptoms experience.
Self-efficacy and patient activation Self-Efficacy for Managing Chronic Disease 6-Item Scale [bib_ref] Effect of a self-management program on patients with chronic disease, Lorig [/bib_ref] This 6-item scale covers several domains that are common across many chronic diseases such as symptom control, role function, emotional functioning and communicating with physicians.
## The cancer behaviour inventory-brief (cbi-b) [40]
A self-efficacy measure specifically designed for assessing coping with cancer. Devised from the full 33 item measure, this brief version has 14 items covering: maintaining activity and independence, seeking and understanding medical information, stress management, coping with treatment related side effects and accepting cancer/maintaining a positive attitude.
## The patient activation measure (pam) [41]
The PAM is a tool for measuring the level of patient engagement in their healthcare. It was designed to assess an individual's knowledge, skill and confidence for selfmanagement. The PAM 13-item scale explores beliefs, knowledge and confidence for engaging in health behaviours. Each item is rated on a four point scale from strongly disagree to strongly agree and an overall score from 0 to 100 can be calculated. These scores can be subdivided to categorise people into one of four activation categories ranging from 1-Low activation to 4-High activation.
## Socio-demographic and clinical process data
Participants complete a baseline questionnaire on sociodemographics and current computer usage. Clinical baseline data are obtained from participants' medical notes and include diagnosis, co-morbidities and planned treatment [fig_ref] Table 2: eRAPID RCT in systemic cancer treatment [/fig_ref].
To determine any association between the eRAPID intervention and improved detection and management of AEs, data is collected from hospital triage forms, medical records, hospital databases to record:
Number of scheduled and unscheduled hospital contacts (admissions, clinic visits, phone calls with staff ) Changes to supportive medications and chemotherapy dose changes Contacts with GP and community services Number of clinician alerts generated from eRAPID severe symptom reports and actions taken by staff eRAPID system performance Throughout the study the eRAPID IT system is monitored for unscheduled server down time (leading to the unavailability of the QTool questionnaire website, eRA-PID website and patient symptom data in PPM). A log of phone calls/feedback from study participants regarding issues/problems surrounding the use of the eRAPID questionnaire or website will be maintained. eRAPID intervention participants are asked to complete the System Usability Scale(SUS). This 10 item instrument assesses subjective views of usability of different systems including hardware, software, mobile devices, websites and applications. The 10 items cover the ease of using the system, its complexity and user confidence. Each item is rated from 1 to 5 and a composite score of overall usability can be calculated ranging from 0 to 100 (higher scores reflect better usability). Intervention participants are also asked to complete a short end of study questionnaire about their experiences with the eRAPID intervention which includes free text boxes for comments and feedback.
## Participant interviews
Between 5 and 10 participants per disease group and study arm will be interviewed at the end of the full trial. Participants will be asked about their treatment experience, how they managed and monitored their symptoms and perceptions of reporting and discussing their symptoms with hospital staff. Intervention arm participants will be asked to describe their thoughts on using the eRAPID system.
## Staff feedback-interviews and questionnaires
At routine chemotherapy review appointments involving eRAPID intervention patients, staff will be asked to provide:
Clinician reports of use of eRAPID patient data during consultations At 6 weeks routine clinic visits clinicians are asked to complete CTCAE scoring form matching those AE completed by patients on the eRAPID questionnaire At the end of the study 5 health professionals from each disease group will be interviewed to determine their views of eRAPID, the perceived value and use of the patient data in clinical practice (e.g. improving the detection, documentation and management of AE, supporting treatment decision-making in routine care).
Perceptions of staff training needs and recommendations for improving the system will also be explored.
## Sample size calculations
The sample size for the full trial is based on the primary patient outcome of better symptom control measured at 18 weeks by the FACT-G. A sample of 176 patients per arm is necessary to detect a 2-point change in the FACT-G Physical Wellbeing score with 80% power and 5% significance, where the population standard deviation is 6.7. This change corresponds to a medium Cohen's effect size (0.3) [bib_ref] Metaanalysis provides evidence-based effect sizes for a cancer-specific quality-oflife questionnaire, the FACT-G, King [/bib_ref].
Allowing for 30% attrition, a minimum of 252 patients per arm (504 total) is required. With potentially >500 eligible patients treated in the cancer centre annually, we expect to recruit 20 patients per month over approximately 24-30 months, allowing for 70% internet access and 70% consent rate.
# Analysis populations
All analyses and data summaries will be conducted on the intention-to-treat (ITT) population which is defined as all participants registered regardless of non-compliance with the protocol or withdrawal from the study.
## Statistical analysis baseline characteristics
Data from the baseline socio-demographic, computer usage and clinical data questionnaires will be tabulated using frequencies and summary statistics for each treatment group and overall for both the pilot phase and full trial.
## Primary outcome
The FACT-G Physical Well-being score will be summarised overall and by treatment arm. Changes in score over time and differences between treatment arms will be explored using a multilevel repeated measures model. The model for each post-randomisation point will be adjusted for baseline score and stratification factors. If there are missing items, subscale scores will be prorated as per the FACT-G scoring manual.
## Secondary outcomes clinical process measures
The number of calls made to the hospital will be summarised overall and by treatment arm. Differences between the two treatment groups will be compared using either Poisson regression or negative binomial regression; the most appropriate model will be chosen after performing post-estimation tests. Models will be adjusted for the stratification factors.
The numbers of weekly/additional AE reports and severe AE alerts generated will be summarised for participants randomised to the eRAPID intervention. The number of telephone calls to hospital staff, acute admissions, contacts with GP and/or community services and number of deaths will be summarised overall and by treatment arm. Any differences between treatment arms will be explored using the most appropriate regression model (either Poisson or negative binomial, to be decided using post-estimation tests) adjusted for stratification factors.
## Patient outcome measures (other than primary)
Changes in scores over time and differences between treatment arms will be explored using a multilevel repeated measures model adjusted for baseline scores and stratification factors. As the sample size was not powered to detect changes in these outcome measures, statistical significance will be assessed at the 1% level.
## Health-economic data
An embedded health-economic study will allow within trial incremental cost-effectiveness analysis (18 weeks) taking the perspective of the service provider including the costs of NHS and Personal Social Services. The analysis will compare usual care with the eRAPID-supported pathway. A secondary analysis will take a societal perspective. Analyses will use quality-adjusted-life-years (QALYs) outcome-measures. Estimation of QALYs requires the production of utility-weights for each health-state observed in the trial population. We will use the EQ-5D-5 L for this purpose [bib_ref] Emergency department waiting times for patients with cancer with febrile neutropenia: a..., Nirenberg [/bib_ref] [bib_ref] Symptom monitoring with patient-reported outcomes during routine cancer treatment: a randomized controlled..., Basch [/bib_ref] collected at baseline, 6, 12 & 18 weeks. We will also use EORTC QLQ-C30 to derive utilities (EORTC QLQ-U10) to calculate QALYs in the same way. This will limit the need to interpolate quality of life between observation points [bib_ref] Design and analysis of pilot studies: recommendations for good practice, Lancaster [/bib_ref]. NHS resource-use associated with each treatment modality will be collected using the process-of-care measures to contribute to a health-economics analysis of additional health financial costs related to treatment and the study. Use of outpatient and community-based health and social care (including, for example, home help or residential care) will be collected from the patient at baseline, 6, 12, and 18 weeks with the Use of Resources questionnaire developed in the Programme Development Grant and tested in the pilot study. Unit financial costs for health services resources will be obtained from national source: the Personal Social Services Research Unit, the British National Formulary and NHS reference cost database. Given the duration of the trial discounting is not required.
Secondary analysis will include costs to participants (travel expenses, over the counter medicines) and productivity losses.
In addition to the analyses at 18 weeks we will undertake an exploratory cost effectiveness analysis (including a planned a-priori sub-group cost-effectiveness analysis at 12 months using a sub-sample of participants for whom we have collected resource use, EQ-5D-5 L and EORTC QLQ-C30 data).
For each analysis we will undertake probabilistic sensitivity analysis using bootstrapping. The results will be presented as the Expected Incremental Cost Effectiveness Ratio, scatter plot on the cost-effectiveness plane and a Cost Effectiveness Acceptability Curve. We will calculate the expected net-benefit assuming lambda has a value of £20,000 [bib_ref] Integrated care pathways for cancer survivors -a role for patient-reported outcome measures..., Warrington [/bib_ref].
## Qualitative data
Interviews will be recorded and transcribed. Data will be managed by NVivo software and analysed using thematic analysis [bib_ref] Understanding distress and distressing experiences in patients living with multiple myeloma: an..., Potrata [/bib_ref] [bib_ref] A qualitative study evaluating causality attribution for serious adverse events during early..., Mukherjee [/bib_ref]. Two researchers will independently look for the emerging themes and code them. Then they will meet, compare the codes/themes and resolve any potential conflicts by consensus.
# Discussion
This paper describes the protocol for the eRAPID RCT in systemic cancer treatment. eRAPID is a unique web based intervention designed to improve the systematic reporting of AE during cancer treatment and improve patient care and experiences. A number of web based PROMs systems have been developed. Since the current trial began Basch and colleagues in the U.S. have published findings from an RCT using the STAR (Symptom Tracking and Reporting) web interface during chemotherapy indicating a positive impact on patients' quality of life, treatment delivery, number of emergency room attendances and 1 year survival [bib_ref] Symptom monitoring with patient-reported outcomes during routine cancer treatment: a randomized controlled..., Basch [/bib_ref]. We believe eRAPID is the first of its kind to allow remote monitoring of symptoms and side effects where patient reported data is accessible alongside standard clinical information in electronic patient records as well as providing patients with immediate symptom management advice. We hypothesise that these features along with alerts for severe symptoms will lead to improved clinical outcomes for participants allocated to the eRAPID intervention and will benefit health care services.
This study can be considered a complex intervention due to the number of active components involved. These include the new technology for patients completing symptom self-reports from home, automatic advice on managing mild symptoms and when to contact the hospital for severe problems, the availability of this patient data for staff to use in clinical practice, alert generation for severe problems and maintaining staff training and engagement. Consequently eRAPID's success relies on the investment of both staff and patient groups in the intervention and the robustness of the IT supporting the system. Although the eRAPID website and the online symptom reporting questionnaire have undergone extensive usability testing, the pilot phase of the RCT was considered vital in order to assess the intervention over a longer time frame and with all participating cancer groups as each differ in terms of the care pathways and staff involved. The decision to perform an internal pilot, rather than a separate pilot study, was motivated by our intention to avoid losing momentum and reduce the time between the end of the pilot and the start of the main trial [bib_ref] Design and analysis of pilot studies: recommendations for good practice, Lancaster [/bib_ref]. This approach aimed to maintain continuity with the staff involved in the eRAPID intervention both in terms of recruitment and utilising the patient AE reports in clinical encounters.
The study is funded as part of 5 year programme, in parallel we are developing multi-centre eRAPID interventions for cancer patients receiving radiotherapy and surgery which will be evaluated in separate pilot studies. If found to have a positive effect on patient wellbeing and use of health care resources, eRAPID has the potential to provide a cost effective enhancement to the standard care of cancer patients. Such an approach could also be extended to long-term survivorship beyond cancer treatment [bib_ref] Integrated care pathways for cancer survivors -a role for patient-reported outcome measures..., Warrington [/bib_ref].
[fig] Figure 1: eRAPID system overview [/fig]
[fig] Figure 3: Trial flow diagram [/fig]
[fig] Figure 4: Stratification factors used in randomising patients in the eRAPID RCT [/fig]
[table] Table 1: eRAPID RCT in systemic cancer treatment: Participant completed primary and secondary outcomes measures • I get support from my friends • I worry that my condition will get worse • I have accepted my illness Secondary outcomes-health economic/clinical process data [/table]
[table] Table 2: eRAPID RCT in systemic cancer treatment: Researcher collected data for secondary outcomes • Baseline data on planned chemotherapy • Changes to treatment delivery and reason [/table]
|
Isolation, characterization and transcriptome analysis of a novel Antarctic Aspergillus sydowii strain MS-19 as a potential lignocellulosic enzyme source
Background: With the growing demand for fossil fuels and the severe energy crisis, lignocellulose is widely regarded as a promising cost-effective renewable resource for ethanol production, and the use of lignocellulose residues as raw material is remarkable. Polar organisms have important value in scientific research and development for their novelty, uniqueness and diversity. Results: In this study, a fungus Aspergillus sydowii MS-19, with the potential for lignocellulose degradation was screened out and isolated from an Antarctic region. The growth profile of Aspergillus sydowii MS-19 was measured, revealing that Aspergillus sydowii MS-19 could utilize lignin as a sole carbon source. Its ability to synthesize low-temperature lignin peroxidase (Lip) and manganese peroxidase (Mnp) enzymes was verified, and the properties of these enzymes were also investigated. High-throughput sequencing was employed to identify and characterize the transcriptome of Aspergillus sydowii MS-19. Carbohydrate-Active Enzymes (CAZyme)-annotated genes in Aspergillus sydowii MS-19 were compared with those in the brown-rot fungus representative species, Postia placenta and Penicillium decumbens. There were 701CAZymes annotated in Aspergillus sydowii MS-19, including 17 cellulases and 19 feruloyl esterases related to lignocellulose-degradation. Remarkably, one sequence annotated as laccase was obtained, which can degrade lignin. Three peroxidase sequences sharing a similar structure with typical lignin peroxidase and manganese peroxidase were also found and annotated as haem-binding peroxidase, glutathione peroxidase and catalase-peroxidase. Conclusions: In this study, the fungus Aspergillus sydowii MS-19 was isolated and shown to synthesize low-temperature lignin-degrading enzymes: lignin peroxidase (Lip) and manganese peroxidase (Mnp). These findings provide useful information to improve our understanding of low-temperature lignocellulosic enzyme production by polar microorganisms and to facilitate research and applications of the novel Antarctic Aspergillus sydowii strain MS-19 as a potential lignocellulosic enzyme source.
# Background
With the growing demand for fossil fuel and the severe energy crisis, lignocellulose is widely regarded as a promising, cost-effective renewable resource for bioethanol production, and the use of lignocellulose residues as a raw material has become remarkable [bib_ref] Fuel ethanol production from lignocellulose: a challenge for metabolic engineering and process..., Zaldivar [/bib_ref] [bib_ref] Fungal bioconversion of lignocellulosic residues; opportunities & perspectives, Dashtban [/bib_ref] [bib_ref] Hydrolysis of lignocellulosic materials for ethanol production: a review, Sun [/bib_ref]. However, there are numerous technological obstacles to the degradation of lignocellulose. Lignin is important organic matter that is widely present in the plant cell wall. Together with cellulose and hemicellulose, lignin forms the main component of the plant skeleton, representing the second most abundant organic regenerative resource after cellulose on earth. Since lignin and cellulose are cross-linked and lignin has complex physical and chemical properties, it represents the restrictive factor for the utilization of lignocelluloses. To effectively utilize cellulose and hemicellulose from lignocellulose raw materials, it is essential to release them from lignin bonds. Cellulose is composed of D-glucose with beta-1 and 4 glycosides, while lignin is a natural amorphous highmolecular-weight polymer. Because of the complex structure of lignin, there are no conclusive assessments of its structure to date, but a consensus has developed that the basic structure of lignin consists of phenyl propane units [bib_ref] Occurrence of naturally acetylated lignin units, Rio [/bib_ref].
In their natural environments, only a small number of microorganisms are capable of degrading lignin. Lignin can be successfully implemented not only by pure cultures of particular microorganisms but also by the application of a variety of lignocellulolytic species and some non-lignocellulolytic microbes that work synergistically to break down the tough lignocellulosic structure [bib_ref] Microbial cellulose utilization: fundamentals and biotechnology, Lynd [/bib_ref] [bib_ref] Lignocellulose biotechnology: issues of bioconversion and enzyme production, Jvre L [/bib_ref] [bib_ref] Lignocellulosic residues: biodegradation and bioconversion by fungi, Sanchez [/bib_ref]. The complete degradation of lignin results from the cooperation of fungi, bacteria and actinomycetes, among which fungi play the most important role. Fungi enter wood materials through hyphae while secreting extracellular enzymes that attack cellulose in the plant cell wall, resulting in the depolymerization and dissolution of lignin and cellulose. According to the type of decay caused in different lignocellulose components, the fungi can be divided into white-rot fungi, brown-rot fungi and softrot fungi [bib_ref] Vanillic acid metabolism by selected soft-rot, brown-rot, and white-rot fungi, Buswell [/bib_ref]. The essence of lignin degradation consists of an oxidative process, with almost equal importance of phenol oxidase conduction. It is generally believed that lignin degradation mainly depends on four enzymes that are secreted by white-rot fungi: Lac (laccase, EC 1.10.3.2) [bib_ref] Lignin-degrading enzyme fromPhanerochaete chrysosporium, Kirk [/bib_ref] , LiP (lignin-peroxidase, EC 1.11.1.14) [bib_ref] Oxidative degradation of non-phenolic lignin during lipid peroxidation by fungal manganese peroxidase, Bao [/bib_ref] , Mnp (manganese peroxidase, EC 1.11.1.16) [bib_ref] Oxidation of non-phenolic substrates: An expanded role for laccase in lignin biodegradation, Bourbonnais [/bib_ref] and VP (versatile peroxidase, EC 1.11.1.16) [bib_ref] Catalytic activity of versatile peroxidase from Bjerkandera fumosa in aqueous solutions of..., Rodakiewicz-Nowak [/bib_ref]. Some lignindegrading fungi do not secrete laccase, including Phanerochete chrysosporium, which indicates that laccase is not necessary for the degradation of lignin but could be involved in the process coordinating with other peroxidases [bib_ref] Crystal structure of lignin peroxidase, Edwards [/bib_ref]. Recently, several Aspergillus fungi have been shown to produce such enzymes, and additional enzymes are indispensable for complete degradation [bib_ref] Molecular biology and structure-function of lignin-degrading heme peroxidases, Martínez [/bib_ref] [bib_ref] Biodegradation of lignocellulosics: microbial, chemical, and enzymatic aspects of the fungal attack..., Martínez [/bib_ref]. Cellulase and hemicellulose are also required during lignin degradation, most of which can be categorized into the glycoside hydrolase (GH) families and carbohydrate esterase (GE) families in the Carbohydrate-Active Enzymes database (CAZy). Apart from the four abovedescribed peroxidases, other particular enzymes also participate in or have a certain impact on lignin degradation, including cellobiose dehydrogenase [bib_ref] Lignin-degrading peroxidases from genome of selective ligninolytic fungus Ceriporiopsis subvermispora, Fernández-Fueyo [/bib_ref]. This unexpected finding may imply that lignin degradation mechanisms vary among species.
The production of bioethanol requires the degradation of cellulose and lignin to glucose, followed by the fermentation of glucose by yeast. The temperature required for yeast fermentation is 30°C. Low-temperature enzymes are defined as those with an optimum temperature of approximately 35°C while maintaining a certain catalytic efficiency at 0°C compared to the most optimum reaction temperature for cellulose ranging from 45°C to 65°C. Hence, if low-temperature cellulase and lignin degradation enzymes could be obtained, synchronized fermentation of the two procedures could be achieved, which would greatly simplify the production process of bioethanol and reduce costs. Although large amounts of research have been published on cellulase and lignin-degrading enzymes, few studies have investigated low-temperature lignocellulose degradation enzymes. Cecil w. Forsberg et al. reported low-temperature glucanase from the rumen thermophilic anaerobic bacteria, Fibrobacter succinogenes S85 [bib_ref] A novel cold-tolerant Clostridium strain PXYL1 isolated from a psychrophilic cattle manure..., Akila [/bib_ref] , and another low-temperature cellulose, CelG, from the Antarctic marine thermophilic bacteria, Pseudoalteromonas haloplanktis, was also discovered [bib_ref] Expression, purification, crystallization and preliminary X-ray crystallographic studies of a psychrophilic cellulase..., Violot [/bib_ref]. To date, no reports on low-temperature lignin degradation enzymes have been identified.
High-throughput sequencing of transcriptomes (RNA-Seq) has provided new routes to study the genetic and functional information stored within any organism at an unprecedented scale and speed. These advances greatly facilitate functional transcriptome research in species with limited genetic resources, including many "nonmodel" organisms with substantial ecological or evolutionary importance [bib_ref] De novo transcript sequence reconstruction from RNA-seq using the Trinity platform for..., Haas [/bib_ref]. Most genomics studies of lignin-degrading fungi have focused on white-rot fungi, brown-rot fungi belonging to the Basidiomycota and filamentous fungi (trichoderma, neurospora, penicillium, among others) belonging to the Ascomycota. The most widely studied white-rot fungi is Phanerochete chrysosporium, the genome sequence of which was published in 2004 [bib_ref] Genome sequence of the lignocellulose degrading fungus Phanerochaete chrysosporium strain RP78, Martinez [/bib_ref]. Analyses of its genome sequence and subsequently of its transcriptional and secretary proteins have provided ample information [bib_ref] Transcriptome and Secretome Analyses of Phanerochaete chrysosporium Reveal Complex Patterns of Gene..., Vanden Wymelenberg [/bib_ref] [bib_ref] Extracellular oxidative systems of the lignin-degrading Basidiomycete Phanerochaete chrysosporium, Kersten [/bib_ref] [bib_ref] Expression analysis of extracellular proteins from Phanerochaete chrysosporium grown on different liquid..., Sato [/bib_ref]. Analyses of the transcriptome suggested that 545 genes or proteins were significantly altered during the lignin degradation process. Some proteins contain signal peptide and carbohydrate (CBM) domains, which may be related to the degradation of lignocelluloses. Martinez D et al. analysed the genome, transcriptome and secretome of Postia placenta, the most well-studied brown-rot fungus. They discovered three groups of peroxidase (LiP, Mnp and VP) and laccasein the fungal genome [bib_ref] Genome, transcriptome, and secretome analysis of wood decay fungus Postia placenta supports..., Diego [/bib_ref] , in accordance with the inability of brown-rot fungi to degrade lignin.
This study aimed to isolate, identify and perform a transcriptome analysis of novel strains of fungi with the potential to degrade the lignocellulosic biomass isolated from the Antarctic Pole, to screen for filamentous fungi capable of producing low-temperature lignin enzymes, and to obtain lignin degradation-related enzymes through RNA-seq, with the goal of gaining insight into the mechanism underlying lignin degradation in fungi and providing a potential lignocellulosic enzyme source for industrial production.
# Methods
## Sample collection and isolation of fungi
Soil, macro-algal rot and sediment samples were collected from Ardley Island -near Fildes Peninsula, Antarctica, during the Chinese 27th Antarctic Scientific Expedition. All samples were placed in sterilized plastic bags or flasks and transported to the laboratory at 4°C for microorganism isolation. One gram of each sample was placed in a 50-mL sterile centrifuge tube containing 10 mL of sterile distilled water and shaken at 120 rpm overnight. The tube was maintained stably overnight. Next, 10 −1 and 10 −2 serial dilutions of each sample suspension were spread as 0.1-mL aliquots on plates containing potato dextrose agar (PDA). The PDA plates were incubated at 12°C for 1-2 weeks, and distinct colonies were picked and sub-cultured for further analysis.
# Phylogenetic analysis
The total DNA from 15 native fungal isolates was extracted according to the method of Gonzalez-Mendoza et al. with minor modifications [bib_ref] A rapid method for isolation of total DNA from pathogenic filamentous plant..., González-Mendoza [/bib_ref]. Each DNA sample was amplified by PCR with Taq DNA polymerase following the manufacturer's instructions (Tiangen, Beijing, China). Next, 2 μL DNA was used as PCR template. The primers used for amplification were ITS 1 forward (TCCGTAGGTGAACCTGCGG) and ITS 4 reverse (TCCTCCGCTTATTGATATGC). PCR amplification was performed using the following protocol: 94°C for 5 min (1 cycle), 94°C for 30 s, 55°C for 30 s and 72°C for 40 s (30 cycles), and 72°C for 10 min(1 cycle). To confirm the quality of the PCR, the amplification products were run on 0.8% Tris acetate EDTA agarose gels, and bands were visualized by staining with ethidium bromide. The PCR products were purified using a universal DNA purification kit (Tiangen, Beijing, China), and the amplified ITS regions were sequenced (Jimei, Shanghai, China) and submitted to GenBank.
The similarities of the 15 native isolate sequences with other known species were investigated by comparisons with sequence data in the National Center for Biotechnology Information (NCBI) database using the BLASTN programme. The phylogenetic analysis was based on BioEdit multiple alignment with sequences from their closest relatives and from common fungi in the Antarctic. A phylogenetic tree based on the ITS region was constructed using MEGA5.1 software with the neighbour-joining method, and the statistical analysis utilized bootstrapping with 1000 replications.
## Enzyme assays
Fresh spores of the 15 native fungi were filtered through three layers of sterile gauze and inoculated into 150-mL flasks containing 50 mL of optimized medium for lignocellulosic enzyme production (glucose 10 g/L, ammonium sulphate 0.2 g/L, KH 2 PO 4 2 g/L, MgSO 4 ·7H 2 O 0.5 g/L,CaCl 2 0.01 g/L,Vitamin B 1 1.0 mg/L). The flasks were incubated at 12°C and shaken at 120 rpm. Supernatant was collected at 2d, 3d, 4d, 5d, 7d and 10d after inoculation, centrifuged at 5000 rpm to obtain cell-free samples and specific enzymatic activities were measured as follows.
Lignin peroxidase (Lip) activity was measured by the oxidation of Azure B in a reaction mixture consisting of 32 μM Azure B, 100 μM H 2 O 2 and 50 mM sodium citrate buffer at pH 4.5 in a final volume of 1 mL. Azure B oxidation was monitored at an absorbance of 651 nm. The unit of enzymatic activity (U) was defined as an OD value reduction of 0.1 OD per mL of supernatant in 1 min, and its linear reaction time can be extended to 20 min [bib_ref] A new assay for lignin-type peroxidases employing the dye azure B, Archibald [/bib_ref].
Laccase (Lac) activity was measured using a colorimetric assay based on the oxidation of 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulphonic acid (ABTS). The reaction mixture consisted of 2 mL of ABTS (0.5 mM), 1 mL of sample and 0.1 mMHAc-NaAc buffer at pH 5 and 25°C in a final volume of 3 mL. The oxidation of ABTS was monitored from 0 to 200 s at an absorbance of 420 nm. The unit of enzymatic activity (U) was defined as 1 μM of ABTS oxidized per mL of supernatant in 1 min (ε = 3.6 * 10 4 M −1 cm −1 ) [bib_ref] Screening for ligninolytic enzymes in autochthonous fungal strains from Argentina isolated from..., Saparrat [/bib_ref].
Manganese peroxidase (Mnp) activity was measured by the oxidation of guaiacol in a reaction mixture added in turn to 2.9 mL of phosphate buffer (50 mM), 1 mL of H 2 O 2 (2%), 1 mL of guaiacol (50 mM) and 0.1 mL of fermentation broth. The enzyme solution was boiled for 5 min and used as a control. The reaction system was incubated in a 34°C water bath for 3 min immediately after addition of the fermentation broth, quickly diluted one-fold, and then monitored for phenol red oxidation at an absorbance of 465 nm once per minute a total of five times. The unit of enzymatic activity (U) was defined as △OD465/t min × 1000.
To evaluate the effects of different temperatures and pH values on the enzymatic activity, Lip and Mnp activities were measured from 0 to 60°C and in buffers with different pH values: Na 2 HPO 4 -KH 2 PO 4 buffer (pH 6.0-7.5), Tris-HCl buffer (pH 7.5-9.0) and Na 2 CO 3 -NaHCO 3 (pH 9.0-11.0). All enzymatic activities were assessed in triplicate using a GE healthcare NanoVue spectrophotometer.
## Growth measurements of aspergillus sydowii ms-19
The growth of Aspergillus sydowii MS-19 in fermentation medium (glucose 10 g/L, ammonium sulphate 0.2 g/L, KH 2 PO 4 2 g/L, MgSO 4 ·7H 2 O 0.5 g/L,CaCl 2 0.01 g/L,Vitamin B 1 1.0 mg/L)was followed based on measurement of the dry weight of mycelia: to evaluate the effect of temperature on Aspergillus sydowii MS-19 growth, fermentation broth was filtered using a moderate speed qualitative filter, and mycelia were washed three times with sterile distilled water and dried at 80°C for 3 h. Fresh spores were inoculated at a 2% inoculum concentration in six 150-mL flasks containing 50 mL of fermentation medium. The flasks were incubated at 0, 3, 10, 20, 37, and 45°C shaken at 120 rpm for 10 days. The dry weights of the mycelia were measured once daily from day two after inoculation. To evaluate the effect of pH on Aspergillus sydowii MS-19 growth, fresh spores were inoculated at a 2% inoculum concentration in six 150-mL flasks containing 50 mL of fermentation medium. The flasks were incubated at pH 4, 5, 6, 7, 8 and 9 and shaken at 120 rpm at an optimized temperature (20°C). The dry weights of the mycelia were measured upon initiation of the logarithmic growth period beginning at 7 d.
To assess the effects of lignin as the sole carbon source on growth, fermentation medium was replaced by medium containing lignin as the sole carbon source (lignin 0.3 g/L, K 2 HPO 4 1.0 g/L, NaCl 0.5 g/L, MgSO 4 ·7H 2 O 0.3 g/L, NaNO 3 2.5 g/L, CaCl 2 0.1 g/L, FeCl 3 0.01 g/L, pH 7.0).
## Rna extraction and transcriptome sequencing
The Aspergillus sydowii MS-19 strain was grown in fermentation medium with shaking for 1 week. Mycelia were harvested by centrifugation and ground in liquid nitrogen. Total RNA samples were isolated using a standard TRIzol method, eluted in RNase-free water and stored at −80°C until further use. Two replicates were performed for library preparation. The integrity of the total RNA was checked on an agarose gel, and its quantity and purity were determined using NanoVue (GE). mRNA was enriched from total RNA using oligo T (dT) beads, and broken into short fragments by the addition of fragmentation buffer.
cDNA was then synthesized using these fragments as template, purified, end-repaired and dA-tailed, and then ligated to sequencing adaptors. The samples were gel sizeselected for the 150-bp fragment size. Size-selected adaptor-ligated cDNA was purified with an AgencourtAM-Pure kit (Beckman Coulter, CA, USA) and used as template for PCR amplification to create the cDNA library. The purified library were profiled using the Agilent Bio analyser and sequenced using the PE100 strategy on the IlluminaHiSeq 2000 platform to yield paired-end reads.
## Assembly and annotation of the transcriptome
Raw reads from the library were filtered to remove lowquality reads as well as adapters and poly-A/T-containing reads. The resulting clean reads were assembled to produce unigenes using the short reads assembling programme Velvet (v1.2.08) [bib_ref] Velvet: algorithms for de novo short read assembly using de Bruijn graphs, Zerbino [/bib_ref] and Osase (v0.2.08) [bib_ref] Oases: robust de novo RNAseq assembly across the dynamic range of expression..., Schulz [/bib_ref] with default parameters. Potentially contaminated sequences were removed using BLAST.
For functional annotations, all unigene sequences were searched against Nr, eggNOG, and KEGG using BLASTX to extract predicted coding region sequences with high sequence similarity to the given unigenes along with their protein function annotations. We use the blastx with cutoff of similarity > = 80%, coverage > = 80% and evalue <1e-5. In addition, the gene ontology (GO) term for each unigeneand GO enrichment analysis was obtained using the Blast2GO programme. Pathway annotation of unigenes was performed according to the KEGG database mapping method. Unigenes were aligned to the KOG database to predict and classify possible functions.
CAZyme annotation of the Aspergillus sydowii MS-19 transcriptome was conducted by searching against the dbCAN database [bib_ref] dbCAN: a web resource for automated carbohydrate-active enzyme annotation, Yin [/bib_ref] , which is a network resource for automatic annotation of carbohydrate active enzymes based on CAZyme tag domains for any submitted protein dataset. For each CAZyme family, a tag domain was defined by combining the search results for the conserved domain database (CDD) and published literature, and a corresponding Hidden Markov Model (HMM) was constructed. Bioedit software (v7.2.5) was used to assign genes with particular activities, including the enzyme commission (EC) number annotations.
# Results
## Fungal isolation
Six compost types were subjected to screening for the isolation and characterization of fungi. A total of 168 isolates were obtained, and their species and genera were determined based on their colony morphology and the microscopic characteristics of the spore apparatus, spore stalks and spores. Fifteen representative isolates with clear differences were selected for further analysis [fig_ref] Table 1: Sampling information and the representative fungal isolates [/fig_ref]. The phylogenetic tree constructed based on the ITS sequences of the 15 native isolates indicated that five isolates were members of the genus Penicillium, two isolates were closely related to Pseudeurotium, Geomyces and Cladosporium, and one isolate matched Bionectria, Aspergillus, Aureobasidium and an unclassified Onygenales (Additional file 1: . The Antarctic region exhibited a rich diversity of fungal species.
Neighbour-joining tree showing the relationship between the ITS sequences from 15 Antarctic native isolates and their closest relatives as well as common fungi in the Antarctic. The bootstrap values for the neighbour-joining analysis with 1000 replicates are shown on the branches. The scale bar represents 0.05 substitutions per amino acid site.
## Aspergillus sydowii ms-19 is able to produce lowtemperature lip and mnp enzymes
To test the ability of the 15 representative fungi to synthesize lignin-degrading enzymes, the strains were grown in liquid medium supplemented with the corresponding substrates. The activities of Lip, Lac and Mnp were measured. Four isolates, PU-01, SS-04, MS-05 and MS-19, showed Lip and Mnp activities [fig_ref] Table 2: Lignin-degrading enzymatic activities of the fungal isolatesFig [/fig_ref]. Lac activity was not detected. Since MS-19 exhibited the highest activities of both Lip and Mnp enzymes, it was chosen as our target fungal strain in this study.
Lip and Mnp activities were measured at different temperatures and pH values to identify the optimum condition for enzymes produced by MS-19. Lip and Mnp maintained a certain activity at 0°C and gradually increased with temperature, showing the highest activity at 30°C and declining sharply above this temperature . Since lowtemperature enzymesare defined to have an optimal temperature of approximately 35°C and to maintain a certain catalytic efficiency at 0°C, the Lip and Mnp enzymes synthesized by MS-19 were considered low-temperature enzymes. The optimal pH for Lip and Mnp was 3.0 and 4.5, respectively. Lip was found to have sufficiently high activity between pH 2.0-4.0, while Mnp showed high activity between pH 4.0-6.0 .
BLAST searches demonstrated that MS-19 was highly similar to Aspergillus (Additional file 1: . To further identify the phylogeny of MS-19, a phylogenetic tree was constructed based on the ITS sequences of MS-19 and its 15 closest relatives. MS-19 displayed 99% identity to Aspergillus sydowii [fig_ref] Figure 2: Phylogeny and growth profile of Aspergillus sydowii MS-19 [/fig_ref]. Combined with the morphological and microscopic characteristics of MS-19 [fig_ref] Figure 2: Phylogeny and growth profile of Aspergillus sydowii MS-19 [/fig_ref] , the genus of the MS-19 isolate was identified as Aspergillus. sydowii. Aspergillus sydowii MS-19 could grow in the temperature range from 3 to 37°C. Its growth was observed at temperatures as low as 3°C. The growth rate was maximal at 20°C and pH 7 [fig_ref] Figure 2: Phylogeny and growth profile of Aspergillus sydowii MS-19 [/fig_ref]. A white mycelial pellet could be observed after 7 days of culturing at 20°C, demonstrating that Aspergillus sydowii MS-19 could utilize lignin as a sole carbon source [fig_ref] Figure 2: Phylogeny and growth profile of Aspergillus sydowii MS-19 [/fig_ref].
## Summary of the rna-seq dataset
To obtain an overview of the Aspergillus sydowii MS-19 transcriptome, the poly (A)-enriched mRNA sample was subjected to high-throughput IlluminaHiSeq sequencing, resulting in 18,453,231 reads with an average length of 101 nt. Assembler Velvet and Oases was employed to complete the assembly and cluster of the Aspergillus sydowii MS-19 transcriptome using default parameters. From the Velvet assembly were obtained 72,387contigs with a total length of 23,862,098 nt. Oases continued to generate longer transcripts, resulting in 27,600 transcripts with a length of 21,976,433 nt. Additional sequence cluster analyses were conducted among all transcript sequences, generating 11,269 unigenes with a mean size of 1130 nt [fig_ref] Table 3: Output statistics for the sequencing and assembly [/fig_ref].
## Go and kog classification
GO assignments were used to classify the unigene functions of Aspergillus sydowii MS-19. Based on the sequence homology, 8199 unigenes were categorized into 55 functional groups. In terms of biological process, the majority of the unigenes were involved in "hydrolase activity" (3143 members), "transferase activity" (2345 members) and "transport" (1918 members). For the cellular component, the majority of the unigenes were involved in "membrane" (2393 members), "nucleus" (1760 members) and "ribosome" (479 members). The investigation of molecular functions revealed that most unigenes were involved in "DNA binding" (1522 members), "RNA binding" (621 members) and "structural molecule activity" (515 members) [fig_ref] Figure 3: Histogram of the gene ontology classification [/fig_ref]. To further evaluate the completeness of the transcriptome and the effectiveness of the annotation process, the annotated sequences were screened for genes involved in KOG classifications. In total, among17767 nr hits, 11,192 sequences had KOG classifications. Among the 25 KOG categories, the cluster for "function unknown" represented the largest group (3,039members), followed by "general function prediction" (1064 members) and "transcription" (818 members). The following categories represented the smallest groups: Extracellular structures (3 members); Cell motility (6 members) and Nuclear structure (31 members) [fig_ref] Figure 4: Histogram of the KOG classification [/fig_ref]. To identify the biological pathways that were active in Aspergillus sydowii MS-19, we mapped all the unigenes to the reference canonical pathways in KEGG, and we found that a total of 3533 sequences could be assigned to 39KEGG pathways. The most representative pathways by the unigenes were "metabolism pathways" (2229 members), "genetic information processing" (1158 members), "cellular processes" (552 members), Organismal Systems (522 members) and "environmental information processing" (196 members) [fig_ref] Figure 5: Histogram of the KEGG classification [/fig_ref]. These annotations provide a valuable resource for investigating specific processes, functions and pathways in Aspergillus sydowii MS-19.
GO has three ontologies: molecular function, cellular component and biological process, indicating the GO functional classification annotation and the number of unigenes in each category. The x-axis indicates the term description, and the y-axis indicates the number of genes annotated in the current group (GO terms with less than 50 annotated genes are not shown). The KEGG pathway annotation provides a mapping of the transcriptomic dataset to the KEGG pathway maps for biological interpretation of higher-level systemic functions. It indicates the KEGG pathway classification annotation and the percentage of each pathway.
Since our main concern was material degradation, we focused on sequences that participate in carbohydrate metabolism and xenobiotic biodegradation pathways (Additional file 2: [fig_ref] Table 2: Lignin-degrading enzymatic activities of the fungal isolatesFig [/fig_ref]. Most fungi adopt the Embden-Meyerhof-Parnas pathway (EMP) for carbohydrate metabolism, while a few fungi such as red yeast utilize the pentose phosphate pathway (HMP). Annotation of the Aspergillus sydowii MS-19 transcriptome revealed enzymes that play important roles in the EMP, such as hexokinase, glucose-6-phosphate isomerase, fructose-phosphate kinase phosphatases, glyceraldehyde-3-phosphate dehydrogenase, pyruvate kinase, acetyl coenzyme synthetase and 6-phosphate glucose isomerase, indicating that the EMP is the main carbohydrate metabolism pathwayin Aspergillus sydowii MS-19.
Regarding the annotated unigenes associated with xenobiotic biodegradation, some enzymes are capable of metabolizing the benzene ring structure. These enzymes may contribute to lignin degradation, which is a class of polymer composed of phenyl propane as the structural unit. Relevant benzene ring-metabolizing enzymes are as follows: 2-haloacid dehalogenase, aldehyde dehydrogenase and S-(hydroxymethyl) glutathione dehydrogenase related to the degradation of chloroalkane and chloroalkene; catechol 1,2-dioxygenase and carboxymethylenebutenolidase related to the degradation of toluene and chlorobenzene; phenylacetate 2-hydroxylase, fumarylacetoacetase, fumarylacetoacetase, amidase, nitrilase and 3-hydroxyphenylacetate 6-hydroxylase related to the degradation of styrene; salicylate hydroxylase related to the degradation of dioxin;glutathione Stransferase; and S-(hydroxymethyl) glutathione dehydrogenase in the cytochrome P450 superfamily related to xenobiotic metabolism, among others. These annotations indicate that Aspergillus sydowii MS-19 not only has the ability to metabolize xenobiotics and environmental pollutants but also may degrade lignin.
## Cazyme expression profiles
Using the CAZy database, a total of701 unigenes were annotated, including 355 glycoside hydrolase (GH), 208 glycosyltransferase (GT), 9 polysaccharide lyase (PL), 92 carbohydrate esterase (CE) and 37 carbohydrate combined structure domain (CBM) proteins [fig_ref] Figure 6: Comparison of the numbers of unigenes belonging to CAZy families [/fig_ref]. The most representative families were GT41 (38 members), CE10 (31 members) and GH5 (24 members).
There were 37 genes encoding proteins containing the cellulose-binding domain (CBM), which may assist in lignocellulose-degrading enzyme attachment to the surface of cellulose and thus facilitate co-degradation of the natural cellulose-hemicellulose network.
CAZyme-annotated genes in Aspergillus sydowii MS-19 and the brown-rot fungi representative species, Postia placenta and Penicillium decumbens, were compared [fig_ref] Figure 6: Comparison of the numbers of unigenes belonging to CAZy families [/fig_ref]. The total number of CAZymes and the number of CAZymes in the GH and GT classes of Aspergillus sydowii MS-19 were much higher than in the other two.
The CAZymes in Aspergillus sydowii MS-19 related to lignocellulose degradation are listed in [fig_ref] Table 4: CAZymes in Aspergillus sydowii MS-19 related to lignocellulose degradation Locus_2944,Locus_4414,Locus_6027,Locus_7420,Locus_15666,Locus_22469,Locus_24089 1015 reached... [/fig_ref]. There were 17 cellulase (eight cellobiohydrolase and nine endo-1,3-beta-glucanase) and 19 feruloyl esteraseencoding genes. In addition, 13 chitinase genes were annotated, and the chitin-degrading ability of Aspergillus sydowii MS-19 is subject to verification. Remarkably, one sequence annotated as laccase, which can degrade lignin, was found at Locus_6288. The BLAST results showed that the nucleic acid sequence homology between this sequence and the laccase sequences from Coccidioides immitis (EAS35187), Grosmannia clavigera (EFX01145), Aspergillus kawachii (GAA87354), and Aspergillus niger (XP_001389525) was 52%, 40%, 54% and 53%, respectively [fig_ref] Figure 7: Alignment of the amino acid sequence of Aspergillus sydowii MS-19 [/fig_ref].
Three peroxidase sequences sharing a structure similar to typical lignin peroxidase and manganese peroxidase were found and annotated as haem-binding peroxidase, glutathione peroxidase and catalase-peroxidase. Among them, the haem-binding peroxidase annotated sequence (Locus_2354) displayed the highest similarity, and the BLAST results showed a nucleic acid sequence homology of 75% between this sequence and Aspergillus niger CBS 513.88 haem-binding peroxidase [fig_ref] Figure 8: Alignment of the amino acid sequences of Aspergillus sydowii MS-19 haem-binding peroxidase-like... [/fig_ref]. It is speculated that this gene may have different names but similar functionalities.
# Discussion
Polar microorganisms have important value in scientific research as well as in application development since they provide novelty, uniqueness and diversity. An increasing number of people are now focused on the study and application of low-temperature enzymes. In recent years, due to the shortage of energy sources, utilization of lignocellulosic material to produce liquid fuels and chemicals has become an important alternative approach for supporting the sustainable development of human society. Low-temperature lignocellulosic enzymes also show great potential applications in industrial production [bib_ref] Harnessing the potential of ligninolytic enzymes for lignocellulosic biomass pretreatment, Masran [/bib_ref]. In this study, we isolated and identified a fungus, Aspergillus sydowii MS-19, from the Antarctic region that is able to produce the low-temperature enzymes, Lip and Mnp. We also adapted high-throughput sequencing technology to characterize the transcriptome features and genes encoding CAZymes in Aspergillus sydowii MS-19. This work will provide useful information to facilitate our understanding of low-temperature lignocellulosic enzyme production by polar microorganisms, as well as further research and the application of a novel Antarctic Aspergillus sydowii strain MS-19 as a potential source of lignocellulosic enzymes.
RNA-seq reads of Aspergillus sydowii MS-19 were assembled and subsequently clustered into 11,269unigenes, among which most of the sequences were annotated against three databases. The Blast2GO framework was used to analyse the annotation results based on transcriptomic research. Among the unigenes, 3.82% of 11,269 accounted for catabolism processes involved in lignin degradation and glycometabolism. Most of the unigenes generated products inside the cell, whereas a few generated extracellular products, potentially due to the different cultivating conditions. In many lignocellulose-degrading microorganisms, lignocellulosic enzyme secretion issignificantly enhanced following the addition of particular inducers were added [bib_ref] Evidence that linker sequences and cellulose-binding domains enhance the activity of hemicellulases..., Black [/bib_ref] [bib_ref] Factors Involved in the Regulation of a Ligninase Activity in Phanerochaete chrysosporium, Faison [/bib_ref]. Among the MFs (molecular functions) of the annotated sequences, 10.48%, 14.12%, 1.14% were responsible for DNA-binding, ion binding and enzyme binding, respectively, which may be related to gene transcription regulation and protein-folding transportation, and can be used for further studies of the synthesis and regulatory mechanisms of lignocellulosic enzymes [bib_ref] Disruption of ten protease genes in the filamentous fungus Aspergillus oryzae highly..., Yoon [/bib_ref]. The largest portion of annotated unigenes against the KEGG database was associated with metabolic pathways. Enzymes involved in these pathways could metabolize cellulose, hemicellulose, and starch, and they were also carbohydrate-active enzymes.
Cellobiohydrolase and endo-1,4-beta-xylanase ensured effective lignocellulose degradation in Aspergillus sydowii MS-19 [bib_ref] Modeling cellulase kinetics on lignocellulosic substrates, Bansal [/bib_ref] [bib_ref] Xylanases, xylanase families and extremophilic xylanases, Collins [/bib_ref]. Enzymes such as 2-hydrochloric acid halide can degrade the benzene ring. Since lignin is a polymer composed of a phenyl propane structural unit, its expression in Aspergillus sydowii MS-19 suggested a great potential ability to degrade xenobiotic and environmental pollutants, even lignin [bib_ref] Aspergillus enzymes involved in degradation of plant cell wall polysaccharides, De Vries [/bib_ref]. Mariana et al. have reported that Aspergillus sydowii Gc12 isolated from sponge can secrete peroxidase to catalyse loop-opening of benzyl glycidyl ether (a lignin monomer analogue).
An interesting sequence located in Locus_6288 was presumed to perform a similar function to laccase. The sequence identity between Locus_6288 in Aspergillus sydowii MS-19 and laccase in Aspergillus niger ATCC We also observed that three sequences, annotated as haem-binding peroxidase, glutathione peroxidase and catalase-peroxidase, were similar to typical lignin peroxidase and manganese peroxidases, all of which had haem binding sites based on a multi-alignment analysis. This result suggests the presence of peroxidase sequences with different names that perform similar functions in Aspergillus sydowii MS-19. Elena Fernández-Fueyo et al. [bib_ref] Lignin-degrading peroxidases from genome of selective ligninolytic fungus Ceriporiopsis subvermispora, Fernández-Fueyo [/bib_ref] screened all peroxidase-encoding genes associated with lignin degradation in Ceriporiopsis subvermispora. Instead of the LiP and VP gene, they found two other peroxidases with similar functions. There is a relatively conserved tryptophan residue exposed outside the protein tertiary structure in the two peroxidases, which plays an important role in electron transportation during lignin degradation. Whether enzymes encoded by haembinding peroxidase genes participate in Aspergillus sydowii MS-19 lignin degradation remain to be elucidated.
At present, Phanerochete chrysosporium is the most comprehensive studied. It shows the best enzyme activity in ligninase research field. According to the unit definition and the measurement method of ligninase proposed by Archibald [bib_ref] A new assay for lignin-type peroxidases employing the dye azure B, Archibald [/bib_ref] and Gleen [bib_ref] Purification and characterization of an extracellular Mn(II)-dependent peroxidase from the lignin-degrading basidiomycete,..., Glenn [/bib_ref] , the optimized Lip and Mnp activities of Phanerochete chrysosporium may reach 300~1000 U/L, and the optimal enzyme activity temperature is as high as 40~45°C. The novel Antarctic Aspergillus sydowii MS-19 we obtained in this study has low temperature ligninase activity, its optimal enzyme activity is 30°C. The maximum enzyme activity of Lip and Mnp are 182.6 U/ L and 159.7 U/L for Aspergillus sydowii MS-19, which are lower than that of Phanerochete chrysosporium. This can be attributed to the most suitable conditions for enzyme production, influencing factors and mutagenesis for Phanerochete chrysosporium that have been studied thoroughly and higher enzyme activity is obtained. Low temperature enzyme such as Aspergillus sydowii MS-19 has advantages over Phanerochete chrysosporium in bio energy applications field in our study. The production of bio ethanol requires degradation of cellulose and lignin to glucose, followed by the fermentation of glucose by yeast. The reaction temperature of conventional cellulase and ligninase are between 45~65°C, while the temperature required for yeast fermentation is 30°C. Thus the two processes are separate, and with the increase of reaction time, the concentration of glucose is feedback inhibited. It not only affects the production efficiency, but also greatly increases the cost of bioethanol production. Hence, if low temperature cellulase and lignin degradation enzymes is obtained, synchronization fermentation of the two procedures would achieve. This will greatly simplify the production process of bio ethanol with lower cost. Excess glucose can also be used for yeast fermentation medium. Synchronization fermentation can also reduce the concentration of glucose, remove feedback inhibition and improve the yield of ethanol. We have obtained the low temperature cellulase (optimal enzyme activity temperature 30°C). Meanwhile, the isolated strain in this study has the ability of producing low temperature ligninase. For further research, these results will provide a solid theoretical basis for the industrial production of bioethanol, and provide a new way of production. In a word, because of the huge difference of reaction temperature, there is no direct comparability between the strains of Aspergillus sydowii MS-19 and industrialized strains.
# Conclusions
In the present investigation, a fungus, Aspergillus sydowii MS-19, with the potential for lignocellulose degradation was screened out and isolated from the Antarctic region. We measured the growth profile of Aspergillus sydowii MS-19 and found that it could utilize lignin as a sole carbon source. We also verified its ability to synthesize low-temperature lignin peroxidase (Lip) and manganese peroxidase (Mnp) enzymes, the properties of which were also investigated. High-throughput sequencing was employed to identify and characterize the transcriptome of Aspergillus sydowii MS-19. Carbohydrate-Active Enzymes (CAZyme)-annotated genes in Aspergillus sydowii MS-19 were compared with those in the brown-rot fungi representative species, Postia placenta and Penicillium decumbens. One sequence annotated as laccase, which can degrade lignin, and three peroxidase sequences sharing a similar structure with typical lignin peroxidase and manganese peroxidase were obtained. The novel Antarctic Aspergillus sydowii strain MS-19 could be utilized as a potential source of lignocellulosic enzymes.
## Additional files
Additional file 1: . Phylogeny of Antarctic fungal isolates. Neighbour-joining tree showing the relationship between the ITS sequences from 15 Antarctic native isolates and their closest relatives as well as common fungi in Antarctic. The bootstrap values of the neighbor-joining analysis with 1000 replications are shown on the branches. The scale bar represents 0.05 substitutions per amino acid site. The isolated strains can be classified into 4 classes: Eurotiomycetes,
[fig] Figure 2: Phylogeny and growth profile of Aspergillus sydowii MS-19. a Neighbour-joining tree showing the relationship between ITS sequences from MS-19 and its 15 closest relatives. Bootstrap values for the neighbour-joining analysis with 1000 replicates are shown on the branches. The scale bar represents 0.01 substitutions per amino acid site. b The morphological and microscopic characteristics of Aspergillus sydowii MS-19. [/fig]
[fig] Figure 3: Histogram of the gene ontology classification [/fig]
[fig] Figure 4: Histogram of the KOG classification. The x-axis indicates 25 KOG groups, and the y-axis indicates the number of genes annotated in the current group [/fig]
[fig] Figure 5: Histogram of the KEGG classification [/fig]
[fig] Figure 6: Comparison of the numbers of unigenes belonging to CAZy families (a) in Aspergillus sydowii MS-19 and (b) in Aspergillus sydowii MS-19, Postia placenta and Penicillium decumbens. a The numbers of unigenes belonging to particular CAZy families in Aspergillus sydowii MS-19, only CAZy families containing 10 or more unigenes are shown. b Distribution of various CAZymes in Aspergillus sydowii MS-19 (outer ring), Postia placenta (middle ring) and Penicillium decumbens (outer ring). GH, glycoside hydrolase; GT, glycosyltransferase; CE, carbohydrate esterase; PL, polysaccharide lyase; CBM, carbohydrate-binding module [/fig]
[fig] Figure 7: Alignment of the amino acid sequence of Aspergillus sydowii MS-19 (A.sydowii) laccase-like gene with the laccase sequences from four other fungal species. The species selected for the alignment were Coccidioides immitis (EAS35187, C.immitis), Grosmannia clavigera (EFX01145, G claviger), Aspergillus kawachii (GAA87354, A.kawachii), and Aspergillus niger (XP_001389525, A.niger) [/fig]
[fig] Figure 8: Alignment of the amino acid sequences of Aspergillus sydowii MS-19 haem-binding peroxidase-like gene with the same gene from Aspergillus niger Leotiomycetes, Sordariomycetes and Dothideomycetes. All the 18S/ITS sequences of isolated fungus have been submitted to GenBank and the GenBank accession No. can be available in the brackets. (PPTX 76 kb) Additional file 2: Table S2. Annotated unigenes associated with carbohydrate metabolism and xenobiotic biodegradation. Annotated unigenes, pathways and corresponding members number associated with carbohydrate metabolism and xenobiotic biodegradation after the CAZyme annotation of A. sydowii MS-19 transcriptome. (DOCX 18 kb) Abbreviations CAZyme: Carbohydrate-active enzymes; CBM: Carbohydrate-binding modules; CE: Carbohydrate esterase; GH: Glycoside hydrolase; GO: Gene ontology; GT: Glycosyltransferase; ITS: Internal transcribed spacer; KEGG: Kyoto encyclopaedia of genes and genomes; Lac: Laccase; Lip: Lignin peroxidase; Mnp: Manganese peroxidase; PL: Polysaccharide lyase; VP: Versatile peroxidase [/fig]
[table] Table 1: Sampling information and the representative fungal isolates [/table]
[table] Table 2: Lignin-degrading enzymatic activities of the fungal isolatesFig. 1 Enzymatic assay of Lip and Mnp produced by MS-19 at different temperatures and pH values. a Enzymatic activities of Lip and Mnp synthesized by MS-19 at different temperatures. b Enzymatic activities of Lip and Mnp synthesized by MS-19 at different pH values [/table]
[table] Table 3: Output statistics for the sequencing and assembly [/table]
[table] Table 4: CAZymes in Aspergillus sydowii MS-19 related to lignocellulose degradation Locus_2944,Locus_4414,Locus_6027,Locus_7420,Locus_15666,Locus_22469,Locus_24089 1015 reached 53%. However, the enzyme activity assay of Aspergillus sydowii MS-19 in this study did not detect laccase activity. These contradictory results may be due to the small amount and low activity of laccase secreted extracellularly. The laccase activity of Aspergillus sydowii MS-19 could by further analysed by increasing the enzyme concentration. [/table]
|
Image-based phenotyping of cassava roots for diversity studies and carotenoids prediction
Phenotyping to quantify the total carotenoids content (TCC) is sensitive, time-consuming, tedious, and costly. The development of high-throughput phenotyping tools is essential for screening hundreds of cassava genotypes in a short period of time in the biofortification program. This study aimed to (i) use digital images to extract information on the pulp color of cassava roots and estimate correlations with TCC, and (ii) select predictive models for TCC using colorimetric indices. Red, green and blue images were captured in root samples from 228 biofortified genotypes and the difference in color was analyzed using L*, a*, b*, hue and chroma indices from the International Commission on Illumination (CIELAB) color system and lightness. Colorimetric data were used for principal component analysis (PCA), correlation and for developing prediction models for TCC based on regression and machine learning. A high positive correlation between TCC and the variables b* (r = 0.90) and chroma (r = 0.89) was identified, while the other correlations were median and negative, and the L* parameter did not present a significant correlation with TCC. In general, the accuracy of most prediction models (with all variables and only the most important ones) was high (R 2 ranging from 0.81 to 0.94). However, the artificial neural network prediction model presented the best predictive ability (R 2 = 0.94), associated with the smallest error in the TCC estimates (root-mean-square error of 0.24). The structure of the studied population revealed five groups and high genetic variability based on PCA regarding colorimetric indices and TCC. Our results demonstrated that the use of data obtained from digital image analysis is an economical, fast, and effective alternative for the development of TCC phenotyping tools in cassava roots with high predictive ability. (2022) Image-based phenotyping of cassava roots for diversity studies and carotenoids prediction. PLoS ONE 17(1): e0263326. https://doi.org/ 10.adaptation to different edaphoclimatic conditions and satisfactory yields, even in soils with low fertility[1,2]. Nigeria, with 59.19 million tons cultivated on 7.21 million hectares, is the world's largest producer, while Brazil ranks fifth with production of 17.49 million tons in an area of 1.19 million hectares, and average yield of 14.70 t ha -1[3].Cassava roots are basically composed of carbohydrates, making them an important source of energy for millions of people around the world[4]. The composition of the roots includes cyanogenic glycosides (linamarin and lotaustralin), present in concentrations ranging from 9.90 to 3927.0 μg g -1[5][6][7], and low nutritional quality, with considerably reduced values of proteins, minerals (Zn, Fe, N, Ca, P, K, and Mg) and vitamins (B1, B2, B3, C, and β-carotene)[7,8].The increase in nutritional content through biofortification is a strategy that has been adding nutritional value to cassava varieties[9]. Biofortification is an international and interdisciplinary initiative that seeks to reduce human malnutrition by increasing the concentration of micronutrients in various staple crops, such as increasing the carotenoids content in cassava[9,10]. In cassava roots, the aim is to increase the β-carotene content (precursor of vitamin A), with the main purpose of combating hidden hunger caused by vitamin A deficiency[9,11]. In addition, biofortification programs seek to adjust the levels of other attributes such as high agronomic performance[12], low levels of cyanogenic compounds and high levels of dry matter in the roots, in order to meet the preferences for sweet cassava market[5,9].Despite important advances in increasing the carotenoids content obtained in several conventional breeding programs[5,13,14], phenotyping to quantify the total carotenoids content (TCC) or β-carotene is quite sensitive, time-consuming, tedious, and costly. Therefore, new rapid phenotyping methods are needed, since biofortification programs require the screening of hundreds of samples in a short period of time[5,10].Recently, predictions based on near-infrared spectroscopy (NIRS) have demonstrated high potential for indirect phenotyping of carotenoid content in cassava[5,11,[15][16][17]. In addition, methodologies based on digital image analysis allow extracting color information [18] due to the strong correlation between digital and visual data[19]. Since the present carotenoids have a strong correlation with the intensity of the yellow color[14], it is assumed that this type of phenotyping is a viable and adequate option for the quantification of the carotenoid content.The main advantages of image-based root phenotyping are savings time, use of commercially available digital cameras that are easy to handle, transport and use open-source software for image processing, demonstrating high potential for use in plant breeding programs [19. 20]. The recorded images also allow researchers to go back and re-examine them whenever doubts arise about the phenotyping process[18]. Furthermore, once the prediction models are calibrated, it is possible to reduce the number of laboratory samples, concentrating efforts only on those of greatest interest, and thus, greatly reducing the cost of the analyses necessary for the selection of genotypes.Rapid imaging phenotyping has already been widely used to study the relationship between color indices extracted from images and carotenoid content in different crops such as carrots (Daucus carota)[21], tomatoes (Solanum lycopersicum L.)[22], olive oil (Olea europaea)[23]and mandarin (Citrus reticulata Blanco)[24]. Typically, images are acquired by a digital device and saved in the three-dimensional red, green and blue (RGB) color space; however, this space is not uniform in percentage and does not represent colors naturally perceived by human vision[25]. Some studies have opted to convert RGB color matrices into International Commission on Illumination (CIELAB) codes[26], to correlate image colors with some specific attribute in plants[27][28][29]. The CIE [26] color model is a predominant choice among researchers [25] because it has a color space with a standardized measurement technique [30]. The CIELAB color space [26] proved to be quite accurate in estimating the carotenoids content PLOS ONE Carotenoid prediction in cassava PLOS ONE | https://doi.org/10.
# Introduction
Cassava (Manihot esculenta Crantz), belonging to the Euphorbiaceae family, is a globally cultivated species, due to its advantageous traits such as tolerance to water deficit, with wide a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Data Availability Statement: All relevant data are within the paper and its Supporting Information files. based on digital images in tomato (Solanum lycopersicum) and in pollen of species from the Brassicaceae, Myrtaceae and Fabaceae families [bib_ref] Rapid determination of lycopene content and fruit grading in tomatoes using a..., Ye [/bib_ref].
In other species, high predictive ability for traits associated with nutritional quality has been observed when using colorimetric indices obtained from digital images. For example, in mandarin (Citrus reticulata Blanco), partial least squares (R 2 = 0.96) and third-order linear regression (R 2 = 0.95) models were able to predict fruit maturation (citrus color index) with high precision [bib_ref] Development of a computer vision system to estimate the colour indices of..., Hadimani [/bib_ref]. In olive oil (Olea europaea), the support vector machine (R 2 = 0.96) and artificial neural networks (R 2 = 0.94) models also showed high predictive ability for carotenoid content in this species [bib_ref] Fusion of dielectric spectroscopy and computer vision for quality characterization of olive..., Sanaeifar [/bib_ref]. Additionally, a multiple linear regression model was able to predict the carotenoids content with high precision (R 2 = 0.89) in pollen of species from the Brassicaceae, Myrtaceae, Fabaceae families.
Considering the need to develop high-performance phenotyping methodologies to optimize the selection of biofortified sweet cassava in the breeding programs, the present study aimed to: (i) evaluate the potential of using digital images to capture information on the pulp color of cassava roots and estimate correlations with TCC; (ii) select predictive models powered by colorimetric indices that have high precision and predictive ability to obtain TCC.
# Material and methods
## Plant material and experimental design
The calibration set was composed of 228 genotypes from the Cassava Germplasm Bank of Embrapa Mandioca e Fruticultura, Cruz das Almas, Bahia, Brazil (Latitude: 12˚40' 12" S, Longitude: 39˚06' 07" W and Altitude: 220m). This germplasm panel is composed of genotypes with high diversity for carotenoid content, cyanogenic compounds, dry matter content of the root and several yield attributes. The clones were cultivated from June 2019 to July 2020 at the Experimental Station of Embrapa Cassava e Fruticultura, using an augmented block design, with 34 local and improved varieties as common controls. The experimental plot consisted of two lines with 10 plants each , with a spacing of 0.90 m between lines and 0.80 m between plants. Planting was carried out with 16 cm stakes in rainfed condition, following the recommendations and agricultural practices for cassava crop [bib_ref] Exigências edáficas da cultura da mandioca, Souza [/bib_ref]. To compose the calibration set, clones with cream and yellow root pulp coloration were previously selected.
The climate in the germplasm growing region is tropical, hot, and humid, without a dry season. With an average annual rainfall of 1,170 mm, the wettest months are from March to August and the driest ones from September to February. The average annual temperature is 24.5˚C, with an average of 80% relative humidity. The predominant soils in the region are Dystrophic Yellow Latosols.
## Obtaining and processing images
A box (dimensions 24×18×19 cm), with white inner walls and the bottom lined with blue plastic material to increase the precision of the separation of the pixels from the root, was used as a phenotyping platform. We used 6500k frequency (Cool White Light) 9W lamps, placed equidistantly inside the box. To obtain the digital images, a tablet with an Android 9.0 system and an 8MP camera was attached to the top of the box. The digital images were captured in an environment with controlled temperature (22˚C±2˚C) to prevent carotenoid degradation.
During harvest, 4-6 commercial roots (>5 cm in diameter and around 12-20 cm in length) of all 228 genotypes were selected. The roots were washed with running water to remove excess adhered soil; the central region was then cut into pieces approximately 2 cm thick. Six pieces of root per genotype were placed in the center of the camera's field of view and three RGB images were captured. Images were saved in Joint Photographic Experts Group (JPEG) format with a resolution of 2448 × 3264 pixels. The images were pre-processed using Gaussian Blur filters to remove the background with the aid of ImageJ software version 1.52a. Colorimetric analysis was performed using the Tomato Analyzer-Color Test version 4.0 software [bib_ref] Tomato analyzer-color test:a new tool for efficient digital phenotyping, Darrigues [/bib_ref]. The initial values of the chromatic components (L � , a � and b � ) of the CIELAB systemwere obtained using the standard calibration of the software. This calibration was performed assuming values of 1.0 for the slope and 0 for the intercepts of a linear regression equation. The excess root matter was processed with the aid of a food multiprocessor and reserved for TCC analysis.
## Root color evaluation
The differences in root color between the different cassava genotypes were evaluated using the CIELAB parameters, which were obtained from the RGB values. The RGB conversion into L � , a � and b � was performed in three steps [bib_ref] Tomato analyzer-color test:a new tool for efficient digital phenotyping, Darrigues [/bib_ref].
1. Initially, the RGB values were scaled to a uniform color space using the following equations: 2. The scaled RGB values were converted to XYZ tristimulus values using the following equations:
[formula] Var R ¼ ð ð R 255 Þ þ 0:055 � � � � 2:4 Þ=1:055Þ � 100 Var G ¼ ð ð G 255 Þ þ 0:055 � � � � 2X ¼ ðVar R � 0:4124Þ þ ðVar G � 0:3576Þ þ ðVar B � 0:1875Þ Y ¼ ðVar R � 0:2126Þ þ ðVar G � 0:7152Þ þ ðVar B � 0:0722Þ Z ¼ ðVar R � 0:0193Þ þ ðVar G � 0:1192Þ þ ðVar B � 0:9505Þ 3. [/formula]
The XYZ values were converted to L^�, a^�, and b^� values using the following equations: The L � coordinate indicates the lightness, from dark (0) to light (100), of the colour, while a � and b � indicate the variation from green to red (-60 to 60) and from blue to yellow (-60 to 60), respectively. For each sample, each record represents an average of six measurements. The cylindrical coordinates, chroma and hue, were calculated from a � and b � . Hue is an angular measurement between 0˚and 360˚that represents the base color. Its value was calculated based on the following equations:
[formula] L � ¼ 116f Y Yn � � À 16 a � ¼ 500½f X Xn � � À fð Y Yn Þ� b � ¼ 200½f Y Yn � � À fð Z Zn Þ� wheref q ð Þ ¼ q 1=3 if q > 0Hue ¼ 180 pi �q � � � � for a� < 0: [/formula]
The chroma value is the relative saturation, represented by the mean of a � and b � . The chroma was calculated using the following equation: chroma ¼ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffiffi
[formula] ða�Þ 2 þ ðb�Þ 2 q [/formula]
The lightness was also calculated, defined by the average amount of lightness (ranging from 0 to 240) in all pixels. Luminance accounts for the human eye's varying sensitivity to radiation at various wavelengths and defines brightness. The lightness was estimated from the RGB value of each pixel using the expression:
[formula] Lightness ¼ ½maximumðR; G; BÞ þ minimumðR; G; BÞ � 240� 2 � 255 [/formula]
## Total carotenoid content analysis
After capturing the root imagens, two samples were collected for the TCC analysis, each containing 10 g, 15 g or 25 g of the crushed roots (depending on the intensity of the pulp color), in addition to a backup sample of 60 g. The samples were placed in glass jars with lids and covered with aluminum foil to avoid contact with light as much as possible, to prevent degradation of the carotenoids. The samples were then frozen for further analysis.
To quantify the carotenoid content, the procedure described by the HarvestPlus Handbook for Carotenoid Analysiswas adopted. The pigments from the crushed roots were extracted by grinding the samples in an Ultra Turrax homogenizer for three minutes, in order to break down the plant cells to then allow the carotenoids' extraction with approximately 50 mL of acetone. The sample and acetone mixture were filtered through a Buchner funnel with the aid of a vacuum pump, and the residue retained in the funnel was washed with acetone until it did not show any color. The extract containing only the pigment and acetone was reserved in the suction flask, then transferred to a separating funnel containing petroleum ether (of which the quantity varied depending on the color of the sample); approximately 250mL of saline solution was added slowly to induce phase separation. The aqueous phase was discarded and the saline washing procedure repeated five times until only petroleum ether and pigment remained. This new extract was filtered through a funnel with glass wool and anhydrous sodium sulfate into an amber volumetric flask, where its volume was supplemented by petroleum ether. An aliquot of each sample was taken for determination by a spectrophotometer (UV-Vis Thermo Scientific, Genesis 10S model), adjusted for an absorbance at 450 nm.
The TCC was calculated using the following formula:
[formula] TCC ug:g À 1 ð Þ ¼ A � VðmlÞ � 10 4 A 1% 1cm � PðgÞ ; [/formula]
where A is the absorbance, V(ml) is the total extract volume in milliliters, P(g) is the sample weight in grams, and A 1% 1cm is equal to the extinction coefficient of β-carotene in petroleum ether (2592).
# Data analysis
Initially, Pearson's correlation analysis was performed between the phenotypic data from the TCC and the colorimetric data from the CIE color system in order to verify the magnitude and direction of the correlation between these parameters. PCA was implemented to verify the phenotypic diversity of genotypes based on TCC and colorimetric indices, as well as to verify the importance of colorimetric indices as predictors of TCC. The grouping of genotypes was determined by the K-means clustering algorithm. These analyses were all performed using the factoextra packageimplemented in the R version 4.03 programming environment.
Then, the colorimetric data were used to develop models to predict TCC. Twelve predictive models derived from regression and machine learning were tested: Linear Regression with [bib_ref] A review of studies on machine learning techniques, Singh [/bib_ref] [bib_ref] A few useful things to know about machine learning, Domingos [/bib_ref]. The predictive models were implemented in the caret package (Kuhn, 2008) from R software version 4.03.
To test the accuracy of the predictive models, we used different cross-validation schemes: 1) random cross-validation without test set (V-Random), where genotypes were randomly split 80/20% into training and validation, respectively; 2) clustering-based, where PCA clustering with k = 5 (IV-Cluster) clusters were assigned to the split sets in order to have a group of clusters in training, validation and testing with no further criteria; respectively; and 3) random cross-validation with test set (IV-Random): where genotypes were randomly split 60/20/20% into training, validation, and test sets respectively. For the EV-Cluster strategy, we considered the population structure of genotypes determined by PCA and k-means clustering. The training and validation sets were composed of genotypes from all clusters, except the cluster used as a test set. For example, in the independent validation of the first cluster, the accessions belonging to the second, third, fourth and fifth clusters were used for training and validation. In all cross-validation schemes, the training and validation sets were formed considering 5-fold increase and 6 repetitions. The process was randomly repeated 30 times.
The performance of the predictive models was evaluated based on the root-mean-square error (RMSE) and the coefficient of determination (R 2 ) values, obtained in each cross-validation partition. The RMSE measures the average magnitude of the estimated errors; the closer its positive value is to 0, the higher the quality of the estimated values. It is calculated according to the following equation:
## Rmse ¼
ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi 1 n
[formula] P n i¼l ðE i À O i Þ 2 r ; [/formula]
where E i and O i are the estimated and observed values, respectively, and n is the number of observations. By contrast, R 2 varies between 0 and 1, and the closer to 1, the better its explanatory power. R 2 is described as the ratio that represents the proportion of the total variation of the dependent variable that is explained by the variation of the independent variable. It is estimated by the following equation:
[formula] R 2 ¼ P n i¼l ðb y i À yÞ 2 P n i¼l ðy i À yÞ 2 ; [/formula]
where P n i¼l ðb y i À yÞ 2 corresponds to the explained variation, and P n i¼l ðy i À yÞ 2 represents the variation not explained by the model. The mean absolute percentage error (MAPE) was also used to estimate the accuracy of the TCC predictions. MAPE was estimated using the following equation:
[formula] MAPE ¼ 1 n X n tÀ 1 j y t À b y t y t j [/formula]
where y t is the observed value, ŷ t is the predicted value, and n is the number of fitted values.
# Results
## Correlation between total carotenoid content and colorimetric indices
The TCC ranged from 0.30 μg g -1 to 13.67 μg g -1 based on the fresh weight of the samples, while the overall mean of the 228 cassava genotypes was 3.56 μg g -1 . Scatter plots were constructed to assess the distribution of data and estimate the Pearson correlation coefficients between the TCC and the colorimetric indices extracted from the digital images (Fig 1, S1 . A strong positive correlation was identified between TCC and the variables b � (r = 0.90) and chroma (r = 0.89), while moderate and negative correlations were identified between TCC and the indices a � (r = -0.53), hue (r = -0.59), and lightness (r = -0.66). On the other hand, the L � parameter was not found to be significantly correlated with TCC.
## Phenotypic diversity and clustering of cassava genotypes based on total carotenoid content and colorimetric indices
Population structuring based on the TCC dataset and colorimetric indices was evaluated using PCA. According to the K-means clustering algorithm, five groups were identified based on the sum of squares within groups [fig_ref] Fig 2: Within-group sum of squares of 228 biofortified cassava genotypes based on phenotypic... [/fig_ref]. The first two main components explained 89% of the total variance of the sample population dataset and, therefore, allow a reasonable representation of the cassava genotype clustering based on the evaluated parameters [fig_ref] Fig 3: Principal Component Analysis [/fig_ref]. There was wide phenotypic variation among cassava genotypes for both TCC and colorimetric indices. The TCC and the colorimetric indices b � and chroma were the variables that contributed most to the dispersion of the genotypes. Similarly to the findings of the correlation analyses, the b � and chroma indices were those that showed the greatest association with TCC, and therefore have high relative importance for the prediction of this phenotype. On the other hand, the L � index contributed the least to the phenotypic variation of the cassava genotypes.
Based on PCA, the cassava genotypes were categorized into five groups. Groups 1, 2, 3, 4 and 5 comprised 76, 34, 26, 21 and 71 genotypes, respectively. Groups 1 and 2 consisted of genotypes with low TCC values (1.91 μg g -1 and 1.82 μg g -1 , respectively) and low b � and chroma, in addition to the highest L � , lightness and hue [fig_ref] Fig 4: Boxplot of total carotenoid content and colorimetric indices for each cluster of... [/fig_ref]. Groups 3 and 5 consisted of cassava genotypes with yellow pulp, but with moderate TCC contents (4.82 μg.g -1 and 3.60 μg.
## Prediction of total carotenoid content using colorimetric indices
The performance of predictive models for TCC in cassava roots using colorimetric data obtained from digital images was evaluated considering the complete model (all variables used for TCC prediction) and the reduced model, in which only the variables with relative importance greater than or equal to 50% for the TCC prediction were retained in the different models evaluated.
The colorimetric indices b � and chroma, each with a relative importance of 100%, were the variables that contributed most to the prediction of TCC in the majority of models [fig_ref] Fig 5: Relative importance of colorimetric indices for predicting total carotenoid content using 12... [/fig_ref]. The exception was the PLS model, in which the b � and a � indices each had a relative importance of 100%. The lightness and hue indices achieved importance between 50% and 70% depending on the model under consideration. For all models, the L � index was below the 50% importance threshold and was hence excluded from the reduced models. Therefore, most of the reduced PLOS ONE models included the variables a � , b � , chroma, hue, and lightness, with the exception of the RF, whose reduced model was represented only by b � , hue and chroma. Despite the reduced models containing only the variables with the greatest predictive ability for TCC, there were no differences in their predictive abilities compared to the complete model ; S1 Fig). Only the RF and PLS models showed a small difference in prediction accuracy, with R 2 = 0.91 and 0.90 (RF) and R 2 = 0.94 and 0.93 (PLS) in the complete and reduced models, respectively.
Regardless of the cross-validation strategy, the predictive ability between the complete and reduced models was similar (p>0.05) for all prediction models, showing that the selection of more important features did not negatively affect the model's performance
[formula] S1 Fig). [/formula]
The error associated with TCC prediction was low (0.04 � MAPE � 0.13) in most models using V-Random cross-validation. The BL, BBL and LRFS models had the lowest MAPE values (0.04) using the reduced model, while the CART model had higher MAPE (0.13) in both the complete and reduced models. In general, the use of reduced models did not negatively affect the prediction accuracy of the TCC values considering that MAPE values were less than or equal to those obtained for most complete models. In the BBL, BL, LRFS and SVM models, there was a slight improvement in accuracy when using the reduced model compared to the complete model.
In general, the accuracy of the predictive models (both complete and reduced) was high, with R 2 ranging from 0.90 (SVM) to 0.94 (ANN) Although the RF and SVM models had high predictive abilities (R 2 = 0.90 for both complete and reduced models), they also had higher RMSE values (0.31 and 0.32 for complete and reduced model, respectively (RF) and 0.33 for both complete and reduced (SVM).
The performance of two additional cross-validation schemes with an independent sample was also analyzed to provide an unbiased evaluation of the goodness fit of the training set. The IV-Cluster and IV-Random routines were performed using the complete and reduced models [fig_ref] Fig 2: Within-group sum of squares of 228 biofortified cassava genotypes based on phenotypic... [/fig_ref]. In general, IV-Random exhibited higher accuracy than IV-Cluster. The accuracy of IV-Random considering the complete and reduced models was similar (0.90 and 0.94), except for the model CART, which had the lowest R 2 value (0.81). On the other hand, the R 2 value to the cross-validation scheme (IV-Cluster) ranged from 0.23 (SVM) to 0.71 (LRFS, LRBS, RR, PLS and BBL).
For the IV-Cluster, the SVM model had the lowest accuracy for both complete (R 2 = 0.23) and reduced models (R 2 = 0.29). Regarding the RMSE, the prediction models of the V-Random and IV-Random strategies showed a best fit for both complete and reduced models (range from 0.24 to 0.43), while the IV-Cluster was the worst one (range from 0.29 to 0.97) [fig_ref] Fig 2: Within-group sum of squares of 228 biofortified cassava genotypes based on phenotypic... [/fig_ref]. The cross-validation scheme IV-Cluster had lower MAPE values compared to IV-Random and V-Random . The MAPE values of the IV-Random were quite similar between the complete and reduced models (0.05 � MAPE � 0.14), while for the IV-Cluster strategy there was only a small MAPE reduction in the reduced model (0.14 � MAPE � 0.77) compared to the complete model (0.13 � MAPE � 0.92) for some prediction models such as RF and SVM.
For both V-Random and IV-Random schemes, the scatterplot of observed and predicted TCC values clearly shows the good fit of the ANN model (R 2 = 0.94) followed by the BBL, BL, GLMSS, LRBS, LRFS, LRSS, PLS, and RR models (R 2 = 0.93) regardless of the fold validation [fig_ref] Fig 6: Relationship between observed and predicted values for total carotenoid content of cassava... [/fig_ref]. On the other hand, the CART (R 2 = 0.81 to 0.82), RF and SVM (R 2 = 0.90) models exhibited greater dispersion of the predicted data and, therefore, less adjustment for the TCC prediction in cassava roots. In general, for most models, the TCC predictions with values between 0.50 and 5.0 μg g -1 were reasonably accurate. However, above this value, the
## Plos one
regression graphs show a greater dispersion of data. One possible explanation is the observation that 82% of genotypes had a TCC below 5.0 μg g -1 , thus representing most of the samples.
In general, the IV-Cluster scheme exhibited the worst prediction parameters regardless the model used , [fig_ref] Fig 8: Relationship between observed and predicted values for total carotenoid content of cassava... [/fig_ref]. The best prediction fits were identified in the BBL, LRBS, LRFS, PLS, and RR models (average R 2 = 0.71), followed by the GLMSS and LRSS models (average R 2 = 0.69). On the other hand, the SVM and CART models presented predictions below 0.30 with the greatest dispersion in the predicted values and the worst prediction adjustment.
In the IV-Cluster scheme, there was wide accuracy variation of the models for predicting the 5 PCA-clusters [fig_ref] Fig 8: Relationship between observed and predicted values for total carotenoid content of cassava... [/fig_ref]. The worst prediction scenario occurred when the genotypes of clusters 1, 2, 3 and 5 were used as the training population (low to medium TCC contents) for the prediction of cluster 4 genotypes (genotypes with high TCC), regardless of the prediction model used. This was possibly due to the fact that the information for predicting genotypes with more than 5 ug g -1 was not incorporated in the training population, thus limiting the predictive capacity in the test population.
# Discussion
## Variation and correlation between total carotenoid content and colorimetric indices
The TCC in the 228 cassava genotypes analyzed ranged from 0.30 μg g -1 to 13.67 μg g -1 . This variation in the TCC is consistent with other reports on cassava, such as the screening of 1789 cassava genotypes in the International Center for Tropical Agriculture (Colombia), whose
## . performance of different prediction models for total carotenoid content in cassava roots using colorimetric indices obtained from digital images considering the complete model (all variables) and reduced model (variables with more than 50% relative importance), using the random cross-validation without test set (v-random), pca clustering-based cross-validation (iv-cluster), and random cross-validation with test set (iv-random).
## Model a v-random iv-cluster iv-random
## Plos one
TCC varied from 1.02 to 10.40 μg g -1 [bib_ref] ariation of quality traits in cassava roots evaluated in landraces and improved..., Chávez [/bib_ref]. However, in breeding populations specifically aimed at increasing TCC, this variation may be even wider, such as the variation from 0.2 μg g -1 to 25.5 μg g -1 observed by [bib_ref] Prediction of carotenoids, cyanide and dry matter contents in fresh cassava root..., Sá Nchez [/bib_ref].
The variation observed in the TCC constitutes a wide range in which to build predictive models with high accuracy for this trait. Therefore, this study analyzed the potential use of
## Plos one
rapid phenotyping based on digital images and the ability to estimate TCC in cassava roots from colorimetric data using the CIELAB color system [bib_ref] Correlation between L* a* b* color space values and carotenoid content in..., Itle [/bib_ref].
The presence of carotenoid pigments contributes to the intensity of the pulp color of cassava roots [bib_ref] Carotenoids in cassava roots, Ceballos [/bib_ref] [bib_ref] Rapid cycling recurrent selection for increased carotenoids content in cassava roots, Ceballos [/bib_ref] , providing a positive correlation between these two traits. Thus, first, the relationship between TCC and variables related to color was evaluated, with a strong
## Plos one
a tendency to increase b � , as verified in several Kinnow mandarin varieties (Citrus reticulata Blanco) [bib_ref] Development of a computer vision system to estimate the colour indices of..., Hadimani [/bib_ref] and in extruded corn samples [bib_ref] Carotenoid and color changes in traditionally flaked and extruded products, Cueto [/bib_ref]. In cassava, it has also been found that higher b � values were associated with cassava samples with higher carotenoid content. This behavior is due to the correlation between the b � parameter and β-carotene [bib_ref] Reflectance colorimetry as asimple method for estimating carotenoid content in maize grain, Kljak [/bib_ref] , since β-carotene is the most abundant carotenoid in cassava [bib_ref] Carotenoids in cassava roots, Ceballos [/bib_ref].
The same pattern of correlation was observed between TCC and chroma (R = 0.89) [fig_ref] Fig 1: Scatter plots of the total carotenoid content and colorimetric indices [/fig_ref]. The chroma parameter indicates the color intensity; therefore, the higher the TCC, the stronger the color intensity of the cassava pulp. Similar correlations have been found between b � and chroma when analyzing the individual carotenoids content and TCC in pumpkins (Cucurbita spp.) [bib_ref] Correlation between L* a* b* color space values and carotenoid content in..., Itle [/bib_ref]. In addition, these authors observed that the hue angle was negatively correlated with TCC (r = -0.83), similarly to what was observed in the present study in cassava (r = -0.59), suggesting that as the angles decrease, carotenoid concentrations increase. Furthermore, the authors of [bib_ref] Correlation between L* a* b* color space values and carotenoid content in..., Itle [/bib_ref] identified strong correlations between colorimetric indices and carotenoid content, and stated that these correlations can be useful in indirect selection for high carotenoid content in improved pumpkin populations, being simultaneously easy to implement and low cost.
## Clustering the phenotypic diversity of total carotenoid content based on colorimetric indices
The two main components explained almost 90% of the phenotypic variance, and thus were able to demonstrate the dispersion of cassava genotypes and the formation of five groups with similar characteristics in terms of TCC and colorimetric indices. Specifically, groups 3 and 4 gathered accessions with yellow pulp and consequently higher levels of TCC, b � and chroma, since these two colorimetric indices were positively correlated with TCC.
The potential for clustering genotypes based on colorimetric indices has been recently analyzed in several species, including in cassava. The authors ofevaluated the clustering potential of different samples of cassava roots based on CIELAB color space indices, obtained by means of a colorimeter. The authors found that samples with higher TCC had higher b � values. They also reported clustering of genotypes via PCA based on root pulp color and TCC, using colorimetric data.
In [bib_ref] Classification tools for carotenoid content estimation in Manihot esculenta via metabolomics and..., Moresco [/bib_ref] , researchers used PCA and other clustering methods from the spectrophotometric data matrix applied in the UV-Vis region (400-500nm) in cassava root samples with cream, yellow and reddish pulp colors. In this strategy, the first two components represented 99.97% of the variance, clearly revealing three groups according to the carotenoid contents. These authors attribute the grouping of genotypes to the carotenoids content and the discrepancy of the reddish pulp genotype to the presence of lycopene in relevant quantities, detected by chromatographic analysis.
The efficiency of PCA in distinguishing different samples was also described in [bib_ref] Application of high-throughput phenotyping tool Tomato Analyzer to characterize Balkan Capsicum fruit..., Nankar [/bib_ref] , which reported the morphometric and colorimetric diversity of fruits from the Balkan pepper collection (Capsicum spp.) and performed grouping via PCA. One of the characteristics evaluated that most contributed to the variability of peppers was the fruit color, estimated by the CIELAB parameters, obtained by the Tomato Analyzer software. The similarities between pomegranate (Punica granatum L.) varieties in antioxidant activity and physiochemical fruit properties have also been reported . These authors verified the significant contribution to the PCA grouping of the a � index measured in the fruit juice and noted its correlation with other important attributes for the crop, such as the total phenolic content and the total anthocyanin content.
## Performance of tcc prediction models
We used different cross-validation approaches (with and without population testing and PCA clustering) to verify the predictions' accuracy. The main findings showed that it is necessary to include the large phenotype amplitude in the training and validation populations, to enable obtaining highly accurate predictions, regardless of the model used. Therefore, cluster-based and cross-validation approaches to genotypes with contrasting TCC did not fulfill this premise, in general tending to exhibit low prediction accuracies.
The various prediction models based on machine learning showed high predictive ability for TCC in cassava roots, indicated by the high R 2 and low RMSE values. In particular, the model based on ANN was the most accurate, providing a better fit for RMSE (0.24) and R 2 (0.94). Similar results were also observed in [bib_ref] Fusion of dielectric spectroscopy and computer vision for quality characterization of olive..., Sanaeifar [/bib_ref] , which identified the high predictive ability of carotenoid content in olive oil (Olea europaea) using ANN (R 2 = 0.94) and SVM (R 2 = 0.96). In mandarins (Citrus reticulata Blanco), the PLS (R 2 = 0.96) and third-order linear regression (R 2 = 0.95) models were able to predict ripeness (citrus color index) with high accuracy [bib_ref] Development of a computer vision system to estimate the colour indices of..., Hadimani [/bib_ref]. Additionally, a methodology has been presented for predicting the individual carotenoid content from digital image analysis parameters of pollen samples of species from the Brassicaceae, Myrtaceae, and Fabaceae families, based on multiple linear regression, with high R 2 (ranging from 0.76 to 0.89).
ANN models use techniques inspired by the brain and the manners in which it learns and processes information . The ANN model, like the others, uses supervised learning, which aims to provide the prediction of an output variable according to known input variables . Machine learning models are an efficient tool and have been widely used in different agriculture approaches to unravel, quantify, and understand data-intensive processes . However, the predictive ability depends on the dataset under analysis; for this reason, it is interesting to evaluate different approaches in order to explore differences in algorithms that may benefit the predictions .
The search for high-throughput phenotyping methodologies has been a priority in cassava breeding programs. Therefore, several groups have been searching for alternatives to shorten selection cycles, increase the number of evaluated genotypes, and reduce costs associated with the phenotyping that are difficult to measure, such as carotenoid content. Even with the implementation of rapid phenotyping methodologies, few studies have been dedicated to the validation of models that predict traits associated with nutritional quality, such as dry matter content and carotenoids in cassava, based on image analysis. Models with good predictive accuracy can facilitate early selection of only the most interesting genotypes, allowing selection optimization.
The use of easy-to-measure predictive variables, such as colorimetric data extracted from digital images, is a key factor in accelerating the evaluation of genotypes in breeding programs aimed at screening thousands of individuals annually. This approach demonstrated that the variables b � and chroma were strongly correlated with TCC in cassava roots [fig_ref] Fig 1: Scatter plots of the total carotenoid content and colorimetric indices [/fig_ref]. This result makes it possible to further simplify the phenotyping protocol for TCC in cassava roots using digital images, allowing for the reduction of variables without the loss of predictive ability.
Previous authors have also used colorimetric techniques to predict TCC in cassava roots. It has been demonstrated that color measurement from colorimetric data using the CIELAB color system can be used as a fast and non-destructive method to calibrate the TCC of roots with acceptable prediction error. The authors used CIELAB data and TCC data determined by spectrophotometry as input to various machine learning models and found R [bib_ref] Model-assisted phenotyping by digital images in papaya breeding program, Cortes [/bib_ref].60. Similarly, the authors of [bib_ref] Classification tools for carotenoid content estimation in Manihot esculenta via metabolomics and..., Moresco [/bib_ref] used the colorimetric technique associated with UV-Vis/ HPLC to predict TCC in cassava roots, finding R 2 > 0.40 and RMSE < 9.99 for the PLS, SVM, and Elastic Net models. Therefore, the approaches used in this study allowed the obtention of greater predictive power and smaller error compared to similar reports in the literature. A possible explanation for this is the fact that colorimeters were used inand [bib_ref] Classification tools for carotenoid content estimation in Manihot esculenta via metabolomics and..., Moresco [/bib_ref] to obtain the CIELAB colorimetric indices. It is speculated that digital images allow for greater precision in color analysis, as they capture all the pixels of root images, while colorimeters measure only a specific portion of the sample. The software used in the present study for analyzing RGB images, Tomato Analyzer, has also showed high potential for quantifying characteristics that are difficult to measure by conventional phenotyping methods, such as fruit wall thickness and pericarp thickness in pepper fruits (Capsicum spp.) evaluated in germplasm collected in different regions of the Balkans [bib_ref] Application of high-throughput phenotyping tool Tomato Analyzer to characterize Balkan Capsicum fruit..., Nankar [/bib_ref]. The Tomato Analyzer allows the estimation of a large number of characteristics associated with the shape and colors derived from digitized images of fruit sections or even roots, as was the case in the current study.
In general, the development of predictive models based on different approaches has shown its great utility and reliability in estimating TCC in cassava. One of the first studies to predict the carotenoids content in cassava was developed based on NIRS data, resulting in predictive abilities of 0.92 and 0.93 for TCC and β-carotene, respectively, using PLS regression [bib_ref] Prediction of carotenoids, cyanide and dry matter contents in fresh cassava root..., Sá Nchez [/bib_ref]. In another study, also using NIRS data, the practical use of carotenoid content predictions for selection in species breeding programs was demonstrated for the first time [bib_ref] High-throughput phenotyping and improvements in breeding cassava for increased carotenoids in the..., Belalcazar [/bib_ref]. Based on the use of LOCAL regression, these authors reported R 2 = 0.74 for predictions of the β-carotene content. More recently, NIRS data has been combined with different predictive models [bib_ref] Genetic correlation, genome-wide association and genomic prediction of portable NIRS Predicted carotenoids..., Ikeogu [/bib_ref] , such as the modified partial least squares and RF approaches. These last authors reported a high predictive ability for β-carotene content (R 2 = 0.99).
In the present study, phenotyping via digital images also demonstrated its high predictive ability (R 2 = 0.94 and RMSE = 0.24) for TCC prediction based on the ANN model. Therefore, this offers a new approach for measuring this trait in cassava, with much lower costs in the acquisition and maintenance of equipment. Already, the implementation of a computer vision system based on RGB images to detect colorimetric characteristics in tomatoes, using low-cost equipment, has showed excellent classification capacity . An application has also been developed [bib_ref] Rapid determination of lycopene content and fruit grading in tomatoes using a..., Ye [/bib_ref] to quickly determine the lycopene content in tomato fruits and classify the different stages of maturation based on color, based on L � , a � , and b � parameters, obtained from RGB images. Therefore, the use of digital image analysis techniques, based on free software and low-cost cameras, enables the construction of accurate and reliable predictive models for the quantification of TCC in cassava. These aspects optimize the analysis time and human and financial resources required without sacrificing the reliability of the results.
## Future perspectives
The potential for implementing rapid phenotyping based on digital images for TCC prediction was demonstrated in this study. In addition, image analysis allows the analysis and incorporation of other characteristics of agronomic importance in cassava cultivation, such as the evaluation of canopy and root yield-related traits, assessment of the linkage between root architecture and micronutrient (zinc and calcium) concentration, and estimation of the growth and nutritional performance of cassava under poor irrigation and potassium fertigation [51-53]. Therefore, these variables can be used in the composition of multi-trait selection indexes for further optimization of cassava breeding programs. Furthermore, the joint use of this approach with genomic selection tools can accelerate the development of cassava varieties with higher TCC.
This study completed the first step in the construction of a digital image database to predict TCC in cassava. The next steps will involve the inclusion of root samples with higher TCC to aid the reduction of predictive error, particularly in samples exceeding 5.0 μg g -1 . In addition, samples of roots grown in different cultivation environments will be evaluated to include the effect of genotype × environment interaction in the predictive models so that they can achieve more widespread use in different breeding programs. Such improvements are predominantly focused on increasing the predictive ability of TCC and the practical implementation of highthroughput phenotyping for screening hundreds of samples daily. The expectation is that image-based phenotyping methodologies will improve the quality and accuracy of data collected in various phenotypic evaluations, thus contributing to an increase in the selection differential and in the heritability coefficient, with a direct effect on genetic gain [bib_ref] Model-assisted phenotyping by digital images in papaya breeding program, Cortes [/bib_ref].
# Conclusion
The use of colorimetric data from the CIELAB space, obtained from the analysis of RGB digital images, is an economical, fast, and efficient alternative that has shown an excellent predictive ability for the TCC in cassava roots. The b � and chroma indices of the CIE color model are strongly correlated with TCC and can be used to accurately assess this characteristic in cassava. These indices were also the variables that most influenced the accuracy of predictive models, and that most contributed to clustering genotypes with higher TCC levels based on PCA.
The ANN model showed the greatest predictive ability of all the algorithms for TCC in cassava roots, even producing accuracy and prediction errors comparable to more sophisticated approaches such as the NIRS. Thus, this first proof of concept demonstrated the high potential of using digital images for the practical implementation of high-throughput phenotyping in cassava.
## Supporting information
[fig] •: Eder Jorge de Oliveira: FAPESB (Fundação de Amparo à Pesquisa do Estado da Bahia). Grant number: Pronem 15/2014 Eder Jorge de Oliveira and Massaine Bandeira e Sousa: UK's Foreign, Commonwealth & Development Office (FCDO) and the Bill & Melinda Gates Foundation. Grant number: INV-007637 The funder provided support in the form of fellowship and funds for the research, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.Competing interests:The authors have declared that no competing interests exist. [/fig]
[fig] Fig 1: Scatter plots of the total carotenoid content and colorimetric indices. Numerical values represent Pearson's correlation coefficients between colorimetric indices and total carotenoid content. The blue line represents the 1:1 isoline. https://doi.org/10.1371/journal.pone.0263326.g001 PLOS ONE Group 4 was composed of cassava genotypes with the highest TCC (mean 10.74 μg g -1 ). Regarding the means of the colorimetric indices, group 4 was characterized by high chroma (56.96) and b � (55.57); low hue values (102.59), a � (-12.38), and lightness (136.04); and moderate values of L � (76.14). [/fig]
[fig] Fig 2: Within-group sum of squares of 228 biofortified cassava genotypes based on phenotypic data of total carotenoid content and colorimetric indices (CIELAB). The grouping criterion was based on the K-means clustering algorithm. https://doi.org/10.1371/journal.pone.0263326.g002 [/fig]
[fig] Fig 3: Principal Component Analysis (PCA) based on phenotypic data of total carotenoid content and colorimetric indices (CIELAB) evaluated in 228 biofortified cassava genotypes. https://doi.org/10.1371/journal.pone.0263326.g003 model produced a TCC prediction accuracy below 0.90. In contrast, the ANN model had the best predictive ability (R 2 = 0.94), associated with the lowest error in the TCC estimates (RMSE = 0.24). The models based on linear regression (LRFS, LRBS, LRSS, and GLMSS), as well as the PLS, BL, BBL, and RR models, performed similarly (RMSE = 0.26; R 2 = 0.93). [/fig]
[fig] Fig 4: Boxplot of total carotenoid content and colorimetric indices for each cluster of 228 biofortified cassava genotypes based on principal component analysis. Different letters represent significant differences between accession groups with p < 0.05 by the Tukey Honest Significant Difference test. https://doi.org/10.1371/journal.pone.0263326.g004 [/fig]
[fig] Fig 5: Relative importance of colorimetric indices for predicting total carotenoid content using 12 prediction models: Linear Regression with Forward Selection (LRFS), Linear Regression with Backwards Selection (LRBS), Ridge Regression (RR), Linear Regression with Stepwise Selection (LRSS), Generalized Linear Model with Stepwise Feature Selection (GLMSS), Random Forest (RF), Partial Least Squares (PLS), the Bayesian Lasso (BL), the Bayesian Blasso (BBL), Artificial Neural Network (ANN), Support vector machine (SVM), and Classification and regression trees (CART). https://doi.org/10.1371/journal.pone.0263326.g005 [/fig]
[fig] Fig 6: Relationship between observed and predicted values for total carotenoid content of cassava roots. The prediction was performed based on 12 different models based on random cross-validation without test set (V-Random) (80/20% in training and validation, respectively). Artificial neural network (ANN), Bayesian Blasso (BBL), Bayesian Lasso (BL), classification and regression trees (CART), generalized linear model with stepwise feature selection (GLMSS), linear regression with backward selection (LRBS), linear regression with forward selection (LRFS), linear regression with stepwise selection (LRSS), partial least squares (PLS), random forest (RF), ridge regression (RR), and support vector machine (SVM) were calculated. Each cross-validation fold is represented in different colors. The numerical data included in the graphs represent: linear equations (y) and coefficient of determination (R 2 ). https://doi.org/10.1371/journal.pone.0263326.g006 [/fig]
[fig] Fig 7: Relationship between observed and predicted values for total carotenoid content of cassava roots. The prediction was performed based on 12 different models based on random cross-validation with test set (IV-Random) (60/20/20% in training, validation, and test set respectively). Artificial neural network (ANN), Bayesian Blasso (BBL), Bayesian Lasso (BL), classification and regression trees (CART), generalized linear model with stepwise feature selection (GLMSS), linear regression with backward selection (LRBS), linear regression with forward selection (LRFS), linear regression with stepwise selection (LRSS), partial least squares (PLS), random forest (RF), ridge regression (RR), and Support vector machine (SVM) were calculated. Each cross-validation fold is represented in different colors. The numerical data included in the graphs represent: linear equations (y) and coefficient of determination (R 2 ). https://doi.org/10.1371/journal.pone.0263326.g007 correlation observed between TCC and b � (r = 0.90). This strong correlation was expected as the positive values of the b � axis correspond to yellow color. This result is in agreement with studies of other species, in which it was verified that the accumulation of carotenoids indicates [/fig]
[fig] Fig 8: Relationship between observed and predicted values for total carotenoid content of cassava roots. The prediction was performed based on 12 different models based on PCA clustering-based with k = 5 (IV-Cluster). Artificial Neural Network (ANN), Bayesian Blasso (BBL), Bayesian Lasso (BL), Classification and Regression Trees (CART), Generalized Linear Model with Stepwise Feature Selection (GLMSS), and Linear Regression with Backward Selection (LRBS). The numerical data included in the graphs represent: linear equations (y) and coefficient of determination (R 2 ). https://doi.org/10.1371/journal.pone.0263326.g008 PLOS ONE Fig 9. Relationship between observed and predicted values for total carotenoid content of cassava roots. The prediction was performed based on 12 different models based on PCA clustering-based with k = 5 (IV-Cluster). Linear Regression with Forward Selection (LRFS), Linear Regression with Stepwise Selection (LRSS), Partial Least Squares (PLS), Random Forest (RF), Ridge Regression (RR), and Support Vector Machine (SVM). The numerical data included in the graphs represent: linear equations (y) and coefficient of determination (R 2 ). https://doi.org/10.1371/journal.pone.0263326.g009 [/fig]
[fig] S1: Table. List of cassava genotypes and the total carotenoid content and colorimetric indices. (CSV) S2 Table. Observed and predicted values for total carotenoid content of cassava roots. based on Linear Regression with Forward Selection (LRFS). Linear Regression with Backwards Selection (LRBS). Ridge Regression (RR). Linear Regression with Stepwise Selection (LRSS). Generalized Linear Model with Stepwise Feature Selection (GLMSS). Random Forest (RF). Partial Least Squares (PLS). the Bayesian Lasso (BL). the Bayesian Blasso (BBL). Support vector machine (SVM). and Classification. regression trees (CART) and Artificial Neural Network (ANN). The cross-validation scheme was performed with 5 repetitions of 6-fold analysis. (CSV) Fig. Boxplot of the performance of the random cross-validation without test set (V-Random) to predict the total carotenoid content in cassava roots using colorimetric indices obtained from digital images considering the complete (all variables) and reduced models (only variables with more than 50% relative importance). Artificial Neural Network (ANN), Bayesian Blasso (BBL), Bayesian Lasso (BL), Classification and Regression Trees (CART), Generalized Linear Model with Stepwise Feature Selection (GLMSS), Linear Regression with Backward Selection (LRBS), Linear Regression with Forward Selection (LRFS), Linear Regression with Stepwise Selection (LRSS), Partial Least Squares (PLS), Random Forest (RF), Ridge Regression (RR), and Support Vector Machine (SVM). (TIFF) S2 Fig. Performance of PCA clustering-based cross-validation scheme (IV-Cluster) to predict the total carotenoid content in cassava roots using colorimetric indices obtained from digital images considering the complete (all variables) and reduced model (only variables with more than 50% relative importance). Artificial Neural Network (ANN), Bayesian Blasso (BBL), Bayesian Lasso (BL), Classification and Regression Trees (CART), Generalized Linear Model with Stepwise Feature Selection (GLMSS), Linear Regression with Backward Selection (LRBS), Linear Regression with Forward Selection (LRFS), Linear Regression with Stepwise Selection (LRSS), Partial Least Squares (PLS), Random Forest (RF), Ridge Regression (RR), and Support Vector Machine (SVM). (TIFF) S3 Fig. Boxplot of the performance of the random cross-validation with test set (IV-Random) to predict the total carotenoid content in cassava roots using colorimetric indices obtained from digital images considering the complete (all variables) and reduced model (only variables with more than 50% relative importance). Artificial Neural Network (ANN), Bayesian Blasso (BBL), Bayesian Lasso (BL), Classification and Regression Trees (CART), Generalized Linear Model with Stepwise Feature Selection (GLMSS), Linear Regression with Backward Selection (LRBS), Linear Regression with Forward Selection (LRFS), Linear Regression with Stepwise Selection (LRSS), Partial Least Squares (PLS), Random Forest (RF), Ridge Regression (RR), and Support Vector Machine (SVM). (TIFF) Author Contributions Conceptualization: Ravena Rocha Bessa de Carvalho, Massaine Bandeira e Sousa, Eder Jorge de Oliveira. Data curation: Ravena Rocha Bessa de Carvalho, Massaine Bandeira e Sousa. Formal analysis: Diego Fernando Marmolejo Cortes, Massaine Bandeira e Sousa.Funding acquisition: Eder Jorge de Oliveira. Investigation: Ravena Rocha Bessa de Carvalho, Diego Fernando Marmolejo Cortes, Massaine Bandeira e Sousa. Methodology: Diego Fernando Marmolejo Cortes, Massaine Bandeira e Sousa, Eder Jorge de Oliveira. 45. Zaouay F, Mena P, Garcia-Viguera C, Mars M. Antioxidant activity and physico-chemical properties of Tunisian grown pomegranate (Punica granatum L.) cultivars. Industrial Crops and Products. 2012; 40: 81-89. 46. Wang S, Summers RM. Machine learning and radiology. Medical image analysis. 2012; 16(5): 933-951. https://doi.org/10.1016/j.media.2012.02.005 PMID: 22465077 47. Liakos KG, Busato P, Moshou D, Pearson S, Bochtis D. Machine learning in agriculture:A review. Sensors. 2018; 18(8): 2674. https://doi.org/10.3390/s18082674 PMID: 30110960 48. Carmo CD, Sousa MB, Pereira JDS, Ceballos H, Oliveira EJ. Identification of waxy cassava genotypes using fourier-transform near-infrared spectroscopy. Crop Science. 2020; 60(2): 883-895. 49. Figàs MR, Prohens J, Raigó n MD, Fernández-De-Có rdova P, Fita A, Soler S. Characterization of a collection of local varieties of tomato (Solanum lycopersicum L.) using conventional descriptors and the high-throughput phenomics tool Tomato Analyzer. Genetic Resources and Crop Evolution. 2015; 62 (2): 189-204. 50. Wan P, Toudeshki A, Tan H, Ehsani R. A methodology for fresh tomato maturity detection using computer vision. Computers and electronics in agriculture. 2018; 146: 43-50. 51. Selvaraj MG, Valderrama M, Guzman D, Valencia M, Ruiz H, Acharjee A. achine learning for highthroughput field phenotyping and image processing provides insight into the association of above and below-ground traits in cassava (Manihot esculenta Crantz). Plant methods. 2020; 16(1): 1-19. 52. Busener N, Kengkanna J, Saengwilai PJ, Bucksch A. Image-based root phenotyping links root architecture to micronutrient concentration in cassava. Plants People Planet. 2020; 2(6): 678-687. 53. Wasonga DO, Yaw A, Kleemola J, Alakukku L, Mäkelä PS. Red-green-blue and multispectral imaging as potential tools for estimating growth and nutritional performance of cassava under deficit irrigation and potassium fertigation. Remote Sensing. 2021; 13(4): 598. [/fig]
|
Clinicians’ views and experiences of prescribing oral anticoagulants for stroke prevention in atrial fibrillation: A qualitative meta-synthesis
BackgroundGlobally, over 33 million people have atrial fibrillation (AF). In eligible patients, oral anticoagulation (OAC) is recommended for stroke risk reduction. Despite recent increases in OAC prescribing, global under-prescription to high-risk AF patients and inappropriate prescription to low-risk patients is leading to unnecessary risk of stroke and haemorrhage. This metasynthesis explored clinicians' beliefs and experiences regarding OAC prescription to AF patients, highlighting barriers to stroke prevention and informing future clinician-focused interventions.
# Introduction
Atrial fibrillation (AF) affects over 33 million people worldwide with prevalence rising globally.Oral anticoagulation (OAC) is recommended for stroke risk reduction by current national and international guidelines for female AF patients with a CHA 2 DS 2 -VASc score �2 and male patients with a score �1.OAC prescription rates are currently around 80% in Europe and the USA,but substantially lower in many low and middle-income countries, including in China (1-4%) and high income Asian countries such as South Korea (32-37%).There are also specific improvements to be made in 'guideline-adherent' OAC prescription. Only 68% of the UK's high-risk AF patients (CHA 2 DS 2 -VASc score �2) are receiving OACs, falling to 48% in the Ukraine and 40% in India.Globally, half of all newly diagnosed AF patients with a CHA 2 DS 2 -VASc score of 0 (low risk) are receiving OACs unnecessarily, putting them at increased risk of major haemorrhage.An understanding of clinicians' views and experiences of prescribing OAC in practice, including what drives their prescribing decisions and their beliefs regarding barriers to guideline-adherent prescribing, and is necessary to inform the design of future effective interventions supporting clinicians in their management of AF patients.
A 2012 qualitative meta-synthesis explored patients' and clinicians' experiences of AF and OACs.Provision of sufficient information; patients' and clinicians' preferences for consultation styles; patients' preconceptions of OACs; and communication between clinicians all affected clinicians' and patients' experience of AF and OACs.In 2017, a qualitative metasynthesis explored AF patients' and clinicians' views specifically of vitamin K antagonist (VKA) oral anticoagulants.Authors reported that available scientific information; concerns about risks related to VKAs; confidence in prescribing; and degree of paternalism were concerns pertinent to clinicians and patients' decision-making processes.In addition, patients placed importance on their own knowledge and understanding of OACs, the impact of OAC on their daily life and satisfaction with their OAC therapy. Despite the continued popularity of VKAs (predominantly warfarin), there has been a considerable rise in NOAC (Non-VKA oral anticoagulants) use, a class of drugs with a different risk profile and very different daily management requirements. This change in prescribing habits resulted in the 2017review presenting an incomplete picture of the current challenges of OAC decision-making and prescription. To explore potential reasons behind the current disparity between clinicians' prescription of OACs for AF patients and guideline recommendations, an up-to-date review is required including all currently licenced OACs for stroke prevention in AF.
## Aim of this review
To identify and synthesise qualitative research concerning the views and experiences of clinicians in prescribing OACs for stroke prevention in AF patients. Our findings will highlight barriers to stroke prevention that may help to inform future interventions supporting clinicians in managing AF patients.
# Methods
## Protocol and registration
This systematic review has followed the reporting guidelines formulated in the 'Enhancing transparency in reporting the synthesis of qualitative research' (ENTREQ) statement.The protocol was registered with PROSPERO prior to conducting literature searches (CRD42016038133).
## Eligibility criteria
Qualitative studies, standalone or within mixed methods designs, investigating the views and experiences of clinicians in the decision-making process concerning OAC prescription for stroke prevention in AF patients. We have defined 'clinicians' as any healthcare professional licenced to prescribe OACs to AF patients.
## Literature searches
A comprehensive, pre-planned search strategy employed search filters and/or index terms relating to "qualitative research". A combination of text words and index terms were used relating to "atrial fibrillation", "oral anticoagulants" and "views, experiences, perceptions and attitudes" (see S1 . No language or date restrictions were applied to searches which included all material published from the inception of each database to 12 th June 2018. MEDLINE, MEDLINE in Process, EMBASE, PsychINFO, CINAHL, Applied Social Sciences Index and Abstracts (ASSIA) were searched for primary studies; Sciences and Social Sciences Citation Index (Web of Science) for citation searching. Grey literature sources were Open Grey, Sociological abstracts, conference proceeding citation index (Web of Science) and the Healthcare Management Information Consortium (HMIC). Citation lists of included studies and relevant reviews were checked. An update of the main Medline search was conducted from inception to 3 rd February 2020.
## Study selection process
Search results were exported into Endnote X8.0.2 and duplicates removed. Two reviewers (JC and DC/RP) independently screened titles and abstracts and assessed full texts of potentially relevant studies against inclusion criteria. Disagreements were discussed with a third reviewer (GT). Authors of relevant conference abstracts were contacted to see if work had been published, with requests for access to unpublished data, where appropriate. The study selection process is illustrated in a PRISMA flow diagram.
## Data extraction
Data from included studies were extracted independently by two reviewers (JC and RP) using a standardised piloted data extraction form containing details regarding the study's year of publication; aim; where it was conducted and in what type of setting; participant demographic characteristics; the qualitative method used; the data analysis method and the findings. Discrepancies were resolved through discussion and comparison with original data. Authors were contacted for clarification as required. The following information was extracted from all included studies: study characteristics and location, aim; participants (e.g. number, recruitment/sampling strategies); qualitative approach, research paradigm, (e.g. grounded theory, ethnography, phenomenology) and methods (e.g. survey); data analysis methods and findings (including quotes and authors' interpretations).
## Quality assessment
Two reviewers (JC and RP) independently assessed the methodological quality of included studies using the Critical Appraisal Skills Programme (CASP) Qualitative Research Checklist.
## Plos one
Oral anticoagulant prescribing: A meta-synthesis of clinicians' experiencesDisagreements were resolved by discussion. Studies were not excluded based on quality or adequacy of reporting.
## Data synthesis
An inductive three-stage thematic synthesis of qualitative data was undertaken.First, lineby-line coding of the findings of primary studies was conducted, according to meaning and content. All findings, in both quote form and author comments, were coded, as recommended by Harden.Coding was conducted independently by two researchers (RP and JC) using NVivo version 12;interpretations were compared and consensus reached through discussion. Translation of concepts between studies was achieved by applying the codes developed through coding initial transcripts, to subsequent transcripts, with new codes being created where necessary. Codes were then categorised into descriptive themes. Finally, analytical themes were developed to describe and understand all previous descriptive themes.
# Results
This review identified 3499 records. After removing duplicates, a further 2867 ineligible records were excluded. The eligibility of 101 full text papers was assessed, with 13 studies being included. For details of excluded full text papers see S1 Table in S1 File.
## Study quality
Included studies varied in rigour according to the CASP qualitative checklist.Ten studies were high quality in relation to the majority (five or more) of the nine graded fields of the qualitative CASP checklist.Poorly described facets included the relationship between researcher and participant;explanation of the study to participants and handling of their data; rigour of data analysis and appropriateness of recruitment strategies.For further details of study quality see S2 Table in S1 File.
## Study characteristics
Characteristics of included studies are reported in detail in S3 Table in S1 File and summarised in. Data were included from at least 248 clinicians. Included studies were conducted in the UK,Finland,Australia,the USAand Canada.Participating clinicians were from primary, secondary and tertiary care.Qualitative methodologies included focus groups,semi-structured interviews,in-depth interviews with a semi-structured guideand interviews and reviews of healthcare file notes.Mixedmethods studies included semi-structured interviews exploring decision-making in relation to vignettes,surveys and interviewsand questionnaires with both qualitative and quantitative responses.
## Summary of themes
Four analytical themes emerged from the data: (i) 'Clinicians' intellectual and emotional responses to the evidence'; (ii) 'Prescribing in primary and secondary/tertiary care' (iii) 'Clinicians' views of how patients' characteristics and opinions influence prescribing'; and (iv) 'Clinicians' views on their interactions with patients'. (For summary of codes, descriptive and analytical themes see for frequency of codes within studies see S4 Table in S1 File)
## Qualitative synthesis
## Clinicians' intellectual and emotional response to the evidence
Clinicians' interpretations of the evidence guiding the change in stroke prevention over the past decade from anti-platelets and VKAs to VKAs and NOACs varied greatly and were influenced by the evidence and the opinions of hospital physicians and patients.Clinicians justified their differing views with selective use of supportive evidence. Clinicians' uncertaintySecondary and tertiary care Semi structured interviews considering treatment decision related to vignettes. 14 senior consultants/specialist registrars: 5 in Cardiology; 9 in general medicine/geriatric medicine. Leeds, UK Conductor of interviews unclear.
Murray et al. 2011Primary and secondary care Open-ended, semi-structured telephone interviews. in prescribing OAC also resulted from a perceived lack of evidence to guide choices between warfarin and the various NOACs.
"The other pro-warfarin health professionals were either uncertain about the efficacy and bleeding risks of NOACs compared with the risks associated with warfarin, or followed specific guidelines and information from continuing medical education that favoured warfarin."The accessibility of the format and content of the guidelines were raised. Some clinicians made a case for creating guidelines specific to different disciplines.
"there should be one A4 page for an emergency department medical assessment unit, a different A4 for cardiology clinics, different A4 for general practice, and different one for general physicians. All probably saying the same thing, but the way it is being put across, the simplicity is the important point. [D9, General physician]"In addition to their considered intellectual response to scientific evidence, some clinicians described their emotional response to the evidence.
"There was some ambiguity in the approach to guidelines. Three respondents were broadly in favour of guidelines (GP5. . .), one applied them 'flexibly' and three were largely hostile, (GP2)."Some clinicians reported discomfort with the rigid nature of guideline criteria. Others reported anxiety, both within themselves and from their patients, when they felt that the evidence called for a change of practice.
"Yes it does make me feel anxious. . .all the BMJs, all the rags. . .these people must be on warfarin. . .With me messing about with his medication and trying to practise evidence based medicine, I found it was making him [patient] feel more anxious."Clinicians also reported feeling anxious about potentially causing harm to patients and being held accountable for such harm.
"I think doctors are anxious about avoiding risks of medications, we worried about causing bleeding. But because of our fear of that we massive under treat patients with AF and we don't prevent stroke when we should be. . .because our failing as a medical profession is about not using these drugs enough. . .I think the disaster is that the patient has a stroke because they had an untreated AF. (G01)"Sometimes safety concerns, such as those regarding falls and haemorrhage, led to inappropriate prescribing, however other concerns did not.
"'I almost feel like it's the reverse, where the more I know, the more fear I have of using it. The sense of first do no harm. ' 'Coumadin is a terrible drug to take, but it's the least of two evils. So we have to do it. (Cardiologist, 10)'"For some this anxiety was based on previous negative experiences with OACs. Guilt due to personal experiences was a strong motivator.
"I actually had two 50 year olds who had strokes from atrial fibrillation because they didn't get warfarin. . . that really hit me."The degree to which stroke and haemorrhage risks were assessed and influenced the prescribing process varied substantially. Stroke risk was often a more important consideration than bleeding risk. The use of guideline-based risk assessment tools such as the CHA 2 DS 2 -VASc and HAS-BLED was not common practice for all clinicians.
"most PCP [primary care physicians] interviewees were unfamiliar with both the CHA 2 DS 2 -VASc score and bleeding risk assessment tools"Some clinicians reported a lack of knowledge of stroke and haemorrhage risks in specific cases and the skills required to individualise treatment.
"If he had a metallic valve. . .I don't think I'd be comfortable to make that decision. I don't know what the risk of having a stroke is. . .I wouldn't discount home being on warfarin, but only because it's a kind of imprint"Clinicians' described different treatment preferences, sometimes based on clinical considerations, sometimes formed through habit.
"it goes back to the culture thing, if you have cardiologist who have been using warfarin for the last 40 years and why they are going to change to NOACs?"Regular warfarin monitoring was sometimes seen as providing a safety net unavailable to patients on NOACs; however, the risk of bleeding and poor anticoagulation control (international normalised ratio, INR) associated with VKAs were off-putting to some clinicians. Some clinicians expressed concern about lack of long-term safety data for NOACs, particularly in older patients, and their lack of reversibility; however, others reported a preference for NOACs, due to their more stable INR, the lack of regular testing and their superior effectiveness and safety compared to warfarin. Often, no general consensus was reached regarding the choice between warfarin and the various NOACs.
"11 physicians considered high-quality warfarin treatment as effective as DOACs, although 11 physicians felt DOACs to be safer than warfarin. In addition 10 physicians mentioned that they did not see great differences between different DOACs or that they have not been directly compared and, therefore, none of them can be said to be better than the others'Different types of clinicians described different opinions of NOACs and warfarin.
"compared with cardiologists and geriatricians, who were cautious about using NOACs, a neurologist and the haematologists had strong opinions about using NOACs as first-line therapy."A perception was reported that GPs were quite reluctant to use warfarin.
"even before the NOACs patients still were not put on to warfarin who had AF, just because they (GPs) thought the risk of bleeding is far too high."Prescribing in primary and secondary/tertiary care A well-developed doctor-patient relationship and the perception of less time-pressured consultations in primary care were thought to enable more thorough treatment discussions, however, specialist knowledge was seen to be lacking compared to secondary care. In contrast, lack of time in secondary care was seen to seriously affect patient education and involvement in decision-making.
"In our chronic disease clinics . . . the people with AF on warfarin will see the same practitioner . Perceived strengths and weaknesses of different healthcare settings were seen to be reflected in the different outlooks of the clinicians.
## "hospital doctors tended to impose a decision 'treating the disease rather than the patient.'. . . the idea that there is a treatment of a condition therefore you must treat patients with this condition with this treatment is what a lot of hospital doctors think . . .whereas i think in general practice people often are prepared to take a step back and say 'well yes, there is a treatment there but is it in the best interests of this patient, is it going to fit in with the rest of you know. . .it's the holistic view'" (gp11) [24]
The availability of medical care was also seen to influence the prescribing process.
"Complications always tended to happen 'over the weekend,' and those practitioners who, for example, did not always have nursing staff to help do blood tests seemed to be less enthusiastic about implementing evidence on anticoagulation"A lack of effective communication between primary and secondary care was reported.
## "one of the biggest problems we have is a lack of communication, too many chefs in the kitchen. . .it's a waste of my time to do all the research and then send patients to somebody who feels uncomfortable with my plan"[26]
Clinicians sometimes reported confusion over where the principal responsibility for OAC prescribing lay, with some clinicians routinely referring their patients for anticoagulation, others for support of their recommendations and some referring only complex cases. This disjointed care pathway was seen to have a serious impact on accountability.
"Decision making for who goes on warfarin is taken often by one person, monitoring of warfarin is taken by another person and in our practice people are monitored in 5 different systems. . .ongoing responsibility for patient education is non-existent. . .the potential risks of warfarin to me are so large in terms of errors basically. (GP2)"GPs reported feeling uncomfortable challenging the decisions of consultants, who they regarded as experts. GPs saw this conflict as being an issue for themselves and their patients, who they felt were more inclined to take their consultants advice.
"the one thing that surprised me looking through this case note review is just how big an influence secondary care is still having on the decision as to what people have, whether they get anticoagulated or not and how difficult it is in practice to change that." (GP5)However, this view of the medical hierarchy was sometimes felt to be mediated by the individual confidence of the clinician with OAC prescription. Amongst emergency department clinicians a relationship was found between overall lack of experience and deference to cardiologists in OAC prescription.
"Physicians with less than 10 years of experience more commonly indicated that they would consult a cardiologist to gain concordance of opinion"
## Clinicians' views of how patient characteristics and opinions influence prescribing
Clinicians' views of how medical factors such as patients' age, mental state, fall risk, other medications and medication adherence influenced prescribing, were widely discussed. Clinicians also acknowledged the burden of INR testing and questioned patients' drive, cognitive ability, mental health, and access to the social and medical structure needed to successfully take OACs. Great emphasis was given to the psychosocial suitability of patients by clinicians.
"I add up their CHADS 2 score and think about what category they fall into. I also am concerned with the social milieu of the patient-do they have the will, the intelligence, how far are they from medical care. Those things factor very strongly into my decisions "Equally, patient refusal often highlighted the importance of non-clinical factors to patients. However, this was reportedly not always taken into account within the consultation discourse.
"Patient stories about refusing warfarin demonstrate that lifestyle needs and choices were significant to their decision. Yet physicians' information provision focused on explaining the physiology of the condition and the stroke risk reduction associated with warfarin."Clinicians sometimes reported conflict between the guidelines and the influence of patients' psychosocial factors. On occasion, a clinician's response to patient's psychosocial factors resulted in guidelines potentially not being followed.
"'Strictly speaking this chap could get by without any anticoagulation but in view of his anxiety. . .I think I would go ahead and offer him long term warfarinization."Some American clinicians described perceiving racial or socioeconomic prejudice among colleagues. It was reported that some clinicians' opinions regarding non-compliance in certain patient populations were not supported with evidence.
"None of the interviewees reported personal racial biases; yet when asked for reasons why African Americans may have a higher incidence of AF-related stroke and related mortality, many responded that it was because of a higher rate of noncompliance in this patient population. However, no specific examples or data supporting this view were provided."For some clinicians their assumptions about patients' lack of understanding about OAC treatment influenced their willingness to prescribe OAC and undertake the required monitoring.
"you have listed all possible risk factors (stroke, bleeding and medication safety risk factors). . . So it is just you would have to constantly measure these from time to time. It is time consuming and detailed and maybe doesn't warrant maybe because of the patient health illiteracy. (P05)"Physicians from countries in which patients were directly or indirectly (through insurance) charged for OAC medication reported that the cost of NOACs could form a prohibitive barrier that outweighed both clinical and social considerations.
"Some patients can't really afford NOACs. I think they're something like 90 bucks a month or something. So I wouldn't switch anyone over basically because of the cost difference."Patient beliefs could also influence the prescribing process. Clinicians noted a dislike of warfarin amongst patients due to its use as a rat poison and the burden of frequent blood test monitoring, prompting a preference for NOACs.
"Taking something like warfarin is incredibly taxing on people. . .with weekly or at least biweekly INR checks'
## Clinicians' views on their interactions with patients
Clinicians employed a wide range of methods to convey the physiology of patients' stroke and haemorrhage risks including risk and benefit language, metaphors, description of physiology, drawings and printed information. However, providing explanations of AF and the associated risks to asymptomatic patients was challenging for some. There was a wide range in the degree of information clinicians felt their patients required.
"two GPs described minimal explanation of risk and benefit to patients. . .five gave some detail and three described detailed and explicit explanation""The only way you can empower the patient to make decisions is to provide the information that they require to make that decision."Discussion of the motivation behind clinicians' interactions with patients revealed a desire to persuade patients of the 'right' decision.
"Some GPs described quite strong levels of persuasion while others were content to explain their view of the risks and benefits and assist the patient to choose. Equally, while some GPs described negotiating the decision, others seemed to be describing exerting pressure on patients to accept responsibility for the decision or, conversely, to agree to the choice the GP thought most beneficial."Persuading the patient to make 'the right treatment decision' sometimes involved careful choice of language.
"The choice of words or the use of metaphors like 'slanting' or 'selling' were mechanisms the doctors used to influence patients to make a decision about their treatment that was consistent with what the doctor had decided was appropriate. Doctors would refer to 'rat poison' when describing warfarin if they felt its use would be difficult or inappropriate, or describe pills as 'having been shown to keep the heart young' when they wanted a patient to agree to treatment."Selective provision of information either, to avoid provoking fear, or capitalise on it in order to encourage patients to make 'the right treatment choice' was described.
"I don't usually warn them about brain haemorrhage . . . perhaps I should. But I don't. [D3, Consultant cardiologist]""I mean generally, if you tell them it's a stroke, they get worried about it and generally as you know, when people think of stroke, they think 'oh my god, I'll become bed bound, very disabled, I wouldn't be able to do it'. So they start to take it very seriously that AF is not as benign as they expected. [D10, General physician]"Clinicians were very aware of the concepts of patient-centred and shared decision-making. Consultation styles varied from strongly patient-centred to strongly paternalistic.
## "gps influencing patients by a combination of persuasion, shared decision making, gp-or patient-centred consulting, and explanation, with the gps' input being modified by their view of the evidence. the weight and quality of each component varied from one gp to another (and possibly from one consultation to another)." [24]
Some clinicians were comfortable allowing patients' views and experiences to guide the treatment decision.
"I think it's a personal decision rather than a right decision. . .their life and how they feel about life, how they feel about death, how they feel about illness and all those sort of things to actually throw that into the equation" (GP8)Clinicians were more able to accept a patient's decision to decline treatment if they believed the decision was well informed.
"GPs with most experience of EBM were also most willing to engage with the agendas of patient-centred consulting and shared decision making, and were most relaxed if patients declined treatment, provided they were convinced that the patient understood the consequences. . .they were willing to question the applicability of guidelines to individual patients and believed that patients would sometimes accept higher levels of risk than suggested by guideline treatment thresholds"For some clinicians, the acceptability of the treatment options to the patient was important, with the final decision being shared. Some clinicians found that patient involvement in the decision making process had beneficial long-term effects.
"three physicians felt that patients adhere better to their treatment when they have participated in the decision-making"Some clinicians began discussing their beliefs regarding the importance of a patient-centred approach; however, their thoughts then led on to the importance of making the right decision and revealed a much more paternalistic style that many patients were reportedly happy with.
"The only way you can empower the patient to make decisions is to provide the information that they require to make that decision. AND if you don't then they make the decision on the basis on inaccurate or inadequate information and they often come to the wrong conclusion. . .but erm the majority just allows me to choose what's best for them. [D1, Consultant cardiologist]"Other clinicians were confident in their opinion and presented the options as a fait accompli.
"My own personal view I would probably avoid warfarin lifelong so the option would either be aspirin or clopidogrel."Some clinicians who were certain in their belief in the right treatment decision, apportioned blame to themselves if their advice was not followed and experienced a sense of failure.
"I've made a mistake in how I've described the risk for that individual patient because they made a decision which I think is probably the wrong decision (GP4)"Discussion This review explored the views and experiences of clinicians prescribing OACs for stroke prevention in AF patients, revealing the barriers to guideline-adherent OAC prescription from the clinician's viewpoint. Clinicians reported challenges around OAC prescription relating to both their intellectual and emotional response to evidence; the primary and secondary/tertiary care settings; the characteristics and opinions of their patients; the cost of NOACs and their interactions with patients during the prescription consultation.
A wide variation in familiarity with risk assessment tools and greater emphasis on stroke risk rather than bleeding risk assessment was evident. A lack of confidence and knowledge was reported in primary care and amongst emergency department clinicians, especially of risk factors in more complex cases and some clinicians were unsure about the relative merits of warfarin and the various NOACs in different clinical scenarios. Further education and training in using appropriate risk assessment tools may be beneficial to clinicians lacking confidence in their application. Guidelines specific to different healthcare settings/types of clinicians may be welcomed. Our findings support a recent quantitative review of interventions to improve OAC prescription for AF patients, which found educational interventions and the implementation of local guidelines may be effective approaches to improving OAC prescription.A reluctance to prescribe NOACs was also found due to concerns regarding a lack of long-term safety data. Further research into the safety profile of NOACs may improve clinicians' confidence with these medications.
Clinicians' treatment preferences were sometimes based predominantly on scientific evidence, such as guidelines, risk assessment scales and knowledge of specific drug risk profiles. Sometimes preferences were based on more subjective considerations, such as perceptions of a patient's likely adherence to monitoring and treatment. Assumptions about patients' adherence to medication based on their socioeconomic or ethnic background influenced prescribing in American research.Sometimes treatment preferences were heavily influenced by clinician's emotional response to treatment decisions, such as fear of causing harm and being held accountable; clinician and patient anxiety at the prospect of new evidence requiring a change in practice; or nervousness about the loss of regular monitoring when moving from warfarin to NOACs. Encouraging greater use of the stroke and haemorrhage risk assessment tools may help to rationalise the prescription process; however, interventions enabling clinicians to explore and challenge their emotional responses to difficult OAC prescribing scenarios may also prove useful.
A perception was reported that primary care fostered a greater understanding of a patients' social context and more thorough treatment discussions, but provided less specialist knowledge than in secondary care. Secondary care clinicians were reportedly more inclined to treat the 'condition'. It is not for this paper to discuss the relative merits of the holistic and more condition-focused outlooks, however, an emphasis on patients' ability to adhere to regular monitoring and maintain a suitable INR may have made some clinicians more cautious in prescribing OACs. Again, encouragement to use suitable risk assessment tools may help to focus prescribing decisions on the evidence. Local interventions to upskill primary care physicians have reported some success in improving rates of appropriate OAC use among AF patients.Clinicians' methods and degree of communication with patients regarding AF itself and OAC treatment varied greatly; some clinicians described seeing patients first prescribed OACs in secondary care who were not well informed on their condition and treatment. Time has been reported as a limiting factor in a previous review.Other clinicians, such as pharmacists, may be well placed to provide further, detailed explanations on OAC therapy or to reinforce key messages over time, to encourage adherence. As in previous research,lack of good communication was noted between primary and secondary care. Poor communication was compounded by lack of agreement on treatment; conflict due to the deference of primary care, other non-cardiologist clinicians and their patients to secondary care expertise and a lack of accountability due to the number of people involved in each patient's care. However, a previous review,despite describing poor communication between primary and secondary care, reported that it was primary care physicians who felt specialists were delegating the responsibility of decision making regarding OAC prescription to them.Greater cohesion of services may be encouraged through interventions that are delivered to interdisciplinary teams from primary, secondary and tertiary care. This finding is supported by current AF guidelinesand a recent quantitative systematic review that suggested interdisciplinary medical care programmes educating both clinicians and AF patients were effective in improving OAC prescription.
# Strengths and limitations
This qualitative meta-synthesis was rigorously conducted in accordance with the ENTREQ statement. Searches were free from language bias and duplicate screening, data extraction, quality assessment and coding were used. Included studies were generally of good quality but with poor descriptions of relationships between the researcher and participants, recruitment processes and rigorousness of analyses. The lack of detailed descriptions of these facets of the qualitative research process should temper the weight given to the conclusions presented. All included studies were conducted in high-income countries; therefore, the findings may not be applicable to low and middle-income countries. Included studies were published between 2001 and 2017; changes in the treatment of AF with OACs during this time should be borne in mind when considering the review findings. |
Selective vulnerability to atrophy in sporadic Creutzfeldt‐Jakob disease
a Authors contributed equally to this work.AbstractObjective: Identification of brain regions susceptible to quantifiable atrophy in sporadic Creutzfeldt-Jakob disease (sCJD) should allow for improved understanding of disease pathophysiology and development of structural biomarkers that might be useful in future treatment trials. Although brain atrophy is not usually present by visual assessment of MRIs in sCJD, we assessed whether using voxel-based morphometry (VBM) can detect group-wise brain atrophy in sCJD. Methods: 3T brain MRI data were analyzed with VBM in 22 sCJD participants and 26 age-matched controls. Analyses included relationships of regional brain volumes with major clinical variables and dichotomization of the cohort according to expected disease duration based on prion molecular classification (i.e., short-duration/Fast-progressors (MM1, MV1, and VV2) vs. long-duration/ Slow-progressors (MV2, VV1, and MM2)). Structural equation modeling (SEM) was used to assess network-level interactions of atrophy between specific brain regions. Results: sCJD showed selective atrophy in cortical and subcortical regions overlapping with all but one region of the default mode network (DMN) and the insulae, thalami, and right occipital lobe. SEM showed that the effective connectivity model fit in sCJD but not controls. The presence of visual hallucinations correlated with right fusiform, bilateral thalami, and medial orbitofrontal atrophy. Interestingly, brain atrophy was present in both Fast-and Slow-progressors. Worse cognition was associated with bilateral mesial frontal, insular, temporal pole, thalamus, and cerebellum atrophy. Interpretation: Brain atrophy in sCJD preferentially affects specific cortical and subcortical regions, with an effective connectivity model showing strength and directionality between regions. Brain atrophy is present in Fast-and Slow-progressors, correlates with clinical findings, and is a potential biomarker in sCJD.
# Introduction
Sporadic Creutzfeldt-Jakob disease (sCJD) is a rapidly progressive dementia with underlying neurodegeneration, gliosis, and vacuolation. [bib_ref] Prion diseases, Geschwind [/bib_ref] Although early observations reported cerebral atrophy in sCJD through clinical neuroimaging, this was usually associated with very late disease course and/or long duration. [bib_ref] Cerebral MR and CT imaging in Creutzfeldt-Jakob disease, Kovanen [/bib_ref] [bib_ref] Serial computed tomographic and electroencephalographic studies in Creutzfeldt-Jakob disease, Hayashi [/bib_ref] [bib_ref] MR imaging of Creutzfeldt-Jakob disease, Finkenstaedt [/bib_ref] Currently, cortical atrophy in sCJD is not considered as a salient feature on visual inspection of MRI, except for cases of long duration. Studies on objective atrophy quantification in sCJD are scant, and there are no data on the presence of atrophy in Slow-versus Fast-progressors based on sCJD molecular classification subtyping 5 or on brain atrophy patterns associated with some common clinical characteristics of sCJD. Previous morphometric studies have focused primarily on isolated brain regions and regions of interests, [bib_ref] MRI detection of the cerebellar syndrome in Creutzfeldt-Jakob disease, Cohen [/bib_ref] and white matter, [bib_ref] Evolution of diffusion-weighted magnetic resonance imaging signal abnormality in sporadic Creutzfeldt-Jakob disease..., Eisenmenger [/bib_ref] [bib_ref] Cerebral white matter disruption in Creutzfeldt-Jakob disease, Lee [/bib_ref] or cohorts which combined sporadic and genetic forms of prion disease. [bib_ref] Distinct neuropsychological profiles correspond to distribution of cortical thinning in inherited prion..., Alner [/bib_ref] [bib_ref] Multiparameter MR imaging in the 6-OPRI variant of inherited prion disease, Vita [/bib_ref] Determination of atrophy patterns in sCJD might provide insights into the mechanisms of disease progression and identify vulnerable regions to track disease progression in both Fast-and Slow-progressors, which could be targets of future network-level-based therapies. The primary goal of this cross-sectional study is to quantify cerebral atrophy in sCJD using 3T brain magnetic resonance imaging (MRI) and voxel-based morphometry (VBM). Based on the proposed mechanisms of transsynaptic propagation of prion particles, [bib_ref] The prion model for progression and diversity of neurodegenerative diseases, Stopschinski [/bib_ref] our hypothesis is that neurodegeneration in sCJD causes selective brain network degeneration in anatomically connected regions, as opposed to indiscriminate global injury that might happen if prion disease was spreading within the brain either randomly or to adjacent regions. Structural equation modeling (SEM) is a statistical technique used in functional MRI and positron emission tomography (PET) studies to test whether activity in brain regions correlated with each other indiscriminately or if activity in one region influences that in another. [bib_ref] Imaging and neural modelling in episodic and working memory processes, Krause [/bib_ref] [bib_ref] Principle of structural equation modeling for exploring functional interactivity within a putative..., De Marco [/bib_ref] We therefore tested this hypothesis-driven approach (SEM) to further investigate the relationship between brain regions vulnerable to sCJD pathology beyond simple correlations.
# Participants and methods
## Patient selection
Participants or caregivers provided informed consent for participation in this study, which was approved by the University of California, San Francisco (UCSF) institutional review board. Participants were evaluated between January 2010 and August 2013 at the UCSF Memory and Aging Center (MAC) rapidly progressive dementia research program. Consecutive sCJD participants who ultimately either met 2007 UCSF clinical criteria for probable sCJD [bib_ref] Rapidly progressive dementia, Geschwind [/bib_ref] (n = 8, 20%) or who had definite sCJD (pathology-proven, n = 32, 80%), [bib_ref] Diagnostic criteria for sporadic Creutzfeldt-Jakob disease, Kretzschmar [/bib_ref] and who had the same standardized MRI protocol of adequate quality (on the same scanner) from their first UCSF visit were included in the study. Forty sCJD participants were identified prior to MRI quality assessment. For the VBM analysis, 18 of these 40 participants were excluded due to severe motion artifact (n = 14) or lack of appropriate magnetization-prepared rapid gradient-echo (MPRAGE) T1-weighted images (n = 4), leaving 22 participants with sufficient quality scan for VBM analyses. All 40 sCJD participants, however, had sufficient quality diffusionweighted imaging (DWI) MRIs for visual assessment (e.g., determination of pattern of involvement by DWI) and were used for analyses not requiring T1 sequences.
Although we had initially intended to also examine longitudinal volume change, of the 14 participants with serial T1 scans, too few (n = 4) were of sufficient quality for cohort longitudinal analysis, so this was not performed. Twenty-six healthy age-and gender-matched participants, from the UCSF MAC Hillblom Healthy Aging Network project, who had MRIs performed with the same protocol on the same scanner were used as healthy controls (Controls). For some analyses, sCJD participants were grouped based on their DWI lesion patterns by visual inspection into 1) cortical-subcortical, 2) cortical-only, and 3) subcortical-only cohorts. [bib_ref] Diffusionweighted MRI hyperintensity patterns differentiate CJD from other rapid dementias, Vitali [/bib_ref] Clinical and cognitive evaluation sCJD participants had a standardized clinical evaluation including: neurological history and examination; Mini-Mental State Examination (MMSE) [bib_ref] A practical method for grading the cognitive state of patients for the..., Folstein [/bib_ref] ; the Barthel index (measuring activities of daily living function) [bib_ref] Functional evaluation: the barther index, Mahoney [/bib_ref] ; a neuropsychological battery (although we only used MMSE); the neuropsychiatric inventory (NPI, 12-item version) [bib_ref] The Neuropsychiatric Inventory: assessing psychopathology in dementia patients, Cummings [/bib_ref] to assess for behavioral symptoms and the presence of hallucinations; motor evaluation with the Unified Parkinson's Disease Rating Scale -Motor (UPDRS-Motor); and routine electroencephalogram (EEG). Controls underwent neurological history and exam, the clinical dementia rating scale, and a neuropsychological battery. Signs and symptoms were recorded from patient records and UCSF research visit summaries until the time of the research brain MRI. EEGs were classified into three categories: normal, slowing, or periodic sharp wave complexes, with the latter two categorized as abnormal for this study. [bib_ref] Clinical overlap between Jakob-Creutzfeldt disease and Lewy body disease, Tartaglia [/bib_ref] CSF total tau (t-tau; performed at the U.S. National Prion Disease Pathology Surveillance Center (NPDPSC), Cleveland, OH) and neuron-specific enolase (NSE; performed at Mayo Laboratories, Rochester, MN) levels, determined with enzymelinked immunosorbent assay, and protein 14-3-3 determined by Western Blot (NPDPSC; reported as either positive, inconclusive, or negative, with inconclusive considered as negative for this study) were available in 17, 13, and 19 sCJD participants, respectively. EEG and CSF biomarkers were obtained within 24 hours of the brain MRI. We also examined the effect of the prion protein gene (PRNP) codon 129 polymorphism on other clinical variables, including t-tau, NSE, protein 14-3-3, EEG, and clinical scales. None of the 40 sCJD participants had a PRNP mutation, and all had codon 129 polymorphism analysis (NPDPSC, Case Western Reserve University, Cleveland, OH). Pathological confirmation was performed at either or both UCSF and NPDSPC, and prion typing was performed through the NPDPSC. The 32 pathology-proven sCJD participants consisted of [bib_ref] A practical method for grading the cognitive state of patients for the..., Folstein [/bib_ref] analysis. One pathology-proven subject without VBM had variably protease-sensitive prionopathy and therefore had no prion typing. [bib_ref] Variably proteasesensitive prionopathy: a new sporadic disease of the prion protein, Zou [/bib_ref] [bib_ref] Prions in variably protease-sensitive prionopathy: an update, Zou [/bib_ref] [bib_ref] Molecular biology and pathology of prion strains in sporadic human prion diseases, Gambetti [/bib_ref] Image acquisition Participants underwent brain MRI at UCSF on a 3T scanner (Siemens TrioTim syngo, Erlangen, Germany) equipped with an eight-channel transmit and receive head coil using an MPRAGE sequence with the following parameters: 160 sagittal slices; slice thickness = 1 mm; field of view = 256 mm; matrix = 256 9 240; voxel size 1.0 9 1.0 9 1.0 mm 3 ; TR/TE/TI = 2300/2.98/900 ms, flip angle = 9°, and a HARDI dataset acquired using a singleshot spin-echo echo-planar imaging (EPI) sequence including 55 contiguous axial slices acquired in an interleaved order with the following parameters: TR/ TE = 8000/109 ms; flip angle = 90°; matrix = 100 9 100; in-plane resolution = 2.2 mm 2 ; slice thickness = 2.2 mm; 64 noncollinear diffusion sensitization directions at b = 2000 s/mm 2 , 1 at b = 0; and integrated parallel acquisition technique acceleration (IPAT) factor = 2.
## Image processing
## Image preprocessing for vbm analyses
All T1-weighted images were visually inspected, and images with excessive motion or artifact were excluded. Processing for T1-weighted sequences was performed using Statistical Parametric Mapping (SPM12, Welcome Trust Center for Neuroimaging, London, UK) running under MATLAB R2014b (MathWorks). Images were bias field-corrected using N3 algorithm and segmented into gray matter, white matter, and CSF compartments using the unified segmentation algorithm from SPM12. A custom template was created from the whole cohort population (sCJD participants and Controls) by nonlinear registration template generation using Large Deformation Diffeomorphic Metric Mapping framework cite. [bib_ref] Diffeomorphic registration using geodesic shooting and Gauss-Newton optimisation, Ashburner [/bib_ref] The images were then spatially normalized to the custom template and then modulated by multiplying the voxel values by the Jacobian determinant derived from the spatial normalization to preserve the relative volumes of gray matter. Finally, the images were smoothed with an 8-mm full-width-at-half-maximum Gaussian kernel. This parcellation process created raw volumetric values for gray matter, white matter, and CSF compartments. The Anatomical Automated Labeling parcellation method was used to generate gray matter volumes, which were used in the SEM analysis and some other analyses to compare with CSF biomarkers (14-3-3, NSE, and total tau). The gray matter images then underwent VBM processing and analysis as discussed below.
## Vbm processing and analysis
For the MRIs with sufficient quality scans to undergo VBM processing, a general linear model was fit at each voxel using FMRIB Software Library v6.0 (FSL, Created by the Analysis Group, FMRIB, Oxford, UK). All comparisons included correction for age and total intracranial volume (TIV; calculated as the sum of gray matter, white matter, and CSF volumes). For comparisons in which we wanted to capture the effect of the speed of disease progression (based on molecular classification dichotomized into Fast-progressors vs. Slow-progressors), we controlled for disease severity by the Barthel index at the time of the MRI. This ensures that the findings are not due to the presence of more advanced disease in one group compared to the other. For comparisons that included clinical severity of symptoms (global cognition based on MMSE, motor symptoms based on UPDRS, and behavioral symptoms based on the NPI), we controlled for the MRI timeratio (the time the MRI was done relative to the total disease duration) to ensure that the findings are not simply due to being either early or late in the total disease process. For comparisons between those subjects with or without certain clinical symptoms (myoclonus, visual hallucinations, and ataxia), we controlled for both disease severity (Barthel index) and MRI time-ratio in order to capture the brain regions associated with these symptoms. For these categorical VBM comparisons, a t-test was performed for each voxel, with accepted threshold p value < .05 after correction for multiple comparisons with the permutation method used by the FMRIB Software Library (FSL), and the number of permutations was set at 5,000 using threshold-free cluster enhancement. [bib_ref] Permutation inference for the general linear model, Winkler [/bib_ref] Resulting statistical maps were normalized to Montreal Neurological Institute space for display. [bib_ref] Unbiased nonlinear average age-appropriate brain templates from birth to adulthood, Fonov [/bib_ref] We ran VBM analyses based on the following clinical variables: 1 sCJD versus Controls. 2 Fast-progressors versus Slow-progressors based on molecular classification (PRNP codon 129 genotype [MM, MV, or VV] and prion typing [type 1 or type 2]). As the median disease duration of three sCJD molecular subtypes (MM1, MV1, and VV2) is less than 7 months (Fast-progressors) and for the remaining three molecular subtypes (MM2, MV2, and VV1) typically is greater than 11 months (Slow-progressors), [bib_ref] Determinants of diagnostic investigation sensitivities across the clinical spectrum of sporadic Creutzfeldt-Jakob..., Collins [/bib_ref] we conducted an analysis dichotomizing the sCJD VBM group into expected Fast-progressors versus Slow-progressors. For this analysis, we did not include sCJD cases with mixed prion types (i.e., 1-2), as such cases often present more heterogeneously, varying along a spectrum between type 1 and type 2, 29,30 nor did we include the single variably protease-sensitive prionopathy case, who by definition had no prion typing. [bib_ref] Variably protease-sensitive prionopathy: a novel disease of the prion protein, Gambetti [/bib_ref] We also compared MM with MV cases (insufficient numbers for VV). Addtionally, we ran a VBM analysis comparing MM2 to MV2 participants (both Slowprogressors) given their clinical and pathological differences,
## Volumetric correlations and structural equation modeling
To investigate the general correlation between the volumetric data in the brain regions that showed atrophy in sCJD compared with Controls, Pearson correlations were performed after correction for age and TIV. We then used SEM to investigate model fit of regional VBM volumetric data. Our hypothesis was that if region atrophy is not random, some regions will influence volumetric changes in another region or regions, and a model that includes the regions most commonly involved in sCJD will pass the strict significance thresholds of SEM only in sCJD but not in regions not commonly involved in sCJD or in Controls. Because previous research that used hypothesisdriven SEM approaches recommends starting with parsimonious models, we tested SEM models in four canonical networks in sCJD participants and Controls first and then added regions commonly affected in sCJD. SEM is a multivariate technique that combines "path analysis" (a statistic used to evaluate causal models) [bib_ref] The method of path coefficients, Wright [/bib_ref] and multiple regression to estimate linear relationships between specific variables in an analysis of the covariance among these variables. One might expect cortical volumes of certain brain regions in sCJD to correlate with each other. This correlation of volumes could occur for at least three reasons: 1.) regional vulnerability; 2.) prions spreading transynaptically to functionally connected regions; or 3.) prions spreading to adjacent areas. SEM extends beyond simple correlations to evaluate the presence of any directional relationship among brain regions, and it examines the strength as well as the directionality of relationships within a network. 14 SEM incorporates the residual variance of the independent variables to estimate the dependences among the observed variables allowing for the validation of a hypothetical model by empirical data using strict goodness-of-fit thresholds. [bib_ref] Imaging and neural modelling in episodic and working memory processes, Krause [/bib_ref] The fundamental concept for the application of SEM for brain network analysis is effective connectivity, which is defined as the influence that one neural system exerts over another and vice versa. It differs from functional connectivity in that it is not solely defined by statistical dependencies between remote events but takes into account the activity-dependent, explicit, and directional coupling between them. Effective connectivity can be estimated from SEM by testing whether a theoretical connectivity model seeking to explain a network of influences can actually fit the influences estimated from observed data. 14 In our case, the theoretical models that we tested were based on regions known to be commonly clinically affected in sCJD (i.e., network of cortico-subcortical regions typically affected by restricted diffusion, including thalamus and striatum) [bib_ref] Multiparameter MR imaging in the 6-OPRI variant of inherited prion disease, Vita [/bib_ref] [bib_ref] Diffusionweighted MRI hyperintensity patterns differentiate CJD from other rapid dementias, Vitali [/bib_ref] [bib_ref] White matter involvement in sporadic Creutzfeldt-Jakob disease, Caverzasi [/bib_ref] or on known anatomical connectivity in healthy participants (i.e., default mode, executive, salience, and motor networks). The DMN analysis was based on , and it included the precuneus, angular gyrus (AG), anterior cingulate cortex (ACC), and mesial and lateral temporal cortical regions. [bib_ref] The brain's default network: Anatomy, function, and relevance to disease, Buckner [/bib_ref] The data used as input for the SEM model were age-and TIV-corrected cortical volumes derived from the VBM analysis. 35 The effective connectivity is described by beta coefficients (called "path coefficients" in SEM), which were generated from interregional volume correlations via a process of iterative data fitting using IBM SPSS Amos graphics software. Based on the evidence that prions propagate transynaptically (i.e., via paths), [bib_ref] The prion model for progression and diversity of neurodegenerative diseases, Stopschinski [/bib_ref] it was hypothesized that atrophy in sCJD occurs selectively and that volumes in brain regions affected would show a directional relationship (i.e., one brain region driving atrophy in another brain region). Nonetheless, given that SEM is a causal modeling statistical tool, we wanted to avoid drawing causative biological conclusion, and hence we only focus on the general model fit rather than on the beta coefficients. Theoretical models being tested were graphically represented by nodes (i.e., brain regions) exerting trophic influences through anatomical pathways (arrows), with the direction of a trophic effect represented by the arrow direction and the connectivity strength (beta coefficient) represented by arrow thickness. Models for right, left, and whole brain were tested separately. Models with optimal data fit have low error (root mean square error of approximation (RMSEA) < .05 and goodness-of-fit index (GFI) > .90). 14
# Results
Fluid, clinical, and neuroimaging biomarkers
The demographics, basic clinical features (including CSF biomarker, EEG, and MRI findings), and scores of our sCJD cohort (n = 40), its various subcohorts, and the Controls cohort (n = 26) are shown in [fig_ref] Table 1: Clinical features of sporadic Creutzfeldt-Jakob patients and Controls [/fig_ref] and in the Supplementary Material. The relationship among CSF results and other outcomes or variables is presented in Supplementary Material.
Comparing volumetric values, derived from image processing prior to VBM, between the 22 VBM sCJD cases and Controls showed significantly less total gray matter volume in sCJD (p < 0.01, Student's t-test) but no differences in total white matter or CSF [fig_ref] Table 1: Clinical features of sporadic Creutzfeldt-Jakob patients and Controls [/fig_ref].
There was significant difference in t-tau levels using dichotomization of sCJD as Fast-versus Slow-progressors (based on the molecular classification). Among the cases with available t-tau and molecular classification (n = 21 of 40 total sCJD participants), the majority of participants with nonelevated t-tau were Slow-progressors (total nine patients with normal tau: six (66%) Slow-progressors, one (11%) Fast-progressor, and two (22%) mixed prion type), whereas the majority of the participants with elevated t-tau were Fast-progressors (total 12 patients with elevated tau: nine (75%) Fast-progressors, one (8%) Slow-progressor, and two (16%) mixed) (v 2 = 10, p = 0.008).
In the total sCJD cohort (n = 40), there were no differences in clinical biomarkers (i.e., CSF biomarkers, signs/ symptoms, clinical scales, and MMSE; see Methods) based on the three common DWI MRI patterns (cortical-only vs. subcortical-only vs. cortico-subcortical). Because in our experience, patients with cortico-subcortical DWI involvement tend to have greater functional impairment, Total disease duration, months, M AE SD (Md, r) MRI-based volume (corrected for TIV) Whole brain (mm [bib_ref] Serial computed tomographic and electroencephalographic studies in Creutzfeldt-Jakob disease, Hayashi [/bib_ref] Percentages might not sum to 100% due to rounding. # = 1 pathology-proven patient had variably protease-sensitive prion disease which by definition has no prion type identified. [bib_ref] Prion diseases, Geschwind [/bib_ref] Includes signs and symptoms up until around the time of UCSF MRI. [bib_ref] Cerebral MR and CT imaging in Creutzfeldt-Jakob disease, Kovanen [/bib_ref] Abnormal value ≥ 1150 ng/mL. [bib_ref] Serial computed tomographic and electroencephalographic studies in Creutzfeldt-Jakob disease, Hayashi [/bib_ref] Abnormal value > 30 ng/mL. Comparisons between all sCJD, sCJD included in VBM, sCJD excluded from VBM, and Controls were done for all the variables in the tables and significant results are noted as below [bib_ref] MR imaging of Creutzfeldt-Jakob disease, Finkenstaedt [/bib_ref] Compared to Controls, P < 0.01 Compared to sCJD excluded from VBM analysis, P < 0.001 [bib_ref] MRI detection of the cerebellar syndrome in Creutzfeldt-Jakob disease, Cohen [/bib_ref] Compared to sCJD excluded from VBM analysis, P < 0.05 [bib_ref] Structural signature of sporadic Creutzfeldt-Jakob disease, Navid [/bib_ref] Compared to sCJD excluded from VBM analysis, P < 0.01 in order to increase the possibility of finding an effect of the DWI MRI pattern of involvement, we also compared this group to a combined group of subcortical-only and cortical-only involvement. This subcohort with cortical and subcortical involvement on DWI had higher t-tau (p = 0.014; Supplemental , NSE (p = 0.005; Supplemental , NPI scores (p = 0.021; not shown), and UPDRS-Motor scores (p = 0.027; not shown), and they were more likely to have a positive 14-3-3 (63% vs. 20%, v 2 = 6, p = 0.017; not shown) than the combined cortical-only and subcortical-only group. There were no statistically significant differences in clinical symptoms, protein 14-3-3, EEG, clinical scales, or biomarkers based on the PRNP codon 129 polymorphism (data not shown). The same analyses were performed for the 22 participants included in the VBM analysis and showed relatively similar results (Supplementary Material). In comparing the subgroup with VBM analysis to the subgroup without VBM analysis, the VBM-excluded participants had shorter disease duration, lower MMSE scores, lower Barthel scores, higher CSF t-tau concentrations, higher prevalence of positive CSF protein 14-3-3, and lower prevalence of ataxia [fig_ref] Table 1: Clinical features of sporadic Creutzfeldt-Jakob patients and Controls [/fig_ref] and Supplementary Material). Most of these findings suggest that the VBM-excluded group had poorer quality MRIs (unable to be used for VBM analysis) due to a higher degree of clinical impairment or disease severity. Interestingly, the VBM group (less impaired) appeared to be much more likely to have only cortical ribboning, whereas the impaired had subcortical involvement, either with or without cortical ribboning. This is consistent with our clinical experience.
Atrophy patterns and brain-phenotype relationships VBM analysis results of the comparisons between 22 sCJD participants and 26 Controls are shown in 3D brain views and in axial views in ,F. VBM analysis revealed gray matter reduction in sCJD compared to Controls in multiple cortical regions, including the bilateral frontopolar, mesial and inferior frontal lobes, mesial and lateral parietal lobes, bilateral lateral temporal, left mesial temporal regions, bilateral insulae, and inferior posterior right occipital regions, as well as bilateral thalami (permutationsbased correction for multiple comparisons P < 0.05). Some areas of relative sparing of atrophy compared to Controls included the motor, much of the occipital, and some dorsolateral and prefrontal cortices as well as bilateral striatum and globus pallidi. Of note, no regions showed higher volumes in sCJD compared to Controls. Interestingly, the regions with atrophy in sCJD included all areas of the functionally defined default mode and salience, possibly some of the executive control (lateral parietal), but none of the motor network hubs. [bib_ref] The brain's default network: Anatomy, function, and relevance to disease, Buckner [/bib_ref] [bib_ref] Basal ganglia circuits as targets for neuromodulation in Parkinson disease, Delong [/bib_ref] [bib_ref] Dissociable intrinsic connectivity networks for salience processing and executive control, Seeley [/bib_ref] VBM analyses revealed atrophy clusters that survived permutations-based correction for multiple comparisons within sCJD subgroups when dichotomizing based on the dementia severity (MMSE) or the presence of visual hallucinations . sCJD participants with visual hallucinations (n = 6; 1 MM1, 1 MV1, 2 MV2, 1 VV2, and 1 MV without typing available) displayed right more than left thalamus, bilateral medial orbitofrontal, rectus gyri, and right fusiform atrophy compared to those without visual hallucinations (n = 15; 1 MM1, 3 MV1, 4 MM2, 2 MV2, 1 VV2, and 4 MV without typing available, one subject excluded because no Barthel score was available to adjust for disease severitysee Methods -VBM processing and analysis). Participants with more severe dementia (MMSE score ≤ 20, n = 12; 1 MM1, 4 MV1, 3 MM2, 2 MV2, 1 VV2, and 1 MV without typing available) had significant regional volume loss in several areas including bilateral mesial and inferior frontal, bilateral insula, thalamus, cerebellum, left orbitofrontal, and right mesial temporal regions compared to those with less severe dementia (MMSE score > 20, n = 7; 1 MM1, 1 MM2, 2 MV2, 1 VV2, and 2 MV without typing available). No effects were observed when MMSE was used as a continuous variable; comparing these groups separately against Controls, however, showed significant differences. No volumetric differences were found when comparing codon 129 MM2 to MV2 (the sample size was small, however: 4 MM2 and 5 MV2) or MM to MV genotypes. Compared to Controls, Fast-progressors (n = 5; 3 MV1 and 2 VV2) had atrophy in bilateral mesial and lateral frontal, bilateral precuneal, middle temporal, postcentral, and occipitoparietal regions as well as bilateral thalamic atrophy. Slow-progressors (n = 9; 4 MM2 and 5 MV2) compared to Controls showed a similar pattern of atrophy as the Fast-progressors compared to Controls, but did not have atrophy of cerebellum, occipital, and most temporal regions . Surprisingly, comparing Fast-progressors directly to Slow-progressors showed no significant differences in regional atrophy (not shown). Participants with ataxia (n = 10; 1 MM1, 3 MV1, 2 MM2, 2 MV2, and 2 VV2) displayed a trend of posterior (left more than right) cerebellar atrophy compared to participants without ataxia (n = 10; 1 MM1, 1 MV1, 2 MM2, 2 MV2, and 4 MV without available type) (uncorrected, p < 0.001; not shown), but did not survive multiple correction. There were no statistically significant differences when comparing participants with myoclonus to participants without myoclonus. There were also no statistically significant differences when comparing participants with higher UPDRS score (n = 11; 1 MM1, 2 MV1, 2 MV2, 2 VV2, 1 MM2, 1 MV1-2, and 2 MV without 1. Regional gray matter atrophy in sporadic Creutzfeldt-Jakob disease. A-E show a 3D rendering, whereas F-J show the same data rendered in axial view. All results shown in color passed permutations-based correction for multiple comparisons p < 0.05. Orientation is neurological (e.g., left side is left brain). Redder colors (A-E) signify higher level of significance (higher t-stat). For F-J (axial views), color bar represents various t-scores. Only regions of t-scores > 2 (i.e., > 2 SD away from the mean) are shown; blue regions color have significantly greater atrophy than the comparison group. Clusters with volume reductions in sCJD compared to Controls were found in the bilateral frontopolar, mesial and inferior frontal, mesial and lateral parietal, bilateral lateral temporal and left mesial temporal, and inferior posterior right occipital regions (A, F). sCJD participants with visual hallucinations had significant volume loss in the bilateral thalami, medial orbitofrontal, rectus gyri, and right fusiform compared to participants without visual hallucinations (B, G). The sCJD group with more severe cognitive impairment (based on dichotomization by the median MMSE score) showed volume reduction in the bilateral mesial and inferior frontal, cerebellum, left orbitofrontal, and right mesial temporal regions compared to the group with less cognitive impairment (C, H). Volume differences between Slow-progressors (based on molecular classification subtype) and Controls were present in the bilateral mesial and lateral frontal, bilateral precuneal, middle temporal, postcentral, and occipitoparietal regions (Slow-progressors = 4 MM2, 5 MV2) (D, J). Volume differences in Fast-progressors, based on molecular classification, and Controls were found in bilateral mesial and lateral frontal, bilateral precuneal, middle temporal, postcentral, and occipitoparietal regions as well as occipital and temporal (Fast-progressors = 3 MV1, 2 VV2) (E, I). No volume differences were found between comparison of Fast-progressors versus Slow-progressors (not shown; see text).
typing available) to those with lower UPDRS score (n = 10; 1 MV1, 3 MM2, 3 MV2, 2 MV1-2, and 1 MV without typing available) or when comparing participants with higher NPI score (n = 9; 1 MM1, 2 MV2, 3 MM2, 1 VV2, 1 MV1-2, and 1 MV without typing available) to those with lower NPI score (n = 8; 1 MV1, 1 MM2, 3 MV2, 1 MM1-2, 1 MV1-2, and 1 MV without typing available). Some comparisons did not include all 22 sCJD participants if subjects were missing a variable being controlled for or relevant to that analysis (e.g., Barthel or sCJD molecular classification).
## Modeling atrophy covariance
Based on our VBM analysis, the atrophy patterns in sCJD involved all areas of the default mode network (DMN), except minimal hippocampal involvement, but also involved other regions outside of the DMN. The DMN is a topographically distant but highly connected network of functionally connected hubs. [bib_ref] The brain's default network: Anatomy, function, and relevance to disease, Buckner [/bib_ref] Although we found high correlations of the volumetric data between these regions, we could not deduce whether these volumetric changes are happening simultaneously or if one region is influencing the atrophy in another (Supplementary [fig_ref] Table 1: Clinical features of sporadic Creutzfeldt-Jakob patients and Controls [/fig_ref] shows the Pearson correlation between some of the brain regions). Because of this and that many of these same areas are preferentially affected with abnormal diffusion in sCJD, [bib_ref] Multiparameter MR imaging in the 6-OPRI variant of inherited prion disease, Vita [/bib_ref] [bib_ref] Diffusionweighted MRI hyperintensity patterns differentiate CJD from other rapid dementias, Vitali [/bib_ref] [bib_ref] Application of quantitative DTI metrics in sporadic CJD, Caverzasi [/bib_ref] we investigated whether a structural equation model would fit the data and, if so, whether these simultaneously implicated atrophic regions are affected due to intrinsic network vulnerability independent of connectivity or if one brain region could be mathematically, although not necessarily biologically, influencing atrophy in another region. To test this hypothesis, we evaluated volumetric interactions in brain connectivity network models using SEM. Models with optimal data fit (RMSEA < .05 and GFI > .90) should fit only in CJD but not Controls, and the beta coefficient from region A to region B in the SEM model would mathematically indicate a directional relationship. 14 Tests of data fitness were done first in four functional networks: 1) DMN [bib_ref] The brain's default network: Anatomy, function, and relevance to disease, Buckner [/bib_ref] ; 2) motor 36 ; 3) executive control 37 ; and 4) salience networks. [bib_ref] Dissociable intrinsic connectivity networks for salience processing and executive control, Seeley [/bib_ref] None of these SEM models met the significance and goodness-of-fit thresholds in either sCJD or Controls (RMSEA > 0.05 and GFI < 0.90). A model, however, based on the DMN model plus the addition of deep nuclei regions (striatum and thalamus)which commonly show restricted diffusion in sCJDresulted in optimal data fit in sCJD but not the Controls . For sCJD and Control groups, left, right, and whole brain volumetric data were tested separately in the model, with only sCJD showing a significant effect but not the Controls (all RMSEA > 0.05 and all GFI < 0.90). The most meaningful model clinically, however, is the whole brain model, as it allows connectivity between hemispheres. The whole brain model showed that volumetric changes in the precuneus influenced the volumetric changes in the anterior cingulate cortex, angular gyrus, and mesial and lateral temporal cortex regions, whereas the connectivity was commensurate between the precuneus and the thalamus. Furthermore, striatal atrophy appeared to cause precuneus atrophy but not vice versa (as RMSEA > .05 and GFI < .90, not shown). Thus, coupled (connected) regions within the network changed their volumes together (i.e., high covariance), and these changes were directional within the network. Thus, there was SEM effective connectivity present in sCJD in a model including the DMN plus the striatum and thalamus, with the precuneus being the main influencer of directionality.
# Discussion
## Summary of major findings
This high-resolution 3T MRI VBM study of 22 sCJD cases provides evidence that regional, but not global, atrophy is a feature of sCJD. Selective regional atrophy was observed in a composite of cortical and deep nuclei regions, including DMN hubs, [bib_ref] The brain's default network: Anatomy, function, and relevance to disease, Buckner [/bib_ref] salience network (SN) hubs, [bib_ref] Dissociable intrinsic connectivity networks for salience processing and executive control, Seeley [/bib_ref] the thalami as well as other regions. Specifically, selective atrophy in sCJD (compared with Controls) was predominant in multimodal association regions, including the bilateral precunei, bilateral lateral parietal, bilateral mesial prefrontal cortices, and bilateral lateral temporal (all DMN regions, with the hippocampus as the only DMN region being only minimally involved), and bilateral anterior cingulate and insula (SN regions), in addition to bilateral lateral and inferior prefrontal cortices, and bilateral thalami . Notably, in sCJD there was relative sparing of atrophy in the dorsolateral prefrontal cortex (part of the executive control network) and no atrophy in the primary motor, sensory, and visual cortices, as well as the striatum and cerebellum. We explored whether certain clinical variables and symptomatology were associated with regions of atrophy. Both Slow-and Fast-progressors (defined based on sCJD molecular classification) 5,29 had significant atrophy compared to Controls but, to our surprise, were not statistically significantly different from each other. The sCJD subgroup with visual hallucinations had significantly greater atrophy in the right fusiform gyrus, right more than left thalamus, and bilateral orbitofrontal areas compared to the subgroup without visual hallucinations. Although MMSE as a continuous variable did not correlate with atrophy, participants with lower MMSE scores (dichotomized by median MMSE of 20) had more frontal, temporal, insula, thalamus, and cerebellum atrophy compared with participants with higher MMSE scores. Finally, SEM of volumetric brain data ª 2021 The Authors. Annals of Clinical and Translational Neurology published by Wiley Periodicals LLC on behalf of American Neurological Association revealed significant covariance and trophic (directional) effects among key functional nodes of the DMN (precuneus, anterior cingulate cortex, angular gyrus, and mesial and lateral temporal cortex) [bib_ref] The brain's default network: Anatomy, function, and relevance to disease, Buckner [/bib_ref] as well as the thalamus and striatum in sCJD but not in Controls. This model of effective connectivity showed trophic influence of subcortical regions on the precuneus and suggests a central role for the precuneus in driving volumetric changes in other cortical regions.
## Atrophy in scjd (vs. controls) and comparison to literature
The pattern of selective brain volume reduction in sCJD overlaps with its typical DWI-restricted diffusion pattern, which involves the same multimodal association regions including the cingulate, precuneus, angular gyrus, and superior and middle frontal gyri, with relative sparing of primary motor and sensory cortices. [bib_ref] Multiparameter MR imaging in the 6-OPRI variant of inherited prion disease, Vita [/bib_ref] [bib_ref] Diffusionweighted MRI hyperintensity patterns differentiate CJD from other rapid dementias, Vitali [/bib_ref] [bib_ref] White matter involvement in sporadic Creutzfeldt-Jakob disease, Caverzasi [/bib_ref] We might expect that areas first involved in sCJD would be the areas to show greatest atrophy later on in the disease. We found five case reports of neurologically normal persons who had brain MRIs for reasons unrelated to sCJD (e.g., carotid bulb tumor, studies assessing utility of annual MRI for standard of clinical care, etc) and who became symptomatic with sCJD between 3 and 14 months later. Overall, these cases showed DWI abnormalities (reduced diffusion) in the presymptomatic phase of disease in association cortices including the bilateral temporo-parietaloccipital junction, lateral parietal, precunei, and mesial . Sporadic Creutzfeldt-Jakob disease selectively changes the effective connectivity between specific cortical and subcortical brain regions that overlap with the default mode network nodes. The figure shows the models of brain effective connectivity when brain volume data are tested in a network of cortical and subcortical regions usually noted by the authors to be commonly affected clinically on diffusion imaging in sCJD, specifically the default mode network plus the striatum and thalamus. Two key take-away points from this figure are (1) the model fit in sCJD but not in Controls, and (2) that the precuneus (PrC) seems to play a central role in influencing volumetric changes in other regions. In the following text, we explain the SEM model and the meaning of the arrows from a mathematical standpoint. The graphs represent anatomical nodes in boxes connected by paths of trophic influence (arrows) that determine the regional volumetric influence on the target nodes. The effective connectivity (i.e., direction of the trophic effect) is represented by the arrow direction. Connectivity strength (i.e., strength of an effect) is represented by path coefficients (i.e., beta coefficient) displayed by the number over each arrow, with higher numbers meaning stronger tropic influence. The thickness of the arrow is a visual representation of the strength of the correlation and the dashed lines representing a negative correlation. Positive values indicate induction of atrophy in the direction of the arrow, whereas negative values indicate induction of increased volume. Goodness-of-fit statistics (GFIs) > .900 are considered significant with the p value equivalent shown by root mean square error of approximation (RMSEA)-only the models in sCJD, and none of the models in Controls, were significant (significant results are indicated with an *). In the whole brain and the right hemisphere models, and partially in the left hemisphere model, the precuneus exerts a large and disproportionate effect on the anterior cingulate (ACC), angular gyrus (AG), and temporal lobe (Temp). For example, in the whole brain model, one-unit change in the precuneus volume results in 1.62, .76, and 1.28 points change in the ACC, AG, and Temp, respectively. Conversely, changes in the ACC, AG, and Temp volumes results in À.23, .17, and .21 unit change, respectively, in the precuneus. Interestingly, compared to the tropic influence of the precuneus, the effects were more balanced between the thalamus (Thal) and the precuneus and were unidirectional from the striatum (Str) to the precuneus. Models that included bidirectional effect between the precuneus and the Str did not meet the goodness-of-fit and the statistical significance parameters. This suggests that the striatum influenced atrophy of the precuneus, but not the reverse. L = left hemisphere, R = right hemisphere, C = combined or bilateral structure. frontal regions, [bib_ref] Long-term preclinical magnetic resonance imaging alterations in sporadic Creutzfeldt-Jakob disease, Zanusso [/bib_ref] [bib_ref] An autopsied case of MM1 + MM2-cortical with thalamic-type sporadic Creutzfeldt-Jakob disease..., Iwasaki [/bib_ref] [bib_ref] Early detection of sporadic CJD by diffusion-weighted MRI before the onset ª..., Satoh [/bib_ref] [bib_ref] Cortical hyperintensity on diffusion-weighted images as the presymptomatic marker of sporadic creutzfeldt-Jakob..., Maeda [/bib_ref] [bib_ref] MRI abnormalities found 1 year prior to symptom onset in a case..., Verde [/bib_ref] similar to regions where we detected atrophy with VBM. Furthermore, our findings are consistent with studies in human genetic prion disease showing atrophy in cortical association regions. [bib_ref] Distinct neuropsychological profiles correspond to distribution of cortical thinning in inherited prion..., Alner [/bib_ref] [bib_ref] Multiparameter MR imaging in the 6-OPRI variant of inherited prion disease, Vita [/bib_ref] Alner et al. 2011, using 1.5T MRI and FreeSurfer imaging analysis, described cortical thickness reduction in 10 patients with a genetic prion disease (gPrD; 6-OPRI PRNP mutation), predominantly in the precuneus, supramarginal, parietal lobule, and lingual regions. [bib_ref] Distinct neuropsychological profiles correspond to distribution of cortical thinning in inherited prion..., Alner [/bib_ref] In a follow-up study also with 1.5T MRI, but using SPM8 VBM analysis, De Vita et al. 2013 described precuneus, perisylvian, lingula, and basal ganglia volume reduction in nine symptomatic patients with the same 6-OPRI PRNP mutation. [bib_ref] Multiparameter MR imaging in the 6-OPRI variant of inherited prion disease, Vita [/bib_ref] Similarly, using 3T VBM analysis in 30 symptomatic predominantly gPrD subjects (10% sCJD), Caine et al. described volume reductions in frontal and parietal gray matter volumes which correlated with predominant frontoparietal dysfunction on neurocognitive testing.These three studies all showed atrophy in regions for which we also found atrophy in our sCJD cohort . In a study using 3T VBM analysis, Grau-Rivera et al. 2015 found significant gray matter loss in 15 CJD cases (13 sCJD and 2 genetic, E200K) specifically in the bilateral thalami, putamen, fusiform gyrus, cerebellum, and left perirolandic cortex. [bib_ref] Quantitative magnetic resonance abnormalities in creutzfeldt-jakob disease and fatal insomnia, Grau-Rivera [/bib_ref] Although we had similar thalamic and fusiform findings, we did not observe basal ganglia (striatum and globus pallidus), perirolandic, or cerebellar volume loss in sCJD vs. Controls. It is interesting that we did not find striatal volume loss despite visible striatal diffusion reductions present in 60% of our sCJD participants [fig_ref] Table 1: Clinical features of sporadic Creutzfeldt-Jakob patients and Controls [/fig_ref]. This parallels findings in a study by [bib_ref] Putaminal volume and diffusion in early familial Creutzfeldt-Jakob disease, Seror [/bib_ref] with 12 patients with E200K genetic prion disease who had striatal-restricted diffusion but did not show quantifiable volume reduction. [bib_ref] Putaminal volume and diffusion in early familial Creutzfeldt-Jakob disease, Seror [/bib_ref] In our experience, for most sCJD cases, striatal involvement on DWI usually occurs after extensive cortical ribboning appears (except in MV2 and VV2 cases which usually have isolated deep nuclei involvement). The fact that we found overlapping but slightly different cortical areas involved than these other studies might be due to other studies using mostly genetic cases and few sporadic cases or to smaller sample sizes in Navid et al. [bib_ref] Structural signature of sporadic Creutzfeldt-Jakob disease, Navid [/bib_ref] (n = 11) and Grau-Rivera et al. [bib_ref] Quantitative magnetic resonance abnormalities in creutzfeldt-jakob disease and fatal insomnia, Grau-Rivera [/bib_ref] (n = 13) compared with 22 sCJD subjects in our study.
## Atrophy in scjd subgroups and comparison to literature
As noted above, we predicted that Fast-progressors might not have sufficient time to develop detectable atrophy compared to Slow-progressors (Fast and Slow based on sCJD molecular classification). Contrary to our prediction, however, each of these subgroups showed significant atrophy when compared to Controls. The atrophy pattern in Fast-progressors compared to Controls and Slow-progressors compared to Controls was relatively similar . On visual inspection of the VBM data, Fast-progressors had more atrophic areas than Slow-progressors, including cerebellum, occipital, and temporal regions. These differences, however, did not reach statistical significance, which might be due to the relatively small sample size. Nevertheless, these findings suggest that brain atrophy is present, even in Fast-progressors, and although usually not detectable by visual assessment, it can be quantified using VBM. To our knowledge, VBM analysis of the Fast-versus Slow-progressors in sCJD based on molecular classification has not been reported previously.
We also identified a number of brain-behavior relationships with VBM in sCJD. Participants with visual hallucinations showed more significant right fusiform, right more than left thalamus, and bilateral medial orbitofrontal volume loss compared with participants without visual hallucinations . The involvement of the fusiform gyrus is consistent with previous studies showing atrophy on VBM and functional (fMRI) abnormalities in visual association areas in patients with neurodegenerative conditions such as Parkinson's disease. [bib_ref] Cerebral basis of visual hallucinations in Parkinson's disease: Structural and functional MRI..., Ibarretxe-Bilbao [/bib_ref] Similarly, orbitofrontal cortex involvement in Parkinson's disease dementia patients experiencing visual hallucinations has been reported in multiple studies, including VBM, [bib_ref] Frontal and associative visual areas related to visual hallucinations in dementia with..., Sanchez-Castaneda [/bib_ref] [bib_ref] Meta-analytic evidence for the plurality of mechanisms in transdiagnostic structural MRI studies..., Rollins [/bib_ref] selective serotonin 2A receptor ligand F-18 PET, [bib_ref] Serotonin 2A receptors and visual hallucinations in Parkinson disease, Ballanger [/bib_ref] and diffusion tensor imaging studies. [bib_ref] Abnormalities in white matter connections between orbitofrontal cortex and anterior cingulate cortex..., Ohtani [/bib_ref] Moreover, fluorodeoxyglucose-PET (FDG-PET) hypometabolism in the orbitofrontal cortex has been found in patients with schizophrenia who have visual hallucinations. [bib_ref] A functional neuroanatomy of hallucinations in schizophrenia, Silbersweig [/bib_ref] Furthermore, recent studies suggest a role for thalamic-DMN decoupling as the mechanism for visual hallucinations in synuclinopathies. [bib_ref] Hallucinations, somatic-functional disorders of PD-DLB as expressions of thalamic dysfunction, Onofrj [/bib_ref] The sCJD group with more severe cognitive impairment, based on dichotomization of MMSE scores had scattered volume loss in the bilateral mesial orbitofrontal, inferior frontal, posterior central, bilateral middle temporal, and bilateral medial precuneus gyri, in addition to more confluent volume loss in the bilateral insula, thalamus, and cerebellum, even after adjusting for the time point of MRI in the disease course (ratio of disease duration at the time of MRI over total disease duration). When we corrected for Barthel, however, this effect went away, probably because Barthel and MMSE were correlated. This atrophy pattern is consistent with the role of the cerebellar-thalamic-cortical connections in a wide range of cognitive functions. [bib_ref] The cerebellar cognitive affective syndrome, Schmahmann [/bib_ref] Although it did not survive correction for multiple comparisons, sCJD participants with ataxia compared with those without ataxia showed a trend toward greater left posterior cerebellar atrophy (not shown), which is consistent with a previous finding of a correlation ª 2021 The Authors. Annals of Clinical and Translational Neurology published by Wiley Periodicals LLC on behalf of American Neurological Association between cerebellar atrophy and ataxia in E200K genetic prion disease. [bib_ref] MRI detection of the cerebellar syndrome in Creutzfeldt-Jakob disease, Cohen [/bib_ref] Perhaps a larger sample size is needed to show this effect in sCJD.
## Cortico-subcortical regions as the epicenter in scjd
From a brain network perspective, our VBM data support the idea that frontoparietal and temporal regions, including but not limited these areas within the DMN, [bib_ref] The brain's default network: anatomy, function, and relevance to disease, Buckner [/bib_ref] as well as several subcortical structures, are at the epicenter of neurodegenerative changes in sCJD. The selective volume reduction we observed in sCJD within most of the DMN (only minimal hippocampal atrophy) and other regions may be explained by both distinctive regional neurometabolic profiles and network connectivity properties in this disease. For example, parietal regions such as the precuneus in particular, but also certain occipital and temporal regions, display the highest state of oxidative neuronal metabolism in the brain. [bib_ref] Regional analysis of FDG and PIB-PET images in normal aging, mild cognitive..., Li [/bib_ref] [bib_ref] Normal patterns of regional brain18F-FDG uptake in Normal aging, Shamchi [/bib_ref] In such regions with high baseline metabolism, oxidative stress may play a crucial role in prion protein misfolding. [bib_ref] Cellular prion protein (PrPC) of the neuron cell transformed to a PK-resistant..., Yuan [/bib_ref] These results are in line with an FDG-PET study of sCJD showing hypometabolism to be common particularly in parietal and temporal regions. [bib_ref] Positron emission tomography with [(18)F]FDG in the diagnosis of Creutzfeldt-Jakob disease (CJD), Henkel [/bib_ref] Networks, such as the DMN, often feature a number of heteromodal association cortices that operate as hub nodes. Such networks are resistant to random node failure but become vulnerable when hub failure occurs, thus facilitating transsynaptic degeneration or other forms of network pathological spread. [bib_ref] Error and attack tolerance of complex networks, Albert [/bib_ref] The biological vulnerability of these hubs is supported by its implication as a pathogenic substrate in various neurodegenerative diseases. For example, Alzheimer's disease (AD), which like prion disease involves protein misfolding, has prominent posterior DMN hypometabolism and atrophy. [bib_ref] Network dysfunction in Alzheimer's disease and frontotemporal dementia: implications for psychiatry, Zhou [/bib_ref] Interestingly, the pattern of cortical atrophy in sCJD, in part targeting DMN hubs, is reminiscent of the involvment of the DMN in AD. [bib_ref] Network dysfunction in Alzheimer's disease and frontotemporal dementia: implications for psychiatry, Zhou [/bib_ref] [bib_ref] Defaultmode network activity distinguishes Alzheimer's disease from healthy aging: evidence from functional..., Greicius [/bib_ref] This also is consistent with literature suggesting shared pathogenic mechanisms and overlap between AD and sCJD. [bib_ref] A unifying role for prions in neurodegenerative diseases, Prusiner [/bib_ref] [bib_ref] Self-propagation of pathogenic protein aggregates in neurodegenerative diseases, Jucker [/bib_ref] [bib_ref] Prion Disease Induces Alzheimer disease-like neuropathologic changes, Tousseyn [/bib_ref] [bib_ref] Evidence for human transmission of amyloid-beta pathology and cerebral amyloid angiopathy, Jaunmuktane [/bib_ref] Furthermore, selective vulnerability of networks containing high-degree hubs is also characteristic of other neurodegenerative disorders such as behavioral variant frontotemporal dementia, semantic dementia, progressive supranuclear palsy, and corticobasal syndrome. [bib_ref] Dissociable intrinsic connectivity networks for salience processing and executive control, Seeley [/bib_ref] [bib_ref] Intrinsic connectivity network disruption in progressive supranuclear palsy, Gardner [/bib_ref] [bib_ref] Intrinsic connectivity networks in posterior cortical atrophy: a role for the pulvinar?, Fredericks [/bib_ref] VBM results implicating the involvement of cortical and subcortical regions in sCJD (through atrophy) were supported using SEM in connectivity models that were specifically tested for the presence of network-level involvement in sCJD. The cortico-subcortical effective connectivity in sCJD can be interpreted as a common trophic fate (i.e., volume reduction) for involved cortical and subcortical regions. A structural connectivity model that involved DMN regions plus subcortical regions consisting of the thalamus and striatum fit in the sCJD group but not in the Controls . In both sCJD and Controls, alternative models involving either the DMN without the addition of subcortical structures, as well as the executive, salience, or motor networks did not fit the model. In the cortico-subcortical network model that met the strict significance thresholds of SEM (GFIs>.90 and RMSEA<.05), the strongest path effect was from the precuneus to the anterior cingulate cortex, mesial and lateral temporal, and angular gyrus regions in the left, right, and combined hemispheres of the sCJD group but not in the Controls. These SEM findings potentially point to a central role for the precuneus in sCJD connectivity-mediated neurodegeneration. This finding is consistent with our clinical experience and literature suggesting that the precuneus and the adjacent posterior cingulate cortex are very commonly involved by visual assessment of the DWI/ADC maps and quantitative mean diffusivity measurement in sCJD. [bib_ref] Diffusionweighted MRI hyperintensity patterns differentiate CJD from other rapid dementias, Vitali [/bib_ref] [bib_ref] White matter involvement in sporadic Creutzfeldt-Jakob disease, Caverzasi [/bib_ref] The precuneus exhibited trophic influences on other cortical regions, whereas its trophic influence on the thalamus was bidirectionally equivalent, and the striatum exhibited a unidirectional trophic influence on the precuneus. We speculate that the bidirectional effects on a group-wise level could be due to the high connectivity between the DMN hubsonce one region is affected it begins influencing atrophy in other highly connected regions. Thus, regardless of where a patient's disease begins, involvement eventually converges on this highly connected network. Some bidirectionality, however, was positive in one direction and negative in another, meaning that atrophy in one region is influencing increased volume in another region (negative arrow in . We specultate that this might occur because astrocytosis, neuroinflammation, and/or possibly edema involved in neurodegeneration might produce volumetric changes that can be interpreted by SEM as increased volumes within the sCJD group, which we simply were not able to detect in our VBM analyses.
Despite the striatum not showing atrophy on VBM, the SEM showed a volumetric trophic effect of the striatum on the precuneus in sCJD, which was not found in Controls. We suspect that, despite the lack of detectable striatal changes on VBM in sCJD, the striatum likely plays an important role on the network-level pathogenesis of sCJD, along a continuum of diffusivity changes preceding atrophy. These findings point to the importance of subcortical regions, such as the striatum and thalamus, in the pathophysiology of sCJD, as similar models including only cortical regions were not solvable and only models including both cortical and subcortical regions were. Furthermore, this SEM model provides mathematical evidence for the hypothesis that the selective atrophy pattern in sCJD is mediated by atrophy spread via anatomical connections, such as those used for the transsynaptic spread of prions, rather than selective atrophy due to intrinsic regional vulnerability alone, independent of connectivity and transsynaptic prion spread. In other words, if atrophy is mediated only by intrinsic regional vulnerability, then we would expect high volumetric correlations in atrophied regions but without a directional relationship as we found in our SEM analysis.
We also identified several associations between various biomarkers and clinical outcomes in both the entire cohort of 40 sCJD participants who had DWI and the subcohort of 22 sCJD participants who also had sufficient quality T1 MRI volumetric analysis. sCJD participants with both cortical and subcortical DWI involvement had higher CSF biomarkers (t-tau and NSE) and more motor and behavioral dysfunction compared with the group combining cortical-only and subcortical-only involvement. This is consistent with our clinical experience in which we find that patients with both cortical and subcortical DWI involvement tend to decline more rapidly, particularly compared to those with cortical-only involvement. Whole brain volumes inversely correlated with CSF t-tau levels, which is generally consistent with the modestly high sensitivity and specificity of an elevated t-tau in sCJD. [bib_ref] Comparing CSF biomarkers and brain MRI in the diagnosis of sporadic Creutzfeldt-Jakob..., Forner [/bib_ref] Similarly, t-tau levels were inversely correlated with total disease duration and MMSE scores (i.e., participants with lower levels had longer disease durations and better cognitive performance), which parallels previously observed changes of t-tau with disease duration [bib_ref] Influence of timing on CSF tests value for Creutzfeldt-Jakob disease diagnosis, Sanchez-Juan [/bib_ref] [bib_ref] CSF tests in the differential diagnosis of Creutzfeldt-Jakob disease, Sanchez-Juan [/bib_ref] [bib_ref] Diagnostic performance of cerebrospinal fluid total tau and phosphorylated tau in Creutzfeldt-Jakob..., Skillback [/bib_ref] [bib_ref] Association of blood and cerebrospinal fluid tau level and other biomarkers with..., Staffaroni [/bib_ref] and cognitive function in prion disease.This study has a number of limitations. Only cross-sectional volumetric quantifications were performed, as we had too few cases with sufficient quality serial data. Longitudinal studies, which are often difficult to obtain in sCJD, particularly of sufficient quality for volumetric analysis, will be needed to better characterize atrophy progression in sCJD. Although our cohort was relatively large for a single-site sCJD imaging cohort, our sample size nevertheless is relatively small, limiting our ability to identify more interactions of clinical variables and volumetric changes or to compare all six main sCJD molecular classification subgroups. The fact that even with our limited sample size, however, we still found any associations suggests the strength of our findings. Although our cohort with DWI (n = 40) included all molecular subtypes and had a plurality of Fast-progressors (MM1, MV1, and VV2) consistent with the general sCJD population, [bib_ref] Determinants of diagnostic investigation sensitivities across the clinical spectrum of sporadic Creutzfeldt-Jakob..., Collins [/bib_ref] [bib_ref] Molecular basis of phenotypic variability in sporadic Creutzfeldt-Jakob disease, Parchi [/bib_ref] Fast-progressors were underrepresented in the subcohort included in the VBM analysis because of difficulties obtaining sufficient quality volumetric scans due to motion artifact. This may represent sampling bias favoring cases with slower progression. Furthermore, we only compared cases whom could easily be classified as Fast-or Slow-progressors based on their sCJD molecular classificaiton, excluding five cases (four with mixed prion types and one variably protease-sensitive prionopathy). This dichotomization of sCJD into Fast-vs. Slow-progressors based on the mean disease duration of each of the six major molecular subgroups has limitations. For example, even though on average a molecular classification subtype is rather homogenous, an individual patient might be more heterogenous. Additionally, the clinical and biomarker data of the VBM-excluded participants suggest that these participants in general had more severe disease, which likely interfered with MRI acquisition and led to VBM analysis of the relatively less severely affected subjects. Lastly, although our cohort might appear to have longer survival than many other large published sCJD cohorts, our 2-day research visit protocol attempts to identify the very first symptoms, which we commonly find to be several weeks and usually months earlier than the medical records or even the family had initially reported. This results in our patients' disease durations appearing longer than those reported in national surveillance cohorts that are largely based on retrospective external record review. [bib_ref] A unifying role for prions in neurodegenerative diseases, Prusiner [/bib_ref] [bib_ref] First symptom in sporadic Creutzfeldt-Jakob disease, Rabinovici [/bib_ref] [bib_ref] Differential diagnosis of Jakob-Creutzfeldt disease, Paterson [/bib_ref] Our cohort in this study had extensive neurological, PRNP codon 129 genotyping, and CSF biomarker characterization and, to our knowledge, represents the first quantitative VBM study exclusively in sporadic prion disease. The results of this study add a valuable dimension to the neuroimaging characterization of sCJD and support the value of unbiased quantification of brain volume as a potential clinical biomarker in sCJD. Trophic SEM analyses of VBM data suggest that atrophy in sCJD might not just be due to regional vulnerability, but that there is a directionality of influence of atrophy in sCJD, which we hypothesize might be influenced by propagation of prions along synaptic pathways.
confirmation of many of our sCJD cases. The authors also express their gratitude to the patients and their loved ones for participating in our research program.
[fig] Figure: 1. Regional gray matter atrophy in sporadic Creutzfeldt-Jakob disease. A-E show a 3D rendering, whereas F-J show the same data rendered in axial view. All results shown in color passed permutations-based correction for multiple comparisons p < 0.05. Orientation is neurological (e.g., left side is left brain). Redder colors (A-E) signify higher level of significance (higher t-stat). For F-J (axial views), color bar represents various t-scores. Only regions of t-scores > 2 (i.e., > 2 SD away from the mean) are shown; blue regions color have significantly greater atrophy than the comparison group. Clusters with volume reductions in sCJD compared to Controls were found in the bilateral frontopolar, mesial and inferior frontal, mesial and lateral parietal, bilateral lateral temporal and left mesial temporal, and inferior posterior right occipital regions (A, F). sCJD participants with visual hallucinations had significant volume loss in the bilateral thalami, medial orbitofrontal, rectus gyri, and right fusiform compared to participants without visual hallucinations (B, G). The sCJD group with more severe cognitive impairment (based on dichotomization by the median MMSE score) showed volume reduction in the bilateral mesial and inferior frontal, cerebellum, left orbitofrontal, and right mesial temporal regions compared to the group with less cognitive impairment (C, H). Volume differences between Slow-progressors (based on molecular classification subtype) and Controls were present in the bilateral mesial and lateral frontal, bilateral precuneal, middle temporal, postcentral, and occipitoparietal regions (Slow-progressors = 4 MM2, 5 MV2) (D, J). Volume differences in Fast-progressors, based on molecular classification, and Controls were found in bilateral mesial and lateral frontal, bilateral precuneal, middle temporal, postcentral, and occipitoparietal regions as well as occipital and temporal (Fast-progressors = 3 MV1, 2 VV2) (E, I). No volume differences were found between comparison of Fast-progressors versus Slow-progressors (not shown; see text). [/fig]
[table] Table 1: Clinical features of sporadic Creutzfeldt-Jakob patients and Controls. ª 2021 The Authors. Annals of Clinical and Translational Neurology published by Wiley Periodicals LLC on behalf of American Neurological Association [/table]
|
Gathering Structured Patient Insight to Drive the PRO Strategy in COPD: Patient-Centric Drug Development from Theory to Practice
We illustrate our experience of gathering patient insights on the most patient-relevant symptoms in chronic obstructive pulmonary disease (COPD) via a structured and systematic approach towards 'patient-centric' drug development, leveraging recent advances in digital technologies using online platforms. The fourstep approach comprised the following: literature search, social media listening (SML) study, online bulletin board (OBB) exercise, and design of an online patient preference study (PPS). The initial online studies (SML and OBB) revealed that, besides dyspnoea and exacerbations,
# Abstract
We illustrate our experience of gathering patient insights on the most patient-relevant symptoms in chronic obstructive pulmonary disease (COPD) via a structured and systematic approach towards 'patient-centric' drug development, leveraging recent advances in digital technologies using online platforms. The fourstep approach comprised the following: literature search, social media listening (SML) study, online bulletin board (OBB) exercise, and design of an online patient preference study . The initial online studies (SML and OBB) revealed that, besides dyspnoea and exacerbations, patients perceive cough and mucus production as equally important aspects of disease management for COPD. To further build and quantify patients' understanding of the importance of these symptoms, an online patient preference survey is underway. Based on these findings, we have elected to include the Cough and Sputum Assessment Questionnaire or CASA-Q, a validated instrument to collect patient-reported outcomes (PRO), besides the use of the COPD assessment test or CAT to assess the severity and impact of COPD in drug development studies for COPD. Additionally, to capture movement and sleep disturbance, we consider the inclusion of actigraphy as a digital evidence-capture end point. Lastly, in a phase II trial, a survey questionnaire on incontinence will be administered to evaluate the importance of this issue among patients. We believe that integrating insights derived from ''online'' studies (SML, OBB, and PPS) into drug development offers an opportunity to truly listen to patients' voices in early product design ensuring relevance of end points selected for the clinical trial program. This approach also has the potential to complement conventional qualitative and quantitative data collection requirements for PRO instrument development. While awaiting final guidance from the US Food and Drug Administration, or FDA, the recently released draft documents on collecting representative patients' input reference social media as a tool to collect qualitative patient preference data and these developments suggest that patient preference data can influence future clinical trial design, end point selection, and regulatory reviews.
Funding: Novartis Pharma AG, Basel.
# Introduction
The healthcare industry is undergoing a paradigm shift towards a more patient-centric approach as patients take centre stage in the contemporary patient-focused drug development (PFDD) process. For successful development of meaningful therapies, it has become vital to determine what matters most to patients living with the respective condition. A growing number of pharmaceutical and device companies are now incorporating patients' perspectives throughout the drug development process. As active patient engagement is increasing, it has become critical to systematize the tools and approaches used to collect patients' perspectives and design a standardized process that can inform drug development based on patient-relevant end points and outcomes. Regulatory authorities do acknowledge the importance of early patient engagement and appreciate and early engagement with authorities on methods to collect patients' insights. The evolution of internet from the static Web to the highly participative and intuitive Web 2.0 has transformed communication channels for patients. Besides seeking traditional sources of healthcare information, patients are increasingly using online health forums to exchange information. Studying such online interactions from social media provides an opportunity to understand expectations of patients related to their disease condition. In addition, in the digital era, it is possible to engage patients virtually without the need for face-to-face interactions.
To leverage the recent advancements in digital technologies and exploit available online platforms to gain insights regarding symptoms that would be most relevant from patients' perspectives in chronic obstructive pulmonary disease (COPD), we considered a four-step approachWe illustrate our experience with this approach to gathering patient insights in a structured and systematic way from patients with COPD and how we are leveraging such early insights to select relevant patient-reported outcome (PRO) end points to use in prospective clinical program for an early drug development project. This article is based on previously conducted studies and does not report any new data with human participants or animals performed by any of the authors.
## Patient insights through social media and online platforms
## Literature review
The first step in our approach was a targeted literature review, performed to understand published literature regarding patient needs and expectations related to COPD. The findings of this literature review revealed that most of the patient preference studies (15 of the total 22 retrieved studies in COPD) have traditionally evaluated key attributes around device preferences. Only two studies focused on the diseaserelated aspects that patients would most likely to alleviate or on an ideal treatment profile for COPD from the patients' perspective. This analysis offered us an understanding of the current knowledge gap concerning preferences of COPD patients and provided the basis for developing the search strategy for the subsequent social media analysis.
## Sml
The next step was an SML study-an analysis of online conversations in open-access platforms among patients with COPD. Analysis of patient interactions on social media is a quick approach to obtain rich information about patient unmet needs, disease management, and delivery of care. Importantly, SML allows researchers to ''listen'' to the open discussions that are already happening online in the patients' own words, without influencing those conversations and imposing any further research burden on the patient community. Although it is convenient to gather patient inputs via studying social media postings, there are inherent limitations due to the nature of SML, such as self-selection bias (not everyone engages online; greater likelihood of negative aspects being posted; may include a narrow representation of clinical/demographic characteristics) limiting generalizability; no verification of identity; and the passive observational nature not allowing further probing to seek clarification on the posted views. At this time there is a lack of guidance surrounding the key ethical issues such as privacy, anonymity, and informed consent (voluntary participation) that may arise in analysing the content posted on social media. To respect patient privacy during the SML study conduct and reporting, we adhered to the data privacy obligations in accordance with the Health Insurance Portability and Accountability Act (HIPAA). All the data for SML study were obtained from publicly accessible sources without accessing password-protected information and only aggregated qualitative findings derived based on the anonymized content were reported. The detailed methodology along with results of COPD SML study has been reported elsewhere. As a follow-up to the SML study, it is important to conduct qualitative research studies with patients to confirm and expand on the findings from SML.
## Obb
Subsequently we conducted an OBB study to further validate and explore the patient insights revealed from SML in greater detail with a qualitative approach. Twenty COPD patients (UK, n = 10; USA, n = 10) participated in the OBB that ran for over 2 weeks. Being an asynchronous, moderated, closed online community platform (similar to a private chat room) for qualitative patient research, an OBB allows participants to answer predefined questions in a comprehensive manner. Information derived from the literature search and SML study formed the basis for the questions developed for the OBB. An OBB offers the interactive benefits of a face-to-face focus group discussion, but without certain individuals dominating the conversation. It also has the potential to generate deep insights, including emotional and sensitive aspects of living with a disease, which may not surface in a traditional interview settings. A key strength of this exercise is that it provides a platform for candid, honest, and shared patient-to-patient interactions while maintaining anonymity through Key steps and corresponding questions/objectivesKey steps Pertinent questions/objectives to understand relevant patient perspective
Step 1: qualitative observational-SML Observe patient conversations
What affects, what motivates the patients?
What are the questions, pains, experiences, concerns?
How do they communicate about their disease?
Step Which trade-offs can be accepted?
What would make a future product attractive to the patient?
OBB online bulletin board, SML social media listening the study. When the participating patients were contacted for debriefing at the end of the OBB session, the patients expressed agreement with the conclusions derived from the discussions and also expressed that they were very engaged throughout their participation in the study. The comments shared by patients on the final day of the OBB study reflected that patients really valued the opportunity to participate in this form of research and share experiences with 'like' patients. The detailed findings from the qualitative online insight gathering approaches (SML and OBB studies) in COPD have been reported in their respective full publications. In short, the results of these studies indicated that besides dyspnoea and exacerbations, cough and mucus production are very important aspects of the disease for COPD patients. The findings from the SML study showed that cough and mucus production were the symptoms affecting quality of sleep, work, and daily activities irrespective of severity of disease. These findings also revealed that relief from cough, mucus production, and shortness of breath would be the most desirable aspects of disease management from a patient perspective.
The results from the OBB study conducted with patients suggested that the symptoms of cough and excess mucus production are particular issues for the patients early in the day, with the effort of mucus expulsion leaving them tired and exhausted for the day ahead. These cough and mucus symptoms also result in the need for continuous breaks during the daytime making most patients exhausted. The symptoms also manifest during the night (phlegm accumulates when lying down and introduces sense of suffocation, together with the cough causing sleep disturbance). Another important observation from the OBB study was that patients with COPD experience urinary incontinence associated with cough but tend to be uncomfortable talking openly about such sensitive symptoms (hence this was not a prominent complaint in social media posts); in fact, two-third of the study participants from OBB who suffered from chronic cough acknowledged incontinence to be a significant problem. To build further on the significance and relative importance of these symptoms to patients (cough and mucus secretion and consequences thereof such as sleep disturbance, fatigue, incontinence) compared with the more commonly measured end points (lung capacity or breathlessness; exacerbations/hospitalizations) in COPD, there was a need to evaluate them in a quantitative patient preference study.
## Web-based patient preference study
Patient preferences reflect patients' choices among different disease state alternatives or particular health interventions (drug or medical device) over other available options, based on. To make a choice or stated preference, patients need to weigh the advantages and disadvantages of each disease state or intervention and compare them with the alternatives. The methods to derive patient preferences focus on understanding the relative importance of attributes by patients and provide an alternative to characterize patient needs and desires. We are evaluating the findings from the SML and OBB studies in a quantitative patient preference study. This study is currently ongoing via a discrete-choice experiment (DCE) online survey in patients with COPD, which will further quantify the relative importance of different symptoms from the patients' perspective. While designing the preference study we approached a health technology assessment (HTA) agency to obtain input on study designa first of its kind 'scientific advice' from an HTA body. We sought input from NICE (The National Institute for Health and Care Excellence) on the COPD patient preference study design and incorporated many suggestions received through this early scientific advice into the study protocol. In particular, considering the recommendations by NICE, the study protocol was updated for eligibility criteria, quality of life questionnaire instrument selection, and total number of patients to be recruited for selected countries.
The patient preference study will use a DCE design, built on the principle of random utility theory, that assumes that disease states can be decomposed into various attributes (symptoms) and the attractiveness/unattractiveness of a disease state depends on patients' relative preferences for the attributes (symptoms) expressed by the frequency with which they choose their preferred disease profile. In our DCE survey, patients with COPD will be asked to choose between two disease states with varying levels of the aforementioned attributes, as expressed by their willingness to accept trade-offs among them. This 'disease state' preference survey is a fairly novel approach as typically the discrete choice is made between two (hypothetical) treatment options. We believe the choice among disease states is a more appropriate exercise for patients when investigating the relative importance of symptoms to them, as it avoids any confusion between attributes/symptoms being incorrectly interpreted as side effects of a product. The results of this ongoing preference study will provide estimates of the relative importance of different symptoms to patients, which in turn will provide evidence about the importance of incorporating these patient-relevant end points in clinical trials. Furthermore, the results will provide a basis for determining the relative value, from the patients' perspective, that future treatments would offer that address these same symptoms to a greater or lesser degree.
## Integration of collective insights to inform choice of pro end points
The integration of insights derived from ''online'' studies (SML, OBB, and the online patient preference study) into drug development for COPD offers an opportunity to truly listen to patients' voices and inform drug development strategy and product design as a step towards a more PFDD-based model. The insights generated from these studies will be of greatest value for taking into account patients' needs and preferences while defining clinical studies in COPD. Collectively, these patient insights and preferences will help assemble hypothetical treatment profiles with specific characteristics and also aid in selecting clinical outcome assessments beyond conventional end points in the COPD drug development program.
Having identified what is important for patients would not serve the purpose until these elements are appropriately integrated into a clinical trial setting in a timely fashion (e.g., identifying relevant end points, selecting an appropriate PRO instrument, and device 'digital' evidence capture strategy). Timely integration of the findings from the patient preference study into drug development is vital for improving (patient-relevant) outcomes in COPD patients. Traditionally, the phase III clinical trials evaluating efficacy and safety of drugs for COPD have been using clinical measures such as FEV1 and exacerbations. We considered the insights derived from qualitative and quantitative studies such as SML, OBB, to inform PRO end point selection. Based on the findings derived from the patient insight studies, it appears that cough and mucus production are additional symptoms where patients with COPD would value new therapeutic options. Therapeutic management of patients with COPD has evolved towards more tailored and individualized treatment from the past where a common therapeutic approach was used for all patients. The role of a more individualized approach for management of COPD has been an emerging strategy for personalized therapy in COPD. A careful evaluation of individual patient needs can lead to improved use of available treatment options and support precision medicine when there is concordance between genotype and phenotypic manifestations of the diseases . With advancement of the therapies that may address symptoms such as cough and mucus in the future, it will be an important step forward for the patient subgroup suffering from these symptoms to have new treatment options to complement the use of bronchodilators and corticosteroids. Using rigorously collected patient insights and the newly available treatment options, coupled with enhanced diagnostic techniques, may result in better outcomes for the individual patients and the overall population. Consequently, it will be appropriate to use a PRO assessment that evaluates the impact of cough and mucus production in the prospective clinical trial for COPD.
We therefore, intend to use the available Cough and Sputum Assessment Questionnaire (CASA-Q), a validated instrument to collect PROs, in the prospective drug development studies for COPD in addition to the use of the COPD assessment test (CAT) to assess the severity and impact of COPD. The preference study results will further illustrate the importance of the other consequences of cough and mucus-sleep disturbance, fatigue, and incontinence, for instance-to patients to justify their inclusion as standard PRO measures in prospective COPD clinical trials. We are considering inclusion of actigraphy as a digital end point in clinical trials to capture movement and sleep disturbances. During the phase II trial, a survey questionnaire on incontinence will be administered to patients to further evaluate the importance of this issue among patients. Once we have the final outputs of the preference study, we also intend to solicit regulatory and HTA scientific advice to inform the design of future clinical studies in COPD and the insights and preferences of patients will form an important input to that advice meeting.
Regulatory guidance recommends the use of a validated PRO instrument in support of a labelling claim, if such an instrument exists, to assess and measure the concepts of interest. In circumstances where patient insights reveal requirements that cannot be assessed using existing tools or otherwise require the development of a new instrument, then it is necessary to follow the formal approach of developing a new PRO instrument recommended by the regulatory agencies. Developing a PRO instrument is an iterative process. PRO instrument development requires qualitative and quantitative studies to identify items and domains of an instrument that are appropriate and comprehensive relative to the intended measurement concept, population, and use. We believe that the approach illustrated here has the potential to complement both qualitative (via SML and OBB studies) and quantitative (by patient preference studies) data collection requirements during the PRO instrument development process. Importantly, if aimed to use for PRO instrument development, such studies need to begin very early in the drug development lifecycle (at latest in phase I) so that the instruments can be validated in phase II to be ready by the start of the pivotal clinical trials for registration.
## Future directions
Regulatory agencies such as the Center for Devices and Radiological Health (CDRH) at the US Food and Drug Administration (FDA) advocate early engagement and frequent discussions regarding design and execution of patient preference studies. Guidance for the incorporation of patient preference measurements into drug development, regulatory, and reimbursement decision-making processes is evolvingand multi-stakeholder initiatives such as IMI PREFER (Innovative Medicines Initiative's The Patient Preferences in Benefit-Risk Assessments during the Drug Life Cycle project) are actively working to fill this gap. There is a need for early engagement in dialogue with regulatory agencies and HTA bodies on scientific advice to discuss patient needs, preferences, and implications thereof for clinical trial design and evidence collection. This step is crucial since, eventually, HTA agencies will be one of the recipients of any product submissions that will be developed through trials using the insights generated through such a PFDD exercise and would need to make reimbursement decisions after recognizing the value that such products offer to patients.
The draft FDA guidance advocates incorporating the patients' voice via collection of patient experience data and other relevant information to leverage the expertise that patients and caregivers can bring to the PFDD model. This FDA draft guidance details data collection methodologies for patient experience data (including social media to collect qualitative data), with an intention to inform clinical trial design and trial end point selection and to facilitate regulatory reviews. The guidance, however, mentions that ''the level of rigor needed to generate such data'' will differ across studies depending on the clinical condition and treatment under investigation and encourages early FDA consultation to obtain feedback. Although the current draft is undergoing stakeholder review and consultation, the availability of this guidance indicates that regulatory agencies encourage stakeholders to explore innovative tools to gain insights into patients' perspectives related to disease burden. The FDA held a workshop intended to develop more guidance documents on methods to identify what is important to patients and select, develop, or modify fit-for-purpose clinical outcome assessments. The final guidance when available from the FDA will clarify how to best integrate patient insights into the PFDD model.
# Conclusions
Gaining patient insights by collecting information on their disease experiences and perspectives should be an ongoing process throughout the product development lifecycle and starts in very early development phases. Conducting ''online'' studies such as the SML, OBB, and online patient preference studies is a relatively new way to generate patient insights. Integrating patient insight gathered via structured and systematic approaches including such ''online'' studies is feasible and an important input to decision making if considered early during drug development.
the authors and do not necessarily represent the views of Novartis.
Compliance with Ethics Guidelines. This article is based on previously conducted studies and does not report any new data with human participants or animals performed by any of the authors.
Data Availability. Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.
Open Access. This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/ by-nc/4.0/), which permits any noncommercial use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. |
RNA-Seq-Based Analysis Reveals Heterogeneity in Mature 16S rRNA 3′ Termini and Extended Anti-Shine-Dalgarno Motifs in Bacterial Species
We present an RNA-Seq based approach to map 39 end sequences of mature 16S rRNA (39 TAIL) in bacteria with single-base specificity. Our results show that 39 TAILs are heterogeneous among species; they contain the core CCUCC anti-Shine-Dalgarno motif, but vary in downstream lengths. Importantly, our findings rectify the mis-annotated 16S rRNAs in 11 out of 13 bacterial species studied herein (covering Cyanobacteria, Deinococcus-Thermus, Firmicutes, Proteobacteria, Tenericutes, and Spirochaetes). Furthermore, our results show that species-specific 39 TAIL boundaries are retained due to their high complementarity with preferred Shine-Dalgarno sequences, suggesting that 39 TAIL bases downstream of the canonical CCUCC motif play a more important role in translation initiation than previously reported.
aSD motif, CCUCC, is widely believed to elevate initation efficiency because of its strong complementarity with SD sequences and conservation across phyla [bib_ref] The 39-terminal sequence of Escherichia coli 16S ribosomal RNA: complementarity to nonsense..., Shine [/bib_ref] [bib_ref] Conservation of primary structure in 16S ribosomal RNA, Woese [/bib_ref] [bib_ref] Identification and characterization of E.coli ribosomal binding sites by free energy computation, Schurr [/bib_ref] [bib_ref] Predicting Shine-Dalgarno Sequence Locations Exposes Genome Annotation Errors, Starmer [/bib_ref] [bib_ref] Translation initiation region sequence preferences in Escherichia coli, Vimberg [/bib_ref] [bib_ref] Dynamic evolution of translation initiation mechanisms in prokaryotes, Nakagawa [/bib_ref]. Yet what constitutes ideal SD:aSD complementarity remains a subject of debate. Some researchers contend that there is weak association between SD:aSD binding affinity and initiation efficiency [bib_ref] The anti-Shine-Dalgarno sequence drives translational pausing and codon choice in bacteria, Li [/bib_ref] , but others suggest that intermediate binding affinities optimize initiation efficiency in Escherichia coli and Bacillus subtilis when a broader range of SD:aSD interactions is considered [bib_ref] Translation initiation region sequence preferences in Escherichia coli, Vimberg [/bib_ref] [bib_ref] Comparison of mRNA features affecting translation initiation and reinitiation, Osterman [/bib_ref] [bib_ref] Leveraging genome-wide datasets to quantify the functional role of the anti-Shine-Dalgarno sequence..., Hockenberry [/bib_ref]. Furthermore, when a SD sequence that binds to the B. subtilis 39 TAIL is substituted with a shorter SD sequence pairing with E. coli's 39 TAIL, interferon plasmids' expression levels decrease drastically [bib_ref] Bacillus subtilis Requires a "Stringent" Shine-Dalgarno Region for Gene Expression, Band [/bib_ref]. These findings emphasize the importance of characterizing the full extent of the 39 TAIL.
The 39 TAIL boundary remains ambiguous for most bacterial species because the precise 39 maturation process of the 16S precursor sequence remains unclear [bib_ref] Multiple exoribonucleases catalyze maturation of the 39 terminus of 16S ribosomal RNA..., Sulthana [/bib_ref] [bib_ref] Twenty years of bacterial RNases and RNA processing: how we've matured, Deutscher [/bib_ref] , and only a few mature 16S rRNA sequences have been experimentally verified [bib_ref] Secondary structure model for bacterial 16S ribosomal RNA: phylogenetic, enzymatic and chemical..., Woese [/bib_ref]. Consequently, determination of the 16S rRNA is frequently automated based on sequence similarity [bib_ref] Questionable 16S ribosomal RNA gene annotations are frequent in completed microbial genomes, Lin [/bib_ref] [bib_ref] Dynamic evolution of translation initiation mechanisms in prokaryotes, Nakagawa [/bib_ref]. However, this process is often unreliable [bib_ref] Predicting Shine-Dalgarno Sequence Locations Exposes Genome Annotation Errors, Starmer [/bib_ref] [bib_ref] Estimating the annotation error rate of curated GO database sequence annotations, Jones [/bib_ref] [bib_ref] RNAmmer: consistent and rapid annotation of ribosomal RNA genes, Lagesen [/bib_ref] [bib_ref] Questionable 16S ribosomal RNA gene annotations are frequent in completed microbial genomes, Lin [/bib_ref] and many such 16S ribosomal RNA sequence annotations have been discontinued in NCBI's Gene database. For example, 16S rRNA entries for Streptococcus pyogenes (NC_002737), Bacillus anthracis (NC_005945), and Legionella pneumophila (NC_005823) are all truncated such that their annotated 39 ends do not encompass the canonical CCUCC motif.
To circumvent the aforementioned problem, we devise strategies to map RNA transcripts from high-throughput RNA sequencing (RNA-Seq) data [bib_ref] Highly integrated single-base resolution maps of the epigenome in Arabidopsis, Lister [/bib_ref] [bib_ref] RNA-Seq: a revolutionary tool for transcriptomics, Wang [/bib_ref] [bib_ref] Count-based differential expression analysis of RNA sequencing data using R and Bioconductor, Anders [/bib_ref] to the 16S rDNA genomic sequence with single base specificity. The feasibility of this approach was shown recently in a study [bib_ref] Elucidating the 16S rRNA 39 boundaries and defining optimal SD/aSD pairing in..., Wei [/bib_ref] where we successfully recovered the E. coli and B. subtilis 39 TAILs documented in literature [bib_ref] The 39-terminal sequence of Escherichia coli 16S ribosomal RNA: complementarity to nonsense..., Shine [/bib_ref] [bib_ref] Conservation of primary structure in 16S ribosomal RNA, Woese [/bib_ref]. Our present objective is to advance our RNA-Seq framework to characterize the 39 TAIL in any bacterial species, especially those that have not been experimentally verified.
The challenge associated with our approach is the limited availability of suitable data. There is a complete lack of publicly available RNA-Seq data in GEO DataSets for many species, such as Acidithiobacillus ferrooxidans, Microcystis aeruginosa, Shigella flexneri, and Yersinia pestis. Furthermore, many experiments remove rRNAs prior to sequencing [bib_ref] Ribosomal RNA depletion for efficient use of RNA-seq capacity, O'neil [/bib_ref] in an effort to enrich the target RNA molecules, such as mRNAs [bib_ref] Purifying mRNAs with a highaffinity eIF4E mutant identifies the short 39 poly(A)..., Choi [/bib_ref]. Fortunately, our findings suggest that ribo-depletion is often incomplete, and enough 16S rRNA reads will persist to allow for 39 TAIL characterization. The inclusion of 13 species studied herein (covering Cyanobacteria, Deinococcus-Thermus, Firmicutes, Proteobacteria, Tenericutes, and Spirochaetes) is thus predicated on the availability of usable RNA-Seq datasets in NCBI's GEO [bib_ref] Gene Expression Omnibus: NCBI gene expression and hybridization array data repository, Edgar [/bib_ref] database (see Materials and Methods for additional details). Additionally, the availability of protein abundance data in PaxDb [bib_ref] PaxDb, a Database of Protein Abundance Averages Across All Three Domains of..., Wang [/bib_ref] [bib_ref] Version 4.0 of PaxDb: Protein abundance data, integrated across model organisms, tissues,..., Wang [/bib_ref] for all species studied allow us to investigate the effect of SD:aSD complementarity on protein production in real genes.
Comprehensive comparative sequence analyses [bib_ref] Dynamic evolution of translation initiation mechanisms in prokaryotes, Nakagawa [/bib_ref] [bib_ref] Comparative genomic analysis of translation initiation mechanisms for genes lacking the Shine-Dalgarno..., Nakagawa [/bib_ref] claim 59-CCUCCU-39 is the functionally constrained 39 TAIL terminus. In other words, the motif is conserved among bacterial species because it pairs with SD sequences effectively. However, several bases further downstream are conserved in the genomic sequences of closely related species. We suspect that this is the result of functional constraint imposed by the SD:aSD interaction further downstream of 59-CCUCCU-39. Accordingly, we hypothesize that downstream bases are retained in 39 TAILs because they effectively interact with speciesspecific SD sequences as previously observed for E. coli and B. subtilis [bib_ref] Bacillus subtilis Requires a "Stringent" Shine-Dalgarno Region for Gene Expression, Band [/bib_ref] [bib_ref] How Changes in Anti-SD Sequences Would Affect SD Sequences in Escherichia coli..., Abolbaghaei [/bib_ref] [bib_ref] Elucidating the 16S rRNA 39 boundaries and defining optimal SD/aSD pairing in..., Wei [/bib_ref].
Our findings corroborate previous studies suggesting that intermediate binding affinity is preferred [bib_ref] Comparison of mRNA features affecting translation initiation and reinitiation, Osterman [/bib_ref] [bib_ref] Leveraging genome-wide datasets to quantify the functional role of the anti-Shine-Dalgarno sequence..., Hockenberry [/bib_ref] [bib_ref] Elucidating the 16S rRNA 39 boundaries and defining optimal SD/aSD pairing in..., Wei [/bib_ref]. The 39 termini downstream of the core CCUCC are heterogeneous among species, but fall within the conserved boundary at the genomic level. Furthermore, terminal bases are preferred in SD:aSD binding in most species, albeit having weaker binding affinity than CCUCC. These findings demonstrate the importance of considering bases downstream of CCUCC in SD:aSD binding.
# Materials and methods
Processing genomic and RNA-Seq data The annotated genomes of 26 species in GenBank formats were retrieved from the National Center for Biotechnology Information (NCBI) database (http://www.ncbi.nlm.nih.gov). Next, the NCBI annotated 16S rRNA was retrieved. In the case where multiple 16S rRNA entries exist, the first one listed is selected.
High-throughput RNA-Seq SRX runs of wildtype species were downloaded from GEO DataSets in FASTQ format. The FASTQ files were first converted to FASTQ+ format using ARSDA 1.1, grouping identical reads under a single ID while also indicating the number of copies (SeqID_# of copies), in order to reduce the size of the datasets prior to adapter trimming. The FASTQ+ data were then processed using CutAdapt 1.17 [bib_ref] Cutadapt removes adapter sequences from highthroughput sequencing reads, Martin [/bib_ref] to trim off the 39 flanking adapter sequences. In experiments that use the oligo(dT)-adapter primer, RNA fragments are first poly-adenylated at the 39 end, we thus set CutAdapt to recognize "AAAAA". In others that use specific sets of primers ligated to random hexamers, we set CutAdapt to recognize all possible adapters in the kits' index, with 10% error rate. Regardless of whether poly-As or barcode adapters were trimmed, we only retained reads that were 25 nt or longer after the trimming process to mitigate bias in expression levels [bib_ref] Trimming of sequence reads alters RNA-Seq gene expression estimates, Williams [/bib_ref]. Next, we used Trimmomatic 0.38 [bib_ref] Trimmomatic: a flexible trimmer for Illumina sequence data, Bolger [/bib_ref] to remove poor quality sequences with average Phred scores lower than 20 (1% probability of a base calling error) [bib_ref] Base-calling of automated sequencer traces using phred. II. Error probabilities, Ewing [/bib_ref]. Since adapters were trimmed after reads were grouped in FASTQ+ format, sequences that were previously unique due to the presence of adapter nucleotides may become identical (such as for SeqGr176560_1 and SeqGr558077_1, [fig_ref] Figure 1: The count of mapped 39 ends of RNA-Seq reads [/fig_ref]. The processed FASTQ+ datasets were subsequently converted into FASTA format for multiple sequence alignment.
Aligning RNA-Seq reads to annotated rRNA sequences We next mapped reads in the FASTA files onto the 16S rDNA genomic sequence. The FASTA+ files were converted into BLAST databases using the "Create BLAST DB" function in ARSDA. The BLAST query sequence was selected using genomic sequences 100 nt upstream and downstream of the core CCUCC motif (205 nt total query length). For each species, the query sequence was searched against BLAST databases using the BLAST function [bib_ref] Basic local alignment search tool, Altschul [/bib_ref] implemented in ARSDA. We used an E-value cutoff of 10 25 (with the exception of Bacillus anthracis, for which we used an E-value of 10 23 due to the relatively shorter average read length and smaller database size) paired with a minimum word length of 12 to balance the quantity and quality of hits, as well as search speed, against the .= 25 nt reads in the ribodepleted datasets. Then, sequence hits were retrieved from the FASTA files using seqtk [bib_ref] The anti-Shine-Dalgarno sequence drives translational pausing and codon choice in bacteria, Li [/bib_ref] and complementary strand sequences were eliminated. Finally, remaining hits were aligned to the query sequence using multiple sequence alignment (Clustal Omega algorithm [bib_ref] Clustal Omega, accurate alignment of very large numbers of sequences, Sievers [/bib_ref] implemented in DAMBE, default parameters).
Determining putative SD sequences based on pairing potential, location, and binding affinity For each species, our characterized 39 TAILs [fig_ref] Table 1: The RNA-Seq corrected 39 TAIL in 13 bacterial species [/fig_ref] were used as the complementary sequence in identifying putative SD sequences. To ensure that determined putative SD sequences are from real genes, we map protein IDs in PaxDb 4.0 [bib_ref] PaxDb, a Database of Protein Abundance Averages Across All Three Domains of..., Wang [/bib_ref] [bib_ref] Version 4.0 of PaxDb: Protein abundance data, integrated across model organisms, tissues,..., Wang [/bib_ref] to Gene IDs in NCBI and only use CDSs that have protein expressions. Using DAMBE7 (Xia 2018b), we followed the methods used in previous studies [bib_ref] Algorithms for Loop Matchings, Nussinov [/bib_ref] [bib_ref] RNA secondary structure: a complete mathematical analysis, Waterman [/bib_ref] : 30 nt upstream of start codon of all CDSs were extracted and matched against the annotated 39 TAIL with 'Analyzing 59UTR' in DAMBE, with minimum SD length =4 nt and maximum SD length = 12 nt. Site-specific observed and expected aSD usage values were retrieved from the DAMBE when SD sequences are determined.
## Data availability
Supplementary file S1 contains RNA-Seq BLAST hits and file S2 contains the list of genes with protein abundance data that were used to determine putative SD sequences in all species studied; [fig_ref] Figure 1: The count of mapped 39 ends of RNA-Seq reads [/fig_ref] contain the 39 TAIL map for the remaining 11 species. Supplemental material available at Figshare: https://doi.org/10.25387/g3.7081094.
# Results and discussion
Characterizing the 39 TAIL in bacteria using an improved RNA-Seq-based approach We improve upon our method of 39 TAIL characterization [bib_ref] Elucidating the 16S rRNA 39 boundaries and defining optimal SD/aSD pairing in..., Wei [/bib_ref] by processing the RNA-Seq data more rigorously. To ensure quality and single-base specificity for reads mapped to a reference genomic sequence, we used CutAdapt [bib_ref] Cutadapt removes adapter sequences from highthroughput sequencing reads, Martin [/bib_ref] to trim adapters flanking raw RNA-Seq reads because these sequences obscure the true end of RNA fragments (see Materials and Methods for more detail). We subsequently filtered out poor quality reads by discarding those with average Phred scores # 20 using Trimmomatic [bib_ref] Trimmomatic: a flexible trimmer for Illumina sequence data, Bolger [/bib_ref] ; in other words, we retained reads with average base-calling error rates of , 1% [bib_ref] Base-calling of automated sequencer traces using phred. II. Error probabilities, Ewing [/bib_ref]. A caveat of using poly-adenylated RNA-Seq datasets for Neisseria meningitidis is that we cannot distinguish between 59-CCUCCUUUCU-39 and 59-CCUCCUUUCUA-39 as the 39 TAIL; it is unclear whether the first adenosine is associated with the 39 TAIL or the poly-A chain [fig_ref] Table 1: The RNA-Seq corrected 39 TAIL in 13 bacterial species [/fig_ref].
To map the 16S rRNA, we generated a BLAST library using the quality filtered datasets and performed ungapped local similarity search using BLAST [bib_ref] Basic local alignment search tool, Altschul [/bib_ref] between RNA-Seq reads and a 205 nt genomic sequence with the canonical CCUCC motif at the center (100 nt extending from each side). We next aligned the BLAST hits by multiple sequence alignment (Clustal Omega algorithm [bib_ref] Clustal Omega, accurate alignment of very large numbers of sequences, Sievers [/bib_ref] implemented in DAMBE (Xia 2018b), default parameters) against the reference genomic sequence. In all species, we define the terminus of the 39 TAIL using two criteria: 1) it must contain the canonical CCUCC, and 2) it is the most mapped site at or near CCUCC. The underlying assumption for the second criterion is that the mature 16S rRNA is more abundant than precursor transcripts, as is the case in E. coli [bib_ref] Depletion of pre-16S rRNA in starved Escherichia coli cells, Cangelosi [/bib_ref] , because precursors are continuously degraded by exoribonucleases [bib_ref] Multiple exoribonucleases catalyze maturation of the 39 terminus of 16S ribosomal RNA..., Sulthana [/bib_ref].
The 39 TAIL termini are heterogeneous but functionally constrained Following our two criteria, we have characterized the 39 TAIL in 13 out of 26 species in PaxDb [fig_ref] Table 1: The RNA-Seq corrected 39 TAIL in 13 bacterial species [/fig_ref]. [fig_ref] Figure 1: The count of mapped 39 ends of RNA-Seq reads [/fig_ref] shows the sequence map and alignments for Lactococcus lactis and Deinococcus deserti. The sequences mapped for the 11 others are present in Supplementary [fig_ref] Figure 1: The count of mapped 39 ends of RNA-Seq reads [/fig_ref]. Two others, E. coli and B. subtilis, were previously determined [bib_ref] Elucidating the 16S rRNA 39 boundaries and defining optimal SD/aSD pairing in..., Wei [/bib_ref]. The remaining 11 species could not be characterized because of the aforementioned absence of data in four species (Acidithiobacillus ferrooxidans, Microcystis aeruginosa, Shigella flexneri, and Yersinia pestis), and because no convincing peaks were observed in the region of interest (up to 30 nt downstream of CCUCC) in the remaining seven species (Bacterioides thetaiotaomicron, Bateonella henselae, Helicobacter pylori, Mycobacterium tuberculosis, Pseudomonas aeruginosa, Staphylococcus aureus, and Shewanella oneidensis), likely due to effective ribo-depletion in their RNA-Seq datasets. We considered a peak to be convincing when the counts mapping to that site were at least 3 fold higher than background (counts of any four flanking sites on either side). Importantly, in the characterized 13 species, we made corrections to annotations in eight species (NC_002662 L. lactis, NC_002163 Clamylobacter jejuni, NC_000911 Synechocystis sp., NC_003112 N. meningitidis, NC_012526 D. deserti, NC_017504 Mycoplasma pneumoniae, NC_003198 Salmonella enterica, NC_ 002937 Desulfovibrio vulgaris), and redefined the 39 TAIL in three others (NC_002737 S. pyogenes, NC_005945 B. anthracis, and NC_ 002942 L. pneumophila) that were certainly mis-annotated due to their failure to incorporate the canonical CCUCC. Resultantly, the annotated 39 TAILs of only two out of 13 species, NC_003210 Listeria n monocytogenes and NC_005823 Leptospira interrogans were left unchanged. In short, the 39 TAILs can variably extend up to six bases downstream of CCUCC in the majority of species studied. The 39 TAILs vary among species, but bases downstream of the CCUCC motif are conserved among bacteria. The 16S genomic sequences are largely conserved for several bases beyond CCUCC: e.g., 59-GAUCACCUCCUUUCUA-39 in Bacilli and 59-GAUCACCUC-CUUA-39 in Beta-and Gammaproteobacteria. This conservation suggests that the regions downstream of CCUCC may also be important in SD:aSD pairing. Importantly, 39 TAIL terminal bases downstream of CCUCC are species-specific, but do not extend past the conserved genomic boundaries, in all species studied except in L. interrogans. In other words, the 39 TAIL falls variably short of 59-GAUCACCUC-CUUUCUA-39. This finding further suggests that both CCUCC and downstream bases are conserved regions that are preferred in SD:aSD pairing in most species.
To offer a plausible reason for the unexpected length of the 39 TAIL in L. interrogans, it is worth noting that the dependence on the SD:aSD interaction for efficient translation is dynamic [bib_ref] Comparative genomic analysis of translation initiation mechanisms for genes lacking the Shine-Dalgarno..., Nakagawa [/bib_ref]. In genes that have strong secondary structure within the TIR, ribosome recruitment is facilitated by RPS1 [bib_ref] Dynamic evolution of translation initiation mechanisms in prokaryotes, Nakagawa [/bib_ref] [bib_ref] Comparison of mRNA features affecting translation initiation and reinitiation, Osterman [/bib_ref]. This protein binds U-rich regions [bib_ref] Ribosome-messenger recognition: mRNA target sites for ribosomal protein S1, Boni [/bib_ref] [bib_ref] AU-Rich Sequences within 59 Untranslated Leaders Enhance Translation and Stabilize mRNA in..., Komarova [/bib_ref] to unfold double-stranded RNA [bib_ref] Ribosomal protein S1 unwinds double-stranded RNA in multiple steps, Qu [/bib_ref] [bib_ref] Escherichia coli Ribosomal Protein S1 Unfolds Structured mRNAs Onto the Ribosome for..., Duval [/bib_ref]. Furthermore, RPS1's domains appear to be under higher functional constraint in species possessing few SD-containing genes, such as L. interrogans [bib_ref] Dynamic evolution of translation initiation mechanisms in prokaryotes, Nakagawa [/bib_ref] [bib_ref] Comparative genomic analysis of translation initiation mechanisms for genes lacking the Shine-Dalgarno..., Nakagawa [/bib_ref] ; the reliance on RPS1 reduces the dependence on a SD:aSD interaction and may relax 39 TAIL boundary constraints.
Notably, four species (D. deserti, M. pneumoniae, L. pneumophila, and L. interrogans) have a secondary peak of mapped reads within 20 nt downstream of CCUCC [fig_ref] Figure 1: The count of mapped 39 ends of RNA-Seq reads [/fig_ref] , [fig_ref] Table 1: The RNA-Seq corrected 39 TAIL in 13 bacterial species [/fig_ref] , [fig_ref] Figure 1: The count of mapped 39 ends of RNA-Seq reads [/fig_ref]. We propose that the secondary peak farther downstream is the pre-16S rRNA; it is too far downstream of CCUCC to be considered as the mature 16S rRNA 39 end based on sequence conservation [bib_ref] Dynamic evolution of translation initiation mechanisms in prokaryotes, Nakagawa [/bib_ref]. The prominence of this second peak may be due to the accumulation of the endoribonuclease cleaved pre-16S rRNA intermediate, because the localization of exoribonuclease to this precursor sequence is a rate limiting step. However, the intermediate sequence is rapidly continuously degraded once it is targeted by these enzymes. This would explain the lack of sequences mapped between the mature 16S rRNA and the intermediate sequence (the two peaks) [fig_ref] Figure 1: The count of mapped 39 ends of RNA-Seq reads [/fig_ref] , [fig_ref] Figure 1: The count of mapped 39 ends of RNA-Seq reads [/fig_ref].
The 39 TAIL terminal bases are preferred in SD: aSD binding We define an aSD site to be preferred if the observed number of times the base is involved in SD pairing is greater than expected. In the absence of SD usage bias, a putative SD sequence of 4 to 12 nt can be expected to pair anywhere within the boundary of the aSD sequence, as long as complete complementarity is achieved. Here, we designate the aSD sequence to constitute the 39 TAIL, beginning with the conserved 59-GAUCA-39, followed by the core motif CCUCC, and extended by variable lengths of terminal bases characterized herein [fig_ref] Table 1: The RNA-Seq corrected 39 TAIL in 13 bacterial species [/fig_ref] , e.g., in L. monocytogenes, the aSD sequence is 59GAUCACCUCCUUUCU-39 and the terminal bases are 59-UUUCU-39). Then, taking L. monocytogenes as example, the maximum number of possible pairs at the first complementary aSD site (aSD_1) by the total pool of 4 nt to 12 nt putative SD sequences is calculated by equation (1), with N m denoting N number of putative SD sequences of length m:
[formula] aSD 1 ¼ X 12 m¼4 N m 15 2 m þ 1(1) [/formula]
However, the number of possible base-pairs resulting in perfect complementarity varies. For example, a 12 nt putative SD sequence may start pairing at the first, but not the sixth, base on a complementary aSD sequence that is 15 nt long, and the maximum usage of the sixth aSD site (aSD_6) is calculated instead by equation (2):
[formula] aSD 6 ¼ 4 · N 4 12 þ 5 · N 5 11 þ 6 · N 6 10 þ 6 · N 7 9 þ 6 · N 8 8 þ 6 · N 9 7 þ 6 · N 10 6 þ 6 · N 11 5 þ 5 · N 12 4 (2) [/formula]
The expected usage is then calculated by taking the relative proportions of maximum usage at each site (adding up to 1) multiplied by the total number of observed putative SD sequences of various lengths. These calculations are implemented in DAMBEunder the 'Analyze 5UTR' function. As defined, a preferred aSD site will have an observed to expected SD: aSD usage ratio (O:E) . 1. Since expected SD:aSD count is calculated in absence of any selection bias in SD usage, an O:E . 1 suggests presence of selection bias in observed SD usage. [fig_ref] Figure 2: Mean ratio of observed over expected SD [/fig_ref] shows an average O:E . 1 at CCUCC for all species, with the exception of L. interrogans. Indeed, CCUCC is preferentially used in SD:aSD pairing. Meanwhile, average O: E is ,1 for 59-GAUCA-39 in all species except M. pneumoniae; hence, this conserved region is avoided by SD:aSD pairing in most species. Lastly, in keeping with expectations, conserved regions downstream of CCUCC have an O:E . 1 in all species except L. monocytogenes. These observations indicate that downstream bases are retained because they are preferred in SD:aSD binding, despite their weaker binding affinity than CCUCC. This result corroborates recent studies suggesting that intermediate SD:aSD complementarity increase initiation efficiency [bib_ref] Comparison of mRNA features affecting translation initiation and reinitiation, Osterman [/bib_ref] [bib_ref] Leveraging genome-wide datasets to quantify the functional role of the anti-Shine-Dalgarno sequence..., Hockenberry [/bib_ref] [bib_ref] Elucidating the 16S rRNA 39 boundaries and defining optimal SD/aSD pairing in..., Wei [/bib_ref].
In this study we present an RNA-Seq based approach to characterize the 39 end of mature 16S rRNA in bacterial species across different lineages. There is weaker 39 TAIL conservation at the RNA level than at the DNA level. Nonetheless, the presence of 39 termini bases downstream of CCUCC falls within the conserved boundary at the genomic level. Furthermore, the usage of terminal bases is favored in SD:aSD binding. Alternatively, RPS1-mediated initiation may relax the functional constraint at the 39 TAIL of L. interrogans, explaining its exceptional length. Our findings complement previous studies investigating the role of CCUCC in translation initiation and suggest that transcribed bases downstream of this canonical motif also play an important role in translation efficiency.
# Acknowledgments
This work was supported by the Discovery Grant of Natural Science and Engineering Research Council of Canada to X.X. , and the Ontario Graduate Scholarship 2018-2019 to Y.W. The manuscript was substantially improved by the comments of two anonymous reviewers, and we are grateful for their insight.
## Literature cited
[fig] Figure 1: The count of mapped 39 ends of RNA-Seq reads (A, C) and sequence alignments (B, D) for Lactococcus lactis and Deinococcus deserti. Mapped regions start with the last C of CCUCC as the first site, extended by 30 nt downstream. The 39 ends of sequence alignments represent local reads mapped to the single major peak in L. lactis and the two major peaks in D. deserti. The complete length of the query genomic sequence is 205 nt long. [/fig]
[fig] Figure 2: Mean ratio of observed over expected SD:aSD complementarity (O:E ratio) in 13 species at conserved 59-GAUCA-39 (blue), and 59-CCUCC-39 motifs (red). The average O:E ratio is also shown for the characterized sequences downstream of CCUCC (green) in the nine bacterial species that have extended ends. [/fig]
[table] Table 1: The RNA-Seq corrected 39 TAIL in 13 bacterial species. RNA-Seq determined 39 TAILs are shaded gray. The NCBI annotated 39 TAILs are in black fonts, extensions revealed by RNA-Seq data are underlined and ambiguities in bold Species [/table]
|
A Systematic Review on the Association Between Tear Film Metrics and Higher Order Aberrations in Dry Eye Disease and Treatment
We systematically reviewed published research on dry eye disease and its association with higher order aberrations (HOAs). The purpose of this review was to first determine if an association between tear film metrics and HOAs existsKey Summary PointsThe rationale for this review was to determine the association between tear film metrics and higher order aberrations (HOAs), and to also determine if treatment of dry eyes improves tear film metrics and HOAs.The literature was systematically reviewed, in that the total of 60 search hits with our key terms were then curated by the authors on the basis of original research and a focus on tear film metrics and HOA parameters, leading to a total of 26 relevant articles for this review.This review has shown clear associations and correlations between dry eyes and HOA parameters. Improvements in HOAs with dry eye interventions provide further evidence to support the intricate relationship between the two.However, despite the clear association between HOAs and dry eye disease, further research is still required in the realm of clinical application as many factors such as patient severity and eye drop viscosity can impact clinical outcomes.
and second to determine if the treatment of dry eyes can improve tear film metrics and HOAs together. A search was conducted in Entrez PubMed on 25 April 2021 using the keywords ''higher order aberrations'' and ''dry eye''. The initial search yielded 61 articles. After publications were restricted to only original articles measuring HOA outcomes in patients with dry eye, the final yield was 27 relevant articles. Of these 27 papers, 12 directly looked at associations and correlations between dry eyes and HOA parameters. The remaining 15 studies looked at dry eye interventions and HOA outcomes and parameters. There is clear evidence demonstrating that dry eyes and HOAs have an association, and that the tear film is one of the most important factors in this relationship. There is also a direct correlation between tear film metrics and HOAs. Improvements in HOAs with dry eye interventions provide further evidence to support the intricate relationship between the two. Despite the clear association between HOAs and dry eye disease, further research is still required in the realm of clinical application as dry eye interventions vary depending on many factors, including patient severity and eye drop viscosity.
Keywords: Dry eye disease; Dry eye treatment; Higher order aberrations; Tear film
# Introduction
The formation of optical images on the retina marks the beginning stages of vision processing [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref]. However, there are many possible avenues of error either through diffraction, errors of focus, and wavefront aberrations [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref] [bib_ref] A review of higher order aberrations of the human eye, Suliman [/bib_ref]. When optical systems are in ''perfect'' settings, rays emanating from an object are refracted to converge to the unique image point as expected from Gaussian theory [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref]. However, when the system suffers from wavefront aberrations, the refracted rays no longer converge to the expected unique image point [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref]. Therefore, the image is defocused or can be distorted. In particular, higher order aberrations (HOAs) have been of growing interest. Although most HOAs do not cause any noticeable degradation in visual acuity below 20/20 vision, laser and other refractive surgeries (including cataract-related ones) have left patients with symptomatic complaints of dry eye related to HOAs [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref] [bib_ref] Effects of 3% diquafosol sodium ophthalmic solution on higher-order aberrations in patients..., Inoue [/bib_ref].
Aberrometers are able to measure wavefront aberrations, i.e., the distance between an actual wavefront and the ideal spherical or ''reference'' wavefront on the ideal image point [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref]. One of the most common designs is the Hartmann-Shack aberrometer which shoots a narrow beam of light into the eye to produce a reflection [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref]. If the eye is free of aberrations, this ideally would result in a parallel ray leaving the eye and the existing wavefronts strike a Hartmann-Shack wavefront sensor comprising an array of identical converging micro lenses [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref]. The overall ideal result is image points reflecting directly in a regular grid pattern. However, when aberrations are present, these image points are displaced by an amount and direction depending on the aberration, resulting in an irregular grid pattern [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref].
An important factor that affects HOAs includes the tear film [bib_ref] A review of higher order aberrations of the human eye, Suliman [/bib_ref] [bib_ref] Wavefront analysis of higher order aberrations in dry eye patients, Montés-Micó [/bib_ref] [bib_ref] Irregular astigmatism and higher-order aberrations in eyes with dry eye disease, Koh [/bib_ref]. The tear film is the first refracting surface that light comes into contact with and thus has a crucial role in maintaining a smooth corneal surface [bib_ref] A review of higher order aberrations of the human eye, Suliman [/bib_ref] [bib_ref] Wavefront analysis of higher order aberrations in dry eye patients, Montés-Micó [/bib_ref] [bib_ref] Irregular astigmatism and higher-order aberrations in eyes with dry eye disease, Koh [/bib_ref]. Changes to the tear film thickness and regularity can introduce additional aberrations [bib_ref] Wavefront analysis of higher order aberrations in dry eye patients, Montés-Micó [/bib_ref]. As the eyes remain open, the tear film eventually breaks up and this also causes irregularity and increased aberrations [bib_ref] A review of higher order aberrations of the human eye, Suliman [/bib_ref] [bib_ref] Wavefront analysis of higher order aberrations in dry eye patients, Montés-Micó [/bib_ref] [bib_ref] Irregular astigmatism and higher-order aberrations in eyes with dry eye disease, Koh [/bib_ref]. Dry eye can be defined as a tear deficiency disorder of the ocular surface [bib_ref] Impaired functional visual acuity of dry eye patients, Goto [/bib_ref]. Several studies have demonstrated that disruptions of tear film caused by dry eye increase HOAs compared with normal eyes which can result in impaired visual function and optical quality [bib_ref] Wavefront analysis of higher order aberrations in dry eye patients, Montés-Micó [/bib_ref] [bib_ref] Impaired functional visual acuity of dry eye patients, Goto [/bib_ref] [bib_ref] Corneal fluorescein staining correlates with visual function in dry eye patients, Kaido [/bib_ref] [bib_ref] Postblink changes in total and corneal ocular aberrations, Montés-Micó [/bib_ref].
One way to address the symptoms and reported impaired optical quality of dry eyes is to use lubricating eye drops that have been shown to improve optical quality and significantly decrease aberrations associated with irregular tear films and dry eyes [bib_ref] A review of higher order aberrations of the human eye, Suliman [/bib_ref] [bib_ref] Changes of corneal wavefront aberrations in dry eye patients after treatment with..., Lu [/bib_ref] [bib_ref] Effect of instillation of eyedrops for dry eye on optical quality, Koh [/bib_ref] [bib_ref] Effect of lubricating eyedrops on ocular light scattering as a measure of..., Diaz-Valle [/bib_ref]. Another strategy or intervention for dry eyes is punctal plugs to block the puncta allowing for increased tear fluid accumulation, thus keeping the tear film intact [bib_ref] Punctal plug: a medical device to treat dry eye syndrome and for..., Yellepeddi [/bib_ref].
Dry eyes are one of the most common ocular disorders and a complication of many common laser and refractive surgeries, including cataract surgeries [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref] [bib_ref] A review of higher order aberrations of the human eye, Suliman [/bib_ref] [bib_ref] Punctal plug: a medical device to treat dry eye syndrome and for..., Yellepeddi [/bib_ref]. Therefore, it is important to understand any potential association or pathophysiology of HOAs and dry eyes, as well as whether interventions that improve dry eyes reduce HOA severity.
# Methods
To determine the primary literature available focussing on HOAs and dry eye association as well as dry eye interventions and HOA outcomes, the following search strategy was used in Entrez PubMed on 25 April 2021. The key terms used were ''higher order aberrations'' and ''dry eyes'' with an ''AND'' between each keyword for a keyword search string of ''higher order aberrations AND dry eyes''. No filters were set. This resulted in a total search hit of 61 articles. [fig_ref] Figure 1: PRISMA [/fig_ref] illustrates the screening process in a flowchart diagram. The articles were curated to determine whether they were original research articles; this excluded 13 articles as they were review articles. The second curation was based on subject relevance-if a paper did not look into dry eyes and HOAs as an outcome parameter, it would not be included. The second curation excluded 21 papers leaving a total of 27 relevant articles for this review [fig_ref] Figure 1: PRISMA [/fig_ref].
This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.
# Results
The search criteria and curation process yielded 27 relevant papers that looked at the primary outcome of HOAs in patients with dry eye. Of these 27 papers, 12 directly looked at any association and correlation of dry eyes and HOAs and HOA parameters. The remaining 15 studies looked at dry eye interventions and HOA outcomes and parameters. The main results of each paper are summarized in [fig_ref] Table 1: Summary of HOAs and its association with dry eyes [/fig_ref].
## Hoas and dry eye association
The first paper retrospectively reviewed patients with chronic ocular graft-versus-host disease (GVHD) after hematopoietic stem cell transplantation (HSCT) with dry eye disease . Dry eye is the main and most common phenotype for GVHD in the eye. Although chronic ocular GVHD does not typically lead to visual loss and recipients generally have good visual acuity, the vision-related quality of life is impaired. The total and anterior HOAs at 4-mm and 6-mm diameters in patients with GVHD were significantly higher compared to both non-GVHD and healthy controls. Total and anterior spherical aberrations (SA) at both 4-mm and 6-mm diameters were significantly lower in patients with GVHD compared to healthy controls. Total and anterior coma aberrations at 4-mm diameters were significantly higher in patients with GVHD compared to patients without GVHD while total coma aberrations at 6-mm were significantly higher in GVHD compared to healthy controls. When the HOAs in the same patients were analyzed and compared to their contralateral eyes, patients with GVHD had significantly higher HOAs compared to patients without GVHD. A significantly positive correlation was found between HOAs and visual acuity (logMAR) as well. Although dry eye is the most common phenotype for GVHD in the eye, the increase in HOAs revealed could also be attributed to the inflammation on the ocular surface as opposed to a direct effect of dry eyes.
One group prospectively assessed HOAs in patients with normal and dry eyes [bib_ref] Wavefront analysis of higher order aberrations in dry eye patients, Montés-Micó [/bib_ref]. The group was able to demonstrate that HOAs were significantly higher in patients with dry eye for both 4-mm and 6-mm diameters. This included coma, spherical, and total aberrations for both 4-and 6-mm diameters. Their quantitative analysis of the Hartmann-Shack images showed a significant change in the wavefront patterns in dry eyes compared to normal eyes. Specifically, these corresponded to reduced optical path length, which indicated relative thinning There was no significant difference in posterior HOAs between groups
There was no significant difference in SA between patients with GVHD and patients without GVHD. There was also no significant difference in posterior SA The correlation between posterior HOAs and ICO severity was significant but not as strong as total or anterior correlations
## Level 3 serial changes in hoas
Total HOAs, FI, and SI Total HOAs and coma aberrations were significantly higher from 5 to 9 s after a blink compared to those immediately after a blink Spherical aberrations were significantly higher from 6 to 9 s after a blink compared to immediately after
The FI and SI in the short TBUT group were significantly higher than those in the normal-stable group
There was no significant difference in FI and SI between the short TBUT group and the normal-sawtooth group Liposome spray had no effect on total aberrations for each eye type. Combination had no effect for each eye type There were no significant changes in spherical aberrations for each intervention type (drops, spray, and combination) for each eye type There were no significant changes in coma for each intervention type (drops, spray, and combination) for each eye type HOAs were higher at baseline in the dry eye group The increase in total aberrations with Hypromellose drops was insignificant compared to baseline LG staining was significantly lower in the DQS group than in the SH group but both groups showed significant improvement compared to baseline after 12 weeks SH treatment significantly lowered HOAs compared to DQS treatment after 12 weeks
There was no significant difference between the two groups for trefoil, coma, tetrafoil, and second astigmatism
Both groups significantly improved UDVA by the 1-week mark postsurgery and sustained this improvement
There was no significant difference in anterior chamber inflammation pre-and postoperatively and at any time points There was no significant difference between the treatment groups and at the 1-week postsurgery there was an increased OSDI score that recovered towards baseline by 4 weeks TBUT at postoperative week 12 was significantly longer than at baseline for DQS treatment whereas SH treatment returned to baseline by 12 weeks There was no significant difference between pre-and postoperative measurements in the DQS group but the SH group never returned to baseline after 12 weeks There was no significant difference in pre-and postoperative HOAs for both groups There was no significant difference between the two groups though inflammation reached its peak 1-week postsurgery and declined to baseline by 12 weeks for both groups of the tear film compared to areas on the lateral sides of the cornea (left and right) and increased thickening of the tear film at the inferior portion of the cornea. Although differences in vertical and horizontal coma aberrations are expected as a result of gravitational effects on the tear layer and recovery from the effects of lid pressure after blinking, these expected differences were significantly higher in dry eyes compared to normal eyes. These changes in tear film in dry eyes may be responsible for the aberrations and irregularities of the corneal surface.
Another group prospectively assessed serial measurements of HOAs after blinking in patients with dry eyes with or without superficial punctate keratopathy (SPK) [bib_ref] Serial measurements of higher-order aberrations after blinking in patients with dry eye, Koh [/bib_ref]. SPK occurs in patients with chronic dry eye disease in which small groups of cells are damaged as a result of inadequate tear production [bib_ref] Superficial punctate keratitis grading for dry eye screening using deep convolutional neural..., Su [/bib_ref]. The total HOAs in the group with SPK were significantly higher than the group without SPK. Similarly, both coma-and spherical-like aberrations were significantly higher in the group with SPK compared to the group without SPK [bib_ref] Serial measurements of higher-order aberrations after blinking in patients with dry eye, Koh [/bib_ref]. In the group without SPK, spherical aberrations were significantly higher at 7 and 9 s compared to 2 s after the blink. In the group with SPK, spherical aberrations were also higher at 9 s compared to at 2 and 4 s after the blink [bib_ref] Serial measurements of higher-order aberrations after blinking in patients with dry eye, Koh [/bib_ref]. The increases in spherical-like aberrations over time between blinks found in both groups were also observed in normal eyes in a previous study by the same group [bib_ref] Serial measurements of higher-order aberrations after blinking in normal subjects, Koh [/bib_ref]. These results suggest that HOAs in dry eye with SPK in the central cornea are higher than those in dry eye without SPK [bib_ref] Serial measurements of higher-order aberrations after blinking in patients with dry eye, Koh [/bib_ref].
As mentioned before, laser and refractive surgeries have been shown to cause dry eyes and an increase in HOAs [bib_ref] Wavefront technology: past, present and future, Charman [/bib_ref] [bib_ref] Effects of 3% diquafosol sodium ophthalmic solution on higher-order aberrations in patients..., Inoue [/bib_ref] [bib_ref] Higher order aberrations of the corneal surface after laser subepithelial keratomileusis, Jung [/bib_ref]. One group prospectively assessed HOAs in patients who underwent laser subepithelial keratomileusis (LASEK) [bib_ref] Higher order aberrations of the corneal surface after laser subepithelial keratomileusis, Jung [/bib_ref] with and without dry eye. At baseline, the dry eye group had significantly lower tear breakup time (TBUT), Schirmer I test values, and tear clearance rate compared to the non-dry eye group [bib_ref] Higher order aberrations of the corneal surface after laser subepithelial keratomileusis, Jung [/bib_ref]. The ocular surface staining and ocular surface disease index (OSDI) were significantly higher in the dry eye group compared to the non-dry eye group at baseline. Postoperatively, at 10 s after blinking, the dry eye group had significantly higher total HOA, coma, and trefoil aberrations compared to the non-dry eye group. The total HOA, coma, and trefoil aberrations were also significantly higher 10 s after a blink compared to immediately after in the dry eye group [bib_ref] Higher order aberrations of the corneal surface after laser subepithelial keratomileusis, Jung [/bib_ref]. Correlation analysis also revealed that total HOA, coma, and trefoil aberrations had a significantly negative correlation with TBUT while the correlation was significantly positive with OSDI score. These results suggest that additional HOAs caused by tear film instability in patients with dry eye after LASEK may lead to a further reduction in optical quality [bib_ref] Higher order aberrations of the corneal surface after laser subepithelial keratomileusis, Jung [/bib_ref]. Another group prospectively assessed HOAs in patients who underwent laserassisted in situ keratomileusis (LASIK) and compared this to patients with dry eyes and normal eyes [bib_ref] Effect of tear film break-up on higher order aberrations of the anterior..., Lin [/bib_ref]. The TBUT was significantly lower for dry eyes compared to both post-LASIK and normal eyes [bib_ref] Effect of tear film break-up on higher order aberrations of the anterior..., Lin [/bib_ref]. There was a significant increase in HOAs from post-blink to BUT in patients with normal and dry eyes and a subsequent decrease in HOAs after saline addition [bib_ref] Effect of tear film break-up on higher order aberrations of the anterior..., Lin [/bib_ref]. Interestingly, post-LASIK eyes showed no significant differences between the phases [bib_ref] Effect of tear film break-up on higher order aberrations of the anterior..., Lin [/bib_ref].
The tear film is subject to dynamic changes and one group assessed the dynamic properties and changes of wavefront aberrations in patients with dry eye and normal eyes [bib_ref] Dynamic wavefront aberrations and visual acuity in normal and dry eyes, Wang [/bib_ref]. Overall, the group noted that changes in wavefront aberrations and visual acuity is dynamic and variable patient-to-patient (in both dry and normal eyes) [bib_ref] Dynamic wavefront aberrations and visual acuity in normal and dry eyes, Wang [/bib_ref]. However, the mean instability of total wavefront aberrations and HOAs were significantly greater in patients with dry eye than in normal patients. Similarly, the mean instability of visual acuity was significantly higher in patients with dry eye compared to the normal patients [bib_ref] Dynamic wavefront aberrations and visual acuity in normal and dry eyes, Wang [/bib_ref]. Interestingly, the instability of visual acuity was dependent on blink rate in the dry eye group, showing that patients with a higher blink rate had more stable visual acuity [bib_ref] Dynamic wavefront aberrations and visual acuity in normal and dry eyes, Wang [/bib_ref]. This could suggest that patients with dry eye may compensate for tear breakup with increased blinking to maintain stable vision [bib_ref] Dynamic wavefront aberrations and visual acuity in normal and dry eyes, Wang [/bib_ref]. Another group similarly assessed tear film aberration dynamics in patients with dry eye disease (Sjögren's syndrome) and normal patients [bib_ref] Tear film aberration dynamics and vision-related quality of life in patients with..., Denoyer [/bib_ref]. At baseline, they found that the patients with dry eye had significantly lower TBUT and Schirmer test values compared to the control group [bib_ref] Tear film aberration dynamics and vision-related quality of life in patients with..., Denoyer [/bib_ref]. Meibomian dysfunction and ocular surface disease index (OSDI) scores were significantly higher for patients with dry eye compared to control [bib_ref] Tear film aberration dynamics and vision-related quality of life in patients with..., Denoyer [/bib_ref]. After blinking, there was a significant increase in total HOAs, coma, and trefoil aberrations in patients with dry eye [bib_ref] Tear film aberration dynamics and vision-related quality of life in patients with..., Denoyer [/bib_ref]. This increase in aberrations was not seen in the control eyes. Total HOAs, coma, and trefoil aberrations were also significantly higher in dry eyes compared to control eyes [bib_ref] Tear film aberration dynamics and vision-related quality of life in patients with..., Denoyer [/bib_ref]. The area under the curve of modulation transfer function (aucMTF), where MTF is the measurement of the ability of an optical system to transfer various levels of detail from object to image, was found to be significantly lower in patients with dry eye disease than controls at 7, 8, 9, and 10 s after blinking [bib_ref] Tear film aberration dynamics and vision-related quality of life in patients with..., Denoyer [/bib_ref]. There was also a significantly negative correlation between TBUT and HOA progression index, and a significantly positive correlation between OSDI scores and HOA progression index [bib_ref] Tear film aberration dynamics and vision-related quality of life in patients with..., Denoyer [/bib_ref]. The study was able to demonstrate that the time course of tear film HOAs correlates with the clinical examination (TBUT) and patient-reported outcomes (OSDI scores) of dry eye [bib_ref] Tear film aberration dynamics and vision-related quality of life in patients with..., Denoyer [/bib_ref].
Dry eye disease has been noted to be more frequent in patients with migraine. One group assessed the impact of dysfunctional tear films and aberrations in patients with migraine and normal eyes. Total aberrations, total HOAs, coma aberrations, spherical aberrations, and objective scatter index (OSI) were significantly higher in normal patients compared to the patients with migraine. Trefoil aberrations and TBUT were not significantly different between the groups. Patients with dry eyes as evaluated with a Schirmer test score \ 10 mm/5 min were excluded. The group's rationale behind this was to prevent overestimation of aberration profile with patients with confirmed dry eyes. On the basis of the results, patients with migraines do not have increased evidence of dry eyes or HOAs.
Another group assessed wavefront aberrations and functional visual acuity as the major causes of visual deterioration in patients with dry eye and Sjögren's syndrome [bib_ref] Corneal fluorescein staining correlates with visual function in dry eye patients, Kaido [/bib_ref]. The group also specifically looked at patients with dry eye and Sjögren's syndrome with and without SPK. The mean visual maintenance ratio (VMR) in the group with SPK was significantly lower than those without SPK and the normal group. The VMR was significantly lower in the group without SPK compared to the normal group [bib_ref] Corneal fluorescein staining correlates with visual function in dry eye patients, Kaido [/bib_ref]. The variation in visual acuity in the group with SPK was significantly higher than in the group without SPK and the normal group. Total HOAs and coma-like aberrations were significantly higher in the SPK group compared to those in the group without SPK and the normal group [bib_ref] Corneal fluorescein staining correlates with visual function in dry eye patients, Kaido [/bib_ref]. Serial changes in HOAs were also tracked and measured over 10 s; however, there were no significant changes in coma, spherical, and total HOAs in any of the groups. A significant negative correlation was found between VMR and both coma and total HOAs [bib_ref] Corneal fluorescein staining correlates with visual function in dry eye patients, Kaido [/bib_ref].
The impact of dry eye disease and functional visual impact in the context of driving was explored by another group [bib_ref] The impact of dry eye disease on visual performance while driving, Deschamps [/bib_ref]. The eyes of patients with dry eye and control patients were compared and total HOAs, coma, and trefoil aberrations varied significantly overtime in patients with dry eye, whereas this variation in aberrations was not found in control patients [bib_ref] The impact of dry eye disease on visual performance while driving, Deschamps [/bib_ref]. To assess the functional impact higher levels of HOA variations have on patients with dry eye, both groups underwent a driving visual performance assessment which required patients to identify targets. Patients with dry eye had a significantly longer response time to identify targets compared to control and were able to detect fewer targets compared to the control group. A significant positive correlation was also found between HOAs and the response time and OSDI symptom score.
In a unique case report, one group reported a woman with Sjögren's syndrome who underwent punctal plug placement to improve her dry eye symptoms [bib_ref] Paradoxical increase of visual impairment with punctal occlusion in a patient with..., Koh [/bib_ref]. However, despite dry eye symptom improvements, the patient reported epiphora and blurred vision immediately after blinking [bib_ref] Paradoxical increase of visual impairment with punctal occlusion in a patient with..., Koh [/bib_ref]. The total HOAs during the few seconds after blinking were spiked higher than before the punctal plug insertion [bib_ref] Paradoxical increase of visual impairment with punctal occlusion in a patient with..., Koh [/bib_ref]. Although these HOAs gradually declined and stabilized, a paradoxical increase in HOAs was observed in this patient after punctal plug insertion [bib_ref] Paradoxical increase of visual impairment with punctal occlusion in a patient with..., Koh [/bib_ref]. The same group also looked into the effects of suppressing blinking on vision quality in 10 patients with no ocular disease but short TBUT [bib_ref] Effects of suppression of blinking on quality of vision in borderline cases..., Koh [/bib_ref]. This group was further divided into a normal stable, normal-sawtooth, and short TBUT group [bib_ref] Effects of suppression of blinking on quality of vision in borderline cases..., Koh [/bib_ref]. The total HOAs and coma-like aberrations were significantly higher from 5 to 9 s after a blink compared to immediately after a blink [bib_ref] Effects of suppression of blinking on quality of vision in borderline cases..., Koh [/bib_ref]. Spherical-like aberrations were significantly higher from 6 to 9 s after a blink compared to immediately after a blink [bib_ref] Effects of suppression of blinking on quality of vision in borderline cases..., Koh [/bib_ref]. When the three groups were compared in terms of HOAs, fluctuation index (FI), and stability index (SI), the short TBUT group had significantly higher FI and SI compared to the normal stable group. There were no significant differences in FI and SI between the normal-sawtooth and short TBUT group [bib_ref] Effects of suppression of blinking on quality of vision in borderline cases..., Koh [/bib_ref].
Overall, the published evidence provides a clear link between dry eye disease and HOAs, in which HOAs are increased with dry eye severity . One of the most important factors to consider is the tear film in the context of dry eyes and HOAs, as many of the studies have demonstrated a significant decrease in TBUT in patients with dry eye [bib_ref] Corneal fluorescein staining correlates with visual function in dry eye patients, Kaido [/bib_ref] [bib_ref] Higher order aberrations of the corneal surface after laser subepithelial keratomileusis, Jung [/bib_ref] [bib_ref] Effect of tear film break-up on higher order aberrations of the anterior..., Lin [/bib_ref] [bib_ref] Tear film aberration dynamics and vision-related quality of life in patients with..., Denoyer [/bib_ref]. Tear film issues should be readily attenuated with lubricant drops, whereas punctate keratitis may not be so readily treated with lubricants. Showing improvements in HOAs after treating dry eyes helps strengthen the relationship and association of HOAs and dry eye severity. Therefore, it is also important to explore whether improvements in dry eye parameters impact HOA severity.
## Dry eye interventions and their effects on tear film metrics and hoa severity
Artificial tears are the first-line treatment for dry eye and seem to have a lubrication effect [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref]. However, the preservatives in artificial tears have been demonstrated to be detrimental to the ocular surface [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref]. Therefore, unpreserved eye drops or alternatives such as liposome sprays are preferred options [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref]. The first paper examined dry eye interventions and HOA outcomes using unpreserved Hypromellose drops, a liposome spray solution (Tears Again TM ), and the combination of the two in normal and dry eyes in an RCT of 48 participants [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref]. There was a significant decrease in HOAs immediately after Hypromellose drop instillation compared to baseline in both normal and dry eyes [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref]. However, this difference was no longer significant 1 h after intervention in both eye types. There was no significant difference in the liposome spray or combination intervention at any time point for both eye types [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref]. Hypromellose drops significantly increased total aberrations immediately after instillation compared to baseline in both normal and dry eyes but there was a similarly sized and significant reduction 1 h after intervention in both eye types [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref]. Liposome and combination intervention had no significant effect on total aberrations at any time point for both eyes [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref]. There were no significant changes in spherical or coma-like aberrations with Hypromellose, liposome, or combination intervention for both eyes [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref]. Despite insignificant and short-lived changes in HOAs, both eye type groups reported higher subjective comfort scores after combination treatment, followed by liposome spray, and Hypromellose drops last [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref].
Diquafosol sodium ophthalmic solution (DQS) promotes the secretion of water and mucin from conjunctival cells and has been shown to be effective in treating dry eyes, improving visual function, and effective even in post-cataract dry eyes [bib_ref] Effects of 3% diquafosol sodium ophthalmic solution on higher-order aberrations in patients..., Inoue [/bib_ref]. This group assessed the effects of 3% DQS and artificial tears (AT) in patients with dry eye after cataract surgery in an RCT of 42 participants [bib_ref] Effects of 3% diquafosol sodium ophthalmic solution on higher-order aberrations in patients..., Inoue [/bib_ref]. Tear film breakup time was significantly decreased in both DQS and AT groups compared to preoperative TBUT baseline values [bib_ref] Effects of 3% diquafosol sodium ophthalmic solution on higher-order aberrations in patients..., Inoue [/bib_ref]. After 4 weeks of DQS instillation there was a significant increase in TBUT, but no significant changes in TBUT were seen with AT [bib_ref] Effects of 3% diquafosol sodium ophthalmic solution on higher-order aberrations in patients..., Inoue [/bib_ref]. Similarly, corneal and conjunctival fluorescein staining was significantly increased in both DQS and AT groups postoperatively but only DQS significantly reduced fluorescein staining after 4 weeks of instillation [bib_ref] Effects of 3% diquafosol sodium ophthalmic solution on higher-order aberrations in patients..., Inoue [/bib_ref]. There were no significant changes in HOAs after 4 weeks post-cataract surgery in both groups, although DQS treatment stabilized HOA patterns, whereas AT treatment continued to have an upward curve pattern [bib_ref] Effects of 3% diquafosol sodium ophthalmic solution on higher-order aberrations in patients..., Inoue [/bib_ref]. The FI and SI were significantly lower in the DQS group compared to the AT group [bib_ref] Effects of 3% diquafosol sodium ophthalmic solution on higher-order aberrations in patients..., Inoue [/bib_ref]. DQS treatment was able to stabilize the pattern of HOAs, while AT treatment continued to show increasing sawtooth patterns [bib_ref] Effects of 3% diquafosol sodium ophthalmic solution on higher-order aberrations in patients..., Inoue [/bib_ref]. Another group assessed the effects of 3% DQS on 16 patients with aqueous-deficient dry eyes in a prospective comparison study [bib_ref] Effect of diquafosol ophthalmic solution on the optical quality of the eyes..., Koh [/bib_ref]. There were significant improvements in dry eye symptoms and corneal staining after treatment [bib_ref] Effect of diquafosol ophthalmic solution on the optical quality of the eyes..., Koh [/bib_ref]. Schirmer test scores and conjunctival staining were not significantly changed after treatment though [bib_ref] Effect of diquafosol ophthalmic solution on the optical quality of the eyes..., Koh [/bib_ref]. HOAs and spherical aberrations were significantly improved after 4 weeks of instillation when compared to baseline values [bib_ref] Effect of diquafosol ophthalmic solution on the optical quality of the eyes..., Koh [/bib_ref]. However, no changes in coma-like aberrations, FIs, and SIs were seen when compared to baseline values [bib_ref] Effect of diquafosol ophthalmic solution on the optical quality of the eyes..., Koh [/bib_ref]. DQS drops were also assessed in 11 patients with dry eye symptoms and short TBUT compared to patients without dry eye symptoms and short TBUT [bib_ref] Effects of diquafosol tetrasodium administration on visual function in short break-up time..., Kaido [/bib_ref]. There was a significant improvement in TBUT in symptom-negative patients treated with DQS but not with symptom-positive patients [bib_ref] Effects of diquafosol tetrasodium administration on visual function in short break-up time..., Kaido [/bib_ref]. The logMAR functional minimal, maximal, and visual acuities were significantly improved in symptom-positive patients, as were HOAs [bib_ref] Effects of diquafosol tetrasodium administration on visual function in short break-up time..., Kaido [/bib_ref].
Sodium hyaluronate (SH) is a glycosaminoglycan that improves tear film stability and decreases washout from the ocular surface [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref]. SH also binds water and resists dehydration to promote corneal epithelial wound healing [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref]. One group assessed the effects of 0.18% SH drops and a 0.9% sodium chloride (NaCl) solution on 50 patients with dry eye in an RCT [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref]. At baseline, there were no significant difference in HOAs in both SH-and NaCl-treated eyes [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref]. HOA was measured over 2 h after SH or NaCl instillation and there was no significant difference in HOAs at any time point in both eye groups [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref]. There was a significant decrease in spherical aberrations 1 min after SH instillation, but this gradually increased to baseline levels at later time points [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref]. Reported dry eye symptoms were significantly reduced in SH-treated eyes compared to NaCl control [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref]. Another group assessed SH at differing concentrations, 0.15% and 0.2%, along with a combination of 0.15% SH and trehalose in patients with dry eye in a prospective comparison study . Trehalose provides a hypo-osmolar effect and has positive contributory effects on the corneal surface and cell survival by stabilizing membrane lipids and proteins during desiccation . Trehalose and SH have also been shown to have synergistic action together in the context of dry eyes . The total root mean square (RMS) values of lower order aberrations (LOAs), HOAs, and spherical aberrations were all significantly decreased compared to their respective baseline values for each treatment group (0.15% SH, 0.2% SH, and 0.15% SH ? trehalose) . However, there was no significant difference amongst the treatment groups . Another group compared and assessed the effects of 0.2% SH and a lipid-based eye drop (Artelac Lipid) on 30 patients with dry eyes in an RCT [bib_ref] Effects of lipid-versus sodium hyaluronate-containing eye drops on optical quality and ocular..., Miháltz [/bib_ref]. There was a significant improvement in Schirmer test, TBUT, ocular staining, and symptom relief after 3 months in both SHand lipid-treated groups compared to baseline values [bib_ref] Effects of lipid-versus sodium hyaluronate-containing eye drops on optical quality and ocular..., Miháltz [/bib_ref]. There was no significant change in HOAs compared to baseline values [bib_ref] Effects of lipid-versus sodium hyaluronate-containing eye drops on optical quality and ocular..., Miháltz [/bib_ref]. In patients with a meibomian gland (MG) dropout rate greater than 50%, lipid treatment was able to significantly improve higher order Strehl ratio and HOAs compared to baseline [bib_ref] Effects of lipid-versus sodium hyaluronate-containing eye drops on optical quality and ocular..., Miháltz [/bib_ref].
The effect of 3% DQS compared to 0.1% SH in 86 patients with dry eyes after cataract surgery was also assessed in an RCT [bib_ref] Clinical effects and safety of 3% diquafosol ophthalmic solution for patients with..., Park [/bib_ref]. Ocular surface disease index (OSDI) was significantly lower in both treatment groups compared to baseline after 12 weeks of treatment, but there was no difference between the treatment groups [bib_ref] Clinical effects and safety of 3% diquafosol ophthalmic solution for patients with..., Park [/bib_ref]. TBUT, Schirmer test, and fluorescein staining were significantly improved in the DQS group compared to the SH group [bib_ref] Clinical effects and safety of 3% diquafosol ophthalmic solution for patients with..., Park [/bib_ref]. HOAs were significantly lowered and stabilized earlier in the DQS group compared to the SH group, although both groups had no significant difference between pre-and postoperative measurements of HOAs [bib_ref] Clinical effects and safety of 3% diquafosol ophthalmic solution for patients with..., Park [/bib_ref]. There was no significant difference in trefoil, coma, tetrafoil, and second astigmatism in both groups after 12 weeks of treatment [bib_ref] Clinical effects and safety of 3% diquafosol ophthalmic solution for patients with..., Park [/bib_ref]. Uncorrected distance visual acuity was significantly improved by the 1-week mark postsurgery in both groups and sustained within the 12-week treatment period [bib_ref] Clinical effects and safety of 3% diquafosol ophthalmic solution for patients with..., Park [/bib_ref]. DQS treatment was superior to SH in improving TBUT, changes in HOAs, Schirmer values, and staining scores throughout a 12-week follow-up period [bib_ref] Clinical effects and safety of 3% diquafosol ophthalmic solution for patients with..., Park [/bib_ref]. Another study similarly assessed the effects of 3% DQS (with and without preservatives) compared to 0.15% SA in 150 patients with pre-existing dry eye disease after cataract surgery in an RCT [bib_ref] Effects of preservativefree 3% diquafosol in patients with pre-existing dry eye disease..., Jun [/bib_ref]. TBUT was significantly improved for the preservative-free 3% DQS and preservativecontaining 3% DQS group compared to baseline measurements 1 month postoperatively [bib_ref] Effects of preservativefree 3% diquafosol in patients with pre-existing dry eye disease..., Jun [/bib_ref]. However, after 3 months, only the preservativefree 3% DQS group was significantly improved compared to baseline. OSDI scores were significantly improved at all time points throughout the study in both the preservative-free and preservative-containing 3% DQS groups. MGD stage was significantly reduced in the preservative-containing DQS group at 1 month postoperatively; however at 3 months, only the preservative-free DQS group had significantly improved MGD stage compared to group 3 [bib_ref] Effects of preservativefree 3% diquafosol in patients with pre-existing dry eye disease..., Jun [/bib_ref]. HOAs were all significantly improved compared to baseline values for all treatment groups compared to baseline and there were no significant differences between DQS (preservativefree and containing) and SH [bib_ref] Effects of preservativefree 3% diquafosol in patients with pre-existing dry eye disease..., Jun [/bib_ref].
The effects of 3% DQS, 0.3% SH, and 2% rebamipide solution were assessed on 15 healthy patients in a prospective comparison study [bib_ref] Effect of instillation of eyedrops for dry eye on optical quality, Koh [/bib_ref]. Rebamipide stimulates mucous secretion on the ocular surface [bib_ref] Effect of instillation of eyedrops for dry eye on optical quality, Koh [/bib_ref]. There was a significant increase in coma, spherical, and total HOAs immediately after instillation of all three drops (3% DQS, 0.3% SH, and 2% rebamipide) when compared to baseline values [bib_ref] Effect of instillation of eyedrops for dry eye on optical quality, Koh [/bib_ref]. Immediately after instillation, SH led to the highest increase in coma, spherical, and total HOAs compared to DQS, rebamipide, and control [bib_ref] Effect of instillation of eyedrops for dry eye on optical quality, Koh [/bib_ref]. However, 5 and 10 min after instillation, there were no significant differences in coma, spherical, or total HOAs among all groups and compared to baseline [bib_ref] Effect of instillation of eyedrops for dry eye on optical quality, Koh [/bib_ref]. There was a significant increase in forward light scatter immediately after rebamipide instillation compared to baseline and compared to DQS and SH [bib_ref] Effect of instillation of eyedrops for dry eye on optical quality, Koh [/bib_ref]. However, 5 and 10 min after instillation, there was again no significant difference in forward scatter among the groups and compared to baseline [bib_ref] Effect of instillation of eyedrops for dry eye on optical quality, Koh [/bib_ref]. The same group assessed the effects of 2% rebamipide on 16 patients with short TBUT type of dry eye in a prospective comparison study [bib_ref] Effect of rebamipide ophthalmic suspension on optical quality in the short break-up..., Koh [/bib_ref]. TBUT was significantly increased at the measured 2-and 4-week time points after treatment when compared to baseline [bib_ref] Effect of rebamipide ophthalmic suspension on optical quality in the short break-up..., Koh [/bib_ref]. Fluorescein staining and Schirmer test scores were not significantly different at 2 and 4 weeks after treatment [bib_ref] Effect of rebamipide ophthalmic suspension on optical quality in the short break-up..., Koh [/bib_ref]. During 10-s recordings of HOAs, there was a significant upward trend at baseline for coma, spherical, and total HOAs [bib_ref] Effect of rebamipide ophthalmic suspension on optical quality in the short break-up..., Koh [/bib_ref]. At the 2-week time point, only spherical aberrations had a significant increase in during the 10-s recording period/post-blink change [bib_ref] Effect of rebamipide ophthalmic suspension on optical quality in the short break-up..., Koh [/bib_ref]. At 4 weeks, there were no significant changes in coma, spherical, or total HOAs, indicating stability in HOAs [bib_ref] Effect of rebamipide ophthalmic suspension on optical quality in the short break-up..., Koh [/bib_ref]. The FI was significantly improved at the 2-and 4-week periods, whereas SI was significantly improved only compared to the 4-week period [bib_ref] Effect of rebamipide ophthalmic suspension on optical quality in the short break-up..., Koh [/bib_ref].
Cyclosporine A is a calcineurin inhibitor and potent immunomodulatory agent that inhibits T cell activation and reduces surface inflammation in the context and usage as an eye drop [bib_ref] Effectiveness and optical quality of topical 3.0% diquafosol versus 0. 05% cyclosporine..., Lee [/bib_ref]. This group assessed the effects of 3% DQS and 0.05% cyclosporine A (Restasis Ò ) in 40 patients with dry eye after cataract surgery in an RCT [bib_ref] Effectiveness and optical quality of topical 3.0% diquafosol versus 0. 05% cyclosporine..., Lee [/bib_ref]. DQS significantly improved TBUT at 1 and 3 months compared to the cyclosporine A group [bib_ref] Effectiveness and optical quality of topical 3.0% diquafosol versus 0. 05% cyclosporine..., Lee [/bib_ref]. Schirmer test scores and corneal staining were significantly improved for both treatment groups, but there was no difference between the groups [bib_ref] Effectiveness and optical quality of topical 3.0% diquafosol versus 0. 05% cyclosporine..., Lee [/bib_ref]. HOAs were significantly improved with cyclosporine at 2 and 3 months compared to baseline values and significantly better than DQS at 3 months after treatment [bib_ref] Effectiveness and optical quality of topical 3.0% diquafosol versus 0. 05% cyclosporine..., Lee [/bib_ref]. Vertical coma was also significantly improved in the cyclosporine A group at 3 months compared to baseline and at both 2 and 3 months when compared to DQS [bib_ref] Effectiveness and optical quality of topical 3.0% diquafosol versus 0. 05% cyclosporine..., Lee [/bib_ref]. DQS was more effective at increasing tear secretions and improving TBUT, but cyclosporine A was more effective at improving HOAs [bib_ref] Effectiveness and optical quality of topical 3.0% diquafosol versus 0. 05% cyclosporine..., Lee [/bib_ref].
Lifitegrast 5% is a pharmacologic eye drop that inhibits the inflammation associated with dry eye disease by blocking T cell adhesion and migration through inhibition of lymphocyte function-associated antigen 1 (LFA-1) to intercellular adhesion molecule 1 (ICAM-1) binding [bib_ref] The effect of lifitegrast on refractive accuracy and symptoms in dry eye..., Hovanesian [/bib_ref] [bib_ref] Safety and tolerability of lifitegrast ophthalmic solution 5.0%: pooled analysis of five..., Nichols [/bib_ref]. Lifitegrast is approved by the US Food and Drug Administration for treating both the signs and symptoms of dry eye disease [bib_ref] The effect of lifitegrast on refractive accuracy and symptoms in dry eye..., Hovanesian [/bib_ref] [bib_ref] Safety and tolerability of lifitegrast ophthalmic solution 5.0%: pooled analysis of five..., Nichols [/bib_ref]. One group assessed the effects of 5% lifitegrast solution in 100 patients with dry eyes undergoing cataract surgery in a prospective comparison study [bib_ref] The effect of lifitegrast on refractive accuracy and symptoms in dry eye..., Hovanesian [/bib_ref]. Eyes that were treated with lifitegrast for 28 days had significantly more accurate predicted spherical equivalent refraction for the lens actually implanted [bib_ref] The effect of lifitegrast on refractive accuracy and symptoms in dry eye..., Hovanesian [/bib_ref]. There was also a significant improvement and reduction in HOAs in 65% (44/77) of patients after the first course of treatment and a significant improvement in HOAs in 54% (27/77) of patients after the full course of treatment when compared to baseline values [bib_ref] The effect of lifitegrast on refractive accuracy and symptoms in dry eye..., Hovanesian [/bib_ref]. Corneal staining, TBUT, and conjunctival redness were all significantly improved after treatment with lifitegrast [bib_ref] The effect of lifitegrast on refractive accuracy and symptoms in dry eye..., Hovanesian [/bib_ref]. Standardized patient evaluation of eye dryness questionnaire also revealed a significant improvement in dry eye symptoms after lifitegrast treatment [bib_ref] The effect of lifitegrast on refractive accuracy and symptoms in dry eye..., Hovanesian [/bib_ref].
Punctal plug procedures are another intervention for dry eyes that block the puncta which increases tear fluid accumulation, thus keeping the tear film intact [bib_ref] Punctal plug: a medical device to treat dry eye syndrome and for..., Yellepeddi [/bib_ref]. One group assessed the effect of punctal plugs in eight patients with dry eye after LASIK in a prospective comparison study. Patients with dry eye after 1 month of post-punctal occlusion had significantly improved RMS error values, lower order wavefront errors, defocus, and astigmatism compared to pre-punctal measurements. There was also a significant improvement in the HOAs, spherical, and coma aberrations after punctal plug insertion.
One group assessed the effects of lubricating eye drops (Blink Intensive Tears) on 20 patients with dry eyes in a prospective comparison study. The TBUT was significantly lengthened at 10 min after instillation when compared to baseline. Spherical, coma, and HOAs were significantly decreased immediately after instillation of eye drops and were significantly maintained after 10 min.
# Discussion
As mentioned before, several studies have demonstrated that disruptions in the tear film caused by dry eye increase HOAs compared to normal eyes which results in impaired visual function and optical quality [bib_ref] Wavefront analysis of higher order aberrations in dry eye patients, Montés-Micó [/bib_ref] [bib_ref] Impaired functional visual acuity of dry eye patients, Goto [/bib_ref] [bib_ref] Corneal fluorescein staining correlates with visual function in dry eye patients, Kaido [/bib_ref] [bib_ref] Postblink changes in total and corneal ocular aberrations, Montés-Micó [/bib_ref]. Of the 12 papers directly looking at dry eyes and HOA outcomes, eight have definitively shown that there is a significant difference and increase in HOAs in patients with dry eyes and that with increasing dry eye severity, defined as patients with SPK, there is significant increase in HOAs as well [bib_ref] Corneal fluorescein staining correlates with visual function in dry eye patients, Kaido [/bib_ref] [bib_ref] Serial measurements of higher-order aberrations after blinking in patients with dry eye, Koh [/bib_ref]. The other five papers had differing results in supporting the association of dry eyes and HOAs. In one paper, although the differences in HOAs between normal and dry eyes were not assessed, TBUT was significantly less in dry eyes compared to normal eyes and total HOAs were decreased after saline drops in both patients with normal and dry eyes [bib_ref] Effect of tear film break-up on higher order aberrations of the anterior..., Lin [/bib_ref]. Patients with migraines are reported to have more frequent dry eye diseaseand this group was able to demonstrate that coma, spherical, and total HOAs were significantly higher in their control group compared to the migraine group. This group specifically excluded patients with actual dry eye disease so as not to overestimate their measured parameters though. The last two case reports by the same group showed an opposite phenomenon where HOAs were increased after blinking in post-punctal plug placement patients and in patients with short TBUT [bib_ref] Paradoxical increase of visual impairment with punctal occlusion in a patient with..., Koh [/bib_ref] [bib_ref] Effects of suppression of blinking on quality of vision in borderline cases..., Koh [/bib_ref]. HOAs tend to increase in normal patients and patients with dry eye post blink due to disruption and changes in the tear film [bib_ref] Wavefront analysis of higher order aberrations in dry eye patients, Montés-Micó [/bib_ref] [bib_ref] Postblink changes in total and corneal ocular aberrations, Montés-Micó [/bib_ref] [bib_ref] Paradoxical increase of visual impairment with punctal occlusion in a patient with..., Koh [/bib_ref] [bib_ref] Effect of tear film break-up on higher-order aberrations measured with wavefront sensor, Koh [/bib_ref]. Before the tear film breaks up, the difference in tear film thickness of the thinner superior and thicker inferior cornea may lead to delayed wavefronts inferiorly and increased coma-like aberrations due to the longer optical path in the inferior tear film compared to the thinner superior tear film [bib_ref] Paradoxical increase of visual impairment with punctal occlusion in a patient with..., Koh [/bib_ref]. Similarly, punctal occlusion in some cases of mild dry eye may cause visual impairment due to the post-blink movement of excessive tear volume [bib_ref] Paradoxical increase of visual impairment with punctal occlusion in a patient with..., Koh [/bib_ref]. A significant negative correlation between HOAs and TBUT was also demonstrated, providing further evidence that as TBUT decreases, and consequently dry eye severity increases, HOAs increases [bib_ref] Higher order aberrations of the corneal surface after laser subepithelial keratomileusis, Jung [/bib_ref] [bib_ref] Tear film aberration dynamics and vision-related quality of life in patients with..., Denoyer [/bib_ref].
As the tear film and TBUT are crucial factors in dry eye disease and HOAs, tear film issues should be readily attenuated with interventions to improve TBUT or help stimulate lipid and aqueous secretions. Therefore, showing improvements in HOAs after treating dry eyes would help strengthen the relationship of HOAs and dry eye severity. From the 15 papers looking at dry eye interventions and HOA outcomes, 12 have shown improvements in HOAs to varying degrees [bib_ref] A preliminary investigation into the effects of ocular lubricants on higher order..., Mcginnigle [/bib_ref] [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref] [bib_ref] The effect of lifitegrast on refractive accuracy and symptoms in dry eye..., Hovanesian [/bib_ref] [bib_ref] Clinical effects and safety of 3% diquafosol ophthalmic solution for patients with..., Park [/bib_ref] [bib_ref] Effectiveness and optical quality of topical 3.0% diquafosol versus 0. 05% cyclosporine..., Lee [/bib_ref] [bib_ref] Effects of lipid-versus sodium hyaluronate-containing eye drops on optical quality and ocular..., Miháltz [/bib_ref] [bib_ref] Effect of diquafosol ophthalmic solution on the optical quality of the eyes..., Koh [/bib_ref] [bib_ref] Effects of diquafosol tetrasodium administration on visual function in short break-up time..., Kaido [/bib_ref] [bib_ref] Effects of preservativefree 3% diquafosol in patients with pre-existing dry eye disease..., Jun [/bib_ref]. In particular, some papers did reveal improvements in HOAs but only for a short time period immediately after instillation of the intervention [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref] , or revealed an improvement in total HOAs but an increase in vertical trefoil and horizontal coma aberrations . The variation in results can be a result of many factors, including the variation in patient dry eye severity as well as the type of eye drop and intervention used. The viscosity of eye drops plays an important role in the impact of aberrations and has been shown to impact aberration parameters [bib_ref] Effect of rebamipide ophthalmic suspension on optical quality in the short break-up..., Koh [/bib_ref]. The type of intervention used also plays a role as one study was able to show that SH was able to significantly improve TBUT over DQS, but DQS was able to improve HOAs significantly more than SH [bib_ref] Clinical effects and safety of 3% diquafosol ophthalmic solution for patients with..., Park [/bib_ref]. Whereas, another study demonstrated that DQS and SH were both able to significantly improve HOAs without any significant difference between the two [bib_ref] Effects of preservativefree 3% diquafosol in patients with pre-existing dry eye disease..., Jun [/bib_ref]. Another paper demonstrated that DQS significantly improved TBUT over cyclosporine A, but cyclosporine A was able to significantly improve HOAs over DQS [bib_ref] Effectiveness and optical quality of topical 3.0% diquafosol versus 0. 05% cyclosporine..., Lee [/bib_ref]. In addition, DQS did not significantly change HOA outcomes and the group postulated that this may be due to 3% DQS causing excessive secretion of fluid that paradoxically disturbs tear film stability [bib_ref] Effectiveness and optical quality of topical 3.0% diquafosol versus 0. 05% cyclosporine..., Lee [/bib_ref]. Related to the viscosity and makeup of the eye drops are the morphological changes and patterns that certain eye drops and artificial tears can produce on the ocular surface, which may be attribute to the variation in clinical success seen [bib_ref] Spectral-domain optical coherence tomography study on dynamic changes of human tears after..., Napoli [/bib_ref]. The turnover of tear film components, in particular the lipid components, plays an important role in the effectiveness of eye drops and is an important aspect to consider when utilizing lipid-based eye drop treatments [bib_ref] A novel technique of contrast-enhanced optical coherence tomography imaging in evaluation of..., Napoli [/bib_ref].
The variation in dry eye disease in patients also played an important role when looking at HOA outcomes. For instance, in one study that assessed the effects of SH in their dry eye groups, HOAs were not significantly changed after instillation [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref] while other studies have shown that SH improved HOAs [bib_ref] Changes in ocular aberrations after instillation of artificial tears in dryeye patients, Montés-Micó [/bib_ref]. The researchers attributed this to their population group having both aqueous tear-deficient and evaporative dry eye with moderate to severe dry eye grading with some cases of SPK [bib_ref] Effects of sodium hyaluronate on wavefront aberrations in dry eye patients, Lekhanont [/bib_ref] , compared to the patients with more mild dry eye in previous studies [bib_ref] Effect of tear film break-up on higher-order aberrations measured with wavefront sensor, Koh [/bib_ref].
One of the limitations of our paper includes the possibility that not all primary articles looking into patients with dry eye and HOAs were captured. However, we have tried to overcome these limitations by utilizing a very broad initial search strategy as mentioned in our ''Methods'' section and curated for relevant papers on the basis of [fig_ref] Figure 1: PRISMA [/fig_ref].
Overall, there is evidence to show that an association between dry eyes and HOAs exists and that treating dry eye disease through interventions including eye drops, secretagogues, and punctal plugs can improve HOA outcomes. As dry eye disease is one of the most common ocular disorders and complications from many common laser and refractive surgeries, the clinical implications of improving HOAs are an important area of study. Factors such as patient variation in dry eye disease and severity as well as eye drop type and viscosity are a few of the many areas of study that need to be further elucidated for better clinical application.
# Conclusion
On the basis of this systematically approached literature review, dry eyes and HOAs have a clear association with the tear film being one of the most important factors in this relationship between the two. Improvements in HOAs with dry eye treatment interventions such as artificial tears, secretagogues, and punctal plugs provide further evidence on this relationship between dry eyes and HOAs. However, there is still much more work to be done in the realm of clinical application as dry eye interventions vary depending on many factors including patient severity, intervention type, eye drop viscosity, and the mechanism of action of intervention. Therefore, although improving HOAs seems to be a viable clinical approach, further evidence and work are required for improved clinical application and outcomes.
## Acknowledgements
Funding. No funding or sponsorship was received for this study or publication of this article. The journal's Rapid Service Fee was funded by the authors.
Authorship. All named authors meet the International Committee of Medical Journal Editors (ICMJE) criteria for authorship for this article, take responsibility for the integrity of the work as a whole, and have given their approval for this version to be published.
Author Contributions. Jess Rhee and Kendrick Co Shih were involved in study design, data collection, data analysis, manuscript writing and editing. Tommy Chung-Yan Chan, Sharon Shu-Wan Chow, Antonio Di Zazzo, Takenori Inomata and Louis Tong were involved in data collection, data analysis, manuscript writing and editing.
Disclosures. The following authors confirm that they have no relevant financial disclosures to declare: Jess Rhee, Tommy Chung-Yan Chan, Sharon Shu-Wan Chow, Antonio Di Zazzo, Takenori Inomata, Kendrick Co Shih, Louis Tong.
Compliance with Ethics Guidelines. This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.
Data Availability. This is a review article and, as such, does not generate new data. The authors agree to make all materials, data and associated protocols promptly available to readers without undue qualifications in material transfer agreements.
## Open access. this article is licensed under a creative
Commons Attribution-Non-Commercial 4.0 International License, which permits any non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creativecommons.org/licenses/by-nc/4.0/.
[fig] Figure 1: PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) flowchart illustrating the selection process of articles [/fig]
[table] Table 1: Summary of HOAs and its association with dry eyes [/table]
[table] Table 2: Summary of dry eye interventions and impact on HOAs [/table]
|
Arthroscopic management for early-stage tuberculosis of the ankle
Background: Due to atypical clinical presentation, wide use of antibiotics, and lack of specificity in diagnosis, diagnosis of tubercular (TB) infection in joints is increasingly difficult, and misdiagnosis is common. The use of arthroscopy for the diagnosis and treatment of early-stage ankle TB has rarely been reported. This case series intended to present the clinical outcomes of arthroscopic management for early-stage ankle TB. Methods: Fifteen patients with chronic synovitis of the ankle and suspicious cause of early-stage ankle TB underwent arthroscopic treatment from April 1, 2010, to March 31, 2016. These cases all failed to confirm diagnosis of TB by ankle arthrocentesis. They included seven males and eight females with an average age of 37.5 (8 to 70) in the study. Among them, five cases had history of pulmonary tuberculosis, and six had history of trauma. The procedure included synovial membrane biopsy and debridement. The diagnosis was confirmed by pathologic examination and culture. The treatment was combined with systemic anti-tuberculous drugs. Follow-up measurements included VAS score, AOFAS score, ESR, CRP, and MRI. Results: After arthroscopic management, 13 cases confirmed TB by pathologic examination and culture, and two cases still remained clinically suspected TB; the rate of confirmed case was 87%. The incision healed well in all cases, and no serious complications were observed. There were significant differences in VAS scores, AOFAS scores, ESR, and CRP between before and after treatment (P < 0.01). Joint swelling disappeared or was relieved after 2 months in most patients. Ankle swelling and pain in one patient was improved after changing anti-tuberculous drugs. MRI suggested that all patients had effusion in the articular cavity, accompanied by bone edema of the distal tibia and talus before the treatment. After the surgery, the effusion was significantly reduced, and the signal of bone edema almost disappeared. No recurrent TB was found during the follow-ups. Conclusion: Arthroscopic management for early-stage ankle TB is minimally invasive, safe, and reliable. It can easily obtain samples from specific area of TB for further confirmation of the diagnosis, while the debridement can also assist in local disease control. For cases of highly suspicious joint TB, arthroscopic biopsy and debridement after transient anti-TB treatment is recommended. Level of evidence: Level IV, therapeutic case series
# Background
According to WHO's 2014 estimation, about 9.6 million people in the world were infected with tuberculosis (TB), resulting in 1.5 million deaths, which made TB the primary cause of infectious diseases [bib_ref] Tuberculosis-advances in development of new drugs, treatment regimens, host-directed therapies, and biomarkers, Wallis [/bib_ref]. In 2015, the annual incidence of tuberculosis in China was 68/100,000. Joint TB accounts for about 3% of the extrapulmonary TB. Occasionally, there would be cases of ankle TB [bib_ref] Tuberculosis of the foot and ankle, Dhillon [/bib_ref] [bib_ref] Arthroscopic operations in knee joint with early-stage tuberculosis, Shen [/bib_ref] [bib_ref] Knee arthrodesis using a unilateral external fixator combined with crossed cannulated screws..., Tang [/bib_ref] [bib_ref] A diagnosis of disseminated tuberculosis based on knee arthroscopic guided synovial biopsy..., Fernandes [/bib_ref]. China has shown a high incidence of TB [bib_ref] Tuberculosis-advances in development of new drugs, treatment regimens, host-directed therapies, and biomarkers, Wallis [/bib_ref] [bib_ref] Utility of real-time quantitative polymerase chain reaction in detecting Mycobacterium tuberculosis, Lv [/bib_ref] , and the problem is becoming even more severe with the emergence of multidrug-resistant TB in recent years [bib_ref] Recurrence after successful treatment of multidrug-resistant tuberculosis in Taiwan, Chen [/bib_ref]. Sinus tract is often concomitant in ankle TB, which would severely affect the motor function of the ankle. The diagnosis of tuberculous infection in a joint is difficult, and misdiagnosis is common due to its atypical clinical presentation, wide use of antibiotics, and lack of specificity in diagnosis.
A few decades ago, ankle arthroscopy was considered unavailable because the joint spaces were too narrow to operate. But in the recent 20 years, it has made astounding advances and been successfully used for osteochondral lesions of the talus, impingement syndrome, ankle arthrodesis, etc. [bib_ref] Ankle and foot tuberculosis: a diagnostic dilemma, Nayak [/bib_ref] [bib_ref] Long-term follow-up results of foot and ankle tuberculosis in Turkey, Gursu [/bib_ref] [bib_ref] Long-term retrospective analysis of surgical treatment for irretrievable tuberculosis of the ankle, Chen [/bib_ref] [bib_ref] Adjuvant arthroscopy for ankle tuberculosis: a report of two cases, Lin [/bib_ref] [bib_ref] Multidrug-resistant tuberculosis of the ankle: case report, Baquero-Artigao [/bib_ref] [bib_ref] Arthroscopy of the ankle and foot, Ferkel [/bib_ref] [bib_ref] Simultaneous ankle arthroscopy and hindfoot endoscopy for combined anterior and posterior ankle..., Miyamoto [/bib_ref] [bib_ref] Trends in ankle arthroscopy and its use in the management of pathologic..., Werner [/bib_ref] [bib_ref] Role of ankle arthroscopy in management of acute ankle fracture, Chan [/bib_ref] [bib_ref] Removal of osteoblastoma of the talar neck using standard anterior ankle Arthroscopy:..., Duan [/bib_ref]. However, there are very few reports on ankle TB treatment [bib_ref] Tuberculosis of the foot and ankle, Dhillon [/bib_ref] [bib_ref] Ankle and foot tuberculosis: a diagnostic dilemma, Nayak [/bib_ref] [bib_ref] Long-term follow-up results of foot and ankle tuberculosis in Turkey, Gursu [/bib_ref] [bib_ref] Long-term retrospective analysis of surgical treatment for irretrievable tuberculosis of the ankle, Chen [/bib_ref] [bib_ref] Adjuvant arthroscopy for ankle tuberculosis: a report of two cases, Lin [/bib_ref] [bib_ref] Multidrug-resistant tuberculosis of the ankle: case report, Baquero-Artigao [/bib_ref] , and arthroscopic management for early-stage ankle TB is even less reported.
The main manifestation of early-stage ankle TB is chronic synovitis. Even though ankle arthrocentesis, bacterial culture, and pathological examination have been performed, it is still difficult to confirm the diagnosis. The authors have been using arthroscopy in treating ankle TB for some years and have achieved satisfying results. In order to further improve the level of diagnosis and treatment, we have followed the clinical outcomes after arthroscopic management for early-stage ankle TB and reported as follows.
# Methods
The inclusion criteria were as follows: (1) patients with chronic ankle synovitis, where inflammatory and traumatic origins have been ruled out, and a strong TB suspicion was present but no infectious origin could be determined; and (2) patients who were able to take oral anti-TB drugs on schedule. The exclusion criteria were as follows: (1) patients who had been diagnosed with ankle TB, preferred conservative treatment, and did not need to obtain samples via arthroscopy; (2) patients who had sinus, sequestrum, and bone destruction; other surgical procedures were often necessary in addition to debridement; (3) patients who had unstable vital signs and were unsuitable for surgery; (4) patients who had low compliance of anti-TB drugs; and (5) patients who had active pulmonary TB, which was contagious and should be transferred to a specialist hospital.
Retrospective study was performed to analyze the patients from April 1, 2010, to March 31, 2016. Fifteen cases were included (seven men and eight women; mean age 37.5 years, range 8-70 years; mean duration of illness 16 months, range 3-36 months). All 15 cases involved unilateral ankle joints, which were swollen and painful. Among them, five cases had history of pulmonary tuberculosis, and six had history of trauma [fig_ref] Table 1: General data of patients [/fig_ref]. They experienced ineffective long-term oral treatment of non-specific anti-inflammatory drugs. Systemic symptoms of TB such as fever, night sweats, and wasting away were not obvious when they transferred to our hospital. Each patient was diagnosed by detailed clinical, radiological, and laboratory evaluations before surgery. All patients had elevated erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) levels preoperatively. The primary diagnosis was chronic synovitis of the ankle. Purulent infection, rheumatoid arthritis, traumatic arthritis, etc. were preliminarily ruled out, and we strongly suspected that the cause was TB. These patients had indications for an ankle lesion biopsy. To reduce complications, all patients received anti-TB drug therapy for at least 2-3 weeks, until the ESR level decreased or became normal. The drugs used were isoniazid (300-450 mg/day for adults; 10 mg/kg/day for children), rifampicin (450-750 mg/day for adults; 10 mg/ kg/day for children), ethambutol (15-25 mg/kg/day for adults and children), and pyrazinamide (20-25 mg/ kg/day). VAS scores and AOFAS scores were assessed prior to the treatment [bib_ref] Surgical strategy for the chronic Achilles tendon rupture, Lin [/bib_ref]. Ankle anteroposterior and lateral radiographs and MRI were used to determine the extent and area of local disease damage.
## Surgical purpose and technique
The purposes of the surgery included two aspects: [bib_ref] Tuberculosis-advances in development of new drugs, treatment regimens, host-directed therapies, and biomarkers, Wallis [/bib_ref] for the suspected cases, to directly observe the lesion under arthroscope and to make relevant examinations with the focal tissue (pathological examination, L-J medium culture, TB-PCR) for further confirmation, as well as to improve the treatment plan with more accurate judgments on the prognosis; (2) arthroscopic debridement, i.e., direct removal of necrotic tissue, which could improve the History of pulmonary tuberculosis (n, %) 5 (33.3%)
History of trauma (n, %) 6 (40.0%)
Time from symptom to surgery (months) 16 effective concentration of anti-TB drugs in the focal tissue and be an adjunct to local disease control. During the surgery, if the area of articular cartilage damage is less than 2 cm 2 and the residual cartilage is stable, it is often effective to perform debridement rather than arthrodesis. Epidural anesthesia or nerve block anesthesia was applied. Patients were placed in supine position, buttocks padded high, and the ankle was maintained in the neutral position. Marker pen was used to mark the superficial peroneal nerve and dorsalis pedis artery. Pneumatic tourniquet was applied with the pressure of 250-300 mmHg. The non-invasive ankle traction was used with the posterior malleolus hanging, which was easy for debridement [bib_ref] Arthroscopy of the ankle and foot, Ferkel [/bib_ref]. Ankle arthrocentesis was performed to drain the joint effusion, and sample of the joint effusion was obtained for acid-fast staining test; saline 10-20 mL was injected into the articular cavity. After incision, a 30°ankle arthroscope (2.7 mm in diameter, Smith & Nephew Endoscopy, Andover, MA) was used. Ankle exploration was carefully performed through the anteromedial incision between the anterior tibialis tendon and the medial malleolus [bib_ref] Arthroscopic arthrodesis for ankle arthritis without bone graft, Duan [/bib_ref] ; further exploration was conducted through the anterolateral incision at the lateral of the peroneus tertius tendon under arthroscope. In general, synovial fluid that is purulent and turbid could be seen under arthroscope, and sometimes fibrous protein and necrotic tissue, hyperplasia of synovial tissue, congestion, and pale areas could also be seen. The diversity of the cartilage damage could be manifested as cartilage degeneration, defects, layering, and subchondral bone exposure. Next, typical pathological tissue and necrotic tissue were taken under the arthroscope as samples for pathological examination, Lowenstein-Jensen medium culture, TB-PCR, etc. Shaver, basket forceps, pituitary rongeur, and other instruments were used to completely remove the necrotic tissue and degenerative synovia of the articular cavity. A drainage tube was placed to infuse 0.1 g isoniazid into the joint cavity and then clamped. The negative pressure drainage could be started 12 h after surgery. The incision was closed with 3-0 sutures. After surgery, all patients were given plaster casts to keep the ankle in the functional position.
Postoperative management for ankle TB is very important. Patients with fever or pain should receive symptomatic treatment. The volume of the negative pressure drainage must be observed carefully, and drain removal is usually performed 24-48 h after surgery. In addition, patients are encouraged to have high-protein diet to improve nutrition. It is an important principle to reduce the movement of ankle joint or to conduct moderate exercise without ankle pain. Weight-bearing exercise could be done 2 weeks postoperatively when the sutures are taken out. Postoperative anti-TB drugs are also very important, with the principles of combined, standardized, adequate, and full course. The quadruple therapy of isoniazid, rifampicin, pyrazinamide, and ethambutol was continued for 3 months, triple therapy of ibuprofen, rifampicin, and pyrazinamide for another 3 months, and then the combination of isoniazid and rifampicin; the total oral administration of anti-TB drugs lasted up to 18 months. The patients could walk with gradual weightbearing after the sutures were taken out 2 weeks after surgery.
## Follow-up and statistical analysis
The patients were followed up at postoperative 2 and 6 weeks and 3, 6, 12, 18, and 24 months. The symptom changes and the laboratory tests such as ESR, CRP, blood routine, liver and kidney function were re-checked. VAS and AOFAS scores were checked at the last followup. The cure criteria were as follows: after 3 months of drug discontinuance, patients could be discharged when they could walk painlessly for 1 km and ESR was normal for more than 3 months postoperatively. The enumeration data was expressed with mean ± standard deviation; statistical analysis was conducted by paired t test with SPSS 16.0; P < 0.05 indicated statistical significance.
# Results
The diagnosis of 13 patients was confirmed postoperatively by pathologic analysis and culture of mycobacterium tuberculosis. The remaining two cases still failed to be diagnosed by biopsy. Therefore, with the numbers available in this study, through pathological examination and bacterial culture of the samples obtained from arthroscopy, 87% of the total cases that could not be diagnosed by ankle arthrocentesis preoperatively could be confirmed. We continued giving all 15 cases anti-TB drugs after arthroscopic management, and their symptoms disappeared completely. All incisions were healed well. There were significant differences in ESR, VAS, and AOFAS scores before and after treatment [fig_ref] Table 2: Comparison of results before treatment and at the last follow-up VAS score... [/fig_ref]. No severe complications were observed, and all patients were cured. MRI examination suggested that all patients had preoperative effusion in the ankle articular cavity, accompanied by bone edema of the distal radius and the talus. After treatment, the effusion was significantly reduced, and the bone edema signal almost disappeared. According to the AOFAS score in the last follow-up, nine cases were excellent, three cases good, and three cases acceptable.
In one case, the symptoms were not significantly improved 3 months postoperatively. After consulting with pharmacologists, anti-TB drugs were adjusted and the Paired t-test P < 0.01 P < 0.01 P < 0.01 P < 0.01 patient was cured in later follow-ups. We chose three typical cases as follows.
## Case 1
A female, 36 years old, had symptoms of pulmonary TB and intracranial TB which were alleviated after using anti-TB drugs, but the swelling and pain of the left ankle lasted for 1 year. Ankle TB was confirmed with the pathological examination of focal tissue after arthroscopy. It showed hyperplasia of synovial tissue, pale areas, regional congestion, scattered fibrous protein, and necrotic tissue, but the residual cartilage remained stable. Arthroscopic debridement was performed to remove the ankle scarring tissue and improve the motor function of the ankle. Re-examination showed she was cured with an AOFAS score of 93 points , Additional file 1: Video S1).
## Case 2
The right ankle of a 64-year-old male gradually became swollen and painful. The results of venous blood test for rheumatoid factor (RF), anti-cyclic citrullinated peptide (CCP) antibody, anti-keratin antibody (AKA), and HLA-B27 were all negative. The patient received treatment for rheumatoid arthritis and was given analgesic drugs orally; no hormone was taken orally. The curative effect was poor. After 6 months of treatment, the ankle joint pain and swelling were aggravated and claudication occurred. Venous blood test was performed again, and the results of T-SPOT.TB test and TB antibody were positive; chest radiograph showed pulmonary TB. Ankle arthrocentesis was conducted; the result of bacterial culture was negative and suspected TB for pathological examination. Quadruple anti-TB therapy (isoniazid, rifampicin, pyrazinamide, ethambutol) was given orally. After 3 weeks of treatment, the swelling of the right ankle joint was relieved. ESR was improved from 42 mm/h before treatment to 25 mm/h. Then ankle arthroscopy was performed. Under arthroscope, a small area of defect in the ankle cartilage was seen, but the cartilage remained stable without looseness; the ankle joint had a large amount of fibrous protein as well as hyperplasia and hyperemia of the synovial tissue. Samples were taken for TB culture and pathological examination, and then ankle debridement was performed. Postoperative pathological examination confirmed ankle TB. Anti-TB treatment was continued for 18 months. At the last follow-up, the symptoms of the ankle joint disappeared, the ESR was 8 mm/h, and the AOFAS score improved from 49 points before treatment to 94 [fig_ref] Figure 2 a: Preoperative right ankle swelling [/fig_ref].
## Case 3
A male, 11 years old, suffered from pain in the left ankle joint after trauma and limitation of motion for 10 years, and the symptoms aggravated with claudication for 1 year. Physical examination showed swelling of the left ankle joint, extensive tenderness, normal skin temperature, and limited range of motion for ankle plantar flexion and dorsi TB was highly suspected. Considering that the patient was still a child, ankle arthrodesis was not suitable in this case. Therefore, we only performed articular cavity debridement after obtaining the tissue sample for biopsy. Pathological examination confirmed ankle TB, and the result was also positive for TB-PCR. The patient was prevented from weight bearing for 6 weeks postoperatively; anti-TB treatment of rifampicin, isoniazid, and pyrazinamide were given orally; his nutrition was strengthened. Regular follow-ups were conducted. The swelling of the posterior malleolus was gradually relieved. After the reexamination at 6 months postoperatively, his anti-TB therapy was adjusted and only rifampicin and isoniazid were continued for maintenance treatment of 12 months. At the last follow-up at 5 years postoperatively, the patient's left ankle swelling and pain disappeared, and the range of motion for ankle plantar flexion and dorsi flexion was basically normal. The AOFAS score improved from 57 points preoperatively to 97, with ESR 1 mm/h. Radiograph and MRI suggested that the ankle joint space was slightly narrow, the surface of tibiotalar joint was not smooth, and the lesion of synovial hyperplasia disappears. The patient was satisfied with the results , Additional file 2: Video S2 and Additional file 3: Video S3).
# Discussion
Tuberculosis, which is mainly found in developing countries, still remains as a major global health problem for 20 years since WHO announced it a global public health issue [bib_ref] Tuberculosis-advances in development of new drugs, treatment regimens, host-directed therapies, and biomarkers, Wallis [/bib_ref] [bib_ref] Tuberculosis of the foot and ankle, Dhillon [/bib_ref] [bib_ref] First insight into the molecular epidemiology of Mycobacterium tuberculosis isolates from the..., Chen [/bib_ref]. In 2010, China published an official guideline for the diagnosis and treatment of pulmonary TB. However, there are no guidelines for joint TB, whose treatment is still referred to guidelines for pulmonary TB [bib_ref] Knee arthrodesis using a unilateral external fixator combined with crossed cannulated screws..., Tang [/bib_ref]. Due to its atypical symptoms, joint TB is often neglected and misdiagnosed [bib_ref] Arthroscopic operations in joint tuberculosis, Titov [/bib_ref] [bib_ref] Osteoarticular tuberculosis-a three years' retrospective study, Arathi [/bib_ref] [bib_ref] Arthroscopically assisted treatment of adolescent knee joint tuberculosis, Guo [/bib_ref]. According to references [bib_ref] Long-term follow-up results of foot and ankle tuberculosis in Turkey, Gursu [/bib_ref] [bib_ref] Arthroscopic synovectomy of synovial tuberculosis of the knee joint, Motycka [/bib_ref] [bib_ref] Tuberculosis of the hand: clinical presentation and functional outcome in 32 patients, Kotwal [/bib_ref] [bib_ref] Tuberculosis of the hip, Babhulkar [/bib_ref] , some patients were given open surgery while receiving anti-TB drugs, but this method would be more traumatic and cause difficulty in postoperative incision healing. Arthroscopy has made great progress in many countries. Yet arthroscopy for ankle TB is still rarely reported, and the effectiveness and safety of the technique is unclear.
Early diagnosis and treatment of joint TB has a decisive significance on the prognosis [bib_ref] Tuberculosis of the foot and ankle, Dhillon [/bib_ref] [bib_ref] Anterior spinal decompression in HIV-positive patients with tuberculosis. A prospective study, Govender [/bib_ref] [bib_ref] Tuberculosis of the foot, Mittal [/bib_ref] [bib_ref] Sacroiliac joint tuberculosis. Classification and treatment, Kim [/bib_ref] [bib_ref] Unusual locations of osteoarticular tuberculosis, García [/bib_ref] [bib_ref] Global epidemiology of tuberculosis. Morbidity and mortality of a worldwide epidemic, Raviglione [/bib_ref] [bib_ref] Characteristics and clinical outcome of bone and joint tuberculosis from 1994 to..., Johansen [/bib_ref] [bib_ref] MRI features of tuberculosis of peripheral joints, Sawlani [/bib_ref]. MRI examination is helpful in the early diagnosis of ankle TB [bib_ref] MRI in ankle tuberculosis: review of 14 cases, Zacharia [/bib_ref]. It can show clear vision of effusion in the articular cavity and sometimes with low-signal focal tissue, and high signal of edema is seen in the distal tibia and the talus (kissing sign). Under MRI, the signal dispersion range of TB is usually greater than osteoarthritis. It is often confined to the lateral of the tibiotalar joint rather than bilateral, which is seen in the case of tumor or early bone necrosis. It is also easy to observe the soft tissue abscess near the ankle and even in the subtalar joint by MRI. Due to its non-invasiveness and sensitiveness, MRI is used as a routine examination of ankle TB. Arthroscopy is of great diagnostic value for cases that cannot be confirmed by ankle arthrocentesis. Through this minimally invasive surgery, typical lesion samples can be obtained for pathological examination, and the positive rate was higher than that by arthrocentesis; besides, the sample volume obtained by arthroscopy was also more, which was very useful for bacterial culture and other related examinations. Arthroscopy can also timely correct misdiagnosed cases, such as in our case 3. Cases that cannot be confirmed before surgery are often quite intractable. In this study, the rate of confirmed cases after arthroscopy is 87%, which we believe is very impressive.
Arthroscopy could not only provide samples for pathological examination and L-J medium culture to confirm the diagnosis, but also be an adjunct to local disease a Preoperative radiograph suggesting unsmooth ankle joint space and epiphyseal injury of the tibia (anteroposterior view). b Preoperative radiograph suggesting unsmooth ankle joint space (lateral view). c Preoperative MRI suggesting ankle joint damage and hyperplasia of synovial tissue (sagittal view). d Preoperative MRI suggesting ankle joint damage and hyperplasia of synovial tissue (coronal view). e Articular cartilage defects, massive synovial hyperplasia, and scar tissue formation seen during arthroscopy. f Radiograph 5 years postoperatively suggesting that the ankle joint is not smooth, but the epiphyseal injury of the tibia is better than before surgery (anteroposterior view). g Radiograph 5 years postoperatively suggesting that the ankle joint is not smooth but is better than before surgery (lateral view). h MRI 5 years postoperatively suggesting that the ankle joint is not smooth (sagittal view) control. The purposes of surgical treatment are first, to further confirm the diagnosis, especially for early joint TB; and second, the removal of necrotic tissue, fibrous protein, and hyperplasia of synovial tissue, which would be conducive to reducing the inflammatory response [bib_ref] The results of nonoperative treatment of craniovertebral junction tuberculosis: a review of..., Arora [/bib_ref] [bib_ref] A posterior versus anterior debridement in combination with bone graft and internal..., Huang [/bib_ref] [bib_ref] Peripheral osteoarticular tuberculosis in children: 106 case-reports, Teklali [/bib_ref]. When the TB gradually stabilizes, the debridement of hypertrophic scarring tissue could further improve the ankle function. There are only a small number of followup cases reported for the surgical technique of ankle TB [bib_ref] Osteoarticular tuberculosis of the foot and ankle, Dhillon [/bib_ref] [bib_ref] A 49-year-old man with fever, erythema nodosum, and ankle swelling. Final diagnosis:..., Truong [/bib_ref] [bib_ref] A case of co-existing pigmented villonodular synovitis and tuberculosis infection of the..., Lui [/bib_ref] [bib_ref] Ankle tuberculosis: a report of four cases in a Japanese hospital, Inoue [/bib_ref] [bib_ref] Surgical treatment of lumbosacral tuberculosis by one-stage debridement and anterior instrumentation with..., Li [/bib_ref] [bib_ref] Surgical treatment of thoracic spinal tuberculosis with adjacent segments lesion via one-stage..., Wu [/bib_ref] [bib_ref] A new classification and guide for surgical treatment of spinal tuberculosis, Oguz [/bib_ref] [bib_ref] Arthroscopically assisted ankle fusion in patients with end-stage tuberculosis, Tang [/bib_ref] [bib_ref] Role of surgery in management of osteoarticular tuberculosis of the foot and..., Dhillon [/bib_ref]. Due to the lack of large sample studies, it is difficult to decide the optimal surgical approach. Traditionally, the surgical treatment for advanced active joint TB includes debridement, arthrodesis, and arthroplasty combined with a certain period of anti-TB therapy [bib_ref] A posterior versus anterior debridement in combination with bone graft and internal..., Huang [/bib_ref]. Of course, intraoperative exploration of articular cartilage damage is the most reliable evidence for determining which surgical procedure to perform. Small area of cartilage damage, especially when the residual cartilage is stable, such as in our case 1, could consider using the arthroscopic debridement to retain ankle function; results show that the patient's prognosis is satisfactory. In large areas of cartilage damage, such as in our case 3, arthroscopic debridement and effective anti-TB drugs will be enough for children to achieve satisfactory prognosis; ankle arthrodesis is not necessary.
Ankle TB is very rare and published data of arthroscopic management for it is rather limited. This study also has some limitations, such as small sample size and relatively short follow-up period. At the same time, due to limited number of cases, it is difficult to conduct comparative studies between arthroscopy and pure drug therapy. Thus, the therapeutic value of arthroscopy is also difficult to determine. Future multi-center controlled studies are expected to address these limitations.
# Conclusion
Arthroscopic management for ankle TB is minimally invasive, safe, and reliable. It can easily obtain samples for further confirmation, and the removal of lesions may be an adjunct to local disease control. Arthroscopy can also help evaluating cartilage damage, providing prognostic information of ankle joint function. During the surgery, if the area of articular cartilage damage is less than 2 cm 2 and the residual cartilage is stable, it is often effective to perform debridement rather than arthrodesis.
[fig] Figure 2 a: Preoperative right ankle swelling. b Preoperative radiograph suggesting no bone destruction in the ankle. c Ankle MRI before rheumatoid arthritis treatment, suggesting ankle effusion and mild bone edema of the talus. d Re-examination of MRI after rheumatoid arthritis treatment, suggesting increased edema of the talus and formation of joint cavity lesions. e Small area of cartilage damage under arthroscope. f Right ankle swelling improved 6 weeks postoperatively [/fig]
[table] Table 1: General data of patients [/table]
[table] Table 2: Comparison of results before treatment and at the last follow-up VAS score AOFAS score ESR (mm/h) CRP (mg/L) [/table]
|
Influence of adiposity and physical activity on the cardiometabolic association pattern of lipoprotein subclasses to aerobic fitness in prepubertal children
Aerobic fitness (AF) and lipoprotein subclasses associate to each other and to cardiovascular health. Adiposity and physical activity (PA) influence the association pattern of AF to lipoproteins almost inversely making it difficult to assess their independent and joint influence on the association pattern. This study, including 841 children (50% boys) 10.2 ± 0.3 years old with BMI 18.0 ± 3.0 kg/m2 from rural Western Norway, aimed at examining the association pattern of AF to the lipoprotein subclasses and to estimate the independent and joint influence of PA and adiposity on this pattern. We used multivariate analysis to determine the association pattern of a profile of 26 lipoprotein features to AF with and without adjustment for three measures of adiposity and a high-resolution PA descriptor of 23 intensity intervals derived from accelerometry. For data not adjusted for adiposity or PA, we observed a cardioprotective lipoprotein pattern associating to AF. This pattern withstood adjustment for PA, but the strength of association to AF was reduced by 58%, while adjustment for adiposity weakened the association of AF to the lipoproteins by 85% and with strongest changes in the associations to a cardioprotective high-density lipoprotein subclass pattern. When adjusted for both adiposity and PA, the cardioprotective lipoprotein pattern still associated to AF, but the strength of association was reduced by 90%. Our results imply that the (negative) influence of adiposity on the cardioprotective association pattern of lipoproteins to AF is considerably stronger than the (positive) contribution of PA to this pattern. However, our analysis shows that PA contributes also indirectly through a strong inverse association to adiposity. The trial was registered 7 May, 2014 in clinicaltrials.gov with trial reg. no.: NCT02132494 and the URL is https://clinicaltrials.gov/ct2/results?term= NCT02132494&cntry=NO.
[formula] a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 [/formula]
Introduction Aerobic fitness (AF) is a strong predictor of cardiometabolic health in adults. High concentration of small low-density lipoprotein (LDL) particles in adults correlates to cardiovascular disease (CVD). A pattern of high concentration of lipoprotein triglycerides (TG), very-lowdensity lipoproteins (VLDL) and large VLDL particles, and low concentration of high-density lipoproteins (HDL), large HDL and large LDL particles, and large average size of VLDL particles and low average size of HDL and LDL particle size correlates to insulin resistance. This condition may ultimately translate into to type 2 diabetes mellitus and progress into CVD.
Associations between AF and cardiovascular risk factors are also observed for childrenand unfavorable associations between adiposity, lipoproteins and cardiovascular risk factors have been verified in studies of children and adolescents. Thus, Suriano et al.found waist circumference to be a predictor of cardiovascular risk in children, while Slyper et al.found that the body mass index (BMI), and several lipoprotein features, correlated with thickening of the intima-media of the carotid artery in a cohort of adolescents. Cardiovascular status as expressed by AF and serum lipoprotein patterns in early years proceeds into adulthood with consequences later in life. It is therefore important to study such associations for children and adolescents. In a previous study, we therefore investigated the association between AF and the lipoprotein pattern with BMI as covariate in 94 prepubertal children. AF correlated positively to a cardioprotective lipoprotein pattern of high concentration of HDL, large and very large HDL particles and average size of HDL particles and to low concentrations of TG, chylomicrons (CM), VLDL, large and medium size VLDL particles, and average size of VLDL particles. BMI had a strong negative association to this pattern implying that control for adiposity is crucial for assessing the independent association pattern of lipoproteins to AF. In a recent investigationof associations of adiposity, physical activity (PA) and lipoprotein subclasses for a cohort of 841 children, we found a similar pattern as for AF. The pattern withstood adjustment for adiposity.
Ekelund et al.observed an independent association of PA and AF to metabolic risk factors in 1709 European children and PA/exercise intervention studies on adultshave shown strong impact of PA on the lipoprotein pattern. Considering those studies, our previous investigation on the association of AF to lipoprotein subclasseswas constrained by not including PA. In intervention studies for adults, increased PA correlated partly to the cardioprotective lipoprotein pattern to AF observed in our study for children, but PA in adults also correlated to reduced concentrations of LDL and the small atherogenic LDL particles and increased concentration of large LDL particles leading to increased average size of LDL particles. The same pattern was observed for physical active versus inactive adults by Kujala et al.and the pattern withstood adjustment for adiposity. A meta-analysis including 10 intervention studies on adultsconfirmed the strong favorable impact of PA on the lipoprotein subclass pattern.
In this work, our aim is to extract the association pattern of AF to lipoproteins in children and to examine the independent and joint influence of adiposity and PA on the association pattern. The same age group as in our previous studyis investigated, but for much larger cohort. The relative influence of adiposity and PA on the association of AF to the lipoproteins is assessed through variance explained in AF and lipoproteins. Our aims present some challenges as adiposity and PA have virtually opposite association pattern to the lipoproteins, and the lipoprotein features are strongly multicollinear, as are the covariates adiposity and PA. The high-resolution PA descriptor used in this work even possessed linear dependency which requires a novel approach to adjustment to achieve the objectives of our study.
# Materials and methods
## Study and participants
1129 5th graders (94% of those invited) from 57 schools in Western Norway participated in this study. Of these, 841 children provided valid baseline data on all relevant variables and were included in the present analysis.
Our procedures and methods conform to ethical guidelines defined by the World Medical Association's Declaration of Helsinki and its subsequent revisions. The South-East Regional Committee for Medical Research Ethics in Norway approved the study protocol (reference number 2013/1893). Prior to all testing, we obtained written informed consent from each child's parents or legal guardian and from the responsible school authorities. The study is registered in Clinicaltrials.gov with identification number: NCT02132494.
## Aerobic fitness test
The Andersen aerobic fitness test, which is a proxy for AF, but less influenced by adiposity than VO2 peak in children, was executed according to the standard procedure. The test measured the total distance covered during a 10-minutes run. Children ran from one endline to another (20 m apart) in a to-and-fro movement intermittently, with 15-second work periods and 15-second breaks.
## Lipoprotein subclasses
Overnight fasting serum samples were obtained and stored at -80˚C according to a standardized protocoland shipped on dry ice to the laboratories doing the analyses.
Serumexcept that we have combined their three subclasses of very small LDL particles and their two subclasses of very small HDL particles. Following the terminology of Ozaki et al., the abbreviations VL, L, M, S and VS imply very large, large, medium, small, and very small particles. Some of the VL subclasses are divided further in accordance with the classification by Okazaki et al. We calculated triglyceride and cholesterol separately and independently for all subclasses using the approach described in the next paragraph but combined them into one subclass representing the total concentration for each subclass of lipoproteins.
The 26 lipoprotein measures were obtained from a partial-least-squares (PLS) regression modelobtained by calibrating proton nuclear magnetic resonance (NMR) spectra to results obtained from high performance liquid chromatography (HPLC). 106 serum samples were used in the calibration. Repeated Monte Carlo resampling was used to optimize the models with respect to predictive performance. The HPLC analyses of the 106 calibration samples were performed by Skylight Biotech (Akita, Japan) as described by Okazaki et al.. Proton NMR was performed at the MR core facility (NTNU, Trondheim) by a standard procedureusing a Bruker Avance III 600 MHz spectrometer, equipped with a QCI Cryo-Probe and an automated sample changer (SampleJet) (Bruker BioSpin GmbH, Karlsruhe, Germany.
## Physical activity descriptor
Raw PA data was obtained using the ActiGraph GT3X+ accelerometerworn at the waist over seven consecutive days, except during water activities (swimming, showering) or while sleeping. Units were initialized at a sampling rate of 30 Hz. Files were analyzed at 1 second epochs using the KineSoft analytical software version 3.3.80 (KineSoft, Loughborough, UK). The use of 1 second epochs were found to be optimal for studying associations between metabolic and PA variables for this cohort. Data were restricted to hours 06:00 to 23:59. In all analyses, consecutive periods of � 60 minutes of zero counts were defined as non-wear time. We applied wear time requirements of � 8 hours/day and � 4 days/week to constitute a valid measurement. We defined a PA descriptor by creating 23 PA variables of total time (min/day) obtained from the vertical axis to capture movement in narrow intensity intervals throughout the spectrum, from 0-99 to � 10000 counts per minute (cpm). This descriptor covers the entire intensity spectrum. The intervals used for the descriptor were 0-99, 100-249, 250-499, 500-999, 1000-1499, 1500-1999, 2000-2499, 2500-2999, 3000-3499, 3500-3999, 4000-4499, 4500-4999, 5000-5499, 5500-5999, 6000-6499, 6500-6999, 7000-7499, 7500-7999, 8000-8499, 8500-8999, 9000-9499, 9500-9999 and � 10000 cpm.
## Adiposity measures
We calculated three measures of adiposity: BMI (kg/m 2 ), waist to height ratio (WC/H), and skinfold thickness. BMI was calculated as mass divided by the squared height. Body mass was measured with an electronic scale (Seca 899, SECA GmbH, Hamburg, Germany). Height was measured with a transportable stadiometer (Seca 217, SECA GmbH, Hamburg, Germany). Waist circumference (WC) was measured twice between the lowest rib and the iliac crest with the child's abdomen relaxed at the end of a gentle expiration using an ergonomic measuring tape (Seca 201, SECA GmbH, Hamburg, Germany). If the difference between measurements was >1 cm, a third measurement was taken. The average of the two closest measurements was used for analyses. Skinfold thickness was measured at the left side of the body using a Harpenden skinfold caliper (Bull: British Indicators Ltd., West Sussex, UK). Two measurements were taken at each position (biceps, triceps, subscapular, and suprailiac). If the difference between measurements was >2 mm, a third measurement was obtained. The total sum of the average of the two closest measurements for each site was used for analysis.
## Transformations and pretreatment of variables
It is not a necessary assumption that the variables are normally distributed, but our method produces more stable models if the variables are approximately normally distributed. All variables, except age, the binary variable for sex, and the Andersen aerobic fitness test (which was approximately normally distributed), were thus log-transformed. Thereafter they were meancentered and standardized to unit variance prior to the statistical analysis. The preprocessed data prior to centering and standardization to unit variance, are provided as supplementary information (S1 . After log transformation, normal probability plots showed that only CM, VLDL and a few of their subclasses in addition to TG still deviated from normal distribution.
## Data sets
Four different data set were created by using a projection approachto adjust all variables for covariates.
Data set 1: Variables adjusted for age and sex (S2 .
Data set 2: Variables adjusted for age, sex, and PA (S3 . Data set 3: Variables adjusted for age, sex, and adiposity (S4 . Data set 4: Variables adjusted for age, sex, PA, and adiposity (S5 . All variables, outcome, covariates, and explanatory variables were adjusted simultaneously and jointly by successive orthogonal projections. Age and sex had only weak correlations to the covariates adiposity and PA and were adjusted for without requiring any special action, but for the linear dependent PA descriptor and the nearly dependent adiposity descriptor of BMI, W/H, and skinfold, we had to follow a different path. The PA descriptor of the 23 intensity variables was orthogonalized using principal component analysis (PCA). Monte Carlo resampling with 100 repetitions leaving out randomly and predicting 25% of the data in each run showed that four principal components, explaining jointly 92.8% of the total variance of the PA variables, contained all the predictive information about PA. We used these four orthogonal PCs to adjust for PA by successive orthogonal projections. The same procedure applied for the three adiposity measures showed that only the first PC, accounting for 88.1% of the total variance in the adiposity variables, contained predictive information and was used to adjust for adiposity.
## Regression models
We used multivariate pattern analysiswith AF as outcome and the lipoproteins as explanatory variables for modelling of data set 1-4. The procedure uses PLS regressionwith the number of PLS components determined by a significance test based on 1000 models calculated by repeated Monte Carlo resampling. Post-processing of the PLS models with target projection (TP)provides a single predictive vector for the lipoproteins quantifying the associations to the predicted AF. For model interpretation and visualization of association patterns, we used selectivity ratio (SR) plots.
# Results
## Descriptive statistics for the variables
The 841 children (50% boys) were (mean ± standard deviation) 10.2 ± 0.3 years old. The result for the Andersen test was 898 ± 102 m, for BMI 18.0 ± 3.0 kg/m2, for WC/H 0.43 ± 0.05, and for skinfold thickness 49.8 ± 26.4 mm. S6 Table provides mean and standard deviation for a range of common anthropometric and blood variables for this cohort. Mean and standard deviation for the 23 variables defining the PA descriptor is provided in S7 Table. S1 File in the present article contains mean and standard deviation for concentrations of lipoproteins and average particle size for VLDL, LDL and HDL.
## Correlations between variables
Correlations coefficients between the variables after adjustments for age and sex are tabulated in S8 Table and briefly summarized in S1 File. The correlation patterns of AF and the adiposity measures to the lipoprotein profile and PA are almost opposite, while AF and high-intensity PA shares a similar association pattern to lipoproteins.shows the association patterns of AF to lipoproteins for the four models.
## Models
Heights of bars show quantitatively the associations of the lipoprotein features to the predicted AF. Bars above and below the vertical line at zero implies positive and negative associations to AF, respectively.summarizes features of data and models for the association between AF and the lipoproteins.shows that adjustment has a larger influence on AF (column 1) than on the lipoproteins (column 2). For instance, adjustment for age and sex removes 8.9% of the variance in AF,
# Discussion
## Plos one
Influence of adiposity and physical activity on the cardiometabolic association pattern but only 1.4% for lipoproteins. Boys were more physical active than girls in our cohort so adjustment for sex explains the reduction in variance in AF. The minor reduction of lipoprotein variance is expected as prepubertal boys and girls have similar lipoprotein profiles. Adjustment for PA or adiposity reduces the variance in AF profoundly and of the same magnitude, but reduction in variance for lipoproteins is twice as large for adjustment of adiposity compared to PA implying a stronger confounding influence of adiposity than PA on the lipoprotein association pattern to AF. The reduction in explained variance in AF after adjustments, column 3) confirms this expectation. Furthermore, adjustment for PA after adjustment for adiposity leads only to a minor additional reduction in explained variance for AF. This observation further implies a correlation between PA and adiposity so that adjustment for adiposity removes variance in PA and vice versa. The strength of this association increases with increasing intensity of PA and peaks with a flat maximum around 7000-7500 counts/minute (S1 File). This pattern resembles the association pattern of PA to other metabolic health variables, but the maximum correlation is shifted towards higher PA intensity. Explained variance in lipoproteins, column 4) relating to AF remains relatively robust to adjustments. The SR plot of the model adjusted for age and sexreveals a positive association of AF to a cardioprotective lipoprotein pattern: Negative associations to concentrations of TG, CM, VLDL, and all their subclasses except VLDL-S, and to LDL and all LDL subclasses except LDL-L, and the small and very small HDL particles, and positive associations to HDL, HDL-VL2, and HDL-L. Association is also negative to average size of VLDL particles and positive to the average size of HDL particles. This pattern withstand adjustment for PA, but adjustment for adiposity weakens the association of AF to the cardiometabolic favorable patterns of HDL particles. Further weakening is observed by adjustment for both adiposity and PA. However, all associations of AF to lipoprotein features found significant before adjustments, are still significant after adjustment as inferred from the 95% limits confidence bands in the SR plot.
The association of AF to the cardioprotective lipoprotein pattern resisted adjustment by both PA and adiposity, but the associations to total HDL and the very large and large HDL subclasses were further weakened. Adiposity appears more important than PA for the association pattern of AF to the lipoproteins. This is consistent with previous investigations. Thus, in a cohort of 262 children (9-10 years old), Hager et al.observed decrease in TG, TC and LDL and increase in HDL with increase in AF, but strong association of the lipoproteins to body fat led to the conclusion that "the goal of favorably altering blood lipids in children should begin with increasing PA and fitness, which in turn will lead to reductions in body fat". This recommendation was supported by Slyper et al.. In a cohort of 61 obese non-diabetic adolescents, they observed a lipoprotein pattern comparable to ours associating to BMI and concluded that "focus of CVD prevention in the adolescent obese should be visceral obesity, and not blood lipids or lipid subclasses." Gutin et al.examined relationships between AF, body fat, and lipoproteins, the latter being expressed as an atherogenic index, in a cohort of 57 children. They observed an inverse relationship between AF and CV risk but increased risk with body fat. In a cohort of 590 children, Hurtig-Wennlöf et al.found positive association of AF to HDL and negative associations to TC and TG and stronger association of AF than PA to CV risk factors. Our finding that the association of AF to the cardiometabolic lipoprotein pattern resists adjustment for PA supports their observation. In a cohort of 1826 adolescents, Bell et al.found similar associations to PA for a comprehensive lipoprotein profile as we observed to PAfor the same cohort as analyzed in the present work and for AF for the smaller cohort of 94 childrenimplying a similar association patterns for AF and PA. However, the strong negative association of LDL and all subclasses of LDL to AF found in the present investigation was not observed for PA in Bell et al.or in Rajalahti et al.. Ekelund et al.found independent effects from AF and PA on metabolic risk factors in children implying that AF associates to lipoproteins partly independently of PA. Several investigations on adultshave found similar association of PA to lipoproteins as we did for AF in the children. Hence, the observed association pattern with AF partly mirrors the correlation between PA and AF. Thus, we observed a decrease in the total LDL concentration and the small and very small LDL particles with increased AF in line with association pattern between PA and lipoproteins in adult populations. However, also concentration of medium and large LDL particles decreases with increase in AF rendering average LDL particle size uncorrelated to AF in our study.
# Strengths and weaknesses
Studies of associations between AF and lipoproteins in children have mostly been limited to the standard lipid panel of TC, LDL and HDL cholesterol, and TG. Resolution into subclasses discriminates between small and large LDL and HDL particles, which associate inversely to cardiometabolic health, allowing an understanding of how AF impacts cardiovascular health through its association to lipoprotein pattern.
Cholesterol levels peak in prepubertal children at approximately 10 years age and then drop during puberty before rising again during adulthood. For children, it is therefore beneficial to constrain such studies to a narrow age range.
Our analytical approach is adapted to handle multicollinear data and enables adjustment for confounders with linear dependency. This provides net association patterns and quantify the influence of confounders on the strength of the patterns which represent a challenge for use of molecular metabolomics descriptors in PA/exercise studies.
Because our analyses were restricted to cross-sectional associations, a limitation is that we cannot infer causality from our findings. Furthermore, our cohort embraces only Norwegian children. This limits the generalization of our study since there are differences in lipoprotein levels between different ethnic groups that may impact on the association to AF. However, in addition to studies discussed above, Okuma et al.observed the same inverse association of adiposity to the cardioprotective subclass pattern of HDL in Japanese schoolchildren as observed in this study. So, the association patterns between lipoprotein and AF and its relation to PA and adiposity extend beyond the ethnic group in our study. Our study lacks information about diet which impacts the lipoprotein distribution. This is a limitation of the study design.
# Conclusion
Since cardiometabolic risk factors carry over from childhood to adulthood, it is crucial to understand the complex relationships between AF, lipoproteins, adiposity, and PA in children. Our study shows that AF associates positively to a cardioprotective lipoprotein pattern but that the strength of this association is strongly influenced by PA and adiposity. PA associates positively to both AF and this pattern, while adiposity associates almost inversely and stronger than PA to this pattern. However, since PA and adiposity are inversely associated, adjustment by adiposity also removes variance of PA shared with AF. Thus, adiposity and PA in childhood influences the cardioprotective lipoprotein pattern directly, but also indirectly through the inverse relation of PA to adiposity. Although adiposity has a stronger independent association than PA to cardiometabolic health, the indirect influence of PA through the inverse relationship to adiposity must be taken into consideration when assessing the relative importance of these two factors on cardiometabolic health. Physical activity has an additional positive effect by preventing increase in adiposity and thus strengthening the cardiometabolic healthy lipoprotein pattern.
Supporting information S1 |
NEPA, a new fixed combination of netupitant and palonosetron, is a cost-effective intervention for the prevention of chemotherapy-induced nausea and vomiting in the UK
Background:The objective was to evaluate the costeffectiveness of NEPA, an oral fixed combination netupitant (NETU, 300 mg) and palonosetron (PA, 0.5 mg) compared with aprepitant and palonosetron (APPA) or palonosetron (PA) alone, to prevent chemotherapy-induced nausea and vomiting (CINV) in patients undergoing treatment with highly or moderately emetogenic chemotherapy (HEC or MEC) in the UK.Scope: A systematic literature review and meta-analysis were undertaken to compare NEPA with currently recommended anti-emetics. Relative effectiveness was estimated over the acute (day 1) and overall treatment (days 1-5) phases, taking complete response (CR, no emesis and no rescue medication) and complete protection (CP, CR and no more than mild nausea [VAS scale <25 mm]) as primary efficacy outcomes. A three-health-state Markov cohort model, including CP, CR and incomplete response (no CR) for HEC and MEC, was constructed. A five-day time horizon and UK NHS perspective were adopted. Transition probabilities were obtained by combining the response rates of CR and CP from NEPA trials and odds ratios from the meta-analysis. Utilities of 0.90, 0.70 and 0.24 were defined for CP, CR and incomplete response, respectively. Costs included medications and management of CINV-related events and were obtained from the British National Formulary and NHS Reference Costs. The expected budgetary impact of NEPA was also evaluated.Findings:In HEC patients, the NEPA strategy was more effective than APPA (quality-adjusted life days [QALDs] of 4.263 versus 4.053; incremental emesis-free and CINV-free days of +0.354 and +0.237, respectively) and was less costly (£80 versus £124), resulting in NEPA being the dominant strategy. In MEC patients, NEPA was cost effective, cumulating in an estimated 0.182 extra QALDs at an incremental cost of £6.65 compared with PA.Conclusion:Despite study limitations (study setting, time horizon, utility measure), the results suggest NEPA is cost effective for preventing CINV associated with HEC and MEC in the UK.
# Introduction
Chemotherapy-induced nausea and vomiting (CINV) is among the most common and feared side effects reported by cancer patients and may appear prior to, during or after chemotherapy administration [bib_ref] International antiemetic guidelines on chemotherapy induced nausea and vomiting (CINV): content and..., Jordan [/bib_ref]. The development of CINV is a complex process involving the activation of several neurotransmitters, which interact with the central nervous system stimulating a physiological response [bib_ref] SEOM clinical guidelines for the treatment of antiemetic prophylaxis in cancer patients..., Gomez [/bib_ref]. The incidence of CINV ranges from less than 10% in patients treated with chemotherapy agents with minimal emetic risk, from 30% to 90% in patients with agents with medium risk, to >90% among those whose regimens contain an agent with a high emetic risk [bib_ref] International antiemetic guidelines on chemotherapy induced nausea and vomiting (CINV): content and..., Jordan [/bib_ref].
Nausea and emesis, particularly when occurring during the delayed phase (days 2-5 of the chemotherapy cycle), cause significant problems for patients with cancer. CINV can lead to severe clinical conditions such as electrolyte imbalance and dehydration, and can affect a patient's quality of life and willingness to continue chemotherapy [bib_ref] Comparative review of 5-HT3 receptor antagonists in the treatment of acute chemotherapy-induced..., Hesketh [/bib_ref]. CINV also exerts a considerable financial burden on the healthcare system. Direct medical costs associated with managing CINV are driven by the cost of rescue medications, unscheduled physician visits, emergency room visits and hospitalisations [bib_ref] The cost of chemotherapy-induced nausea and vomiting in Italy, Ballatori [/bib_ref] [bib_ref] Resource utilization and costs associated with chemotherapyinduced nausea and vomiting (CINV) following..., Burke [/bib_ref] [bib_ref] Clinical and economic burden of chemotherapyinduced nausea and vomiting among patients with..., Craver [/bib_ref] [bib_ref] The cost of antiemetic therapy for chemotherapy-induced nausea and vomiting in patients..., Hamada [/bib_ref] [bib_ref] The impact of delayed chemotherapy-induced nausea and vomiting on patients, health resource..., Ihbe-Heffinger [/bib_ref] [bib_ref] An assessment of chemotherapy-induced nausea and vomiting direct costs in three EU..., Turini [/bib_ref].
Several anti-emetic therapies are used to prevent CINV. The Multinational Association of Supportive Care in Cancer (MASCC)/European Society for Medical Oncology (ESMO), American Society of Clinical Oncology (ASCO) and National Comprehensive Cancer Network (NCCN) guidelines are used to direct clinical practice and treatment in the UK. It has been suggested that with the correct use of anti-emetics, CINV could be prevented in 70%-80% of patients; however, suboptimal CINV control remains an issue in clinical practice. Many patients receiving highly or moderately emetogenic chemotherapy (HEC or MEC) still experience acute and delayed nausea, vomiting or both. In particular, current treatments provide insufficient CINV control during the delayed phase [bib_ref] Guidelines for antiemetic treatment of chemotherapy-induced nausea and vomiting: past, present, and..., Jordan [/bib_ref].
While guidelines for preventing CINV are widely available, clinical uptake of guidelines remains low [bib_ref] A prospective observational study of chemotherapy-related nausea and vomiting in routine practice..., Molassiotis [/bib_ref]. The reasons for non-adherence to clinical guidelines are multifactorial. A key aspect involves individual clinicians disagreeing with the key recommendations of the guidelines [bib_ref] Obstacles to the implementation of antiemetic guidelines, Grunberg [/bib_ref] [bib_ref] Antiemetic guidelines: are they being used?, Kaiser [/bib_ref]. Non-adherence may also reflect economic constraints of hospitals and government payers considering the higher costs of branded anti-emetic therapies.
NEPA is indicated in Europe for the prevention of the acute and delayed nausea and vomiting associated with HEC and MEC.
## Citation
Cawston H, Bourhis F, Eriksson J, Ruffo P, D'Agostino P, Two trials have been conducted examining NEPA, including one phase II dose ranging study, one phase III efficacy and one phase III safety trial.
The primary objective of this study was to assess, from the United Kingdom (UK) payer perspective, the cost-effectiveness of NEPA compared to the extemporaneous combination aprepitant plus palonosetron for patients receiving HEC and to palonosetron alone for patients receiving MEC.
# Methods
## Systematic review
A systematic literature review was undertaken to identify all relevant clinical trials examining the efficacy of NEPA and its comparators in the prevention of CINV for patients undergoing HEC or MEC as treatment for cancer. The systematic literature review focused on English-language publications with the methodological approach in line with established systematic review procedures.
The scope of the literature review was defined in terms of Population, Interventions, Comparators, Outcomes and Study design (PICOS). Eligible studies included human adult (≥18 years) cancer patients receiving highly or moderately emetogenic chemotherapy that assessed the efficacy or safety of one of the anti-emetics (e.g. 5 HT-3s, NK1s, other) by comparing the intervention with either placebo or an active comparator. Studies were required to contain at least complete response and other outcomes like complete protection, partial response, complete control, total control, time to first emetic episode, time to use of rescue medication or time to treatment failure, and to be blinded, randomised controlled trials (≥ Phase II) with more than 50 patients (Appendix 1). Studies were excluded if they were duplicate, animal or in vitro studies, narrative reviews, editorials, case reports or letters, meta-analyses, were not in English language or if they had the wrong scope, population, intervention, or did not contain the correct study type or outcome of interest. Any disagreements between authors were resolved by a third party.
The electronic search was performed in MEDLINE, EMBASE, the Cochrane Collaboration, and HTA in Cochrane. The search was conducted with limits of 'English language' with the date limit from 2000 to 28/08/2013 for all searches except olanzapine, which was included in an updated search which ranged up to 09/01/2014. Although olanzapine was not recommended in ASCO or MASCC as a preferred preventative treatment for CINV, it was later included in the search because of its inclusion in the NCCN guidelines at the time of the updated search. A data table was produced, with key information extracted from the eligible papers including study authors, date, title, study design, treatment strategies, patient characteristics, chemotherapy received, results/outcomes and adverse events. In addition, the quality assessment and evaluation of each study was undertaken using the National Institute for Health and Clinical Excellence (NICE) checklist for randomised controlled trials.
# Meta-analysis
Following the systematic literature review and a feasibility assessment, which included an evaluation of available evidence and a heterogeneity assessment, it was considered feasible to undertake an indirect comparison within a frequentist framework in the MEC indication and a Bayesian Mixed Treatment Comparison (MTC) in the HEC and AC-MEC indication.
Three efficacy outcomes were analysed: complete response (CR), complete protection (CP) and total control (TC). CR is the efficacy outcome most commonly reported in CINV treatment studies, defined as no nausea, no vomiting, and no use of rescue medication (with studies not defining CR in this way excluded). The definition of CP was no emetic episodes, no use of rescue medication, and no more than mild nausea (defined as Visual Analogue Scale, VAS <25 mm). A subgroup of patients who achieve CR also achieve TC, defined as no emesis, no use of rescue medication, and no nausea (<5 mm on the VAS). For each outcome, where data were available, results were presented for two different time periods: the acute phase (day 1) and the overall phase (days 1-5).
In the MEC population, an indirect treatment comparison of CR was performed using fixed effect models, as the inclusion of random effects was considered not appropriate given the low number of studies available and that CR was the only outcome sufficiently reported. MTC models were performed in the HEC population using both fixed and random effects where the model best fitting the data was selected based on the deviance information criterion (DIC) [bib_ref] Meta-analysis in clinical trials, Der Simonian [/bib_ref] [bib_ref] Evidence synthesis for decision making in healthcare, Welton [/bib_ref]. The MTC allowed the comparative effect size of treatments on a specific outcome scale to be estimated, as well as the treatment-specific probability of being the best for some characteristics. The odds ratio was presented as the primary efficacy outcome measure for each analysis. The uncertainty associated with the measure has been reported via the 95% confidence interval (CI) for the indirect comparison in the MEC population, and the credible intervals (CrI) for the MTC analyses in the HEC population.
## Economic evaluation
Cost-utility Markov models were developed to estimate the expected cost and health outcomes after HEC or MEC administration, to compare NEPA against aprepitant combined with palonosetron (APPA) in the HEC population and against palonosetron in the MEC population [fig_ref] Figure 1: A [/fig_ref].
The target patient population was cancer patients receiving prophylactic anti-emetics for the management of HEC or MEC. The selection of relevant comparators to NEPA was based on international clinical guidelines. The perspective was that of the UK National Health Service (NHS). The time horizon for the model was five days that consist of the first day (acute phase) and from the first to fifth day (overall phase) and was run for one cycle of chemotherapy [bib_ref] Guidelines for antiemetic treatment of chemotherapy-induced nausea and vomiting: past, present, and..., Jordan [/bib_ref].
All patients entered the Markov model on the first day after the chemotherapy administration, for HEC and MEC regimens, respectively. In the acute phase (day 1), patients had different probabilities of achieving complete response and incomplete response, depending on the efficacy of the administered antiemetic, obtained from the meta-analysis. Complete response was divided into two subcategories: complete protection and complete response at best (complete response without complete protection). The same approach was taken to assess the patient distribution in each state at the end of day 5 (i.e. response rate in overall phase). To determine the number of patients in each state between the second and the fourth day, the patient flow on days 2-4 was calibrated using linear interpolation between both the acute and overall phase-response rates.
The response rates were obtained from NEPA trials [fig_ref] Table 1: Response rates of NEPA [/fig_ref]. The efficacies of the comparators were obtained by combining the response rates of NEPA and the ORs calculated in the metaanalysis as detailed in the equation below:
[formula] Response rate Response Rate OR (1 Response Rate ) Response Rate T NEPA NEPAvsT N EPA N EPA = × − + [/formula]
where T is a treatment comparator of NEPA and OR NEPAvsT is the odds ratio of NEPA versus this treatment, as obtained from the meta-analysis.
The odds ratios of each considered treatment regimen compared to NEPA are presented in [fig_ref] Table 2: Estimated odds ratios of the comparators in HEC and MEC [/fig_ref] for the HEC and MEC populations. In previously published economic models for CINV, an anchor utility of 0.90 was used for chemotherapy without nausea and vomiting and a utility of 0.20 was used for the 'incomplete response' health state, in which emesis and/or nausea are present [bib_ref] Cost-effectiveness analysis of aprepitant in the prevention of chemotherapy-induced nausea and vomiting..., Annemans [/bib_ref] [bib_ref] Health outcomes and cost-effectiveness of aprepitant in outpatients receiving antiemetic prophylaxis for..., Lordick [/bib_ref]. A value of 0.24 for the incomplete response state was adjusted based on clinical expert feedback. Given that 'chemotherapy without nausea and vomiting' anchor value is 0.90, the utility for the mild nausea state from Borjeson et al. (0.752) was assumed to be 0.70 for the complete response at best [bib_ref] Similarities and differences in assessing nausea on a verbal category scale and..., Borjeson [/bib_ref]. The utility value for total control was assumed to be 0.95 in order to linearise between perfect health (1.00) and complete protection (0.90). Finally, utilities of 0.90, 0.70 and 0.24 were defined for CP, CR and incomplete response, respectively [bib_ref] Cost-effectiveness analysis of aprepitant in the prevention of chemotherapy-induced nausea and vomiting..., Annemans [/bib_ref] [bib_ref] Health outcomes and cost-effectiveness of aprepitant in outpatients receiving antiemetic prophylaxis for..., Lordick [/bib_ref].
Treatment costs of the prophylactic anti-emetic comparators were calculated based on recommended doses from international guidelines and local unit costs in the UK as found in the British National Formulary (BNF) 2014, the Personal Social Services Research Unit (PSSRU) 2013 and the NHS reference costs 2012-13. The costs of NEPA were calculated upon the price assumption £69 per package of 300 mg, as published by UK Monthly Index of Medical Specialities (MIMS). The cost per patient per cycle of health resources used due to emetic events were based on a study among cancer patients receiving a broad range of MEC in a trial setting in the UK [bib_ref] Cost-effectiveness of an aprepitant regimen for prevention of chemotherapy-induced nausea and vomiting..., Humphreys [/bib_ref]. The different costs used in the model are detailed in [fig_ref] Table 1: Response rates of NEPA [/fig_ref] REVIEW -Cost-effectiveness of netupitant and palonosetron in CINV drugsincontext.com
The primary outcome measure used in the analysis was the incremental cost-effectiveness ratio (ICER) defined as the ratio of the incremental difference in total cost to the incremental difference in benefits between two treatment strategies. It can be expressed as the cost per life-year gained or by quality-adjusted life years (QALYs) by multiplying the years of life with a weight, measured in utilities, reflecting quality of life.
Sensitivity analyses were conducted in order to assess the impact of parameter and structural uncertainty on results. The Deterministic Sensitivity Analysis (DSA) showed the extent to which the results are affected by sources or assumptions and continuous variables (e.g. efficacy, costs, and utilities) increasing and reducing by 25% of the values. In addition to the DSAs, a probabilistic sensitivity analysis (PSA) was conducted in the model. A PSA allows capturing interactions between several inputs by running the model a large number of times (typically at least 1000 times) as a Monte Carlo simulation, using randomly drawn sets of inputs. The possible inputs are defined by assigning a statistical distribution around the mean value of each uncertain parameter, reflecting the extent and nature of dispersion around the mean. For each run, values are sampled at random from each distribution, and the decision tree is rolled back using these values to obtain a (costeffectiveness) pair.
# Results
## Systematic review and meta-analysis
The search of randomised trials examining anti-emetics in adult cancer patients returned a total of 2060 results across the four screened databases. After the removal of duplicates, 1749 abstracts were reviewed and of these 208 full-text articles were evaluated. Titles, abstracts and full texts were screened by two independent reviewers. Discrepancies were resolved by a third reviewer. A total of 37 studies were deemed eligible for inclusion. Details on counts and reasons for exclusion are reported in [fig_ref] Figure 2: Literature review flow chart [/fig_ref]. In MEC patients, although the cost per CINV episode was less costly in the NEPA arm, the incremental drug cost resulted in NEPA being more costly overall. Compared to palonosetron (PA), the incremental cost of NEPA was £6.65 while the QALD gain was 0.18 days (=0.0005 QALYs), which gives an ICER of approximately £13,318. Given a threshold of WTP of £30,000, the NB of NEPA against palonosetron was £8 with an acceptability of 88%.
The results of the meta-analysis were directly used as efficacy inputs for the comparators. They are summarised in Appendixes 2 and 3 for HEC and MEC respectively.
## Cost-effectiveness
In the base case analysis, NEPA was evaluated against the relevant comparators. The results of the analyses are presented in [fig_ref] Table 4: Base case analysis of NEPA compared to other treatments in HEC and... [/fig_ref] for HEC and MEC.
In HEC patients, given the price assumption for NEPA, the treatment-acquisition costs and the cost per CINV episode management were lower in the NEPA arm. The quality-adjusted In the one-way sensitivity analysis, the effect of positively and negatively varying the continuous input variables by 25% and selecting alternative sources and assumptions were explored. The results of the sensitivity analyses are presented in a tornado diagram in [fig_ref] Figure 3: Tornado diagram of one-way sensitivity analysis in HEC and MEC [/fig_ref]. The considered parameter variations resulted in NEPA always being the dominant strategy with negative incremental costs and positive incremental QALYs.
A probabilistic sensitivity analysis (PSA) sampling 1000 simulations was conducted. The results of the PSA are presented as scatterplots of incremental effects and costs [fig_ref] Figure 4: Scatterplot of incremental effects and costs in HEC and MEC [/fig_ref] for the HEC and MEC populations. In HEC patients, NEPA was a dominant strategy in 89.2% of simulations against APPA and was cost saving but less effective in 10.4% of cases. Similarly, NEPA was a dominant strategy in 80.4% of simulations against palonosetron and cost saving but less effective in 0.1% of cases in the MEC population.
# Discussion
The primary objective of the analysis was to evaluate the cost-effectiveness of NEPA for the treatment of CINV in terms of the UK payer perspective. The evaluation was implemented by emetogenic level of chemotherapy, that is, HEC and MEC indications. The most relevant comparator to NEPA was APPA for HEC and PA for MEC.
The cost-effectiveness model was developed within a Markov cohort structure, which is well validated and has been used in published CINV models [bib_ref] Cost-effectiveness analysis of aprepitant in the prevention of chemotherapy-induced nausea and vomiting..., Annemans [/bib_ref]. The single-cycle (5-day) structure allows the model to focus on CINV-related consequences and to prevent the outcomes from being influenced by disease characteristics and other toxicities of chemotherapy than emetogenicity. Thus, the results from the analysis can be applicable to CINV management regardless of cancer type, cancer stage and toxicity of chemotherapy except emetogenicity.
Clinical trials have demonstrated that palonosetron is effective for the treatment of CINV as a single agent, especially following cisplatin-based chemotherapy [bib_ref] Improved prevention of moderately emetogenic chemotherapy-induced nausea and vomiting with palonosetron, a..., Eisenberg [/bib_ref] [bib_ref] A systematic review and meta-analysis of intravenous palonosetron in the prevention of..., Likun [/bib_ref] [bib_ref] A phase III, double-blind, randomized trial of palonosetron compared with ondansetron in..., Aapro [/bib_ref]. Palonosetron is therefore now the gold standard for treating CINV, either as a single agent in MEC or in combination with an aprepitantbased regimen in HEC. Aprepitant-based regimens have been widely used in combination with 5-HT3 RAs, such as palonosetron, ondansetron or granisetron, for the treatment of CINV in the HEC population and are recommended by evidencebased, European and US consensus guidelines.
This economic evaluation shows that netupitant forms a better combination with palonosetron compared with aprepitant, in terms of both cost saving and efficacy in the HEC indication. Netupitant plus palonosetron is also more effective in terms of emesis days avoided and increased health-related quality of life, which is an increasingly important outcome for medical interventions, particularly in the area of supportive care. In the MEC indication when NEPA is compared against palonosetron, the current standard of care, the ICER of £13,318 is below the accepted threshold for being considered cost effective in the UK (£20,000-30,000 per QALY gained). In the UK, the use of NEPA for the treatment of CINV in both MEC and HEC can therefore be considered a cost-effective treatment option. The strength of these results were supported by the consistency of the results across all outcomes and populations in the model and in the sensitivity analyses. Several studies have demonstrated the need for therapies that are effective in preventing both acute and delayed CINV [bib_ref] Patient perceptions of the side-effects of chemotherapy: the influence of 5HT3 antagonists, De Boer-Dennert [/bib_ref]. CINV control in clinical practice remains an issue, especially in the delayed phase; delayed CINV occurred in 58.4% of patients receiving anti-emetic therapy and nausea affected 60.7% of patients treated with current anti-emetics.
Owing to the dual component of NEPA in addressing both acute and delayed CINV, this economic evaluation demonstrates that guideline-adherent prophylaxis of CINV may be regarded as a cost-effective investment of resources. Together with the results of clinical trials that demonstrate the efficacy and safety of NEPA, these cost-effectiveness results demonstrate the clinical utility of this combination therapy [bib_ref] A randomized phase III study evaluating the efficacy and safety of NEPA,..., Aapro [/bib_ref] [bib_ref] A phase III study evaluating the safety and efficacy of NEPA, a..., Gralla [/bib_ref] [bib_ref] Efficacy and safety of NEPA, an oral combination of netupitant and palonosetron,..., Hesketh [/bib_ref]. These findings may therefore enhance clinicians' individual acceptance of and adherence to guideline recommendations, thereby leading to the optimisation of CINV treatment strategy within clinical practice.
The analysis had the following limitations. First, the results may vary based on different chemotherapy cycles, since the efficacy of treatments may differ. Second, the analysis was based on a five-day time horizon. Any cost-effectiveness, healthcare consumption, and quality-of-life data, which were acquired beyond the model's five-day time horizon were, therefore, not evaluated. Third, the generic utility measure used may not be very sensitive to CINV-specific states and therefore might not have comprehensively captured the differences between treatment strategies. Fourth, the results may have to be interpreted with caution when the model is applied to a specific group of patients, for example, young and/or female cancer patients, for whom different response rates may be observed compared to the general cancer population.
Further studies are required to evaluate the use of netupitant in combination with palonosetron in the UK for preventing CINV associated with both HEC and MEC over multiple cycles. Ideally, the allocation of resources should be informed by a comprehensive assessment of all relevant direct healthcare costs and indirect costs. To further assess the impact of CINV from a societal perspective, it would be of interest to also incorporate indirect costs, especially costs associated with missed work days for patients and their caregivers, as well as healthcare consumption data that were captured after the fiveday time horizon used in this economic model.
In conclusion, this economic evaluation demonstrates that NEPA is a dominant (in HEC) and cost-effective (in MEC) treatment alternative to current anti-emetic standards of care in the UK during the first five days of chemotherapy treatment in cancer patients.
# Contributions:
[fig] Figure 1: A [/fig]
[fig] Figure 2: Literature review flow chart. [/fig]
[fig] Figure 3: Tornado diagram of one-way sensitivity analysis in HEC and MEC. comparator, complete response in over all phase, MEC Efficacy data for NEPA, MEC NETU 0818: complete response in over all phase NETU 0818: complete response in acute phase Utility value for complete protection Cost of in patient care To include Total Control Utility value for incompelete response Source of healthcare utilisation due to CINV OR against comparator, complete protection in overall phase, chemotherapy induced nausea and vomiting; HEC: highly emetogenic chemotherapy; ICER: incremental cost-effectiveness ratio; MEC: moderately emetogenic chemotherapy; NEPA: netupitant+palonosetron+dexamethasone; OR: odds ratio; PO: per os (by mouth). ISSN: [/fig]
[fig] Figure 4: Scatterplot of incremental effects and costs in HEC and MEC. -e ectiveness Scatter Plot, MEC Base Case HEC: highly emetogenic chemotherapy; MEC: moderately emetogenic chemotherapy; QALY: quality-adjusted life-years. ISSN: [/fig]
[table] Table 1: Response rates of NEPA (95% CI). [/table]
[table] Table 2: Estimated odds ratios of the comparators in HEC and MEC. [/table]
[table] Table 3: Cost of treatment and healthcare resource use (£, 2013). [/table]
[table] Table 4: Base case analysis of NEPA compared to other treatments in HEC and MEC. [/table]
|
Fiber burden and asbestos-related diseases: an umbrella review
Objective: What are the levels of asbestos exposure that cause each type of health effect? The objective of this study was to review the available scientific evidence on exposure levels for asbestos and their relationship to health effects.Method: An umbrella review of English-language reviews and meta-analyses, from 1980 to March 2021 was conducted. We included reviews involving quantified asbestos exposures and health outcomes. The review has been adapted to the indications of the PRISMA declaration. Methodological quality of the selected studies was assessed using the AMSTAR instrument.Results:We retrieved 196 references. After applying the search strategy and quality analysis, 10 reviews were selected for in-depth analysis. For lung cancer, the highest risk was observed with exposure to amphiboles. Longer, thinner fibers had the greatest capacity to cause lung cancer, especially those > 10 μm in length. For mesothelioma, longer and thinner fibers wereThis is an open access article under the CC BY license Prohibir toda exposición al asbesto es la mejor medida para prevenir sus efectos negativos para la salud. Las revisiones y metaanálisis de más alta calidad respaldan que hay escaso riesgo de cáncer de pulmón y de mesotelioma con niveles de exposición diaria por debajo de 0,1 f/ml.
also more pathogenic; amphiboles ≥ 5 μm are especially associated with increased mesothelioma risk. No studies observed an increased risk for lung cancer or mesothelioma at asbestos exposure levels < 0.1 f/ml. No reviews provided information on exposure concentrations for pulmonary fibrosis. Currently, there is limited evidence in humans to establish the causal relationship between gastrointestinal cancer and asbestos exposure.
Conclusions: Banning all asbestos exposure remains the best measure to preventing its negative health effects. The highest quality reviews and meta-analyses support that there is little risk of lung cancer or mesothelioma at daily exposure levels below 0.1 f/ml.
## Resumen
Revisar la evidencia científica disponible sobre los niveles de exposición al asbesto y su relación con los efectos sobre la salud.
Se realizó una revisión de revisiones sistemáticas y metaanálisis en inglés, desde 1980 hasta marzo de 2021. Se incluyeron revisiones que involucran exposiciones cuantificadas al asbesto y resultados de salud. La revisión se adaptó a las indicaciones de la Declaración PRISMA. La calidad metodológica de los estudios seleccionados fue evaluada mediante el instrumento AMSTAR.
Se recuperaron 196 referencias y tras aplicar la estrategia de búsqueda y analizar la calidad se seleccionaron 10 revisiones para un análisis en profundidad. Para el cáncer de pulmón, se observó mayor riesgo con la exposición a anfíboles. Las fibras más largas y delgadas presentaron mayor capacidad de causar cáncer de pulmón, especialmente aquellas de longitud > 10 μm. Para el mesotelioma, las fibras más largas y delgadas también fueron más patógenas; los anfíboles ≥ 5 μm se asociaron con un mayor riesgo de mesotelioma. Ningún estudio observó mayor riesgo de cáncer de pulmón o de mesotelioma con niveles de exposición al asbesto < 0,1 f/ml. Ningún estudio proporcionó información sobre concentraciones de exposición para la fibrosis pulmonar. Actualmente existe evidencia limitada en humanos para establecer la relación causal entre la exposición al asbesto y el cáncer gastrointestinal.
# Background
Asbestos is the generic name for a group of fibrous silicates present in nature, used in industrial processes and in the production of several products for domestic and environmental use. [bib_ref] Exposición al amianto y sus efectos sobre la salud, Agudo [/bib_ref] [bib_ref] Enfermedades respiratorias ocupacionales y medioambientales Fundamentos para su investigación clínico-epidemiológica, Rego Fernández [/bib_ref] [bib_ref] Cohorte poblacional de trabajadores expuestos a amianto Navarra, Artieda [/bib_ref] According to the chemical composition and physical properties, asbestos fibers are classified as serpentine (chrysotile or white asbestos), and amphibole comprising crocidolite (blue asbestos), amosite (brown asbestos), anthophyllite (yellow asbestos), tremolite (gray asbestos), and actinolite. [bib_ref] Enfermedades respiratorias ocupacionales y medioambientales Fundamentos para su investigación clínico-epidemiológica, Rego Fernández [/bib_ref] It is well known that exposure to asbestos fibers can lead to diseases such as asbestosis (or diffuse interstitial pulmonary fibrosis), diffuse pleural fibrosis, rounded atelectasis (Blesovski's syndrome), malignant mesothelioma and also ovary, lung and laryngeal cancer. Currently, there is limited evidence in humans to establish the causal relationship between gastrointestinal cancer and asbestos exposure. [bib_ref] Exposición al amianto y sus efectos sobre la salud, Agudo [/bib_ref] [bib_ref] Asbestos-associated diseases: a review, Delclos [/bib_ref] [bib_ref] Identificación de los riesgos laborales asociados a enfermedad sospechosa de posible origen..., Delclos [/bib_ref] The health hazard of asbestos depends on several factors such as its concentration in the air, the exposure time, the type and the size of the fibers, the respiratory rate associated with physical exertion and thermo-hygrometric conditions, the anatomical and functional conditions of exposed workers and there exists a tobacco smoke modulating effect.The unit measures used for asbestos exposure may vary. The ACGIH establishes threshold limit values in occupational health.On the one hand, fibers per cubic centimeter (f/cc or f/ cm 3 ), equivalent to fibers per milliliter (f/ml), is commonly used to measure exposure during a working day (about 8 hours of duration) or during short-term exposures. These measures are compared with the permissible exposure limits (PEL, OSHA), the threshold limit values (TLV, ACGIH), the recommended exposure limits (REL, NIOSH) or the indicative occupational exposure limit values (IOELV, EU). On the other hand, fibers per milliliter and year (f/ml-y) is an exposure measure that represents the cumulative exposure to asbestos over a working life and is the measure that best reflects exposure intensity (exposure intensity equals the average concentration of asbestos in air multiplied by the duration of the exposure). Expressions such as fibers/milliliter-year (f/ml-y), fibers-year/milliliter (f-y/ml), fibers/cubic centimeter-year (f/cc-y) and fibers-year/cubic centimeter (f-y/cc) can be found in the literature. In order to analyze the effects of cancer, pulmonary fibrosis, mesothelioma and other pathologies, epidemiological studies tend to use this measure of cumulative exposure. Also, exposure units in millions of particles per cubic foot per year (mppcf-y) have been used.
The OSHA has set a PEL for asbestos at 0.1 fiber per cubic centimeter of air as an eighthour time-weighted average (TWA), with an excursion limit (EL) of 1.0 asbestos fibers per cubic centimeter over a 30-minute period.Likewise, in Europe the 2009/148/EC Directive of the European Parliament and of the Council establishes the airborne concentration of asbestos in excess of 0.1 fibers per cubic centimeter as an 8-hour time-weighted average (TWA).For several years, this exposure limit has been adopted, with some variations, by several European countries. 12,13 debate continues on the existence of a minimum level of intensity of exposure to asbestos below which exposure is safe and above which there is a likelihood of developing health damage. For that reason, continually reviewing and updating information on the healtheffects exposure levels of asbestos is important. The aim of this study is to contribute to this debate, reviewing the best and the most recent scientific evidence available in the international literature on exposure levels for asbestos fibers and their relationship with established asbestos health effects.
# Method
An umbrella review 14 was performed, retrieving systematic reviews and meta-analyses available in MEDLINE/PubMed, Google Scholar Academics bibliographic repository, the International Agency for Research on Cancer (IARC),Several search strategies were tested using the usual connectors for keywords (AND, OR, NOT, etc.), obtaining final search engines to capture as many references as possible, filtered by "review AND systematic review AND meta-analysis". The search was restricted to articles published in English from 1980 to March 2021. We included both human and experimental studies in animals.
An initial search strategy was carried out with a wide perspective and using search engines to capture all articles that analyzed the association between asbestos exposure and any known asbestos disease, including lung cancer, mesothelioma, asbestosis, diffuse interstitial pulmonary fibrosis, pleural plaques, diffuse pleural fibrosis, larynx and gastrointestinal cancer. After-wards, different key words were incorporated to the search engines to restrict them to those articles that also incorporated variables that quantified exposure (fiber concentration, dose-exposure, exposure-response or dose-response). The syntaxes used in this review are shown in Supplemental text (syntaxes) in online Appendix.
Each of these syntaxes was applied independently and duplicate studies were eliminated. Titles and abstracts were screened using independent peer-review. A third expert resolved discrepancies and independently decided final inclusion for full text analysis. Only systematic reviews and meta-analyses evaluating exposure to asbestos fibers and its relationship to health effects were included in this umbrella review.
A first selection was based on reading the title, including those with the words "asbestos" and "pathology", excluding those that were not of interest for the purpose of the study or were doubtful. The second selection was based on reading the abstracts, excluding those studies that did not analyze associations between asbestos exposure and the selected pathologies, or were not systematic reviews or meta-analyses. In a third phase, full texts were screened and articles that did not refer to the purpose of the study were excluded.
In a final stage, an evaluation of the methodological quality of the selected studies was performed by means of the AMSTAR (Assessment of Multiple SysTemAtic Reviews) instrument. The instrument is a reliable and valid measure for the evaluation of the methodological quality of systematic reviews and has proven good face and content validity. To report the results of this review, the evidence-based set of items for reporting systematic reviews stated by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) have been followed. This study has been registered at PROSPERO (https://www.crd.york.ac.uk/prospero/ display_record.php?RecordID=185349).
# Results
## Selection of articles
Initially, 207 references were retrieved (170 from PubMed/MEDLINE and 37 from international agencies), and after removing duplicates (n = 122), there were 85 articles left. Of these, 24 articles were excluded after reading the title and 32 after reading the abstract, obtaining 29 articles. After reading the full text of these studies, 13 were excluded because they were not systematic reviews or meta-analysis and six because they did not have measures of exposure-effect association, obtaining a total of 10 articles for in-depth analysis [fig_ref] Figure 1: Figure 1. [/fig_ref] and tables II, III and IV in online Appendix).
After evaluating their methodological quality with the AMSTAR instrument, two of them obtained a global score of 10, one a score of 8 and the rest of the studies obtained global scores varying from four to six (table I in online Appendix). However, none of these last 10 articles was excluded, because they provided valuable information .
## Characteristics of the studies
The main characteristics of the selected articles are listed in . Articles included 22-31 were published between 1997 and 2017, but most (80%) were issued from 2008 onward. Most of the studies (n = 7) were meta-analyses published by researchers from institutions mainly in the U.S., Canada and the Netherlands. 22-25,28,29,31 One of the reviews 31 incorporates a multicenter case-control study 32 with participation from several countries. The reviews and meta-analyses included cohort and case-control studies of reasonably good or very good quality. One of the reviews included is based on experimental animal studies. Nine systematic reviews/meta-analyses assessed the association between asbestos and lung cancer, 22-30 six between asbestos and pleural mesothelioma, 23,25-27,30,31 one between asbestos and pulmonary fibrosis, 30 and one between asbestos and other gastrointestinal cancers. 27
## Association between asbestos exposure and lung cancer by fiber type
There was not a clear pattern in the association measures used to analyze the effects, which included the standardized mortality ratio (SMR), relative risk (RR), odds ratio (OR) and percentage of expected mortality per fiber/ml-year of exposure (RL) . In general, we observed that the effect of chrysotile on lung cancer is characterized by being weaker than the effect of amphibole fibers or mixed fibers. 22-29 Specifically, the highest risk for lung cancer was observed with exposure to amphibole fibers, followed by mixed-fibers and finally chrysotile. The length and durability of the fibers was associated with the carcinogenicity potency. Fibers with a length > 10 μm are more carcinogenic than those < 10 μm in length, and chrysotile has been found to be less carcinogenic than amphibole fibers. The thinner the fiber, the higher its capacity to cause lung cancer, as the fiber can better penetrate the lung tissue. In terms of exposure time, only one study 23 showed that exposure to 2 or more years to amosite fibers had a higher SMR (SMR = 11.7) than exposure to less than one month (SMR = 2.64). However, one of the studies 26 concluded that the highest cumulative exposure to chrysotile without an effect of lung cancer is 25 f/ml-y, which means that at this concentration of chrysotile fibers, lung cancer has not yet been observed.
## Association between asbestos exposure and pleural mesothelioma by fiber type
In general, we observed some inconsistency in the results from the different studies regarding exposure-response . In one study, 23 the proportionality in the expected mortality risk for mesothelioma was 1:100:500 due to exposures to chrysotile, amosite and crocidolite, respectively. In the same study, the dose-response correlation for amphibole fibers suggested a nonlinear relationship for pleural mesothelioma, and that short-term exposures to high fiber concentrations were at higher risk than exposure to low fiber concentrations with long exposure time.
According to the highest cumulative exposure level at which no effect was observed (NOAEL), results evaluating the exposure-response among the studies were also inconsistent. Two cohorts included in the review by Pierce et al. did not observe increased risk with the highest cumulative exposures: NOAEL > 400 and ≥ 112 f/cc-y (latency 20 years); and two cohorts included in the same review 26 observed a NOAEL risk for mesothelioma at 800-1599 f/cc-y and < 15 f/cc-y.
One of the studies 26 concluded that the highest level of cumulative exposure to chrysotile without the effect of mesothelioma is 15 f/ml-y, meaning that at this concentration of chrysotile fibers no risk of mesothelioma was observed. In addition, the meta-analysis with the highest methodological quality 31 concludes that little risk of mesothelioma would be observed with asbestos exposure below 0.1 f/ml.
It is noteworthy that in general, longer and thinner fibers are more pathogenic than short ones, in particular amphiboles ≥5 μm have been associated with increased mesothelioma RR and OR. 25,30 One study examined non-occupational exposures. That study showed that amphibole fibers were those with the highest capacity to produce mesothelioma since exposure to amphibole fibers showed a meta-RR 2.5-3.2 times greater than exposure to mixed fibers, and a meta-RR 5.3-5.6 times greater than exposure to pure chrysotile fibers. shows the main results of the association between asbestos exposure and the development of pulmonary fibrosis and other types of cancer by type of asbestos fiber. One of the studies, 30 which reviewed experimental animal articles, concluded that the length and durability of the fibers is the factor most associated with the potency of carcinogenicity. Long fibers (>5 μm) are associated with pulmonary fibrosis (asbestosis) and cancer (mesothelioma and lung cancer), with no evidence of pathogenicity for pulmonary fibrosis when exposed to fibers with a length ≤5 μm.
## Association between asbestos exposure and pulmonary fibrosis by fiber type
## Association between asbestos exposure and other cancers by fiber type
The study by Gamble 27 found that the association between gastrointestinal cancer and asbestos exposure does not exist or it is very low (RR, OR, and SMR around 1). The only association with gastric cancer was observed with very high exposures (>140 f/ml-y) to chrysotile (SMR = 1.43) and to amosite (SMR = 3.21) at exposures > 1000 mppcf-y). The same would be true for colon cancer, with an association found at exposures > 140 f/ml-y (SMR = 2.31). The studies included in this review do not provide information about exposure levels of asbestos and other cancers, such as larynx or ovarian cancers.
## Association between asbestos exposure and other asbestos related pathologies
Information on exposure levels of asbestos associated with other well-established pathologies (such as pleural plaques, pleural thickening, diffuse pleural fibrosis, effusion and rounded atelectasis) were outside the scope of this review and were not included.
# Discussion
The first finding is that all asbestos fibers have been associated consistently with lung cancer, mesothelioma and pulmonary fibrosis. In relation to laryngeal and ovarian cancer the causality of asbestos has also been demonstrated, but we did not find any systematic reviews or meta-analyses on these two pathologies to be able to draw conclusions about exposure levels. The evidence is, to date, less conclusive for gastrointestinal tumors. 27 A second finding is that the risk, varies depending on the type of asbestos, the physicochemical characteristics of these fibers, the intensity of exposure and, for some pathologies, co-exposures with other carcinogens, especially tobacco. The studies clearly point that the greatest risk exists with exposure to amphibole fibers, followed by mixed fibers (amphibole and chrysotile), and finally, chrysotile. However, when analyzing the risk that occurs depending on the intensity of asbestos exposure, some studies suggest that for mesothelioma and lung cancer there may be a threshold below which there is no significantly increased risk of suffering the pathology, whereas other studies cannot conclude the existence of a threshold. What seems quite evident is that the exposure intensity required to produce mesothelioma may be lower than for lung cancer or pulmonary fibrosis (asbestosis).
The joint assessment of the analyzed systematic reviews and meta-analyses leads us to the conclusion that there is little risk of lung cancer or mesothelioma at daily exposure levels below 0.1 f/ml (daily environmental exposure limit value). However, most times the measures are reconstructions made decades after the exposure has taken place, which can have a large margin of error when applied to individual patients. [bib_ref] Relation between lung asbestos fibre burden and exposure indices based on job..., Takahashi [/bib_ref] Our findings could be explained at least in part by the carcinogenetic inflammatory mechanisms of asbestos. As a recent review has shown, 34 asbestos and other fibers remain in the affected tissue for months to years, triggering a chronic inflammatory process and consequent release of high mobility group protein B1 and other cytokines that maintain this process which may ensue in cancer. This pathogenic mechanism could support the fact that low amounts of fibers, probably below 0.1 f/ml, are less likely to elicit a chronic inflammatory process. Other naturally occurring fibers are present in the environment and some of them (e.g. erionite) are as or more carcinogenic than asbestos. [bib_ref] Mesothelioma: scientific clues for prevention, diagnosis, and therapy, Carbone [/bib_ref] [bib_ref] Erionite exposure in North Dakota and Turkish villages with mesothelioma, Carbone [/bib_ref] In any case, the most recent cutting-edge research shows that "safe" levels could vary greatly from one individual to another depending on the genetics of each individual. In this regard, for example, heritable mutations of the germline BAP1 and other tumor suppressor genes have been reported to increase susceptibility to asbestos carcinogenesis. It is estimated that 12% of mesotheliomas occur in carriers of these mutations. [bib_ref] BAP1 and cancer, Carbone [/bib_ref] [bib_ref] Tumour predisposition and cancer syndromes as models to study gene-environment interactions, Carbone [/bib_ref] The occupational exposure limit values proposed by internationally renowned agencies have been drastically reduced over time. The ACGIH has adopted the current TLV of 0.1 f/ml for all types of asbestos. [bib_ref] Exposure to asbestos: past, present and future, Pira [/bib_ref] The limit values proposed by other agencies such as OSHA,NIOSH,CCOHS, 40 HSE 41 and UE-OSHA 42 currently set an exposure limit value for eight-hour time-weighted average of 0.1 f/ml (with certain particularities). However, as both the ILO and the WHO have stated, there is no safe level of exposure to a carcinogen and, in concordance with them, we advocate for the global ban and eradication of all types of asbestos and demand the complete elimination of asbestos-related diseases as a global public health priority.This review, analyzed and synthesized the main results of 10 systematic reviews and available meta-analyses of the last 40 years that have examined asbestos exposure values that pose a threat to human health. Likewise, the fact that the first article included in the review was published in 1997 and the last in 2017 ensures that our review of reviews included all the reviews and meta-analyses published over the last 40 years.
None of the selected reviews scored low in the assessment of their methodological quality; some of them obtained intermediate quality scores (4 ≥ P ≤ 5) [22-27,30] and 50% of the reviews obtained middle 25,26 or high quality scores (P ≥ 8), 28,29,31 supporting a reasonably good quality of the included studies.
As any scientific study this one may have some limitations. It is possible in this type of studies, based on systematic search for scientific publications, to overlook unpublished studies (publication bias). Nevertheless, the fact that we included in our search those studies available on the websites of the leading institutions world-wide which have evaluated research on the health consequences of asbestos exposure, including IARC (WHO), NIOSH (USA), ACGIH (USA), ATSDR (USA), EPA (USA), IWH (Canada) and HSE (UK), it is unlikely that we have left out any major review or meta-analyses performed in recent years.
# Conclusion
In summary, several studies provide a value for asbestos exposure below which no risk of lung cancer or mesothelioma would be observed 26,28,31 and scientific evidence reported in the highest quality reviews and meta-analyses identified in this umbrella review (those with a 10 on the quality assessment), support that there is only limited evidence of the risk of lung cancer or mesothelioma at daily exposure levels below 0.1 f/ml (daily exposure environmental limit value). However, following the recommendations of the ILO and WHO, the best measure to prevent the negative health effects of asbestos exposure is its banning and eradication following strict cleaning protocols.What is known about the topic?
Although the exposure limit values for asbestos have been lowered over the years, coinciding with the advances in scientific evidence on the harmful effects of asbestos, widespread social debate continues on the existence of a minimum level of intensity of exposure to asbestos below which exposure is safe and above which there is a likelihood of developing health damage.
## What does this study add to the literature?
This study contributes to this debate, reviewing the best and the most recent scientific evidence available in the international literature on exposure levels for asbestos fibers and their relationship with established asbestos health effects.
## What are the implications of the results?
There is limited evidence of the risk of lung cancer or mesothelioma at daily exposure levels below 0.1 f/ml (VLA-ED). However, the best measure to prevent negative health effects of asbestos exposure is its banning and eradication following strict cleaning protocols. Article selection flowchart Main characteristics of the selected articles and exposure data .
Author, year (country) Author, year (country) Main results on lung cancer from the selected studies.
## Type of study n° of studies included in the analyses
## Type of study n° of studies included in the analyses
Author, year Cross-contamination with other fibers should be considered.
There was no increase in risk of lung cancer in the 4 highest exposure categories (≥112 fibers/cc-y), due to a possible confounding factor in smoking. In other studies, the risk increased from exposures >25-1600-3200 fibers/cc-y.
in Cross-contamination with other fibers should be considered. Main results on lung cancer according to fiber type. a Exposure values from which the risk will be evaluated. The minimum and maximum range, if any, are detailed. The unit fiber-year/milliliter (f-y/ml), fibers/milliliter-year (f/ml-y), fibers-year/cubic centimeter (f-y/cc) and fibers/cubic centimeter-year (f/cc-y) are equivalent; General: fibers without exact measurement of exposure. b R L : risk of lung cancer; excess risk of lung cancer risk per fiber/ml-year; equivalent to KL; RL = 100 (SMR-1)/Expos (f/ml-y). c Different RRs depending on the applied coefficient: following EPA models (RR = 1 + 0.01 cumulative exposure), Quebec mines (RR = 1 + 0.0006 accumulated exposure), industry (RR = 1 + 0.00025 accumulated exposure). The latter would be most appropriate for the indicated study. NOAEL: highest cumulative exposure level at which no effect was observed; RR: relative risk; OR: odds ratio; SMR: standardized mortality ratio. Some results have no confidence interval (95% CI). Main results on pulmonary fibrosis and other cancers from the selected studies, according to fiber type.
## Other observations and results
## Roggli v, 2015
Chrysotile <8 μm 23 million chrysotile fibers >5 μm in animals and 272 million ≤5 μm with 18-24 months of exposure Increased risk of alveolitis but not pulmonary fibrosis.
No increased risk of fibrosis, RR and OR not specified.
Long fibers (>5 μm) are associated with a higher risk of pulmonary fibrosis (asbestosis) than short ones.
There is no evidence of pathogenicity in ≤5 μm fibers.
Amphibole: crocidolite Intratracheal injection of short fibers Intratracheal injection of long fibers Increased risk of minimal lung injury or fibrosis.
Increased risk of severe fibrosis (similar to human asbestosis).
Other non-asbestiform long fibers have also been associated with fibrosis and tumors.
## Gamble j, 2008
Chrysotile <15 f-y/ml; >40 f-y/ml 10-49 f/ml-y; 100-356 f/ml-y <8.4 f/ml-y; >140 f/ml-y <300 mppcf-y; >1000 mppcf-y RR = 1.0; RR = 3.4 (1. The association between gastrointestinal cancer and exposure to asbestos is very weak.
Only association with gastric cancer was observed with very high exposures.
Possible cross-contamination with other fibers should be considered.
Amphibole:
amosite Mixed <6 f/ml-y; >250 f/ml-y <10 years exposed; >10 years exposed SMR = 1.66 (0.71-3.26); SMR = 1.96 (0.24-7.08) for gastrointestinal cancer SMR = 0.42; SMR = 0.45 for colon cancer. SMR = 0.52; SMR = 0.82 for rectal cancer. SMR = 1.50; SMR = 0.92 for gastric cancer.
a Exposure values from which the risk will be evaluated. The minimum and maximum range, if any, are detailed. The units fiber-year/milliliter (f-y/ml), fibers/milliliter-year (f/ml-y), fibers-year/cubic centimeter (f-y/cc) and fibers/cubic centimeter-year (f/cc-y) are equivalent.
b RR: relative risk; OR: odds ratio; SMR: standardized mortality ratio. Some results have no confidence interval (95% CI).
[fig] Figure 1: Figure 1. [/fig]
[table] , 44: Supplementary MaterialRefer to Web version on PubMed Central for supplementary material.12. Ministerio de Trabajo, Migraciones y Seguridad Social. Instituto Nacional de Seguridad y Salud en el Trabajo (INSST). Límites de exposición profesional para agentes químicos en España. 2019 [Accessed 2021 April 01]. Available at: https://cutt.ly/BrywB2I. 13. El Yamani M, Boulanger G, Nerrière-Catelinois E, et al. Revision of French occupational exposure limits of asbestos and recommendation of measurement method: can the dimensional characteristics of the asbestos fibers (long, thin, short) be taken into account? Critical Reviews in Environmental Science and Technology. 2012;42:1441-84. 14. Hunt H, Pollock A, Campbell P, et al. An introduction to overviews of reviews: planning a relevant research question and objective for an overview. Syst Rev. 2018;7:39. [PubMed: 29490699] 15. Agency for Toxic Substances and Disease Registry (ATSDR). Asbestos. [Updated 2011 March 3]. [Accessed 2021 April 01]. Available at: https://www.atsdr.cdc.gov/substances/toxsubstance.asp? toxid=4. 16. United Estates Environmental Protection Agency (EPA). Asbestos. [Updated 2019 April 16]. [Accessed 2021 April 01]. Available at: https://www.epa.gov/asbestos. 17. Health and Safety Executive (HSE). Asbestos. [Accessed 2021 April 01]. Available at: http:// www.hse.gov.uk/asbestos/information.htm. 18. Institut for Work and Health. Asbestos. [Updated 2019 December]. [Accessed 2021 April 01]. Available at: https://cutt.ly/LryrzTB. 19. Shea BJ, Grimshaw JM, Wells GA, et al. Development of AMSTAR: a measurement tool to assess the methodological quality of systematic reviews. BMC Med Res Methodol. 2007;7:10. [PubMed: 17302989] 20. Shea BJ, Hamel C, Wells GA, et al. AMSTAR is a reliable and valid measurement tool to assess the methodological quality of systematic reviews. J Clin Epidemiol. 2009;62:1013-20. [PubMed: 19230606] 21. Moher D, Liberati A, Tetzlaff J, et al. The PRISMA Group Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med. 2009;6:e1000097. [PubMed: 19621072] 22. Lash TL, Crouch EA, Green LC. A meta-analysis of the relation between cumulative exposure to asbestos and relative risk of lung cancer. Occup Environ Med. 1997;54:254-63. [PubMed: 9166131] 23. Hodgson J, Darnton A. The quantitative risks of mesothelioma and lung cancer in relation to asbestos exposure. Ann Occup Hyg. 2000;44:565-601. [PubMed: 11108782] 24. Berry G, Gibbs GW. An overview of the risk of lung cancer in relation to exposure to asbestos and of taconite miners. Regul Toxicol Pharmacol. 2008;52 1 Suppl:S218-22. [PubMed: 17998152] 25. Berman DW, Crump KS. A meta-analysis of asbestos-related cancer risk that addresses fiber size and mineral type. Crit Rev Toxicol. 2008;38 Suppl 1:49-73. [PubMed: 18686078] 26. Pierce JS, McKinley MA, Paustenbach DJ, et al. An evaluation of reported no-effect chrysotile asbestos exposures for lung cancer and mesothelioma. Crit Rev Toxicol. 2008;38:191-214. [PubMed: 18324516] 27. Gamble J Risk of gastrointestinal cancers from inhalation and ingestion of asbestos. Regul Toxicol Pharmacol. 2008;52 1 Suppl:S124-53. [PubMed: 18078700] 28. Lenters V, Vermeulen R, Dogger S, et al. A meta-analysis of asbestos and lung cancer: is better quality exposure assessment associated with steeper slopes of the exposure-response relationships? Environ Health Perspect. 2011;119:1547-55. [PubMed: 21708512] 29. van der Bij S, Koffijberg H, Lenters V, et al. Lung cancer risk at low cumulative asbestos exposure: meta-regression of the exposure-response relationship. Roggli VL. The so-called short-fiber controversy: literature review and critical analysis. Arch Pathol Lab Med. 2015;139:1052-7. [PubMed: 26230599] 31. Marsh GM, Riordan AS, Keeton KA, et al. Non-occupational exposure to asbestos and risk of pleural mesothelioma: review and meta-analysis. Magnani C, Agudo A, González CA, et al. Multicentric study on malignant pleural mesothelioma and non-occupational exposure to asbestos. Br J Cancer. 2000;83:104-11. [PubMed: 10883677] [/table]
|
Surface Enhanced Raman Spectroscopy Detection of Biomolecules Using EBL Fabricated Nanostructured Substrates
Fabrication and characterization of conjugate nano-biological systems interfacing metallic nanostructures on solid supports with immobilized biomolecules is reported. The entire sequence of relevant experimental steps is described, involving the fabrication of nanostructured substrates using electron beam lithography, immobilization of biomolecules on the substrates, and their characterization utilizing surface-enhanced Raman spectroscopy (SERS). Three different designs of nano-biological systems are employed, including protein A, glucose binding protein, and a dopamine binding DNA aptamer. In the latter two cases, the binding of respective ligands, D-glucose and dopamine, is also included. The three kinds of biomolecules are immobilized on nanostructured substrates by different methods, and the results of SERS imaging are reported. The capabilities of SERS to detect vibrational modes from surface-immobilized proteins, as well as to capture the protein-ligand and aptamer-ligand binding are demonstrated. The results also illustrate the influence of the surface nanostructure geometry, biomolecules immobilization strategy, Raman activity of the molecules and presence or absence of the ligand binding on the SERS spectra acquired.
# Introduction
Capabilities to develop and characterize conjugate nano-biological systems interfacing solid nanostructures and biological polymers are becoming increasingly important to further advances in next-generation bio-sensing and bio-actuation technologies [bib_ref] Analyzing Nanomaterial Bioconjugates: A Review of Current and Emerging Purification and Characterization..., Sapsford [/bib_ref] [bib_ref] Nanomaterials for Bio-Functionalized Electrodes: Recent Trends, Walcarius [/bib_ref]. This involves multidisciplinary studies across a number of research fields, such as the fabrication of pertinent solid-state components (micro-or nano-electrodes, nano-engineered coatings, nanowires, or nanoparticles) [bib_ref] Nanomaterials for Bio-Functionalized Electrodes: Recent Trends, Walcarius [/bib_ref] [bib_ref] Applications, Techniques, and Microfluidic Interfacing for, Kim [/bib_ref] [bib_ref] Lithography-Based Nanoelectrochemistry, Rassaei [/bib_ref] ; immobilization of biomolecules on the surfaces to create desired bioconjugates [bib_ref] Selective Covalent Protein Immobilization: Strategies and Applications, Wong [/bib_ref] [bib_ref] Immobilization of Histidine-Tagged Proteins on Electrodes, Ley [/bib_ref] [bib_ref] Protein Immobilization Techniques for Microfluidic Assays, Kim [/bib_ref] ; and monitoring nano-biological interfaces [bib_ref] Analyzing Nanomaterial Bioconjugates: A Review of Current and Emerging Purification and Characterization..., Sapsford [/bib_ref]. In most cases, the selection of optimal fabrication, bio-functionalization, and characterization methods is strongly inter-related. Clearly, the choice of nanofabrication techniques would be driven by the requirements of the solid state components of the system, being largely dependent on the detection method, which in turn is determined by the nature of the biopolymers involved and the purpose of monitoring the interface.
Out of a broad variety of techniques applied to characterize bioconjugate systems [bib_ref] Analyzing Nanomaterial Bioconjugates: A Review of Current and Emerging Purification and Characterization..., Sapsford [/bib_ref] [bib_ref] Applications, Techniques, and Microfluidic Interfacing for, Kim [/bib_ref] , surface enhanced Raman spectroscopy (SERS) has emerged as a highly promising method for the detection of chemical and biological species on surfaces [bib_ref] Biosensing with Plasmonic Nanosensors, Anker [/bib_ref] [bib_ref] Recent Progress in SERS Biosensing, Bantz [/bib_ref] [bib_ref] Creating, Characterizing, and Controlling Chemistry with SERS Hot Spots, Kleinman [/bib_ref]. SERS employs inelastic scattering of monochromatic light by surface-immobilized biomolecules [fig_ref] Figure 1: Scheme of surface-enhanced Raman spectroscopy [/fig_ref] allowing the capture of unique signatures corresponding to molecular vibrations. This capability to distinguish among different molecules without involving labels, complex chemistry, or time-consuming steps, makes SERS a potentially very efficient method of bio-detection. Another important advantage of SERS is its high sensitivity. The excitation of localized surface plasmons by light interacting with noble metal nanostructures (SERS substrates) increases dramatically the intensity of Raman scattering by the analyte, allowing the detection of very small amounts of molecules, from monolayers down to the singlemolecule limit [bib_ref] Biosensing with Plasmonic Nanosensors, Anker [/bib_ref] [bib_ref] Recent Progress in SERS Biosensing, Bantz [/bib_ref] [bib_ref] Creating, Characterizing, and Controlling Chemistry with SERS Hot Spots, Kleinman [/bib_ref]. Finally, most biomolecules require aqueous solutions to be stable. Because water often has limited Raman activity, background signal from aqueous samples is minimized [bib_ref] Recent Progress in SERS Biosensing, Bantz [/bib_ref]. Applications of SERS have exhibited an exponential increase over the last decade. However, a much discussed challenge of SERS is that the electromagnetic enhancement of Raman scattering depends critically on the size, shape, and spacing of metal nanostructures where plasmonic waves are induced Numerous methods employed to fabricate SERS substrates [bib_ref] Creating, Characterizing, and Controlling Chemistry with SERS Hot Spots, Kleinman [/bib_ref] [bib_ref] A Review on the Fabrication of Substrates for Surface Enhanced Raman Spectroscopy..., Fan [/bib_ref] [bib_ref] Engineering Metal Nanostructure for SERS Application, Cao [/bib_ref] can be roughly classified into bottom-up and top-down methods. Methods of the first type employ various processes of self-assembly or directed chemical synthesis to produce nanostructures. Often addressed examples include immobilization of monodisperse nanoparticles on solid supports [bib_ref] Creating, Characterizing, and Controlling Chemistry with SERS Hot Spots, Kleinman [/bib_ref] [bib_ref] A Review on the Fabrication of Substrates for Surface Enhanced Raman Spectroscopy..., Fan [/bib_ref] [bib_ref] Engineering Metal Nanostructure for SERS Application, Cao [/bib_ref] , thermal, sputter, or electrochemical deposition of roughened metal films [bib_ref] Creating, Characterizing, and Controlling Chemistry with SERS Hot Spots, Kleinman [/bib_ref] [bib_ref] A Review on the Fabrication of Substrates for Surface Enhanced Raman Spectroscopy..., Fan [/bib_ref] , and various chemical synthesis methods [bib_ref] Engineering Metal Nanostructure for SERS Application, Cao [/bib_ref]. Although such techniques tend to be relatively simple and inexpensive, most of them are challenged by a lack of control over the location of the structures, and limited sample-to-sample reproducibility.
In contrast, top-down lithography techniques employ manipulable instruments such as particle beams to create desired patterns on surfaces. One of the most often used nanolithography methods, electron beam lithography (EBL), offers superb control over features down to below 10 nm and also a flexibility to allow for different substrate designs on solid supports [bib_ref] Creating, Characterizing, and Controlling Chemistry with SERS Hot Spots, Kleinman [/bib_ref] [bib_ref] A Review on the Fabrication of Substrates for Surface Enhanced Raman Spectroscopy..., Fan [/bib_ref]. In EBL, a beam of electrons focused down to a spot of a few nanometers in diameter scans across a surface of an electron sensitive material (resist) causing a chemical change in exposed regions. For positive tone resists such as polymethylmethacrylate (PMMA), electron beam exposure results in scission of the polymer chains composing the resist, leading to an increased solubility in an appropriate solvent (developer). The process of electron-beam lithography includes spin-coating of a uniform layer of resist on a substrate; exposure of the targeted resist material in a vacuum chamber with an electron beam; and development of the sample to remove the soluble regions.
Dielectric supports underneath metallic nanostructures, such as fused silica, have been shown to significantly increase the intensities in SERS due to localization of plasmonic waves compared to other materials such as silicon [bib_ref] Dielectric-Substrate-Induced Surface-Enhanced Raman Scattering, Glembocki [/bib_ref] [bib_ref] Surface Enhanced Spectroscopy with Gold Nanostructures on Silicon and Glass Substrates, Merlen [/bib_ref]. However EBL patterning on dielectric substrates, especially at the nanoscale, involves significant challenges due to charge build-up during exposure. Previously, we have shown [bib_ref] Nanopatterning of PMMA on Insulating Surfaces with Various Anticharging Schemes Using 30..., Muhammad [/bib_ref] [bib_ref] Study of Multilayer Systems in Electron Beam Lithography, Peters [/bib_ref] that these difficulties can be overcome by placing conductive polymer layers above the resist. shows a schematic of the overall fabrication process using EBL exposure and development followed by metal deposition and liftoff to produce metallic nanostructures on fused silica supports. Please click here to view a larger version of this figure.
## Fabrication of nano-patterned pmma masks using electron beam lithography (ebl)
1. Spin-coat the PMMA resist and conductive layers on the substrates.
1. Use a wafer spinner with a vacuum chuck and place samples individually in the center on the chuck. Place 1 drop of polymethylmethacrylate (PMMA) resist on the center of the samples using a glass pipette and spin at 3,500 rpm for 60 sec with a 2 sec ramp time. 2. Bake the substrates at 180 °C for 3-5 min. After baking the substrates, cool the samples to RT. 3. With the substrates cooled and returned to the spinner chuck, spread a drop of conductive polymer on the substrate. Spin the substrate for 40 sec at 3,000 rpm with a 2 sec ramp time. Bake samples at 80 °C for 1 min.
2. Perform EBL exposure according to the standard procedure [bib_ref] Nanopatterning of PMMA on Insulating Surfaces with Various Anticharging Schemes Using 30..., Muhammad [/bib_ref] [bib_ref] Study of Multilayer Systems in Electron Beam Lithography, Peters [/bib_ref] [bib_ref] Application of EBL Fabricated Nanostrucutred Substrates for SERS Detection of Protein A..., Gutierrez-Rivera [/bib_ref]. 1. Using manufacturer's instructions, prepare an exposure design employing feature doses from with the smallest possible beam step size. 2. Load the sample into the electron beam lithography chamber. If the EBL system does not have autofocusing, use a small scratch away from where the pattern is to be exposed and away from the bead edge for focusing. 3. Using manufacturer's instructions, perform the required focusing and astigmatism correction as well as write field alignment as appropriate, and expose the sample. To allow for proper exposure profile and best pattern quality, use a 30 keV electron beam energy and 7.5 µm aperture for the exposures.
3. Remove the conductive polymer and develop exposed samples.
1. Load the samples upside down into the electron beam evaporator system to allow for the evaporated metal to be deposited on the front face of the samples. Deposit a 10 nm thick Au layer onto the samples for Designs 1 and 3, and a 10 nm thick Ag layer for Design 2 at a rate of approximately 0.1 nm/sec. 2. Fill a sonication system to the recommended height with water and fill a separate beaker with acetone. Place a sample face up in the bottom of the beaker and allow the sample to soak for 10 min. Holding the beaker, place it into the water bath and allow the height of the acetone to match the height of the water and turn on the sonication system. Allow sonication to occur for up to 60 sec. 3. Using the same procedure as detailed in step 3.1, prepare uniform Au and Ag pad substrates by deposition of 10 nm thick metal films on FS (Designs 1 and 3) and Ni-coated FS (Design 2) substrates skipping step 2. Incubate for 1 min to activate the self-assembled monolayer (SAM) of MUA. 5. Place a 100 µl drop of protein A solution (47 µM) on the same area of the substrate and store the sample for 24 hr at 5 °C in a multicompartment Petri dish with 1 ml of DI water in another compartment, and with a sealed cover. 6. Rinse the sample in DI water 3 times by continuously stirring the samples in separate beakers for 20 sec in each beaker. Do not let the samples dry after rinsing or during the rinse. 7. Proceed to step 5.
## Bio-functionalization of substrates
## Raman spectroscopy
1. Place each sample in a water-proof chamber to avoid evaporation by laser exposure.
1. Fill a plastic syringe with chemically inert high vacuum grease, place samples on glass slides and dispense a few millimeters of grease surrounding the samples without touching the samples. 2. Place a microscope coverslip on top of the substrates and gently press down to form a seal, creating a thin liquid interface between the substrates and the coverslips without allowing the buffer to come in contact with the vacuum grease.
# Representative results
Collecting control Raman spectra for the main components, including free proteins in solution and free ligands in solution or in powder form without using metal-containing substrates, is important to enable a proper comparison as well as for interpretation purposes. [fig_ref] Figure 5: Control Raman spectrum of protein A in DI water obtained at 532... [/fig_ref] presents a typical Raman spectrum for free protein A in DI water on a glass slide without nanostructured substrates. Two bands with the highest Raman intensity, the band at 2,931 cm -1 and at 1,091 cm , correspond to vibrations involving C-H and C-S bonds, respectively. Other bands with a lower Raman intensity such as 563 cm , can be attributed to a superposition of vibration modes [bib_ref] Application of EBL Fabricated Nanostrucutred Substrates for SERS Detection of Protein A..., Gutierrez-Rivera [/bib_ref] [bib_ref] Band Assignment and Qualitative Analysis of Biomedical Compounds with Cardiovascular Activity, Bright [/bib_ref] [bib_ref] Amide Modes and Protein Conformation, Bandekar [/bib_ref] [bib_ref] What Vibrations Tell About Proteins, Barth [/bib_ref]. Control Raman spectra for the ligand free GPB in buffer solution with three different concentrations, 0.3, 0.9 and 1.3 mM, are shown in [fig_ref] Figure 5: Control Raman spectrum of protein A in DI water obtained at 532... [/fig_ref]. In the figure, the broad band around 3,400 cm -1 corresponds to the solvent [bib_ref] Current Applications and Future Challenges, Chrimes [/bib_ref] , whereas the band at 2,935 cm -1
represents vibrations involving C-H bonds of the protein [bib_ref] Band Assignment and Qualitative Analysis of Biomedical Compounds with Cardiovascular Activity, Bright [/bib_ref] [bib_ref] Amide Modes and Protein Conformation, Bandekar [/bib_ref]. [fig_ref] Figure 5: Control Raman spectrum of protein A in DI water obtained at 532... [/fig_ref] shows the high wavelength Raman spectrum for D-glucose in buffer solution for different concentrations: 1, 6, 100, 200, and 400 mM. When the concentration of glucose is increased, C-H bonds vibration bands arise at 2,890 cm In order to obtain SERS spectra for surface-immobilized biomolecules, substrates comprising metallic nanostructures on fused silica supports are fabricated as described in steps 1-3. The quality of fabricated substrates is monitored using scanning electron microscopy (SEM). The standard SEM procedures are described elsewhere [bib_ref] Nanopatterning of PMMA on Insulating Surfaces with Various Anticharging Schemes Using 30..., Muhammad [/bib_ref] [bib_ref] Study of Multilayer Systems in Electron Beam Lithography, Peters [/bib_ref] [bib_ref] Application of EBL Fabricated Nanostrucutred Substrates for SERS Detection of Protein A..., Gutierrez-Rivera [/bib_ref] and not included in the present protocol. shows representative SEM
## Figure 7. scheme of water-proof chamber for raman imaging of bio-functionalized samples in solution.
In Design 1, recombinant protein A is immobilized on the substrates functionalized by a self-assembled monolayer of 11-mercaptodecanoic acid (MUA) in DI water [bib_ref] Application of EBL Fabricated Nanostrucutred Substrates for SERS Detection of Protein A..., Gutierrez-Rivera [/bib_ref]. The substrates in this design comprise three arrays of Au dots with a 50 nm pitch and varying inter-dot distances on fused silica (see . The process of protein immobilization starts with the formation of a SAM on the substrates. To obtain covalent binding between the SAM and the protein, the carboxylic acid groups of SAMs are transformed into amine reactive NHS-ester by treatment with a mixture of N-ethyl-N'- propyl) carbodiimide (EDC) solution and N-hydroxysuccinimide (NHS) solution in DI water. Immobilization of protein A occurs by displacement of the NHS group by lysine residues of the protein [bib_ref] Immobilization of Protein A on SAMs for the elaboration of immunosensors, Briand [/bib_ref]. An example of imaging samples for Design 1 with protein A immobilized on Au nanostructures is shown in [fig_ref] Figure 9: SERS imaging of substrate-immobilized protein A in Design 1 18 [/fig_ref]. [fig_ref] Figure 9: SERS imaging of substrate-immobilized protein A in Design 1 18 [/fig_ref] presents an optical microscope image of the samples, which comprise three arrays of bio-functionalized Au nanodots with different inter-dot gaps (see also , and [fig_ref] Figure 9: SERS imaging of substrate-immobilized protein A in Design 1 18 [/fig_ref] shows the Raman spectral mapping over these arrays. It can be seen that the highest Raman intensities are found for Array I where the inter-dot gaps are the narrowest, whereas lower intensities are obtained for Array III with the widest inter-dot gaps. This can be explained by a stronger plasmon coupling effect produced by higher electric fields in the narrow spaces between the dots [bib_ref] Application of EBL Fabricated Nanostrucutred Substrates for SERS Detection of Protein A..., Gutierrez-Rivera [/bib_ref]. [fig_ref] Figure 9: SERS imaging of substrate-immobilized protein A in Design 1 18 [/fig_ref] shows the strongest SERS spectra obtained for Arrays I and II. The spectra exhibit several bands (1,630 cm ) in proximity to the Raman modes of free protein A in solution seen in [fig_ref] Figure 5: Control Raman spectrum of protein A in DI water obtained at 532... [/fig_ref]. Attributable to vibrations of various bonds found in proteins, these bands either appear at similar locations in both immobilized protein and in solution, or are slightly shifted to somewhat higher wavenumbers when immobilized. In contrast, SERS spectra of similar nanostructured substrates functionalized by MUA SAM without the protein show an entirely different pattern [bib_ref] Application of EBL Fabricated Nanostrucutred Substrates for SERS Detection of Protein A..., Gutierrez-Rivera [/bib_ref] , confirming that [fig_ref] Figure 9: SERS imaging of substrate-immobilized protein A in Design 1 18 [/fig_ref] represents SERS mapping of surface-immobilized protein A. Please click here to view a larger version of this figure. In Design 2, recombinant glucose binding protein (GBP)complexed with D-glucose (ligand) is immobilized on the appropriate substrates in potassium phosphate buffer solution. Samples with immobilized ligand-free GBP are also prepared for comparison. In this design, glucosebinding protein is attached to surface by means of a histidine tag, which binds well to Ni but not to noble metals [bib_ref] Immobilization of Histidine-Tagged Proteins on Electrodes, Ley [/bib_ref]. Since the substrates comprise arrays of Ag nano-dots, nano-hexagons, and unstructured Ag pads on Ni-coated FS , respectively), one can expect most of immobilized protein molecules to be located in gaps between Ag nanostructures where Ni coating is available. The Raman spectra obtained for immobilized glucose-free and glucose-bound GBP are shown in [fig_ref] Figure 1: Scheme of surface-enhanced Raman spectroscopy [/fig_ref] , respectively. All these spectra exhibit a broad band at approximately 3,300 cm , which corresponds to the buffer solution [bib_ref] Current Applications and Future Challenges, Chrimes [/bib_ref]. The spectra obtained with an unstructured Ag pad contain only this single band and do not show any protein vibration modes, confirming that immobilized protein is not found on the Ag surface as expected. In contrast, the spectra obtained with arrays of Ag nano-dots and nano-hexagons exhibit bands around 1,550 cm -1 and 2,900 cm -1 , which represent the analyte [bib_ref] Band Assignment and Qualitative Analysis of Biomedical Compounds with Cardiovascular Activity, Bright [/bib_ref] [bib_ref] Amide Modes and Protein Conformation, Bandekar [/bib_ref]. In particular, the broad band around 1,550 cm , known as the amide II band, is attributable to peptide bonds vibrations in proteins [bib_ref] Amide Modes and Protein Conformation, Bandekar [/bib_ref] [bib_ref] What Vibrations Tell About Proteins, Barth [/bib_ref]. In the case considered, this band represents a superposition of the vibration modes from GBP immobilized on Ni surface between Ag features, and is indicative of SERS enhancement of these modes in the vicinity of noble metal nanostructures when the substrates containing nano-dots or nano-hexagons are used. This band is very weak for the protein in solution in the absence of SERS enhancement [fig_ref] Figure 5: Control Raman spectrum of protein A in DI water obtained at 532... [/fig_ref] and absent on Ag pads without Ni surface available for the protein binding, but it is well pronounced for nanostructured substrates with some Ni surface accessible for the protein to bind. However, even more important for the present study are the other, narrower bands around approximately 2,900 cm -1 that can be attributed to C-H bond wibrations [bib_ref] Band Assignment and Qualitative Analysis of Biomedical Compounds with Cardiovascular Activity, Bright [/bib_ref] [bib_ref] Amide Modes and Protein Conformation, Bandekar [/bib_ref]. The spectrum of glucose-free GBP shows a pronounced band at 2,933 cm -1 with the nano-dots substrate, and a weak but discernible band at a similar wavelength with the nanohexagons substrate [fig_ref] Figure 1: Scheme of surface-enhanced Raman spectroscopy [/fig_ref]. Distinct from the case of glucose free protein, the SERS spectra of glucose-bound GBP shown in [fig_ref] Figure 1: Scheme of surface-enhanced Raman spectroscopy [/fig_ref] exhibit two bands corresponding to C-H bonds vibrations regimes, at 2,850 cm -1 and 2,910 cm . The bands are well pronounced in the spectrum of glucose-bound GBP on nano-hexagons substrate, and they also can be seen in the spectrum of GBP on nano-dots substrate. The band at 2,850 cm -1 is reasonably close to the 2,890 cm -1 one in the control Raman spectrum from D-glucose in solution, and therefore it can be attributed to glucose bound to the protein, whereas the other band (at 2,910 cm -1
) is attributable to C-H bond vibrations of both the protein and glucose. One can conclude that difference of SERS signatures from glucose-free and glucose-bound substrate-immobilized GBP is observable in this region, and C-H bond vibrations of protein-bound glucose are detectable employing the design described. In Design 3, the customized dopamine binding aptamer (DBA) with thiol termination [bib_ref] Retention of Function in the DNA Homolog of the RNA Dopamine Aptamer, Walsh [/bib_ref] is immobilized on the substrate in tris(hydroxymethyl)aminomethane (TRIS) ethylenediaminetetraacetic acid (EDTA) buffer solution, and the dopamine is then bound to the immobilized aptamer. The substrates for this design contain arrays of Au nano-hexagons on FS . Unstructured Au pads are also used for control purposes. Since DNA is intrinsically fluorescent, 780 nm excitation wavelength is used in Design 3 to reduce this factor. In this design, the recognition element (aptamer) is not Raman active in the region of Raman shifts considered in [fig_ref] Figure 1: Scheme of surface-enhanced Raman spectroscopy [/fig_ref] , whereas dopamine shows a significant Raman activity in this region. Since signal from samples exposed to only dopamine without immobilized aptamer shows no resultant dopamine bands
# Discussion
SERS is gaining a recognition as an extremely powerful technique of bio-detection offering many unique advantages. The relation with molecular vibrations allows selectively identifying "fingerprints" of specific analytes from SERS spectra, whereas the extremely high sensitivity makes it possible detecting very small amounts of the analyte [bib_ref] Recent Progress in SERS Biosensing, Bantz [/bib_ref] [bib_ref] Creating, Characterizing, and Controlling Chemistry with SERS Hot Spots, Kleinman [/bib_ref] [bib_ref] Current Applications and Future Challenges, Chrimes [/bib_ref]. Furthermore, SERS is a nondestructive technique that is also relatively insensitive to water, and thereby it is very well suited for probing biological materials in their natural aqueous environment [bib_ref] Recent Progress in SERS Biosensing, Bantz [/bib_ref]. The results presented emphasize these advantages as well as further demonstrate strong potential of SERS as a very flexible label-free technique of bio-detection. In three designs employing monolayers of different substrate-immobilized biomolecules, Raman modes have been detected that could be confidently attributed to the particular analytes. That the detection of these biomolecules, or their respective ligands, have been demonstrated employing planar surfaces of fused silica as the support for SERS substrates, makes the designs compatible with current electronics and microfluidics settings, promising numerous applications in relation with emerging bio-electronic architectures interfacing biological materials with surfaces of electronic and electrochemical devices [bib_ref] Nanomaterials for Bio-Functionalized Electrodes: Recent Trends, Walcarius [/bib_ref] [bib_ref] Applications, Techniques, and Microfluidic Interfacing for, Kim [/bib_ref]. Importantly, in two of three designs SERS detection has been demonstrated for specific binding of small molecules, such as glucose and dopamine, employing monolayers of the surface-immobilized protein and aptamer, respectively, as the recognition elements.
However, several aspects should be taken care of in order to achieve an efficient SERS bio-detection in the "on-chip" setting. First of all, a wellknown challenge that is common for most biomolecules is their propensity to degrade, particularly when exposed to non-natural conditions such as dry environment or intense laser light. Throughout the protocol, we have emphasized the importance of always keeping the bio-functionalized samples immersed in appropriate solutions during the entire experiment, from preparation of the samples to the acquisition of Raman spectra. For the latter, a custom water-proof chamber has been designed to avoid evaporation of the liquid during laser exposures. The duration of exposure and laser intensity should also be limited as described in step 5.3 of the protocol to avoid damage of the samples.
The outcomes of the SERS detection are found sensitive to the geometry of the substrate employed, and particularly the inter-feature separation of the metallic nanostructures. As it follows from [fig_ref] Figure 9: SERS imaging of substrate-immobilized protein A in Design 1 18 [/fig_ref] , the SERS intensity of Design 1 samples depends strongly on the width of the gaps between Au nano-dots on fused silica. Out of three arrays of Au nanodots tested in this design , the highest Raman intensity is achieved with Array I, which has the narrowest gaps between the Au features and therefore it provides more efficient electromagnetic field enhancement. As [fig_ref] Figure 9: SERS imaging of substrate-immobilized protein A in Design 1 18 [/fig_ref] illustrates, control of inter-feature separations at the level of 10-20 nm or less is required. Employing EBL for fabricating SERS substrates, as demonstrated here, provides an efficient resolution specifically for controlling the widths of inter-feature gaps. With a positive-tone EBL resist such as PMMA, the size of holes in PMMA masks can be varied by simply changing the exposure doses. After lift-off this results in different sizes of fabricated metal dots, and the width of gaps between the dots may be tuned as desired by selecting proper EBL exposure doses [bib_ref] Application of EBL Fabricated Nanostrucutred Substrates for SERS Detection of Protein A..., Gutierrez-Rivera [/bib_ref].
The other challenge is optimization of SERS substrate geometry for specific bio-detection application. Although the enhancement effect increases with a decrease of the inter-feature gaps, the relatively large size of biological molecules imposes limitations on how narrow the gaps may be. This is evident from the results for Design 2, where the immobilization method is such that the protein efficiently binds only to the surface between noble metal dots, but not to the dots themselves (see . As it follows from [fig_ref] Figure 1: Scheme of surface-enhanced Raman spectroscopy [/fig_ref] , the SERS spectra for unstructured Ag pads do not show any bands from the analyte. Although the pads exhibit a nano-crystalline structure with very thin inter-island gaps (see these gaps are too narrow to accommodate a protein molecule. Yet another dimension of complexity is added when protein-ligand binding has to be detected. In [fig_ref] Figure 1: Scheme of surface-enhanced Raman spectroscopy [/fig_ref] , the SERS C-H bands are more pronounced in the spectra from ligand-bound GBP than in the ligand-free one, which may be hypothetically explained by a change in the GBP conformation upon binding of D-glucose [bib_ref] Periplasmic Binding Proteins: a Versatile Superfamily for Protein Engineering, Dwyer [/bib_ref] [bib_ref] Design of Bioelectronic Interfaces by Exploiting Hinge-Bending Motions in Proteins, Benson [/bib_ref] , resulting in a more rigid structure with increased Raman activity. If one compares the two nanostructured substrates, the C-H band from ligand-free protein is stronger in SERS spectra obtained with the nano-dots substrate, whereas both the protein and glucose C-H bands from ligand-bound protein are more pronounced with the nano-hexagons substrate. Two factors are expected to result in these differences, the availability of space between Ag features where the GBP could bind to Ni, and the susceptibility of the ligand-bound and ligand free protein to the electromagnetic enhancement of the Raman scattering in "hot spots" between these features. On one hand, the nano-dots pattern offers a relatively larger inter-feature area where Ni coating is available for the protein to bind, which may explain a more pronounced C-H band observed for glucose-free GBP on Ag nano-dots substrate. On the other hand, due their non-uniform structure (see , Ag nano-hexagons might be prone to show a stronger electromagnetic enhancement in narrow gaps between Ag islands within nano-hexagons resulting in stronger C-H vibration bands from glucose-bound GBP on the nano-hexagons substrate. Some details of this interplay require further verification, and optimization of SERS substrates for complex analytes involving large proteins such as the GBP is still in the pipeline.
Clearly, SERS detection of ligand binding employing immobilized biomolecules as a recognition element is facilitated when only the ligand is Raman active in a selected region, whereas the other components are not. This is the case of Design 3, where pronounced SERS bands of aptamer-bound dopamine are obtained [fig_ref] Figure 1: Scheme of surface-enhanced Raman spectroscopy [/fig_ref]. The aptamer-dopamine pair exhibits excellent specificity and the SERS spectrum comprises pronounced bands without any significant background signal.
Future advance of the label-fee SERS technology would involve extensive tests of biomolecules' SERS signal enhancement with a broad range of different surface nanostructure designs. The usage of direct-write electron beam lithography to fabricate various nanostructures with a superb level of control over size, shape, and inter-feature separation, combined with the sample preparation protocols presented here, would facilitate comparison and cross-validation of the results obtained by different research groups. This would address the major challenge of reproducibility when SERS substrates are fabricated employing alternative "bottom up" methods [bib_ref] Creating, Characterizing, and Controlling Chemistry with SERS Hot Spots, Kleinman [/bib_ref] [bib_ref] A Review on the Fabrication of Substrates for Surface Enhanced Raman Spectroscopy..., Fan [/bib_ref] [bib_ref] Engineering Metal Nanostructure for SERS Application, Cao [/bib_ref] , allowing for a better control of metal nanostructure's size and position toward a reliable identification of optimal substrate design for a broad variety of applications. Scalability of these techniques may subsequently be improved by combining EBL with complementary nanolithography methods such as nanoimprint lithography 20 toward future mass-production of nanoscale designs optimized employing the tunable EBL techniques.
# Disclosures
The authors have nothing to disclose.
[fig] Figure 1: Scheme of surface-enhanced Raman spectroscopy. [/fig]
[fig] .: Prepare a 1 mM solution of 11-mercaptodecanoic acid (MUA) in ethanol at RT Sonicate for 10 min 2 Immerse proper nanostructured substrate in the solution of MUA for 48 hr Rinse the sample with ethanol three times and dry for 5 min at RT 3 Prepare a 75 mM solution of N-ethyl-N'-(3-(dimethylamino) propyl) carbodiimide (EDC) in DI water Prepare a 15 mM solution of Nhydroxysuccinimide (NHS) in DI water 4 Using a micropipette deposit 100 µl of NHS on the Au on the substrate, and immediately add 100 µl of EDC on the same area [/fig]
[fig] Figure 5: Control Raman spectrum of protein A in DI water obtained at 532 nm excitation wavelength 18 (A); Raman spectra of ligand-free glucose binding protein in buffer solution obtained at 532 nm excitation wavelength (B); Raman spectra of D-glucose in buffer solution obtained at 532 nm excitation wavelength (C); and spectra of dopamine powder obtained at 532 nm and 780 nm excitation wavelengths 20 (D). All the spectra are regular Raman spectra of solutions (a,b,c) and powder (d) on glass slides without nanostructured substrates. In (D), the assignment of Raman shift regimes to various molecular vibrations was done using General Atomic and Molecular Electronic Structure System (GAMESS)32 and MacMolPlt 31 software as detailed elsewhere20. Reprinted panel (a) with permission from18 American Vacuum Society. Please click here to view a larger version of this figure. [/fig]
[fig] Figure 9: SERS imaging of substrate-immobilized protein A in Design 1 18 . (A) Optical microscope image of the sample, comprising three arrays of Au nanodots with 50 nm pitch and different inter-dot gaps on a fused silica substrate (see also Figure 6), bio-functionalized as shown in Figure 3A. (B) Raman mapping of the sample. (C) SERS spectra from dot Array I and II. In panel (B), the vertical axis represents the distance across the substrate, the horizontal axis represents the Raman shift, and the legend bar indicates the Raman intensities. Vertical dashed lines in panels (B) and (C) represent benchmark Raman bands of free protein A in solution and (*) in panel (C) indicates SERS bands from Array I. The Raman spectra were obtained 532 nm excitation wavelength. Reprinted with permission from 18 American Vacuum Society. Please click here to view a larger version of this figure. [/fig]
|
Role of peptide processing predictions in T cell epitope identification: contribution of different prediction programs
Proteolysis is the general term to describe the process of protein degradation into peptides. Proteasomes are the main actors in cellular proteolysis, and their activity can be measured in in vitro digestion experiments. However, in vivo proteolysis can be different than what is measured in these experiments if other proteases participate or if proteasomal activity is different in vivo. The in vivo proteolysis can be measured only indirectly, by the analysis of peptides presented on MHC-I molecules. MHC-I presented peptides are protected from further degradation, thus enabling an indirect view on the underlying in vivo proteolysis. The ligands presented on different MHC-I molecules enable different views on this process; in combination, they might give a complete picture. Based on in vitro proteasomeonly digestions and MHC-I ligand data, different proteolysis predictors have been developed. With new in vitro digestion and MHC-I ligand data sets, we benchmarked how well these predictors capture in vitro proteasome-only activity and in vivo whole-cell proteolysis, respectively. Even though the in vitro proteasome digestion patterns were best captured by methods trained on such data (ProteaSMM and NetChop 20S), the in vivo whole-cell proteolysis was best predicted by a method trained on MHC-I ligand data Electronic supplementary material The online version of this article ((NetChop Cterm). Follow-up analysis showed that the likely source of this difference is the activity from proteases other than the proteasome, such as TPPII. This non-proteasomal in vivo activity is captured by NetChop Cterm and should be taken into account in MHC-I ligand predictions.
# Introduction
The proteasome degrades intracellular proteins, marked for degradation by the ubiquitination pathway [bib_ref] The ubiquitin system for protein degradation, Hershko [/bib_ref]. Protein degradation, i.e., proteolysis, is important to remove miss-folded proteins, to regulate cellular processes such as the cell-cycle or for the production of MHC-I ligands [bib_ref] Immunoproteasomes preserve protein homeostasis upon interferon-induced oxidative stress, Seifert [/bib_ref] [bib_ref] To cell cycle, swing the APC/C, Van Leuken [/bib_ref] [bib_ref] The spindle checkpoint, APC/CC(dc20), and APC/CC(dh1) play distinct roles in connecting mitosis..., Clijsters [/bib_ref] [bib_ref] Antigen processing by the proteasome, Kloetzel [/bib_ref]. Peptide fragments that result from proteolysis are rapidly degraded by cytosolic aminopeptidases [bib_ref] Peptide diffusion, protection, and degradation in nuclear and cytoplasmic compartments before antige..., Reits [/bib_ref]. However, few peptides escape this degradation and are transported to the endoplasmatic reticulum (ER) by the transporter associated with antigen processing (TAP), where they can form peptide-MHC-I complexes (pMHCs) . pMHCs are presented on the cell surface to enable immune surveillance by T cells.
Most cells express the constitutive proteasome, which is a barrel-shaped multi-subunit protein complex, composed of two α-and two β-rings, where each ring contains seven subunits. In the β ring of the constitutive proteasome, three proteins are present that have proteolytic capacity: β1, β2, and β5 [bib_ref] Proteasome and peptidase function in MHC-class-I-mediated antigen presentation, Kloetzel [/bib_ref]. Under the influence of interferon-γ (IFNγ ), these subunits can be substituted by β1 i , β2 i , and β5 i , respectively, to form the so-called immunoproteasome [bib_ref] Interferon-gamma induces different subunit organizations and functional diversity of proteasomes, Aki [/bib_ref]. Whereas the constitutive proteasome has a preference to cleave hydrophobic, acidic, and basic amino acids, the immunoproteasome is more efficiently cleaving after hydrophobic and basic amino acids [bib_ref] Gamma-interferon and expression of MHC genes regulate peptide hydrolysis by proteasomes, Gaczynska [/bib_ref] [bib_ref] Discrete cleavage motifs of constitutive and immunoproteasomes revealed by quantitative analysis of..., Toes [/bib_ref] [bib_ref] Bioinformatic analysis of functional differences between the immunoproteasome and the constitutive proteasome, Kesmir [/bib_ref]. Other proteasome types can be formed by a combination of constitutive and immunoproteasomal subunits [bib_ref] Two abundant proteasome subtypes that uniquely process some antigens presented by HLA..., Guillaume [/bib_ref] , or with the β5 t subunit that is only expressed in cortical thymic epithelial cells [bib_ref] Regulation of CD8+ T cell development by thymus-specific proteasomes, Murata [/bib_ref]. These different proteasome types largely overlap in their cleavage preferences [bib_ref] Two abundant proteasome subtypes that uniquely process some antigens presented by HLA..., Guillaume [/bib_ref] [bib_ref] Regulation of CD8+ T cell development by thymus-specific proteasomes, Murata [/bib_ref] [bib_ref] Activity-based profiling reveals reactivity of the murine thymoproteasome-specific subunit beta5t, Florea [/bib_ref] , though the efficiency can differ at different cleavage sites which has an influence on the repertoire of MHC-I presented peptides [bib_ref] Mice completely lacking immunoproteasomes show major changes in antigen presentation, Kincaid [/bib_ref].
Two main approaches have been taken to study proteolytic activity: in vitro digestion experiments and in vivo MHC-I-ligand elutions. In an in vitro digestion experiment, a protein is incubated with proteasomes. The peptide fragments that are formed during the digestion can be detected by mass spectrometry, and cleavage sites can be inferred from the fragments [bib_ref] The human 26 S and 20 S proteasomes generate overlapping but different..., Emmerich [/bib_ref] [bib_ref] Quantitative analysis of prion-protein degradation by constitutive and immuno-20S proteasomes indicates differences..., Tenzer [/bib_ref] [bib_ref] Discrete cleavage motifs of constitutive and immunoproteasomes revealed by quantitative analysis of..., Toes [/bib_ref]. So far, the cleavage sites of only three proteins, i.e., β-casein, enolase, and prion protein, have been determined in such in vitro assays [bib_ref] The human 26 S and 20 S proteasomes generate overlapping but different..., Emmerich [/bib_ref] [bib_ref] Quantitative analysis of prion-protein degradation by constitutive and immuno-20S proteasomes indicates differences..., Tenzer [/bib_ref] [bib_ref] Discrete cleavage motifs of constitutive and immunoproteasomes revealed by quantitative analysis of..., Toes [/bib_ref]. Alternatively, in vivo proteolytic activity can be measured by the analysis of digestion fragments that form pMHCs; these fragments can be eluted from a cell and identified by mass spectrometry. The C-terminus of the MHC-I presented peptide is generated by proteolytic activity and reflects an in vivo cleavage site in the protein from which the MHC-I ligand was derived [bib_ref] Antigen processing by the proteasome, Kloetzel [/bib_ref]. However, as many cleavage sites will result in fragments that do not become MHC-I ligands, only a small subset of all cleavage sites can be detected via this approach. In addition, other peptidases such as ACE, TPPII, and Nardilysin [bib_ref] A giant protease with potential to substitute for some functions of the..., Geier [/bib_ref] [bib_ref] Expression of angiotensin-converting enzyme changes major histocompatibility complex class I peptide presentation..., Shen [/bib_ref] [bib_ref] Antigen processing by nardilysin and thimet oligopeptidase generates cytotoxic T cell epitopes, Kessler [/bib_ref] can influence the C-terminus of MHC-I ligands. Therefore, the MHC-Iligand data is more likely to reflect the proteolytic activity of all cellular proteases, rather than the activity of just the proteasomes or one proteasome-type.
To study proteolyis and to aid MHC-I ligand predictions, different proteolyis predictors have been developed [bib_ref] A theoretical approach towards the identification of cleavage-determining amino acid motifs of..., Holzhutter [/bib_ref] [bib_ref] A kinetic model of vertebrate 20S proteasome accounting for the generation of..., Holzhutter [/bib_ref] [bib_ref] Prediction of proteasome cleavage motifs by neural networks, Kesmir [/bib_ref] [bib_ref] The role of the proteasome in generating cytotoxic T-cell epitopes: insights obtained..., Nielsen [/bib_ref] [bib_ref] Modeling the MHC class I pathway by combining predictions of proteasomal cleavage,..., Tenzer [/bib_ref] [bib_ref] Protein degradation and protection against misfolded or damaged proteins, Ginodi [/bib_ref] [bib_ref] An algorithm for the prediction of proteasomal cleavages, Kuttler [/bib_ref] [bib_ref] PAProC: a prediction algorithm for proteasomal cleavages available on the WWW, Nussbaum [/bib_ref] [bib_ref] A kinetic model of vertebrate 20S proteasome accounting for the generation of..., Holzhutter [/bib_ref] , ProteaSMM [bib_ref] Modeling the MHC class I pathway by combining predictions of proteasomal cleavage,..., Tenzer [/bib_ref] , PAProC [bib_ref] An algorithm for the prediction of proteasomal cleavages, Kuttler [/bib_ref] [bib_ref] PAProC: a prediction algorithm for proteasomal cleavages available on the WWW, Nussbaum [/bib_ref] , and PepCleave [bib_ref] Protein degradation and protection against misfolded or damaged proteins, Ginodi [/bib_ref] , have been trained on the in vitro proteasome digestion data from β-casein and enolase [bib_ref] The human 26 S and 20 S proteasomes generate overlapping but different..., Emmerich [/bib_ref] [bib_ref] Discrete cleavage motifs of constitutive and immunoproteasomes revealed by quantitative analysis of..., Toes [/bib_ref]. NetChop 20S [bib_ref] Prediction of proteasome cleavage motifs by neural networks, Kesmir [/bib_ref] [bib_ref] The role of the proteasome in generating cytotoxic T-cell epitopes: insights obtained..., Nielsen [/bib_ref] and the so-called enhanced versions of ProteaSMM are trained on the in vitro proteasome digestion data from β-casein, enolase, and the prion-protein [bib_ref] The human 26 S and 20 S proteasomes generate overlapping but different..., Emmerich [/bib_ref] [bib_ref] Discrete cleavage motifs of constitutive and immunoproteasomes revealed by quantitative analysis of..., Toes [/bib_ref] [bib_ref] Quantitative analysis of prion-protein degradation by constitutive and immuno-20S proteasomes indicates differences..., Tenzer [/bib_ref]. Unlike the other predictors, NetChop Cterm is trained on in vivo MHC-I ligand data [bib_ref] Prediction of proteasome cleavage motifs by neural networks, Kesmir [/bib_ref] [bib_ref] The role of the proteasome in generating cytotoxic T-cell epitopes: insights obtained..., Nielsen [/bib_ref]. Besides the different data sets that were used for training the methods, different computational techniques were used to construct the predictors. For instance, ProteaSMM models the cleavage pattern with a stabilized matrix method (SMM) using six amino acids C-terminal and four amino acids N-terminal of a potential cleavage site, and NetChop is based on a neural network that uses nine amino acids C-terminal and eight amino acids N-terminal of a potential cleavage site.
In 2005, [bib_ref] Modeling the MHC class I pathway by combining predictions of proteasomal cleavage,..., Tenzer [/bib_ref] bench-marked FragPredict, PAProC, NetChop-2.0, and ProteaSMM on several data sets, and showed that ProteaSMM best predicted in vitro proteasome digestion cleavage patterns, whereas NetChop-2.0 Cterm best predicted the cleavage patterns based on MHC-I ligands. [bib_ref] Modeling the MHC class I pathway by combining predictions of proteasomal cleavage,..., Tenzer [/bib_ref] argued that the increased performance of NetChop-2.0 Cterm on the MHC-I ligand data was due to a recognition of TAP-transportable peptides. After this study, NetChop was updated to version 3.0 [bib_ref] The role of the proteasome in generating cytotoxic T-cell epitopes: insights obtained..., Nielsen [/bib_ref] and a new method, PepCleave, was developed [bib_ref] Protein degradation and protection against misfolded or damaged proteins, Ginodi [/bib_ref]. Unfortunately, PepCleave cannot be compared to the other predictors as it predicts fragments and not cleavages [bib_ref] Protein degradation and protection against misfolded or damaged proteins, Ginodi [/bib_ref]. Therefore, we have chosen to compare ProteaSMM and the newest version of NetChop on new in vitro proteasome digestion data sets, and a new benchmark set of MHC-I ligands. Next to benchmarking, our analysis shines light on the nature of the difference between in vitro proteasome-only and in vivo whole-cell proteolytic activities, suggesting an important role for proteases other than the proteasome.
# Results
## Predicting in vitro cleavage patterns
To compare proteasome predictors, we generated a new independent data set. This data set was based on in vitro digestions of 17-30 amino acids long HIV-1 peptides; the products of these digestions were analyzed using mass spectrometry to determine cleavage and non-cleavage sites (see "Methods" section). Digestions were performed with either constitutive or immunoproteasomes. Of 368 possible cleavage sites, 150 (41 %) were used by the constitutive proteasomes, and 148 by the immunoproteasomes, 103 sites (of the 148 cleavage sites) were cleaved by both proteasometypes [fig_ref] Table 1: Predictor performances on in vitro proteasomal cleavage pattern predictions [/fig_ref]. Thus, even though the different proteasomes can target the different sites with varying efficiencies, the set of cleavage sites that is identified in this assay largely overlaps.
The prediction performance of ProteaSMM and NetChop-3.0 was analyzed using receiver operator characteristic (ROC) curves, where the number of correct and false predictions is plotted for every prediction threshold [bib_ref] Measuring the accuracy of diagnostic systems, Swets [/bib_ref]. The area under a ROC-curve (AUC) is a performance measure of the predictor, and is widely used because it is threshold independent [bib_ref] Measuring the accuracy of diagnostic systems, Swets [/bib_ref]. For each predictor (and different versions of the predictors), the AUCs were determined on both constitutive and immunoproteasomal cleavage patterns obtained from the in vitro digestions [fig_ref] Table 1: Predictor performances on in vitro proteasomal cleavage pattern predictions [/fig_ref]. In general, the methods performed better in predicting the immunoproteasomal cleavage pattern. This could be explained by the more biased cleavage preference of immunoproteasomes, that cleave after hydrophobic and basic amino acids with greater, and after acidic amino acids with lesser efficiency [bib_ref] Gamma-interferon and expression of MHC genes regulate peptide hydrolysis by proteasomes, Gaczynska [/bib_ref] [bib_ref] Discrete cleavage motifs of constitutive and immunoproteasomes revealed by quantitative analysis of..., Toes [/bib_ref] [bib_ref] Bioinformatic analysis of functional differences between the immunoproteasome and the constitutive proteasome, Kesmir [/bib_ref]. Such a more biased cleavage pattern might be easier to predict. The immunoproteasomal cleavage pattern was best predicted by proteaSMMimmuno and proteaSMM-constitutive (ROC-comparison test: p < 0.001; [fig_ref] Table 1: Predictor performances on in vitro proteasomal cleavage pattern predictions [/fig_ref] , and the constitutive cleavage pattern was best captured by proteaSMM-constitutive and NetChop-3.0 20S (ROC-comparison test: p < 0.001; [fig_ref] Table 1: Predictor performances on in vitro proteasomal cleavage pattern predictions [/fig_ref]. Surprisingly, the enhanced ProteaSMM versions did not perform better, even though they are trained on extra data from proteasomally digested prion protein [bib_ref] Quantitative analysis of prion-protein degradation by constitutive and immuno-20S proteasomes indicates differences..., Tenzer [/bib_ref]. NetChop-3.0 20S is also trained on prion data, but no version of this method is available that is not trained on prion data, to test if prion data negatively affects the performance of NetChop-3.0 20S.
In summary, the methods that have been trained on in vitro proteasome digestion data (proteaSMMs and NetChop-3.0 20S) outperformed the method that has been trained on in vivo MHC-I ligand data (NetChop-3.0 Cterm), which agrees with previous observations [bib_ref] Modeling the MHC class I pathway by combining predictions of proteasomal cleavage,..., Tenzer [/bib_ref] [bib_ref] Predicting proteasomal cleavage sites: a comparison of available methods, Saxova [/bib_ref] and the expectation that methods trained on in vitro data can best predict proteasome-only cleavage patterns.
## Predicting in vivo cleavage patterns
In vivo proteolytic activity can be rather different from pure proteasomal activity, if other peptidases e.g., ACE, TPPII, or Nardilysin [bib_ref] A giant protease with potential to substitute for some functions of the..., Geier [/bib_ref] [bib_ref] Expression of angiotensin-converting enzyme changes major histocompatibility complex class I peptide presentation..., Shen [/bib_ref] [bib_ref] The carboxypeptidase ACE shapes the MHC class I peptide repertoire, Shen [/bib_ref] [bib_ref] Antigen processing by nardilysin and thimet oligopeptidase generates cytotoxic T cell epitopes, Kessler [/bib_ref] contribute to the in vivo proteolysis. As a result, the ability of different proteasome predictors to predict in vivo proteolysis might be different from their ability to predict in vitro proteasome-only cleavages. To test and compare the in vivo proteolysis prediction performances, we inferred in vivo cleavage sites from non-redundant MHC-I ligands that have been identified from 2005 on, after NetChop Cterm was last updated (n = 832; see and "Methods" section). A data set of in vivo non-cleavage sites was derived in two ways: (1) by shuffling, 100 noncleavage sites were made by shuffling the 19 amino acids flanking a cleavage site (the area used by NetChop for predictions plus one N-terminal and one C-terminal extension, see [fig_ref] Figure 2: Constructing non-cleavage sites by shuffling [/fig_ref].
(2) By assuming that all sites in the source protein of the MHC-I ligand that are not identified as cleavage sites are non-cleavage sites ("Methods" section). The predictors were assessed for their capacity to discriminate cleavage sites from non-cleavage sites, by comparing AUC values. Not surprisingly, NetChop-3.0 Cterm most accurately captured the in vivo cleavage pattern irrespective of the non-cleavage data set (ROC-comparison test: p < 0.001; . This is expected as NetChop Cterm has been trained on in vivo cleavage patterns inferred from MHC-I ligands.
As in vivo proteolysis is inferred from MHC-I ligand data and NetChop Cterm is trained on such data, [bib_ref] Modeling the MHC class I pathway by combining predictions of proteasomal cleavage,..., Tenzer [/bib_ref] noted in an earlier benchmark study that the superior performance of NetChop might be due to a biased recognition of peptides with a high TAP affinity. To exclude this effect, the performance of the different proteasome predictors was tested in combination with a The prediction performance was determined for constitutive and immunoproteasomal cleavage patterns (second and third column, respectively), as AUC of ROC-curves. In boldface the predictors are indicated that performed better than the other predictors (ROC-comparison test: p< 0.001). These predictors were not significantly different in their performance (ROC-comparison test: p> 0.001)
## Fig. 1
Constructing the MHC-I ligand data set. MHC-I ligands and source proteins, that were discovered in elution studies, were derived from the SYFPHEITI database [bib_ref] SYFPEITHI: database for MHC ligands and peptide motifs, Rammensee [/bib_ref] and the IEDB database [bib_ref] The immune epitope database 2.0, Vita [/bib_ref]. The data sets were combined and nonredundant ligands that were not published before 2005 were selected. Every MHC-I ligand in its source protein represents a cleavage site; non-cleavage sites were derived by either shuffling an area of 19 amino acids around the cleavage site [fig_ref] Figure 2: Constructing non-cleavage sites by shuffling [/fig_ref] and "Methods" section), or by defining all other sites in the source proteins of MHC-I ligands as non-cleavage sites ("Methods" section) TAP transport predictor [bib_ref] Identifying MHC class I epitopes by predicting the TAP transport efficiency of..., Peters [/bib_ref]. Therefore, we first followed the approach from [bib_ref] Modeling the MHC class I pathway by combining predictions of proteasomal cleavage,..., Tenzer [/bib_ref] by summing TAP-transport and proteasome cleavage scores into a single score. For both non-cleavage site definitions, NetChop-3.0 Cterm outperformed the other predictors, even when the TAP transport scores were differently weighted prior to summation .
In an AUC-analysis, one can test the predictive performance of a single set of scores. However, we wanted to test the performance of a combination of two scores, i.e., proteasome cleavage and TAP transport scores, as an alternative to the additive model proposed by as Tenzer et al. did. Therefore, we developed a new method to measure the performance of these two scores simultaneously. In this method, for every TAP binding threshold, the performance of the cleavage predictor was measured on cleavage and non-cleavage sites exceeding the threshold. Next, an integration over all the performance scores was combined in a score called volume under the plane (VUP; see "Methods" section). For both non-cleavage definitions, NetChop-3.0 Cterm outperformed the other proteasome predictors based on VUP-scores (ROC-comparison test: p < 0.001; , again indicating that its higher performance is not due to a biased recognition of TAP ligands. Taken together, NetChop Cterm seems to predict in vivo proteolysis better than the other predictors that are trained on proteasome-only in vitro proteolysis data. This suggests that the proteolytic activity in vivo that underlies MHC-I ligand production is markedly different from in vitro proteasome-only proteolysis.
## Comparing in vitro and in vivo proteolysis activity
To better understand why NetChop-3.0 Cterm predicts in vivo proteolysis better than the other predictors, even though these predictors better predict proteasome-only in vitro proteolysis, we examined for each predictor which cleavage sites were given a low prediction score. The cleavage sites with a bottom 5 % prediction score were selected for further analysis. A striking difference between NetChop-3.0 Cterm and the other predictors was observed at position P1' of these poorly predicted cleavage sites (i.e., the C-terminus of the MHC-I ligand; [fig_ref] Figure 2: Constructing non-cleavage sites by shuffling [/fig_ref]. Whereas the amino acids at position P1' were equally distributed for NetChop-3.0 Cterm, a Lysine was found in at least 50 % of the cases for the other predictors [fig_ref] Figure 5: Proteolytic activity after Lysine residues is only predicted by NetChop-3 [/fig_ref]. In other words, the predictors based on in vitro proteasomal cleavage data fail to capture the in vivo cleavage after Lysine residues. This fits with the described proteolytic preferences of TPPII and Nardilysin [bib_ref] A giant protease with potential to substitute for some functions of the..., Geier [/bib_ref] [bib_ref] Antigen processing by nardilysin and thimet oligopeptidase generates cytotoxic T cell epitopes, Kessler [/bib_ref] , and the suggested role of these proteases in the generation of MHC-I ligands, for instance for HLA-A*03 and HLA-A*11 [bib_ref] An essential role for tripeptidyl peptidase in the generation of an MHC..., Seifert [/bib_ref] [bib_ref] Generation of major histocompatibility complex class I antigens: functional interplay between proteasomes..., Kloetzel [/bib_ref] [bib_ref] Proteasome and peptidase function in MHC-class-I-mediated antigen presentation, Kloetzel [/bib_ref]. In addition, other proteases such as ACE have been shown to influence the generation of MHC-I ligands [bib_ref] Expression of angiotensin-converting enzyme changes major histocompatibility complex class I peptide presentation..., Shen [/bib_ref] [bib_ref] The carboxypeptidase ACE shapes the MHC class I peptide repertoire, Shen [/bib_ref] and their proteolytic activity could be captured by NetChop-3.0 Cterm. Taken together, these results
## Fig. 3 predicting in vivo proteolysis. proteasome cleavage predictors
were tested as a stand-alone predictor, or in combination with a TAP predictor, and performance was assessed using AUC and VUP, respectively (see "Methods" section). The performance was tested using either non-cleavage data sets derived by the shuffling method (a) or by taking other sites from the source protein as non-cleavage sites (b) (see and "Methods" section). Examples of the AUC and the VUP analyses are shown in the upper part, AUC and VUP scores are given in the lower part. In all analyses, NetChop-3.0 Cterm showed the highest performance (ROC-comparison test: p< 0.001). The performance of NetChop-3.0 Cterm is shown in red lines, NetChop-3.0 20S in black, ProteaSMM Immuno in yellow, ProteaSMM Immuno enhanced in green, ProteaSMM constitutive in blue, and ProteaSMM constitutive enhanced in magenta lines suggest that NetChop-3.0 Cterm incorporates the activity of all different proteases that make a substantial contribution to in vivo proteolysis, thereby can predict in vivo proteolysis better.
# Discussion
In this study, we analyzed how well different methods can predict the cleavage patterns in proteolysis. In vitro cleavage patterns were shown to be best captured by methods trained on in vitro proteasome digestion data, i.e., ProteaSMM and NetChop-3.0 20S [fig_ref] Table 1: Predictor performances on in vitro proteasomal cleavage pattern predictions [/fig_ref]. Similarly, in vivo proteolysis was best predicted by the method that is trained on MHC-I ligand data, NetChop-3.0 Cterm . Furthermore, we showed that the better prediction of in vivo proteolysis was not due to an embedded recognition of TAP transportable peptides .
There can be two explanations for the difference between in vitro and in vivo proteolysis: First, the proteolytic activity of proteasomes in vitro might be different from their in vivo activity. This difference might result from the interactions with other molecules such as PA28 or the 19S cap regulatory particle [bib_ref] PA28 and the proteasome immunosubunits play a central and independent role in..., De Graaf [/bib_ref] [bib_ref] The human 26 S and 20 S proteasomes generate overlapping but different..., Emmerich [/bib_ref]. Second, other proteases such as TPPII, ACE, or Nardilysin might make a substantial contribution to the in vivo proteolysis [bib_ref] A giant protease with potential to substitute for some functions of the..., Geier [/bib_ref] [bib_ref] Expression of angiotensin-converting enzyme changes major histocompatibility complex class I peptide presentation..., Shen [/bib_ref] [bib_ref] The carboxypeptidase ACE shapes the MHC class I peptide repertoire, Shen [/bib_ref] [bib_ref] Antigen processing by nardilysin and thimet oligopeptidase generates cytotoxic T cell epitopes, Kessler [/bib_ref]. The best described example of in vivo proteolytic activity that is not observed in vitro is the cleavage after Lysine residues. This activity is required to generate ligands for HLA-A*03 and HLA-A*11 that bind peptides with a Lysine at the C-terminus [bib_ref] An essential role for tripeptidyl peptidase in the generation of an MHC..., Seifert [/bib_ref] [bib_ref] Generation of major histocompatibility complex class I antigens: functional interplay between proteasomes..., Kloetzel [/bib_ref] [bib_ref] Proteasome and peptidase function in MHC-class-I-mediated antigen presentation, Kloetzel [/bib_ref]. A welldescribed example of such peptides is the HIV Nef-derived epitope at positions 73 to 82 with a Lysine at its C-terminus, and it was shown that the generation of this peptide depends on TPPII activity [bib_ref] An essential role for tripeptidyl peptidase in the generation of an MHC..., Seifert [/bib_ref]. However, it is not yet known how dominant this endopeptidase activity is within the TPPII enzyme complex [bib_ref] A giant protease with potential to substitute for some functions of the..., Geier [/bib_ref] , and therefore it is not yet clear whether TPPII is responsible for all the activities creating the peptides with a Lysine at its C-terminal. More recently, a more detailed analysis of the Predicting in vivo proteolysis by combining proteolysis and TAP transport predictor scores. For the different proteasome cleavage predictors, the proteasome cleavage prediction score was added to the TAP transport prediction score (as proposed by [bib_ref] Modeling the MHC class I pathway by combining predictions of proteasomal cleavage,..., Tenzer [/bib_ref]. Prediction performance was measured as the AUC of an ROC-curve (Y-axis), using the shuffled sequences as non-cleavage sites (see [fig_ref] Figure 2: Constructing non-cleavage sites by shuffling [/fig_ref] and "Methods" section). When combining the scores, the weight of the TAP transport score was changed by the factor W (on the X-axis). The combined score (C), based on the TAP transport (T ) and proteolysis (P ) score would be C = W * T + P . As a result, the proteasome cleavage or the TAP transport predictor has a larger influence in the combined score if W is smaller or larger, respectively. See for color coding substrate specificity of TPPII has been published , which suggests that the endopeptidase activity of TPPII is very much dependent on the length of the substrate and thus is not likely to be a very general enzymatic activity of TPPII. We show that only NetChop-3.0 Cterm captures this hallmark of in vivo proteolysis [fig_ref] Figure 5: Proteolytic activity after Lysine residues is only predicted by NetChop-3 [/fig_ref]. As this activity has not been contributed to the proteasome, we conclude that NetChop Cterm has learned to incorporate non-proteasomal proteolytic activity.
A biased recognition of TAP transportable peptide is not explaining the increased performance of NetChop Cterm on the prediction of in vivo cleavage sites derived from MHC-I ligand data [fig_ref] Figure 5: Proteolytic activity after Lysine residues is only predicted by NetChop-3 [/fig_ref]. Similarly, one could think that a bias to recognize MHC-I presented ligands should be controlled. NetChop Cterm, was trained on in vivo cleavage sites derived from a set of pMHCs with a homogenous distribution of MHC-I molecules with various binding preferences [bib_ref] The role of the proteasome in generating cytotoxic T-cell epitopes: insights obtained..., Nielsen [/bib_ref] to minimize such a bias that would be due to the recognition of MHC-I binding peptides. In addition, the in vivo cleavage/non-cleavage site data sets in this study are derived from peptides that were not used to train NetChop Cterm and that were eluted from many different MHC-I molecules.
The evaluation of different proteasome cleavage predictors depends on the construction of a set of non-cleavage Proteolytic activity after Lysine residues is only predicted by NetChop-3.0 Cterm. For every proteasome cleavage predictor, 5 % of the true cleavage sites with the lowest prediction scores were determined. The amino acid profile at P1' (i.e., the C-terminus of the presented MHC-I ligand) of these cleavage sites with a low prediction score was analyzed. The height of the letters represents their frequency in the amino acid profile sites, as the performance on these and on the true cleavage sites needs to be compared. Unfortunately, a substantial set of true non-cleavage sites is not available, and therefore we have to rely on assumptions when compiling a set of non-cleavage sites. To prevent a bias as a result of such assumptions, we have followed two different sets of assumptions when constructing the non-cleavage sites. First, non-cleavage sites were made by shuffling the sequence around a cleavage site to destroy any motif that is used by the proteasome while keeping the same distribution of amino acids. Second, we considered other positions in the source protein as non-cleavage sites. Although identical conclusions were drawn from the analyses with the different sets non-cleavage sites, identification of true in vivo noncleavage sites is required to permanently settle this issue or to describe sequence motifs that truly inhibit proteasomal cleavage.
The development of proteasome predictors serves two goals. First, to understand the specificity and biochemical processes that underly proteolysis. Second, to predict and understand how this process influences the MHC-I ligandome. With respect to the first goal, we show that profound differences between proteasome activity in vitro and cellular proteolysis in vivo exist, suggesting a non-negligible role of non-proteasomal proteases. Evidently, the specificity of these additional proteases should be taken into account for optimal MHC-I ligand predictions. Therefore, we conclude that NetChop Cterm or future proteolysis predictors trained on in vivo data should be used in MHC-I ligandome predictions.
# Methods
## Data collection
Proteasomal in vitro cleavage patterns were derived from a digestion of HIV-1 peptides with constitutive or immunoproteasomes, as explained in [bib_ref] Assessment of proteasomal cleavage probabilities from kinetic analysis of time-dependent product formation, Peters [/bib_ref]. Sixteen peptides from the HIV-1 proteins GAG and TAT, with a length of 17 to 30 amino acids, were degraded. After 0, 1, 2, 4, 8, and 24 h of degradation, peptide fragments were analyzed using mass spectrometry (as in [bib_ref] Assessment of proteasomal cleavage probabilities from kinetic analysis of time-dependent product formation, Peters [/bib_ref]. To avoid analyzing secondary cleavage products, peptide fragments found after 4 h of degradation were used to infer cleavage sites. Of 368 possible cleavage sites, 150 were efficiently cleaved by the immunoproteasome after 4 h and 148 were efficiently cleaved by the constitutive proteasome; 103 sites (69 %) were shown to be cleaved by both proteasome subtypes (Supplementary [fig_ref] Table 1: Predictor performances on in vitro proteasomal cleavage pattern predictions [/fig_ref]. The Pro-teaSMM proteasome cleavage predictors require six amino acids N-terminal and four amino acids C-terminal of a possible cleavage site. Therefore, cleavage predictions cannot be made at the beginning and end of a peptide sequence. As a result of this limitation, only 240 (of the 368) sites could be used to compare the different proteasome predictions. Of these 240 sites, 99 were efficiently cleaved by the immunoproteasome and 99 were efficiently cleaved by the constitutive proteasome; 68 sites were put in both sets.
In vivo cleavage sites were inferred from MHC-I ligand data. Ligands that were identified in MHC-I elution studies were downloaded from the SYFPHEITI database [bib_ref] SYFPEITHI: database for MHC ligands and peptide motifs, Rammensee [/bib_ref] and the IEDB database [bib_ref] The immune epitope database 2.0, Vita [/bib_ref]. Source proteins of the MHC-I ligands were downloaded from the NCBI via links that were provided by the SYFPHEITI and IEDB databases. The C-terminal residue of an MHC-I ligand was regarded as position P1' of a cleavage site [fig_ref] Figure 2: Constructing non-cleavage sites by shuffling [/fig_ref]. In total, 3076 MHC-I ligands with their source protein were derived from the SYFPHEITI database and 457 MHC-I ligands with their source protein were derived from the IEDB database. Identical peptides, or peptides that were either a C-or N-terminal extension of each other, were regarded as redundant. In addition, the ligands and their corresponding source proteins that were published before 2005, or which were redundant/identical to an MHC-I ligand published before 2005, were excluded because they could have been used for training of NetChop-3.0 Cterm. This filtering resulted in 832 MHC-ligands and their source proteins, of which every MHC-I ligand corresponds to a peptide fragment that is generated by in vivo proteolytic activity .
Detecting in vivo non-cleavage sites based on the absence of a peptide in the MHC-I ligand databases is not possible, as many other reasons might underlie the absence of an MHC-I ligand, e.g., further degradation of the fragment or low affinity to MHC-I molecules. Therefore, non-cleavage sites were generated in two ways: (1) by shuffling of an area of 19 amino acids around the cleavage site (the longest flanking region used by a proteasome predictor method plus one extra amino acid on each side, as indicated in [fig_ref] Figure 2: Constructing non-cleavage sites by shuffling [/fig_ref]. After shuffling, the middle position, previously corresponding to the cleavage site, was assigned as a noncleavage site. For every cleavage site, 100 non-cleavage sites were generated, i.e., in total 83.200 non-cleavages sites . The advantage of this method is that the amino acid frequencies of cleavage and non-cleavage sites remain identical.
(2) All sites in the source proteins of the MHC-I ligands, that were not assigned as a cleavage site were assumed to be non-cleavage sites (N = 507.538, .
## Prediction performance measures
Proteasome cleavage and TAP transport predictions were performed as suggested by the developers of the different prediction methods [bib_ref] Identifying MHC class I epitopes by predicting the TAP transport efficiency of..., Peters [/bib_ref] [bib_ref] Modeling the MHC class I pathway by combining predictions of proteasomal cleavage,..., Tenzer [/bib_ref] [bib_ref] The role of the proteasome in generating cytotoxic T-cell epitopes: insights obtained..., Nielsen [/bib_ref]. The different proteasome predictors were assessed for their performance in discriminating cleavage from non-cleavage sites. First, the performance of the proteasome predictors was tested using receiver operator characteristic (ROC) curves [bib_ref] Measuring the accuracy of diagnostic systems, Swets [/bib_ref]. In a ROC curve, true positive proportions (TPP) and false positive predictions (FPP) are plotted on the y-and x-axis, respectively, for every threshold. The area under the ROC curve (AUC) is a measure of the predictor's performance. If a predictor performs well, the TPPs increase faster than the FPP, and the AUC becomes larger than 0.5; the maximal AUC is 1.0.
The AUC can only be determined on a single set of prediction scores. However, we aimed to compare the prediction performance of the proteasome predictors in combination with the TAP transport predictor. Therefore, we developed an alternative performance measure: for every TAP transport prediction value, based on the cleavage and non-cleavage sites that exceeded the TAP transport value (T ), the AUC was determined (AU C T ). If less than 25 cleavage sites or non-cleavage sites exceeded the TAP threshold, it was discarded. A score was derived by integrating over all the AUCs with respect to the TAP threshold values and subsequent normalization by the range of TAP thresholds (1). The resulting score ranges between 0 and 1, a random predictor would score 0.5 and a perfect predictor would score 1, similar to the scores obtained in an AUC analysis. This score reflects the predictive performance of the proteolysis predictor for different data sets which have been selected over a range of possible TAP values. We call this performance measure volume under the plane (VUP):
[formula] V UP = n i=1 (T i−1 − T i ) × AU C T i−1 + (T i−1 −T i )×(AU C T i−1 −AU C T i ) 2 Max(T ) − Min(T )(1) [/formula]
## Statistics
Statistical tests were performed using the stats-package from the scipy-module in Python. The difference between AUC/VUP performance measures was determined by deriving AUCs/VUPs on 50 new data sets that were generated by bootstrapping the original data set. The derived AUCs/VUPs were compared using a paired two-tailed t test; p values less than 0.001 were considered significant (as in [bib_ref] Modeling the MHC class I pathway by combining predictions of proteasomal cleavage,..., Tenzer [/bib_ref]. We refer to this test as the ROC-comparison test.
[fig] Figure 2: Constructing non-cleavage sites by shuffling. The C-terminus of an MHC-I ligand (between P1' and P1) is defined as a cleavage site. An area of 19 amino acids (from P10' to P9) around a cleavage site was shuffled and the middle position was assigned as a non-cleavage site. For every cleavage site, 100 non-cleavage sites were constructed. The positions that are used by NetChop-3.0 (P9' to P8) and ProteaSMM (P6' to P4) for predicting cleavage probabilities are indicated [/fig]
[fig] Figure 5: Proteolytic activity after Lysine residues is only predicted by NetChop-3.0 Cterm. For every proteasome cleavage predictor, 5 % of the true cleavage sites with the lowest prediction scores were determined. The amino acid profile at P1' (i.e., the C-terminus of the presented MHC-I ligand) of these cleavage sites with a low prediction score was analyzed. The height of the letters represents their frequency in the amino acid profile [/fig]
[table] Table 1: Predictor performances on in vitro proteasomal cleavage pattern predictions [/table]
|
Nuc2p, a Subunit of the Anaphase-Promoting Complex, Inhibits Septation Initiation Network Following Cytokinesis in Fission Yeast
# Introduction
The eukaryotic cell cycle is composed of an invariant sequence of events, in which DNA replication precedes mitosis and mitosis in turn precedes cytokinesis [bib_ref] Checkpoints: Controls that ensure the order of cell cycle events, Hartwell [/bib_ref]. Cells also possess mechanisms to ensure that DNA replication, chromosome segregation and cytokinesis occur only once per cell cycle [bib_ref] Mechanisms restricting DNA replication to once per cell cycle: MCMS, pre-replicative complexes..., Romanowski [/bib_ref] [bib_ref] Uncontrolled septation in a cell division cycle mutant of the fission yeast..., Minet [/bib_ref]. While much progress has been achieved in understanding the temporal regulation of DNA synthesis and chromosome segregation, the mechanisms by which cytokinesis is restricted to once per cell cycle has not been fully explored.
In recent years, the fission yeast Schizosaccharomyces pombe has emerged as an attractive organism for the study of cytokinesis and its relation to the rest of the cell cycle [bib_ref] Timing is everything: Regulation of mitotic exit and cytokinesis by the MEN..., Mccollum [/bib_ref]. S. pombe cells, like animal cells, divide utilizing an actomyosin based contractile ring [bib_ref] A new tropomyosin essential for cytokinesis in the fission yeast S. pombe, Balasubramanian [/bib_ref] [bib_ref] The Schizosaccharomyces pombe cdc3þ gene encodes a profilin essential for cytokinesis, Balasubramanian [/bib_ref] [bib_ref] Isolation and characterization of new fission yeast cytokinesis mutants, Balasubramanian [/bib_ref] [bib_ref] a protein required for cytokinesis in fission yeast, is a component of..., Chang [/bib_ref] [bib_ref] Rng2p, a protein required for cytokinesis in fission yeast, is a component..., Eng [/bib_ref] [bib_ref] The S. pombe cdc15 gene is a key element in the reorganization..., Fankhauser [/bib_ref] [bib_ref] Schizosaccharomyces pombe cdc4þ gene encodes a novel EF-hand protein essential for cytokinesis, Mccollum [/bib_ref] [bib_ref] Fission yeast Rng3p: An UCS-domain protein that mediates myosin II assembly during..., Wong [/bib_ref] [bib_ref] Growth polarity and cytokinesis in fission yeast: The role of the cytoskeleton, Marks [/bib_ref] [bib_ref] The S. pombe rlc1 gene encodes a putative myosin regulatory light chain..., Goff [/bib_ref] [bib_ref] Type II myosin regulatory light chain relieves auto-inhibition of myosin-heavy-chain function, Naqvi [/bib_ref]. The actomyosin ring is assembled upon entry into mitosis and prior to chromosome segregation and it contricts after chromosome segregation and mitotic spindle disassembly [bib_ref] Cytokinesis in fission yeast: A story of rings, rafts and walls, Rajagopalan [/bib_ref] [bib_ref] Spatial and temporal pathway for assembly and constriction of the contractile ring..., Wu [/bib_ref].
In fission yeast, a signaling pathway known as Septation Initiation Network (SIN) is a key determinant of cytokinesis [bib_ref] Timing is everything: Regulation of mitotic exit and cytokinesis by the MEN..., Mccollum [/bib_ref] [bib_ref] Isolation and characterization of new fission yeast cytokinesis mutants, Balasubramanian [/bib_ref] [bib_ref] Events at the end of mitosis in the budding and fission yeasts, Simanis [/bib_ref]. While loss of SIN function leads to an inability of cells to undergo cytokinesis, ectopic activation of SIN allows cytokinesis to proceed even prior to entry into mitosis [bib_ref] Uncontrolled septation in a cell division cycle mutant of the fission yeast..., Minet [/bib_ref] [bib_ref] Isolation and characterization of new fission yeast cytokinesis mutants, Balasubramanian [/bib_ref] [bib_ref] Genetic interactions in the control of septation in Schizosaccharomyces pombe, Marks [/bib_ref]. SIN is a GTPase-driven signaling cascade that comprises a small GTPase, Spg1p, three protein kinases: Cdc7p, Sid1p, Sid2p, and their associated factors: Cdc14p, Mob1p [bib_ref] The Spg1p GTPase is an essential, dosage-dependent inducer of septum formation in..., Schmidt [/bib_ref] [bib_ref] Sid2p, a spindle pole body kinase that regulates the onset of cytokinesis, Sparks [/bib_ref] [bib_ref] The role of the sid1p kinase and cdc14p in regulating the onset..., Guertin [/bib_ref] [bib_ref] The cdc7 protein kinase is a dosage dependent regulator of septum formation..., Fankhauser [/bib_ref] [bib_ref] Mob1p interacts with the Sid2p kinase and is required for cytokinesis in..., Hou [/bib_ref] [bib_ref] The S. pombe orthologue of the S. cerevisiae mob1 gene is essential..., Salimova [/bib_ref]. This pathway is negatively regulated by a twocomponent GTPase Activating Protein (GAP), Cdc16p and Byr4p [bib_ref] A novel suppressor of ras1 in fission yeast, byr4, is a dosage-dependent..., Song [/bib_ref] [bib_ref] Simanis V (1993) The S. pombe cdc16 gene is required both for..., Fankhauser [/bib_ref] [bib_ref] Byr4 and Cdc16 form a two-component GTPase-activating protein for the Spg1 GTPase..., Furge [/bib_ref] [bib_ref] Regions of Byr4, a regulator of septation in fission yeast, that bind..., Furge [/bib_ref]. Components of the SIN localize to the spindle pole body (SPB) by association with the scaffold proteins Cdc11p and Sid4p [bib_ref] Sid4p is required to localize components of the septation initiation pathway to..., Chang [/bib_ref] [bib_ref] Sid4p-Cdc11p assembles the septation initiation network and its regulators at the S...., Morrell [/bib_ref] [bib_ref] The spindle pole body protein Cdc11p links Sid4p to the fission yeast..., Tomlin [/bib_ref] [bib_ref] S. pombe cdc11p, together with sid4p, provides an anchor for septation initiation..., Krapp [/bib_ref]. In addition, two of the SIN components, Sid2p and Mob1p, also localize to the cell division site during cytokinesis [bib_ref] Mob1p interacts with the Sid2p kinase and is required for cytokinesis in..., Hou [/bib_ref] [bib_ref] The S. pombe orthologue of the S. cerevisiae mob1 gene is essential..., Salimova [/bib_ref]. Although most components of the SIN are detected at the SPB throughout the cell cycle, Cdc7p and Sid1p are detected at the SPB only during mitosis and cytokinesis [bib_ref] Events at the end of mitosis in the budding and fission yeasts, Simanis [/bib_ref] [bib_ref] The role of the sid1p kinase and cdc14p in regulating the onset..., Guertin [/bib_ref] [bib_ref] The control of septum formation in fission yeast, Gould [/bib_ref] [bib_ref] Asymmetric segregation on spindle poles of the Schizosaccharomyces pombe septuminducing protein kinase..., Sohrmann [/bib_ref]. In particular, since the localization of Sid1p to the SPB depends on cyclin B proteolysis and cyclin dependent kinase (CDK) inactivation, it has been proposed that the SIN might link cytokinesis to mitotic exit [bib_ref] The role of the sid1p kinase and cdc14p in regulating the onset..., Guertin [/bib_ref]. Proteolysis of cyclin B (and thereby the inactivation of CDK activity) is triggered by a multisubunit E3 ubiquitin ligase, termed the anaphase-promoting complex/ cyclosome (APC/C) [bib_ref] A 20S complex containing CDC27 and CDC16 catalyzes the mitosis-specific conjugation of..., King [/bib_ref] [bib_ref] The cyclosome, a large complex containing cyclin-selective ubiquitin ligase activity, targets cyclins..., Sudakin [/bib_ref]. After completion of cytokinesis, the SIN pathway is inactivated, as characterized by the presence of Cdc16p and Byr4p on the SPB and the concomitant loss of Cdc7p and Sid1p from the SPB [bib_ref] Events at the end of mitosis in the budding and fission yeasts, Simanis [/bib_ref] [bib_ref] Byr4 localizes to spindlepole bodies in a cell cycle-regulated manner to control..., Li [/bib_ref] [bib_ref] Asymmetry of the spindle pole bodies and spg1p GAP segregation during mitosis..., Cerutti [/bib_ref]. How the SIN pathway is inactivated following cytokinesis and how its precocious activation in interphase is prevented, while the CDK activity is low, remain poorly understood.
In this study, we uncover a function of Nuc2p, a Tetratricopeptide repeat (TPR)-domain containing subunit of APC/C, in preventing inappropriate cytokinetic events following cell division. While loss of Nuc2p function leads to uncontrolled septation, the current study and a previous study have shown that overproduction of Nuc2p leads to inhibition of cytokinesis [bib_ref] Fission yeast TPR-family protein nuc2 is required for G1-arrest upon nitrogen starvation..., Kumada [/bib_ref]. Nuc2p appears to exert its effects on cytokinesis by modulating the nucleotide state of the Spg1p-GTPase and thereby down regulating the SIN.
# Results
## Nuc2-663 cells undergo multiple septation events at the restrictive temperature
Previous studies have shown that cells overproducing Nuc2p are not defective for mitotic exit, but die as elongated multinucleate cells [bib_ref] Fission yeast TPR-family protein nuc2 is required for G1-arrest upon nitrogen starvation..., Kumada [/bib_ref]. Based on this observation it has been suggested that Nuc2p, a TPR-domain containing subunit of APC/C, is an inhibitor of septation [bib_ref] Fission yeast TPR-family protein nuc2 is required for G1-arrest upon nitrogen starvation..., Kumada [/bib_ref]. To understand how Nuc2p functions in cytokinesis, we investigated the organization and function of the cytokinetic machinery in the nuc2-663 mutant. As previously reported, nuc2-663 cells, upon shift to the restrictive temperature, undergo cytokinesis in the absence of chromosome segregation [bib_ref] A temperature-sensitive mutation of the Schizosaccharomyces pombe gene nuc2þ that encodes a..., Hirano [/bib_ref]. Interestingly, we observed that approximately 28% of nuc2-663 cells (n ¼ 112/ 404) displayed multiple septa, indicating that the cytokinetic machinery might be constitutively active in these cells [fig_ref] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant [/fig_ref] and 1B).
Since hyperactivation of SIN signaling also results in multiseptated cells, we tested the possibility that SIN signaling is up-regulated in the nuc2-663 mutant. Previous studies have shown that Cdc7p and Sid1p are localized to the SPBs in mitotic cells and are lost from the SPBs upon completion of cytokinesis in fully septated cells [bib_ref] Events at the end of mitosis in the budding and fission yeasts, Simanis [/bib_ref]. Accordingly, in wildtype cells, Cdc7p and Sid1p were detected at the SPBs in cells undergoing cytokinesis and were not detected at the SPBs in fully septated cells [fig_ref] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant [/fig_ref] , wt panel). Interestingly, in nuc2-663 cells shifted to the restrictive temperature, Cdc7p and Sid1p persisted at the SPBs even after completion of septation [fig_ref] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant [/fig_ref] , nuc2-663 panel). In addition, Sid1p was routinely detected on both SPBs in nuc2-663 cells shifted to the restrictive temperature [fig_ref] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant [/fig_ref] , indicated by arrowhead). This effect of persistent localization of Cdc7p and Sid1p in septated cells was similar to that observed in cells defective for Cdc16p function [bib_ref] The role of the sid1p kinase and cdc14p in regulating the onset..., Guertin [/bib_ref] [bib_ref] Asymmetric segregation on spindle poles of the Schizosaccharomyces pombe septuminducing protein kinase..., Sohrmann [/bib_ref]. The protein kinase Sid2p and its binding partner Mob1p appear to constitute the most downstream elements of the SIN. In wild type cells, Sid2p and Mob1p localize to the SPBs throughout the cell cycle and are also detected at the cell division site during cytokinesis [bib_ref] Mob1p interacts with the Sid2p kinase and is required for cytokinesis in..., Hou [/bib_ref] [bib_ref] The S. pombe orthologue of the S. cerevisiae mob1 gene is essential..., Salimova [/bib_ref]. Interestingly, in the nuc2-663 mutant, Sid2p was detected at the division site in fully septated cells [fig_ref] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant [/fig_ref] , nuc2-663 panel; indicated by arrow). Furthermore, additional ring like structures containing Sid2p were also detected in these cells [fig_ref] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant [/fig_ref]. Similar localization of Sid2p was not seen in wild type cells, in which the medial localization of Sid2p was lost upon completion of septation [fig_ref] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant [/fig_ref] , wt panel). The persistent localization of Sid1p and Cdc7p to the SPBs and Sid2p at the division site in nuc2-663 cells suggested that Nuc2p might be involved in down regulation of SIN function following cytokinesis.
## Nuc2p prevents inappropriate cytokinesis after septum assembly
The nuc2-663 mutant has been shown to be capable of partially proteolyzing the mitotic B-type cyclin, Cdc13p [bib_ref] Study of cyclin proteolysis in anaphase-promoting complex (APC) mutant cells reveals the..., Chang [/bib_ref]. Since cyclin proteolysis as well as CDK inactivation is sufficient for completion of cytokinesis, it was possible that the multiseptated phenotype we observed was purely due to partial proteolysis of cyclin B and/or re-entry into a subsequent round of mitosis. To address if Nuc2p played a role in the regulation of cytokinesis after execution of its function in mitotic exit, we inactivated Nuc2p function after passage through anaphase. To this end, we generated a nuc2-663 nda3-KM311 strain so as to allow for the inactivation of Nuc2p function following anaphase. The nda3 þ gene encodes the b-subunit of the tubulin heterodimer and the nda3-KM311 mutant results in cold-sensitivity and lethality [bib_ref] The NDA3 gene of fission yeast encodes beta-tubulin: A cold-sensitive nda3 mutation..., Hiraoka [/bib_ref]. The nda3-KM311 allows the synchronization of cells at metaphase due to the activation of the spindle checkpoint, caused by the loss of b-tubulin function. The product of the cold-sensitive allele nda3-KM311 resumes its ability to polymerize into microtubules within 6 minutes of return to the permissive temperature [bib_ref] The NDA3 gene of fission yeast encodes beta-tubulin: A cold-sensitive nda3 mutation..., Hiraoka [/bib_ref] and allows progression through chromosome segregation and mitotic exit.
The nuc2-663 nda3-KM311 and nda3-KM311 (as a control) cells were first cultured at 19 8C to inactivate Nda3p function. Under these conditions, at least 50% of cells arrested at metaphase due to the activation of the spindle assembly checkpoint. Subsequently, these cells were shifted to 32 8C to inactivate Nuc2p function and to reactivate Nda3p function [fig_ref] Figure 2: Ectopic Actomyosin Ring and Septum Formation in the nuc2-663 Mutant After Septation [/fig_ref]. Since reactivation of Nda3p function (based on the immediate assembly of the mitotic spindle and 40% of total cells formed an elongated anaphase spindle after 45 min at 32 8C) occurred more rapidly than the inactivation of Nuc2p, nuc2-663 nda3-KM311 (and the control nda3-KM311) underwent anaphase and septation. Interestingly, maintenance of nuc2-663 nda3-KM311 at 32 8C led to the accumulation of cells that either formed multiple septa, or
## Author summary
Cytokinesis is the process by which a mother cell is physically partitioned into two daughter cells. Cytokinesis is well coordinated with segregation of the genetic material to ensure that the genome is not damaged by the cell division apparatus. How untimely cytokinesis is prevented is not fully understood, and is a topic of current interest. Studies of the mechanisms of segregation of the genetic material and cytokinesis have benefited extensively from the use of the fission yeast Schizosaccharomyces pombe. In this study, we make the discovery that fission yeast Nuc2p, a protein previously known to form part of a multi-protein machine that regulates genome segregation, has a second function in regulating cytokinesis. Nuc2p appears to dampen the septation initiation network, which is an important signaling pathway that is essential for cytokinesis. Thus, Nuc2p prevents the occurrence of cytokinetic events prior to segregation of the genetic material and thereby contributes to genome stability. Since the multi-component machinery that Nuc2p forms part of, as well as Nuc2p itself, has relatives in essentially all eukaryotic cells, a similar mechanism might operate in other cells as well. [fig_ref] Figure 2: Ectopic Actomyosin Ring and Septum Formation in the nuc2-663 Mutant After Septation [/fig_ref] and 2C, cells i-iii). Cells with mis-oriented ectopic septa were also frequently observed [fig_ref] Figure 2: Ectopic Actomyosin Ring and Septum Formation in the nuc2-663 Mutant After Septation [/fig_ref] , cell iv). These effects were observed less frequently in the control nda3-KM311 cells [fig_ref] Figure 2: Ectopic Actomyosin Ring and Septum Formation in the nuc2-663 Mutant After Septation [/fig_ref]. In addition to division septa, additional actomyosin rings (as visualized with antibodies against the myosin light chain Cdc4p) were also detected in nuc2-663 nda3-KM311 cells [fig_ref] Figure 2: Ectopic Actomyosin Ring and Septum Formation in the nuc2-663 Mutant After Septation [/fig_ref]. These rings were either orthogonally placed or mis-oriented [fig_ref] Figure 2: Ectopic Actomyosin Ring and Septum Formation in the nuc2-663 Mutant After Septation [/fig_ref] , Cdc4p panel). The fact that cells with actomyosin rings contained interphase microtubule arrays [fig_ref] Figure 2: Ectopic Actomyosin Ring and Septum Formation in the nuc2-663 Mutant After Septation [/fig_ref] , tubulin panel), suggested that the assembly of additional actomyosin rings was not linked to re-entry into mitosis.
The formation of excessive septum after one round of cytokinesis raised the possibility that SIN activation in nuc2-663 mutant might be able to trigger cytokinesis in interphase cells as has been observed in cdc16-116 mutants [bib_ref] Uncontrolled septation in a cell division cycle mutant of the fission yeast..., Minet [/bib_ref] [bib_ref] Asymmetry of the spindle pole bodies and spg1p GAP segregation during mitosis..., Cerutti [/bib_ref]. To test whether this is the case, nuc2-663 mutant and cdc16-116 mutant were treated with hydroxyurea to block cells in S phase and shifted to restrictive temperature to heat inactivate Nuc2p and Cdc16p functions. As previously reported, heat inactivation of Cdc16p function led to formation of septated cells during interphase as indicated by the interphase microtubule organization and presence of septum in these cells [fig_ref] Figure 2: Ectopic Actomyosin Ring and Septum Formation in the nuc2-663 Mutant After Septation [/fig_ref] , cdc16-116 panel, and 2F). The nuc2-663 mutant at restrictive temperature, however, was similar to wild-type cells and did not assemble division septa [fig_ref] Figure 2: Ectopic Actomyosin Ring and Septum Formation in the nuc2-663 Mutant After Septation [/fig_ref] , wt and nuc2-663 panels, and 2F). The observation that interphase arrested nuc2-663 cells did not assemble division septa suggested that Nuc2p might play an important role in preventing additional rounds of cytokinetic events following septation.
Since Nuc2p is a subunit of the APC/C, it was possible that the entire APC/C might function to inhibit inappropriate cytokinesis. Alternatively, it was possible that Nuc2p regulated cytokinesis in a manner independent of other subunits of the APC/C. To distinguish between these possibilities, we assayed the ability of cut9-665 and lid1-6 (two essential components of the APC/C [bib_ref] Distinct subunit functions and cell cycle regulated phosphorylation of 20S APC/cyclosome required..., Yamada [/bib_ref] [bib_ref] The schizosaccharomyces pombe dim1(þ) gene interacts with the anaphase-promoting complex or cyclosome..., Berry [/bib_ref] mutants to accumulate multiple septa. As before, we tested for the presence of multiple and excessive septa upon shift of synchronous nda3-KM311 cut9-665 and nda3-KM311 lid1-6 to conditions that inactivated the APC/C components. In these experiments, nda3-KM311 cut9-665 and nda3-KM311 lid1-6 behaved similar to nda3-KM311 cells and did not accumulate multiple and excessive septa [fig_ref] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant [/fig_ref]. Furthermore, activation of APC/C by overexpression of Slp1p [fig_ref] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant [/fig_ref] or Ste9p [bib_ref] Fission yeast Ste9, a homolog of Hct1/Cdh1 and Fizzy-related, is a novel..., Kitamura [/bib_ref] [bib_ref] A WD repeat protein controls the cell cycle and differentiation by negatively..., Yamaguchi [/bib_ref] did not lead to septation defects, indicating that APC/C activation per se does not lead to septation defects. Collectively, these experiments suggested that the inhibition of inappropriate cytokinesis by Nuc2p might not require the other subunits of the APC/C.
## Cells overexpressing nuc2p assemble, but do not maintain, actomyosin rings
We have found that loss of Nuc2p function leads to persistent localization of SIN components, such as Cdc7p, Sid1p, and Sid2p, even after completion of septation.
Previous studies have shown that overproduction of Nuc2p leads to defective cytokinesis, although the basis of this effect remained unknown [bib_ref] Fission yeast TPR-family protein nuc2 is required for G1-arrest upon nitrogen starvation..., Kumada [/bib_ref]. To gain insights into the mechanism by which Nuc2p overproduction inhibits cytokinesis, a strain of yeast was generated in which the thiamine-repressible promoter, nmt1, was used to replace the endogenous nuc2 promoter (hereafter referred to as nmt1-nuc2). The nmt1-nuc2 strain was used in all overexpression experiments. The nmt1-nuc2 strain resembled wild type cells in morphology upon growth in medium supplemented with thiamine. To analyze the overexpression phenotype more thoroughly, we generated an nmt1-nuc2 strain that expresses the nuclear marker Uch2p-GFP (Ubiquitin C-terminal hydrolase fused to GFP). When this strain was grown in medium containing thiamine, cells were found to contain either a single nucleus or two nuclei [fig_ref] Figure 3: Actomyosin Rings Are Assembled, but Not Maintained, at the Division Site in... [/fig_ref] , depending on the cell cycle stage. Upon removal of thiamine from the medium (leading to overexpression of Nuc2p), a high proportion of cells accumulated two or more nuclei [fig_ref] Figure 3: Actomyosin Rings Are Assembled, but Not Maintained, at the Division Site in... [/fig_ref]. In particular, after 18 h of derepression of the nmt1 promoter, more than 40% of the cells contained 4 or more nuclei, while another 40% of the cells contained two nuclei, most of which were of a postmitotic configuration. We also attempted to overproduce Nuc2p in synchronous cultures, but these experiments did not lead to significant synchrony at the point of maximal induction of nmt1 promoter, due to the long amount of time required to fully derepress the nmt1 promoter (unpublished data).
It is known that cytokinesis in fission yeast requires the function of an actomyosin ring [bib_ref] Cytokinesis in fission yeast: A story of rings, rafts and walls, Rajagopalan [/bib_ref]. We considered the possibility that overexpression of Nuc2p affects the assembly of actin and/or myosin components at the cell division site. To address this question, we grew the nmt1-nuc2 cells in medium lacking thiamine to overexpress Nuc2p and stained these cells with phalloidin. In cells overexpressing Nuc2p, Factin assembled into ring structures in mitotic cells, as in control cells, suggesting that the recruitment and assembly of F-actin into the actomyosin ring was not affected [fig_ref] Figure 3: Actomyosin Rings Are Assembled, but Not Maintained, at the Division Site in... [/fig_ref]. In addition, we tested the effect of Nuc2p overexpression on the localization of the FCH domain protein, Cdc15p, which is essential for actomyosin ring maintenance and septum assembly [bib_ref] The S. pombe cdc15 gene is a key element in the reorganization..., Fankhauser [/bib_ref] [bib_ref] The PCH family protein, Cdc15p, recruits two F-actin nucleation pathways to coordinate..., Carnahan [/bib_ref] [bib_ref] Cell cycle-dependent roles for the FCH-domain protein Cdc15p in formation of the..., Wachtler [/bib_ref]. As in the case of F-actin, cells were able to assemble Cdc15p rings upon Nuc2p overexpression [fig_ref] Figure 3: Actomyosin Rings Are Assembled, but Not Maintained, at the Division Site in... [/fig_ref].
To observe the dynamics of ring assembly and constriction process, we performed time-lapse microscopy in nmt1-nuc2 cells expressing Rlc1p-GFP (regulatory light chain of myosin fused to green fluorescent protein) and Uch2p-GFP as the markers for actomyosin ring and nucleus, respectively. Under conditions of Nuc2p overexpression, cells assembled Rlc1p rings overlying the position of the interphase nuclei [fig_ref] Figure 3: Actomyosin Rings Are Assembled, but Not Maintained, at the Division Site in... [/fig_ref] , ÀT panel). However, these rings were not stable and collapsed in late anaphase and failed to constrict, leading to failure of septation [fig_ref] Figure 3: Actomyosin Rings Are Assembled, but Not Maintained, at the Division Site in... [/fig_ref] , ÀT panel). In contrast, the Rlc1p rings in control cells constricted normally at the end of mitosis leading to the formation of a division septum [fig_ref] Figure 3: Actomyosin Rings Are Assembled, but Not Maintained, at the Division Site in... [/fig_ref] , þT panel). Collectively, these studies established that the presence of excess Nuc2p led to defects in maintenance of fluorescence microscopy. Arrowheads indicate mitotic cells with Sid1p at both SPBs. Arrows indicate cells in which Sid2p-GFP signal persisted at the division site after completion of cytokinesis. Scale bar, 5 lm. doi:10.1371/journal.pgen.0040017.g001 Overexpression of Nuc2p Leads to Loss of Cdc7p and Sid1p from the SPB Previous studies have shown that the mutants defective in the septation initiation network (SIN) assemble actomyosin rings, but these rings collapse late in anaphase, leading to a defect in division septum assembly [bib_ref] Cytokinesis in fission yeast: A story of rings, rafts and walls, Rajagopalan [/bib_ref] [bib_ref] The control of septum formation in fission yeast, Gould [/bib_ref]. Since we observed a phenotype similar to that seen in SIN mutants in cells overexpressing Nuc2p and a complementary phenotype in nuc2-663 cells, it was possible that overexpression of Nuc2p might affect SIN signaling. To determine whether this was the case, we examined the localization of SIN components Sid4p, Cdc11p, Spglp, Cdc7p, Sid1p, and Sid2p upon Nuc2p overexpression. Of these proteins, Sid4p, Cdc11p, Spg1p, and Sid2p are known to localize to the SPBs throughout the cell cycle, while Cdc7p and Sid1p localize during cytokinesis to one of the two SPBs (which has been shown to contain GTP-bound Spg1p) [bib_ref] Events at the end of mitosis in the budding and fission yeasts, Simanis [/bib_ref]. The localization of Sid4p, Cdc11p, and Spg1p to the SPBs was unaltered upon overexpression of Nuc2p [fig_ref] Figure 4: Overexpression ÀT, bottom panel [/fig_ref]. Interestingly, the localization of Cdc7p and Sid1p was severely affected in mitotic cells overexpressing Nuc2p. Cdc7p was detected at the SPBs in 12/61 cells with actomyosin rings, while Sid1p was detected in 6/71 cells with rings [fig_ref] Figure 4: Overexpression ÀT, bottom panel [/fig_ref] and 4D, ÀT and Induced panels), compared to uninduced cells in which at least 80% of cells with actomyosin rings contained Cdc7p-GFP and GFP-Sid1p at the SPBs [fig_ref] Figure 4: Overexpression ÀT, bottom panel [/fig_ref]. Furthermore, in Nuc2p overproducing cells that did display Cdc7p and Sid1p at the SPBs, the signal of Cdc7p and Sid1p was significantly reduced compared to that observed in control cells [fig_ref] Figure 4: Overexpression ÀT, bottom panel [/fig_ref]. Consistent with the loss of Cdc7p and Sid1p from the SPBs, the fluorescence signal of the most downstream kinase of the SIN, Sid2p, was dramatically reduced upon overexpression of Nuc2p [fig_ref] Figure 4: Overexpression ÀT, bottom panel [/fig_ref] and quantification of Sid2p-GFP fluorescence in relation to Sid4p-mRFP is shown in [fig_ref] Figure 4: Overexpression ÀT, bottom panel [/fig_ref]. Next, we determined if overexpression of Nuc2p affected the recruitment or the maintenance of SIN components at the SPBs. To this end, we performed time lapse microscopy to observe the localization of Cdc7p during cytokinesis. We found that Cdc7p was recruited to one of the SPBs during actomyosin ring constriction in cells grown in the presence of thiamine [fig_ref] Figure 4: Overexpression ÀT, bottom panel [/fig_ref] , top panel; indicated by arrowhead). In contrast, in cells overexpressing Nuc2p, Cdc7p was not detected at the SPBs throughout the cell division cycle and the actomyosin ring disassembled at the end of mitosis [fig_ref] Figure 4: Overexpression ÀT, bottom panel [/fig_ref] , bottom panel). Taken together, these studies suggested that overexpression of Nuc2p affects the recruitment of the SIN components, Cdc7p and Sid1p to SPB.
Overexpression of Nuc2p Affects the Binding of Spg1p and Cdc7p, but Not the Steady-State Levels of Cdc7p, Sid1p, and Sid2p
Since Nuc2p is a component of the APC/C, and participated in degradation of molecules regulating mitosis, we considered the possibility that overproduction of Nuc2p might lead to instability and degradation of one or more of the SIN components. We therefore tested the steady state levels of the SIN components Cdc7p, Sid1p, and Sid2p in cells overexpressing Nuc2p. We found that the steady state levels of Cdc7p, Sid1p, and Sid2p were not significantly altered upon overexpression of Nuc2p [fig_ref] Figure 5: Overexpression of Nuc2p Does Not Affect the Steady-State Levels of Cdc7p-3HA and... [/fig_ref] ; and unpublished observations on Sid1p). Thus, Nuc2p appears to affect the SPB localization of the SIN kinases but not their stability, consistent with our findings that other subunits of APC/C were not required for the inhibition of septation by Nuc2p.
It has been shown that the small GTPase Spg1p binds to Cdc7p and recruits this protein kinase to SPBs during mitosis. Cdc7p binds preferentially to GTP-bound Spg1p, which is thought to be the activated form of this GTPase [bib_ref] Asymmetric segregation on spindle poles of the Schizosaccharomyces pombe septuminducing protein kinase..., Sohrmann [/bib_ref] [bib_ref] Identification of functional domains within the septation initiation network kinase, Cdc7, Mehta [/bib_ref]. The fact that the recruitment of Cdc7p to the SPBs was affected in Nuc2p overexpressing cells suggested that the binding between Spg1p and Cdc7p might be affected upon Nuc2p overproduction. To test if this was the case, we overproduced Nuc2p in cells expressing Spg1p-GFP and Cdc7p-3HA and performed co-immnunoprecipitation experiments to look for a physical interaction between Spg1p and Cdc7p. Control immunoprecipitation experiments were carried out from lysates prepared from cells expressing Cdc7p-3HA (but not Spg1p-GFP). Immunoblotting with HA antibodies showed that the level of Cdc7p was comparable in both strains used as well as under both conditions (þ orthiamine) used [fig_ref] Figure 5: Overexpression of Nuc2p Does Not Affect the Steady-State Levels of Cdc7p-3HA and... [/fig_ref]. Immunecomplexes generated with GFP antibodies contained Cdc7p when cells were grown in medium containing thiamine [fig_ref] Figure 5: Overexpression of Nuc2p Does Not Affect the Steady-State Levels of Cdc7p-3HA and... [/fig_ref] , þT of IP panel). Interestingly, immunecomplexes generated with GFP-anti- Arrowhead indicates localization of Cdc7p-GFP at the SPB. Bottom panel, cell grown in medium without thiamine to overexpress Nuc2p. Rlc1p-GFP was used in these cells to label actomyosin ring. Cells were grown on agarose pad containing growth medium and were imaged by laser scanning microscopy. Scale bar, 5 lm. doi:10.1371/journal.pgen.0040017.g004 bodies from cells overproducing Nuc2p contained very little or no Cdc7p-HA [fig_ref] Figure 5: Overexpression of Nuc2p Does Not Affect the Steady-State Levels of Cdc7p-3HA and... [/fig_ref] , ÀT of IP panel). Immune complexes generated with GFP antibodies from cells expressing Cdc7p-HA, but not Spg1p-GFP, did not contain Cdc7p-HA, establishing the specificity of the immunoprecipitation procedure. These experiments established the binding between Cdc7p-3HA and Spg1p-GFP was interrupted in cells overproducing Nuc2p.
## Genetic evidence that nuc2p affects byr4p-cdc16p function
We have shown that cells overproducing Nuc2p are able to localize Sid4p, Cdc11p, and Spg1p to the SPBs. In contrast, the localization of Cdc7p, Sid1p and Sid2p-Mob1p was significantly reduced/altered upon overproduction of Nuc2p. Cdc7p has been shown to preferentially bind to GTP-bound Spg1p [bib_ref] Asymmetric segregation on spindle poles of the Schizosaccharomyces pombe septuminducing protein kinase..., Sohrmann [/bib_ref] , and we have shown that overproduction of Nuc2p leads to a reduction/failure of physical interaction between Spg1p and Cdc7p. Thus it was possible that overproduction of Nuc2p led either to the inactivation of putative guanine nucleotide exchange factors (GEF) for Spg1p or to the persistent activation of the two-component GAP (Byr4p-Cdc16p), leading to the maintenance of Spg1p in its inactive GDP bound form. Since proteins related to the budding yeast Lte1p (functions as a GEF for the Tem1p-GTPase, which is related to fission yeast Spg1p) [bib_ref] The yeast TEM1 gene, which encodes a GTP-binding protein, is involved in..., Shirayama [/bib_ref] , have not been identified in fission yeast, we considered the possibility that overproduction of Nuc2p might cause activation of Byr4p-Cdc16p. We tested this idea by overproduction of Nuc2p in cdc16-116 mutants and assayed the ability of these cells to localize Cdc7p to the SPB and to assemble division septa [fig_ref] Figure 6: Inactivation of Cdc16p Function Promotes Localization of Cdc7p to SPBs and Allows... [/fig_ref]. Inactivation of Cdc16p function in cells overexpressing Nuc2p resulted in the recruitment/maintenance of Cdc7p on SPBs [fig_ref] Figure 6: Inactivation of Cdc16p Function Promotes Localization of Cdc7p to SPBs and Allows... [/fig_ref] , whereas this was not detected in the presence of Cdc16p function [fig_ref] Figure 6: Inactivation of Cdc16p Function Promotes Localization of Cdc7p to SPBs and Allows... [/fig_ref]. More than 90% of the tetranucleate cells were able to undergo cytokinesis and septation (n ¼ 190/203, [fig_ref] Figure 6: Inactivation of Cdc16p Function Promotes Localization of Cdc7p to SPBs and Allows... [/fig_ref] and 6D). In contrast, less than 25% of tetranucleate cells in which Cdc16p was functional underwent cytokinesis and septation (n ¼ 44/204, [fig_ref] Figure 6: Inactivation of Cdc16p Function Promotes Localization of Cdc7p to SPBs and Allows... [/fig_ref]. Although other interpretations are possible, we favour the idea that overproduction of Nuc2p might lead to activation of Byr4p-Cdc16p, thereby to the inability to maintain SIN function and septation.
## Nuc2p functions independently of chfr-related dma1p to inhibit the sin
Previous studies have shown that overproduction of Dma1p, a fission yeast protein related to human CHFR, leads to inhibition of SIN function and defective cytokinesis [bib_ref] The fission yeast dma1 gene is a component of the spindle assembly..., Murone [/bib_ref] [bib_ref] Dma1 prevents mitotic exit and cytokinesis by inhibiting the septation initiation network..., Guertin [/bib_ref]. Since overproduction of Nuc2p also led to defects in SIN signaling, we tested if the cytokinesis-inhibitory effect upon Nuc2p overproduction depended on Dma1p. To this end, Nuc2p was overproduced in cells lacking Dma1p. Cytokinesis defects were observed in control (dma1 þ ) cells as well as dma1D cells overproducing Nuc2p [fig_ref] Figure 7: Nuc2p Acts Independently of Dma1p to Regulate Cytokinesis [/fig_ref]. This experiment suggested that the SIN-inhibitory effect caused by overproduction of Nuc2p was independent of Dma1p. To firmly establish if this was the case, double mutants defective in nuc2 and dma1 were constructed. Interestingly, whereas nuc2-663 and dma1D cells were able to form colonies at 26 8C, the double mutants were unable to do so [fig_ref] Figure 7: Nuc2p Acts Independently of Dma1p to Regulate Cytokinesis [/fig_ref]. Staining of DNA and septum material revealed that the nuc2-663 dma1D, but not nuc2-663, cells displayed an aberrantly septated phenotype, similar to that observed in nuc2-663 cells at higher temperatures [fig_ref] Figure 7: Nuc2p Acts Independently of Dma1p to Regulate Cytokinesis [/fig_ref]. Immunostaining of nuc2-663 dma1D cells with antibodies against Cdc4p and tubulin revealed that 65.3% of septated cells contained segregated DNA, interphase microtubules and actomyosin cables/rings [fig_ref] Figure 7: Nuc2p Acts Independently of Dma1p to Regulate Cytokinesis [/fig_ref] , type I). Furthermore, 34.7% of the septated cells contained unsegregated DNA, interphase microtubules, and actomyosin rings/cables [fig_ref] Figure 7: Nuc2p Acts Independently of Dma1p to Regulate Cytokinesis [/fig_ref] , type II). Since the multi-septate phenotype is largely detected in cells in which chromosome segregation is not affected, we conclude that the effect of septation in the nuc2-663 dma1D double mutants is not purely due to synthetic effects on APC/ C function. Collectively, these studies established that Nuc2p and Dma1p inhibited the SIN by different mechanisms.
# Discussion
Previous studies have characterized the role of Nuc2p, a TPR-containing subunit of APC/C, in the regulation of progression through events of mitosis [bib_ref] Distinct subunit functions and cell cycle regulated phosphorylation of 20S APC/cyclosome required..., Yamada [/bib_ref] [bib_ref] Fission yeast APC/cyclosome subunits, Cut20/Apc4 and Cut23/Apc8, in regulating metaphase-anaphase progression and..., Yamashita [/bib_ref] [bib_ref] Bypassing anaphase by fission yeast cut9 mutation: Requirement of cut9þ to initiate..., Samejima [/bib_ref]. Previous studies have also proposed that Nuc2p might function as an inhibitor of septation, although the mechanism of this inhibition was not investigated [bib_ref] Fission yeast TPR-family protein nuc2 is required for G1-arrest upon nitrogen starvation..., Kumada [/bib_ref]. In the current study we show that Nuc2p prevents septum assembly by inhibiting the septation initiation network (SIN).
Cells defective for nuc2 display a ''cut'' phenotype, characterized by septum assembly in the absence of proper segregation of chromosomes [bib_ref] A temperature-sensitive mutation of the Schizosaccharomyces pombe gene nuc2þ that encodes a..., Hirano [/bib_ref]. Interestingly, we have found that prolonged incubation of nuc2-663 mutants at the restrictive temperature leads to the accumulation of cells with multiple septa. The specific effects of Nuc2p on septation are particularly clear in nuc2-663 mutants shifted to the restrictive temperature after execution of Nuc2p function in mitosis. The multiseptate phenotype of nuc2-663 is reminiscent of that displayed by cells in which the SIN is constitutively activated, such as in cells defective in Cdc16p, which together with Byr4p functions as a GTPase activating protein for the Spg1p-GTPase [bib_ref] Byr4 and Cdc16 form a two-component GTPase-activating protein for the Spg1 GTPase..., Furge [/bib_ref] [bib_ref] Regions of Byr4, a regulator of septation in fission yeast, that bind..., Furge [/bib_ref]. As in the case of cdc16-116 cells, the SIN components Cdc7p and Sid1p are retained at the SPBs even after septum assembly in nuc2-663 cells. In addition, Sid2p, considered to be the most downstream element of the SIN, is retained at the cell cortex after septum assembly in cdc16-116 and nuc2-663 mutants. Interestingly, overproduction of Nuc2p has been shown to cause defects in cytokinesis [bib_ref] Fission yeast TPR-family protein nuc2 is required for G1-arrest upon nitrogen starvation..., Kumada [/bib_ref] and this study). Furthermore, overproduction of Nuc2p leads to defects in actomyosin ring maintenance and localization of Cdc7p, Sid1p, and Sid2p to the SPB, suggestive of defective SIN signaling. Thus, while nuc2 mutants phenocopy cdc16 mutants (in which SIN is constitutively active), overproduction of Nuc2p leads to a phenotype indistinguishable from that displayed by SINdefective mutants. These results imply that Nuc2p might be a bona fide inhibitor of the SIN. Interestingly, whereas activation of SIN, by loss of Cdc16p function, promotes events of cytokinesis (actomyosin ring and division septum assembly) in premitotic interphase arrested cells, the loss of Nuc2p function in premitotic interphase does not lead to actomyosin ring and septum assembly. Thus, although Nuc2p appears to be an inhibitor of the SIN, Nuc2p might inhibit SIN specifically after cell division.
How does Nuc2p inhibit the SIN? Cells overexpressing Nuc2p phenocopy mutants defective in SIN function, which is a Spg1p GTPase driven signaling cascade. Interestingly, the localization of the upstream components of SIN, Cdc11p, Sid4p and Spg1p are not altered upon overproduction of Nuc2p. The increased level of Nuc2p, however, specifically affects the localization of downstream kinases: Cdc7p, Sid1p and Sid2p. Since Spg1p, but not its effectors Cdc7p and Sid1p, remains at the SPB, after Nuc2p overexpression, it can be speculated that the Spg1p in cells overexpressing Nuc2p is GDP-bound. Consistent with this, we have found that the physical interaction between Cdc7p and Spg1p is dramatically reduced in cells overproducing Nuc2p. Two possibilities can be envisaged for the mechanisms that maintain Spg1p in GDP-bound form. First, it is possible that overexpression of Nuc2p inhibits a putative guanine nucleotide exchange factor (GEF) for Spg1p, thereby preventing the loading of GTP onto Spg1p and the activation of SIN signaling. The second possibility is that Nuc2p promotes the activation of the two-component GTPase activating protein (GAP), Cdc16p-Byr4p, leading to the conversion of GTP-Spg1p into GDP-Spg1p. We do not favour the first possibility since no guanine nucleotide exchange factor for Spg1p has been identified to date. Our genetic analysis (restoration of Cdc7p localization and septum assembly in cdc16-116 cells overproducing Nuc2p) points to the second possibility that Nuc2p might function as an activator of Cdc16p-Byr4p. However, it also remains possible that Nuc2p might prevent the conversion of GDP-Spg1p to GTP-Spg1p by acting as a GDP dissociation inhibitor (GDI). Future studies should test these possibilities.
What is the physiological function of the inhibition of SIN by Nuc2p? Both cell division and cell growth utilize actin cytoskeleton and its modulators. Persistent SIN signaling might sequester actin cytoskeleton at the division site and thereby block actin remodeling at the growth sites. Thus, prevention of cytokinesis after cell division might ensure proper growth polarity establishment. Several mechanisms have been uncovered in fission yeast that prevent cytokinesis by inhibiting SIN [bib_ref] Timing is everything: Regulation of mitotic exit and cytokinesis by the MEN..., Mccollum [/bib_ref]. During metaphase, the human CHFR protein homolog in fission yeast, Dma1p, inhibits SIN signaling to prevent precocious cytokinesis [bib_ref] The fission yeast dma1 gene is a component of the spindle assembly..., Murone [/bib_ref] [bib_ref] Dma1 prevents mitotic exit and cytokinesis by inhibiting the septation initiation network..., Guertin [/bib_ref]. Prior to mitotic exit (in metaphase and anaphase A), the inhibition of SIN function by high CDK activity ensures the coordination of mitosis and cytokinesis [bib_ref] The role of the sid1p kinase and cdc14p in regulating the onset..., Guertin [/bib_ref]. Our studies show that Nuc2p inhibits SIN function after completion of cytokinesis. Genetic analysis suggests that Nuc2p and Dma1p, both inhibitors of SIN, might function independently of each other. Our study also points to an additional function for Dma1p after cell division in the prevention of SIN activation. Thus, it appears that high CDK activity and Dma1p inhibit SIN during metaphase, while Nuc2p and Dma1p are required to prevent SIN activation and inappropriate cytokinesis after completion of cell division [fig_ref] Figure 8: Model Summarizing the Temporal Regulation of Cytokinesis in S [/fig_ref].
The current study shows that Nuc2p inhibits SIN signaling. However, it remains unclear whether Nuc2p acts independently or requires a specific set of APC/C components in order to inhibit septation. We have not been able to detect a multiseptate phenotype upon inactivation of function of two additional subunits of the APC/C, namely Cut9p and Lid1p, after passage through anaphase. These observations might suggest that Nuc2p, rather than the entire APC/C, is required for the inhibition of SIN following cell division. In addition, hyperactivation of APC/C by overexpression of Slp1p (this study) and Ste9p [bib_ref] Fission yeast Ste9, a homolog of Hct1/Cdh1 and Fizzy-related, is a novel..., Kitamura [/bib_ref] [bib_ref] A WD repeat protein controls the cell cycle and differentiation by negatively..., Yamaguchi [/bib_ref] does not lead to septation defects, suggesting that activation of APC/C does not lead directly to abnormal septation. In contrast, we have found that overproduction of Cut23p gives weak cytokinesis defects [bib_ref] ORFeome cloning and global analysis of protein localization in the fission yeast..., Matsuyama [/bib_ref] and our unpublished observation), suggesting a role for the entire APC/C in prevention of inappropriate cytokinesis. Collec-tively, our present studies lean toward a specific role for Nuc2p (rather than the entire APC/C) in the inhibition of cytokinesis. However, additional studies with stronger alleles of other APC/C components, which inactivate faster than the currently available alleles, might be required to firmly conclude if Nuc2p inhibition of the SIN requires the entire APC/C.
# Materials and methods
S. pombe strains, media, and reagents. S. pombe strains used in this study are listed in [fig_ref] Table 1: S [/fig_ref]. YES medium or Minimal medium with appropriate supplements were used to culture fission yeast cells. Strains were constructed by either random spore germination method or by tetrad dissection. The nmt1-nuc2 strain was created by transforming a DNA fragment generated by Polymerase Chain Reaction (PCR) using the forward primer: 59CAATAACAACCACC TGTTTGTACCCACATGTTTTTGTTGACATTAACTCCCATCGTTT CCAAAACTTTAATAGATTTGTCGAATTCGAGCTCGTTTAAAC39 and the reverse primer: 59CGTTCTGAATAAAAAATTGAATT ATCATAATTCTGATTATCAATGCAATACCATATTAAACATTTCA ATCGATCTGTCATCATGATTTAACAAAGCGACTATA39. The pos-itive clone was selected using Geneticin (Sigma) and confirmed by PCR. To overexpress Nuc2p in the nmt1-nuc2 strain, cells were first grown in minimal medium containing 15 lM thiamine. The culture was then washed three times and re-inoculated into medium without thiamine, to induce Nuc2p expression.
To arrest cells in S phase, cells were treated with 12 mM hydroxyurea (HU; Sigma) for 6 h prior to any experimental manipulation. To synchronize nuc2-663 mutant at metaphase, nda3-KM311 (control) and nuc2-663 nda3-KM311 mutants were first grown at 27.5 8C in YES medium supplemented with 1.2 M sorbitol. Cells were washed three times with YES before shifting to 19 8C in YES medium to achieve metaphase-arrest.
Gene tagging. To fuse Rlc1p and Sid4p with monomeric red fluorescence protein (mRFP), DNA fragments containing either the rlc1 or sid4 were first amplified by PCR and were then cloned into pJK210-mRFP. The products were linearized using restriction enzymes and transformed into wild type S. pombe by the lithium acetate method [bib_ref] Highfrequency transformation method and library transducing vectors for cloning mammalian cDNAs by..., Okazaki [/bib_ref]. DNA fragments containing C-terminal sequences of rlc1 and sid4 were amplified by PCR with primer pairs MOH461 59GAGAGCTGGTACCTGAATGTTCTCTTCGAAGGAA39 and MOH462 59GAGAGTGCCCGGGATTGCTATCTTTTGACCC39; MOH2460 59CGGGGTACCTAAGGAGATGAATGCCACAATACA ATC39 and MOH2461 59TCCCCCGGGCAAACTACGTTTTTTAAG CTCCC39, respectively, for cloning.
Immunoprecipitation and Western blotting. Immunoprecipitation and Western blotting was performed as described [bib_ref] The novel fission yeast protein Pal1p interacts with Hip1-related Sla2p/End4p and is..., Ge [/bib_ref]. Briefly, cell extracts were prepared by glass bead disruption and solubilised in buffer containing 1% Triton X-100, 150 mM NaCl, 2 mM EDTA, 6 mM Na 2 HPO 4 , 4 mM NaH 2 PO 4 , and complete protease inhibitors (Roche Diagnostics). Cell extracts were then clarified by centrifugation at 14,000 rpm for 10 min at 4 8C. To immunoprecipitate protein complex, 500 ll of soluble protein was incubated with 5 ll of -GFP antibodies for 1-2 h at 4 8C. Protein A-Sepharose beads (100 ll, Amersham Biosciences) were then added to the antigen-antibody immunecomplex and incubated for 45 min at 4'8C. After six washes with buffer containing 1% Triton X-100, the beads were resuspended in SDS-PAGE loading buffer and heated at 95 8C for 5 min. The Protein A-Sepharose beads were spun down at 14,000 rpm for 5 min and the supernatants were subjected to SDS-PAGE. To detect GFP, Myc, or HA-tagged proteins, antibodies recognizing GFP, Myc (Sigma), or HA (Sigma) were used to probe the PVDF membranes containing separated proteins.
Microscopy. To visualize the F-actin cytoskeleton, cells were fixed with 7% formaldehyde and stained with Alexa Fluor-488 phalloidin (Molecular Probes). Septum/cell wall and DNA were stained with aniline blue (Sigma) and 49, 6-diamidino-2-phenylindole (DAPI), respectively. For immunofluorescence studies, cells were fixed either with formaldehyde or with methanol. Cells were then processed as described [bib_ref] Cytokinesis in fission yeast Schizosaccharomyces pombe, Balasubramanian [/bib_ref]. Antibodies against Cdc4p and b-tubulin were used to stain the actomyosin ring and microtubules, respectively. Images were captured using an Olympus IX71 microscope equipped with a Photometrics CoolSNAP ES camera. All images were processed with MetaMorph 6.1. For confocal imaging, Zeiss LSM 510 confocal microscope equipped with a 633/1.4NA PlanApo objective lens was used.
Supporting Information [fig_ref] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant [/fig_ref]. Analysis of Septation in Other APC/C Mutants, cut9-665 and lid1-6, and Overexpression of Slp1p (A) Cells were first synchronized at metaphase using cold sensitive allele of b-tubulin, nda3-KM311. The cultures were shifted to 36 8C to release cells from the metaphase block and to inactivate Cut9p and Lid1p functions. Cells were then collected after 1 to 2 h release from the metaphase block and scored for septation phenotype (normal versus excessive septum) by staining the cells with DAPI and aniline blue.
(B) The APC/C activator Slp1p was overexpressed from nmt1 promoter in wild-type cells. Shown are DAPI stained images of cells grown in the presence (repressing) and absence (inducing) of thiamine. Scale bar, 5 lm. Found at doi:10.1371/journal.pgen.0040017sg001 (1.7 MB TIF).
[fig] Figure 1: Hyperactivation of SIN Signaling in nuc2-663 Mutant (A) nuc2-663 cells were stained with DAPI and aniline blue to visualize nuclei and septum material, respectively. The left panel shows nuc2-663 cells at permissive temperature. The right panel shows nuc2-663 cells at restrictive temperature. (B) Quantification of multiseptated cells in the nuc2-663 mutant. (C) Wild-type and nuc2-663 cells expressing GFP tagged versions of Cdc7p, Sid1p, and Sid2p were grown at 36 8C for 4 h and were visualized using [/fig]
[fig] Figure 2: Ectopic Actomyosin Ring and Septum Formation in the nuc2-663 Mutant After Septation (A) The diagram schematically illustrates the experimental design. (B) Quantification of cells with normal or excessive septa after 1-2 h release from metaphase block. At least 200 cells were counted in each category. (C) Examples of septum patterns in nuc2-663 nda3-KM311 cells. Four examples of cells with excessive septum material are shown. i and ii, multiseptated cells; iii, excessive deposition of septum material at division site; iv, cell with ectopic misoriented septum. (D) Visualization of actomyosin rings in nuc2-663 nda3-KM311 cells by immunofluorescence microscopy. Top, middle, and bottom panels show cells with actomyosin rings that were formed straight, close to, or misoriented with respect to the previous division site, respectively. Microtubules were stained with TAT-1 antibody, and the actomyosin ring was stained with anti-Cdc4p antibody. (E) Septum assembly in S-phase-arrested cells. Wild-type, nuc2-663, and cdc16-116 cells arrested in S phase by hydroxyurea treatment were either formaldehyde fixed for DAPI and aniline blue staining or methanol fixed for immunostaining with TAT-1. (F) Quantification of septated cells versus non-septated cells upon hydroxyurea treatment. At least 300 cells were counted for each category. Scale bar, 5 lm. doi:10.1371/journal.pgen.0040017.g002actomyosin rings in late anaphase, leading to failure of division septum assembly. [/fig]
[fig] Figure 3: Actomyosin Rings Are Assembled, but Not Maintained, at the Division Site in Cells Overexpressing Nuc2p (A) Quantification of the number(s) of nuclei/cell and the frequency of their appearance upon overexpression of Nuc2p. Cells expressing Uch2p-GFP as a nuclear marker were used in the experiment. (B) Visualization of F-actin. Cells overexpressing Nuc2p were fixed and stained with phalloidin and DAPI to visualize the F-actin cytoskeleton and nuclei, respectively. (C) Visualization of Cdc15p. Cells carrying Cdc15p-GFP were induced to overexpress Nuc2p and visualized by fluorescence microscopy. þT/ÀT indicates medium supplemented with or without thiamine. (D) Time-lapse fluorescence microscopy to image the dynamics of actomyosin ring assembly and constriction. Cells were grown in medium with or without thiamine and imaged by time-lapse microscopy. The nucleus and actomyosin ring were marked by Uch2p-GFP and Rlc1p-GFP, respectively. Scale bar, 5 lm. doi:10.1371/journal.pgen.0040017.g003 [/fig]
[fig] Figure 4: Overexpression ÀT, bottom panel) of thiamine. The actomyosin ring (visualized with Rlc1p-GFP) was used as the marker for cells undergoing cytokinesis. (B) Fluorescence signal of Sid2p is reduced in cells overproducing Nuc2p. nmt1-nuc2 cells expressing Sid2p-GFP were visualized using fluorescence microscopy upon growth on medium supplemented with or without thiamine. SPBs and actomyosin ring were marked by Sid4p-mRFP and Rlc1p-mRFP, respectively. (C) Quantification of the relative fluorescence intensities of Sid2p-GFP and Sid4p-mRFP in Nuc2p-uninduced and -induced cells. (D) Quantification of actomyosin ring containing cells with Cdc7p-GFP or GFP-Sid1p at the SPBs. At least 80 cells were counted for each category. (E) Cdc7p is not recruited to the SPBs in cells overproducing Nuc2p. Top panel, control cell grown on minimal medium supplemented with thiamine. [/fig]
[fig] Figure 5: Overexpression of Nuc2p Does Not Affect the Steady-State Levels of Cdc7p-3HA and Sid2p-13Myc, but Disrupts the Binding of Spg1p-GFP and Cdc7p-3HA (A) Protein levels of Cdc7p-3HA and Sid2p-13Myc in nmt1-nuc2 cells grown in medium with or without thiamine. Tubulin serves as the loading control. (B) Lysates from cells expressing Cdc7p-3HA and Spg1p-GFP in the presence or absence of Nuc2p overexpression were immunoprecipitated with GFP antibodies and immunoblotted with antibodies against the HA epitope. Lysates from a strain without Spg1p-GFP was included as the control. Tubulin serves as the loading control. Asterisk indicates heavy chain of IgG. doi:10.1371/journal.pgen.0040017.g005 [/fig]
[fig] Figure 6: Inactivation of Cdc16p Function Promotes Localization of Cdc7p to SPBs and Allows Septation in Cells Overexpressing Nuc2p (A) Diagram schematically illustrates the experimental design. Nuc2p was induced for 24 h at 26 8C, and the culture was shifted to 36 8C to inactivate Cdc16p and examined after 3 h of incubation. (B) The localization of Cdc7p-GFP in cdc16-116 cells overexpressing Nuc2p was visualized by fluorescence microscopy at the permissive or restrictive temperatures. (C) Septum assembly is restored upon inactivation of Cdc16p in cells overexpressing Nuc2p. DAPI and aniline blue staining of formaldehyde fixed cells is shown. (D) Quantification of septated cells versus non-septated cells in nmt1-nuc2 and nmt1-nuc2 cdc16-116. Note that only cells with four nuclei were counted. Scale bar, 5 lm. doi:10.1371/journal.pgen.0040017.g006 [/fig]
[fig] Figure 7: Nuc2p Acts Independently of Dma1p to Regulate Cytokinesis (A) Overexpression of Nuc2p in wild-type or dma1D mutant. (B) Wild-type, nuc2-663, dma1D, and nuc2-663 dma1D strains were diluted serially and spotted on YES agar and incubated at 24 8C or 26 8C. (C) Septum patterns in nuc2-663 and nuc2-663 dma1D mutants grown at 28 8C. Cells were fixed and stained with DAPI and aniline blue. (D) Visualization of actomyosin rings and microtubules in nuc2-663 dma1D cells by immunofluorescence microscopy. Microtubules were stained with TAT-1 antibody and the actomyosin ring was stained with antibodies against Cdc4p. About 150 cells were counted in each category. Arrowhead indicates septum assembled in previous cell division. doi:10.1371/journal.pgen.0040017.g007 [/fig]
[fig] Figure 8: Model Summarizing the Temporal Regulation of Cytokinesis in S. pombe doi:10.1371/journal.pgen.0040017.g008 [/fig]
[table] Table 1: S. pombe Strains Used in This Study nuc2::kan R rlc1-GFP::leu þ uch2-GFP::ura4 þ This study MBY2942 nmt1-nuc2::kan R sid4-GFP::kan R h þ This study MBY2944 nmt1-nuc2::kan R cdc15-GFP::kan R This study MBY3057 nmt41-nuc2-GFP::ura4 þ h À This study MBY3355 nmt1-nuc2::kan R cdc7-GFP::ura4 þ rlc1-GFP::leu1 þ h À This study MBY3356 nmt1-nuc2::kan R GFP-sid1::ura4 þ rlc1-GFP::leu1 þ h þ This study MBY3397 nmt1-nuc2::kan R cdc7-3XHA::ura4 þ h þ nuc2::kan R sid2-GFP::ura4 þ rlc1-mRFP::ura4 þ sid4-mRFP::ura4 þ nuc2::kan R cdc16-116 cdc7-GFP::ura4 þ This study doi:10.1371/journal.pgen.0040017.t001 [/table]
[table] Table S1: S. pombe Strains Used in This Study Found at doi:10.1371/journal.pgen.0040017st001 (26 KB DOC). [/table]
|
Advances of Brain Transcriptomics
Published: 11 October 2022Published: 11 October 202210.3390/genes13101831Received: 20 September 2022 Accepted: 26 September 2022Citation: Redina, O.E.; Babenko, V.N. Advances of Brain Transcriptomics. Genes 2022, 13, 1831. https://
Advancements in RNA sequencing technology in past decade have underlined its power for elucidating the brain gene networks responsible for various stressful factors, as well as the pathologies associated with both genetically determined neurodegenerative diseases and those acquired during the lifespan. As these exciting studies have continued to grow in recent years, we present a series of research papers reporting on progress and new findings based on the analysis of brain transcriptional activity. The brain transcriptome is the most evolved among tissues in terms of transcriptome plasticity and responses to various stimuli. In this Special Issue, we address a spectra of studies, including, but not limited to, animal models of social stress response and various brain-disease-related data/models, in which the identification of features in gene transcription profiles aids our understanding of the molecular mechanisms of socially significant disease development.
Publicly available databases allow researchers to test hypotheses and uncover additional developmental mechanisms of a variety of neuronal degenerative disorders that are actively studied, but remain poorly understood and undefeated. A group of bioinformaticians from Lomonosov Moscow State University, using RNA sequencing data freely available from the Sequence Read Archive, tested a hypothesis about the role of adenosineto-inosine mRNA-editing patterns in the development of Parkinson's disease (PD) [bib_ref] Differential Analysis of A-to-I mRNA Edited Sites in Parkinson's Disease, Pozdyshev [/bib_ref]. They analyzed potential editing sites and showed decreased editing levels in the brains of patients with PD.
Autism spectrum disorder (ASD) is a neurodevelopmental pathology that impedes patients' cognition, social skills, speech, and communication. In recent years, the prevalence of ASD has been steadily increasing; therefore, the identification of the molecular mechanisms underlying ASD occurrence and development is a socially important task. Since ASD is characterized by high heterogeneity and-accordingly-diverse etiology and clinical manifestations, the molecular mechanisms of its development have not yet been fully studied. Lee et al. [bib_ref] Gene Regulation Analysis Reveals Perturbations of Autism Spectrum Disorder during Neural System..., Li [/bib_ref] present a comprehensive study of the molecular basis of this pathology, carried out using integrated multilayer genomics data. The authors compared the RNA sequencing (RNA-seq) expression profiles of induced pluripotent stem cells (iP-SCs), neural progenitor cells (NPCs), and neuron cells from samples from patients with and without ASD; their results suggest that ASD development occurs at the early stage of neural system development. The study allowed the authors to uncover variations in the regulatory cascades between samples from patients with and without ASD, as well as to determine a set of novel disease-associated genes and gene interactions; in particular, they highlighted the functional roles of ELF3 and the interaction between STAT1 and lncRNA ELF3-AS 1 in ASD development.
Spinocerebellar ataxias (SCAs) are a heterogeneous group of dominantly inherited ataxias characterized by progressive cerebellar atrophy mainly due to the dysfunction in and loss of Purkinje cells. Szilvia E. Mezey and colleagues used a SCA14 mouse model to investigate Spinocerebellar ataxia type 14 (SCA14), which is a rare variant of SCAs caused by missense mutations or deletions in PRKCG gene encoding protein kinase C gamma (PKCgamma) [bib_ref] Transcriptome Profile of a New Mouse Model of Spinocerebellar Ataxia Type 14..., Mezey [/bib_ref]. Comparison of transcription profiles of the cerebella from control mice and PKCgamma-A24E mice-which were heterozygous and homozygous for a knock-in pseudo-substrate domain mutation in PKCgamma-allowed the authors to identify key genes and show involved metabolic pathways; these findings expanded our understanding of how mutated PKCgammas might be involved in SCA14 pathogenesis.
It is well known that human lifespans are full of different social conflicts, which can lead to the development of various neurological disorders. An animal model of chronic social confrontation shows that daily agonistic interactions lead male mice to form alternative patterns of social behavior depending on victories and defeats. This model of chronic social confrontation has proven to be an effective experimental approach in elucidating differences in the molecular mechanisms underlying the excitation of the brain neurons in chronically winning mice (winners) and chronically defeated mice (losers) compared with intact controls.
This Special Issue presents the results of a comparative analysis of gene expression profiles in the midbrain raphe nuclei (MRNs) of control male mice compared with winners and losers. It is known that MRNs contain a large number of serotonergic neurons associated with the regulation of numerous types of psycho-emotional states and physiological processes. The article by Redina and colleagues focuses on an analysis of the co-expression of DEGs and Tph2-encoding tryptophan hydroxylase 2, the rate-limiting enzyme of the serotonin synthesis pathway [bib_ref] Correlation of Expression Changes between Genes Controlling 5-HT Synthesis and Genes Crh..., Redina [/bib_ref]. The results showed that Tph2 expression correlated with the expression of many DEGs, associated with behavior, learning, memory, and synaptic signaling. Many common DEGs correlated with Tph2 expression in the MRNs of both aggressive and defeated mice were downregulated. Based on the results obtained, it has been suggested for the first time that CRH and TRH (locally produced in MRNs)-which control the synthesis of corticotrophin-and thyrotropin-releasing hormones, respectively-may be involved in the serotonergic regulation of brain processes during chronic social conflict.
Another article in this Special Issue discusses the contribution of the altered expression profiles of genes encoding the solute carrier (SLC) transporters, which may serve as markers of altered brain metabolic processes and neurotransmitter activities in psychoneurological disorders [bib_ref] Reduced Expression of Slc Genes in the VTA and NAcc of Male..., Smagin [/bib_ref]. The article compares the expression profiles of the Slc genes in the ventral tegmental area (VTA), the nucleus accumbens (NAcc), and the prefrontal cortex (PFC) of both control and aggressive male mice with psychosis-like behavior, induced by repeated experience of aggression accompanied with wins in daily agonistic interactions. The results demonstrated that altered Slc gene activity may lead to a restructuring of metabolic and neurotransmitter processes in a way which is specific for each brain region in the male mice with repeated experience of aggression.
As a continuation of these studies, we include a report by the same group of authors on changes in the level of gene expression in the dorsal striatum in aggressive (winners) and aggression-deprived (AD) male mice [bib_ref] Dorsal Striatum Transcriptome Profile Profound Shift in Repeated Aggression Mouse Model Converged..., Babenko [/bib_ref]. Earlier work by the authors showed that AD mice displayed a higher aggressive behavior score compared with the aggressive group. The results presented in this Special Issue show for the first time that a number of gene-encoding transcription factors did not restore transcription levels within 14 days of aggression deprivation. The authors outlined specific TF-activator gene networks associated with transcriptional repression in affected species compared with controls and concluded that aggressive phenotype selection with a positive reward effect (winning) manifested in an addiction model, which featured a distinct opioid-related withdrawal effect in the AD group.
A study on the impact of social stress on brain plasticity is presented by a team of authors from the University of Illinois at Urbana-Champaign [bib_ref] Terpenoid Backbone Biosynthesis among Pig Hippocampal Pathways Impacted by Stressors, Rymut [/bib_ref]. Using a porcine maternal immune activation (MIA) model, it was shown that activation of maternal immunity during pregnancy and social stress during weaning can alter brain plasticity in the offspring's hippocampus. An analysis of differences in the expression profiles of hippocampus genes in groups of control and experimental animals allowed the authors to conclude that the terpenoid backbone biosynthesis and cocaine addiction pathways played important roles in a three-way interaction between sex, maternal immune activation, and weaning stress. This Special Issue presents the results of another experiment by the same group of authors, which aimed to study the impact of proinflammatory challenges caused by maternal immune activation (MIA) and postnatal exposure to drugs of abuse (morphine) on the prefrontal cortex molecular pathways [bib_ref] Prefrontal Cortex Response to Prenatal Insult and Postnatal Opioid Exposure, Rymut [/bib_ref]. The range of interacting effects on the prefrontal cortex pathways were uncovered, providing new insights into the interplay between the effects of MIA and morphine.
Another article in this Special Issue is devoted to the study of transcription profiles in the brain of Takifugu rubripes fish under hypoxia and normoxia to identify pathways involved in regulating brain metabolism under chronic hypoxic stress. The results of the work also provided new insights into the adaptive molecular mechanisms that arise when the brain responds to hypoxic stress [bib_ref] Transcriptome Analysis Identifies Key Metabolic Changes in the Brain of Takifugu rubripes..., Shang [/bib_ref].
Long noncoding RNAs (lncRNAs) play an important role in the control of many physiological and pathophysiological processes. Differential hypothalamic expression of several lncRNAs was found to be associated with hypertension and the behavior/neurological phenotypes of ISIAH rats; this was characterized as a rat model for a stress-sensitive form of hypertension [bib_ref] Identification of Hypothalamic Long Noncoding RNAs Associated with Hypertension and the Behavior/Neurological..., Fedoseeva [/bib_ref].
In summary, the current Special Issue "Advances of Brain Transcriptomics" uncovers the molecular mechanisms related to transcriptional changes that accompany the development of neurodegenerative diseases or those involved in response to various social or environmental stressors. This knowledge paves the way for the next steps of further research in this field, which will expand the boundaries of our understanding of the processes occurring in the brain under conditions of pathological changes or under the influence of adverse factors.
Conflicts of Interest:The authors declare no conflict of interest. |
Microbe to Microbiome: A Paradigm Shift in the Application of Microorganisms for Sustainable Agriculture
# Introduction
The health of soil plays an essential role in the ability of plants to produce food, fuel, and fiber for a growing world population. To keep pace, the total area of cultivated land worldwide has increased over 500% in the last five decades with a 700% increase in fertilizer use and a several-fold increase in pesticide use. In addition to being the world's largest agricultural producers and exporters, the EU, Brazil, United States, and China also are some of the world's largest pesticide users -each using 827 million, 831 million, 1.2 billion, and 3.9 billion pounds of pesticides, respectively, in 2016. However, these numbers are not sustainable from either a supply-chain or environmental perspective. Thus, because natural resources are limited and their overuse pollutes the environment, the continued use of fertilizers and water to meet the demand of future global food requirements is not sustainable. Of relevance here is that agricultural intensification with high resource use and low crop diversity can negatively affect soil-and plant-associated microbiota (the so-called "phytobiome") with subsequent impacts on critical ecosystem services.
There is growing evidence that aboveground plant diversity supports belowground microbial biodiversity, primarily through root exudation and rhizo-deposition. These more simple carbohydrates released into the soil primarily feed bacteriaand are the most abundant near the root surface and diffuse along a gradient as distance from the root increases. The microbial composition is more abundant and complex in the rhizosphere, the narrow zone surrounding plant roots, with up to 10 9 cells per gram in typical rhizospheric soil, comprising up to 10 6 taxa. The more complex carbohydrates (e.g., lignin, cellulose) are largely degraded by decomposer fungi that break down these recalcitrant compounds into forms that can be used by other microbes. This conversion is largely decoupled from conventional agricultural practices, wherein the organic matter content is often lost to the system, and the carbon flux is at least partially unregulated in this regard. Again, defining nutrient fluxes with techniques like Stable Isotope Labeling (SIP) holds great potential to define and construct resilient, functioning and beneficial microbiomes that can contribute to future holistic agriculture. Thus, applying an efficient and diverse soil microbiome backed by these new technologies can facilitate and promote sustainable agriculture and can effectively contribute to meet the triple requirements of economic, social and environmental sustainability.
Historically, microorganisms that promote plant growth and nutrient acquisition have been used largely as single strains in agriculture to offset such fertilizer inputs as nitrogen and phosphorous. However, studies of natural populations suggest that groups of microbes with distinct function niches play pivotal roles in adhering and desorbing inorganic nutrients to physical surfaces, as well as breaking down organic residues and incorporating them into the soil. Conceptually, such observations support the idea of the microbiome as a second genome or an extended genome of the plant. It is now evident that improving plant performance in a sustainable manner is beyond the binary interaction between a specific microbe or a consortium of beneficial microbes and a targeted host plant. This is a much more complex set of interactions than previously thought that requires modeling for improving predictable outcomes. In this review, we will highlight the current state of the art for the incorporation of specific plant growth-promoting microorganisms and discuss the principles and management practices for next-generation, microbiome-based approaches for sustainable agriculture.
## Application of beneficial microbes in sustainable agriculture: past, present and future
Since the early 1800s, the United States Department of Agriculture has recommended the use of certain rhizobacteria to improve nitrogen fertility in leguminous crops. Since that time, a great deal of research has been conducted on this relationship between legumes and these bacteria, now termed rhizobia, that inhabit unique structures, the nodules, that form on the roots. Rhizobia infecting these nodules are now capable of "biological nitrogen fixation, " whereby dinitrogen is fixed into forms that can be used by the plant. Symbiotically, the bacteria trade these nitrogenous compounds to the host plant in exchange for photosynthetically derived carbon. Despite these limited applications, much remains to be learned regarding both the functional and taxonomic diversity of these symbiotic bacteria and their host plants, the role they play in the global nitrogen cycle, and ultimately, how they can best be harnessed for improving plant productivity. This is particularly true for marginal lands that are not suited for row crop production but will need to be incorporated into global food and forage production approaches moving forward. Further, such degraded lands must but regenerated with the goal of restoring soil health and productivity. Any successful endeavor in this regard must include a characterization of the soil microbiome, both taxonomically and functionally. Attempts currently are underway to fix nitrogen in such non-legumes as wheat, corn and other staple crops that produce the bulk of human food by engineering symbiotic relationships using synthetic biology approaches. Such approaches would significantly impact global food supplies, and may function adequately to reduce the arable land required to meet productivity goals.
Plant growth-promoting microbes not only play critical and diverse roles in growth promotion per se, but also in improving various aspects of plant resilience against a wide array of biotic and abiotic stresses. In this context, researchers globally have worked over the last several decades on plant growthpromoting microorganisms, such as root-associated mycorrhizal fungi, across a broad range of crops and encompassing a wide range of agro-climatic conditions. For perspective,recently reviewed 135 years of mycorrhizal research and reported that merely 8% of the vascular plants are non-mycorrhizal, suggesting that plant families associating with mycorrhizae have been very successful over the evolution of the plant kingdom.
Traditionally, agricultural application of beneficial microorganisms involves a few types of well-characterized microbes, such as mycorrhizal fungi or rhizobia bacteria, for which the mechanisms underlying the plant growth promotion effects are well understood. Further, most of these studies focused solely on the ability of the applied microorganisms to facilitate such specific plant growth-promoting traits as phosphate solubilization, nitrogen fixation, ACC deaminase production, siderophore production, biofilm formation, plant hormone production, biotic, and abiotic stress tolerance or resistance, among others. While these beneficial microorganisms can impart considerable benefits to plant growth and fitness, they are typically documented in simple, one-on-one studies, often conducted in sterile soils in greenhouse conditions. As a consequence, the effects found in such simplified conditions often fail to translate to more complex field situations. Soil in field plots have more complex microbial environments that are presumably adapted to the local eco-environment.
In recent years, next-generation sequencing has revolutionized our understanding of microbial community composition and function, and together with improved culturing methodologies has greatly facilitated the use of biologicals in the field. Specifically, metagenomics-based approaches have uncovered vast, previously unrecognized populations of microbes that may have new or enhanced properties that could be used for agriculture, bioremediation, and human health. For example, comparative analyses of rhizosphere metagenomes from resistant and susceptible tomato plants enabled the identification and assembly of a flavobacterial genome that was far more abundant in the resistant plant rhizosphere microbiome than in that of the susceptible plants. Such findings certainly reveal a role for native microbiota in protecting plants from phytopathogens, and pave a way forward for the development of probiotics to ameliorate plant diseases akin to human health. In another study, a 16S rRNA gene amplicon sequencing analysis of maize root microbiome led to the identification of bacteria that promote growth under low temperature conditions. Additionally, principles of consortium design that rely on cross-talk, cross-feeding and/or substrate channeling between different microorganisms offer new opportunities for "intelligent" consortia design. We propose that the manipulation of the plant microbiome holds tremendous potential for agricultural improvement . Through recent years of research, it is elucidated how microbes worked in nature before, and how decades of chemical fertilizer use have silenced their ability to improve plant fitness and soil health. Therefore, designing a microbial consortium that carefully weighs and evaluates the relationship between inoculants and the resident microbiome would substantially improve the plant growth-promoting potential and resilience of agricultural biologicals to boost plant growth. In this review, we will discuss the key considerations that would improve the likelihood of microbial products to improve crop yield, decrease disease severity and/or ameliorate abiotic stress response. Further, it is likely that such considerations would reduce the inconsistency between the performances of beneficial microbes from controlled greenhouse conditions and more natural environments.
## Microbes for plant growth promotion: a reductionist approach
Sustainable agriculture primarily focuses on reducing the dependency of plants on chemical fertilizers and improving their ability to grow on marginal soil types. For such purposes, individual microorganisms for plant growth-promotion have largely focused on those that facilitate growth and development by enhancing acquisition of nutrient resources from the environment, including fixed nitrogen, iron and phosphate, or modulating growth by altering plant hormone levels. Another approach aimed at reducing yield losses to disease relies on microbes that decrease or prevent the deleterious effects of plant pathogens by several different mechanisms, i.e., by acting as a biocontrol agent. Microbe-based plant growth-promoting products, more popularly marketed as biofertilizer, has been commercially available in many countries since the 1950s. Application of such plant growth-promoting microbes in agricultural context and more specifically as inoculants has been nicely reviewed by. However, under certain cases, the results obtained in the laboratory could not be reproduced in the field primarily due to the presence of many crop species and crop varieties, variable environmental conditions between fields,, occasionally due to the low quality of the inocula, and their inability to compete with the indigenous population. In that context, it is important to consider the fact that there is always greater likelihood of success by introducing mixed cultures of compatible microorganisms, rather than single, pure cultures. This is simply because each strain in the multi-strain consortium can compete effectively with the indigenous rhizosphere population and enhance plant growth with its partners. For example, sequential inoculation of nitrogen fixing bacterium Azotobacter vinelandii, followed by plant growth-promoting root-endophytic fungus Serendipita indica demonstrated better growth in rice. Dual inoculation of S. indica and Mycolicibacterium strains boosted the beneficial effects in tomato (deland that of arbuscular mycorrhizal fungus with plant growthpromoting bacteria Bacillus subtilis demonstrated better growth in wheatas compared to the singly inoculated plants. There also are numerous other reports that showed two strains used in a consortium promoted plant growth in a more effective manner. Nevertheless, to unlock the full potential of soil microbes for such nutrient cycling as nitrogen or phosphorus and providing plant protection against biotic and abiotic stress microbiomes, it is necessary to develop strategies to comprehend the functional capabilities TABLE 1 | List of recent publications in plant and soil microbiome focusing on plant fitness and productivity.
## Category salient findings reference
Plant growth promotion Chilling temperatures critically affects growth of Maize in N. hemisphere. This study reported enrichment of Comamonadaceae and the Pseudomonadaceae in the root endosphere of maize grown under chilling conditions. Additionally two bacterial strains were identified from the root endosphere that could boost maize growth under chilling conditions.
## Beirinckx et al., 2020
A root endophytic bacteria Sphingomonas sp. Cra20 improved growth of Arabidopsis thaliana under drought stress by stimulating the growth of lateral roots and root hairs. Additionally, the relative abundance of Sphingomonas increased in the rhizosphere bacterial community in the water-deficit treatment, suggesting the role of Sphingomonas sp. Cra20 in alleviating drought induced stress.
## Luo et al., 2019
A community-based culture collection (CBC) approach was undertaken to isolate bacteria from the stalks and rhizosphere of Sugarcane. Subsequently, a synthetic community was designed by cross-referencing the CBC with the sugarcane microbiome profile that comprised of highly abundant bacterial groups from roots and stalks. The synthetic community could successfully improve the biomass of sugarcane, and was found to displace the native rhizosphere bacteria community.Willows (Salix spp.) were grown in gamma-irradiated petroleum-contaminated soils. Plants were inoculated with contaminated rhizosphere soil from a willow that grew well, or with contaminated bulk soil in which the plants had died. Willows inoculated with bulk soil performed better than those inoculated with rhizosphere soil.
Microbiomes of different treatments were divergent at the beginning, but had converged at the end of the study, suggesting lasting effect of inoculated microbiome on plant growth, but not on the rhizosphere microbiome.
## Yergeau et al., 2015
Plant defense response Microbiome structure of Banana endosphere in the roots and shoot tips were investigated during plant growth and wilting processes. The keystone bacterial species belonging to the family Enterobacteriaceae family were isolated and further engineered to express ACC deaminase. Plants inoculated with engineered Enterobacteriaceae strains increased resistance to the Fusarium wilt disease. The findings illustrate that the keystone species in the banana microbiome plays functional role in the wilt resistance.
## Liu et al., 2019
Composts represent a sustainable way to suppress diseases and improve plant growth. Compost derived microbial communities enriched in the rhizosphere of Tomato were analyzed for antifungal activity against soil-borne fungal pathogens. Subsequently, microbial synthetic communities (SynComs) were designed with an overarching aim to improve plant fitness. SynComs were found to promote tomato growth as well as suppressed Fusarium wilt symptoms in controlled conditions.
## Tsolakidou et al., 2019
Tomato variety Hawaii 7996 is resistant to the soil-borne pathogen Ralstonia solanacearum, whereas the Moneymaker variety is susceptible to the pathogen. Rhizosphere microbiome analysis revealed clear differences in community profile of these two varieties. Transplantation of rhizosphere microbiota from resistant plants suppressed disease symptoms in susceptible plants. Additionally, a flavobacterium strain isolated from resistant plant rhizosphere microbiome was found to suppress R. solanacearum in susceptible plants.A simplified synthetic bacterial community based on maize rhizosphere microbiome was designed representing most dominant phyla to study their functional attributes in maize seedlings. This synthetic community inhibited the phytopathogenic fungus Fusarium verticillioides, both in planta and in vitro. This study indicates how community profile information can be utilized to design beneficial microbial consortia for improving plant fitness and productivityAbiotic stress response and nutrient use efficiency Recognition of microbes by plant immune system mediated by phosphate stress was investigated. A representative synthetic bacterial community (SynCom) was designed that comprised of 35 bacteria isolated from the roots of Brassicaceae. SynCom enhanced the activity of PHR1, the master transcriptional regulator of the PSR, in Arabidopsis thaliana grown under limited phosphate. Additionally PHR1, repressed plant's immune system in phosphorous starved regime, validating plant's prioritization of nutritional stress over defense.Nitrogen-use efficiency of indica varieties of rice is superior to that of japonica varieties. Root microbiome analysis of these two varieties revealed that microbiota of indica and japonica highly distinct. Further, it was found that this distinctness was associated with a rice nitrate transporter and sensor NRT1.1B. Based on microbiome analysis, microbial synthetic communities (SynComs) were designed. It was found that indica-enriched SynCom improved rice growth in organic nitrogen conditions compared with a japonica-enriched SynCom.
## Zhang et al., 2019
Root associated microbiome of drought-sensitive pepper plant (Capsicum annuum L.) were analyzed focusing on role of microbes conferring plant growth under water limitation. Subsequently pepper root associated culturable bacteria were isolated and evaluated for plant growth promotion and drought tolerance abilities. The composition of the cultivable community associated to rhizosphere and root surrounding soil fractions shared a high similarity. Most of these isolates were able to promote plant growth and alleviate drought-induced stress with enhanced abilities observed in Bacillus and the Rhizobacteria strains.
## Marasco et al., 2012
Community profiling microbiome associated with superior halo-tolerant seepweed Suaeda salsa revealed that rhizospheric and endophytic bacterial community were enriched in genes responsible for salt stress acclimatization. This suggest that S. salsa preferentially recruit halotolerant taxa to confront soil salinity. Based on root endosphere core microbiota, halotolerant bacterial and fungal strains belonging to Pseudomonadales and Montagnulaceae were isolated. It was demonstrated that these core microbiome members were successfully able to improve growth and salt tolerance in the non-host rice plant.Frontiers in Microbiology | www.frontiersin.org of soil microbial communities. Irrespective of the approach, persistence is the first and foremost principle underlying the design of a successful microbial consortium for conferring plant growth promotion. This is not surprising, as the survival and activity of microbes in any soil system face a monumental task of competing with the myriad of microbes naturally adapted to that same soil. Thus, in addition to establishment of a compatible interaction with the host, a successful microbial inoculant has to subsequently compete and persist in the context of indigenous microbes as well as local abiotic conditions. It has been reported that bacterial inoculations can persist in soil up to 7 weeks, but whether this inoculum also can provide plant growth benefits is not clear. While persistence or resilience of any microbial inoculum is more dependent on biotic components of a specific soil type, their persistence can be improved by inoculating crops with consortia rather than single strains. Thus, it can arguably be stated that the diversity of a microbial inoculum, in addition to its plant growth-promoting traits, is critical for enhancing productivity and longevity.
To improve the likelihood of success for such a management strategy, a priori knowledge of indigenous microbial populations competing with the introduced plant growth-promoting agent(s) is critical. While a reductionist approach can define the currency of individual plant-microbe interactions, the concepts of microbial community survival and functioning require, a more holistic, microbiome-based approach empowered by next-generation sequencing technology to study plant-microbe interactions at the community level . Indeed, this will enable researchers to design more robust, synthetic microbial consortia capable of reliably enhancing agricultural productivity.
## Microbiomes for plant growth promotion: the holistic approach
Soil is a vastly heterogeneous growth medium, providing a wide spectrum of ecological niches for microorganisms that enable diverse strains to coexist and form complex microbial communities. When the earliest plants extended their roots into primordial soils, they encountered a habitat already teeming with bacterial and fungal life. Since that early time, plants have interacted with rhizosphere microbes, evolving strategies to forge beneficial alliances with some while keeping others at bay. Such early associations certainly had consequences on plant growth and development. Therefore, a more holistic approach is needed to understand better these microbes and the roles they play in the overall health of plant and soil. Again, recent advances in next-generation sequencing technology and the decreasing costs associated with that technology now allow us to evaluate how microbial populations fluctuate in both space and time or to identify core microbiomes that appear conserved among host genotypes or species. Thus, although culture-independent methods have contributed tremendously to our understanding of plant-associated fungal and bacterial community structures, the study of microbiome functions remains challenging because of the inherent noise of plantassociated microbial communities. It is now well known that there are core sets of microbes that, depending on the host, are recognized as keystone taxa that consistently associate with healthy plants . Consequently, researchers working with specific plant-microbe interactions have increasingly acknowledged the mitigating impact these larger microbial communities have on individual plant-microbe outcomes for plant growth promotion or fitness. Now, plantassociated fungal and bacterial stains from various plant species are being isolated, which will provide in the near future an inestimable resource for assembling taxonomically defined microbial communities with increasing complexity. Therefore, it is now imperative to take advantage of this knowledge to design consortia of microbes to maintain a sustainable rhizosphere community, with key functional properties that include plant protection, nutrient acquisition, and alleviating biotic and abiotic stress responses. From that perspective, synthetic community (SynCom) approaches can provide functional and mechanistic insights into how plants regulate their microbiomes. Not surprisingly, recent culture-independent analyses thus have paved the way for developing SynComs more often.
Mycorrhizal fungi, at least the arbuscular type, were early symbiotic partners of most land plant species, improving nutritional conditions through soil exploration and pathogen resistance of host plants. In reward for the essential physiological services, they receive ca. 20% of net photosynthetic products from plants. Other mycorrhizal systems may have different nutritional benefits and costs, as has been proposed for the serendipitous system. Additionally, third-party partners can modulate the outcome of the tripartite interaction, such as the case of mycorrhizal helper bacteria , fungal endobacterialike Candidatus Moeniiplasma glomeromycotorum within the spores and hyphae of Glomeromycotina, Rhizobium radiobacter within Serendipita indica (Guo FIGURE 2 | Responsiveness as the% gain in plant fitness attribute in response to symbiosis over un-colonized cohorts. This figure illustrates a hypothetical situation wherein genotype A loses less biomass (-20%) in response to soil nutrient limitation than does genotype B (-40%). However, if genotype B for its inherently associated rhizosphere microbiome responds optimally to a mycorrhizal symbiont, then it may be that it loses the least biomass (-10%) due to soil nutrient limitation, if the symbiont is present.% denotes loss in biomass due to soil nutrient limitation. , and N 2 -fixing endobacteria Pseudomonas stutzeri inside basidiomycetes yeast endophyte Rhodotorula mucilaginosa. Hence, it is imperative to consider the composition and functioning of these microbe-microbe interactions to understand plant-microbiome associations in a holistic manner.
## Principles and management of rhizosphere microbiomes for sustainable agriculture
Competence and Resilience of the Rhizosphere Microbiome: Impact of Introduced Microbes on Native Microbiomes
In 1904, the German agronomist and plant physiologist Lorenz Hiltner coined the term rhizosphereto describe the area around a plant root inhabited by a unique population of microorganisms. Since then, numerous studies have been undertaken to decipher the interplay between plants and rhizosphere microorganisms, encompassing a wide variety of plant growth-promoting bacteria, fungi, insects, protozoans, viruses, etc.. The majority of these studies have traditionally followed a simple principle for maximizing successful host infection by pre-inoculation onto the targeted crop of choice to provide a competitive advantage for a desired microbe. Conceptually, this increases the relative abundance of a given beneficial microbe in the rhizosphere, at least temporarily, to achieve the desired benefit. Such studies typically take place in a controlled, artificial condition, such as a defined growth medium in a greenhouse, where competition from a native rhizosphere community is relatively low or nonexistent. As mentioned above, this approach occasionally has failed once field application is attempted or the benefits are dramatically reduced in amplitude and/or endurance.
As an example, Lekberg and Helgason (2018) conducted a literature survey of research papers published on mycorrhizal functioning spanning a 30-year period . The most striking finding of this survey was that less than 5% of the work scientifically manipulated mycorrhizal abundance in the field. While we are not arguing the merit of greenhouse-based studies where the number of variables can be controlled and accounted for, yield gains in field conditions will continue to be modest with such an approach. Rhizosphere competence must be evaluated in a field situation if the true power of this approach is to be realized.
Over the last few decades, mycorrhiza-based bio-fertilizers containing one or several species of fungi were developed in forestry and agriculture. These inoculants are generally effective in plant growth promotion under controlled lab and greenhouse conditions. However, few targeted efforts have been made to measure interactions between the introduced microbe(s) and the native mycorrhizal community, let alone the more complex rhizosphere microbiome. To optimize outcomes from these interactions, targeted research must be undertaken to understand how such mycorrhiza-based biofertilizer integrate themselves within the context of the native microbiome.
## Integration of rhizosphere microbiomes in plant-microbe-nutrient relationships
The soil microbial community often assists plants by weathering minerals from rock surfaces and degrading recalcitrant soil organic matter whereby soil microbes break down soluble and insoluble organic matter and convert it into inorganic, plantavailable forms. Soil organic matter turnover is thus considered a net positive, as it liberates the nutrients locked up in organic matter. For this reason, conventional farming has always relied heavily on soil tillage, along with such other intensive agricultural practices as usage of inorganic fertilizers, herbicides and pesticides. However, it is already clear that such practices have negative consequences on the functional diversity of soil microbiomes. Long-term chemical fertilization has been shown to dramatically decrease the soil pH, which leads to a decrease in bacterial diversity and other changes in microbial community structure. This was well documented in the work of, who showed that long-term application of high doses of inorganic nitrogenous fertilizers severely reduces relative abundance, diversity and structure of diazotrophs, which play a key role in converting atmospheric N 2 to plant-available ammonium.
As mentioned above, soil bacterial communities play a pivotal role in soil organic matter decomposition. In particular, soil carbon and nitrogen are critical factors for bacteria that rely on soil organic C and N decomposition to obtain energy. Further, different types of soil C selectively manipulate soil microbial community composition, resulting in changes in such belowground ecosystem functions as decomposition and nutrient transfer and creating feedbacks that may affect overall plant growth and productivity. For example, bacteria belonging to the genera Chloroflexi, Nitrospirae, and Planctomycetes preferentially feed on recalcitrant organic C, whereas Proteobacteria and Bacteroidetes prefer labile organic C present in the soil. For this reason, amending the soil with such organic fertilizers as compost or manure contributes to higher microbial diversity and biomass compared to mineral-fertilized soils, which in turn positively impacts soil health. Unfortunately, only a few agroecosystem experiments exist that compare organic and conventional management strategies over an extended period for evaluation of impact on soil health and restoration.took a metagenomics approach to assess microbial diversity of soil in response to more than 20 years of continuous organic and conventional farming. Not surprisingly, they found that organic farming increased richness, decreased evenness, and shifted the structure of the soil microbiota when compared with conventionally managed soils under mineral fertilization. There also are reports of significant alterations in the microbial community composition of both summer maize and winter wheat in response to increased nitrogen fertilization dose. Clearly, a better understanding of the interactions between the soil microbiome and conventional agricultural practices is crucial for the development of sustainable management of soil fertility and crop production.
## Managing the rhizosphere microbiome to induce disease suppression in soil
Disease suppressive soils were originally defined byas "soils in which the pathogen does not establish or persist, establishes but causes little or no damage, or establishes and causes disease for a while but thereafter the disease is less important, although the pathogen may persist in the soil." Disease suppressive soils are the best examples of microbiome-mediated protection of plants against root infections by soil-borne pathogens. Such disease-suppressive soils have been described for various soil-borne pathogens, including fungi, bacteria, oomycetes, and nematodes. To date, several microbial genera have been proposed as key players in disease suppressiveness of soils, but the complexity of the microbiome, as well as the underlying mechanisms and microbial traits, remain elusive for most disease suppressive soils.
Recently,showed that upon pathogen invasion, members of the Chitinophagaceae and Flavobacteriaceae became enriched within the plant endosphere. They proposed that this bacterial population shift led to the induction of enzymatic activities associated with fungal cell-wall degradation, as well as secondary metabolite biosynthesis, all aimed at accelerating and augmenting the plant defense response(s). Although the disease suppressive abilities of certain soils can be at least partially attributed to their physico-chemical properties, the capacity of a soil to suppress disease progression is more often attributed to agri-management practices and crop rotation. In classic studies byand, the authors have shown soil to become disease suppressive after mono-culturing wheat over time. More recently, a comparative metatranscriptome analysis of wheat rhizosphere microbiome grown in fields suppressive and nonsuppressive to the plant pathogen R. solani AG8 clearly revealed distinct dominant taxa in these two soil types. Additionally, suppressive samples showed greater expression of polyketide cyclase, terpenoid biosynthesis, and cold shock proteins. While development of probiotics for the human gut microbiome has already been an established field of research, the use of probiotics that comprises naturally occurring bacterial antagonists and competitors that suppress pathogens has recently emerged as a promising strategy for disease suppression in soil. A study on application of probiotic consortia that comprised predefined Pseudomonas species reported suppression of the bacterial plant pathogen Ralstonia solanacearum in the tomato rhizosphere microbiome. In another study, amendment of Metarhizium, an insect-pathogenic fungus that is commonly employed as biological control agents against crop pests, in the rhizosphere of common bean (Phaseolus vulgaris) significantly increased the relative abundance of plant growth promoting such taxa as Bradyrhizobium, Flavobacterium, Chaetomium, and Trichoderma while suppressing the root rot disease symptoms Fusarium solani. Soil suppressive properties are mostly derived from the biological functions of soils. Therefore, elucidation of microbial functions in suppressive soils by a next-generation sequencing approach will facilitate the development of effective, consistent and durable disease management tools.
## Impact of agriculture management practices on the soil microbiome
One important context for plant-microbe interactions is soil structure, as it can vary greatly depending on land-use history, plant species composition and successional stage. Besides playing pivotal roles in soil organic matter decomposition, carbon cycling, nutrient mobilization, etc., saprotrophic fungi also are involved in creating soil structure through the secretion of extracellular compounds and physical binding of soil via hyphal networks. Interestingly, studies on the impact of tillage on the soil fungal communities have shown mixed results. Reports in no-till systems have varied from increased ratios of fungal to bacterial biomassto decreased ratios, as well as no change at all. More recent studies have shown that soil fungal communities are negatively impacted by tillage, as they typically would be responsible for degrading crop residue left on the surface with no-till. More specifically, soil bacterial communities were primarily found to be structured by tillage, whereas soil fungal communities responded mainly to management type with additional effects by tillage. Additionally, it is acknowledged that organically managed systems increased taxonomic and phylogenetic richness, diversity and heterogeneity of the soil microbiota when compared with conventional farming systems. In a simple definition, organic farming system consists of low-input agro-ecosystem farms in which plant productivity and ecosystem functionality are based on the natural availability of plant nutrients. A study aimed at comparing the soil microbiome in conventional and organic farming systems in central Europe revealed no major differences among the main phyla of bacteria between the two farming styles, whereas another study that investigated the effects of 12 years of organic farming on soil microbiomes in northern China reported shifting of the community composition of dominant phyla and significant alterations of functional groups associated with ammonia oxidation, denitrification and phosphorus recycling when compared to conventional farming systems.
In addition to tillage, crop rotation also plays a pivotal role in increasing belowground microbial diversity compared to intensive mono-cropping practices. Although the United States Department of Agriculture has advocated [via the Conservation Reserve Program (CRP)] crop rotation to improve eroded land as early as 1985, its benefit on soil health has only been recognized recently. Several studies reported increases in such soil quality parameters as organic matter content, microbial biomass and respiration under crop rotation management when compared with a mono-cropping system. A meta-analysis of 122 studies that examined crop rotation revealed similar findings, namely that adding one or more crops in rotation to a monoculture substantially increased the soil microbial biomass along with increases in total soil C and N, respectively. In another study, soil microbial communities of corn and switchgrass in mono-cropping systems when compared with mixed prairie grasses demonstrated that bacterial and fungal biomass, especially arbuscular mycorrhizal fungi, were higher in plots with mixed prairie grasses. A 16S amplicon-based metagenomic analysis of an almost 20year-old field trial in Bernburg, Germany revealed a significant effect of tillage practice and the preceding crop on prokaryotic community structuresCover crops are typically unharvested crops planted between cash crops that augment C provisioning to the soil system not only via unharvested residues, but also as root exudates that can support many rhizosphere microbes during the active growing season of the cover crop. Other benefits attributed to cover cropping include improved N fertility by incorporating legumes as a cover crop, reduced soil compaction via deep-rooted plants, and reduced erosion by keeping a plant and its root system in the field year round. Of various crop rotation management practices, those that include cover crops sustain soil quality and productivity by enhancing soil C, N and microbial biomass, making them a cornerstone for sustainable agroecosystems. Nonetheless, very few studies have assessed the relationship between cover crop stands and their associated belowground microbial communities. Early research in unfertilized grasslands demonstrated that fungal communities respond positively to plant-derived C inputs, suggesting that inclusion of cover crops in a rotation may promote fungal community development. More recently, a field study tested this hypothesis by specifically examining the impact on soil microbial communities of eight fall-sown cover crop species grown singly and in multispecies mixtures following a spring oats (Avena sativa L.) cropping season and found that certain cover crops selectively favored particular microbial functional groups. Arbuscular mycorrhizal fungi were more abundant beneath oat and cereal rye (Secale cereale L.) cover crops, while non-AM fungi were positively associated with hairy vetch (Vicia villosa L.). Beyond positively affecting soil C and increasing the diversity of such beneficial fungi as arbuscular mycorrhiza, clover as a cover crop is often reported to suppress the relative abundance of pathogenic fungi. Contrarily, in a 2-year field study, cover crops reportedly increased overall phylogenetic diversity of fungi but did not change the relative abundance of saprophytes, symbionts or pathogens, implying that cover cropping does not always appear to contribute to functional changes in the fungal community.
## Reassessment of plant responsiveness to symbiosis
It is now increasingly evident that plants employ fine-tuned mechanisms to shape the structure and function of their microbiome, with different genotypes of the same plant species growing in the same soil yet associating with distinct microbial communities. This is demonstrated in the findings of, who clearly demonstrated the importance of intraspecific host variation in the association of chickpea cultivars with AM and non-AM fungi. Therefore, specific traits of a plant that modulate its microbiome should be considered as a trait for plant breeding.
Despite the obvious importance of beneficial microorganisms for plant growth and fitness, and the impact of plant genotype on shaping their microbiome composition, plant germplasm is typically screened in the absence of microbes, and the selection of best breeding lines made solely based on the interaction between plant genotype and performance under various abiotic factors. We propose that an a priori examination of the interaction between a plant genotype(s) and the symbiotic microbes upon which it likely depends is an important factor in the selection of plant breeding lines. It seems very likely that a subset of rejected germplasm could outperform others, but only when coupled with a beneficial microbe or microbiome . Arguably, current breeding and selection efforts most likely result in decoupling of the soil microbiome from plant fitness. As a result, modern varieties may have lost their ability to support diverse microbiomes and thus, fail to gain the most from these interactions.
It is now acknowledged that transitioning from a highly intensive mono-cropping system to a more diversified cropping system consisting of multiple host genotypes leads to increased bacterial and fungal diversity. Hence, future plant breeding efforts should incorporate plant characteristics that are related to microbiome diversity. For example, efforts focusing on manipulating plant root exudates likely play a critical role in selective recruitment of the rhizosphere microbiome . In support of this notion, it has been shown that plants can select which microbial populations receive the lion's share of root exudates, demonstrating a capacity by the host to refine its microbial composition. Hence, an unbiased screening of plant genotypes for responsiveness in the presence of a beneficial microbe or microbiome can set forth a new and potentially transformative paradigm in selecting microbes for plant growth promotion .
## Significance of mycorrhizas: a critical component of healthy soil rhizospheres
Mycorrhizae are mutualistic associations between soil fungi and plant roots that gradually evolved to be reciprocally beneficial to both partners. The benefits are generally assumed to involve an exchange of photosynthetically derived carbon from the host plant in exchange for soil nutrients provided by the foraging mycorrhiza. While likely true of arum-type arbuscular types of mycorrhizae, there are other types that can derive carbon from organic matter in the soil, or even "steal" it from one host plant to supply to another. A recent study has reported that in contrast to Arum maculatum, in which carbon is entirely derived from photo-assimilation, the green leaves of Paris quadrifolia contain a striking 50% carbon of fungal origin. Such partial mycoheterotrophy could thus potentially be widespread among the roughly 100,000 plant species that are known to develop a Paris-type AM, with far-reaching implications for our understanding of C trading in plantmicrobe communities. Exactly what the mycorrhiza gains from this interaction is still under debate, but benefits may involve a safe haven from the open, more competitive soil space and a second, more reliable carbon source.
Mycorrhizae not only shape plant communities, but they also affect the functional diversity of their cohabitants in the rhizospheric microbiome. The mycelium of mycorrhizal fungi transports plant-derived carbon into the soil in the form of sugars, amino acids and polyols to help sustain the microbiome. More recent studies focusing on soil microbial ecology revealed that mycorrhizal fungi mediate many diverse interactions within the soil "mycorrhizosphere, " including pathogens and mutualists that fix atmospheric nitrogen, take up phosphorus, produce vitamins, and/or protect against antagonists. The "ectomycorrhizosphere, " which forms a very specific interface between soil and many trees, hosts a large and diverse community of microorganisms that likely play roles in mineral weathering and solubilization processes. This carbon-rich mycorrhizosphere also supports large communities of root-associated microorganisms that further accelerate weathering of minerals by excreting organic acids, phenolic compounds, protons, and siderophores.
Similarly, the extraradical hyphae of arbuscular mycorrhiza provide a direct pathway for the translocation of photosynthetically derived carbon to the soil, leading to the development of nutrient-rich niches for other soil microorganisms, particularly bacteria. A quantitative real-time PCR method detected significantly higher 16S rDNA abundance in both the bulk and the rhizosphere soils of zucchini (Cucurbita pepo L.) inoculated with Acaulospora laevis and Glomus mosseae. Additionally, arbuscular mycorrhizae have been reported to increase the relative abundance of Firmicutes, Streptomycetes, Comamonadaceae, and Oxalobacteraceae inhabiting the mycorrhizosphere. While there is clear evidence that microbial communities in the rhizosphere function cohesively with their mycorrhizal partner in nutrient mobilization from soil minerals, nitrogen cycling and protection of plants against root pathogens, such bidirectional synergy is not always universal. There are reports that indicate suppressive effects of bacterial communities on mycorrhizal functioning and vice versa. While one study reportedthat soil with a higher abundance of Acidobacteria suppresses the normal functioning of extra-radical mycelium in arbuscular mycorrhizae, another study found that Glomus intraradices and Glomus mosseae suppressed most of the associated soil microbial community.
## A novel type of endophytic symbiont: the serendipitaceae
A diverse group of fungi in the Basidiomycota, the Serendipitaceae (formerly Sebacinales Group B)encompasses endophytes and lineages that repeatedly evolved ericoid, orchid and ectomycorrhizal types. Accordingly, in many natural ecosystems these fungi form mycorrhizal symbioses with an astounding variety of host plants -every mycorrhizal type, in fact, except for arbuscular. Previous research performed in our lab with a strain of this group, Serendipita vermifera, demonstrated plant growth-promoting properties in a variety of plants. Unfortunately, the agronomic utility of these fungi is hampered by the paucity of strains available, the large majority isolated from Australian orchids. We have begun to address this constraint by isolating the first North American strain of Serendipita, named Serendipita vermifera subsp. bescii NFPB0129, from the roots of a switchgrass plant in Ardmore, Oklahoma.
As mentioned above, soil organic matter has a tremendous influence on the biological, chemical, and physical properties of soils, making it a vital component of healthy agricultural systems. Whether a natural soil or an agricultural one, the release of the nutrients locked within SOM requires decomposers, primarily insects, fungi, and bacteria, to secrete organic acids and enzymes that can loosen and break down the cellulose and the recalcitrant lignin into nutritive forms that can be used by other microbes and plants. Unlike arbuscular mycorrhizae, which exchange inorganic, mineralized nutrients mined from the soil for carbon derived from host photosynthesis, members of the Serendipitaceae studied thus far have a complete arsenal of carbohydrate-active enzymes (CAZymes), representing approximately 4% of the entire gene set and rivaling the more well-studied saprophytic white and brown wood rotters, and much more than other symbiotic fungi. Additionally, genome analysis of S. bescii and S. vermifera suggests that Serendipitaceae fungi have the metabolic capacity to assimilate N from organic forms of N-containing compounds . We hypothesize that this carbohydrate-degrading enzyme complement endows these Serendipitaceae fungi with saprotrophic abilities. Unlike free-living decomposers that maintain a solitary lifestyle, seeking only dead or dying plant tissues as their source of subsistence, Serendipitaceae fungi seem to maintain a largely symbiotic lifestyle with the roots of living host plants. It currently is unclear whether there is expression of CAZymes while strains of Serendipita are in symbiosis with host plants, and if so, whether there is spatial or temporal separation from more mutualistic traits. Still, the capacity of some strains to form mycorrhizal relationships with orchids, where the seeds require carbon from the fungus for germination and often well into the plant's lifespan, suggests that these Serendipitaceae symbionts may be less of a carbon cost to their host plant. Presumably, this saved carbon could potentially be used for other symbiotic relationships or developmental processes. In any case, these intriguing fungi and their seemingly unlimited host range provide a novel symbiosis that could be used in a broad variety of cropping systems.
# Conclusion
Soil-dwelling microorganisms are critical components of soil health, itself a determinant of plant productivity and stress tolerance. Deploying microbes to improve agriculture productivity is an extremely attractive approach that is non-transgenic and can be viewed collectively as the extended plant genome. Because these same microbes can contribute to restoring soil health and productivity, they have a bright future in low-input, sustainable agriculture that extends beyond more classically defined plantmicrobe symbioses.
# Author contributions
PR and KC conceived and planned the overall idea of the review manuscript. PR, VL, JL, and KC wrote the manuscript. All authors contributed to the article and approved the submitted version. |
Is Preventative Long-Segment Surgery for Multi-Level Spondylolysis Necessary? A Finite Element Analysis Study
ObjectiveFor multi-level spondylolysis patients, surgeons commonly choose to fix all the segments with pars interarticularis defect even those without slippage and not responsible for clinical symptoms. In this study, we tried to study the necessity of the preventative long-segment surgery for the defected segment without slippage in treatment of multi-level spondylolysis patients from a biomechanical perspective.MethodWe established a bi-level spondylolysis model with pars defects at L4 and L5 segments, and simulated posterior lumbar interbody fusion (PLIF) and pedicle screw fixation at L5-S1 level. Then we compared the biomechanical changes at L4 segment before and after surgery in neutral, flexion, extension, lateral bending and axial rotation position.ResultsThe stress on L4 pars interarticularis was very similar before and after surgery, and reached the highest in axial rotation. The L3-L4 intradiscal pressure was almost the same, while L4-L5 intradiscal pressure changed a little in lateral bending (increase from 1.993 to 2.160 MPa) and axial rotation (decrease from 1.639 to 1.307 MPa) after surgery. The PLIF surgery caused a little increase of range of motion at adjacent L4-L5 and L3-L4 levels, but the change is very tiny (1 degree).ConclusionThe PLIF surgery will not cause significant biomechanical change at adjacent segment with pars defect in multi-level spondylolysis. On the contrary, excessive long-segment surgery will damage surrounding soft tissues which are important for maintaining the stability of spine. So a preventative long-segment surgery is not necessary for multi-level spondylolysis as long as there are no soft tissue degeneration signs at adjacent level.
# Introduction
Spondylolysis is defined as a defect in the pars interarticularis (pars for short) of the vertebral arch. The morbidity of spondylolysis is about 3-6% in general population [bib_ref] Spondylolysis: a critical review, Standaert [/bib_ref] , and even higher in athletes [bib_ref] Incidence and etiology of lumbar spondylolysis: review of the literature, Sakai [/bib_ref]. The most vulnerable segments are L5-S1(85-95%) and L4-L5 (5-15%) [bib_ref] Spondylolysis: a critical review, Standaert [/bib_ref]. When the defected vertebra develops to a forward displacement over its inferior vertebra, it's called spondylolisthesis. Spondylolisthesis is a common cause of low back pain, and surgery is the main treatment for spondylolisthesis with severe symptoms.
Multi-level spondylolysis is comparatively rarer. Ravichandran [bib_ref] Multiple lumbar spondylolyses, Ravichandran [/bib_ref] reported an incidence of about 1.5% amongst symptomatic patients. Sakai et al. [bib_ref] Incidence of lumbar spondylolysis in the general population in Japan based on..., Sakai [/bib_ref] reported an incidence of 0.03% in general Japanese population. However, a couple of papers introduced their experience of treating multi-level spondylolysis patients [bib_ref] Rehabilitation of a patient with a rare multi-level isthmic spondylolisthesis: a case..., Wong [/bib_ref] [bib_ref] Surgical treatment of double-level isthmic spondylolisthesis, Song [/bib_ref] [bib_ref] Multiple-level lumbar spondylolysis and spondylolisthesis, Liu [/bib_ref] [bib_ref] A case report of 3-level degenerative spondylolisthesis with spinal canal stenosis, Moo [/bib_ref]. Notably, Song et al.reported the surgical treatment of 54 multi-level spondylolisthesis patients in 8 years. So it is still an issue worth paying attention to. Some of the multi-level spondylolysis patients may only have one segment develops to spondylolisthesis and causes symptoms (responsible segment), the adjacent pars defects are just found casually by preoperative examination and were not responsible for the symptoms (innocent segment). Surgeons used to fix all the defected segments at the same time. It is reasonable to fix the responsible segment, but is a preventative fixation for the innocent segment without slippage necessary? There have been no biomechanical and clinical reports supporting previous surgeons' choice.
By reviewing previous literatures, we found that the probability for spondylolysis to develop to spondylolisthesis which needs to be treated by surgery is quite low [bib_ref] The natural history of spondylolysis and spondylolisthesis: 45-year follow-up evaluation, Beutler [/bib_ref] [bib_ref] Spondylolysis and spondylolisthesis in children and adolescents: I. Diagnosis, natural history, and..., Cavalier [/bib_ref]. So we assume that as long as the fixation of the responsible segment does not apply extra stress on the adjacent innocent segment, it can maintain its stability and does not need a preventative fixation.
It is difficult to set up a randomized controlled clinical trial due to the low incidence of multi-level spondylolysis. So we used finite element analysis (FEA) in this study to investigate the biomechanical influence of short-segment posterior lumbar interbody fusion (PLIF) on adjacent innocent segment in bi-level spondylolysis.
# Methods
A three dimensional (3D) FEA model of L3-S1 segments which had been established in our previous study was used, the validation of this model had been documented before [bib_ref] Finite element analysis on transforaminal lumbar interbody fusion treatment, Zhang [/bib_ref]. The model was modified to simulate two situations. Before surgery: bilateral pars defects were simulated at L4 and L5 segments to establish a bi-level spondylolysis model; Short-segment PLIF: short-segment surgery (PLIF coupled with bilateral pedicle screw fixation) was performed at L5-S1 level on the bi-level spondylolysis model. By comparing these two models, we could see whether there would be biomechanical changes at the defected L4 segment after fixation of L5-S1 level in bi-level spondylolysis.
## Finite element model
The Computed Tomography scan images of a normal male adult's lumbar spine (slice width 0.625 mm) were imported into Mimics 10.01 (Materialise NV, Belgium). A 3D model of L3-S1 vertebras and intervertebral discs was established after processes including thresholding, region growing, 3D modeling, and so on. The vertebras consisted of anterior centrum (cancellous and cortical bone) and posterior vertebral arch. The intervertebral discs consisted of anulus fibrosus and nucleus pulposus.
The transpedicular screw-rod system (screw: diameter 5.5mm, length 40.0mm; rod: diameter 6.0mm, length 51.0mm) and interbody fusion cage (dimension: 22.0mm×10.0mm×9.0mm) models were established by Pro/ENGINEER 2.0 (Parametric Technology Corporation, USA). Then the data were imported into Mimics.
Pars defects were simulated by creating gaps with width of 1mm [bib_ref] Buck's direct repair of lumbar spondylolysis restores disc stresses at the involved..., Sairyo [/bib_ref] [bib_ref] Three dimensional finite element analysis of the pediatric lumbar spine. Part II:..., Sairyo [/bib_ref]. Both the fractured ends could transmit stresses when they contacted each other during motion. The screw-rod system, interbody fusion cage and spine were assembled and Boolean operated by simulation in Mimics.
The spine was imported into ANSYS 10.0 (ANSYS, Inc. USA) for mesh generation. Then, it was imported back into Mimics to define the material properties referring to previously published literatures. The screw-rod system and cage were directly imported into ANSYS to mesh and define property [bib_ref] A biomechanical study of lumbar spondylolysis based on a threedimensional finite element..., Chosa [/bib_ref] [bib_ref] Biomechanical comparison of unilateral and bilateral pedicle screws fixation for transforaminal lumbar..., Chen [/bib_ref] [bib_ref] Finite element analysis of the lumbar spine with a new cage using..., Zhong [/bib_ref] [bib_ref] Biomechanical analysis of two-step traction therapy in the lumbar spine, Park [/bib_ref]. [fig_ref] Table 1: Material properties/values of the FEA models [/fig_ref] Two 3D finite element models were established through above process [fig_ref] Fig 1: Three-dimensional finite element models [/fig_ref]. Before surgery model consisted of 353587 total elements and 581704 total nodal points; Short-segment PLIF model consisted of 536318 total elements and 872853 total nodal points.
## Boundary and loading condition
All interfaces in the FEA models were assumed to be bonded except the facet joints which were applied with a surface-to-surface condition of which the friction coefficient was set as 0.1 [bib_ref] Biomechanical comparison of unilateral and bilateral pedicle screws fixation for transforaminal lumbar..., Chen [/bib_ref] [bib_ref] Biomechanical comparison of transforaminal lumbar interbody fusion with 1 or 2 cages..., Xu [/bib_ref]. The nodes of the inferior surface of S1 were completely fixed in all directions [bib_ref] Biomechanical evaluation of three surgical scenarios of posterior lumbar interbody fusion by..., Xiao [/bib_ref]. Loading force of 500N [bib_ref] Biomechanical evaluation of three surgical scenarios of posterior lumbar interbody fusion by..., Xiao [/bib_ref] [bib_ref] Applying a follower load delivers realistic results for simulating standing, Rohlmann [/bib_ref] was applied on superior surface of L3. Torque of 10NÁm was applied to simulate physiological activity in 5 directions: neutral, flexion, extension, lateral bending and axial rotation.
# Results
## Stress on l4 pars
As indicated in [fig_ref] Table 2: Comparison of stress on L4 pars, adjacent intradiscal pressure [/fig_ref] , the stress on L4 pars was smallest in neutral position, and rose a little in flexion, extension and lateral bending position. The L4 pars suffered the highest stress when the spine rotated. There was no significant difference observed after PLIF surgery was performed at L5-S1 level.
## Adjacent intradiscal pressure
The intradiscal pressure of L3-L4 and L4-L5 discs were listed in [fig_ref] Table 2: Comparison of stress on L4 pars, adjacent intradiscal pressure [/fig_ref]. The L3-L4 intradiscal pressure significantly increased when spine flexed forward, and decreased a lot when the spine extended. This change tendency was in accordance with the load distribution shift of spine in vivo, which demonstrated the validation of our model. The change pattern of L4-L5 intradiscal pressure was different from that of L3-L4 disc. The highest pressure presented in lateral bending, and the pressure in flexion and extension were close to each other. After PLIF surgery at L5-S1 level, the L3-L4 intradiscal pressure is very close to that of before surgery. The L4-L5
## Range of motion
The segmental angles and range of motion (ROM) of L3-L4, L4-L5 and L5-S1 were listed in and illustrated in [fig_ref] Fig 2: Comparison of displacement distribution in flexion and extension position [/fig_ref] The segmental angle is defined as the intersection angle of the superior end plate of two adjacent vertebras in sagittal plane. The ROM equals to the difference value of segmental angles between flexion and extension position. The segmental angel and ROM had a gradual increase trend from cranial to caudal level. After PLIF surgery, the ROM of L5-S1 disappeared, and that of L3-L4 and L4-L5 both increased only 1 degree.
# Discussion
The stability system of vertebral column consists of vertebras, intervertebral discs, ligaments and paravertebral muscles. In spondylolysis, the pars are fractured and detached, so there will almost be no force transferred when the two fractured ends have no contact. The stability of the defected segment will be mainly maintained by the intervertebral disc, ligaments and paravertebral muscles instead. As a chronic disease, adaptive changes might be developed to maintain its stability when the pars fractures. Beutler et al. [bib_ref] The natural history of spondylolysis and spondylolisthesis: 45-year follow-up evaluation, Beutler [/bib_ref] followed up 30 people diagnosed of pars defect by imageological examination for 45 years. Twelve of them never experienced slippage throughout the follow up period. For those who developed to spondylolisthesis, the average slippage degree is less than 25% (Meyerding grade I), and only three of them underwent lumbar spine surgery. At the final follow-up, the function and pain evaluation of these people had no significant difference with general population of matched age. It indicates that not all of pars defect will develop to spondylolisthesis, and only a few of them will cause severe symptoms that need surgery. Beutler's report indicates that healthy soft tissues (intervertebral disc, ligaments and paravertebral muscles) are commonly enough to stabilize the lumbar spine. So if the fixation and fusion of the displaced segment will significantly increase the stress on defected pars, intradiscal pressure and ROM of the adjacent segment which has pars defect but no displacement, then a preventative fixation of the adjacent segment for protection is needed. Otherwise, a short-segment fixation and fusion of the displaced segment is enough. The results of our study indicated that the biomechanical changes at L4 segment were little after a PLIF surgery was performed at L5-S1 level in bi-level spondylolysis. First, the stress on defected L4 pars didn't have a significant change in all directions of motion after surgery. So it means the surgery will not apply more stress on L4 pars. Second, the intradiscal pressure of discs either superior or inferior to the L4 vertebra didn't significantly increase after surgery. So it will not lead to higher risk of adjacent disc degeneration. Third, the changes of the ROM of L3-L4 and L4-L5 were very tiny after surgery. So it would not impose a heavier burden on the . Comparison of segmental angels and ROM of L3-L4, L4-L5 and L5-S1 (degree).
## L3-l4
L4-L5 L5-S1
Before surgery Short-segment PLIF Before surgery Short-segment PLIF Before surgery Short-segment PLIF intervertebral ligaments and paravertebral muscles to maintain stability. All these comparisons demonstrated that the fixation and fusion of L5-S1 level would not increase the risk for the defected L4 segment to develop to spondylolisthesis. Interestingly, the L4-L5 intradiscal pressure and ROM didn't increase a lot after a PLIF surgery was performed at L5-S1 level. This does not agree with previous studies which stated that the interbody fusion will increase the adjacent level intradiscal pressure and ROM [bib_ref] Effects of lumbar spinal fusion on the other lumbar intervertebral levels (three-dimensional..., Goto [/bib_ref] [bib_ref] Analysis of the effect of lumbar spine fusion on the superior adjacent..., Chosa [/bib_ref] [bib_ref] Stress analysis of the disc adjacent to interbody fusion in lumbar spine, Chen [/bib_ref]. We think it is because that previous studies were based on intact spine, while our study is the first to investigate the influence of PLIF on an adjacent segment with pars defect. As discussed above, the rigid bony connection will be lost when the pars fractures, the alternative soft tissue connection is resilient. Maybe this resilient connection mechanism can avoid stress concentration at the adjacent level. This hypothesis is partly supported by both experiment and clinical trials. It has been observed that if the adjacent segment above a rigid instrumented level was fixed with a semi-constrained instrument, the rise of ROM and intradiscal pressure at this adjacent segment could be reduced. A similar clinical approach which is called "Toppingoff technique" has been demonstrated to restrict the hyper-extension movement of adjacent segments, prevent back and forth movement of proximal vertebrae, and decrease loads of intervertebral disc and facet joints [bib_ref] Topping-off technique prevents aggravation of degeneration of adjacent segment fusion revealed by..., Zhu [/bib_ref]. However, the pars of the non-rigid fixed adjacent segment were intact in these studies, so the validation of our hypothesis needs to be further investigated.
According to the results of our study, if adjacent disc has no apparent degeneration sign in preoperative examinations, a short-segment PLIF should be recommended for the multi-level spondylolysis patients. Short-segment fusion has many advantages compared with long-segment fusion. It can preserve more ROM, has less chance of adjacent level degeneration, less surgery time and blood loss, less cost, et al. Some surgeons chose direct repair instead of interbody fusion in order to preserve the segmental mobility, but the bony union rate of direct repair is not satisfying [bib_ref] Direct repair surgery with screw fixation for young patients with lumbar spondylolysis:..., Lee [/bib_ref]. The most negative influence of long-segment surgery is that it will cause unnecessary iatrogenic injury of ligaments and paravertebral muscles, which are very important stabilizing structures especially in spondylolysis [bib_ref] Role of ligaments and facets in lumbar spinal stability, Sharma [/bib_ref]. Additionally, postoperative rehabilitation exercises should be emphasized, so as to enhance the strength of paravertebral muscles.
Our study justified the short-segment fixation for multi-level spondylolysis patients for the first time. In previous reports, all surgeons chose to fix all the defected segments including those without displacement. We think they concerned more about risk aversion while making this choice. Surgeons will prone to overtreatment in case of operative complications. Our study provides biomechanical proof to support short-segment PLIF which can maximize the benefit of these patient. It will be helpful for surgeons to choose proper surgical method for these patients in the future.
There are a few things that should be noted. First, this is purely a FEA study. Although this method is wildly accepted for studying the biomechanical effects in vivo, the parameters or boundary conditions may not perfectly mimic the real properties, which may cause biases. We will try to use biomechanical test on cadavers to verify our FEA results later on. Second, we didn't include a long-segment PLIF model, because the purpose of our study is not to determine whether long and short-segment PLIF is better than the other. Third, segmental slippage was not simulated in our study. Because the slippage severity is highly individualized, the influence of slippage is very complicated for comparison and is beyond the content of this study.
# Conclusion
In this study, we established a FEA model with bi-level spondylolysis at L4 and L5 segments, then simulated a PLIF surgery at L5-S1 level. We performed three dimensional finite element analysis to study the biomechanical changes before and after surgery on this model. Neutral, flexion, extension, lateral bending and rotation position were simulated in both models. The stress on L4 pars, L3-L4 and L4-L5 intradiscal pressure, ROM of L3-4, L4-L5 and L5-S1 levels were evaluated in each position. The results showed that the short-segment PLIF at L5-S1 level would not cause significant biomechanical changes at adjacent L4 segment with pars defect in the bilevel spondylolysis FEA model. So a preventative long-segment fixation is not necessary in multilevel spondylolysis as long as there are no soft tissue degeneration signs at adjacent levels.
[fig] Fig 1: Three-dimensional finite element models. (a) Before surgery model; (b) Short-segment PLIF model. doi:10.1371/journal.pone.0149707.g001 [/fig]
[fig] Fig 2: Comparison of displacement distribution in flexion and extension position. (a) Before surgery model; (b) Short-segment PLIF model. doi:10.1371/journal.pone.0149707.g002 Is Preventative Long-Segment Surgery for Multi-Level Spondylolysis Necessary? PLOS ONE | DOI:10.1371/journal.pone.0149707 February 26, 2016 [/fig]
[table] Table 1: Material properties/values of the FEA models. [/table]
[table] Table 2: Comparison of stress on L4 pars, adjacent intradiscal pressure (MPa). [/table]
|
MRI Findings of Coexistence of Ectopic Neurohypophysis, Corpus Callosum Dysgenesis, and Periventricular Neuronal Heterotopia
Ectopic neurohypophysis is a pituitary gland abnormality, which can accompany growth hormone deficiency associated with dwarfism. Here we present magnetic resonance imaging (MRI) findings of a rare case of ectopic neurohypophysis, corpus callosum dysgenesis, and periventricular neuronal heterotopia coexisting, with a review of the literature.
# Introduction
Ectopic neurohypophysis may be accompanied by midline brain defects and some degree of endocrine disorders. The associated endocrine disease varies from isolated growth hormone (GH) deficiency to multiple anterior pituitary hormone deficiencies, but posterior pituitary function remains unchanged. Anterior pituitary hormone deficiencies may present with varied clinical manifestations like growth failure, central hypothyroidism, or underdeveloped secondary sexual characters. At the same time, the disease starting with GH deficiency can progress to pan hypopituitarism with time. [bib_ref] Hypothalamic-pituitary dysfunction ingrowth hormone-deficient patients with pituitary abnormalities, Maghnie [/bib_ref] Neurohypophysis is observed as a bright dot behind normal posterior lobe of adenohypophysis on T1-weighted magnetic resonance (MRI) images. Ectopic neurohypophysis is specifically monitored at the level of the median eminence of the hypothalamus in imaging studies. Because of its rareness, we present radiological findings of a case that showed coexistence of ectopic neurohypophysis, corpus callosum dysgenesis, and periventricular neuronal heterotopia, with accompanying literature review.
## Case report
A 21-year-old female patient was admitted to the neurology clinic of our hospital with complaints of forgetfulness, twitching of eyes, and short stature. The patient's height was 135 cm and weight was 45 kg. Her past medical history was unremarkable. Laboratory measurements for the pituitary hormones uncovered a very low level of GH (0.06 ng/ml), while the other hormones and their levels were normal [ [fig_ref] Table 1: Serum levels of anterior pituitary hormones in the patient [/fig_ref] ]. In MR images of the brain and pituitary (1.5-T system, GE Medical Systems Signa, Milwaukee, WI), the neurohypophysis was seen at the ectopic site as a bright dot at the level of the optic chiasm on coronal T1 images. Infundibular part was thinner than usual [ [fig_ref] Figure 1: 21-year-old female patient with complaints of forgetfulness, twitching of eyes, and short... [/fig_ref] ]. At the same time, partial dysgenesis of corpus callosum and periventricular neuronal heterotopia were observed on the MR images [ [fig_ref] Figure 2: 21-year-old female patient with complaints of forgetfulness, twitching of eyes, and short... [/fig_ref] ]. Ophthalmologic examinations for a possible involvement of the optic discs, including visual acuity test, visual field test, and dilated fundoscopy were completely normal. Also, the systemic evaluation of the patient for other possible associations was normal.
# Discussion
The pituitary gland consists of two portions: The adenohypophysis and the neurohypophysis. The neurohypophysis comprises the posterior pituitary lobe, the infundibulum, and the median eminence of the hypothalamus. The adenohypophysis and the neurohypophysis differ from each other, both embryologically and histologically. The neurohypophysis is formed by the evagination of neural tissue from the floor of the third ventricle. It consists of the distal axons of the hypothalamic magnocellular neurones that shape the neurohypophysis. After its downward migration, it is encapsulated together with the ascending ectodermal cells of Rathke's pouch, which form the adenohypophysis. The pituitary gland acts as an endocrine gland under control of the hypothalamus via the infundibular stalk. [bib_ref] Developmental abnormalities of the posterior pituitary gland, Di Iorgi [/bib_ref] Regardless of the underlying cause, the ectopic neurohypophysis results from incomplete downward growth and fusion of the neurohypophysis with the adenohypophysis in the sella. Possible pathophysiologic mechanisms for ectopic neurohypophysis development are transection of the pituitary stalk followed by hypertrophy of the proximal axons with subsequent reorganization. Also trauma of the pituitary stalk due to vascular accidents, anoxia, or compression injuries have been suggested as acting through similar mechanisms leading to reorganization of the proximal neurons of the neurohypophysis. [bib_ref] Pituitary Stalk and ectopic hyperintense T1 signal on magnetic resonance imaging, Ultmann [/bib_ref] Although early studies describe a traumatic cause in the etiology of the ectopic neurohypophysis, more recent studies favor a genetic basis, supported by rare familial cases of ectopic neurohypophysis. [bib_ref] MR findings in growth hormone deficiency: Correlation with severity of hypopituitarism, Kornreich [/bib_ref] [bib_ref] Growth hormone deficiency with ectopic neurohypophysis: Anatomical variations and relationship between the..., Chen [/bib_ref] The ectopic neurohypophysis cases have been reported in idiopathic GH deficiency or in subjects with HESX1, LHX4, and SOX3 gene mutations. [bib_ref] Developmental abnormalities of the posterior pituitary gland, Di Iorgi [/bib_ref] Again, a few earlier reports have suggested that ectopic neurohypophysis may share a similar pathogenesis as septo-optic dysplasia, which classically has features of optic nerve hypoplasia, hypothalamic-pituitary dysfunction, and agenesis of the septum pellucidum. The HESX1 homebox gene has been implicated in both of these conditions. [bib_ref] Ectopic posterior pituitary lobe and periventricular heterotopia: Cerebral malformations with the same..., Mitchell [/bib_ref] The other well-known malformations associated with ectopic neurohypophysis include Chiari 1 malformation, agenesis of the corpus callosum, persistent cranio-pharyngeal canal, Kallmann syndrome, basilar impression, medial deviation of the carotid arteries, microcephaly, cerebellar atrophy, and vermian dysplasia. Associated ophthalmic and midline facial abnormalities, such as a single central incisor tooth, are typical, and non-midline somatic conditions, including cardiac and musculoskeletal abnormalities, have been noted. Breech presentation and neonatal hypoglycemia are also commonly reported. [bib_ref] MR findings in growth hormone deficiency: Correlation with severity of hypopituitarism, Kornreich [/bib_ref] [bib_ref] Growth hormone deficiency with ectopic neurohypophysis: Anatomical variations and relationship between the..., Chen [/bib_ref] [bib_ref] Ectopic posterior pituitary lobe and periventricular heterotopia: Cerebral malformations with the same..., Mitchell [/bib_ref] As a result of ectopic neurohypophysis and associated stalk abnormalities, factors released by the hypothalamus, which usually travel down the portal circulation to the anterior pituitary can only reach their target via the general circulation. The patients with an ectopic neurohypophysis located at the median eminence and no visible pituitary stalk present a more severe form of the disease, with multiple anterior hormone deficiencies. Whereas visibility of the pituitary stalk and an ectopic neurohypophysis located at some point along the stalk is associated with isolated GH deficiency. [bib_ref] Pituitary Stalk and ectopic hyperintense T1 signal on magnetic resonance imaging, Ultmann [/bib_ref] [bib_ref] Ectopic posterior pituitary lobe and periventricular heterotopia: Cerebral malformations with the same..., Mitchell [/bib_ref] In our patient, there were radiological findings of coexistence of ectopic neurohypophysis, corpus callosum partial dysgenesis, and periventricular neuronal heterotopia. Dysgenesis of the corpus callosum represents an in utero developmental anomaly, which may be complete (agenesis) or partial. It may occur in isolation or in association with other central nervous system or systemic malformations.
Isolated partial dysgenesis of the corpus callosum is often asymptomatic.Periventricular heterotopia is the most common form of gray matter heterotopia and is characterized by nodules of gray matter located immediately beneath the ependyma of the lateral ventricles. For this reason, it is also known as subependymal heterotopia. They are most commonly seen in the region of the trigones and occipital horns. They are most frequently associated with epilepsy and sometimes developmental delay. [bib_ref] Disorders of cortical formation: MR imaging features, Razek [/bib_ref] Previous reports of cases with coexistent periventricular heterotopia and ectopic neurohypophysis suggest both conditions can share a common underlying causative mechanism, even though pituitary development occurs earlier in fetal life than does neuronal migration. [bib_ref] Ectopic posterior pituitary lobe and periventricular heterotopia: Cerebral malformations with the same..., Mitchell [/bib_ref] Another study has found HESX1 mutations in only 5 of 93 patients with ectopic neurohypophysis, indicating that other unrecognized genes or local environmental factors are likely to be involved in the ectopic neurohypophysis development. [bib_ref] Molecular effects of novel mutations in Hesx1/HESX1 associated with human pituitary disorders, Brickman [/bib_ref] In ectopic neurohypophysis cases, normal neurohypophysis bright spot is absent in the sella and there is high T1 signal at the median eminence (floor of third ventricle) or along infundibuler stalk, instead of normal sellar position, on unenhanced T1-weighted images. Infundibuler stalk may be tiny or absent. Although controversy still exists on the source of this distinct signal, it has most recently been attributed to phospholipids within the walls of the secretory vesicles containing the arginine vasopressin neurophysin complex. [bib_ref] Posterior pituitary ectopia: Another hint toward a genetic etiology, Maintz [/bib_ref] Distinction of the ectopic neurohypophysis from dermoid tumor or lipoma can be confirmed by images obtained using fat saturation. [bib_ref] Molecular effects of novel mutations in Hesx1/HESX1 associated with human pituitary disorders, Brickman [/bib_ref] [bib_ref] Posterior pituitary ectopia: Another hint toward a genetic etiology, Maintz [/bib_ref] In this article, because of its rareness, we present MRI findings of concurrence of ectopic neurohypophysis, dysgenesis of the corpus callosum, and periventricular neuronal heterotopia.
No systemic association was detected in our case.
# Conclusion
Ectopic neurohypophysis can be detected by MRI.
Radiologists should carefully evaluate MR images in patients with epilepsy or hormone deficiency for the possibility of ectopic neurohypophysis and neuronal heterotopia. In the presence of ectopic neurohypophysis, other associated cerebral malformations should be thoroughly investigated. : 21-year-old female patient with complaints of forgetfulness, twitching of eyes, and short stature later diagnosed with coexistence of ectopic neurohypophysis, corpus callosum dysgenesis, and periventricular neuronal heterotopia. Axıal T2-weighted MRI image reveals an appearance that is compatible with periventricular-subependymal heterotopic neuronal tissue (arrows).
[fig] Figure 1: 21-year-old female patient with complaints of forgetfulness, twitching of eyes, and short stature later diagnosed with coexistence of ectopic neurohypophysis, corpus callosum dysgenesis, and periventricular neuronal heterotopia. A coronal non-contrast T1-weighted MRI image reveals a hyperintense ectopic posterior pituitary adjacent to the optic chiasm, at the level of the median eminence of the hypothalamus (arrows). The bright dot in the ectopic site indicates neurohypophysis. [/fig]
[fig] Figure 2: 21-year-old female patient with complaints of forgetfulness, twitching of eyes, and short stature later diagnosed with coexistence of ectopic neurohypophysis, corpus callosum dysgenesis, and periventricular neuronal heterotopia. Sagıttal T2-weighted MRI image reveals partial dysgenetic view of the corpus callosum and protrusion of cingulated gyrus (arrow). [/fig]
[table] Table 1: Serum levels of anterior pituitary hormones in the patient [/table]
|
Gender-, Age- and Region-Specific Characterization of Vertebral Bone Microstructure Through Automated Segmentation and 3D Texture Analysis of Routine Abdominal CT
Purpose: To identify long-term reproducible texture features (TFs) of spinal computed tomography (CT), and characterize variations with regard to gender, age and vertebral level using our automated quantification framework.Methods:We performed texture analysis (TA) on baseline and follow-up CT (follow-up duration: 30-90 days) of 21 subjects (8 females, 13 males, age at baseline 61.2 ± 9.2 years) to determine long-term reproducibility. TFs with a long-term reproducibility error D rel <5% were further analyzed for an association with age and vertebral level in a cohort of 376 patients (129 females, 247 males, age 62.5 ± 9.2 years). Automated analysis comprised labeling and segmentation of vertebrae into subregions using a convolutional neural network, calculation of volumetric bone mineral density (vBMD) with asynchronous calibration and TF extraction. Variance global measures the spread of the gray-level distribution in an image while Entropy reflects the uniformity of gray-levels. Short-run emphasis (SRE), Long-run emphasis (LRE), Run-length non-uniformity (RLN) and Run percentage (RP) contain information on consecutive voxels of a particular greylevel, or grey-level range, in a particular direction. Long runs (LRE) represent coarse texture while short runs (SRE) represent fine texture. RLN reflects similarities in the length of runs while RP reflects distribution and homogeneity of runs with a specific direction.Results: Six of the 24 extracted TFs had D rel <5% (Variance global , Entropy, SRE, LRE, RLN, RP), and were analyzed further in 4716 thoracolumbar vertebrae. Five TFs (Variance global ,SRE,LRE, RLN,RP) showed a significant difference between genders (p<0.001), potentially being caused by a finer and more directional vertebral trabecular microstructure in females compared to males. Variance global and Entropy showed a significant increase from the thoracic to the lumbar spine (p<0.001), indicating a higher degree and earlier initiation of trabecular microstructure deterioration at lower spinal levels. The four higher-order TFs showed significant variations between spine regions without a clear directional gradient (p ≤ 0.001-0.012). No TF showed a clear age dependence. vBMD differed significantly between genders, age groups and spine regions (p ≤ 0.001-0.002).Conclusion: Long-term reproducible CT-based TFs of the thoracolumbar spine were established and characterized in a predominantly older adult study population. The gender-, age-and vertebral-level-specific values may serve as foundation for osteoporosis diagnostics and facilitate future studies investigating vertebral microstructure.
# Introduction
Texture analysis (TA) is an emerging subfield of radiomics, representing a non-invasive and objective method for the quantitative assessment of medical images. Texture features (TFs) can be used to quantitatively characterize image properties, such as uniformity, heterogeneity and randomness, as well as repetitive image patterns. TA has the potential to enable the extraction of additional diagnostic, predictive and prognostic information beyond what is visually perceptive [bib_ref] Radiomics and Deep Learning in Clinical Imaging: What Should We do?, Choi [/bib_ref]. Traditionally, applications of TA include neuroimaging, musculoskeletal and oncological imaging, e.g. to assess tumor heterogeneity [bib_ref] Imaging Intratumor Heterogeneity: Role in Therapy Response, Resistance, and Clinical Outcome, O'connor [/bib_ref]. TA has been performed across different anatomical regions and modalities, with CT being the modality used most frequently, arguably due to its multitude of clinical applications, broad availability and data quality. Overall, musculoskeletal applications have been based on radiographic, magnetic resonance imaging (MRI) [bib_ref] Vertebral Bone Marrow Heterogeneity Using Texture Analysis of Chemical Shift Encoding-Based MRI:..., Dieckmeyer [/bib_ref] and CT data [bib_ref] Vertebral Body Insufficiency Fractures: Detection of Vertebrae at Risk on Standard CT..., Muehlematter [/bib_ref] [bib_ref] Opportunistic Osteoporosis Screening in Multi-Detector CT Images via local classification of textures, Valentinitsch [/bib_ref].
Osteoporosis is a systemic disorder of bone metabolism characterized by reduced bone mineralization and microarchitectural deterioration of osseous tissue [bib_ref] The Pathophysiology and Treatment of Osteoporosis, Drake [/bib_ref]. Bone health and the closely associated vertebral fracture risk are dependent on bone mineral density (BMD) and bone microstructure which is primarily defined by the threedimensional (3D) trabecular bone architecture as suggested by high correlations with micro-CT as reference [bib_ref] Osteoporosis Imaging: State of the Art and Advanced Imaging, Link [/bib_ref] [bib_ref] X-Ray-Based Quantitative Osteoporosis Imaging at the Spine, Loffler [/bib_ref]. Volumetric BMD (vBMD) can be measured on a vertebraspecific level with high spatial resolution using quantitative CT (QCT), while CT-based TA enables the evaluation of trabecular bone microstructure. Therefore, osteoporosis represents one of the most promising clinical applications of TA. In a small study cohort, showed that TA is feasible for opportunistic osteoporosis screening and can discriminate subjects with and without VFs accurately while exhibiting acceptable long-term scan-rescan reproducibility. However, the size of the used study population does not enable the derivation of TF reference values [bib_ref] Feasibility of Opportunistic Osteoporosis Screening in Routine Contrast-Enhanced Multi Detector Computed Tomography..., Mookiah [/bib_ref]. Muehlematter et al. used machine learning algorithms on CT-based TA to better predict incident VFs [bib_ref] Vertebral Body Insufficiency Fractures: Detection of Vertebrae at Risk on Standard CT..., Muehlematter [/bib_ref]. Through the combination of 3D-TA and regional vBMD, Valentinitsch et al. could improve the classification of prevalent VFs. In this study, the eventually analyzed TFs were selected from a variety of features, using an exponential search based on the Gini Index and subsequently used for classification of VFs which represent the most relevant clinical outcome of osteoporosis. However, the actual TF values were not reported [bib_ref] Opportunistic Osteoporosis Screening in Multi-Detector CT Images via local classification of textures, Valentinitsch [/bib_ref]. In spine imaging, commonly used TFs include first-order, second-order and higher-order statistical features, and reliable extraction of specific TFs was derived from sagittal reformations of up to 3 mm slice thickness in routine abdominal multidetector CT (MDCT) scans with administration of intravenous contrast medium (IVCM) [bib_ref] Feasibility of Opportunistic Osteoporosis Screening in Routine Contrast-Enhanced Multi Detector Computed Tomography..., Mookiah [/bib_ref]. However, TA for the evaluation of the osseous microstructure of the spine and ensuing bone health has not been standardized, yet.
One essential and time-consuming part of the TF extraction process is the segmentation to create individual regions of interest (ROIs) of the vertebral bodies, which has usually been performed manually on single slices. However, recent advancements in deep learning, in particular in convolutional neural networks, enable a standardized 3D segmentation by a fully automated pipeline [bib_ref] VerSe: A Vertebrae Labelling and Segmentation Benchmark for Multi-Detector CT Images, Sekuboyina [/bib_ref] [bib_ref] A Vertebral Segmentation Dataset With Fracture Grading, Loffler [/bib_ref] [bib_ref] Verse: A Vertebrae Labelling and Segmentation Benchmark for Multi-Detector CT Images, Sekuboyina [/bib_ref]. In order to discriminate TF values indicative of osteoporosis from normal values, and to enable longitudinal analyses of vertebral microstructure, the characterization of reproducible TF values is needed and can therefore be considered a prerequisite for diagnostic and therapeutic decisions in osteoporosis that use TFs.
Therefore, purposes of our study were (i), to identify CTbased TFs of vertebral bone with high long-term reproducibility which would be particularly important for longitudinal studies in the context of intervention monitoring, and (ii), to generate an automated standardized pipeline for segmentation and TA of the spine to establish characteristic TF values and determine physiological variations of the identified highly reproducible TFs with regard to gender, age and vertebral level in a mainly older adult population.
# Materials and methods
## Subjects
First, for the assessment of long-term reproducibility, a cohort of 21 Caucasian patients who received baseline (BL) and follow-up (FU) routine abdominal MDCT scans with IVCM and a FU duration between 30 and 90 days were identified. Clinical indications were oncologic staging for the BL scans, and ruling out postoperative complications (e.g. anastomotic insufficiencies, fistulas, infections) for the FU scans. For each TF and vertebra, the relative difference between BL and FU measurement D rel was calculated as a measure of long-term reproducibility,
[formula] D rel = 2 TF FU − TF BL TF FU + TF BL , [/formula]
with TF FU and TF BL denoting the measurements of a certain TF at FU and BL, respectively, and averaged across all vertebrae (T4 -L5) for each patient. All TFs showing a reproducibility error D rel < 5% were investigated further with regard to gender, age and vertebral level in a larger patient cohort as described in the following.
Second, 376 Caucasian patients who received routine abdominal MDCT scans with IVCM were analyzed. Clinical indication was oncologic follow-up to rule out tumor recurrence. For each of these patients, all vertebrae from T4 to L5, completely included in the field of view (FOV), were analyzed.
All patients were retrospectively identified in our hospital's digital picture archiving and communication system (PACS). Inclusion criteria were the availability of an MDCT scan including the spine at the same scanner with a specific protocol as outlined below. Exclusion criteria were VFs, osteoporosis, history of bisphosphonate or other bone metabolism-influencing therapy, osseous metastases as well as hematological or other metabolic bone disorders. Osteoporotic vertebrae were defined as vBMD < 80 mg/cm 3 according to the cutoff values suggested for spine QCT measurements by the American College of Radiology (ACR). The present study was approved by the local institutional review board. The requirement of the written consent was waived due to the retrospective nature of the study.
## Ct image acquisition
All CT data was acquired on the same 64-row MDCT scanner (Somatom Sensation Cardiac 64; Siemens Medical Solutions, Erlangen, Germany). A standardized protocol with the following dedicated scanning parameters was applied. Average tube voltage: 120 kVp, adapted tube load: 200 mAs (averaged), minimum collimation of 0.6 mm. An intravenous contrast agent (Imeron 400, Bracco, Konstanz, Germany) was administered for each examination following a standard protocol using a high-pressure injector (Fresenius Pilot C; Fresenius Kabi, Bad Homburg, Germany) with a delay of 70 s, a flow rate of 3 mL/s, and a body weight-dependent dose (80 mL for body weight ≤ 80 kg, 90 mL for body weight > 80 kg and ≤ 100 kg, and 100 mL for body weight > 100 kg). Oral contrast medium (1000 mL of Barilux Scan; Sanochemia Diagnostics, Neuss, Germany) was given before each CT scan. Sagittal reformations of the spine with a slice thickness of 3 mm and reconstructed with a standard bone kernel were used for TA.
## Automated image segmentation
The vertebrae T4 to L5 were automatically segmented in the MDCT images using a deep learning-driven framework (https:// anduin.bonescreen.de). The freely available tool identifies and labels each vertebra in a fully automated process and creates corresponding segmentation masks. Additionally, for each vertebra, a defined set of segmentation masks of subregions, including the trabecular compartment of the vertebral body among others, is created [fig_ref] FIGURE 1 |: Exemplary illustration of the automated labeling and segmentation [/fig_ref] [bib_ref] VerSe: A Vertebrae Labelling and Segmentation Benchmark for Multi-Detector CT Images, Sekuboyina [/bib_ref] [bib_ref] A Vertebral Segmentation Dataset With Fracture Grading, Loffler [/bib_ref]. The fully automatically generated labels and segmentation masks of all vertebrae were checked visually by a radiologist with two years of experience in spine imaging, and manually corrected if necessary. In total, labels and segmentations in 53 vertebral bodies of the 376 analyzed patients were manually corrected. Causes of imperfect labeling and segmentation were Schmorl nodes (33 vertebrae), severe degenerative changes [bib_ref] Axial QCT: Clinical Applications and New Developments, Link [/bib_ref] , partial block vertebrae (3), hemivertebrae (1), or thoracolumbar and lumbosacral transitional vertebrae (7).
# Texture analysis
All TFs were calculated for the ROI corresponding to the trabecular compartment of each segmented vertebral body. The extracted TFs included three global features, also referred to as first-order statistical moments, which are computed by gray-level histogram analysis, eight second-order features, based on graylevel co-occurrence matrix (GLCM) analysis, and 13 higherorder features, based on gray-level run-length matrix (GLRLM) analysis. In total, 24 TFs were extracted which quantify textural patterns (e.g. fine, coarse, smooth, or irregular) in an image. Each GLCM entry represents the probability of gray-level co-occurrence between voxel pairs at a given vector direction and a fixed length of 1 between the voxel pair. Each GLRLM entry is the probability of voxel occurrences of a specific gray-level for a possible run-length. A gray-level run is a set of successive voxels with identical gray-level values arranged collinearly along a given vector direction, and the run length is defined as the number of voxels in it. GLRLM features are computed based on the occurrence and distribution of such runs in a given image and measure the directional changes in the GLCM [bib_ref] Texture Analysis Methods for Medical Image Characterisation, Nailon [/bib_ref]. In total, there are 13 direction vectors with a displacement of (dx, dy, dz). The matrices of all 13 directions were averaged and normalized before calculating the second-order GLCM and higher-order GLRLM indices. GLCM is an (n × n) matrix where n is the number of gray-levels in each image. GLRLM is an (n × m) matrix, where n is the number of gray-levels in the scan, and m is the run length.gives an overview of the analyzed TFs together with descriptions of the quantified image properties.
In order to generate isotropic volumes of the image datasets necessary for comparable TA results, cubic interpolation was used. To prevent sparseness, gray-level quantization was performed using the normalized gray-levels (scale 0 to 1) of the ROI corresponding to the trabecular compartment of each vertebral body. All steps of the TA were performed with MATLAB (version R2021a; MathWorks Inc., Natick, MA, USA) using a modified version of a publicly available radiomics toolbox (https://github.com/mvallieres/radiomics) [bib_ref] El Naqa I. A Radiomics Model From Joint FDG-PET and MRI Texture..., Vallieres [/bib_ref].
vBMD Extraction vBMD was calculated for the ROI corresponding to the trabecular compartment of each segmented vertebral body applying asynchronous calibration to convert CT attenuation, measured in Hounsfield units (HU), to vBMD. Asynchronous calibration uses bone-equivalent density phantoms to generate HU-to-BMD conversion equations for a specific CT scanner and acquisition protocol. Previously established HU-to-BMD conversion equations were applied in this study using a phantom with hydroxyl-apatite inserts of known density in mg/cm³ (Anthropomorphic Abdomen Phantom, QRM Quality Assurance in Radiology and Medicine) [bib_ref] VerSe: A Vertebrae Labelling and Segmentation Benchmark for Multi-Detector CT Images, Sekuboyina [/bib_ref]. Linear correction equations for the presence of IVCM in portal-venous phase were applied to avoid vBMD bias [bib_ref] Bone Mineral Density Estimations From Routine Multidetector Computed Tomography: A Comparative Study..., Kaesmacher [/bib_ref].
# Statistical analysis
Statistical analysis was performed with SPSS (SPSS Inc., Chicago, IL, USA) and MATLAB (version R2021a; The Mathworks, Natick, MA, USA) using a two-sided level of significance of 0.05 for all statistical tests. The Kolmogorov-Smirnov test indicated normally distributed data for the analyzed TFs and vBMD, irrespective of gender. The association between vBMD and the analyzed TFs was determined using Pearson correlation coefficient (r). Independent t-tests were used to test for significant gender-dependent differences of TFs and vBMD. Analysis of variance (ANOVA) was used to test for significant differences across age groups and vertebral levels, respectively. For all tests, Bonferroni correction was applied to adjust for multiple comparisons.
# Results
## Long-term reproducibility of tfs
For each of the 21 included patients (8 females, 13 males, mean age at BL = 61.2 ± 9.2 years, age range at BL = 43.7 -71.6 years), T4 to L5 were analyzed, resulting in a total of 294 analyzed vertebrae. The mean follow-up duration, which is the time interval between the BL and FU scan, was 59 ± 11 days (range: 34 -77 days). Of the 24 analyzed TFs, four features showed a relative difference between FU and BL measurements (D rel ) below 1%, and two features between 1% and 5% [fig_ref] TABLE 2 |: Relative difference D rel between baseline and follow-up measurement of the analyzed... [/fig_ref]. In total, six TFs showed a reproducibility error D rel < 5% (Variance global , Entropy, SRE, LRE, RLN, RP) and were further analyzed with regard to gender, age and vertebral level.
## Tf variations with regard to gender, age and vertebral level
In total, 376 patients (129 females, 247 males, mean age = 62.5 ± 9.2 years, age range = 39.0 -88.0 years) were included, and stratified into four age groups for each gender, respectively: <50 years, 50-59 years, 60-69 years, and ≥70 years [fig_ref] FIGURE 2 |: Patient age vs vBMD [/fig_ref] and [fig_ref] TABLE 3 |: Patient demographics stratified by age groups [/fig_ref]. In total, vertebral bodies from T4 to L5 were included. In total, labels and segmentations were manually corrected in 53 of the 4716 vertebral bodies, resulting in a correction ration of 1.1%. The analyzed vertebral bodies were stratified into three spine regions: mid thoracic spine (T4-T8), lower thoracic spine (T9-T12) and lumbar spine (L1-L5). TF values with regard to gender, age and vertebral level are summarized in Tables 4-7. All of the six analyzed TFs, except Entropy, showed a statistically significant difference between genders [fig_ref] TABLE 8 |: Texture features and vBMD, averaged from T4-L5 and grouped by gender [/fig_ref].
Variance global and Entropy showed a statistically significant increase in cranio-caudal direction along the spine across all subjects as well as for both genders separately and [fig_ref] TABLE 5 |: Mean and standard deviation [/fig_ref]. All other TFs (SRE, LRE, RLN and RP) showed statistically significant differences between spine regions across all subjects [fig_ref] TABLE 9 |: Pairwise significant differences of texture features between spine regions across all subjects [/fig_ref]. SRE, LRE, RLN and RP showed also statistically significant differences between spine regions for both genders separately, except for LRE, RLN and RP between T9-T12 and L1-L5 in females and [fig_ref] TABLE 5 |: Mean and standard deviation [/fig_ref].
For both males and females, certain TFs showed a statistically significant difference between age groups, however without a clear directional trend and [fig_ref] TABLE 6 |: Mean and standard deviation [/fig_ref].
vBMD showed a statistically significant difference between genders [fig_ref] TABLE 8 |: Texture features and vBMD, averaged from T4-L5 and grouped by gender [/fig_ref] , a statistically significant decrease with age as well as statistically significant differences between spine regions for both males and females .
Linear correlation analysis revealed statistically significant (p < 0.05) low positive associations (r vBMD,Entropy = 0.39, r vBMD,SRE = 0.31, r vBMD,RLN = 0.31, r vBMD,RP = 0.31), and very low to low negative associations (r vBMD,Varianceglobal = -0.19, r vBMD,LRE = -0.31) between the analyzed TFs and vBMD, respectively.
# Discussion
The present study performed automated spine segmentation and TF extraction of vertebral bodies in routine abdominal MDCT scans in a large-scale patient cohort. In a preceding analysis, long-term , gray-level co-occurrence matrix (GLCM)-based, and gray-level run-length matrix (GLRLM)-based texture features and descriptions.
## Category
Texture feature Description (What is quantified)? Reference reproducibility of a set of commonly utilized TFs was evaluated. Six TFs with high reproducibility were identified and selected for further analysis. To the best of our knowledge, this is the first study combining fully automated segmentation, vBMD mapping and long-term reproducibility analysis to establish gender-, age-, and region-specific characteristic values and variations of vertebral TFs. This data could serve as foundation for a reliable differentiation of values indicative of osteoporosis from normal values across genders, age groups and spine regions. In particular, it could be a prerequisite for the longitudinal comparison of TFs. Variance global and Entropy showed a significant increase in cranio-caudal direction along the spine for both males and females. The other four TFs, which are all GLRLM-derived features, also showed significant region-dependent differences, except for females between the lower thoracic and lumbar spine. Five of the six TFs showed a significant difference between genders. None of the TFs showed a significant age-dependence. vBMD showed a significant decrease with age as well as statistically significant differences between genders and spine regions, confirming results from previous studies [bib_ref] X-Ray-Based Quantitative Osteoporosis Imaging at the Spine, Loffler [/bib_ref] [bib_ref] Axial QCT: Clinical Applications and New Developments, Link [/bib_ref] [bib_ref] Regional Analysis of Age-Related Local Bone Loss in the Spine of a..., Valentinitsch [/bib_ref]. Dimension reduction and feature selection is a commonly performed step in TA studies to reduce redundancy, and to ensure that selected features fulfill certain conditions and have high relevance. However, the applied approaches are manifold and vary substantially between studies. Among many other methods, GLCM-and GLRLM-derived features can be restricted to or averaged across directions and distances [bib_ref] Normative Values for CT-Based Texture Analysis of Vertebral Bodies in Dual X-Ray..., Mannil [/bib_ref]. Another popular approach is to only select reproducible features with high inter-and intra-reader agreement [bib_ref] Normative Values for CT-Based Texture Analysis of Vertebral Bodies in Dual X-Ray..., Mannil [/bib_ref]. Mannil et al. additionally performed correlation analysis and excluded highly correlating TFs to reduce redundancy [bib_ref] Normative Values for CT-Based Texture Analysis of Vertebral Bodies in Dual X-Ray..., Mannil [/bib_ref]. The emergence of machine learning has further extended the repertoire of feature reduction methods. Using a random forest classifier to identify vertebral fractures, Valentinitsch et al. optimized the number of features based on the Gini importance and classification performance in an exponential search [bib_ref] Opportunistic Osteoporosis Screening in Multi-Detector CT Images via local classification of textures, Valentinitsch [/bib_ref]. We chose a more clinically driven approach and selected TFs based on a preceding long-term reproducibility analysis to identify features particularly suited for longitudinal comparisons.
Our study determined characteristic TF values based on CT images of vertebral bodies. The found values for Variance global show substantial differences to the values reported by , who performed the only similar study we could find in previous literature. This discrepancy, however, is potentially due to the limited comparability of the two studies. While Mannil et al. used manually prescribed ROIs of mid-sagittal planes of the vertebral body for two-dimensional (2D) TA (30), we utilized 3D-ROIs of the entire volume of the trabecular compartment of the vertebral body with subsequent 3D-TA. As a result, number and intra-vertebral localization of the included voxels as well as their gray value distribution vary remarkably between the two studies. Differences in CT image acquisition and reconstruction as well as slice thickness (2 mm vs. 3 mm) may additionally contribute to the discrepancy of the reported TF values. Fully automated labeling and 3Dsegmentation of the spine represent major improvements in terms of time efficiency, reproducibility and utilization of available CT data when compared to manually performed 2D-segmentations. Therefore, the segmentation approach used in the present study should be considered as the standard for TA of the spine.
Characteristic values of TFs should be used and compared with great care and ideally only if acquisition and reconstruction parameters as well as postprocessing methods are identical. Therefore, we used images acquired at the same MDCT scanner with a standardized protocol, a fully automated and standardized segmentation method, as well as careful preparation of CT data prior to TF computation, including isotropic resampling and graylevel normalization.
The TFs Variance global and Entropy showed significant regional differences, increasing from the mid-thoracic to the lumbar spine, irrespective of gender. Along the spine, more inferior vertebral bodies have larger size as well as a higher prevalence and degree of degenerative changes [bib_ref] Regional Analysis of Age-Related Local Bone Loss in the Spine of a..., Valentinitsch [/bib_ref] [bib_ref] Pain in the Lumbar, Thoracic or Cervical Regions: Do Age and Gender..., Leboeuf-Yde [/bib_ref]. This could be potential explanations for the cranio-caudal increase in Variance global since it reflects the spread of gray-level distribution. The GLCM feature Entropy is a measure of randomness in an image. However, it is not trivial to define what image properties it actually represents. According to Haralick et al., Entropy is associated with gray-level number, complexity and degree of structure in an image [bib_ref] Textural Features for Image Classification, Haralick [/bib_ref]. Therefore, the found region-dependent variations could indicate a gradient in architectural complexity along the spine, FIGURE 3 | Texture features and vBMD grouped by spine region for females (red) and males (blue). Asterisks (*) mark pairwise significant differences between spine regions. Bonferroni correction was applied to adjust for multiple comparisons. Error bars indicate ± 2 standard deviations. SRE, short-run emphasis; LRE, long-run emphasis; RLN, run-length non-uniformity; RP, run percentage; vBMD, volumetric bone mineral density.
FIGURE 4 | Texture features and vBMD grouped by age group for females (red) and males (blue). Asterisks (*) mark pairwise significant differences between consecutive age groups. Bonferroni correction was applied to adjust for multiple comparisons. Error bars indicate ± 2 standard deviations. SRE, short-run emphasis; LRE, long-run emphasis; RLN, run-length non-uniformity; RP, run percentage; vBMD, volumetric bone mineral density.
potentially suggesting variability in trabecular microstructure and skeletal integrity [bib_ref] Anorexia Nervosa: Analysis of Trabecular Texture With CT, Tabari [/bib_ref]. The TFs SRE, LRE, RP and RLN also showed gender-specific differences, as well as region-dependent differences, irrespective of gender for most features. These GLRLM features are highly associated with each other and, according to their defining equations, reflect the predominating length of gray-level runs, the distribution of gray-level run lengths, as well as the extent of linear structures [bib_ref] Texture Analysis Using Gray Level Run Lengths, Galloway [/bib_ref]. The variations we found may therefore indicate gender-and region-specific variability in trabecular microstructure, in line with previous ex vivo results [bib_ref] Variability of Trabecular Microstructure is Age-, Gender-, Race-and Anatomic Site-Dependent and Affects..., Yeni [/bib_ref].
There are several limitations to our study. First, we performed a retrospective analysis with limited availability of weight and height as well as other bone health-related data of the study cohort. Therefore, we did not adjust our results for variations in body mass index (BMI) or other potential confounders, such as smoking status, which are known to be correlated with BMD and could also have an influence on TFs. Second, n = 376 is relatively small sample size for establishing reference values, and the age range of the study is limited, in particular the age group under 40 years is not represented in our cohort. However, primary osteoporosis, as the most important condition for clinical application of the herein used approach and derived characteristic TF values, usually does not start before the age of 40 years. In contrast to values derived from a population of young individuals with normal bone health, age-, gender-and region-specific may also represent a viable approach for the differentiation of physiological and pathological bone alterations of the spine. Nevertheless, extending the analysis to an increased study population with a broader age range should be an aim of future investigations, in order to firmly establish a reference database. Third, the initial set of TFs we considered in our analysis was limited, and additional TFs (e.g. based on local binary patterns, wavelets or other transforms) could have been used [bib_ref] Multiresolution Gray-Scale and Rotation Invariant Texture Classification With Local Binary Patterns, Ojala [/bib_ref] [bib_ref] Bone Texture Characterization Based on Contourlet and Gabor Tranforms, Ngo [/bib_ref]. Fourth, to maximize comparability of the computed values, CT data acquisition and postprocessing as well as TA were performed with homogenous and standardized settings. The restriction to a specific CT scanner and acquisition protocol may reduce the utility of the findings to other investigations using different settings. However, TF values, in particular for GLCMand GLRLM-derived features, can only be reliably used for comparisons in studies with identical settings.
# Conclusion
In conclusion, we established highly reproducible TF values for CT-based 3D-TA of vertebral bodies in a predominantly older study population, using an automated segmentation and quantification pipeline. These characteristic TF values and the found gender-, age-, and vertebral-level-specific variations can be considered a foundation for the reliable differentiation of physiological and pathological alterations and may be particularly important for future longitudinal studies.
# Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
# Ethics statement
The studies involving human participants were reviewed and approved by Ethikkommission der Technischen Universität München. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.
# Author contributions
MD wrote the manuscript. MD organized the database. MD performed the formal and statistical analysis. MD contributed to the concept and design of the study. MD developed major parts of the used software routines. NS, MH, AS, ML, CZ, JK, KS, and TB critically revised the manuscript. MH, AS, and ML contributed to the development of the software used in the study. TB conceptualized and designed the study. CZ, JK, KS, and TB provided scientific supervision and resources. JK and TB provided funding. TB contributed to the generation and organization of the database. All authors contributed to the article and approved the submitted version.
[fig] FIGURE 1 |: Exemplary illustration of the automated labeling and segmentation. (A) individually labeled and segmented vertebrae from T4-L5. (B, C) vertebral subregions, including the trabecular (white) and cortical (green) compartment of the vertebral body. [/fig]
[fig] FIGURE 2 |: Patient age vs vBMD (averaged across all analyzed vertebrae for each patient) for females (red) and males (blue); vBMD, volumetric bone mineral density. [/fig]
[table] TABLE 2 |: Relative difference D rel between baseline and follow-up measurement of the analyzed TFs. TFs with D rel < 1% and D rel < 5% are marked with ** and *, respectively. [/table]
[table] TABLE 3 |: Patient demographics stratified by age groups. Mean ± standard deviation (SD) for age (in years).TABLE 4 | Mean and standard deviation (SD) of the six analyzed texture features and vBMD of females, grouped by spine region. [/table]
[table] TABLE 6 |: Mean and standard deviation (SD) of the six analyzed texture features and vBMD of females, grouped by age group. [/table]
[table] TABLE 5 |: Mean and standard deviation (SD) of the six analyzed texture features and vBMD of males, grouped by spine region. [/table]
[table] TABLE 7 |: Mean and standard deviation (SD) of the six analyzed texture features and vBMD of males, grouped by age group. [/table]
[table] TABLE 8 |: Texture features and vBMD, averaged from T4-L5 and grouped by gender. [/table]
[table] TABLE 9 |: Pairwise significant differences of texture features between spine regions across all subjects.Bonferroni-corrected p-values corresponding to a level of significance of 0.05 are given. SRE, short-run emphasis; LRE, long-run emphasis; RLN, run-length non-uniformity; RP, run percentage. [/table]
|
Increasing prevalence of HIV infection among first time clients in Italian drug treatment services – is it sexual transmission?
Background: Over the last two decades, the proportion of people who inject drugs among newly reported HIV cases in Italy has been continuously declining. This trend is reflected in the prevalence of HIV infection among problem drug users followed in drug treatment services. We report nationwide trends in the prevalence of HIV and HCV among tested clients in charge to drug addiction services from 2005 to 2011. Methods: Data on the prevalence of HIV and HCV among drug users from public drug treatment services across Italy were collected and analyzed for the period from 2005 to 2011. Prevalence of HIV and HCV were compared between clients returning to treatment and those entering treatment for the first time, and by gender. Due to the high percentage of missing data, the "inverse probability weight" method was used. Trends in testing uptake were also analysed. Results: A significant decrease of HIV and HCV prevalence is observed among all PDUs entering treatment (from 14.7% to 11.1% and from 61.6% to 50%, respectively, in 2005-2011). By contrast, among those entering the services for the first time, after an initial decline the prevalence of HIV infection steadily increased in both sexes, from 2.2% in 2009 to 5.3% in 2011. Self-reported injecting rates in this group decreased over time, and in 2011 the proportion reporting drug injecting was lower among new clients than in people returning to services (14.5 vs. 34.4%). We also observed a progressive and significant reduction in HIV and HCV testing in drug treatment services. Conclusions: Changes in injection practice and type of drugs used, coupled with a concurrent reduction in HCV prevalence, do not support drug injection as the main explanation for an increased HIV transmission in people entering drug treatment services for the first time. While reductions in testing rates raise concerns over data quality, the possibility of increased sexual transmission needs to be considered.
# Background
In some European countries, particularly in East Europe, HIV infection in people who inject drugs (PWID) is reemerging as a serious public health problem. Since 2001 and through 2011, data on newly diagnosed HIV infections and HIV prevalence suggested falling infection rates in PWID in the European Union [bib_ref] HIV infection and AIDS in the European Union and European Economic Area, Likatavicius [/bib_ref] [bib_ref] Trends in HIV and hepatitis C virus infections among injecting drug users..., Wiessing [/bib_ref]. Also in Italy the proportion of PWID among those newly diagnosed with AIDS has been continuously decreasing, from 62.4% before 2001 to 17.6% in 2011/12 [bib_ref] Aggiornamento delle nuove diagnosi di infezione da HIV e dei casi di..., Sanità [/bib_ref]. This may, at least partly, follow from the increased availability of prevention, treatment and harm-reduction measures, including opiate substitution treatment and needle and syringe programmes [bib_ref] Opiate substitution treatment and HIV transmission in people who inject drugs: systematic..., Macarthur [/bib_ref]. Other factors, such as the decline in injecting drug use that has been reported in some countries (e.g., United Kingdom, Norway), may also have played an important role [bib_ref] Injecting drug use in Europe: stable or declining, Wiessing [/bib_ref]. Recent data, however, suggest that there is a continuing potential for outbreaks of HIV infection among PWID in some European countries, such as Greece and Romania [bib_ref] HIV among injecting drug users in Europe: increasing trends in the East, Wiessing [/bib_ref] [bib_ref] Human immunodeficiency virus in injecting drug users in Europe following a reported..., Pharris [/bib_ref] [bib_ref] Human immunodeficiency virus among people who inject drugs: Is risk increasing in..., Hedrich [/bib_ref].
Drug injecting has been associated with HIV transmission since the beginning of the epidemic in the early 80' [bib_ref] HIV infection among intravenous drug users: epidemiology and risk reduction, Jarlais [/bib_ref] ; however the role of non-injection drug use in the spread of HIV is still unclear [bib_ref] Parenteral and sexual transmission of human immunodeficiency virus in intravenous drug users:..., Nicolosi [/bib_ref] [bib_ref] Sexual transmission of HIV-1 among injection drug users in San Francisco, USA:..., Kral [/bib_ref] [bib_ref] Sex differences in risk factors for hiv seroconversion among injection drug users:..., Strathdee [/bib_ref]. Drug use by any route (not just injection) can be associated with unsafe sexual practices, potentially putting people at risk for acquiring or transmitting HIV and other sexually transmitted infections. In particular, stimulants, such as cocaine, crack cocaine and methamphetamine, have been linked to high-risk sexual behaviour [bib_ref] HIV and other sexually transmitted infections in injection drug users and crack..., Jarlais [/bib_ref] [bib_ref] Crystal methamphetamine, its analogues, and HIV infection: medical and psychiatric aspects of..., Urbina [/bib_ref] [bib_ref] Drug use and sexual risk behavior among gay and bisexual men who..., Colfax [/bib_ref]. Importantly, individuals who engage in high-risk sexual practices (e.g. sex workers, men who have sex with men) may have a higher prevalence of drug use than the general population so that drug use may be associated with high HIV prevalence even without being causally linked to high-risk sexual behaviour [bib_ref] Drug use and sexual risk behavior among gay and bisexual men who..., Colfax [/bib_ref] [bib_ref] HIV and female sex workers, Estébanez [/bib_ref].
In Italy, as in many other countries, HIV prevention programmes for problem drug users (PDUs) have mainly focused on HIV transmission through unsafe injecting practices with less attention to sexual risk behaviours. It has been noted that in many countries, including Italy, Netherlands and USA, the prevalence of HIV infection among non-injecting PDUs has increased (with a simultaneous decrease among PWID) and this increase has been correlated with sexual transmission [bib_ref] Continued high prevalence of HIV, HBV and HCV among injecting and noninjecting..., Camoni [/bib_ref] [bib_ref] Des Jarlais DC. Sexual and other noninjection risks for HBV and HCV..., Neaigus [/bib_ref] [bib_ref] The transition from injection to non-injection drug use: long term outcomes among..., Jarlais [/bib_ref] [bib_ref] Decline in HIV incidence and injecting, but not in sexual risk behaviour,..., Lindenburg [/bib_ref].
We analyze data obtained from the national network of drug treatment centres in Italy to investigate the potential role of HIV transmission through sexual contact among non-injecting drug users and to identify HIV trends in specific drug-using subgroups. Where available, results of the HIV, HBV and HCV screening tests were also analyzed.
# Methods
There are 550 drug treatment services located throughout Italy, offering free drug treatment, medical care and psychological assistance on an outpatient basis. Attendance is voluntary and, as defined by law, screening for HIV, HBV and HCV should be offered to all new clients entering the drug treatment services and to returning clients. Data are continuously collected in each service using standardised data forms, regarding: the number of new clients and returning clients, gender, substance(s) of abuse and, where available, results of the HIV, HBV and HCV screening tests. Data are sent to the Ministry of Health and to the Antidrug Policy Department in aggregate format on an annual basis.
For the purpose of this study, we used aggregated data on HIV, HCV and HBV prevalence among people registered in drug treatment services. Screening for HIV, HBV and HCV is performed with enzyme-linked immunosorbent assays on venous blood specimens. In case of HIV and/or HCV seropositivity, subjects are referred to an Infectious diseases or Internal Medicine specialist. Where available, results of HBV serology are interpreted in order to detect subjects candidates to vaccination (anti-HBs and anti-HBc negatives) and those with previous HBV vaccine or infection. Prevalence of HIV and HCV were compared between clients returning to treatment and those entering treatment for the first time, and by gender. Trends in testing uptake were analyzed.
Data were analyzed from 2005 to 2011 using annual reports from the Antidrug Policy Department. Utilization of data collected in the study for scientific purposes was approved by the Bioethics Committee of the Department for antidrug policies.To compare proportions, we used the chi-2 test for trend with double-side p-values. Due to the high percentage of missing data on the raw prevalence of HIV and HCV infection, and in order to reduce the impact of the missing data on the final estimates, the "inverse probability weight" method was used [bib_ref] Review of inverse probability weighting for dealing with missing data, Seaman [/bib_ref].
# Results
From 2005 to 2011 the Italian treatment monitoring system documented between 162,005 and 174,156 clients entering drug treatment services yearly [fig_ref] Table 1: Prevalence of HIV and HCV infection among clients registered in Drug addiction... [/fig_ref]. Due to inconsistencies in reporting serologic markers of HBV, data on HBV infection are not reported. We observed a progressive decrease in HIV and HCV screening coverage in drug addiction services, and these differences were statistically significant (p < 0.0001 by χ2 for trend) [fig_ref] Figure 1: Temporal trends in HCV and HIV screening in Drug addiction services in... [/fig_ref]. The percentage of clients tested for HIV has decreased from 40.6% in 2005 to 30.5% in 2011, and that of clients tested for HCV from 46.5% in 2005 to 40.6% in 2011 (Table 1, [fig_ref] Figure 1: Temporal trends in HCV and HIV screening in Drug addiction services in... [/fig_ref]. Of note, in 2011 only 21.9% of female and 21.1% of male new clients were screened for HIV, with 8.7% (86 women) and 5.5% (340 men) respectively found to be HIV positive.By comparison, in 2005 35.5% of female and 32.4% of male new clients were screened for HIV, with 3.7% of female and 3.3% of male found positive. Due to the high percentage of missing data, the inverse probability weight method was used to analyze data. Overall, prevalence of HIV infection in patients registered in drug addiction services decreased significantly, from 14.7% in 2005 to 11.1% in 2011. However, among those entering the services for the first time (new clients), after an initial decline the prevalence of HIV infection steadily increased in both sexes, from 2.2% in 2009 to 5.3% in 2011 (Table 1, [fig_ref] Figure 2: HIV and HCV prevalence among new clients in Drug addiction services in... [/fig_ref]. Data on HCV prevalence are shown in [fig_ref] Figure 1: Temporal trends in HCV and HIV screening in Drug addiction services in... [/fig_ref] and [fig_ref] Table 1: Prevalence of HIV and HCV infection among clients registered in Drug addiction... [/fig_ref]. HCV prevalence in patients attending drug treatment services is high, but a steady and significant decline has been observed during the last years both in the overall population and in new clients. HCV infections decreased significantly from 61.6% to 50.0% in the overall population tested (p < 0.0001 by χ 2 for trends), and from 31.8% to 21.6% in new clients (p <0.0001).
Among registered clients in 2011, 79.8% reported opioids as the primary drug of abuse, followed by cocaine in 12.6% and other substances (mostly cannabis) in 7.6%. A different distribution was identified among new clients: 38.7% reported opioids as the primary drug, 30.8% used cocaine, and 25.4% cannabis. Moreover, drug injection was reported by 14.5% of new clients (42.4% of those reporting opioids as primary drug, and 3.9% of those reporting cocaine) compared to 34.4% (58.1% of opioids users, 7.3% of cocaine users) of overall clients. By comparison the 2005 data show a higher proportion of opioids use (45.7%) and a lower cocaine use (26.5%) in new clients, and an overall higher proportion of injective use (22.8% among new clients, and 67.3% among the overall population).
# Discussion
We report a marked increase in the prevalence of HIV coupled with a decrease of HCV prevalence among people entering drug treatment services for the first time in Italy.
Changes in injection practice and type of drugs used, coupled with a concurrent reduction in HCV prevalence, do not support increases in drug injection risks as an explanation for the increase in HIV transmission in this group. A slightly higher HIV prevalence was observed among women compared to men, possibly due to the low number of women in the sample. However, the proportion of HIV-positive women to HIV-positive men (1:2.5) is similar to that reported in the surveillance system for new HIV diagnoses (1:3) in 2011 [bib_ref] Aggiornamento delle nuove diagnosi di infezione da HIV e dei casi di..., Suligoi [/bib_ref]. Although our monitoring system has national coverage and as such provides key data on infectious diseases in PDU we identify important weaknesses that need to be overcome to improve their relevance for public health. Among injecting drug users, HIV is transmitted sexually and parenterally, but HCV is transmitted primarily parenterally [bib_ref] Is the HCV-HIV co-infection prevalence amongst injecting drug users a marker for..., Vickerman [/bib_ref]. Recent evidence suggests that HIV transmission in European countries (Greece, Romania) reporting new outbreaks of HIV infection among injecting drug users was preceded by increases in the prevalence of the HCV [bib_ref] Trends in HIV and hepatitis C virus infections among injecting drug users..., Wiessing [/bib_ref]. Rising HCV prevalence may act as an early indicator of increases in injecting risks among injecting drug user populations, possibly before HIV has started to spread [bib_ref] Is the HCV-HIV co-infection prevalence amongst injecting drug users a marker for..., Vickerman [/bib_ref] [bib_ref] Can hepatitis C virus prevalence be used as a measure of injection-related..., Vickerman [/bib_ref] [bib_ref] Determinants of persistent spread of HIV in HCV-infected populations of injecting drug..., De Vos [/bib_ref]. Given that the increase in HIV prevalence among the new clients in our data is not associated with an increase in HCV prevalence, our data thus seem not to support increases in injection risks as an explanatory factor for the observed increase in HIV.
Data from the Italian National Institute of Health collected in 2005 in a randomly selected sample of 1330 individuals with problematic drug use (defined according to EMCDDA [bib_ref] Estimating prevalence of problem drug use at national level in countries of..., Kraus [/bib_ref] as 'injecting drug use or long-duration/ regular use of opioids, cocaine and/or amphetamines), attending public drug treatment services, showed that the prevalence of HIV was 14.4% among drug injectors and 1.6% among non-injectors; HCV prevalence was 83.2% among injecting drug users and 22.0% among noninjectors [bib_ref] Continued high prevalence of HIV, HBV and HCV among injecting and noninjecting..., Camoni [/bib_ref]. However, the percentage of individuals tested for HIV in those centres has also decreased significantly, from 40.6% in 2005 to 30.5% in 2011 (p < 0.0001 by χ 2 for trends).
The progressive reduction of HIV and HCV testing is also one of our main findings, based on a monitoring system with nationwide coverage of drug addiction population in charge to drug addiction services. Results of a survey conducted in Italy in drug treatment services underline the importance of facilitating access to testing and removing obstacles that can lead to the drug user's refusal to undergo testing. The failure to undergo testing among injectors was associated with a shorter history of drug use (<5 years) and with drug centres in central or southern Italy; these associations were also found among non-injectors, with the addition of low level of education. Other reasons for this decline in testing over time include a progressive reduction of resources for public drug treatment services, and a decreased proportion of drug users undergoing pharmacological-psychological treatment (which involves a more complete monitoring program) [bib_ref] Few Italian Drug Users Undergo HIV Testing, Camoni [/bib_ref]. Moreover, main reasons for not undergoing HIV testing were an individual's refusal and lacking of a blood drawing facility within the service. These results underline the importance of facilitating access to testing, of providing the drug dependency centres with the necessary resources for taking blood samples at the centres themselves, of making access to testing uniform throughout the country, and of removing obstacles that can lead to the drug user's refusal to undergo testing.
In Italy, the government response to preventing infections among PDUs is mainly based on opioid substitution treatment (OST), detoxification and psychosocial interventions, while needle and syringe programs (NSP) are mainly implemented through NGOs in the setting of outreach programmes with the training for PWIDs and coverage is very low. Opioid agonist treatments are associated with reductions in the frequency of opioid use, fewer injections and injection-related risk behaviours and lower rates of HIV/HCV prevalence and incidence [bib_ref] Oral substitution treatment of injecting opioid users for prevention of HIV infection, Gowing [/bib_ref] [bib_ref] Reference Group to the UN on HIV and Injecting Drug Use. HIV..., Mathers [/bib_ref] [bib_ref] Patients receiving opioid maintenance treatment in primary care: successful chronic hepatitis C..., Seidenberg [/bib_ref]. On the other hand, progressively less attention has been given to periodic testing for HIV, HBV and HCV.
The low percentage of PWID tested for HIV observed in drug treatment services leads to a high proportion of subjects with late HIV diagnosis and presentation (late presenters), correlated with a high probability of transmitting HIV and worse clinical outcome [bib_ref] Aggiornamento delle nuove diagnosi di infezione da HIV e dei casi di..., Sanità [/bib_ref] [bib_ref] Estimating sexual transmission of HIV from persons aware and unaware that they..., Marks [/bib_ref] [bib_ref] Risk factors and outcomes for late presentation for HIV-positive persons in Europe:..., Mocroft [/bib_ref]. In Italy in 2012, PWID showed the highest proportion (61.0%) of late presenters for HIV infection as compared to heterosexuals or men who have sex with men [bib_ref] Aggiornamento delle nuove diagnosi di infezione da HIV e dei casi di..., Sanità [/bib_ref]. This finding confirms that HIV testing is not routinely performed on a year basis among PWID but rather at a late stage when HIV-indicator diseases are diagnosed, leading to a higher risk of death among PWID with AIDS [bib_ref] Risk of hepatitis-related mortality increased among hepatitis C virus/HIV-coinfected drug users compared..., Smit [/bib_ref].
Likewise, screening for HCV in drug services remains of utmost importance, since HCV prevalence among injecting drug users remains high, and the availability of improved HCV treatment is expected to reduce the disease burden. HIV/HCV co-infected drug users remain at increased risk of dying from liver-related death in the HAART era compared with HCV-mono-infected drug users [bib_ref] AIDS and injecting drug use: survival determinants in the highly active antiretroviral..., Zucchetto [/bib_ref]. Moreover, recent reports show that among Italian PWID liver disease is the most frequent non-AIDS related cause of death [bib_ref] AIDS and injecting drug use: survival determinants in the highly active antiretroviral..., Zucchetto [/bib_ref].
Another point that deserves attention is the changing epidemiology of problem drug use. In Europe, treatment demand for stimulants (cocaine and amphetamine-type stimulants) and cannabis have increased over time, in parallel with a decline in treatment demands for opiates. Moreover, it has been suggested that the increasing use of non-injecting drugs, such as recreational drugs, may lead to an increased spread of infections through sexual contact [bib_ref] Des Jarlais DC. Sexual and other noninjection risks for HBV and HCV..., Neaigus [/bib_ref] [bib_ref] Patients receiving opioid maintenance treatment in primary care: successful chronic hepatitis C..., Seidenberg [/bib_ref]. Such developments may partly explain our findings for Italy. Important limitations exist regarding the quality and completeness of our data. First, the progressive reduction in HIV and HCV testing in drug treatment services is of concern, as it can result in unpredictable biased prevalence rates for viral infections, making prevalence trends more difficult to interpret. Second, until present only aggregate data on HIV prevalence are available, which provide summary information but do not allow more detailed epidemiological analysis, such as stratifying seroprevalence rates by type of client, type of substance used, or by route of drug administration.
# Conclusions
Changes in injection practice and type of drugs used, coupled with a concurrent reduction in HCV prevalence, do not support drug injection as the main explanation for an increased HIV transmission in people entering drug treatment services for the first time. While reductions in testing rates raise concerns over data quality, the possibility of increased sexual transmission needs to be considered. Evaluation of this hypothesis and planning of possible effective interventions for reducing HIV transmission via sexual behaviours among PDUs (irrespective of injective or non-injective use) is required.
Abbreviations PWID: People who inject drugs; PDUs: Problem drug users; HIV: Human Immunodeficiency Virus; HCV: Hepatitis C Virus; HBV: Hepatitis B Virus; EMCCDA: European Monitoring Centre for Drugs and Drug Addiction.
## Competing interests
Authors declare that there are not any financial competing interests in relation to this manuscript.
Authors' contributions MC, LW, BS contributed to interpretation of data and drafted the submitted article. GS, BG, AA, CI contributed to design, acquisition and analysis of data; MZ contributed to data analysis and changes in the revised manuscript. All the Authors provided final approval of the version of the submitted article.
[fig] Figure 1: Temporal trends in HCV and HIV screening in Drug addiction services in Italy. A significant decrease in HIV and HCV testing in clients of drug addiction services (p < 0.0001 by χ 2 for trends) was observed. [/fig]
[fig] Figure 2: HIV and HCV prevalence among new clients in Drug addiction services in Italy. [/fig]
[table] Table 1: Prevalence of HIV and HCV infection among clients registered in Drug addiction services in Italy [/table]
|
Noble Metal‐Based Multimetallic Nanoparticles for Electrocatalytic Applications
Noble metal-based multimetallic nanoparticles (NMMNs) have attracted great attention for their multifunctional and synergistic effects, which offer numerous catalytic applications. Combined experimental and theoretical studies have enabled formulation of various design principles for tuning the electrocatalytic performance through controlling size, composition, morphology, and crystal structure of the nanoparticles. Despite significant advancements in the field, the chemical synthesis of NMMNs with ideal characteristics for catalysis, including high activity, stability, product-selectivity, and scalability is still challenging. This review provides an overview on structure-based classification and the general synthesis of NMMN electrocatalysts. Furthermore, postsynthetic treatments, such as the removal of surfactants to optimize the activity, and utilization of NMMNs onto suitable support for practical electrocatalytic applications are highlighted. In the end, future direction and challenges associated with the electrocatalysis of NMMNs are covered.
## Doi: 10.1002/advs.202104054
strength of reaction intermediates on active sites with catalytic activity, suggests that the interaction should neither be too strong nor too weak.It is widely accepted that noble metals (mainly Pt, Pd, Ir, Rh, Ru, and Au) have great potential as electrocatalysts due to their optimal sorption properties in volcano-shaped activity trends for many representative electrocatalytic reactions that comprise fuel cells, water splitting, and CO 2 electrolysis.These metals are usually utilized in various forms of nanoparticles to take advantage of their maximized surface area and defect-rich surfaces. The surface properties are well known to strongly influence the overall catalytic performance of noble metal nanoparticles.However, monometallic systems are suffering from an intrinsically limited activity enhancement as well as instability under operating conditions.Furthermore, scarcity and high demand together made noble metals not affordable. Therefore, noble metal-based monometallic catalysts are considered less promising for industrial applications, which typically require excellent activity, chemical and electrochemical stability, and cost-effectiveness.
Over the past few decades, noble metal-based multimetallic nanoparticles (NMMNs) synthesized by incorporating secondary metal elements into noble metal systems have emerged as alternatives. These nanoparticles have garnered tremendous attention because of their superior catalytic activity and durability compared to their monometallic counterparts.In addition to the improvement in the overall catalytic performance, the adoption of non-noble metals reduces the load on expensive noble metals, enabling economically feasible processes for manufacturing nanocatalysts. Pt-based multimetallic nanoparticles are representative examples of enhanced catalytic activity and durability for the oxygen reduction reaction (ORR) compared to commercial Pt catalysts.With the growing demand for electrocatalysts with superior performance, design principles have been more sophisticated by both empirical and theoretical methodologies to meet the need for multifunctionality. As shown in the previous studies, the physicochemical properties and catalytic performance of NMMNs are largely affected by the atomic distribution, size, composition, and morphology.However, synthesizing NMMNs as per the design is challenging because the mechanism of the chemical synthesis of nanoparticles is complex and it is largely unclear. Nevertheless, more advanced and complicated synthetic procedures have been extensively developed to precisely modulate the structural factors, thereby achieving the desired properties of NMMNs.
Although there have been great achievements in the synthesis of nanoparticles, the as-prepared nanoparticles themselves are not efficient electrocatalysts, leaving several more factors to be considered to successfully translate them into practical electrocatalysts. In the general synthesis of nanoparticles, surface capping agents (ligands) are prerequisite in achieving uniform size and controlled shapes with excellent dispersion. As the ligands can either have (usually) detrimental or (rarely) promotive effects on electrocatalytic processes, it is essential to consider them for better utilization of NMMNs.On the other hand, catalytically active nanoparticles are usually anchored on the surface of support materials before being applied in catalytic reactions to prevent the agglomeration of nanoparticles and also to ensure sufficient electrical conductivity.Besides, the support materials can also affect the catalytic properties of nanoparticles through metal-support interactions,and the stability of supports becomes significant, especially when the operating conditions are harsh or long-term operation is required.In this review, recent developments in the synthesis of NMMNs, which are categorized based on their structural features, are highlighted. For each structure, representative synthetic methods with exemplary works are provided. Along with the notable early contributions, major current synthetic challenges and some of the recent attempts to tackle them are discussed. Afterward, postsynthetic treatments/modifications that need to be considered to deal with the practical electrocatalytic applications of NMMNs, such as the removal of ligands and selection of the supports are covered.
## Synthetic methods for various nmmns
Based on theoretical and empirical insights from various studies, design principles for high-performance electrocatalysts have been developed. The performance of nanocatalysts is sensitive to their physical and chemical characteristics, which necessitates employing appropriate synthetic procedures with reliability and reproducibility. This section describes the synthetic approaches for various kinds of NMMNs, major unsolved issues in the synthesis, and recent progress. We have classified NMMNs into six categories to present the principal synthetic methods in a more organized manner: random alloys, single-atom alloys (SAAs), high-entropy alloys (HEAs), ordered-intermetallics, core-shell structures, and other heterostructures. Representative synthetic methods for a specific structure are described in each section with exemplary cases.
## Random alloys
A random alloy refers to a solid solution comprising multiple metal components with a random distribution in the crystal structure. This section includes the synthetic methods prevalently used for random alloy nanoparticles, followed by the modulation of the size, composition, and morphology of these nanoparticles. The last topic of this section concerns strategies to prevent phase separation of alloys to obtain homogenous single phases.
## Common synthetic routes
This section discusses the representative routes to synthesize random alloy nanoparticles together with exemplary works. Although the boundaries of the methods are being blurred as they have been modified and combined with each other as per the requirement, this section may help understand the synthetic factors that mainly affect the physical and chemical properties of nanoparticles and how they can be carefully controlled.
Co-reduction: Simultaneous reduction of metal precursors is a straightforward route to prepare random alloy nanoparticles, especially when the alloyed metal elements have similar reduction potentials. In this process, supersaturation of monomers is induced by the introduction of an appropriate reducing agent, such as hydrazine, NaBH 4 , or polyol, into the precursor solution, followed by the nucleation and growth processes. For instance, displays monodispersed PtFe nanoparticles obtained via coreduction of FeCl 2 and Pt(acac) 2 (acac = acetylacetonate) by introducing a reductant, LiBEt 3 H, with oleic acid and oleylamine as stabilizing agents.Another exemplary work is the synthesis of Pd-based alloys,where a solution of mixed metal acetylacetonate precursors was injected into a preheated solution containing a reductant, such as borane morphine or borane tertbutylamine, which reduces the precursors to PdCo and PdCu alloy nanoparticles.
Thermal Decomposition: In this approach, the synthesis of random alloy nanoparticles is triggered by the decomposition of organometallic compounds to monomers at high temperatures followed by nucleation of the supersaturated monomers. The hotinjection and heat-up processes fall under the category of thermal decomposition.The primary difference between these two processes is the way of achieving high supersaturation levels of monomers for a burst nucleation. In the hot-injection method, the organometallic complexes are rapidly injected into an already heated reaction solution to induce burst nucleation. On the other hand, in the heat-up method, precursors are mixed with all the other chemicals before heating up, and the supersaturation of monomers is achieved by continuous elevation of temperature. For this reason, hot-injection method features flexibility for versatile syntheses, including the use of labile precursors, while heatup method features convenience by the absence of external operation during the reaction. As this technique requires high temperatures, the organic solvents of high boiling points are usually employed. The high temperature used in this method has mer-its in the selection of elements, enabling the use of metals such as Fe, Co, and Ni, which have relatively lower reduction potentials than precious metals. A representative case is the synthesis of PtFe nanoparticles introduced by Sun and co-workers.The preparation of monodisperse PtFe nanoparticles was accomplished by the decomposition of Fe(CO)by injecting it into a solution containing a Pt precursor at high temperature and simultaneous reduction of the Pt precursor by diol .
Solvothermal Synthesis: Solvothermal synthesis generally takes place at temperatures higher than the boiling point of the solvent and at pressures above atmospheric pressure to induce the reaction of the precursors. The reaction is usually performed in a Teflon-lined autoclave, a closed vessel that can withstand high pressure and temperature. This method is useful when the synthesis of nanocrystalline materials requires better control over size and shape. For example, the synthesis of PtCu 3 nanoparticles possessing an excavated rhombic dodecahedral shape could be accomplished via the reaction of acetylacetonate precursors in an autoclave in the presence of n-butylamine and cetyltrimethylammonium chloride (CTAC) .The authors suggested that the amine group in n-butylamine critically contributes to the stabilization of the {110} facet of the nanoparticles, which is known to be the least stable among low-index facets. Through the solvothermal technique, nanocrystals with metastable phases could also be prepared. The synthesis of Pt-Ni excavated nano-multipods with metastable hexagonal closepacked structures using formaldehyde and a Ni-rich precursor is one of the examples .Incipient Wetness Impregnation: Incipient wetness impregnation (IWI) is one of the simplest and most commonly employed methods for preparing metal nanoparticles anchored on support materials. A solution containing metal precursors is first mixed with support, and then the composite is completely dried to form supported nanoparticulates upon annealing. Despite the limited controllability of the size and composition of individual nanoparticles, the method has been used to prepare supported alloy nanoparticles owing to the ease and scalability. Recently, Abruña and co-workers reported a generalized synthesis of RuM (M = Co, Ni, Fe) nanoparticles via wet impregnation of precursors into carbon support, followed by thermal annealing.Furthermore, the method can be extended to the synthesis of trimetallic nanoparticles by using precursors of three different metals.Seed-Mediated Synthesis: Seed-mediated synthesis involves a two-step process comprising the preparation of seed nanoparticles and incorporation of secondary (or more) metal species on seeds forming random alloy nanoparticles. The prerequisites to achieving homogeneous random mixing of metals are similar physicochemical properties of the constituent metals and sufficiently high temperatures. Seed-mediated synthesis methods can be divided into two types: direct diffusion of secondary metals into seed particles and conversion of core-shell nanoparticles into random alloys. Sun and co-workers successfully synthesized PdCu nanoparticles by controlling the diffusion of Cu into Pd nanoparticles at a high temperature (280°C).Murray and co-workers reported a generalized approach for thermal conversion of core-shell structures into random alloys starting from Au seeds .Interestingly, the temperature required for the complete diffusion of each secondary metal (Ag, Pt, Hg, Sn, or Cd) varies from metal to metal. Similarly, Ag@Auand Au@Agcore@shell nanoparticles can be converted into randomly alloyed nanoparticles.
## Control over size and composition
The catalytic performance is largely affected by the size and composition of nanoparticles. These two parameters are sensitive to the reaction conditions, such as the concentration of precursors, stabilizing agent, solvent, reductant, and reaction temperature.
The size of the nanoparticles is primarily determined in the nucleation and growth stages, especially by the number of nuclei and the total amount of precursor .That is, the regulation of reaction rates gives rise to the variation in size. The effect of various control factors on the size of nanoparticles was extensively investigated in the synthesis of Pt 3 Co nanoparticles as a model system .The alloying of Pt with Co can be achieved by injecting cobalt carbonyl into a preheated Pt precursor solution, and the high temperature induces the thermal decomposition of cobalt carbonyl, triggering the nucleation. The size of the nanoparticles was influenced by control factors such as injection temperature, the ratio of precursor to stabilizing agent, and the ratio of two precursors. As the injection temperature increased from 145 to 220°C, the size decreased from 10 to 3 nm as more nuclei are produced. Similarly, the nucleation rate could be regulated by adjusting the amount of stabilizing agent as more stabilization decreases the supersaturation and suppresses the nucleation, resulting in the larger particles. Interestingly, the molar ratio between the precursors also affects the size. The smaller particle size obtained by a larger amount of Co is ascribed to the larger number of nuclei formed due to the increased nucleation rate. On the contrary, an increased concentration of Pt leads to the formation of larger particles as the precursor amount is increased. On the other hand, the composition of nanoparticles can be readily tuned by regulating the molar ratio of metal precursors. For example, Sun and co-workers successfully synthesized composition-controlled PdPt nanoparticles by varying the molar ratio of Pd and Pt precursors .This strategy has been applied to different synthetic routes that control the composition of alloy nanoparticles.Unfortunately, the size and composition are coupled to each other and are not independently tuned in most of the cases because of the intricate interactions among several control factors.
## Facet control
The synthesis of multimetallic nanoparticles with well-controlled morphology is important in many catalytic reactions since the differences in exposed facets affect the sorption properties of reactants, intermediates, and products, which change the overall catalytic performance.Typically, the evolution of a specific facet is determined in the growth stage, and the surface energy is critical for the resultant shape of the nanoparticles. If a facet possesses lower stability than other facets, the growth or coalescence over the facet takes place with fast kinetics for minimizing the overall surface energy of the nanoparticles, leading to diminution or even elimination of the facet.As a result, surfaces with lower surface energies tend to remain and to be exposed dominantly. The orientation of the facet can be guided by the chemicals involved in the reaction that have specified affinity toward certain facets, and the extent of facet dominance can also be controlled by the reaction temperature and time.
Yan and co-workers studied the facet-selective synthesis of PtPd alloy nanoparticles and investigated the effect of chemical species.Although poly(vinylpyrrolidone) (PVP) acts as a surface-stabilizing agent to prevent the aggregation of nanoparticles, it shows lacking facet-directing ability. Instead, halide ions (for cubic nanoparticles) and a combination of HCHO with C 2 O 4 2− (for tetrahedral nanoparticles) worked critically to determine the facet. Similarly, in another report, the presence of HCHO with C 2 O 4 2− induced the formation of (111) . Control over the size and composition of NMMNs. a) Schematic representation of a relationship between nucleation rates and size of resultant nanoparticles in the absence of Ostwald ripening. b) Dependence of the size of Pt 3 Co on the reaction temperature, the ratio of surfactant to precursor, and the molar ratio between precursors. Reproduced with permission.Copyright 2003, American Chemical Society. c) TEM images of Pd and PdPt nanoparticles with controlled composition, and d) correlation between the proportion of Pt precursor and atomic percent of Pt in the alloy nanoparticles. Reproduced with permission.Copyright 2011, American Chemical Society.
facet-dominant (icosahedral) nanoparticles, verifying the facetspecificity of the chemicals.Although C 2 O 4 2− commonly induces the exposure of (111) facet as in the case of PtPd tetrahedra, the evolution of multiply-twinned icosahedral shape, instead of single-crystalline tetrahedral shape, was observed. The deceleration in the reduction rate by the decreased amount of HCHO promoted the formation of multiply twinned metal nanocrystals (icosahedra). The role of halide ions in the shape control of PdPt alloy nanoparticles was further studied by Huang and coworkers.In this study, NaCl and NaI were added to the reaction solution, and the addition resulted in nanoparticles terminated withand (100), respectively. Interestingly, when halide ions (Cl − , Br − , I − ) were mixed together, the iodide ion produced cu-bic nanoparticles regardless of the other halide ions. Another shape-directing agent investigated systematically is CO, which has been commonly used for Pt-based nanoparticles with controlled shapes.In the synthesis of Pt 3 Ni cubes and octahedrons, the shape of the nanoparticles became relatively irregular without CO, implying the significance of CO than oleic acid and oleylamine for shape control.In another study conducted by Peng and co-workers, the amount of Ni precursor was adjusted to obtain octahedral-shaped PtNi nanoparticles by applying the preferential adsorption of CO for (100) on Pt and (111) on Ni.Among the shape-controlled nanocatalysts, multimetallic nanoparticles with high-index facets (HIFs) have attracted much attention owing to their low-coordinated surface atoms that Reproduced with permission.Copyright 2011, American Chemical Society. d) Electron microscopy images of HOH AuPd nanoparticles synthesized in the presence of OTAC and Cu 2+ ions. Reproduced with permission.Copyright 2011, American Chemical Society. e) Scheme for inducing HIF on nanoparticles via alloying/dealloying route. Reproduced with permission.Copyright 2019, AAAS. f) Control over concavity of Pt 3 Co nanocubes by the amount of deionized water. Reproduced with permission.Copyright 2017, American Chemical Society. exhibit enhanced activity for several electrocatalysis.However, the instability of the facets makes it difficult for the facets to survive under reaction conditions. While extensive progress has been made in the fabrication and systematic studies on monometallic systems with HIF, there are very few studies for multimetallic systems, owing to the higher complexity compared to monometallic systems.
The most prevalent strategies for the fabrication of HIF-bound NMMNs are introducing facet-directing agents and controlling the reaction kinetics of the precursors. One pioneering work is the synthesis of hexoctahedral (HOH) AuPd nanoparticles.The synthesis was aided by the presence of octadecyltrimethylammonium chloride (OTAC) and Cu 2+ ions, which facilitated the formation of the HOH shape and guaranteed the homogeneous phase by underpotential deposition, respectively. Further, OTAC is used as a capping agent in the synthesis of rhombic dodecahedral and trisoctahedral AuPd nanoparticles.During the synthesis of PtNi, Sun and co-workers identified the role of glycine in the HIF structure in the presence of PVP, and obtained the morphological evolution from concave nanocubes to nanocubes to hexoctahedra with the increase in glycine content.They attributed the formation of HIFs to the lowered reduction rate of metal ions through strong coordina-tion with glycine, thereby changing the nucleation and growth rates. Moreover, the combination of glycine and PVP has been utilized to synthesize other Pt-based bimetallic and trimetallic alloy nanoparticles with controlled HIF.Mirkin and co-workers discovered that the exposed facets of gold nanoparticles could be diversified from low-index to high index by adjusting the added amount of shape-directing silver species in solution-phase synthesis.Recently, they extended the use of foreign shape-directing elements for the evolution of HIF on mono-and bimetallic nanoparticles in a general and scalable manner via solid-state synthesis.The whole procedure consists of two steps: colloidal synthesis of noble-metal-based nanoparticles with irregular shapes followed by thermal treatment in the presence of foreign metals, such as Sb, Bi, Pb, and Te, to induce shape directing via alloying/dealloying.
It has been reported that selective etching is an effective strategy.Shen and co-workers reported the construction of deeply excavated Pt 3 Co nanocubes via oxidative etching by Cl − /O 2 pairs during a solvothermal synthesis.The introduction of water into the reactor during the synthesis is critical for excavating surfaces because the presence of water breaks the kinetic equilibrium of surface capping CTAC and provides more O 2 , enhancing the corrosion rate. www.advancedsciencenews.com www.advancedscience.com
## Random alloys with hollow structures
One special morphology that has garnered much attention of researchers is the hollow (or open) structure due to its intrinsically large surface area, which makes it easy to maximize the atomic efficiency for catalysis. The fabrication of hollow nanoparticles is usually accomplished by template-mediated routes. The template can be either presynthesized or in situ formed during the synthesis, and later removed by etching, replacement, or migration by the Kirkendall effect, leaving a cavity inside a nanoparticle.
Etching is a popular method for removing templates to create voids.Xia and co-workers utilized a chemical etching route to fabricate PtPd nanocages with cubic and octahedral shapes.An aqueous solution containing FeCl 3 and HCl was used for the selective etching of the Pd core in Pd@Pt nanoparticles. The method was further applied to the synthesis of IrPd nanocages by etching Pd@Ir nanoparticles.On the other hand, composition-segregated nanoparticles were employed as precursors instead of core@shell nanoparticles. For example, a polyhedral PtNi nanoframe was produced by etching the Ni-rich phase with acetic acid to form nanoparticles with composition segregation.Similarly, PtCo nanoframes with rhombic dodecahedral morphology were obtained by chemical etching of Co in phase-segregated PtCo nanoparticles using nitric acid.The galvanic replacement reaction refers to the oxidation of less noble metal atoms coupled with the reduction of more noble metal ions. The process is thermodynamically driven based on the difference between the reduction potentials of the metals involved. Because less noble metals are sacrificed during the process, the galvanic replacement has been exploited to achieve nanoparticles with hollow shapes. For example, in situ formed Co nanoparticles were utilized as sacrificial templates for galvanic replacement with Pt ions.In the synthesis, Co remains and participates as a composing metal after being consumed as a template. It is suggested that, in the presence of PVP, which is known to trap metal cations, the excessive reducing environment by a remnant reductant allowed the formation of PtCo alloy through the coreduction of Pt ions and the leached Co ions. The shapes of the hollow nanoparticles were also readily tuned by a slight modification of the reaction conditions in the reactor. Lee and co-workers synthesized a series of RhCu nanoframes with various morphologies via galvanic replacement of the Rh precursor onto Cu nanoparticles.Control of reductants or temperature during galvanic replacement affects the shapes of the templates and the resulting nanoframes. Xia and co-workers demonstrated that the thickness of a hollow nanostructure could be precisely controlled via regeneration of a template, followed by galvanic replacement, using Ag-Au cubic nanocages as a model case.First, Ag-Au nanocages were prepared by galvanic replacement over sacrificial Ag nanocubes. The hollow structure was refilled with Ag core by l-ascorbic acid as a reductant, forming an Ag@Ag-Au core-shell structure, and then further galvanic replacement was conducted using these core-shell particles as a new template. After repeated reactions, the thickness of the nanoparticles could be regulated, and this technique was successfully applied to other combinations of metals and trimetallic systems.
The Kirkendall effect, driven by the difference in the diffusion rates of metals at the interface, can be utilized to fabri-cate hollow NMMNs, when the diffusion rate of core metal is higher than that of shell component. Yang and co-workers investigated the structural evolution of Cu@Pt core-shell nanoparticles in toluene for three weeks under ambient conditions with the controlled kinetics.Over time, the Cu atoms diffused toward the shell at a higher rate than the inward diffusion of Pt, leaving voids inside the nanoparticles and converting them into PtCu 3 nanoframes. The Kirkendall effect can be accelerated by increasing the diffusion rate of core metals at a sufficiently high temperature. For example, hollow cavities in the PtFe nanoparticles were constructed via the thermal treatment (800°C for 2 h) of Pt nanoparticles covered by an Fe 3+ -chelating polydopamine layer.The higher outward diffusion of Pt than the inward diffusion of Fe resulted in hollow PtFe nanoparticles. The simultaneous application of galvanic replacement and the Kirkendall effect could produce interesting hollow structures. The synthesis of concentric double-walled AuAg nanoboxes was achieved by sequential galvanic replacement and the Kirkendall effect.The voids inside the Ag templates were produced by the leaching of Ag + ions by galvanic replacement with Au ions at the outer surface. Once the voids formed, galvanic replacement between Au and Ag also occurred at the inner voids, constructing two separate Au layers on the inside and outside surfaces of the hollow Ag frame, which finally evolved into double-walled nanoboxes by the Kirkendall effect.
## Prevention of phase segregation
Generally, precursors with similar reduction rates easily form homogenous alloy nanoparticles. For example, PdAg alloy nanoparticles were readily formed because of the similar reduction potential of Ag with Pd (+0.8 V for Ag + /Ag and +0.9 V for Pd 2+ /Pd).Unfortunately, not all combinations of metals are eligible because a huge difference in the reduction rate leads to a sequential reduction of metals, producing phase-segregated products. This problem can be tackled in two ways: introducing additional chemicals or using bimetallic complexes.
As the precursor reduction rates are closely related to the reduction potential and stability of metal ions, they have been effectively regulated by the introduction of additional chemicals. Xia and co-workers showed that the presence of KBr significantly affected the compositional homogeneity of Pd-Pt nanoparticles after coreduction.Cubic-shaped homogeneous PdPt random alloys were produced with a KBr additive, while octahedral-shaped Pd@Pt nanoparticles were formed in the absence of KBr. In the presence of KBr, PtCl2− , and PdCl 4 2− transformed into PtBr 4 2− and PdBr 4 2− because of the stronger binding strengths of Br − than Cl − toward the metal ions. The ligand exchange narrowed the gap of the reduction rates between the Pt and Pd precursors by slowing them down. Kuo and co-workers have demonstrated that the reduction of HAuCl 4 and H 2 PdCl 4 by sodium citrate can be controlled by varying the amounts of CTAC and cetyltrimethylammonium bromide (CTAB) additives.By changing the ratio of Br − to Cl − (in CTAB and CTAC), core-shell nanoparticles were produced in the Cl − -rich case, while random-alloy nanoparticles were prepared in the Br − -rich case . The above results indicate that additives can play pivotal roles in the synthesis of random alloys by controlling the reduction rates of theCopyright 2015, AAAS. f) Ideal model structure of a RuCu nanoframe and the temporal TEM images obtained at g) 3 min, h) 6 min, i) 13 min, and j) 30 min of reaction. Reproduced under the terms of the Creative Commons CC BY license.Copyright 2015, The Authors. Published by Wiley-VCH. k) Structural evolution of AuAg dual-walled nanoboxes over reaction time, and corresponding l) HAADF-STEM, and m) EDS mapping images. Reproduced with permission.Copyright 2011, AAAS. . a) Schematic diagram for describing an aspect of nanoparticle formation in the presence of halide species with different ratios. EDS mapping images of b) AuPd core-shell nanoparticles in Cl−rich environment and c) random alloy in Br−rich one. Reproduced with permission.Copyright 2016, American Chemical Society. d) Structures of dual metal precursors with controlled metal ratio, and e,f) TEM images of corresponding nanoparticles produced by thermal decomposition (scale bar: 100 nm). Reproduced with permission.Copyright 2009, American Chemical Society.
precursors. Moreover, the reduction rates of metal precursors can also be affected by their concentrations. Sun and co-workers reported that similar rates for the nucleation and growth of Ag and Au precursors could be achieved by using a much larger amount of silver precursor (20 times) than Au, which compensated for the much higher reduction potential of Au (1.5 V for Au 3+ /Au).Another strategy for overcoming this difference is the use of a bimetallic precursor instead of a mixture of monometallic precur-sors. Lukehart and co-workers first reported the formation of PtFe nanoparticles using a bimetallic precursor.The Fe 3 Pt 3 (CO)complex, prepared by reacting Fe(CO)with Pt(C 7 H 10 ) 3 , underwent ultrasonication to produce precipitates of PtFe nanoparticles with nearly 1:1 composition. Thanh and co-workers extended the synthesis of PtFe and PtFe 4 nanoparticles by the thermal decomposition of [Fe 3 Pt 3 (CO)] 2− and [Fe 4 Pt(CO) 16 ] 2− , respectively (Figure 6d-f).www.advancedsciencenews.com www.advancedscience.com
## Single-atom alloys
SAA is a type of single-site catalyst in which a trace amount of catalytically active metal is atomically dispersed on the surface of the host metal. Owing to the low coordination, isolation, and well-defined nature of active sites, SAAs have shown excellent selectivity as well as outstanding atomic efficiency, attracting attention from researchers as good model catalysts for various kinds of reactions.In addition to their role as model systems, SAAs in the form of nanoparticles have recently been utilized in the fields of practical applications, such as electrochemical water splitting and fuel cell catalysis.The synthesis of SAA nanoparticles with various routes has been investigated over the past several years to fully exploit the above-mentioned advantages of SAA nanoparticles. In fact, the representative synthetic methods for SAA nanoparticles do not vary much from those for random alloys and core-shell nanoparticles. For the synthesis of SAA nanoparticles, the amount of the dopant species should be much lower than that of the host so that the dopant atoms are completely isolated as single atoms on the host. On the other hand, the synthesis of random alloys or core-shell nanoparticles requires sufficiently higher amount of dopants than that of SAA nanoparticles.
Toshima and co-workers reported the first synthesis of SAA nanoparticles using galvanic replacement.They prepared isolated Au sites on Pd clusters by adding an aqueous solution of HAuCl 4 ⋅4H 2 O into dispersed Pd clusters. As the surface free energies of Pd atoms on Pd clusters differed according to their positions, Au preferentially displaced the Pd atoms at specific sites. Copper is one of the most intensively used hosts in the synthesis of SAA nanoparticles via galvanic replacement. Flytzani-Stephanopoulos and co-workers published a series of papers on the preparation of Pd 1 /Cuand Pt 1 /Cunanoparticles by galvanic replacement of Cu. Wei and co-workers also used Cu nanoparticles as a host for synthesizing Pt 1 Cu SAA nanoparticles.Besides spherical nanoparticles, Cu nanosheets and nanocubes were also used as host materials for galvanic replacement with Pd.With the different morphologies of the hosts, the catalytic properties of Pd 1 Cu were significantly adjusted according to the different exposed facets. Wasserscheid and co-workers used Ga nanoparticles as host materials and successfully synthesized Pdand Rhsingle-atom alloys on a Ga host using (NH 4 ) 2 [PdCl 4 ] and RhCl 3 ·H 2 O as precursors, respectively. Similarly, the synthesis of SAA using Ni as a host was also achieved by Zeng and co-workers.It is important to note that the precursor concentration of a doped metal should be carefully adjusted. Sykes and co-workerscontrolled the concentration of Pt in the solution for galvanic replacement with Cu nanoparticles. From the EXAFS analysis, the formation of isolated Pt singleatomic sites was confirmed by the absence of Pt-Pt bond for the sample prepared with low Pt concentration, while one with higher Pt concentration showed the presence of Pt-Pt bond.
Unfortunately, the galvanic replacement method can be used only when the dopant metal is more noble than the host nanoparticles. On the other hand, a sequential reduction can be a suitable alternative if a noble metal host is to be used. In sequential reduction, one metal species is reduced first to form seed nanoparticles, followed by the reduction of a secondary metal onto the surface of the seeds. To date, most reports have used Au nanoparticles as the seed material, with Pd as a dopant. Various combinations of precursors and reductants have been employed for sequential reduction: NaBH 4,ethylene glycol,ascorbic acid,and oleylaminefor Pd(NO 3 ) 2 , formic acid for H 2 PdCl 4 ,and H 2 bubbling for PdCl 2. All the aforementioned methods require a multistep procedure in the solution phase, with complexity, cost, and time. Coreduction provides a facile route for obtaining SAA nanoparticles. A representative example was reported by Shishido and coworkers.They synthesized Pd-Au alloy nanoparticles through coreduction method, and obtained Au 1 /Pd SAA, PdAu random alloy, Pd 1 /Au SAA nanoparticles by simply adjusting the ratio of Pd precursor to Au precursor under the same reaction conditions. The synthesis of several SAA nanoparticles such as Pd 1 /Ag,Cu 1 /Pd,Pd 1 /Au, and Rh 1 /Au,has been achieved by simply adjusting the molar ratio of precursors with suitable reductants.
As in the case of random alloys, incipient wetness impregnation also provides a simple way to synthesize SAA nanoparticles by controlling the ratio of impregnated metal precursors. Zhang and co-workers reported a series of Pd-based SAA nanoparticles with different Pd/Ag ratios on silica support by adjusting the molar ratio of Pd/Ag precursors from 0.005 to 0.025.Additionally, they reported the synthesis and catalytic application of Pd 1 /Cu nanoparticles on SiO 2 .They changed the ratio of Pd to Cu with fixed Cu loading or fixed Pd loading. In the same manner, -Al 2 O 3 was also used as a support material.Although IWI is a facile technique to prepare SAA nanoparticles, it should be noted that the resultant products usually have inhomogeneity in sizes and compositions from particles to particles, which can be a potential problem in particular cases where high uniformity is required.
Lastly, the atomic layer deposition (ALD) technique can also be used to prepare SAA nanoparticles. Lu and co-workers reported the synthesis of Pd 1 /Ni nanoparticles on SiO 2 support using ALD.With an increase in the number of ALD cycles, the loading of Pd in the nanoparticles increased, eventually losing the single atomic feature. Although the metal atoms deposited by ALD cycles can form isolated atomic sites due to the steric hindrance by bulky ligands, aggregation of metal atoms often occurs after multiple ALD cycles.To prevent the aggregation of the deposited metal species, the Pd ALD was performed on Ni/SiO 2 at a relatively low temperature (150°C). Sun and co-workers studied the formation of Pt single-atomic sites on Pd nanoparticles using the ALD technique.Interestingly, isolated Pt atoms were embedded on octahedral Pd nanoparticles. At the same time, a few atomic layers of Pt were formed on cubic Pd nanoparticles under the same operating conditions, mainly due to the different surface energies of Ptand Pt (100).
## High entropy alloys
HEAs usually refer to alloys of multiple (five or more) elements with maximized entropy from nearly equal contents.Recently, HEAs have come into the spotlight, and the synthesis of HEAs in the nanometer size region has become important for maximizing the utilization of surface atoms in catalysis with various kinds of atomic ensembles.Copyright 2011, Nature Publishing Group. c) Schematic illustration of sequential reduction of Au and Pd for the fabrication of Pd 1 /Au SAA nanoparticles with varied Pd loading, and d) the TEM images of corresponding resultants. Reproduced with permission.Copyright 2018, American Chemical Society. e) Schematic illustration of selective deposition of Pd atoms on a Ni nanoparticle through ALD technique. f) HAADF-STEM image for Pd 1 /Ni SAA nanoparticles (Scale bar: 2 nm). g) The profile of line X-Y shows the single atomic feature. h) Pd loading on the Pd/Ni samples as a function of ALD cycle number. Reproduced under the terms of the Creative Commons CC BY license.Copyright 2019, The Authors, published by Springer Nature.
Unfortunately, traditional wet-chemical routes used to synthesize bi-and trimetallic nanoparticles tend to produce phasesegregated products of polyelemental systems,because of the differences in the reactivity of multiple metal precursors and immiscibility of metal elements. Therefore, ingenious synthetic methods featuring rapid heating/cooling have been developed to prepare polyelemental solid solutions with a reliably homogeneous phase. The rapid heating ensures simultaneous reduction of metal precursors with different chemical potentials, while rapid cooling also plays a pivotal role in preventing the phase separation among immiscible elements, thereby preventing a compositional inhomogeneity in the prepared HEA nanoparticles.Carbothermal shock (CTS) synthesis was shown to be an effective technique for synthesizing HEA nanoparticles, which was proposed by Hu and co-workers.First, carbon nanofiber was immersed in a precursor solution for loading metal species. Then, a thermal shock (≈2000 K) was applied to the precursor-loaded carbon nanofiber using Joule heating by an electrical pulse. They reported the successful synthesis of welldispersed nanoparticles comprising up to eight elements using this method. In the control experiments, prolonged duration at high temperatures produced larger particles, while a slower cooling rate led to phase segregation. A subsequent study compared the products synthesized using the CTS method with those from the IWI method.Although the morphology and composition of the two cases were not very different, the former showed superior stability to the latter during catalysis, while HEAs with a larger number of components further enhanced the stability. The reliability and generalizability of the technique have facilitated the synthesis of other HEA nanoparticles comprising various combinations of elements.Lu and co-workers suggested that pyrolysis using a fastmoving bed is a reliable synthetic method for synthesizing HEA nanoparticles on granular supports.In this method, a boat containing precursor-loaded support was moved into the center of the furnace preheated to 923 K with a propulsion speed of ≈20 cm- s −1 . When the fast-moving bed passed through the heating zone, the temperature of the boat rapidly increased within only 5 s, inducing simultaneous pyrolysis of the precursors and formation of denary HEA nanoparticles. Contrary to the case of the fast-moving bed, phase-separated nanoparticles were obtained when using a fixed or slow-moving bed because the metal precursors were reduced sequentially as the temperature increased. It is worth noting that this technique is generally adopted for various supports, such as carbon black, graphene oxide (GO), zeolite, and alumina.
The aerosol droplet-mediated technique is a continuous process for the fabrication of HEA nanoparticles with high scalability. In this process, a solvent containing the precursors was nebulized and carried into a heating zone using a carrier gas. While passing through the heating zone, HEA nanoparticles were formed quickly, followed by rapid quenching after their removal from the zone. Zachariah and co-workers first reported this method with excellent compositional homogeneity in a single nanoparticle.They were able to form alloys of several combinations of immiscible metals in a particle. Hu and co-workers successfully synthesized hollow HEA nanoparticles by introducing a gas-blowing agent.Citric acid dissolved in ethanol rapidly decomposes to CO 2 and H 2 O when the droplet passes through a heating zone, forming hollow cavities inside the nanoparticles. The aforementioned approaches collectively show that rapid heating and cooling induce simultaneous reactions of metal precursors, which otherwise would react at different rates.
Compared to the methods using rapid heating and cooling, the wet-chemical route is disadvantageous because a gradual change in temperature induces a sequential reaction of precursors, which results in the inhomogeneous elemental distribution in each particle. Nevertheless, wet-chemical approaches for the synthesis of HEA nanoparticles have been steadily investigated. Iversen and co-workers prepared binary to quinary nanoparticles with homogeneous phases via a solvothermal method.They observed a single homogeneous phase with a face-centered cubic (fcc) structure (which is unusual for Ru-based alloys) by using metal acetylacetonate precursors. Additionally, as the reaction temperature exceeded a certain threshold, products with hcp structures were produced. An ultrasonication-assisted method was also applied to synthesize PtAuPdRhRu HEA nanoparticles with diameters less than 3 nm.Under intense ultrasonication, acoustic cavitation conditions are generated with the explosion of bubbles transiently, creating extremely high pressures (≈2000 atm) and high temperatures (5000°C) in a local region. With the massive amount of energy, the precursors were coreduced, forming HEA nanoparticles. Skrabalak and coworkers demonstrated the synthesis of core-shell mediated HEA nanoparticles.They used PdCu@PtNiCo nanoparticles as a precursor for HEAs. The core-shell nanoparticles were deposited on carbon support to prevent coalescence and then annealed at high temperatures to promote the diffusion and intermixing of metal elements.
As in the case of monometallic and other heterostructured nanoparticles, it can be inferred that the size and exposed facets influence the catalytic performance of HEA nanoparticles. However, the control over the morphology of HEA nanoparticles has not been systematically studied, and the above-mentioned routes still have intrinsic limitations in morphological control, because most of the current methods apply extremely high temperature, which makes it difficult to control the size and shape of HEAs. Therefore, a more advanced synthetic method is required for a better morphology control.
## Ordered-intermetallics
Alloys with a regular atomic arrangement between constituent metals of more than two different types of elements are termed intermetallic alloys.Atomically ordered structures, such as L1 0 , L1 1 , L1 2 , and B2 phases, feature highly effective electronic (d-d orbital) interactions between different metal elements and stabilized metal elements (especially 3d transition metals) that are vulnerable to oxidation or leaching. Over the last two decades, tremendous efforts have been devoted to prepare intermetallic nanoparticles that are highly ordered and precisely tuned in composition and morphology.Herein, we describe the developments in synthesizing intermetallic alloy nanoparticles that provide significant insights into a realization of desirable intermetallic nanomaterials. Our focus is not on listing possible synthetic methods but on describing . Reproduced with permission.Copyright 2018, AAAS. b) Schematic illustration of fast-moving bed approach, and comparison with the case using fixed-bed. Reproduced under the terms of the Creative Commons CC BY license.Copyright 2020. Published by Springer Nature. c) Schematic illustration of the fabrication of hollow HEA nanoparticles via aerosol-mediated synthesis (top) and the EDS mapping images of a representative nanoparticle (bottom, scale bar: 100 nm). Reproduced with permission.Copyright 2020, Wiley-VCH. www.advancedsciencenews.com www.advancedscience.com key advancements in precisely controlling the structural properties of intermetallic nanoparticles. We begin our discussion on Pt intermetallics and then proceed to Pd and Au intermetallics.
## Pt intermetallics
Among all noble metal-based intermetallic alloy nanoparticles, Pt-Fe intermetallics are most common and well approved. Sun and co-workers first reported the precise synthesis of spherical PtFe nanoparticles and annealing of their assembly that has had a profound impact on the emergence of several succeeding works on intermetallic nanoparticles with desired properties.The spherical PtFe nanoparticles were self-assembled into hexagonal close-packed or cubic packed structures on a substrate , followed by thermal annealing in an inert gas that allowed the rearrangement of the constituent metal atoms. Eventually, it led to atomically ordered PtFe nanoparticles with the L1 0 phase (face-centered tetragonal) . Since then, the postannealing process on presynthesized random alloy nanoparticles has been the main strategy to obtain ordered Pt-alloy nanoparticles, while the size of these nanoparticles often increases due to aggregation.Various types of protecting layers have been adopted to prevent aggregation that led to uniformly sized intermetallic nanoparticles upon annealing. SiO 2 -coatingand NaCl-blendingsuccessfully protected PtFe nanoparticles from aggregation and were easily removed after annealing. Sun and co-workers developed an effective stabilization strategy by coating random-alloyed PtFe nanoparticles with MgO in benzyl ether, thereby maintaining a uniform size distribution after annealing to obtain ordered PtFe nanoparticles .In a subsequent study, the authors employed this approach to obtain fully ordered PtFe nanoparticles supported on carbon by annealing MgO-coated PtFe-Fe 3 O 4 dumbbell-shaped nanoparticles at 700°C (5% H 2 in Ar).The Hyeon group contributed to coating random-alloy PtFe nanoparticles with polydopamine to in situ form N-doped carbon shells during annealing. The N-doped carbon coating protected the nanoparticles from coalescence while exposing the active catalytic surface of the intermetallic alloy nanoparticles toward the electrolyte without removing the shell .Despite its less uniformity, the synthesis of intermetallic PtFe nanoparticles has recently been achieved by a wet-chemical approach. Hou and co-workersincluded halide ions (Cl − , Br − , or I − by using NH 4 Cl, KBr, or KI, respectively) during the formation of PtFe nanoparticles in oleylamine so that the formation of the ordered L1 0 -phase became thermodynamically favorable. The above-mentioned strategies can readily be applied to the preparation of intermetallic Pt 3 Fe or PtFe 3 nanoparticles of the L1 2phase by controlling the atomic ratio between Pt and Fe in initial random-alloy nanoparticles.Interestingly, adding a third metal species to the Pt-Fe system to obtain doped or trimetallic PtFeM nanoparticles has been proven to promote atomic ordering and tune the strain of the Pt surface. Sun and co-workers demonstrated that the addition of Au to the PtFe system facilitated the formation of the L1 0 phase when annealed at 600°C.The embedded gold atoms were segregated toward the surface and generated internal vacancies enabling the effective transformation of random-alloy nanoparti-cles into intermetallic PtFe nanoparticles with Au-enriched surfaces . Wang and co-workers introduced Ag into the wet-chemical synthesis of L1 0 -PtFe nanoparticles.By controlling the atomic percentage of Ag, they observed at a certain point (≈23 at%) Ag chemically segregates on PtFe with the simultaneous formation of an ordered L1 0 -PtFe phase without a postannealing process, and the optimum point for the most significant ordering was at 29 at% of Ag. A trimetallic PtFeCu system with controllable atomic composition was systematically studied by Sun and co-workers.Starting from presynthesized PtFe nanoparticles, a trimetallic solid solution was prepared by the reduction of Cu(acac) 2 and diffusion of Cu into PtFe random alloy in 1-octadecene and oleylamine at 240°C. Although the inclusion of Cu in PtFe was detrimental for atomic ordering upon annealing, Cu played an important role in modulating the compressive strain applied to the L1 0 -PtFe intermetallic nanoparticles, thus improving the catalytic activity. On the other hand, the substitution of 10% of Pt in Pt 3 Fe with Pd was recently found to promote atomic ordering in the L1 2 phase through a simple impregnation method.The role of Pd was not limited to that, but also included the formation of PdH x to tune the electron density of the intermetallic nanoparticles.
Analogous to the Pt-Fe intermetallic system, there has been a great advancement in the synthesis of Pt-Co intermetallic nanoparticles. Although much attention has been paid to the importance of L1 0 -PtCo or L1 2 -Pt 3 Co nanoparticles for oxygen reduction electrocatalysis,precise synthesis of L1 0 -PtCo nanoparticles on carbon support was recently realized by Sun and co-workers .Randomly alloyed A1-PtCo (fcc) nanoparticles were prepared by thermal reduction of Pt(acac)and Co(acac) 2 in oleylamine, which were loaded on a carbon support and annealed without any protecting layer. No sign of aggregation was found on the intermetallic nanoparticles owing to the low loading of 8 wt% and the high resistance of the PtCo system against aggregation. Similar to the case of PtFe, where Au was added to aid atomic ordering, PtCoAu nanoparticles with Auenriched L1 0 -PtCo structure have been studied recently.The A1-PtCoAu nanoparticles underwent annealing at 650°C to be converted to the L1 0 core and PtAu alloy shell of ≈2 atomic-layer thickness. Furthermore, various trimetallic L1 0 -PtCoM (M = Fe, Ni, Cu or Zn) nanoparticles were prepared by substituting half the amount of Co(acac) 2 with M(acac) 2 or M(acac)and adding a borane-tert-butylamine complex in the colloidal synthesis of random trimetallic alloys that underwent annealing.Li and co-workers applied several dopants (W, Ga, or Zn) to L1 0 -PtCo nanoparticles with a size of ≈3 nm by mixing a small amount of W(CO), Ga(NO 3 ) 3 , or Zn(acac) 2 with Co(acac) 2 in the course of CoO shell formation on presynthesized Pt nanoparticles.After annealing the M-Pt/CoO core-shell at 700°C with MgO protection, the MgO shell on the intermetallic core was removed and the nanoparticles supported on carbon were further annealed at 400°C without severe aggregation.
The intermetallic Pt-Ni system is less familiar than Pt-Fe and Pt-Co because of the low tendency of atomic ordering between Pt and Ni even with annealing. Early efforts on the synthesis of PtNi intermetallic nanomaterials have tried different metal precursors and reducing agents to develop a novel set of reagents.Reduction of Li 2 NiCl 4 and Pt(COD)Cl 2 (COD = cyclooctadiene) by potassium triethylborohydride (KBH(Et)produced small- . a,b) TEM images of assemblies of monodispersed 6 nm PtFe nanoparticles and c) XRD patterns of 4 nm PtFe nanoparticles annealed to form L1 0 -PtFe phase. Reproduced with permission.Copyright 2000, AAAS. Schematic illustration of d) MgO-coating approach. Reproduced with permission.Copyright 2009, Wiley-VCH. e) Schematic illustration of N-doped carbon coating approach. Reproduced with permission.Copyright 2015, American Chemical Society. f) Schematic illustration of structural evolution of PtFeAu upon annealing. Reproduced with permission.Copyright 2012, American Chemical Society. g) Schematic and h) STEM image of an L1 0 -PtCo nanoparticle. Reproduced with permission.Copyright 2019, Elsevier. i) STEM image and j) XRD pattern of intermetallic PtNiN/KB. Reproduced with permission.Copyright 2020, American Chemical Society. k) HAADF-STEM image (left), enlarged HAADF-STEM image (right top) and simulated HAADF-STEM image (right bottom) of intermetallic PtCuN nanoparticle. Reproduced with permission.Copyright 2021, American Chemical Society. l) Schematic illustration of template-mediated synthesis of L1 2 -Pt 3 Co nanowires and their m) TEM (top) and HAADF-STEM (bottom) images. Reproduced with permission.Copyright 2019, American Chemical Society. n) TEM image of PtSnBi nanoplates and schematic illustration of its atomic arrangement (inset). Reproduced with permission.Copyright 2019, Wiley-VCH.
sized PtNi nanoparticles, which were converted to an intermetallic phase with severe aggregation by vacuum annealing at 500°C. Li and co-workers recently developed the synthesis of small-sized L1 0 -PtNi by annealing Pt/NiO x core/shell nanoparticles on a carbon support.The initially prepared ≈5 nm sized core/shell nanoparticles did not undergo severe aggregation upon annealing at 600°C because of their fine dispersion on the carbon support. Sasaki and co-workers introduced nitrogen into intermetallic PtNi nanoparticles by simply annealing carbon support, impregnated with Pt(acac)and Ni(acac) 2 , at 560°C under an ammonia atmosphere.Despite the high Pt loading of 18.3 wt% in the final product, highly dispersed ≈4.7 nm L1 0 -PtNiN nanoparticles were formed on the carbon support .
The remaining 3d late transition metals, Cu and Zn, were alloyed with Pt in an atomically ordered way to produce L1 1 -PtCu, L1 2 -PtCu 3 , L1 2 -Pt 3 Zn, and L1 0 -PtZn. Intermetallic L1 1 -PtCu has better electronic interaction between Pt and Cu than the L1 0 or L1 2 phases of Pt-Fe, Pt-Co, and Pt-Ni systemsalthough it has mostly been developed recently with only a few reports.Xin and co-workers prepared L1 1 -PtCu nanoparticles on carbon by first impregnating Pt(acac)and Cu(acac) 2 on carbon, followed by annealing at 800°C under 5% H 2 in Ar flow.Additionally, they doped nitrogen into the intermetallic PtCu alloy by flowing ammonia for 2 h at 500°C before the temperature was raised to 800°C . Moreover, a similar impregnation method was also used to prepare L1 2 -PtCu 3 nanoparticles on carbon by Abruña and co-workers through two-step annealing in flowing H 2 .In the intermetallic PtZn nanoparticles, Murray and coworkers reported monodisperse Pt 3 Zn random-alloy nanoparticles by colloidal synthesis and annealed them at 600°C to get L1 2 -Pt 3 Zn nanoparticles with slight aggregation for the first time.Huang and co-workers synthesized L1 0 -PtZn nanoparticles on multiwalled carbon nanotubes (MWNTs), and small-sized intermetallic nanoparticles less than 5 nm were obtained using a silica shell.The introduction of Zn into Pt/MWNT@mSiO 2 was achieved by the thermal reduction of Zn(acac) 2 in oleic acid and oleylamine at 330°C, and the final intermetallic phase was obtained after annealing at 600°C.
Other transition metals, such as Ti, V, Cr, Mn, and Sn, have been alloyed with Pt in an atomically ordered manner with limited control over their size, composition, and shape. The formation of L1 2 -Pt 3 M nanoparticles of 3d early transition metals such as Ti,V,Cr,and Mnwas achieved in a similar manner to those already described by annealing presynthesized colloidal random alloys with or without protecting agents. On the contrary, syntheses of small intermetallic PtSn (P63/mmc) and L1 2 -Pt 3 Sn nanoparticles have been possible without further annealing,although the formation of L1 2 -Pt 3 Sn often requires a postannealing step.Hard-template and wet-chemical approaches have mainly been employed to obtain shaped intermetallic nanoparticles. In a recent approach by Joo and co-workers, for 1D intermetallic nanostructure of Pt 3 Co, mesoporous silica (SBA-15) was infiltrated with a controlled amount of metal precursors followed by a typical annealing process under H 2 flow.Although the HF washing step was necessary to remove the silica template after annealing, this hard-template method uniquely provided a highly crystalline intermetallic Pt 3 Co phase with a controlled 1D morphology . Moreover, the use of a 3D sil-ica template (KIT-6) also generated a regular 3D network of intermetallic Pt 3 Co by the same process. Less controlled but more efficient method to obtain 1D L1 2 -Pt 3 Co nanowires was reported by Huang and co-workers using an oil bath to simply reduce metal precursors in oleylamine in the presence of a surfactant (CTAC) and a reducing agent (glucose).The same group recently reported 1D L1 2 -Pt 3 Sn through similar synthetic conditions by reducing metal precursors in oleylamine with a surfactant (stearyltrimethylammonium chloride, STAC) and Mo(CO).They controlled the aspect ratio of Pt 3 Sn nanofibers, from 22.5 to 13.4, with an average diameter of 1.7 nm, by changing the type of surfactant or solvent. 2D PtPb nanoplates with a hexagonal phase (P63/mmc) were also prepared similarly to 1D Pt-intermetallics.Metal acetylacetonate precursors were reduced with oleylamine as a surfactant and ascorbic acid in octadecene. The initially formed Pb 3 (CO 3 ) 2 (OH) 2 transformed into intermetallic PtPb nanoplates by the reduction and interdiffusion of Pt. Quan and co-workers recently discovered a facile wet-chemical route to prepare trimetallic PtSnBi nanoplates with the same hcp structure (P63/mmc).Three different metal precursors were mixed and thermally decomposed in an oleylamine/octadecene mixture with a surfactant (CTAB) and ascorbic acid. The molar ratio of Sn to Bi was readily tuned by controlling the amount of metal precursors. As can be seen in the aforementioned works, where Sn and Pb can form intermetallic phases with Pt by wet-chemical approaches, octahedronshaped PtPb (P63/mmc) with Ni dopingand cube-shaped L1 2 -Pt 3 Snnanoparticles were prepared by simply controlling the amount of ascorbic acid (PtPb) and via the hot-injection method (Pt 3 Sn). For 3D intermetallic nanoparticles, Joo and coworkers successfully synthesized very recently intermetallic L1 1 -PtCu nanoframes without losing the 3D frame morphology by silica protection.The initially formed PtCu randomly alloyed nanoframes on a carbon support transformed into an ordered intermetallic state at 600°C with a silica layer. The silica coating was then etched with HF, and the final form of the supported L1 1 -PtCu nanoframe with a Pt-skin layer was obtained after annealing at 300°C.
Lastly, recent novel approaches using the heat-induced formation of Pt-intermetallic nanoparticles have garnered attention due to several advantages of simultaneous formation of nanoparticles and their atomic ordering in a single annealing step, along with high potential scalability for practical applications. One important advancement is the application of zeolitic imidazolate framework (ZIF)-derived carbon (ZIF-C) as a source of both carbon support and transition metals.Similar kinds of ZIF materials, composed of Co and Zn metals, each coordinated by four imidazolate groups, were first pyrolyzed above 700°C to obtain highly porous carbons in the presence of Co. After loading a certain amount of Pt nanoparticles on them, an additional annealing process was applied to form intermetallic L1 0 -PtCoor L1 2 -Pt 3 Conanoparticles on ZIF-C.
A significant but more facile approach that can directly convert well-defined metal precursors into multimetallic nanoparticles has recently attracted a lot of attention because of the enhanced uniformity in size and composition compared to conventional impregnation methods.Even though the overall synthetic procedures are often simple, the precise design of the structure of multimetallic precursors itself or on support is critical to. a) HAADF-STEM image and high resolution TEM image (inset) of ZIF-derived supported L1 2 -Pt 3 Co nanoparticles. Reproduced with permission.Copyright 2018, AAAS. b) Schematic illustration of the synthesis of L1 2 -Pt 3 M nanoparticles on rGO by annealing and c) TEM image of the prepared Pt 3 Mn/rGO sample with STEM image (inset). Reproduced with permission.Copyright 2020, Wiley-VCH. d) Schematic illustration of the synthesis of L1 0 -PtFe nanoparticles on rGO by decomposition of bimetallic compound and e) TEM images of PtFe/rGO. Reproduced with permission.Copyright 2020, American Chemical Society. f,g) HAADF-STEM images of highly dispersed Pt 3 Y nanoparticles on mesoporous zeolite, h) their proposed atomic arrangement, and i) HAADF-STEM image of atomically dispersed La atoms by silanol nests of zeolite. Reproduced with permission.Copyright 2020, The Authors, under exclusive licence to Springer Nature Limited. the product obtained after annealing.Cui and co-workers provided a general strategy to prepare Pt-intermetallic nanoparticles on reduced graphene oxide (rGO) by impregnating metal precursors on GO in the presence of poly(vinyl alcohol) (PVA). PVA forms hydrogen bonds with GO and ensures uniform coating of metal precursors on it before freeze-drying and annealing at 700°C for 12 h to obtain ultrasmall (≈3 nm) L1 2 -Pt 3 M (M = Cr, Mn, Fe, Co) nanoparticles on rGO.The Hyeon group recently designed a synthetic concept to prepare intermetallic Pt nanoparticles with 1:1 composition by uniform decomposition of bimetallic compounds with the formula [M(bpy) 3 ] 2+ [PtCl 6 ] 2− (M = Fe, Co, or Ni; bpy = 2,2′-bipyridine).Bimetallic compounds with sizes of several hundreds of nanometers, simply prepared by mixing separate solutions of M 2+ , bpy, and PtCl2− , were each attached to GO and freeze-dried. The composite underwent thermal annealing up to 700°C, during which the surface of GO was uniformly coated with a thin layer of decomposed bimetallic compounds (≈500°C) and intermetallic nanoparticles grew while confined in the in situ formed Ndoped carbon shell surrounding them. Despite the 6 h annealing period at 700°C, the L1 0 -PtFe, L1 0 -PtCo, and PtNi nanoparticles maintained their uniform sizes, while the constituent metal atoms migrated enough to form intermetallic phases (for PtFe and PtCo).
Among many efforts to obtain intermetallic Pt-alloy nanoparticles with low reduction potential metals,Ryoo and co-workers successfully prepared highly dispersed and atomically ordered L1 2 -Pt 3 Y or L1 2 -Pt 3 La nanoparticles in mesoporous zeolite by creating silanol nests (through degallation of mesoporous gallosilicate zeolite) that could anchor platinum and rare-earth elements (Y or La) as single atomic species upon impregnation.The atomic dispersion of Y and La, as confirmed www.advancedsciencenews.com www.advancedscience.com by HAADF-STEM images, was considered to be critical for their diffusion into Pt and the formation of intermetallic phases despite their low chemical reduction potential. In summary, the above-mentioned recent references demonstrate that novel approaches to thermally control the nucleation and growth processes of intermetallic nanoparticles are now actively being discussed and provide valuable insights to obtain desirable nanomaterials (whether uniform or uncommon structures) via simple and scalable methods.
## Pd intermetallics
Unlike Pt-intermetallics, the domain of ordered Pd-intermetallic nanoparticles has been quite limited until recently. The most studied category is the Pd-Cu system, as the B2-PdCu (bodycentered cubic) phase is readily formed by annealing below 400°C. Synthesis of monodisperse A1-PdCu nanoparticles with sizes of ≈5 nm was carried out in an oil bath by Huang and co-workers.After annealing at 350°C (or 375°C), the A1-PdCu nanoparticles evenly distributed on a carbon support were transformed into ordered B2-PdCu without aggregation. They also added a third metal species such as Co or Ni (of about 20 at% in trimetallic alloy) into the PdCu system and obtained monodisperse B2-PdCuM (M = Co or Ni) nanoparticles on carbon after annealing.Skrabalak and coworkers successfully synthesized B2-PdCu via a wet-chemical approach that produced monodisperse B2-PdCu nanoparticles from the co-reduction of PdBr 2 and Cu(ac) 2 (ac: acetate) at 235°C on presynthesized disordered A1-PdCu seeds.The authors attributed the phase transition from A1 to B2 to the increase of size from 6.8 nm (of A1-PdCu seeds) to 11.9 nm (B2-PdCu nanoparticles), which made each nanoparticle overcome the activation barrier for atomic ordering. In the case of PdZn, an intermetallic phase is manifested as L1 0 -PdZn, which can be simply prepared at 500°C by impregnation method.Zhang and co-workers reported the colloidal synthesis of nearly spherical L1 0 -PdZn nanoparticles and nanosheets of L1 0 -PdZn, L1 0 -PdCd, and L1 0 -PdZnCd phases using Mo(CO) 6.Metal acetylacetonate precursors were coreduced at 325°C in the presence of Mo(CO), which produced atomically ordered nanosheets with thicknesses of less than 5 nm. Moreover, Chen and co-workers recently adopted a simple impregnation method on ZIF-C to prepare sub-2 nm L1 0 -PdZn nanoparticles after annealing at 400°C.The ZIF-C (obtained by calcining ZIF-8 at 650°C) was impregnated with controlled amounts of Pd and Zn precursors to obtain 1.2, 1.8, 2.7, and 10 nm-sized ordered PdZn nanoparticles with restricted aggregation due to the rich porous structure of ZIF-C.
Similar to several reports on Pt intermetallic nanoparticles,core-shell Pd@FeO x nanoparticles were also prepared for conversion into ordered L1 0 -PdFe nanoparticles of uniform sizes by annealing at 600°C on a carbon support.Moreover, carbon-supported L1 2 -Pd 3 Fe nanoparticles were directly synthesized by heat-induced formation from KCl-protected precursors on carbonor cyanogel-coated carbon.On the contrary, Pd-Ni or Pd-Co systems are much less known than those already described because of the greater difficulty in the formation of an intermetallic phase through annealing.On the other hand, L1 2 -Pd 3 Pb nanomaterials have been recently developed via facile oil bath methods, providing nanoplates and 3D nanowire networks.Finally, the intermetallic Pd-Bi system covers a wide range of ordered phases; however, control over size and shape has been largely limited.Recent efforts on this highly complicated system have offered supported nanoparticles of different intermetallic phases, such as Pd 3 Bi nanoparticles, by electrochemical dealloying of carbon-supported -PdBi 2 ,-PdBi nanoparticles prepared via impregnation reduction at 600°C,and Pd 31 Bi 12 nanoparticles electrochemically deposited on carbon support at room temperature.
## Au intermetallics
The scope of Au intermetallic nanomaterials is narrower than that of Pt and Pd intermetallic nanomaterials. Early works by Schaak and co-workers provided valuable insights into the formation of L1 0 -AuCu and L1 2 -AuCu 3 nanoparticles.Although the obtained nanoparticles were polydisperse and irregularly formed, they found that the ordered L1 0 -AuCu phase starts to nucleate at 200°C, and the phase is maintained between 200 and 400°C upon annealing in Ar atmosphere, while the L1 2 -AuCu 3 phase nucleates at ≈300°C by diffusion of Cu into the AuCu intermediate.Thereafter, they reported a wet-chemical approach for the preparation of L1 0 -AuCu and L1 2 -AuCu 3 nanoparticles and nanowire networks by using polyols to reach up to 310°C.Recent works successfully prepared monodisperse L1 0 -AuCu via seed-mediated growth in wet-chemical systems using oleic acid or oleylamine. Yang and co-workerstuned the degree of ordering in an AuCu nanoparticle by controlling the temperature (200-300°C) and time (50-100 min), while Zhu and co-workersobtained disordered and ordered AuCu nanoparticles by controlling the temperature (210 or 290°C), both starting from presynthesized Au seed nanoparticles. Another recent work published by Skrabalak and coworkers started from disordered A1-AuCu nanoparticles to prepare monodisperse L1 0 -AuCu at 280°C with an increase in size from 6.5 to 10.3 nm.The seed-mediated approach was also used to prepare intermetallic AuSn (NiAs-type) and AuSn 2 (orthorhombic) nanoparticles via the diffusion of Sn into presynthesized Au nanoparticles upon hydride reduction.By controlling the amount of Sn precursor, single-phased or multiphased AuSn/AuSn 2 intermetallic nanoparticles could be readily prepared. Other Au intermetallic nanoparticles such as ′-AuZn, R1-Au 3 Zn,L1 2 -Au 3 M (M = Fe, Co or Ni),and L1 2 -Au 3 Cuhave been possible through colloidal syntheses at temperatures between 200 and 250°C, mainly developed by Schaak and coworkers.
## Core-shell structures
## Seed-mediated synthesis
The core-shell structure has gained much attention in electrocatalysis due to its ability to precisely tune the surface properties of active metal species by the core and the reduced amount of. a) TEM images of B2-PdCu/C and A1-PdCu nanoparticles (inset). Reproduced with permission.Copyright 2020, Wiley-VCH. b) XRD pattern of B2-PdCu nanoparticles prepared from A1-PdCu seeds and c) their TEM image with proposed atomic arrangement (inset). Reproduced with permission.Copyright 2016, American Chemical Society. d) XRD pattern of L1 0 -PdZn phase obtained by impregnation method. Reproduced with permission.Copyright 2019, American Chemical Society. e) TEM image of L1 0 -PdZn nanosheets with proposed atomic arrangement (inset). Reproduced with permission.Copyright 2019, American Chemical Society. f) HAADF-STEM image of supported L1 0 -PdZn nanoparticles on ZIF-C. Reproduced with permission.Copyright 2018, Wiley-VCH. g) TEM images and h) XRD patterns of AuCu nanoparticles with tuned degree of ordering. Reproduced with permission.Copyright 2017, American Chemical Society.
precious noble metals used, which contributes to higher mass activity. Until recently, most of the works that produced novel coreshell nanoparticles with elaborate control over morphologies relied on a seed-mediated approach.Heterogeneous nucleation of secondary metal species on seeds is important to obtain only core-shell products, which is typically achieved by low concentra-tions of secondary metal precursors during shell growth. The epitaxial growth of secondary metal shells usually follows the atomic arrangements on specific facets of core nanocrystals, thereby retaining the initial morphologies of the cores.The two well-known combinations of metals are Pt-Pd and Au-Ag, as they both have lattice mismatches of less than 1%.Copyright 2007, Nature Publishing Group. c) HAADF-STEM image of a Pd@Pt nanoparticle and d) a schematic illustration of control of shell thickness. Reproduced with permission.Copyright 2014, American Chemical Society. e) TEM images and the proposed structure of Pd-(Ni-Pt) 2 nanoparticles and f) their magnified TEM image. Reproduced with permission.Copyright 2020, Wiley-VCH. g) Schematic illustration of epitaxial Cu shell formation on 4H (or 4H/fcc) Au nanorods. Reproduced with permission.Copyright 2020, American Chemical Society. h) Schematic diagram (up) and HAADF-STEM images (down) of pristine (left), acid-etched (center, Pt-skeleton), and thermally annealed (right, Pt-skin) PtNi catalysts. Reproduced with permission.Copyright 2011, American Chemical Society. i) Reversible change in surface composition of RhPd nanoparticles under different gas environments. Reproduced with permission.Copyright 2008, AAAS.
The greater the lattice mismatch between two metal elements, the lower the likelihood of shell formation and the poorer control over shell thickness, which has been identified in efforts to deposit 3d transition metals on Pd or Au seeds.Owing to the low to moderate lattice mismatches with Pd, the shells of Pt, Ir, Rh, and Ru have been prepared on various structures of Pd nanoparticle seeds with a controlled num-ber of shell atomic layers. Layer-by-layer conformal deposition of Pt atomic layers on Pd nanocubes was reported in 2014 by the slow injection rate of Pt at 200°C, which allowed enough time and energy for Pt atoms to spread onto the entire surface of Pd.The same synthetic concept has been adopted with little modification to deposit several atomic layers of Pt shells on Pd nanocrystals with twinned decahedral,www.advancedsciencenews.com www.advancedscience.com twinned icosahedral,and octahedralshapes. Reaction kinetics have also been controlled recently by the amount of reductant (glucose) for the deposition of a uniform shell or islands of Pt on Pd nanocubes.Moreover, Qiu and co-workers recently provided tetrahedral core-shell nanoparticles by deposition of Pt, Ru, or Rh on tetrahedral Pd seeds, obtaining less than five atomic layers of shells.For deposition of Ir shells on Pd nanoparticles, Xia and co-workers reported layer-by-layer deposition of Ir shells rather than island formation through slow injection of Ir solution into a mixture of reagents containing ethylene glycol, PVP, ascorbic acid, and Pd seeds (cubes or octahedra).Recently, the method was used to precisely control the number of atomic layers of the Ir shell and correlated it with the electrocatalytic activity and durability.Although Ag has a large lattice mismatch of ≈4.5% with Pd, Xia and co-workers successfully prepared Ag shells on Pd nanocubes by a fast reduction kinetics, which produced Ag coating within 2 min.The authors attributed the anomalous shell formation to the small size of Pd seeds (18 nm), which enabled nucleation and epitaxial growth of Ag shells on the {100} facets. On the other hand, deposition of Au on Pd seeds generally forms PdAu shells of a few atomic layers by galvanic replacement.Another widely used core is Au, which usually promotes durability against electrochemical leaching, thereby providing unique stability to the shell.The deposition of Pdor Ptshells on Au seeds was carried out under mild conditions in which the addition of Pd or Pt precursor solution and reductant resulted in uniform shells on Au. Furthermore, Sun and co-workers have reported the synthesis of alloy shells of various compositions, such as PtCuand PdCu,simply by placing Au seeds before coreduction of two metal precursors with controlled atomic ratios. This approach has enabled the production of various alloy shells on Au or Pd cores such as Au@PtFe,Au@PtNi,Pd@PtFe,Pd@Au@PtFe,and Pd@PdFe core@shell nanoparticles.Recently, Tsung and co-workers reported an effective strategy to prepare intermetallic alloy shells on Pd cores through thermal intermixing of Pt and Ni from epitaxially grown Pt and Ni shells.Compared to thicker Ni-Pt shells on Pd, the 4-layered Pd-(Ni-Pt) 2 (Pd-(Ni-Pt-Ni-Pt)) core-shell structureinduced strain and provided a shorter diffusion length between Pt and Ni, so that only 400°C was sufficient for full mixing of Pt and Ni to produce intermetallic PtNi 3 shells on Pd nanocubes, while 2-layered Pd-(Ni-Pt) with thicker shells of Pt and Ni showed incomplete intermixing after annealing. Finally, epitaxial shell growth on unconventional phases has recently been explored by Zhang and co-workers, starting from 4H/fcc Au nanorods, 2H Pd nanoparticles, and 4H(or 4H/fcc) Au nanorods to prepare 4H/fcc Au@Pd,fcc-2H-fcc Pd@M (M = Au, Ag, or Pt),and 4H(or 4H/fcc) Au@Cu nanoparticlesunder mild conditions, which are already well known for shell growth.
## One-step synthesis
As discussed in Section 2.1.5, a huge gap between the reduction rates of the constituent metals leads to phase-segregated nanoparticles. As Au@Pd nanoparticles have a large gap in reduction potential, they have been frequently demonstrated as representative examples for the one-step synthesis of core-shell structures. In 2009, Han and co-workers reported the one-step synthesis of Au@Pd via the coreduction of HAuCl 4 and K 2 PdCl 4 in the presence of CTAC.Temporal investigation of the metal ratio in a nanoparticle indicated that the formation of Au octahedra takes place first, and a Pd layer forms on them due to the huge difference in the reduction rates of the metal precursors. In a control experiment, the use of ascorbic acid, which is a stronger reductant, resulted in AuPd alloy nanoparticles, owing to the rapid reduction of the precursors. The synthesis of other core-shell nanoparticles with various combinations of metals, such as Rh@Feand Cu@IrPd@PdPt,could also be achieved via this one-pot approach through appropriate selection of chemicals and reaction conditions. A recent study by Shao and co-workers attempted to control the shell thickness by varying the ratio of metal precursors, although one-pot synthesis provides less control over the thickness and composition of shells than the seed-mediated approach.The one-pot synthesis of a more complex core-shell structure was demonstrated by the use of IrCl 3 and Ir(acac) 3 , which are known to have quite different reduction potentials. Zhuang and co-workers took advantage of the dual precursor approach to produce IrNi@Ir alloy core-shell nanoparticles.The synthesis proceeded by the simultaneous reduction of Ir(acac)and Ni(acac) 2 to form an alloy core, followed by successive reduction of IrCl 3 on the surface of the IrNi alloy, forming an Ir shell. Likewise, Lee and co-workers synthesized alloy core@alloy shell nanoparticles by reducing dual Ir precursors (Ir(acac) 3 and IrCl 3 ) and dual transition (Ni and Cu) metal precursors to obtain double-layered nanoframes.When only one Ir precursor, Ir(acac) 3 or IrCl 3 , was used, only single-layered nanoframes were produced.
## Transformation from alloy
Apart from the afore-mentioned approaches, the core-shell structure can be derived from alloy structures. Owing to the lower reduction tendency of non-noble metals than noble metals, they are relatively vulnerable to chemical/electrochemical leaching under corrosive conditions. This difference in intrinsic stability provides another opportunity for the preparation of core-shell nanoparticles by dealloying the surface of alloy nanoparticles via chemical and electrochemical etching.
In recent years, the dealloying route has been intensively investigated for the preparation of noble metal-based core-shell nanomaterials. In particular, nanostructures with Pt-rich shells have attracted enormous interest owing to their greatly enhanced activity and stability toward oxygen reduction catalysis.Chemical, electrochemical, or even combined etching has been conducted to dealloy the surface of Pt-based random alloys, such as PtCu,PtCo,PtNi,PtAg,and PtCuCo,to produce Pt-rich shells on Pt-alloy cores. Recently, this approach has been further enhanced by the use of additional thermal treatment after selective dealloying. The resultant structural transformation from the Pt skeleton to the Pt-skin surfacealtered the sorption properties by reducing the lowcoordinated surface sites, resulting in enhanced activity and durability. The post-treatment of intermetallic nanoparticles re-www.advancedsciencenews.com www.advancedscience.com cently produced L1 0 -FePt@Ptand L1 0 -CoPt@Ptfrom the corresponding intermetallic alloys.
In addition to Pt-based materials, the synthesis of core-shell nanoparticles based on other noble metals, such as Ir and Au, has also been investigated for specific electrochemical reactions. IrNi@IrO x core-shell nanoparticles were prepared using two different dealloying procedures.The formation of IrO x on the surface was promoted either by sequential dealloying and oxidation or by coupled dealloying/oxidation. AuNi/Au core-shell nanoparticles were also produced by electrochemical dealloying of AuNi nanoparticles.Another approach to obtain core-shell structures from alloys is thermodynamically driven phase segregation by reducing the surface energy. Phase segregation from homogeneously alloyed nanoparticles can be induced in the presence of proper surface adsorbates because of their different binding affinities for each metal atom. Somorjai and co-workers conducted a model study using Rh 0.5 Pd 0.5 and Pt 0.5 Pd 0.5 nanoparticles to investigate the effect of binding affinity between gases and metals on the elemental distribution in a nanoparticle.As the composition of gases changed from oxidizing to reducing conditions, the richness of metal on the surface was reversed in the case of Rh 0.5 Pd 0.5 , implying that the gaseous environment is important for the elemental distribution of metal atoms in a particle. On the other hand, the annealing temperature also affected the surface property of products with a constant environment.Similar to the above studies, the PdCo@Pd core-shell structure could be induced by thermal annealing of PdCo alloy nanoparticles under H 2 flow.The absorbate-induced transformation can also be achieved by an electrochemical process.PtCo@Pt core-shell nanoparticles were fabricated via electrochemical annealing of PtCo alloy nanoparticles in a CO-saturated electrolyte for an hour. Owing to the stronger binding of CO on Pt than Co, the surface of the nanoparticles was enriched by Pt, forming PtCo@Pt coreshell nanoparticles.
## Other heterostructures
Unlike alloys that contain more than two different metal elements in a single phase, the heterostructure features separate domains of metal elements. Most attempts to successfully obtain well-defined and fine-tuned heterostructured nanoparticles have relied on seed-mediated syntheses. The large lattice mismatches between Au and 3d transition metals can easily deliver Au-M dimers by the deposition of Cu or Fe on single-crystalline Au seeds.However, because of the similar atomic radii of Ag and Au, anisotropic growth of Ag on Au core has been achieved by the use of twinned Au nanoparticles (e.g., decahedron),which also proved effective for obtaining Au-Cu dimer nanoparticles through Cu deposition on twinned icosahedral Au seeds rather than single-crystalline Au seeds.On the other hand, Wang and co-workers employed an irregular distribution of surfactant molecules on shaped nanoparticles to selectively deposit Pt at the tips of Au nanotriangles, where fewer CTAB molecules were covered.Recently, for the deposition of noble metals on Au seeds, polymeric materials were deposited on Au to cover certain portions of the surface, followed by selective loading of other noble metals (Pd, Pt, or Ag) to form coaxialor dimericheterostructures. Furthermore, Chen and co-workers precisely controlled how polymeric shells cover the surface of Au nanoparticles by forming polystyrene-blockpoly(acrylic acid) shells on Au in the presence of different types of ligands and their heat-induced transformation.Pd and Ag were sequentially deposited on Au, covering only the exposed surface of the Au seeds after the polymeric shells were thermally contracted or dissociated.
Similarly, various heterostructured nanoparticles have been prepared from Pt or Pd seeds. The anisotropic growth of Au on Pt is well established because of their large lattice mismatch, while the deposition of Pd results in an epitaxial shell on the Pt core.Shevchenko and co-workers reported a general nucleation and growth model of Au on Pt or Pt-alloy seeds (Pt 3 Co or PtFe) by controlling the type of Au precursor (AuCl and AuCl 3 ) and the use of foreign ions (Co 2+ or Pb 2+ ). They observed that Au + ions are essential for the initial nucleation of Au on Pt, and the foreign ions promote the reduction of Au 3+ to Au + for nucleation.Further, they systematically uncovered three distinct periods for the formation of seed/Au dumbbell structures, which comprise prenucleation, nucleation, and growth.Surprisingly in the prenucleation period, Au formed an epitaxial shell on the seed, exerting lattice expansion on the seed by high stress of ≈2.4 GPa. A strain-free domain of Au developed after the prenucleation period to relieve the lattice strain on the seed, leading to heterostructured dumbbell nanoparticles. Notwithstanding the similar lattice parameters of Pt and Pd, avoiding shell formation by localized epitaxial growth of Pd on Pt seeds was demonstrated by increasing the reduction rate under high pH conditions (pH ≈ 9).Likewise, dendritic Rh nanostructures, rich in (100) facets, on Pt nanocubes were prepared by thermal decomposition of Rh(acac) 3 at 200 and 170°C, even in epitaxial growth.The formation of Pt branches on Pd nanocrystals was accomplished by the reduction of K 2 PtCl 4 using ascorbic acid in the presence of Pd seeds.A more direct route to obtain an analogous structure was subsequently reported, which reduced both K 2 PtCl 4 and Na 2 PdCl 4 at 30°C using Pluronic P123 as a structural template.Xia and co-workers elaborated the seeded growth of Pt on Pd by introducing Br − ions to transform PtCl 6 2− (or PtCl2− ) to PtBr 6 2− (or PtBr 4 2− ), which promoted selective galvanic replacement on the vertices of Pd nanocubesand icosahedra by slow reduction kinetics.Manipulation of reaction kinetics also played a key role in controlling the nucleation and growth of Ag or Au on Pd nanocubes, enabling growth at only one facet at a slow injection rate of Ag (or Au) and complete coverage in one-shot injection.Lastly, heat-induced formation of heterostructured nanoparticles on a substrate has been actively studied by Mirkin and co-workers using a technique called scanning-probe block copolymer lithography (SPBCL).They used atomic force microscopy probes to transfer controlled amounts of polymer inks composed of poly(ethylene oxide)-block-poly(2-vinylpyridine) and combinations of metal precursors. Upon two-step thermal annealing under Ar and H 2 flow, metal precursors were first aggregated and then reduced to a single multimetallic nanoparticle for each polymer dome on a substrate. When the metal compositions of polymer inks were chosen to combine immiscible metals, heterostructured nanoparticles formed after annealing such as AuCo, AuNi, or AgCuCo.They fur-. TEM images and schematic illustrations of a,b) Au nanotriangles and c,d) Au-Pt heterostructure. Reproduced with permission.Copyright ther extended to study how Pd-Sn alloy system interacts with other multiple additional metals (Au, Ag, Co, Ni, or Cu) to determine design rules to modulate the number and types of phase boundaries in polyelemental nanoparticles with different tunable compositions.A recent study discussed a li-brary of heterostructured nanoparticles with HIFs by the presence of Bi powders upstream of the tube furnace during the annealing of spin-coated metal salts.They reported uniform heterostructured nanoparticles of HIFs using the SPBCL technique.
## Considerations for electrocatalytic applications of nmmns
As discussed in earlier sections, the designed synthesis of nanoparticles has achieved great success in controlling the size, shape, composition, and multifunctionalities of nanoparticles. However, several points remain to be considered for effective electrochemical energy conversion applications. To begin with, although surfactants (ligands) play pivotal roles in precise control over the structure and prevention of agglomeration during the synthesis of NMMNs, they hinder surface reactions. Covering the surface with surfactants hinders the effective utilization of principal active sites on the surface of nanoparticles, leading to poor catalytic performance. Therefore, it is necessary to establish efficient approaches to remove or tune the state of the surfactants.It is necessary to carefully consider whether the original structural features of nanoparticles, such as facets, composition, strain, and intended functionalities, can be maintained after the removal processes. Second, we should consider how to deposit and stabilize NMMNs on support materials for the efficient utilization of prepared nanoparticles in electrocatalytic applications. In order to overcome the intrinsic instability of nanoparticles and to maximize their surface utilization, their dispersion on solid supports has been extensively investigated in the field of heterogeneous catalysis. There are several points to be considered for optimal realization of nanoparticle/support system: high electrical conductivity for efficient electron transport, high chemical and electrochemical stability for use at various pH and electrochemical potentials, strong interactions between the support and catalyst, and efficient mass transfer. Herein, we summarize typical strategies to address surfactant issues (ultravioletozone (UVO) treatment, thermal, chemical, and electrochemical cleaning processes) as shown inand utilization of support materials (use of stable supports, nanoparticle-support interaction, and encapsulation) and discuss future-scope.
## Removal of surface capping agents
## Uvo treatment
UVO treatment is an effective strategy to remove various contaminants from the surface of nanoparticles. This strategy utilizes a combination of UV light and ozone, which can oxidize www.advancedsciencenews.com www.advancedscience.com carbon-containing organics into gaseous carbon dioxide. Sum frequency generation spectroscopy was used to confirm that tetradecyltrimethylammonium bromide surfactant molecules on platinum nanoparticles were removed after 15 min of UVO treatment.After the treatment, the shape and size of the nanoparticles were preserved, confirming the effectiveness of this treatment. This approach has been successfully extended to various nanoparticles, including Pt,Au,Pd,and PtFenanoparticles coated with various organic ligands such as oleylamine,dodecanethiol,and dioctadecyldimethylammonium ligands.
## Thermal annealing
Thermal annealing is commonly applied to remove ligands by decomposition or desorption at appropriate temperatures. The annealing conditions should be carefully chosen to prevent agglomeration or sintering of nanoparticles and preserve the original size and morphology, which are detrimental to the catalytic activity.Stamenkovic and co-workers compared the effectiveness of thermal annealing with an acid treatment and UVO treatment on surfactant removal.Among them, the catalyst annealed at 185°C showed a more exposed electrochemical surface than the others, which was analyzed by hydrogen underpotential deposition (H upd ). Thermal gravimetric analysis confirmed that most of the surfactant was removed (with ≈45% weight loss) after 5 h of thermal annealing at 185°C, leaving only 1% loss when the temperature was further increased to 900°C
. The electrochemical surface area (H upd ) showed the following trend: thermally annealed > acid-treated > UV-ozone treated > untreated catalysts. Not surprisingly, the ORR activity was highly affected by the surfactant removal protocols, highlighting the importance of proper removal of surfactants for electrochemical performance.
## Chemical treatment
Chemical treatment is used to remove the organic surfactants. Acetic acid is a widely used chemical for the removal of various organic surfactants at 60-90°C.Unfortunately, the method is insufficient to remove the ligands with strong binding, such as phosphines. Recently, ligand exchange of surfactants with an N-heterocyclic carbene (NHC) ligand (with strong -donors and -acceptors) was investigated as an efficient way to remove surfactants after mild acid treatment.As NHC is a stronger electron-donor than phosphines, they readily replace phosphine ligands. The vibrational bands of NHC-bound Pt, Pd, and Au confirmed the presence of oxidized NHC, indicating that the phosphine ligands were fully exchanged with NHC. The NHC ligands were subsequently removed by acetic acid treatment without heating. After treatment with acetic acid and washing, the diffuse reflectance infrared Fourier transform spectra of the samples showed no or tiny signs of organics on the surface; however, the initial sizes and morphologies of the noble metal nanoparticles (Pt, Pd, and Au) were well preserved.
# Electrochemical method
Electrochemical strategies have also been applied to remove various ligands from the surfaces of nanoparticles.Recently, Fan and co-workers reported a facile and universal electrochemical method to eliminate organic surfactants.] They applied the following steps: 1) removal of surfactants by forming strong O or H bonds on the metal surface (M) by electrochemical reactions, and 2) recovering the initial metallic state by reduction or oxidation to remove O and H. In detail, Pt tends to be oxidized to PtO to form Pt-O covalent bond over 1.0 V (vs reversible hydrogen electrode (RHE)). Once Pt is oxidized to PtO, oleylamine can no longer coordinate with Pt because of the robust covalent bonding between Pt and O. The PtO is reduced to Pt, as the potential decreases below 0.7 V (vs RHE). Thus, the desorbed oleylamine cannot reabsorb onto the Pt surface because of its low concentration in the electrolyte due to its low solubility in aqueous solution. In turn, the hydrogen evolution reaction removes the oleylamine ligands on the nanoparticles by strong M-H bonds in the same manner as the M-O bond. In the hydrogen evolution region, Moleylamine bonds are substituted by the M-H bond because the M-H bond is stronger than M-oleylamine. As the hydrogen desorption potential goes beyond 0.3 V (vs RHE), a clean M surface is exposed, which can then be utilized for desired catalytic reactions.
## Outlook
Various methods have been successfully used to remove surfactants. Furthermore, various analytical techniques have been utilized to prove the removal of surfactants; however, there is still a lack of careful consideration of how the physicochemical properties of nanoparticles change during the removal processes. Owing to the high surface energy of the nanoparticle surface, they may easily be reconstructed during the treatments and lose their designed functions. Therefore, monitoring the structural evolution during processes with reliable surface-sensitive protocols and real-time (in situ) monitoring is highly recommended. Finally, recent reports suggest that surfactants or organic molecules are no longer merely undesirable but can be utilized as positive promoters for certain reaction conditions to guide reaction selectivity by steric hindranceor electronic effects.Therefore, it is important to redefine the roles of surfactants on the nanoparticle surface as promotional or detrimental effects and how to control their functionalities according to the target catalytic reactions.
## Effect of support materials
## Nanoparticle-support interaction by the electronic effect
There are several requirements that support materials should satisfy to achieve ideal catalytic performance. First, the high electrical conductivity of the supports provides a complete electrical network for high-rate electrochemical reactions. Second, supports should be (electro)chemically stable against corrosive environments in electrolytes as nanoparticles rely largely on sup-. Carbon corrosion issues for conventional nanoparticle/support systems and recent approaches to overcome the problems. a) Carbon corrosion followed by nanoparticle detachment. b) (Electro)chemically inert metal oxide supports. Inset: TEM image of Pt dispersed on Nb-doped TiO 2 . Reproduced with permissionCopyright 2016, Elsevier. c) Nanoparticle-support interaction to improve activity and stability. Inset: TEM image of Pt dispersed on porous TiN. Reproduced with permission.Copyright 2016, American Chemical Society.
ports. Third, strong interactions between nanoparticles and supports are highly recommended for the stabilization of nanoparticles. Finally, support materials with a high surface area allow nanoparticles to be adsorbed on them with proper interparticle distances. Considering the above criteria, high-surface-area carbon materials, including microporous carbons, carbon nanotubes, and graphene, are commonly utilized as supports for the application of nanoparticles in various electrochemical reactions. Although the oxidation of carbon can occur over 0.207 V (vs RHE) to produce CO 2 , as follows
[formula] C + 2H 2 O → CO 2 + 4H + + 4e −(1) [/formula]
It does not significantly affect the electrochemical reactions that occur below 1.5 V (vs RHE) because the kinetics of carbon corrosion are not very fast.Therefore, carbon-based materials are generally considered as ideal support materials with high electrical conductivity and (electro)chemical stability.
However, under realistic operating conditions in devices such as start-up/shutdown events in fuel cells and high-potentialbased applications such as water oxidation (oxygen evolution reaction), carbon-based supports suffer degradation by particle detachment and alteration of pore structure, resulting in performance losses,making them no longer an ideal platform for nanoparticle utilization.Furthermore, the interaction between nanoparticles and carbon supports is not strong enough to stabilize nanoparticles, despite the limited discussion on this issue.Metal oxides are often suggested as alternatives to overcome the carbon-based support limitationsas metal oxides are generally stable in electrochemical operation windows. However, the number of possible oxide candidates is limited owing to the issue of electrical conductivity.One typical example is titanium-based oxide materials. While the electrical conductivity of bare TiO 2 is not high enough to fully support electrochemical reactions, various dopants including Ru, Mo, Ta, V, Cr, and Nb could significantly increase the conductivity that made the doped-TiO 2 highly applicable to electrochemical applications.For example, nanostructured Ti 0.7 Mo 0.3 O 2 was utilized as a superior support material for ORR.Compared to Pt/C, Pt deposited on Ti 0.7 Mo 0.3 O 2 was more stable due not only to the stability of the support materials itself but also to the strong interaction between nanoparticles and support by electronic effect, the so-called strong metal-support interaction. Similar to TiO 2 , SnO 2 -based materials have also been highlighted as potential alternatives.By doping with In, Sb, or F atoms into SnO 2 , the obtained indium-doped tin oxide, antimony-doped tin oxide (ATO), and fluorine-doped tin oxide were utilized under various electrochemical conditions with high chemical stability and electrical conductivity. However, recent in situ studies on electrochemical stability revealed that tin oxide-based materials are also vulnerable to dissolution in a broad electrochemical potential window,raising questions about whether oxides can be "actual" alternatives as stable support materials for electrochemical applications.
Not surprisingly, metal oxide-based support materials are highly accentuated in oxygen evolution reaction. As we discussed already, carbon corrosion is an extremely severe problem in the www.advancedsciencenews.com www.advancedscience.com high potential region (>1.5 V vs RHE). Especially, the surface oxidation of carbon followed by passivation can abruptly prohibit catalyst utilization.Therefore, carbon-based supports are no longer considered in OER applications. The Strasser group successfully introduced and validated the importance of durable support materials using SnO 2 , doped with various elements (Sb, In, and F).Among the candidates, mesoporous ATO showed superior physicochemical properties such as high surface area, high electrical conductivity, and superior electrochemical stability. During the potential cycling (10 000 cycles with 500 mV s −1 ), the carbon support (Vulcan) lost ≈ 50% of its surface capacitance while ATO showed negligible capacitance loss.Furthermore, the superiority of the ATO support was further demonstrated when it was used to support Ir nanoparticles. Not only was the support itself stable, but also were the Ir nanoparticles effectively stabilized without a size growth due to the strong electronic effect of the support on the active metal catalyst. This system has been further improved by various metal-support combinations with mechanistic studies,morphology controls,surface treatments,and selection of various oxides.In addition to the stabilization effect on active metal nanoparticles, there have been many reports on the promotive effects of metal oxide supports in the increased reaction kinetics, such as in methanol oxidation reaction (MOR).Various multimetallic catalysts showed enhanced performance in combination with different oxide supports, which has been explained by two main reasons.First, OH groups on the oxide surface facilitate the removal of CO ad on the metal sites via bifunctional mechanism and increase MOR activity, because CO ad (reaction intermediate of MOR) is strongly adsorbed and blocks (poison) the active surface of metal nanoparticles. Second, the electronic effect between metal nanoparticles and metal oxide supports improves charge transfer between them. In ethanol oxidation reaction (EOR), the dissociation of C-C bond has been a major challenge to oxidize ethanol completely to CO 2 , prohibiting further development of alcohol-based direct fuel cells. To solve this issue, the Adzic group designed a SnO 2 -supported multimetallic PtRh system as a synergistic EOR catalyst.SnO 2 strongly adsorbs water and provides OH species to oxidize CO on Rh sites while Pt facilitates the ethanol dehydration. Furthermore, SnO 2 also modifies the electronic state of Rh nanoparticles to afford optimum binding of intermediate species, enhancing the overall reaction kinetics.In order to complement the limited range of multimetallic nanoparticle design, many combinations of metal-support structures have been studied, tuning the compositionsand metal-support interfaces.In contrast to oxide-based systems, nitrides and carbides have expanded their applicability owing to their high electrical conductivity, among which TiN has received much attention due to the higher conductivity (40 kS cm −1 in bulk phase).TiN supports with various morphologies, including particulate, tubular, hollow, and 3D structures, were found to stabilize noble metalbased catalysts better than carbons.Furthermore, the introduction of Nb, Cr, or Ni into TiN has proved to provide more active and stable electrochemical performance.Despite the promising aspects of their applicability, the issue of surface oxidation continues to be pointed out as the primary limitation of nitride-based supports. In addition, most studies based on nitrides are limited to half-cell level tests, so whether they can be applied to actual devices, such as membrane electrode assembly, is an important remaining challenge. On the other hand, light element-based carbides and nitrides, such as hexagonal boron nitride and boron carbide (BC), have recently been studied to have strong electronic interactions with catalysts having long-term durability.In Pt/BC systems, strong dipole interactions between metal nanoparticles and supports reduced the mobility of Pt nanoparticles, thus mitigating agglomeration. Furthermore, carbide-supported nanoparticles are considered to be slightly less prone to form oxide surfaces, which is directly linked to the mitigation of the anodic and cathodic Pt dissolution during electrochemical potential cycles.
## Nanoparticle-support interaction by the geometric effect
Simple physical attachment is a conventional approach to stabilize nanoparticles on supports. As one of the major degradation mechanisms of nanoparticles is particle-particle agglomeration caused by weak interaction between them with supports, various efforts have been devoted to enhancing the interaction by heteroatom doping, strong dipole interaction, and functional groups.Recently, two different approaches have been widely adopted to overcome the instability of catalysts on supports: 1) geometric confinement of nanoparticles via encapsulation with stable overlayersand 2) geometric confinement of nanoparticles in nanoscale pore structures to mitigate their migration. First, encapsulation of nanoparticles is usually performed by confinement in carbon nanotubes or the formation of carbon overlayers.Particularly, the use of a very thin carbon layer is attracting attention in terms of being able to stabilize nanoparticles while maintaining mass transfer effectively.When the protection layer is too thick, it is difficult for the reactants to penetrate, while nanoparticles cannot be protected if the layer is too thin. Therefore, it is important to control and balance the thickness of the protection layer to simultaneously determine the optimal conditions for achieving high activity and stability at the same time. This strategy has been successfully extended to various types of nanoparticles (metal oxides, nitrides, and phosphides)and different carbon layer formation strategies have been proposed.Second, geometric confinement in porous structures can also stabilize nanoparticles against migration on supports and detachment from the supports.Using porous structures, nanoparticles are geometrically confined in pore space and maintained their size during electrochemical measurements without agglomeration and particle detachment, minimizing activity losses.Recently, encapsulation of nanoparticles in porous structures was further extended to the preconfinement of nanoparticles in porous matrices such as metal organic frameworks.After encapsulation of nanoparticles in a porous frame, the thermal annealing process can confine nanoparticles in porous carbon supports. Furthermore, the surface of nanoparticles can be wrapped with very thin-layer carbon upon annealing, achieving a dual stabilization step with thin carbon-layer encapsulation and geometric confinement in the porous structure. In addition, attempts to simultaneously achieve nanoparticle synthesis and stabilization effects by introducing metal precursors to porous carbons or frameworks have been recently reported. Moreover, highly advanced nanopar-. Nanoparticle degradation mechanism originated from migration of nanoparticles on support and recent approaches to overcome the problem. a) Nanoparticle agglomeration via weak interaction with support. b) Geometrical encapsulation method to stabilize nanoparticles. Inset: TEM image of carbon-shell-encapsulated PtFe nanoparticles. Reproduced with permission.Copyright 2015, American Chemical Society. c) Geometric confinement in porous structures. Inset: TEM image of geometrically pore-confined Pt 3 Ni-Mo octahedra. Reproduced with permission.Copyright 2018, American Chemical Society.
ticle design strategies for their stable utilization have been continuously proposed to overcome the limitations of conventional nanoparticle/support systems.
## Summary and perspectives
NMMNs have been an indispensable part of active research in electrocatalysis because of their intrinsic electronic properties that give close-to-optimal binding energies of reacting species, such as reaction intermediates. However, as the demand for higher electrocatalytic performance has increased, researchers have developed different approaches and strategies to produce desired or novel multimetallic nanostructures based on noble metals. Considering the myriad of studies done so far on the synthesis of NMMNs, although we still have far more to go, we attempted to provide a comprehensive but concise review of the controlled synthesis of NMMNs after classifying their structures into random alloys, single-atom alloys, high-entropy alloys, ordered-intermetallics, core-shell structures, and other heterostructures.
The current status of each structure since last two decades can be summarized as follows. 1) For random alloys, control of size and composition relies primarily on reaction kinetics that affects the nucleation and growth of nanoparticles, while recent state-of-the-art catalysts take advantage of novel shapes that can be tuned by facet control (including HIFs), etching, galvanic replacement, or the Kirkendall effect. 2) As subgroups of alloys, single-atom alloys, and high-entropy alloys have unusual struc-tural features, while controlled synthetic methods are still underdeveloped. The key points to obtain SAA and HEA are the effective deposition of only one separate atom and fast simultaneous nucleation of multiple metal elements, respectively. 3) Although various intermetallic phases of Pt, Pd, and Au have been constructed as nanoparticles, precise control over the size and composition is still limited to a few metal combinations, calling for modified or novel synthetic strategies. 4) Core-shell nanoparticles have been effectively prepared by seed-mediated routes in a controlled manner, while one-pot synthesis and transformation of alloys into core-shell nanoparticles have also been adopted as alternatives. 5) Other heterostructured NMMNs have also been prepared by seed-mediated approaches, revealing the important role of surfactants in the selective anisotropic deposition of secondary metal species by their nonuniform binding on different surface sites of seeds. Additionally, the lattice mismatch and immiscibility between metal elements also contribute to the formation of heterostructures.
In addition to the discussed issues regarding the designed synthesis of various kinds of NMMNs, their subsequent translation into the active form of electrocatalysts should be viewed as equally important. Although the presence of surfactants plays a vital role in controlling the size and shape of NMMNs, they should generally be removed to expose the active metal surfaces to achieve desirable performance fully. Representative surface treatments were introduced in this review, among which physical treatments using ultraviolet ozone or thermal annealing have been widely used. One the other hand, careful selection and modification of the support materials are required to enhance the long-term stability of NMMNs. Protection of active nanoparticles is achieved by surface encapsulation (usually by thin carbon) or geometric confinement of each particle in the pores of the supports.
Furthermore, future research should focus on overcoming the following challenges in the preparation of active, selective and durable electrocatalysts. First, controlled synthesis of SAAs and HEAs with carefully tuned active sites can provide well-defined catalytic active sites (by SAAs) or a wide range of surface atomic ensembles (by HEAs) for an unprecedented activity and selectivity in a target reaction. Development of more generalized and reliable synthetic strategies would stimulate in-depth investigations in either empirical or theoretical ways. Secondly, synthesis of uniformly sized, more atomically ordered and shape-controlled intermetallic nanoparticles may resolve the current stability issues of NMMN-based electrocatalysts that impede their practical applications. Efforts to discover and analyze intermetallic nanomaterials of unexplored compositions would expand our options to find the most suitable practical catalyst. Third, monitoring the structural evolution of NMMNs upon surfactant removal would enable a precise activation of as-synthesized surface active sites, preventing their unwanted deformation into less active forms. In situ microscopic and spectroscopic techniques are highly promising for the detailed observation of structural changes during surfactant removal steps. Fourthly, systematic comparison of the effect of support structures on the activity and stability of NMMNs is still limited, which becomes more important when NMMNs are applied for practical devices. Finally, cost-effective, large-scale production of supported NMMN-based catalysts is indispensable for their industrial-level uses but scaling up the synthesis usually compromises their unique structural properties and performances. Thus, novel or modified synthetic methods are to be developed to obtain active and durable supported NMMNs in a gram-scale or a kilogram-scale. |
Hybridoma technology: is it still useful?
A B S T R A C TThe isolation of single monoclonal antibodies (mAbs) against a given antigen was only possible with the introduction of the hybridoma technology, which is based on the fusion of specific B lymphocytes with myeloma cells. Since then, several mAbs were described for therapeutic, diagnostic, and research purposes. Despite being an old technique with low complexity, hybridoma-based strategies have limitations that include the low efficiency on B lymphocyte-myeloma cell fusion step, and the need to use experimental animals. In face of that, several methods have been developed to improve mAb generation, ranging from changes in hybridoma technique to the advent of completely new technologies, such as the antibody phage display and the single B cell antibody ones. In this review, we discuss the hybridoma technology along with emerging mAb isolation approaches, taking into account their advantages and limitations. Finally, we explore the usefulness of the hybridoma technology nowadays.
# Introduction
Monoclonal antibodies (mAbs) are universal highly specific binding proteins that were envisioned for a long time as "magic bullets" in the fight against diseases, and also important tools for other biological uses, including diagnosis and research [bib_ref] Hybridoma technology a versatile method for isolation of monoclonal antibodies, its applicability..., Parray [/bib_ref]. These applications were only possible with the advent of methodologies that allow the isolation of individual antibodies. The hybridoma technology was the pioneer on that. Indeed, this technique revolutionized the therapeutic and research scenario, which was further recognized by the 1984 Nobel Prize in physiology or medicine [bib_ref] The birth of monoclonal antibodies, Leavy [/bib_ref]. Other strategies have been developed for the same purpose. In this review, we explore the relevance of the hybridoma technology nowadays, how it has evolved with time, and its advantages and limitations compared with other methods that further come out.
## Hybridoma technology
The hybridoma technology, described by Georges K€ ohler and Cesar Milstein in 1975 [fig_ref] Figure 1: Timeline of important events in the generation of monoclonal antibodies [/fig_ref] , is based on the immunization of animals with the desired antigen, followed by the fusion of specific B lymphocytes with "immortal" myeloma cells. The generated hybrid cells, called hybridomas, are then cloned to obtain stable monoclonal cell lines [bib_ref] Continuous cultures of fused cells secreting antibody of predefined specificity, K€ Ohler [/bib_ref]. After selecting the antibody-secreting clones of interest, the cells are transferred to large-scale culture setups to produce the antibody in the desired amounts (Holzl€ ohner and [bib_ref] Generation of murine monoclonal antibodies by hybridoma technology, Holzl€ Ohner [/bib_ref]. B lymphocyte-myeloma cell fusion is often obtained by using the chemical compound polyethylene glycol (PEG). However, this agent can be cytotoxic at some level, and non-specific membrane fusion may occur [bib_ref] Hybridoma technologies for antibody production, Tomita [/bib_ref] [bib_ref] Use of human hybridoma technology to isolate human monoclonal antibodies, Smith [/bib_ref]. Fusogenic viruses, such as the Sendai and the vesicular stomatitis viruses, are alternatives that bypass the cytotoxic effects of PEG [bib_ref] Use of human hybridoma technology to isolate human monoclonal antibodies, Smith [/bib_ref]. Another possibility is the pearly chain method, through which the fusion occurs with the aid of an electric field and laser radiation. In this case, the contact cell surface is irradiated with pulsed laser beams to make a small perforation in the cell membrane, which enhances the chance to promote cell fusion [bib_ref] New technique for producing hybridoma by using laser radiation, Ohkohchi [/bib_ref] [bib_ref] Hybridoma technologies for antibody production, Tomita [/bib_ref]. Although the pearly chain method has advantages over the PEG-mediated strategy, Abbreviations: ASC, antibody-secreting cell; BCT, B cell targeting; cDNA, complementary DNA; CDR, complementarity determining region; Fab, antigen-binding fragment; mAb, monoclonal antibody; PEF, pulsed electric field; PEG, polyethylene glycol; scFv, single-chain variable fragment; SST, stereospecific targeting; VH, heavy chain variable domain; VL, light chain variable domain.
it still cannot selectively control the fusion of a specific B lymphocyte with myeloma cell [bib_ref] Hybridoma technologies for antibody production, Tomita [/bib_ref].
Since the introduction of the hybridoma technology, mAbs have had a profound impact on medicine, providing an almost limitless source of therapeutic, diagnostic, and research reagents [bib_ref] Historical development of monoclonal antibody therapeutics, Nissim [/bib_ref] [bib_ref] From the discovery of monoclonal antibodies to their therapeutic application: an historical..., Ribatti [/bib_ref]. Given the universality and usefulness of mAbs, many discoveries came as a result of hybridoma technology, allowing the generation of antibodies directed against an antigen or even different antibodies against the same antigen [bib_ref] Hybridoma technology a versatile method for isolation of monoclonal antibodies, its applicability..., Parray [/bib_ref]. Among the advantages of this technique, we can list the highly reproducible mAb obtainment, once the hybridoma clones are established, the preservation of the native pairing of the combination of genes of the antibody variable and constant regions, and the in vivo antibody affinity maturation [bib_ref] Hybridoma technology: the preferred method for monoclonal antibody generation for in vivo..., Zaroff [/bib_ref] [fig_ref] Table 1: Advantages and disadvantages of technologies used to generate monoclonal antibodies [/fig_ref].
Muromonab-CD3, also called orthoclone (OKT3), was the first mAb approved by the Food and Drug Administration (FDA), in 1986, for therapeutic use in humans [fig_ref] Figure 1: Timeline of important events in the generation of monoclonal antibodies [/fig_ref] [bib_ref] The therapeutic monoclonal antibody market, Ecker [/bib_ref]. That is a murine hybridoma-derived mAb targeting CD3 on mature peripheral T cells to avoid organ allograft rejection [bib_ref] Laboratory monitoring of therapy with OKT3 and other murine monoclonal antibodies, Colvin [/bib_ref]. However, the occurrence of a human anti-mouse immune response has limited the clinical applicability of murine mAbs in humans [bib_ref] Minimizing the immunogenicity of antibodies for clinical application, Gonzales [/bib_ref]. The most appropriate strategy for obtaining therapeutic mAbs would come with the use of human hybridomas, but attempts to obtain these hybrid cells failed, mostly due to their genetic instability [bib_ref] Use of human hybridoma technology to isolate human monoclonal antibodies, Smith [/bib_ref]. On the other hand, technological advances allowed the structural modification of these molecules, and the first achievements on that made feasible the removal of antibody murine markers, giving rise to chimeric mAbs containing fragments of variable regions of the murine antibody light and heavy chains linked to human immunoglobulin constant regions. The chimeric mAbs are originated from mouse myeloma cells transfected with chimeric genes, producing antibodies with human features and the same antigen specificity of the antibody originally generated in mice [bib_ref] Chimeric human antibody molecules: mouse antigen-binding domains with human constant region domains, Morrison [/bib_ref]. Abciximab (c7E3 Fab) was the first chimeric antibody approved by the FDA, in 1994, to inhibit platelet aggregation in high-risk angioplasty cases [fig_ref] Figure 1: Timeline of important events in the generation of monoclonal antibodies [/fig_ref] [bib_ref] Platelet glycoprotein IIb/IIIa receptor inhibitors in ischemic heart disease, Lefkovits [/bib_ref]. Following studies led to a process known as antibody humanization, which grafts non-human antibodies complementarity determining regions (CDR) into human antibody scaffolds. That is obtained using non-human antibody framework regions as CDR graft acceptors [bib_ref] Replacing the complementarity-determining regions in a human antibody with those from a..., Jones [/bib_ref] [bib_ref] Antibody humanization methods -a review and update, Safdari [/bib_ref]. In 1997, the FDA approved the first humanized antibody, called daclizumab [fig_ref] Figure 1: Timeline of important events in the generation of monoclonal antibodies [/fig_ref] , which is indicated for prophylaxis of acute organ rejection in patients who received a kidney transplant and, subsequently, it was also allowed for the treatment of adults with recurrent forms of multiple sclerosis [bib_ref] Daclizumab: development, clinical trials, and practical aspects of use in multiple sclerosis, Baldassari [/bib_ref]. In the next decade, a great advance happened with the obtainment of appropriate transgenic animals for generating fully human mAbs [bib_ref] Antigen-specific human antibodies from mice comprising four distinct genetic modifications, Lonberg [/bib_ref]. This achievement was possible due to several methodological advances that allowed the integration of the human immunoglobulin gene loci into the mouse genome in a stable way, along with the inactivation of the endogenous murine immunoglobulin genes [bib_ref] High-affinity IgG antibodies develop naturally in Ig-knockout rats carrying germline human IgH/Igκ/Igλ..., Osborn [/bib_ref] [bib_ref] Mice with megabase humanization of their immunoglobulin genes generate antibodies as efficiently..., Murphy [/bib_ref]. Other transgenic animals, such as cattle, rabbits, and rats, can also be exploited for the biological production of human antibodies [bib_ref] Efficient immunoglobulin gene disruption and targeted replacement in rabbit using zinc finger..., Flisikowska [/bib_ref] [bib_ref] High-affinity IgG antibodies develop naturally in Ig-knockout rats carrying germline human IgH/Igκ/Igλ..., Osborn [/bib_ref] [bib_ref] Triple immunoglobulin gene knockout transchromosomic cattle: bovine lambda cluster deletion and its..., Matsushita [/bib_ref]. The genetic manipulation of the genome was made such that the transgenic animal immunization with the antigen of interest turns possible the generation of murine hybridomas secreting human mAbs. The first hybridoma-derived human mAb isolated from transgenic animalspanitumumabwas approved for therapeutic use in 2006 [bib_ref] From XenoMouse technology to panitumumab, the first fully human antibody product from..., Jakobovits [/bib_ref] [fig_ref] Figure 1: Timeline of important events in the generation of monoclonal antibodies [/fig_ref].
The hybridoma technology has remained at the forefront of the mAb generation field (Zaroff and Tan 2019). Currently, more than 90% of the -Antibody chimerization and humanization methods and transgenic animals can be used to obtain mAbs for therapeutic use in humans. -Antibodies undergo in vivo affinity maturation.
-Known and available antigen targets are needed.
-Low efficiency on cell fusion and hybridoma isolation.
-It is required a relatively long period for generating the cell line and the selection of a specific hybridoma.
-Hybridoma cell lines may be genetically unstable.
-Constant risk of cell culture contamination. B Cell Targeting -More efficient cell fusion compared to the original hybridoma technique.
-Use only B lymphocyte selected by antigen.
-Possibility to simultaneously generate at least 3 specific mAbs against different antigens, using a single mouse.
-Electrostatic field applications might be challenging.
-High technical expertise is needed.
## Stereospecific targeting
-More efficient cell fusion compared to the original hybridoma technique.
-Generation of mAbs that recognize native antigen conformations, instead of linear structures.
-DNA immunizations are cheaper than the original hybridoma technique and allow the generation of antibodies against complex or non-conventional antigens.
-Might be more time-consuming and expensive than the previous techniques, particularly if cell lines for immunization, cell fusion, and screening steps are not available. -As it is necessary to perform an electric fusion, it also has the disadvantages of the BCT technique.
## Antibody phage display
-Animal host is not required.
-The screening of a large number of clones increases the chances of generating good mAbs.
-Potential to isolate mAbs against toxic and non-immunogenic antigens.
-Possibility to redesign natural CDRs for generating mAbs of improved specificity and affinity.
-Display libraries are commercially available.
-The diversity of the phage library depends on the bacterial transformation efficiency.
-Antibody formats are limited to scFv and Fab.
-Building a phage display library is expensive.
Single B cell -High efficiency in obtaining specific mAbs, compared to hybridoma technology.
-Possibility to isolate mAbs from vaccinated or naturally immunized human subjects.
-Isolation of native mAbs with the preservation of natural cognate VH and VL pairing.
-No need to culture B cells.
-Potential to isolate functional mAbs against conformation determinants that are difficult to emulate in vitro. antibodies approved for therapeutic use were generated by this technology, most of them in chimeric or humanized versions [bib_ref] Hybridoma technology a versatile method for isolation of monoclonal antibodies, its applicability..., Parray [/bib_ref]. However, the dominance of this method is accompanied by its low efficiency. Hybridoma-based mAb generation is marked by long screening processes, suboptimal selection of specific mAb-secreting cells, a mAb validation that is rarely possible at an early stage, not to mention that the availability of the purified antigen target is needed. To optimize antibody generation, several variants of this technology have been developed over the years. Examples are the B Cell Targeting and the Stereospecific Targeting techniques, which are described below.
## B cell targeting (bct)
The B Cell Targeting (BCT) method, also known as Pulsed Electric Field (PEF), was described by [bib_ref] Monoclonal antibody production by receptor-mediated electrically induced cell fusion, Lo [/bib_ref]. It is based on two central points: the preselection of B lymphocytes recognizing the antigen of interest, and the further B lymphocyte fusion with myeloma cells by using direct current electrical pulses [bib_ref] Hybridoma technologies for antibody production, Tomita [/bib_ref]. Briefly, specific biotin-labeled antigen binds to the corresponding B lymphocytes, which are subsequently recovered by using streptavidin, giving rise to a B lymphocyte-antigen-biotin-streptavidin complex [bib_ref] Hybridoma technologies for antibody production, Tomita [/bib_ref] [bib_ref] Electro cell fusion for hybridoma production, Greenfield [/bib_ref]. Then, such B lymphocyte complexes are co-cultured with biotin-labeled myeloma cells and the resulting mixture is exposed to PEF to promote cell fusion [bib_ref] Monoclonal antibody production by receptor-mediated electrically induced cell fusion, Lo [/bib_ref].
This last step, the most critical one, is characterized by the cell membrane destabilization after electrostatic field exposure, which eases the occurrence of fusion between cell membranes [bib_ref] Electro cell fusion for hybridoma production, Greenfield [/bib_ref]. For that, a strong electric field is formed vertically between electrodes arranged in parallel and guides the alignment of the B lymphocyte-myeloma cell complexes along with it, favoring the fusion of the membranes close to each other. No electrical fusion occurs in complexes arranged in any other direction [bib_ref] Hybridoma technologies for antibody production, Tomita [/bib_ref]. Different research groups have explored the application of electrostatic pulses for generating hybridomas [bib_ref] Hybridoma production by simplified avidinmediated electrofusion, Wojchowski [/bib_ref] [bib_ref] The use of peptidemediated electrofusion to select monoclonal antibodies directed against specific..., Werkmeister [/bib_ref] [bib_ref] The use of an electroporation apparatus for the production of murine hybridomas, Hewish [/bib_ref]. In general, the cell fusion mediated by electric field was found more efficient than the achieved with PEG, a cytotoxic agent [bib_ref] Selective production of hybridoma cells: antigenic-based pre-selection of B lymphocytes for electrofusion..., Tomita [/bib_ref] [bib_ref] Orientation of biotinbinding sites in streptavidin adsorbed onto the surface of polythiophene..., Awsiuk [/bib_ref] , with improvements not only in the number of fused cells but also in the hybridoma growth rate. The BCT technique demonstrated five-to-ten times greater efficiency in the formation of hybridoma cells secreting the antibodies of interest, in comparison with the PEG-mediated method. However, based on the reported data, such fusion efficiency does not seem to go far beyond 20% [bib_ref] Antigenbased immunofluorescence analysis of B-cell targeting: advanced technology for the generation of..., Tomita [/bib_ref] , and the BCT protocol is more complex than the original hybridoma one. Another point to note is that the electrofusion yields are low when the fusion partner cells have different sizes, although this is a limitation that can be overcome with the use of nanosecond pulse electroporation [bib_ref] Cell electrofusion using nanosecond electric pulses, Rems [/bib_ref].
The BCT method can also be used for the simultaneous generation of at least three to five mAbs against different antigens, using a single mouse [bib_ref] Orientation of biotinbinding sites in streptavidin adsorbed onto the surface of polythiophene..., Awsiuk [/bib_ref] , which reduces not only the laboratory work but also the number of animals needed for isolating mAbs. This procedure, known as multitargeting, is based on mouse immunization with multiple antigens, followed by the immunoglobulin B-cell receptor-guided selection of B lymphocytes sensitized by each of the desired antigens. As a disadvantage, immunosuppression caused by immunization with several antigens may occur [fig_ref] Table 1: Advantages and disadvantages of technologies used to generate monoclonal antibodies [/fig_ref] [bib_ref] Hybridoma technologies for antibody production, Tomita [/bib_ref] [bib_ref] Orientation of biotinbinding sites in streptavidin adsorbed onto the surface of polythiophene..., Awsiuk [/bib_ref].
## Stereospecific targeting (sst)
Early descriptions of conformation-specific mAbs were published in the 1960s [bib_ref] Synthetic antigens composed exclusively of L-or Damino acids. I. Effect of optical..., Janeway [/bib_ref] , highlighting the characteristic of these antibodies in specifically recognizing only one type of stereoisomer of a given chemical compound. It is known that stereospecific mAbs have high specificity for their ligands, which is helpful for diagnostic and therapeutic approaches. However, the generation of these mAbs is technically challenging, particularly in the case of highly structured and well-preserved targets. Examples are extracellular loops or domains of multi-transmembrane proteins, such as membrane-bound receptors [bib_ref] An improved and robust DNA immunization method to develop antibodies against extracellular..., Hazen [/bib_ref]. The Stereospecific Targeting (SST) method was proposed to address this problem [bib_ref] A new technology for intact antigen-based and receptor-mediated generation of novel monoclonal..., Tomita [/bib_ref] [fig_ref] Figure 1: Timeline of important events in the generation of monoclonal antibodies [/fig_ref] and consists of four phases.
A modification in the original hybridoma technology was performed already in the first step, the animal immunization. The immunogen is administered intramuscularly in the DNA form [bib_ref] A new technology for intact antigen-based and receptor-mediated generation of novel monoclonal..., Tomita [/bib_ref] , which guides the expression of the antigen in its native form. Thereby, the chances of inducing the production of functional mAbs are greater, even against the most challenging targets . Compared to protein inoculation, gene immunization allows the efficient testing of different designs of immunogens, does not require purification of proteins from a pathogen, circumvents the difficulty of expressing and purifying antigens in large quantities, and can also be used to obtain antibodies against several proteins at the same time through immunization with several nucleic acid sequences that encode different proteins or different subunits of the same protein , which are relevant advantages for generating high-quality mAbs. Although the DNA immunizations can be considered not very immunogenic in some cases, the use of immunomodulators, if necessary, does not interfere negatively in the conformation of the antigen. Also, among the options of entry pathways for DNA immunization, the intrasplenic administration may be still more efficient, since a single dose of DNA is sufficient to generate the desired antibody responses, with reduced immunization period and technique cost, compared to the traditional protein administration [bib_ref] Hybridoma technology a versatile method for isolation of monoclonal antibodies, its applicability..., Parray [/bib_ref]. On the other hand, the antigen glycosylation pattern, that differs from the occurring in humans, as well as the possibility of inducing immune tolerance and generating anti-DNA antibodies may be problems when using this approach [bib_ref] DNA vaccines: roles against diseases, Khan [/bib_ref]. The transduction of myeloma cells to express the antigen is a limitation that sums to those described for BCT. In a recent update, an additional intraperitoneal injection containing cells that express the target antigen has been proposed to increase the humoral response and ensure the recognition of antigenic structures. The idea is to promote a further stage in the B cell maturation. Indeed, an increase in serum antibody titers, when compared to the results of gene immunization only, could be observed [fig_ref] Table 1: Advantages and disadvantages of technologies used to generate monoclonal antibodies [/fig_ref] [bib_ref] Optimization of stereospecific targeting technique for selective production of monoclonal antibodies against..., Yamasaki [/bib_ref]. The second step involves the preselection of conformational epitope-recognizing B cells. For this, isolated splenic cells are incubated for a short period with myeloma cells transduced with a vector carrying the antigen gene for the formation of B lymphocyte-myeloma cell complexes [bib_ref] Optimization of gene transfection in murine myeloma cell lines using different transfection..., Shabani [/bib_ref]. The third step is the cell fusion itself, which occurs by using electrical pulses as described for the BCT method. The screening of hybridomas secreting the desired mAbs, the fourth step, makes use of the native antigen targets expressed on a cell surface. The clone selection may include an additional step to discard the undesirable clones by using recombinant protein, which may contain partially denatured structures [bib_ref] Optimization of stereospecific targeting technique for selective production of monoclonal antibodies against..., Yamasaki [/bib_ref]. The SST method provides more than 50% positivity for B lymphocyte-myeloma cell fusion, and more than 24% of the generated clones were found to secrete the desired mAbs [bib_ref] Optimization of stereospecific targeting technique for selective production of monoclonal antibodies against..., Yamasaki [/bib_ref].
## Antibody phage display technology
The antibody phage display technology, initially reported in 1990 [bib_ref] Phage antibodies: filamentous phage displaying antibody variable domains, Mccafferty [/bib_ref] , is considered a powerful tool to generate mAbs [fig_ref] Figure 1: Timeline of important events in the generation of monoclonal antibodies [/fig_ref]. The methodology, based on the phage display concept described by [bib_ref] Filamentous fusion phage: novel expression vectors that display cloned antigens on the..., Smith [/bib_ref] , consists in the development of a combinatorial antibody phage librarythat is, a huge collection of phages displaying antibody fragmentsand the subsequent screening of the antibodies that recognize the antigen of interest.
To generate an antibody phage library, firstly it is necessary to clone antibody gene fragments into vectors. Both filamentous M13 phage and phagemid, which combines the characteristics of plasmids and phages [bib_ref] Phage display on the base of filamentous bacteriophages: application for recombinant antibodies..., Tikunova [/bib_ref] , can be used as vectors. Comparatively, while the first one has all the ability to produce phage particles and display antibody, the phagemid needs to infect bacteria with a helper phage, that is required to package the phagemid as single-strand DNA into virion particle [bib_ref] Assembly of combinatorial antibody libraries on phage surfaces: the gene III site, Barbas 3rd [/bib_ref] [bib_ref] Phage display libraries for antibody therapeutic discovery and development, Almagro [/bib_ref]. In both cases, vectors are used to transform E. coli by electroporation. After obtaining the phage display library, the antibodies displayed on the vector surface are screened through a process called biopanning [bib_ref] Advancement and applications of peptide phage display technology in biomedical science, Wu [/bib_ref]. It should be noted that the antibodies are most often displayed in single-chain variable fragment (scFv) or antigen-binding fragment (Fab) forms.
There are four types of antibody display libraries: immune, naïve, semisynthetic, and synthetic. The immune libraries are obtained from immunized animals or humans and are mostly used to discover antibodies against infectious pathogens [bib_ref] Functional characterization of two scFv-Fc antibodies from an HIV controller selected on..., Trott [/bib_ref] or antigenic targets in cancer patients [bib_ref] Rise and fall of an anti-MUC1 specific antibody, Thie [/bib_ref] [bib_ref] Phage display-derived human antibodies in clinical development and therapy, Frenzel [/bib_ref]. This library contains a restricted antibody repertoire that underwent antigen-driven in vivo selection [bib_ref] Assembly of combinatorial antibody libraries on phage surfaces: the gene III site, Barbas 3rd [/bib_ref] [bib_ref] Efficient method for constructing comprehensive murine Fab antibody libraries displayed on phage, Orum [/bib_ref] [bib_ref] Phage display-derived human antibodies in clinical development and therapy, Frenzel [/bib_ref] , which differs from the other phage display libraries, known as "universal", that theoretically provide binders for all possible antigen structures [bib_ref] Phage display-derived human antibodies in clinical development and therapy, Frenzel [/bib_ref]. The naïve antibody libraries are generated from a pool of B lymphocytes of non-immunized donors, and one successful example is the scFv library licensed from Cambridge Antibody Technology (CAT; now part of MedImmune/AstraZeneca) [bib_ref] Ramucirumab: successfully targeting angiogenesis in gastric cancer, Javle [/bib_ref] [bib_ref] Phage display libraries for antibody therapeutic discovery and development, Almagro [/bib_ref]. While the naïve libraries are derived from natural antibody gene repertoires, the synthetic ones are entirely based on in silico design to obtain individual antibody amino acid sequences [bib_ref] Synthetic antibodies as therapeutics, Fuh [/bib_ref] , bypassing the need to isolate antibody genes. The semisynthetic libraries, on the other hand, are created using both naturally and synthetically (in silico) randomized CDRs. In this library type, it is possible to redesign natural CDRs to improve the chance of finding antibodies with high specificity and affinity [bib_ref] Efficient method for constructing comprehensive murine Fab antibody libraries displayed on phage, Orum [/bib_ref] [bib_ref] Synthetic antibodies as therapeutics, Fuh [/bib_ref].
Building the phage display library is the most important step of this technology. There is a directly proportional relationship between the size of the antibody library and the probability of finding a particular antibody [bib_ref] An improved phage display vector for antibody repertoire cloning by construction of..., Burioni [/bib_ref] [bib_ref] Phage display libraries for antibody therapeutic discovery and development, Almagro [/bib_ref]. The Next-Generation Sequencing (NGS) is an important tool to analyze the variability, the sequence composition, and the size of antibody phage display libraries [bib_ref] Next-generation sequencing of antibody display repertoires, Rouet [/bib_ref]. The construction of a phage display library is more expensive than generating hybridomas after animal immunization. However, the antibody screening step of the phage display method is faster and cheaper [bib_ref] Monoclonal antibody generation by phage display: history, state-of-the-art, and future, Hentrich [/bib_ref].
The first antibody discovered by phage display (CAT library) as well as the first human antibody approved for therapy was adalimumab (Humira®) [fig_ref] Figure 1: Timeline of important events in the generation of monoclonal antibodies [/fig_ref] [bib_ref] Adalimumab: long-term safety in 23 458 patients from global clinical trials in..., Burmester [/bib_ref]. It is an IgG1 mAb that binds tumor necrosis factor-alpha (TNF-α) and prevents the interaction of this inflammatory cytokine with the corresponding receptor. Having been discovered from an scFv phage library, gene manipulation was needed to obtain the final IgG format [bib_ref] Adalimumab -a new TNF-alpha antibody for treatment of inflammatory joint disease, Machold [/bib_ref]. This antibody has been used for the treatment of patients with moderate to severe rheumatoid arthritis, among other autoimmune diseases.
Although the phage display library is a promising technology for the development of antibodies, it has limitations. The diversity of the phage library depends on the bacterial transformation efficiency and is limited to the 10 10 -10 11 variant antibody maximum repertoire of the phage display library. This restriction can be overcome by mRNA and ribosome display strategies, which are in vitro cell-free methods having a bigger library size and a higher displayed antibody diversity (10 14 variants) [bib_ref] Engineered antibodies, Hudson [/bib_ref] [bib_ref] Ribosome display technology: applications in disease diagnosis and control, Kunamneni [/bib_ref]. It should be also considered that phage display-selected mAbs are generated in E. coli and therefore are not glycosylated; the use of eukaryotic display platforms, like yeast [bib_ref] Therapeutic antibody engineering by high efficiency cell screening, Doerner [/bib_ref] and mammalian expression systems [bib_ref] New mammalian expression systems, Zhu [/bib_ref] , is a possibility to circumvent that. Other antibody phage display methodology disadvantages are the propensity to generate biased repertoires and the loss of information of antibody natural pairing [bib_ref] Antibody isolation from immunized animals: comparison of phage display and antibody discovery..., Saggy [/bib_ref] [fig_ref] Table 1: Advantages and disadvantages of technologies used to generate monoclonal antibodies [/fig_ref].
## Single b cell antibody technology
Several technological platforms have been proposed to generate mAbs from hybridomas. An inherent characteristic of these methods is the need to fuse B lymphocytes with myeloma cells [bib_ref] Continuous cultures of fused cells secreting antibody of predefined specificity, K€ Ohler [/bib_ref] and this was, for a long time, a required step to isolate single antibodies of known specificity. In the last few decades, technical advances have allowed the detection and isolation of single functional B lymphocytes from heterogeneous primary cell populations, as well as the antibody gene amplification and cloning without the need to immortalize the selected antibody-secreting cell (ASC). These single B lymphocyte approaches, collectively known as "single B cell antibody technology" [fig_ref] Figure 1: Timeline of important events in the generation of monoclonal antibodies [/fig_ref] [bib_ref] A novel strategy for generating monoclonal antibodies from single, isolated lymphocytes producing..., Babcook [/bib_ref] , revealed attractive and useful to generate neutralizing mAbs in a rapid way for several applications [bib_ref] Efficient generation of monoclonal antibodies from single human B cells by single..., Tiller [/bib_ref] , including the management of emerging pathologies. Indeed, an increasing number of mAbs against infections caused by viral agents, such as HIV [bib_ref] A method for identification of HIV gp140 binding memory B cells in..., Scheid [/bib_ref] , Dengue, MERS-CoV [bib_ref] Importance of neutralizing monoclonal antibodies targeting multiple antigenic sites on the Middle..., Wang [/bib_ref] , and SARS-Cov-2 [bib_ref] Potent neutralizing antibodies against SARS-CoV-2 identified by high-throughput single-cell sequencing of convalescent..., Cao [/bib_ref] , were obtained with such technology. The following items briefly describe the basic concepts and benefits of the single B cell antibody technology.
## Identification and isolation of single b cells
The screening and isolation of ASC can occur in a random or antigenspecific manner, from peripheral blood or lymphoid tissue samples. For random selection, B cells can be recovered by flow cytometry [bib_ref] Rapid generation of fully human monoclonal antibodies specific to a vaccinating antigen, Smith [/bib_ref] or can be picked from tissues by micromanipulation [bib_ref] Tracing B cell development in human germinal centres by molecular analysis of..., Küppers [/bib_ref]. For antigen-specific selection, multi-parameter flow cytometry or other fluid-based approaches are generally used [bib_ref] The rapid generation of recombinant functional monoclonal antibodies from individual, antigen-specific bone..., Clargo [/bib_ref] [bib_ref] Efficient generation of monoclonal antibodies from single rhesus macaque antibody secreting cells, Meng [/bib_ref] [bib_ref] Recombinant human B cell repertoires enable screening for rare, specific, and natively..., Rajan [/bib_ref]. Flow cytometry systems are efficient to recover single cells [bib_ref] Single cell sorting and cloning, Battye [/bib_ref] and an example is their successful use to isolate IgG þ memory B lymphocytes reactive to gp140 from donors with HIV [bib_ref] A method for identification of HIV gp140 binding memory B cells in..., Scheid [/bib_ref] [bib_ref] Broad diversity of neutralizing antibodies isolated from memory B cells in HIV-infected..., Scheid [/bib_ref]. In this case, anti-CD19 and anti-IgG antibodies, along with biotinylated gp140, were used to select the desired cell subset. Such methodology led to the generation of anti-gp140 mAbs with different antigen neutralization activities [bib_ref] Broad diversity of neutralizing antibodies isolated from memory B cells in HIV-infected..., Scheid [/bib_ref].
It should be noted that antigen-specific IgG þ B cells comprise just a small percentage of circulating cells and, to identify and isolate them, reagents targeting B cell surface markers are desirable. A variety of antibodies are available to detect human B lymphocytes, which makes it even possible to distinguish cells at different stages of development and differentiation. This is an advantage of the single B cell technology over the original hybridoma technique. On the other side, the scenario is not the same when it comes to isolating non-human subsets. Indeed, we do have antibodies against mouse B lymphocyte markers [bib_ref] Generation of recombinant monoclonal antibodies from immunised mice and rabbits via flow..., Starkie [/bib_ref] , such as CD45R and CD19, but the sorting of B cells from most of the other species (rabbit and guinea pig, for example), although feasible [bib_ref] Generation of recombinant monoclonal antibodies from immunised mice and rabbits via flow..., Starkie [/bib_ref] [bib_ref] Antigen-specific single B cell sorting and monoclonal antibody cloning in Guinea pigs, Lei [/bib_ref] , becomes challenging due to the low or absent repertoire of appropriate B cell-targeting antibodies. Another point that should be considered is related to cost: the use of expensive sorting devices integrates an important part of the procedures to isolate antigen-specific single B lymphocytes from a polyclonal mixture. Alternatively, other strategies can be used, including antigen-coated magnetic beads [bib_ref] Rare, high-affinity anti-pathogen antibodies from human repertoires, discovered using microfluidics and molecular..., Adler [/bib_ref] , cell-based microarrays [bib_ref] Rapid isolation of antigen-specific antibody-secreting cells using a chip-based immunospot array, Jin [/bib_ref] , and soft lithographic methods for micro engraving [bib_ref] A microengraving method for rapid selection of single cells producing antigenspecific antibodies, Love [/bib_ref]. The downside? These techniques are also costly or require extensive knowledge.
## Single-cell immunoglobulin gene transcript amplification, cloning, and expression
Having isolated single B cells, the next step is the immunoglobulin gene amplification. The cells are lysed, the cDNA is synthesized by reverse transcription of total mRNA, and the full-length immunoglobulin genes for the variable and constant regions of the light and heavy chains are amplified by PCR [bib_ref] Efficient generation of monoclonal antibodies from single human B cells by single..., Tiller [/bib_ref]. The obtained fragments are cloned into linear expression cassettes to further generate the immunoglobulin domains in cell-based expression systems (mammalian or bacterial cells). In scenarios without the cultivation of the recovered B cells, the cDNA is synthesized from single-cell material. The antibodies are typically expressed in Fab form [bib_ref] The rapid generation of recombinant functional monoclonal antibodies from individual, antigen-specific bone..., Clargo [/bib_ref] , but it is also possible to express them in other formats, including full-length IgG and single-chain variable fragment (scFv) [bib_ref] Efficient generation of monoclonal antibodies from single rhesus macaque antibody secreting cells, Meng [/bib_ref] [bib_ref] Recombinant human B cell repertoires enable screening for rare, specific, and natively..., Rajan [/bib_ref].
These procedures summarize a common protocol route for protein expression. However, more robust and sophisticated systems are also available. That is the case of the "single-cell RT-PCR-linked in vitro expression" (SICREX) platform, through which the antibodies are expressed outside a cell unit [bib_ref] A novel strategy for generation of monoclonal antibodies from single B cells..., Jiang [/bib_ref] [bib_ref] In vitro generation of rabbit anti-Listeria monocytogenes monoclonal antibody using single cell..., Ojima-Kato [/bib_ref]. In this system, the protein synthesis occurs in a mixture containing the transcription/translation machinery from E. coli, and therefore the gene-cloning, transformation, and cultivation procedures are not needed. As a consequence, the time to generate the antibodies is greatly reduced to just a few days. Here we also have a drawback: incorrect folding of the antibody domains sometimes occurs.
From a broad perspective, the single B cell antibody technology, just like the other methods discussed in this review, has its advantages balanced by downsides, revealing a singular panel that characterizes it. Compared with the current hybridoma technology, though, the single B cell approaches have some positive points that stand out and even exceed those exposed above. It can be included here the potential to (a) isolate mAbs reactive to conformational determinants that are difficult to emulate in vitro; and (b) in experimental studies, collect multiple samples after the immunization period without the need to euthanize the animals [bib_ref] Efficient generation of monoclonal antibodies from single human B cells by single..., Tiller [/bib_ref] [bib_ref] Generation of recombinant monoclonal antibodies from immunised mice and rabbits via flow..., Starkie [/bib_ref] [bib_ref] Recombinant human B cell repertoires enable screening for rare, specific, and natively..., Rajan [/bib_ref]. But the biggest advantage of single B cell approaches is the possibility to isolate neutralizing mAbs from vaccinated or naturally immunized human subjects, as well as from those with autoimmune diseases. The high-throughput screening of individual ASC repertoire based on phenotypic and genotypic features allows the analysis of the human immune response to pathogens [bib_ref] More than one antibody of individual B cells revealed by single-cell immune..., Shi [/bib_ref] , accelerates the search for neutralizing mAbs of therapeutic relevance, and also provides insights for a rational vaccine design strategy [bib_ref] Broad diversity of neutralizing antibodies isolated from memory B cells in HIV-infected..., Scheid [/bib_ref].
Overall, the recent advances in the single B cell field trace a path that was out of reach when C esar Milstein and Georges K€ ohler found on the hybridoma creation the magic solution to isolate mAbs [bib_ref] Continuous cultures of fused cells secreting antibody of predefined specificity, K€ Ohler [/bib_ref]. [fig_ref] Table 1: Advantages and disadvantages of technologies used to generate monoclonal antibodies [/fig_ref] summarizes some of the advantages and drawbacks of the single B cell antibody technology, in comparison with the hybridoma and phage display techniques.
# Discussion
Given the foregoing, we can consider that the choice of the method to be used for obtaining an antibody must be guided by the purpose of the demand. The first demonstration that mAbs could be isolated came with the hybridoma technology, which made feasible the use of these molecules for a variety of biological applications. The task was revealed to be not as practical as it might seem, though. Hybridoma-derived immunoglobulins are of animal origin and, to be used as therapeutic tools, need to be converted into human mAbs. Such protein structural change can be currently achieved with established antibody chimerization and humanization protocols or the use of appropriate transgenic animals, in strategies that were crucial for the obtainment of the therapeutic mAb repertoire available today but are known to be costly, time-consuming, and technically challenging [bib_ref] Antibody humanization methods -a review and update, Safdari [/bib_ref]. The limitations are not restricted to that. The low efficiency of the B lymphocyte-myeloma cell fusion and the further hybridoma cell isolation are important bottlenecks of this technology, not to mention the constant risk of cell culture contamination and the genetic instability of the generated hybridoma cell lines.
Since the mid-1980s, several methods have been developed to work around these limitations, starting with changes in hybridoma technology. Examples are the proposed BCT and SST protocols, that brought relative improvements in the B lymphocyte-myeloma cell fusion efficiency, but instead turned the hybridoma technique more complex and hardworking, compared with the original methodology. Alterations in the other steps of this technology, such as the selection of the desired antibody-secreting cells, have been also described [bib_ref] Analysis and sorting of live cells according to secreted molecules, relocated to..., Manz [/bib_ref] [bib_ref] Antibodies and selection of monoclonal antibodies, Hanack [/bib_ref] [bib_ref] A novel selection strategy for antibody producing hybridoma cells based on a..., Listek [/bib_ref] ; however, despite indeed accelerating the mAb identification process, the need to generate hybridomas remains. Based on different principles, the antibody phage display method emerged as the first alternative to the hybridoma technology. It brings important advantages, such as the potential to isolate mAbs against toxic and non-immunogenic antigens, and the possibility to generate, for the first time, antibodies without using experimental animals. On the other hand, an important limitation is the need to have an available and previously identified target antigen, which is also valid for the hybridoma technology.
Despite improvements in the hybridoma technology, and the development of antibody display [bib_ref] Making antibodies by phage display technology, Winter [/bib_ref] , chimerization and humanization strategies [bib_ref] Man-made antibodies, Winter [/bib_ref] , a major advance came with the discovery of tools to isolate mAbs directly from single B cells. Besides not strictly depending on B cell culture and the use of experimental animals, the single B cell antibody technology allows a simple and rapid generation of mAbs with therapeutic potential without the need to previously know the target and have it available. This is a promising technique with the potential for even isolating functional mAbs against conformation determinants that are difficult to emulate in vitro but, currently, it still has low accessibility, particularly compared to the hybridoma methodology.
Overall, all the technologies discussed above revealed useful for obtaining therapeutic antibodies against several disorders, including infectious diseases. More than a hundred mAbs described against the Ebola virus illustrate that and, among them, some hybridoma-derived antibodies were used to develop therapeutic cocktails, such as ZMapp, composed of three chimeric mAbs [bib_ref] Characterization of Zaire ebolavirus glycoprotein-specific monoclonal antibodies, Qiu [/bib_ref] [bib_ref] Reversion of advanced Ebola virus disease in nonhuman primates with ZMapp, Qiu [/bib_ref] [bib_ref] CHO cell production and sequence improvement in the 13C6FR1 anti-Ebola antibody, Pettit [/bib_ref] , and REGN-EB3, comprising three human mAbs generated by using appropriate transgenic mice [bib_ref] Development of clinical-stage human monoclonal antibodies that treat advanced Ebola virus disease..., Pascal [/bib_ref]. Other examples of antibodies generated toward the Ebola virus are the phage display-derived mAb KZ52 [bib_ref] Ebola virus can be effectively neutralized by antibody produced in natural human..., Maruyama [/bib_ref] , and the single B cell-isolated antibody Mab114, obtained from a human survivor [bib_ref] Protective monotherapy against lethal Ebola virus infection by a potently neutralizing antibody, Corti [/bib_ref]. But when considering emerging diseases, the hybridoma technology does not seem to be the most appropriate, particularly taking into account the need to obtain therapeutics in a short time. In this situation, the single B cell antibody technology seems to better respond to the urgent demand for functional mAbs, which is illustrated by the experience in the recent COVID-19 pandemic. In a period less than one year, at least 14 single B cell-derived human mAbs or mAb cocktails were obtained against SARS-CoV-2, the causative agent of this disease, and five of them entered Phase 2/3 clinical trials [bib_ref] Anti-SARS-CoV-2 neutralizing monoclonal antibodies: clinical pipeline, Tuccori [/bib_ref]. Another positive point of the single B cell antibody technology is the possibility to isolate the desired mAbs without previously knowing the antigen target, which could be particularly helpful in infectious disease cases. However, all that does not exclude the potential application of other methodologies in the fight against emerging pathogens. Indeed, a panel of neutralizing mAbs elicited against SARS-CoV-2 was obtained from phage display libraries [bib_ref] A panel of human neutralizing mAbs targeting SARS-CoV-2 spike at multiple epitopes, Noy-Porat [/bib_ref] , and even hybridoma-based strategies have been explored for that purpose .
So, is the hybridoma technology still useful? The reported data so far indicate yes. Beyond being a pioneer, this methodology is very popular. Several of the most recently generated mAbs were discovered on murine hybridomas [bib_ref] Anti-bevacizumab idiotype antibody vaccination is effective in inducing vascular endothelial growth factor-binding..., Sanches [/bib_ref] [bib_ref] Hybridoma technology a versatile method for isolation of monoclonal antibodies, its applicability..., Parray [/bib_ref] , including some of the most successful FDA-approved antibodies, such as the immune checkpoint inhibitors nivolumab (anti-programmed cell death protein 1; anti-PD-1) [bib_ref] Anti-programmed-death-receptor-1 treatment with pembrolizumab in ipilimumab-refractory advanced melanoma: a randomised dosecomparison cohort..., Robert [/bib_ref] and atezolizumab (anti-programmed cell death protein ligand 1; PD-L1) [bib_ref] Atezolizumab versus docetaxel for patients with previously treated non-small-cell lung cancer (POPLAR):..., Fehrenbacher [/bib_ref] , used in the management of non-small cell lung carcinomas, head and neck cancers, melanomas, renal cell carcinomas, and several other tumors [bib_ref] Hybridoma technology a versatile method for isolation of monoclonal antibodies, its applicability..., Parray [/bib_ref]. Despite the emergence of new promising technologies for generating mAbs, it seems that none of them was able to provoke a technological shift up to now, remaining the hybridoma-based strategies in a leadership position.
Credit author statement T.M.P. and F.B.V.C. wrote the "Hybridoma technology" section. C.B. and G.S. wrote the "B Cell Targeting (BCT)" section. B.H. and E.R.S. wrote the "Stereospecific Targeting (SST)" section. J.H.O., J.Z.M., and R.B.A. wrote the "Antibody phage display technology" section. R.B.A. wrote the "Single B cell antibody technology" section and prepared the timeline figure. All authors contributed critically to the review preparation, discussed the covered topics, and approved the final text. R.B.A. and J.Z.M wrote the discussion section, revised all the text, and answered the reviewer.
## Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
[fig] Figure 1: Timeline of important events in the generation of monoclonal antibodies. (A) Milestones related to hybridoma technology (boxes in green) and the obtainment of mAbs similar to those produced by humans (boxes in gray). (B) Landmarks related to hybridoma technology alternatives: display library techniques (boxes in orange) and single B cell antibody technology (box in purple). [/fig]
[table] Table 1: Advantages and disadvantages of technologies used to generate monoclonal antibodies. [/table]
|
Unique Case of Congenital Duodenal Atresia and a Choledochal Cyst and the Hypothesis of Their Embryological Evolution
The concomitant occurrence of duodenal atresia (DA) and a choledochal cyst (CC) has rarely been reported. Knowledge of both the presentation and management of this rare co-occurrence is imperative in avoiding potential complications and sequelae, such as biliary metaplasia. Herein we describe a female infant born at 32 weeks gestational age who was diagnosed with duodenal atresia and annular pancreas postnatally, who had subsequent findings of malrotation and a choledochal cyst, as seen from contrast imaging. Uncomplicated repair of the DA and obstruction was performed at 4 days of life. She re-presented 2 years later with non-bloody, nonbilious emesis and was found to have elevated amylase, lipase and liver enzymes. Imaging revealed dilated intra-hepatic ducts, a distended gallbladder and a large choledochal cyst. She underwent a cholecystostomy tube placement followed by a definitive choledochal cyst excision with immediate improvement following surgery and full resolution of symptoms before discharge.
# Introduction
Congenital obstruction of the duodenum can be complete or partial and due to an intrinsic or extrinsic etiology. Intrinsic obstruction is often caused by duodenal atresia (DA) and occurs in approximately 1 in 5000 to 1 in 10,000 live births. Extrinsic obstruction can be caused by intestinal malrotation, annular pancreas, gastro-duodenal duplication anomalies and pseudocysts of the pancreas and biliary tree. The incidence of congenital choledochal cysts (CC) is reported as 1 in 15,000 live births. The concomitant occurrence of a congenital choledochal cyst with duodenal atresia is very rare, with only 12 cases reported. We report a case of concomitant CC with DA in a preterm 32-week gestational age infant and postulate a hypothesis for its embryological evolution.
## Case summary
A female infant weighing 1700 g was born at 32 weeks gestational age to a 42-year-old mother. The pregnancy was complicated by gestational diabetes, with a non-immune rubella titer and prenatal ultrasound (US) revealing polyhydramnios and an echogenic bowel. Imaging after birth revealed a double-bubble sign typical for duodenal atresia.A subsequent upper gastrointestinal contrast study revealed obstruction in the first part of the duodenum, malrotation, and a choledochal cyst.A follow-up abdominal ultrasound (US) showed reversal of the expected superior mesenteric artery and superior mesenteric vein with no swirling of the vascular pedicle, as expected in the setting The postoperative course was significant for elevated direct bilirubin levels, presumably secondary to cholestasis due to prolonged parenteral nutrition. Abdominal US, on postoperative day 14, revealed a persistent, focally dilated bile duct and extrahepatic biliary ducts without the associated symptoms.Bilirubin normalized and the patient was discharged at a corrected gestational age of 38 weeks with adequate oral intake and weight gain.
Children 2020, 7, x FOR PEER REVIEW 2 of 7 pedicle, as expected in the setting of intestinal malrotation. The infant underwent an uncomplicated DA repair and Ladd's procedure at 4 days of life. Of note, an annular pancreas was visualized intraoperatively. Portal structures were not explored at this time in favor of repeat imaging of the biliary tree when the infant was older. The postoperative course was significant for elevated direct bilirubin levels, presumably secondary to cholestasis due to prolonged parenteral nutrition. Abdominal US, on postoperative day 14, revealed a persistent, focally dilated bile duct and extrahepatic biliary ducts without the associated symptoms.Bilirubin normalized and the patient was discharged at a corrected gestational age of 38 weeks with adequate oral intake and weight gain. At two years of age the patient presented to the emergency department with feeding intolerance, persistent emesis, elevated liver transaminases (AST 244, ALT 288), elevated amylase >1446 U/L and lipase 465 IU/L. Magnetic resonance imaging (MRI) revealed significantly dilated intra-and extra-hepatic ducts, a distended gallbladder and a large choledochal cyst. The physical exam was significant for ascites and jaundice. She underwent an urgent laparotomy, which found bilious ascites requiring subsequent placement of a cholecystostomy tube. A cholangiogram was then performed due to the persistent feeding intolerance. This re-demonstrated the intra-and extra-hepatic ductal dilation and incomplete biliary obstruction.She underwent a cholecystectomy, choledochal cyst excision and hepatico-duodenostomy, which she tolerated well. By post-operative day 7 the patient demonstrated substantial clinical improvement and was deemed appropriate for discharge. On a follow-up examination at 3 years of age, the patient had remained asymptomatic with normal growth and liver function. At two years of age the patient presented to the emergency department with feeding intolerance, persistent emesis, elevated liver transaminases (AST 244, ALT 288), elevated amylase >1446 U/L and lipase 465 IU/L. Magnetic resonance imaging (MRI) revealed significantly dilated intra-and extrahepatic ducts, a distended gallbladder and a large choledochal cyst. The physical exam was significant for ascites and jaundice. She underwent an urgent laparotomy, which found bilious ascites requiring subsequent placement of a cholecystostomy tube. A cholangiogram was then performed due to the persistent feeding intolerance. This re-demonstrated the intra-and extra-hepatic ductal dilation and incomplete biliary obstruction.She underwent a cholecystectomy, choledochal cyst excision and hepatico-duodenostomy, which she tolerated well. By post-operative day 7 the patient demonstrated substantial clinical improvement and was deemed appropriate for discharge. On a follow-up examination at 3 years of age, the patient had remained asymptomatic with normal growth and liver function.
# Discussion
This case demonstrates a case of antenatally formed DA and CC with concomitant anomalies requiring surgical intervention due to an obstructed hepatobiliary system 2 years after the initial findings. The embryologic origin of the duodenal atresia centers on the failure of the recanalization of the duodenum after the 7th week of gestation and can be associated with other congenital anomalies such as Down's syndrome, which is present in 25-40% of cases. Other associated
# Discussion
This case demonstrates a case of antenatally formed DA and CC with concomitant anomalies requiring surgical intervention due to an obstructed hepatobiliary system 2 years after the initial findings. The embryologic origin of the duodenal atresia centers on the failure of the recanalization of the duodenum after the 7th week of gestation and can be associated with other congenital anomalies such as Down's syndrome, which is present in 25-40% of cases. Other associated anomalies include VACTERL (vertebral, imperforate anus, cardiac, tracheo-esophageal fistula, renal anomalies and limb abnormalities), malrotation, annular pancreas, biliary tract abnormalities as well as cardiac and mandibulofacial anomalies. Two of these were described in our patient case. Though uncommon, the etiology of both DA and CC occurring during the congenital period and their embryologic association have been largely debated. One such theory by Boyden et al. proposes that a "local traffic jam" occurs between the duodenum and the entry of the hepatopancreatic and accessory pancreatic ducts during the 7th week of embryogenesis. This "traffic jam" hinders the recanalization of the duodenum and leads to a pancreaticobiliary maljunction (PBM, the congenital fusion of pancreatic and biliary ducts outside the Ampulla of Vater). In 1999, Ando et al. described PBM as the link between the development of a CC and DA. His own description of a PBM involved an abnormal fusion of the right ventral pancreatic duct with the bile/pancreatic ducts during embryogenesis leading to "an oddly shaped union that could present at birth with or without bile duct dilatation".
Similarly, pediatric cases reported in the literature since the early 1990s have included findings of PBM associated with duodenal obstruction and subsequent choledochal cysts. The first case of CC and its surgical removal was reported by . Iwai et al. followed, reporting 9 cases of CC-associated duodenal obstruction. Iwai's case review describes 9 patients with annular pancreas and either duodenal stenosis or atresia. He reported that each patient required surgical intervention a minimum of two years after the findings, due to the symptomatic development of choledochal cysts. More recently, a case report describing a 35-week gestational age male born with DA post duodenoduodenostomy, which required surgical intervention at 2 months of age due to a dilated common bile duct whose pathology was consistent with a Type I CC. In his case report, Shih et al. noted an increased presence of annular pancreas in their review of 6 reported cases of duodenal atresia or stenosis and choledochal cysts. This suggests an increased risk of duodenal obstruction and PBM is present with embryologic events involving the ventral pancreas rotation, leading one to support the association between annular pancreas and choledochal cyst formation.
This same finding was described by Iwai et al. involving a 37-week gestational age female infant born with DA repaired at birth, who presented at 4 years of age with jaundice, vomiting and was found to have a choledochal cyst and dilatation of the left intrahepatic duct requiring surgical resection. Iwai theorized the reason that no choledochal cysts were identified at the initial duodenal repair was secondary to only minimal common bile duct dilatation with a gradual increase in size over time after surgery. Sugimoto et al. similarly suggests that the common bile duct in their case, became gradually dilated after DA repair because of a "congenital pancreaticobiliary anomaly". Our patient was found to have a choledochal cyst during the initial upper GI study on day-of-life 2, implying a congenital presence. The asymptomatic nature of our patient until years later, may suggest a possible secondary insult leading to its later presentation or a gradual dilatation of the bile duct after duodenal repair as hypothesized above.
Despite the rarity of this association, the proposed mechanisms have all seemingly involved an anomalous formation of the pancreaticobiliary junction. Early case reports have nearly identical imaging findings, including an anomalous pancreaticobiliary ductal junction found during the intraoperative cholangiogram, annular pancreas and CC. The later presentation of patients described in previous cases and findings of intrahepatic and biliary duct dilatation in the setting of duodenal repair years earlier parallels our case and supports the hypothesized claims of gradual symptomatic CBD dilatation secondary to a congenital presence of PBM.
# Conclusions
The simultaneous occurrence of congenital duodenal atresia and a choledochal cyst, although rare, have common anatomical findings in all case reports throughout the years. These findings offer an underlying connection between an anomalous arrangement of the pancreaticobiliary system, duodenal obstruction, annular pancreas and the gradual dilation of bile ducts into symptomatic choledochal cysts. This emphasizes the importance of early evaluation in infants presenting with DA, to identify abnormalities such as CC and annular pancreas to prevent the development of dangerous sequelae, such as chronic cholangitis, pancreatitis, biliary metaplasia or the need for invasive surgical interventions in later childhood.
# Financial disclosures
No financial relationships relevant to this article to disclose.
Author Contributions: B.D. and R.I.K. contributed to the initial conceptual design and manuscript and final review. M.Y.B. and F.P.M. were involved in the review editing and part of the initial design and F.M. also sought and retrieved the ideal images along with its input. All authors have read and agreed to the published version of the manuscript.
# Funding:
No funding was secured for this study. |
Allele Loss and Down-Regulation of Heparanase Gene Are Associated with the Progression and Poor Prognosis of Hepatocellular Carcinoma
Objectives: The role of heparanase (HPSE) gene in cancers including hepatocellular carcinoma (HCC) is currently controversial. This study was aimed at investigating the impact of genetic alteration and expression change of HPSE on the progression and prognosis of HCC.Methods:The HPSE gene was studied in three different aspects: (1) loss of heterozygosity (LOH) by a custom SNP microarray and DNA copy number by real-time PCR; (2) mRNA level by qRT-PCR; and (3) protein expression by immunohistochemistry. The clinical significances of allele loss and expression change of HPSE were analyzed.Results: Microarray analysis showed that the average LOH frequency for 10 SNPs located within HPSE gene was 31.6%, three of which were significantly correlated with tumor grade, serum HBV-DNA level, and AFP concentration. In agreement with SNP LOH data, DNA copy number loss of HPSE was observed in 38.74% (43/111) of HCC cases. HPSE mRNA level was notably reduced in 74.1% (83/112) of tumor tissues compared with non-tumor liver tissues, which was significantly associated with DNA copy number loss, increased tumor size, and post-operative metastasis. HPSE protein level was also remarkably reduced in 66.3% (53/80) of tumor tissues, which was correlated with tumor grade. Patients with lower expression level of HPSE mRNA or protein had a significantly lower survival rate than those with higher expression. Cox regression analysis suggested that HPSE protein was an independent predictor of overall survival in HCC patients.Conclusions:The results in this study demonstrate that genetic alteration and reduction of HPSE expression are associated with tumor progression and poor prognosis of HCCs, suggesting that HPSE behaves like a tumor suppressor gene and is a potential prognostic marker for HCC patients.
# Introduction
Heparanase (HPSE) is an endoglycosidase that cleaves side chains of heparan sulfate (HS), a linear polysaccharide found on the cell surface and extracellular matrix (ECM), which plays critical roles in cell-cell and cell-matrix interactions [bib_ref] Heparanase, a potential regulator of cell-matrix interactions, Dempsey [/bib_ref]. HS also tethers a multitude of growth factors, chemokines, cytokines and enzymes to the ECM and cell surface [bib_ref] Functions of cell surface heparan sulfate proteoglycans, Bernfield [/bib_ref]. Hence, HPSE not only participates in degradation and remodeling of the ECM, but also releases HS-bound biological molecules by cleavage of HS side chains [bib_ref] Regulation, function and clinical significance of heparanase in cancer metastasis and angiogenesis, Ilan [/bib_ref]. Aside from the well-studied catalytic features of the enzyme, non-enzymatic functions of HPSE include enhancement of cell adhesion [bib_ref] Regulation, function and clinical significance of heparanase in cancer metastasis and angiogenesis, Ilan [/bib_ref] and inducing phosphorylation of p38 [bib_ref] Heparanase induces vascular endothelial growth factor expression: correlation with p38 phosphorylation levels..., Zetser [/bib_ref] , Akt [bib_ref] Heparanase induces Akt phosphorylation via a lipid raft receptor, Ben-Zaken [/bib_ref] and VEGF [bib_ref] Heparanase augments epidermal growth factor receptor phosphorylation: correlation with head and neck..., Cohen-Kaplan [/bib_ref]. Altogether, HPSE may have extensive and complex effects on wide variety of biological activities.
Due to its important and extensive biological activities, HPSE also plays a critical role in cancer development and progression. Many studies have shown that HPSE is up-regulated in a variety of primary human tumors, which is correlated with higher incidence of lymph node and distant metastasis, increased micro-vessel density and reduced post-operation survival of cancer patients [bib_ref] Heparanase expression in primary and metastatic pancreatic cancer, Koliopanos [/bib_ref] [bib_ref] Heparanase is overexpressed in lung cancer and correlates inversely with patient survival, Esti [/bib_ref]. These studies suggested that HPSE behaves like an oncogene or tumor promoter. However, other studies showed opposite results. For example, studies on clinical tumor samples indicated that the up-regulated HPSE in the cell nucleus was correlated with a favorable outcome in patients with esophageal squamous cell carcinomas [bib_ref] Localization of heparanase in esophageal cancer cells: respective roles in prognosis and..., Ohkawa [/bib_ref] , gastric carcinomas [bib_ref] Heparanase expression correlates with invasion and poor prognosis in gastric cancers, Takaoka [/bib_ref] , head and neck carcinomas [bib_ref] Heparanase localization and expression by head and neck cancer: correlation with tumor..., Doweck [/bib_ref] and lung cancer [bib_ref] Heparanase is overexpressed in lung cancer and correlates inversely with patient survival, Esti [/bib_ref]. Conflicting results were also reported in hepatocellular carcinoma (HCC) [bib_ref] Heparanase and hepatocellular carcinoma: promoter or inhibitor?, Dong [/bib_ref]. Therefore, it remains unclear whether PHSE is a suppressor or promoter of human cancers, especially for HCC [bib_ref] Heparanase and hepatocellular carcinoma: promoter or inhibitor?, Dong [/bib_ref] , which is possibly related to the extensive and complex functions of HPSE.
In a previous study, we performed a genome-wide analysis of loss of heterozygosity (LOH) in 104 HCCs with 382 microsatellite markers and found that the LOH rate of D4S2964 on 4q21.1 was as high as 50% [bib_ref] Genome-wide analyses on loss of heterozygosity in hepatocellular carcinoma in Southern China, Li [/bib_ref]. This result was consistent with Bando's report in 1999, which found this locus with 41.5% LOH in HCC [bib_ref] Identification of a 1-cM region of common deletion on 4q35 associated with..., Bando [/bib_ref] , and Nishimura's study in 2006, which reported that a deletion region containing D4S2964 occurred in 47% of HCC patients [bib_ref] Genomewide semiquantitative microsatellite analysis of human hepatocellular carcinoma: discrete mapping of smallest..., Nishimura [/bib_ref]. In addition, other genetic studies on HCC showed that chromosome 4q21, where the D4S2964 locus was located, was a common deleted region in HCC [bib_ref] Investigation of chromosomal aberrations in hepatocellular carcinoma by fluorescence in situ hybridization, Huang [/bib_ref] [bib_ref] Allelic loss of chromosome 4q21 approximately 23 associates with hepatitis B virus-related..., Yeh [/bib_ref]. All of these evidences indicated that the D4S2964 locus might contain a tumor suppressor gene(s) in HCC. In order to identify the gene(s) involved in this LOH region, we performed a fine-scale LOH analysis with 440 SNP markers located in 49 genes surrounding D4S2964 locus in 112 paired HCC and adjacent non-tumor liver tissues using a custom SNP microarray, and found a high frequency of LOH in HPSE gene [bib_ref] LOH analysis of genes around D4S2964 identifies ARD1B as a prognostic predictor..., Huang [/bib_ref]. Our results suggest that HPSE is a tumor suppressor gene based on the fact that tumor suppressor gene usually has LOH in carcinogenesis. Combined with observations from our previous study and by others, we hypothesized that HPES was a tumor suppressor gene in HCC. To support the tumor suppressor role of this gene, we further investigated the genetic alterations and expression changes of the HPSE gene in HCCs and evaluated their clinical implications. Our results show that the allele loss and reduced HPSE expression are indeed closely correlated with tumor progression and poor prognosis of HCC patients.
# Methods
## Patients and tissue samples
All 112 patients with HCC received hepatectomy between 2004 and 2007 at the Department of Hepatobiliary Oncology at the Sun Yat-Sen University Cancer Center. HCC and corresponding non-tumor liver tissues were collected at the time of hepatectomy.
There were about 1563 patients with HCC undergoing hepatectomy between 2004 and 2007 in the cancer center. However, only patients who met all of the following criteria were included in this study: 1) did not receive any other anti-cancer therapies before the surgery, such as chemoembolization, chemotherapy, etc; 2) underwent curative resection for complete removal of the tumor without macroscopic evidence of residual cancer tissues; 3) diagnosed with HCC by pathology; 4) had frozen tumor tissues available in the Tissue Bank of the cancer center. The 112 HCC patients included 94 males and 18 females with a median age of 45.5 years (range, 13-72 years). No patients had extrahepatic metastasis when they underwent hepatectomy. Postoperative metastasis meant the extrahepatic metastasis to distant organs and recurrence was the intrahepatic recurrence after hepatectomy. The patients were followed up every 2 to 4 months in the first two years, and thereafter annually, with a median follow-up time of 36 months. Patients who returned to the hospital were detected by Computed tomography (CT) or Ultrasound B-mode scanner and AFP test in the follow-up or contacted by mail or call. The primary endpoint was overall survival from the date of hepatectomy to patient death or the last follow-up. Written informed consent was obtained from all the participants or guardians on the behalf of the children participants. This study was reviewed and approved by the Committee for the Conduct of Human Research of the Sun Yat-Sen University Cancer Center.
The fresh samples were immediately immersed in RNAlater (Ambion, Inc., USA) after surgical resection, and stored at 4uC overnight to allow thorough penetration of the tissues. The samples were then frozen at 280uC until RNA and DNA extraction. Total RNA and DNA were extracted sequentially using TRIzol reagent (Invitrogen, USA) according to the manufacturer's instructions.
## Snp genotyping by microarray and data processing
The procedure of selecting 440 SNPs in 49 genes was described by Huang et al [bib_ref] LOH analysis of genes around D4S2964 identifies ARD1B as a prognostic predictor..., Huang [/bib_ref]. The polymorphic sites of the SNPs were all transition variations (A/G or C/T) to facilitate microarray analysis by using two fluorescent colors (Cy-3 and Cy-5). The SNP microarray genotyping system described by Wang HY et al [bib_ref] A genotyping system capable of simultaneously analyzing .1000 single nucleotide polymorphisms in..., Wang [/bib_ref] was used with minor modifications. The microarray consisting of probes for the 440 SNPs was printed on cleaned slides by a SmartArray TM -136 printer, hybridized with the amplified singlestrand DNAs containing the SNP sites in a BioMixer TM II hybridization oven and scanned by a LuxScan TM -10 K Scanner (CapitalBio Inc., Beijing, China). Scanned images of microarrays were analyzed using GenPix Pro 6.0 software (Axon Instruments, Foster City, CA, USA). The hybridization signals were presented as median intensity for each feature. Data normalization, lowsignal filtering, background subtraction and genotyping were processed by our developed SNP genotype analysis program called ''AccuTyping'' [bib_ref] AccuTyping: new algorithms for automated analysis of data from highthroughput genotyping with..., Hu [/bib_ref]. Genotype call for each SNP was determined based on the signal intensity ratio of the two fluorescences [bib_ref] LOH analysis of genes around D4S2964 identifies ARD1B as a prognostic predictor..., Huang [/bib_ref].
# Loh analysis
The purpose of SNP genotyping in this study was for LOH analysis. If a SNP was typed as heterozygous in non-tumor tissue of a patient, this SNP in the patient was defined as informative SNP. Only informative SNPs were included in LOH analysis. SNP LOH was defined when a SNP was heterozygous in a non-tumor liver tissue and homozygous in the corresponding HCC tissue. If a SNP was heterozygous in both paired non-tumor liver and tumor tissue, it was defined as retention. Frequency of SNP LOH for each SNP site was equal to the ratio of the number of cases with the SNP LOH to the number of cases with the informative SNPs for a SNP in all of the cases. If any one or more SNPs had LOH in the HPSE gene in one case, it was defined as Gene LOH in this case. Informative case for a gene was defined as a case with any informative SNPs in this gene. Frequency of Gene LOH for a gene was equal to the ratio of the number of cases with the Gene LOH to the number of cases with informative cases.
## Quantitative pcr (qpcr)
HPSE DNA copy number quantification by qPCR described by Liu et al [bib_ref] Interactions between MDM2 and TP53 Genetic Alterations, and Their Impact on Response..., Liu [/bib_ref] was performed using Platinum SYBR Green qPCR SuperMix-UDG reagents (Invitrogen, USA) in Applied Biosystems PRISM 7900HT instruments. Briefly, a 142-bp amplicon of HPSE was amplified with a pair of primers (forward: 59-GTT TGG CTT TGA GCT TTG CTT-39 and reverse: 59-ATC GTG CTT GCT GCT TTT TAT C-39). Another 149-bp amplicon of the LINE1 sequence was used as an internal control [bib_ref] Gene copy number for epidermal growth factor receptor (EGFR) and clinical response..., Moroni [/bib_ref]. Real-time qPCR was performed in a 15-ml reaction mixture containing 10 ng genomic DNA, 0.25 ml of 10 mM each primer (forward and reverse primer) and 7.5 ml of 26 SYBR Green PCR Master Mix. The PCR was cycled 45 times at 95uC for 30 s, 60uC for 1 min, after preheating at 95uC for 10 min. All reactions were run in duplicate. The HPSE copy number was normalized to that of LINE1 to obtain a ratio (R1) for each tissue sample. Sample R1 was then normalized to the average R1 of 10 normal liver samples (the samples were obtained from the normal liver tissues of the resected edges of hepatoadenomas, none of them had hepatitis B) to get the second ratio (R2), and the HPSE copy number for each tissue was calculated by doubling R2. Based on the values of normal liver tissues, copy number change between 1.76 and 2.24 would be considered as normal, less than 1.76 defined as a DNA copy number loss and greater than 2.24 defined as a DNA copy number gain. Quantitative RT-PCR (qRT-PCR)
Reverse transcription reaction was carried out with MMLV reverse transcription kit according to the manufacturer's protocol (Promega, USA). Two micrograms of total RNA were used for reverse-transcription reaction to generate cDNA at 42uC for 60 min. Then qPCR assay was performed with Platinum SYBR Green qPCR SuperMix-UDG reagents (Invitrogen, USA) in Applied Biosystems PRISM 7900HT instruments according to the manufacture's protocol. The reactions with 0.5 ml of cDNA, 7.5 ml of 26 SYBR Green qPCR SuperMix, 0.25 ml each of 10 mM forward and reverse primers at 15 ml volume were carried out in a 96-well plate at 95uC for 10 min, followed by 40 cycles of 95uC for 30 sec and 60uC for 1 min. Each sample was run in duplicate. The primers for 11 genes with higher LOH and a housekeeping gene, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), were listed in [fig_ref] Table 1: LOH frequency of 10 SNPs in HPSE gene at D4S2964 on Chromosome... [/fig_ref]. The comparative Ct method (DDCt) was used for quantification of the HPSE gene expression and relative quantification (RQ) was calculated as 2 2DDCt .
## Immunohistochemistry (ihc)
Formalin-fixed and paraffin-embedded tissues were cut into 4mm sections and mounted onto the polylysine-coated slides. After treated with routine procedures, the sections were incubated for 1 hour at 37uC with rabbit anti-HPSE polyclonal antibody (Santa Cruz Biotechnology, Inc., USA) diluted 1:200 in blocking solution followed by another reaction with HRP-conjugated secondary antibody (ChemMate Envision Detection Kit, Dako) according to the manufacturer's instruction. After washes, color was developed with 3,39-diaminobenzidine tetrahydrochloride (DAB), and then all of the sections were counterstained with hematoxylin. For negative controls, tissue sections were incubated without anti-HPSE antibody under the same experimental conditions. HPSE staining was scored according to its intensity (0, no staining; 1, weak staining; 2, moderate staining; 3, strong staining) and the percentage of tumor cells that were stained (0, ,5% of tumor cells stained; 1, 5%-25% of tumor cells stained; 2, 25-50% of tumor cells stained; 3, .50% of tumor cells stained). The final expression score was calculated from 'intensity score' multiplied by 'percentage': I stands for scores 0-1, II for scores 2-3, III for scores 4-6 and IV for scores .6. For statistical analysis, we combined the cases scored as I and II (low score) to compare with the cases with scored as III and IV (high score). Immunohistochemistry analysis was performed on 80 available cases of the 112 patients.
## The process for identification of genes that might be affected by loh
In order to identify the genes involved in the LOH of D4S2964, all of 49 genes in this locus were detected by SNP microarray and analyzed by LOH method. In this step, the LOH frequency for all of the genes in the locus would be determined. If a gene has LOH, the mRNA level of this gene usually will be reduced. Thus, the mRNA level of the genes in the top-10 list of LOH and another gene were measured by qRT-PCR in 50 pairs of HCC and corresponding non-tumor liver tissues, and the relationships between the mRNA levels of the 11 genes and patient's survival were analyzed by Kaplan-Meier curve and Log-rank test. The genes that were significantly associated with poor survival (P,0.1) were further investigated by qRT-PCR and Kaplan-Meier curve with Log-rank test in all of the 112 HCC cases. If any gene still was
# Statistical analysis
The chi-square test, Fisher's exact test and Mann-Whitney U test were used to analyze the correlations between clinicopathological features with the HPSE LOH status, DNA copy number or expressions. The overall survivals, cumulative metastasis rates and cumulative recurrence rates in different groups were estimated using the Kaplan-Meier analysis and Log-rank test. In order to identify the parameters (clinicopathological features and HPSE expressions) that might affect the survival or metastasis or recurrence, we first performed univariate Cox regression analysis on all of parameters. Then, those parameters would be moved into the multivariate Cox regression analysis with an Enter procedure if they displayed statistical significance (P,0.05) in the univariate analysis. The probability for stepwise variable selection was set at 0.05 for entry and 0.10 for removal. The SPSS version 16.0 software package and GraphPad Prism were used for the statistical analysis and data plotting.
# Results
## Analysis of snp loh and mrna level suggests that hpse is involved in the loh of d4s2964
A total of 13,322 genotypes out of 49,280 SNPs were determined as heterozygous in 112 non-tumor liver tissues for the 440 SNPs in the 49 genes surrounding D4S2964 locus, and 25 genotypes of them were verified by sequencing [bib_ref] LOH analysis of genes around D4S2964 identifies ARD1B as a prognostic predictor..., Huang [/bib_ref]. This microarray data and related information had been deposited into Gene Expression Omnibus (GEO) database (http://www.ncbi. nlm.nih.gov/geo/). The accession number is GSE36277. In the LOH analysis of SNPs, only cases (informative) with heterozygous SNP in non-tumor liver tissue were included. In the present study, the number of informative cases for each SNP ranged from 0 to 65 with an average of 24.2. Sixty-three SNPs located in 32 genes had a LOH frequency greater than 30%, and LOH of the 49 genes had been reported elsewhere [bib_ref] LOH analysis of genes around D4S2964 identifies ARD1B as a prognostic predictor..., Huang [/bib_ref].
In general, if a gene had LOH, its mRNA expression would be reduced. Thus, we selected 10 genes (PPEF2, PRDM8, SDAD1, CXCL9, CCDC158, PRKG2, HPSE, AGPAT9, HELQ, NUP54) with the highest frequency of Gene LOH [fig_ref] Table 2: Correlation of HPSE expression with clinicopathological features [/fig_ref] and another gene (CCNG2) based literature reports, which suggested CCNG2 as a tumor suppressor, to detect mRNA expression level in 50 pairs of HCC and the matched liver samples from the 112 cases. The result indicated that the expression level of most genes was much lower in tumors than in the matched non-tumor liver tissues [fig_ref] Table 2: Correlation of HPSE expression with clinicopathological features [/fig_ref]. In order to identify key gene(s) in this LOH region, we conducted a correlation analysis on the mRNA expression level of the 11 genes and patients' prognosis. The result indicated that mRNA expression level for five (PPEF2, CCDC158, CCNG2, HPSE, AGPAT9) of these genes was significantly associated with patients' prognosis in the 50 HCC cases [fig_ref] Table 2: Correlation of HPSE expression with clinicopathological features [/fig_ref]. Then, mRNA expression level for the five genes was further measured in the others of the 112 HCC patients, and survival analysis of the 112 cases was performed on the mRNA expression levels. The result showed that the low expression levels of four out of the five genes were associated with poor survival of HCC patients [fig_ref] Table 2: Correlation of HPSE expression with clinicopathological features [/fig_ref]. Based on the above results, we decided to further investigate the clinical significances and/or functions of the four genes (CCDC158, CCNG2, HPSE and AGPAT9). Some of this work are still ongoing. HPSE was one of the four genes. These data implied that HPSE is a tumor suppressor gene, which was contradictory with the notion that HPSE is an oncogene or tumor promoter. Therefore, it is necessary to further investigate this gene and its clinical significance in HCC because it should help to clarify its clinical significance, which is of obvious importance to the medical field, especially to HCC biology and therapy.
The high frequency of SNP LOH in HPSE is associated with serum AFP, HBV-DNA and tumor grade Of the 12 SNPs located in the HPSE gene, 10 were genotyped as heterozygous in 10 or more cases (informative cases) among the 112 samples. The average LOH frequency for the 10 SNPs was 31.6% with a range from 14.3% (5/35 for rs9991877) to 48.3% (14/29 for rs4364254) [fig_ref] Table 1: LOH frequency of 10 SNPs in HPSE gene at D4S2964 on Chromosome... [/fig_ref]. Correlation analysis between the LOH frequency of each SNP in HPSE gene and clinical features (serum AFP level, serum HBV-DNA level, HBsAg status, HBeAg status, tumor grade, tumor size, metastasis and recurrence) revealed that LOHs at three SNPs, rs4364254, rs6535458 and rs4568236, were significantly correlated with serum AFP level, tumor grade and serum HBV-DNA level (all P,0.05), respectively. Out of 84 informative cases, 41 (48.8%) had Gene LOH (see LOH analysis in Method). However, no significant correlation was found between the Gene LOH and clinicopathological features.
## Hpse dna copy number loss in hcc is correlated with snp loh and mrna level
Real-time PCR is a common and efficient method for measuring DNA copy number. In the present study, we employed real-time PCR to detect the HPSE DNA copy number alteration. In 111 HCCs with successful detection, the median of DNA copy number was 1.84 with a range from 1.12 to 2.64. DNA copy number loss was observed in 38.74% (43/111) of tumor tissues, and copy number gain in 12.61% (14/111) [fig_ref] Figure 1: HPSE DNA copy number variation and correlation with mRNA expression in HCC [/fig_ref]. The correlation analysis showed that DNA copy number and Gene LOH were significantly associated in 84 cases (r = 0.242, P,0.028). To explore the correlation between HPSE DNA copy number and mRNA expression, HPSE mRNA was quantified by real-time RT-PCR. The result showed a small but significant correlation between HPSE copy number and mRNA levels in tumor tissues (r = 0.23, P = 0.013, [fig_ref] Figure 1: HPSE DNA copy number variation and correlation with mRNA expression in HCC [/fig_ref]. The HPSE mRNA levels in HCC tissues with DNA copy number less than 2 were significantly lower than those with DNA copy number more than 2 (P = 0.0055, [fig_ref] Figure 1: HPSE DNA copy number variation and correlation with mRNA expression in HCC [/fig_ref].
## Reduced hpse mrna level is related with tumor size and metastasis in hcc
We next compared mRNA expression levels between paired HCC samples and non-tumor liver tissues, and found that the HPSE mRNA level in HCC was significantly lower than that in the paired non-tumor liver tissues (P = 0.0005, [fig_ref] Figure 2: Relative expression level of HPSE mRNA in HCC samples with different clinical... [/fig_ref]. The mRNA level of HPSE in 83 (83/112 = 74.1%) cases was reduced in tumor tissues compared with non-tumor liver tissues. To investigate the significance of the reduced HPSE mRNA level in HCC progression, we performed a correlation analysis with clinicopathological parameters. The result indicated that HPSE mRNA expression was significantly reduced in tumors with greater than 5 cm compared with tumors equal to or smaller than 5 cm (P = 0.0015, [fig_ref] Figure 2: Relative expression level of HPSE mRNA in HCC samples with different clinical... [/fig_ref] , and in patients with post-operative metastasis compared with patients without (P = 0.0336, [fig_ref] Figure 2: Relative expression level of HPSE mRNA in HCC samples with different clinical... [/fig_ref]. No correlation was found between HPSE mRNA level and other clinical features by this analysis.
## Down-regulation of hpse protein in hcc is correlated with tumor grade
In general, the reduced mRNA expression of a gene will cause down-regulation of corresponding protein. To confirm this situation in HCC, HPSE protein in 80 cases of the 112 HCCs were detected by IHC. The result exhibited a lower staining score of HPSE protein in 53 (66.3%) HCC tissues than in their matched non-tumor adjacent tissues [fig_ref] Figure 3: Immunohistochemical detection of the HPSE protein expression in HCC [/fig_ref]. Interestingly, HPSE protein was detected in both the cytoplasm and nucleus of liver cancer cell [fig_ref] Figure 3: Immunohistochemical detection of the HPSE protein expression in HCC [/fig_ref] , which is in sharp contrast to previous studies that found HPSE protein exclusively in either the cytoplasm or nucleus of cancer cells [bib_ref] Heparanase is overexpressed in lung cancer and correlates inversely with patient survival, Esti [/bib_ref] [bib_ref] Localization of heparanase in esophageal cancer cells: respective roles in prognosis and..., Ohkawa [/bib_ref] [bib_ref] Heparanase expression correlates with invasion and poor prognosis in gastric cancers, Takaoka [/bib_ref] [bib_ref] Heparanase localization and expression by head and neck cancer: correlation with tumor..., Doweck [/bib_ref]. The x 2 test indicated that the staining score of HPSE protein was significantly correlated with tumor grade (P = 0.012, [fig_ref] Table 2: Correlation of HPSE expression with clinicopathological features [/fig_ref] , and that the correlation between HPSE protein expression and recurrence status was marginally significant as well (P = 0.056, [fig_ref] Table 2: Correlation of HPSE expression with clinicopathological features [/fig_ref] , which suggests that HPSE expression might be significantly correlated with recurrence if the sample size was big enough. No correlation was found between HPSE protein staining score and other clinical features by this analysis.
## The impact of hpse mrna and protein levels on recurrence and metastasis in hcc patients
The above analysis showed that there was a significant or marginally significant correlation between HPSE expression and recurrence or metastasis in chi-square test. We further explored whether HPSE expression (mRNA and protein levels) was a predictor for recurrence or metastasis in HCC patients using Kaplan-Meier curve, Log-rank test and Cox regression analysis. First, we conducted Kaplan-Meier curve analysis and Log-rank test, which showed that the group with lower mRNA level of HPSE had significantly higher cumulative metastasis rate and recurrence rate than the group with high expression level (Logrank test, P = 0.033 and P = 0.023, respectively, [fig_ref] Figure 4: Cumulative metastasis rate and recurrence rate of HCC patients with high and... [/fig_ref] ; Similarly, the group with lower protein level of HPSE had significantly higher cumulative recurrence rate than the group with high expression level (Log-rank test, P = 0.035, [fig_ref] Figure 4: Cumulative metastasis rate and recurrence rate of HCC patients with high and... [/fig_ref]. However, no difference was found in the cumulative metastasis rates between the groups with low or high level of HPSE protein (Log-rank test, P = 0.199, [fig_ref] Figure 4: Cumulative metastasis rate and recurrence rate of HCC patients with high and... [/fig_ref] , which is likely to be due to the relatively small number of metastasis (only 10 cases) and sample size (only 80 cases) analyzed by IHC. Next, Cox regression analysis was conducted to find the independent predictors for recurrence and metastasis in HCC patients. The univariate Cox regression analysis on metastasis indicated that only HPSE mRNA level was significantly associated with metastasis (P = 0.045, [fig_ref] Table 3: Multivariate Cox regression analysis of impacts of variables affecting recurrence [/fig_ref] , and that HPSE mRNA level, protein expression score and serum AFP concentration were significantly associated with recurrence [fig_ref] Table 4: Multivariate Cox regression analysis of variables affecting early recurrence [/fig_ref]. With an Enter procedure, multivariate Cox regression analysis suggested that HPSE mRNA level and serum AFP were the independent predictors for recurrence in HCC (P = 0.031 and P = 0.046, respectively, [fig_ref] Table 3: Multivariate Cox regression analysis of impacts of variables affecting recurrence [/fig_ref].
Since the causes and mechanisms of early recurrence (within two years after hepatectomy) and late recurrence (in more than two years after hepatectomy) might be very different, we tested whether HPSE mRNA expression impacted the early or late recurrence in HCC patients. Kaplan-Meier plots and Log-rank test indicated that the patients with lower HPSE mRNA level had higher early recurrence rate (Log-rank test, P = 0.025, [fig_ref] Figure 5: Cumulative early and late recurrence rate of HCC patients with high or... [/fig_ref] , but no difference in the late recurrence rates between the groups with low or high level of HPSE mRNA [fig_ref] Figure 5: Cumulative early and late recurrence rate of HCC patients with high or... [/fig_ref]. Both univariate and multivariate Cox regression models revealed that HPSE mRNA level and serum AFP were the significant predictors for early recurrence in HCC (P = 0.031 and P = 0.016, respectively in univariate analysis, [fig_ref] Table 5: Multivariate Cox regression analysis of variables affecting overall survival [/fig_ref] ; P = 0.020 and P = 0.012, respectively in multivariate analysis, [fig_ref] Table 4: Multivariate Cox regression analysis of variables affecting early recurrence [/fig_ref] , and no predictor was identified for late recurrence in HCC (all P.0.05, [fig_ref] Table 6: Multivariate Cox regression model analysis of variables affecting overall survival in the... [/fig_ref].
## The impact of hpse protein and mrna levels on hcc patients' survival
To evaluate the importance of HPSE expression in patients' survival, we carried out the Kaplan-Meier analysis and Log-rank test on the gene's expression level. The median expression value of mRNA was used as the cut-off to separate patients into groups with high or low levels. Based on the protein expression score, patients also were divided into groups with high (score III or IV) or low level (score I or II). The analysis showed that the group with lower HPSE mRNA or protein level had significantly worse overall survival rate than did the group with high expression level (Log-rank test, P = 0.009, P = 0.024, for mRNA and protein, respectively; see [fig_ref] Figure 6: Overall survival curves of HCC patients with high or low expression levels... [/fig_ref]. This indicates that the reduced level of mRNA or protein of HPSE is strongly correlated with poor survival of HCC patients. We next performed univariate Cox regression analysis to find all of the factors with potential prognostic significance in HCC patients. The analysis demonstrated that mRNA level and protein staining score of HPSE were both significantly associated with overall survival rate [fig_ref] Table S7: Univariate Cox regression analysis of variables affecting overall survival [/fig_ref]. The univariate analysis also indicated that clinical parameters including tumor size, tumor grade, number of tumor nodules and serum AFP were significantly associated with the overall survival rate [fig_ref] Table S7: Univariate Cox regression analysis of variables affecting overall survival [/fig_ref]. Finally, multivariate Cox regression analysis was used to identify the independent prognostic predictors. With an Enter procedure, multivariate Cox regression analysis was implemented on all of the variables that showed prognostic significances in the univariate analysis. The result demonstrated that HPSE protein staining score, tumor size and nodule number were the independent predictors for overall survival in HCC (P = 0.041, P = 0.019, P = 0.002, respectively, [fig_ref] Table 5: Multivariate Cox regression analysis of variables affecting overall survival [/fig_ref].
## Stratified analysis of the role of hpse mrna and protein levels on survival in subgroups of bclc stages
Survival analysis showed that HPSE mRNA and protein levels were associated with survival of HCC patients, but only protein expression was the independent predictor for HCC patients. To explore whether the HPSE mRNA or protein had different roles in the survivals of early stage and late stage of HCC, we performed stratified analysis for patients with BCLC stage 0-A and stage B-D (BCLC Staging system, 2010). Kaplan-Meier curve analysis and Log-rank test showed that in the subgroup of BCLC stage 0-A, patients with lower level of HPSE mRNA but not protein level had significantly worse overall survival rate than those with high expression level (Log-rank test, P = 0.008, P = 0.214, for mRNA and protein, respectively; see [fig_ref] Figure 7: Overall survival curves of HCC patients with high or low expression levels... [/fig_ref] and 7B); and in the subgroup of stage B-D, patients with lower level of HPSE protein but not mRNA level had significantly worse overall survival rate than those with high expression level (Log-rank test, P = 0.502, P = 0.004, for mRNA and protein, respectively; see [fig_ref] Figure 7: Overall survival curves of HCC patients with high or low expression levels... [/fig_ref] and 7D).
Cox regression analysis was next conducted to identify independent predictors for overall survival in the subgroup analysis. The univariate Cox regression analysis indicated that only HPSE mRNA level was significantly associated with overall survival in the subgroup of BCLC stage 0-A (P = 0.013, [fig_ref] Table S8: Univariate Cox regression analysis of variables affecting overall survival in the subgroup... [/fig_ref] , and that HPSE protein expression score and tumor grade were significantly associated with overall survival in the subgroup of BCLC stage B-D (P = 0.011 and P = 0.022, respectively, Table S9). With an Enter procedure, multivariate Cox regression analysis suggested that only HPSE protein expression score was the independent predictor for overall survival in the subgroup of BCLC stage B-D (P = 0.020, [fig_ref] Table 6: Multivariate Cox regression model analysis of variables affecting overall survival in the... [/fig_ref].
# Discussion
As the mentioned above, we and others found a high frequency of LOH at D4S2964 in HCC. To identify the gene(s) involved, we performed another LOH analysis on 440 SNPs in 49 genes surrounding the D4S2964 locus in HCC by a custom SNP microarray and found that several genes had higher LOH rate [bib_ref] LOH analysis of genes around D4S2964 identifies ARD1B as a prognostic predictor..., Huang [/bib_ref]. In the present study, our result displayed a high frequency of LOH (average 31.6%) at 10 SNP loci located in the HPSE gene [fig_ref] Table 1: LOH frequency of 10 SNPs in HPSE gene at D4S2964 on Chromosome... [/fig_ref]. Further analyses showed that LOH at 3 SNPs in HPSE was correlated with serum AFP, HBV-DNA level and tumor stage. At the same time, we found HPSE DNA deletion in 38.74% HCC patients using real-time PCR. As expected, the reduced expression levels of HPSE mRNA and protein were observed in HCC tissues and associated with poor outcomes in HCC patients. Therefore, those results suggest that HPSE is affected by the LOH at D4S2964 and is a potential tumor suppressor gene.
The evidences obtained from this study are consistent with previous reports by other investigators. In 2002, Ikeguchi et al found that the mRNA level of HPSE detected in HCC was reduced when compared with corresponding non-tumor liver tissues [bib_ref] Quantitative analysis of heparanase messenger RNA expression in hepatocellular carcinoma, Ikeguchi [/bib_ref]. These authors further reported that decreased HPSE gene expression was significantly associated with a poor diseasefree survival of the HCC patients [bib_ref] Heparanase gene expression and its correlation with spontaneous apoptosis in hepatocytes of..., Ikeguchi [/bib_ref]. A significant positive correlation between HPSE expression level and apoptotic index of hepatocytes was also observed [bib_ref] Heparanase gene expression and its correlation with spontaneous apoptosis in hepatocytes of..., Ikeguchi [/bib_ref] , which suggested that high level of HPSE expression might be associated with increased apoptosis of liver cells. In addition, they reported similar situations in esophageal squamous cell carcinoma [bib_ref] Quantitative analysis of heparanase gene expression in esophageal squamous cell carcinoma, Ikeguchi [/bib_ref]. These observations are in agreement with our present study, indicating that HPSE gene is a potential tumor suppressor gene in HCC.
On the other hand, several other studies showed the mRNA expression levels of HPSE gene were significantly higher in primary HCC tissues compared with the non-cancer tissues and/ or normal controls, and the increased expression of HPSE mRNA was correlated with larger tumor size, poor tumor grade, portal vein invasion, tumor microvessel density, and post-operative metastasis in HCC [bib_ref] The clinicopathological significance of heparanase and basic fibroblast growth factor expressions in..., El-Assal [/bib_ref] [bib_ref] Heparanase expression in hepatocellular carcinoma and the cirrhotic liver, Xiao [/bib_ref]. These reports are inconsistent with the results from Ikeguchi and the present study. However, the patient number in most of these studies was significantly less (fewer than 56 cases in contrast to 112 cases in our study). Therefore, our result is deemed to be more reliable.
With respect to the conflicting report on HPSE in HCC, Dong and Wu had discussed possible reasons for the contradictory results obtained from different HCC studies [bib_ref] Heparanase and hepatocellular carcinoma: promoter or inhibitor?, Dong [/bib_ref]. They speculated that the inconsistent results from different reports on HPSE in HCC might be resulted from the different protein subcellular locations (nucleus, cytoplasm and cell surface), expression levels (low, moderate and high) and activities (enzymatic and nonenzymatic activities) of HPSE because it could have different effects in different sites, expression levels and activities [bib_ref] Heparanase and hepatocellular carcinoma: promoter or inhibitor?, Dong [/bib_ref]. It should be mentioned here that HPSE protein was located in both the cytoplasm and nucleus of liver cancer cell in the present study, and the protein in the nucleus was reported to be favorable to the outcome of cancer [bib_ref] Localization of heparanase in esophageal cancer cells: respective roles in prognosis and..., Ohkawa [/bib_ref] [bib_ref] Heparanase localization and expression by head and neck cancer: correlation with tumor..., Doweck [/bib_ref]. We also noticed that 12.6% of patients had HPSE DNA copy number gain (.2.24), which suggested that HPSE gene not only was deleted, but also occasionally amplified. However, the clinical significance of the small number of DNA copy number gain was not clear in this study. In addition, HPSE transcript has different splice variants, which possessed different functions from the wild one [bib_ref] Alternatively spliced Spalax heparanase inhibits extracellular matrix degradation, tumor growth, and metastasis, Nasser [/bib_ref] [bib_ref] Adaptive evolution of heparanase in hypoxia-tolerant Spalax: gene cloning and identification of..., Nasser [/bib_ref]. For example, HPSE splice variant 6 from Spalax showed inhibition of HS degradation, suppression of tumor growth and no enzymic activity, while splice variant 7 could enhanced tumor growth but had no enzymatic activity [bib_ref] Alternatively spliced Spalax heparanase inhibits extracellular matrix degradation, tumor growth, and metastasis, Nasser [/bib_ref]. Conceivably, one of the reasons causing the conflicting results might be that different splice variants of HPES were detected in different reports. These factors may explain the contrary reports on HPSE gene in HCC and other cancers. For more than twenty years, HPSE was thought to function as an oncogene and regarded as a therapeutic target [bib_ref] Heparanase: a target for drug discovery in cancer and inflammation, Mckenzie [/bib_ref]. Consequently, many HPSE inhibitors including chemically modified natural products, small molecule inhibitors, sugar and Neutralizing antibodies, have been developed with some tested in clinical trials [bib_ref] Heparanase: a target for drug discovery in cancer and inflammation, Mckenzie [/bib_ref] [bib_ref] Development of heparanase inhibitors for anti-cancer therapy, Miao [/bib_ref]. However, none of the HPSE inhibitors has showed favorable clinical outcome. For example, PI-88, a sugar inhibitor for HPSE, had been tested in phase I/II clinical trials for melanoma, prostate cancer and HCC [bib_ref] Heparanase inhibitor PI-88 as adjuvant therapy for hepatocellular carcinoma after curative resection:..., Liu [/bib_ref] [bib_ref] A phase II study of the heparanase inhibitor PI-88 in patients with..., Lewis [/bib_ref] [bib_ref] Multicentre phase I/II study of PI-88, a heparanase inhibitor in combination with..., Khasraw [/bib_ref] , but no significant improvement for patients had been observed. In fact, HCC patients with higher PI-88 dose (250 mg/day) even had worse outcome in the phase II clinical trial [bib_ref] Heparanase inhibitor PI-88 as adjuvant therapy for hepatocellular carcinoma after curative resection:..., Liu [/bib_ref].
The poor therapeutic effects with HPSE inhibitors beg for a fresh thinking on the role of HPSE in tumor progression. Many studies pointed to the fact that HPSE is not simply an oncogene because of its complex biological functions. More importantly, Zetser et al observed that high HPSE expression level could inhibit tumor growth [bib_ref] Heparanase affects adhesive and tumorigenic potential of human glioma cells, Zetser [/bib_ref]. Nobuhisa et al proposed that translocating HPSE protein into cell nucleus could be as a new strategy for anti-cancer therapy, which was based on their finding that HPSE protein localized in the nucleus of HL-60 cell nucleus caused cell differentiation [bib_ref] Translocation of heparanase into nucleus results in cell differentiation, Nobuhisa [/bib_ref]. The evidences from these studies and our present study support our hypothesis that HPES can serve as a tumor suppressor gene. Accordingly, it is prudent that caution be taken for HSPE inhibitors as anticancer therapeutics until more clear evidence is available regarding HPSE's biological functions and clinical significances.
## Supporting information
Table S1
The primer sequences for quantitative RT-PCR assay. (DOC)
[fig] Figure 1: HPSE DNA copy number variation and correlation with mRNA expression in HCC. (a), HPSE DNA copy number in 111 HCC samples. The normal range was 1.76-2.24, which was normalized by the normal liver tissues. (b), Correlation between HPSE DNA copy number and mRNA expression levels in 111 HCC samples, r = 0.242, P,0.028. (c), The mRNA expression level of HPSE in HCC patients with HPSE DNA copy number ,2 was significantly lower than that in those with DNA copy number .2, P = 0.005. doi:10.1371/journal.pone.0044061.g001 [/fig]
[fig] Figure 2: Relative expression level of HPSE mRNA in HCC samples with different clinical features. (a), mRNA level in HCC (Tumor) was lower than that in corresponding non-tumor tissues (NAT), P = 0.0005. (b), mRNA level in tumors with size .5 cm was lower than that in tumors with size #5 cm, P = 0.0015. (c), mRNA in patients with post-operative metastasis was lower than that in those without, P = 0.0336. doi:10.1371/journal.pone.0044061.g002 correlated with poor survival of the 112 patients, it would be selected for further study. [/fig]
[fig] Figure 3: Immunohistochemical detection of the HPSE protein expression in HCC. The tissue sections were developed with DAB and counterstained by hematoxylin. (a), Well differentiated hepatocellular carcinoma (2006), HPSE staining was scored as positive 3; (b), moderately differentiated hepatocellular carcinoma (2006), scored as positive 2; (c), poorly differentiated hepatocellular carcinoma (2006), scored as positive 1; (d), Immunostaining of HCC and adjacent surrounding non-cancerous liver tissues, HCC part (left and below) with weak staining, and the adjacent liver tissue with strong staining (2006). doi:10.1371/journal.pone.0044061.g003 [/fig]
[fig] Figure 4: Cumulative metastasis rate and recurrence rate of HCC patients with high and low expression levels of HPSE analyzed by Kaplan-Meier Curve. (a), Cumulative metastasis curves for patients with high or low HPSE mRNA levels (Log-rank test, P = 0.033). (b), Cumulative recurrence curves for patients with high or low HPSE mRNA levels (Log-rank test, P = 0.023). (c), Cumulative metastasis curves for patients with high or low HPSE protein staining scores (Log-rank test, P = 0.199). (d), Cumulative recurrence curves for patients with high or low HPSE protein staining scores (Log-rank test, P = 0.35). doi:10.1371/journal.pone.0044061.g004 [/fig]
[fig] Figure 5: Cumulative early and late recurrence rate of HCC patients with high or low expression levels of HPSE by Kaplan-Meier Curve. (a), Cumulative early recurrence (within 2 years) curves for patients with high or low HPSE mRNA levels (Log-rank test, P = 0.025). (b), Cumulative late recurrence (in more than 2 years) curves for patients with high or low HPSE mRNA levels (Log-rank test, P = 0.582). doi:10.1371/journal.pone.0044061.g005 [/fig]
[fig] Figure 6: Overall survival curves of HCC patients with high or low expression levels of HPSE by Kaplan-Meier Curve. (a), Overall survival curves by HPSE mRNA levels (Log-rank test, P = 0.009). (b), Overall survival curves by HPSE protein staining scores (Log-rank test, P = 0.024). doi:10.1371/journal.pone.0044061.g006 [/fig]
[fig] Figure 7: Overall survival curves of HCC patients with high or low expression levels of HPSE in the subgroups of BCLC stages. (a), Overall survival curves by HPSE mRNA levels in the subgroup of BCLC stage 0,A (Log-rank test, P = 0.008). (b), Overall survival curves by HPSE protein staining scores in the subgroup of BCLC stage 0,A (Log-rank test, P = 0.214). (c), Overall survival curves by HPSE mRNA levels in the subgroup of BCLC stage B,C,D (Log-rank test, P = 0.502). (d), Overall survival curves by HPSE protein staining scores in the subgroup of BCLC stage B,C,D (Log-rank test, P = 0.004). doi:10.1371/journal.pone.0044061.g007 [/fig]
[table] Table 1: LOH frequency of 10 SNPs in HPSE gene at D4S2964 on Chromosome 4q21. [/table]
[table] Table 2: Correlation of HPSE expression with clinicopathological features. [/table]
[table] Table 3: Multivariate Cox regression analysis of impacts of variables affecting recurrence. [/table]
[table] Table 4: Multivariate Cox regression analysis of variables affecting early recurrence. [/table]
[table] Table 5: Multivariate Cox regression analysis of variables affecting overall survival. In univariate analysis, both mRNA and protein of HPSE were significant for overall survival. In this multivariate analysis, only protein score with smaller p value was included. doi:10.1371/journal.pone.0044061.t005 [/table]
[table] Table 6: Multivariate Cox regression model analysis of variables affecting overall survival in the subgroup of BCLC stage B,C,D. [/table]
[table] Table S2: LOH frequencies and mRNA expressions of 11 genes and their relationships with prognosis. (DOC) [/table]
[table] Table S3: Univariate Cox regression analysis of variables affecting metastasis. (DOC) Table S4 Univariate Cox regression analysis of impacts of variables affecting recurrence. (DOC) Table S5 Univariate Cox regression analysis of variables affecting early recurrence. (DOC) Table S6 Univariate Cox regression analysis of variables affecting late recurrence. (DOC) [/table]
[table] Table S7: Univariate Cox regression analysis of variables affecting overall survival. (DOC) [/table]
[table] Table S8: Univariate Cox regression analysis of variables affecting overall survival in the subgroup of BCLC stage 0, A. (DOC)Table S9 Univariate Cox regression analysis of variables affecting overall survival in the subgroup of BCLC stage B, C, D. (DOC) [/table]
|
Metformin Is Associated with a Lower Incidence of Benign Brain Tumors: A Retrospective Cohort Study in Patients with Type 2 Diabetes Mellitus
# Introduction
Meningioma is the most common benign brain tumor (BBT) (53.2%)and may be associated with metabolic syndromes, obesity, diabetes mellitus, hypertension and lack of physical activity. There is probably no association with smoking, alcohol drinking or dietary factors. On the other hand, the roles of hormone use, ionizing radiation, cell phone use and some occupational exposures are not clear.
Metformin is an old oral antidiabetic drug that has been used since the 1940s. It inhibits hepatic glucose output from gluconeogenesis and stimulates glucose uptake into skeletal muscle. The mechanism of these metabolic effects of metformin is mediated by the inhibition of mitochondrial respiratory chain complex 1, which in turn activates the 5 adenosine monophosphate-activated protein kinase (AMPK). Because of the potential risk of fatal lactic acidosis, its use was banned in the USA and Australia until 1995. Its use has gained momentum after 1998 when a reduced risk of cardiovascular events in obese/overweight patients with type 2 diabetes mellitus (T2DM) was demonstrated by the United Kingdom Prospective Diabetes Study. Metformin is now recommended as the first-line treatment for patients with T2DM because abundant studies have shown that it has various beneficial effects beyond glucose-lowering, such as anti-cancer, anti-aging, anti-inflammation and even antibiotic.
We recently found that patients with T2DM who had used metformin for more than two years had a significantly lower risk of malignant brain tumors. However, epidemiological data evaluating the effect of metformin on the protection against BBT are still rare. A recent matched case-control study using the UK Clinical Practice Research Datalink showed a higher but not significant risk of meningioma associated with metformin use, with an estimated odds ratio of 1.64 (95% confidence interval: 0.89-3.04). Because millions of diabetes patients are being treated with metformin, it is clinically important to clarify whether metformin can really increase the risk of BBT. The present study aimed at clarifying the effect of metformin on BBT in patients with T2DM.
# Materials and methods
The Taiwan government implemented a universal and unique health care system, which has been called the National Health Insurance (NHI) since 1 March 1995. The NHI is characterized by a high coverage rate of 99.6% of Taiwan's population and a high rate of involvement of medical providers (93% of all medical settings and all in-hospitals). The data that have to be submitted for reimbursement purpose include disease diagnoses, prescribed medications and performed procedures. Researchers can submit proposals for ethics review by using the database for academic research. The present study was approved and supported by the National Science Council (NSC 102-2314-B-002-067) and was reviewed by the National Health Research Institutes with an approval number of NHIRD-102-175 to provide the related database. The database has been described in more detail previously.
During the whole study period, the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) was the coding system for disease diagnoses in the database. Accordingly, the ICD-9-CM codes for diabetes mellitus were 250.XX and the codes for BBT were 225.X (benign neoplasm of brain and other parts of nervous system). The code 225.2 (benign neoplasm of cerebral meninges) was also used to identify a more specific outcome of cerebral meningioma among the diagnoses of BBT.
The step-by-step procedures inwere used to create a cohort of patients with T2DM enrolled in the study. First, we identified 423,949 patients who were newly diagnosed with diabetes mellitus between 1999 and 2005 and had received two or more times a prescription of antidiabetic drugs in the outpatient clinics. Ever users of metformin were defined as patients who had received metformin as the first antidiabetic drug. Therefore, 183,837 patients who had received a prescription of any other antidiabetic drug before metformin was initiated were first excluded. Ineligible patients who fitted the following criteria were then excluded: (1) 2062 patients who had been diagnosed as having type 1 diabetes mellitus, (2) 423 patients who had missing data, (3) 26,808 patients who had been diagnosed with any cancer before the entry date or within a short period of six months after diabetes diagnosis, (4) 1643 patients who had a previous diagnosis of BBT before enrollment or within 6 months of diabetes diagnosis, (5) 9260 patients aged <25 years, (6) 26,988 patients aged >75 years and (7) 4632 patients who were followed up for a duration of <180 days. As a result, we identified 152,176 ever users and 16,120 never users of metformin for the study.
The prescription information of metformin in the longitudinal database was used to calculate cumulative duration of metformin therapy, expressed in months, for each patient. A potential dose-response effect was evaluated by analyzing the risk in patients categorized according to the tertiles of cumulative duration of metformin therapy.
Potential confounders were classified into demographic and basic data, diabetesrelated complications, major comorbidities, antidiabetic drugs and drugs commonly prescribed to patients with diabetes mellitus. The demographic and basic data included variables of age, sex, living region and occupation. The living region and occupation were detailed elsewhere. In brief, the living region was classified as Taipei, Northern, Central, Southern, and Kao-Ping/Eastern. Occupation was classified as class I (civil servants, teachers, employees of governmental or private businesses, professionals and technicians), class II (people without a specific employer, self-employed people or seamen), class III (farmers or fishermen) and class IV (low-income families supported by social welfare, or veterans).
Diabetes-related complications included nephropathy, eye disease, stroke, ischemic heart disease and peripheral arterial disease. Major comorbidities included hypertension, dyslipidemia, obesity, chronic obstructive pulmonary disease, tobacco abuse, alcohol-related diagnoses and ocular pterygium (used as a surrogate of UV sunlight exposure). The ICD-9-CM codes for the above diagnoses have been described previously.
Antidiabetic drugs were classified as insulin, sulfonylurea, meglitinide, acarbose, rosiglitazone and pioglitazone. Drugs commonly prescribed to patients with diabetes mellitus were angiotensin-converting enzyme inhibitor/angiotensin receptor blocker, calcium channel blocker, statin, fibrate and aspirin.
The standardized difference was calculated for each of the potential confounders according to Austin and Stuart. A value of >10% was considered as a threshold for the indication of a potential confounding effect from the variable.
Statistical analyses were conducted for outcomes of any BBT and for cerebral meningioma, respectively. The incidence density of BBT/cerebral meningioma was calculated for the following subgroups of metformin exposure: never users, ever users and ever users classified according to the first, second and third tertile of cumulative duration of metformin therapy. The case number of new-onset BBT/cerebral meningioma diagnosed during The demographic and basic data included variables of age, sex, living region and occupation. The living region and occupation were detailed elsewhere. In brief, the living region was classified as Taipei, Northern, Central, Southern, and Kao-Ping/Eastern. Occupation was classified as class I (civil servants, teachers, employees of governmental or private businesses, professionals and technicians), class II (people without a specific employer, self-employed people or seamen), class III (farmers or fishermen) and class IV (low-income families supported by social welfare, or veterans).
Diabetes-related complications included nephropathy, eye disease, stroke, ischemic heart disease and peripheral arterial disease. Major comorbidities included hypertension, dyslipidemia, obesity, chronic obstructive pulmonary disease, tobacco abuse, alcoholrelated diagnoses and ocular pterygium (used as a surrogate of UV sunlight exposure). The ICD-9-CM codes for the above diagnoses have been described previously.
Antidiabetic drugs were classified as insulin, sulfonylurea, meglitinide, acarbose, rosiglitazone and pioglitazone. Drugs commonly prescribed to patients with diabetes mellitus were angiotensin-converting enzyme inhibitor/angiotensin receptor blocker, calcium channel blocker, statin, fibrate and aspirin.
The standardized difference was calculated for each of the potential confounders according to Austin and Stuart. A value of >10% was considered as a threshold for the indication of a potential confounding effect from the variable.
Statistical analyses were conducted for outcomes of any BBT and for cerebral meningioma, respectively. The incidence density of BBT/cerebral meningioma was calculated for the following subgroups of metformin exposure: never users, ever users and ever users classified according to the first, second and third tertile of cumulative duration of metformin therapy. The case number of new-onset BBT/cerebral meningioma diagnosed during the follow-up duration was the numerator of the incidence density. The denominator of the incidence density was the time of follow-up expressed as per 100,000 person-years. Follow-up started on 1 January 2006 and ended on the date of any of the following events whichever occurred first: a new-onset BBT/cerebral meningioma, death or the last reimbursement record, until 31 December 2011. We ended the follow-up by 2011 because the Bureau of NHI started to introduce and promote the use of ICD-10-CM to contracted hospitals and medical settings since 2012. This might have caused a mixture of the use of two disease coding systems.
Propensity score was created by logistic regression by treating all the variables listed inand the entry date as independent variables. To reduce confounding by indication, hazard ratios and their 95% confidence intervals were estimated through Cox regression incorporated with the inverse probability of treatment weighting (IPTW) using the propensity score, as proposed by Austin. The following three sensitivity analyses were conducted to examine the consistency of the findings: (1) patients receiving any two consecutive prescriptions of metformin spanning a period of more than four months were excluded. (2) patients having been treated with incretins during follow-up were excluded; and (3) patients having been treated with insulin were excluded. Because the Bureau of the NHI does not allow a prescription of medications for more than 3 months in each outpatient visit, the first sensitivity analyses excluded most patients without receiving regular drug refills. Because incretin-based therapies had not been introduced into Taiwan until after the enrollment of the patients, the second sensitivity analyses were aimed at avoiding the potential influence of these therapies that could have happened after the enrollment of the patients. Patients having been treated with insulin were excluded in the third sensitivity analyses because never users of metformin were characterized by a higher proportion of insulin use (8.34% versus 2.33%,and insulin is a growth factor for cell proliferation.
We used SAS statistical software (version 9.4, SAS Institute, Cary, NC, USA) to analyze the data. A p-value < 0.05 was considered statistically significant.
# Results
The characteristics between never users and ever users of metformin are compared in. The following variables had values of standardized difference >10%: age, nephropathy, eye diseases, dyslipidemia, insulin, meglitinide, acarbose, rosiglitazone, pioglitazone, statin and fibrate, suggesting potential risk of confounding from these variables. shows the incidence rates of BBT and cerebral meningioma, respectively, in different subgroups of metformin exposure; and the hazard ratios comparing the exposed subgroups to never users. While comparing metformin ever users to never users, a significantly (50%) lower risk was observed for both BBT and cerebral meningioma. The findings in the tertile analyses were also very similar for BBT and cerebral meningioma, showing a lower risk associated with metformin use in a dose-response pattern. A significant risk reduction could be seen only after 2 years of metformin use as shown in the second and third tertiles of cumulative duration of metformin therapy. . Incidence rates of benign brain tumors by metformin exposure and hazard ratios comparing exposed to unexposed subgroups. Sensitivity analyses are shown in. The results are very similar to those of the main analyses in . * Defined as "patients who had received two consecutive metformin prescriptions spanning a period of more than four months." ** Incretinbased therapies were not available in Taiwan before the starting date of follow-up.
## Metformin
# Discussion
The findings of the present study suggested that metformin use was associated with a significantly lower risk of BBT and cerebral meningioma in patients with T2DM. The risk reduction showed a dose-response pattern and was significant after a cumulative duration of metformin therapy of two or more years.
The mechanisms of the potential protective effect of metformin on BBT and/or cerebral meningioma remain unknown and await further investigation. Some biological effects of metformin may explain such a protective effect. Metformin may alter the gut microbiota leading to an increased production of butyrate, which may in turn reduce insulin resistance and obesity, the important risk factors of meningioma. Meningiomas are characterized by activation of multiple growth factor signaling pathways involving excess expression of membrane receptors of insulin-like growth factor receptor, platelet-derived growth factor receptor and vascular endothelial growth factor receptor. Metformin improves insulin effect and may reverse the proliferative effects of these growth factors, via AMPK activation and inhibition of the mammalian target of rapamycin pathway.
There are some clinical implications in the present study. First, together with the finding of a protective effect of metformin on malignant brain tumors seen in our previous study, metformin might also provide a protective effect on the development of BBT, in terms of cerebral meningioma and/or other types of BBT. These observations suggested that there might be some common pathophysiological pathways involved in the development of either malignant or benign brain tumors. This extra benefit of metformin further strengthened the recommendation of metformin as the firstline therapy for T2DM. Second, the significant risk reduction observed after two years of metformin use and the dose-response pattern provided good reasons to consider the continuous use of metformin when additional antidiabetic drugs are required to more adequately lower blood glucose levels during the course of treatment. Third, several clinical trials are being conducted to evaluate the efficacy of metformin on the treatment of malignant brain tumors. However, the usefulness of metformin on the prevention and treatment of BBT has not gained similar attention. Although BBT causes less severe clinical problems and is less life-threatening, the findings of the present study provide good rationale for designing and conducting clinical trials to investigate metformin's efficacy as a preventive agent for BBT and probably also other benign diseases characterized by cellular proliferation.
The present study has carefully addressed the potential methodological limitations commonly seen in pharmacoepidemiological studies that use big databases. These limitations may include selection bias, prevalent user bias, immortal time bias and confounding by indication.
It is believed that the problem of selection bias could be avoided in the present study because the healthcare system of the NHI covers nearly the whole population. Prevalent user bias can result from the enrollment of prevalent users rather than new users of a medication under investigation. Two types of bias can be introduced. First, "prevalent users are 'survivors' (healthy users) of the early period of pharmacotherapy, which can introduce substantial selection bias if the risk varies with time." Second, "covariates for drug use at study entry are often influenced by the previous intake of the drug". Therefore, prevalent user bias might have existed in the earlier study conducted in the UK that used a matched case-control study design. To mitigate such a problem, a "new user design" is recommended. In the present study, we carefully addressed this problem by enrolling patients at the time of diabetes diagnosis and only new users of metformin were defined in the user group. Additionally, to exclude the potential carry-over effect of other antidiabetic drugs, we enrolled only ever users of metformin who had never been previously treated with other antidiabetic drugs when the patients were first prescribed metformin.
The follow-up period when the outcome cannot happen is considered as the immortal time. When the treatment status is not appropriately assigned or when the follow-up time is not appropriately calculated, immortal time bias can be introduced. By enrolling patients who had two or more times a documented prescription of antidiabetic drugs from the nationwide NHI database, inappropriate assignment of treatment status is not likely. The follow-up time could be simply and probably accurately calculated from the database. The following periods of "immortal time" were not calculated in the follow-up time of the patients: (1) during the initial period of follow-up for <180 days; (2) between diabetes diagnosis and the start of the use of antidiabetic drugs; and (3) follow-up period of patients without use of any antidiabetic drugs. Another potential source of immortal time is the waiting period between the prescription and the dispense of drugs that may happen when a patient is discharged from an admission. Although this may be commonly seen in many other countries, this does not happen in Taiwan's NHI healthcare system because when the patient is discharged from the hospital, he/she can obtain all discharge medications immediately from the hospital.
Confounding by indication may happen when a risk factor of the outcome is associated with the indication of a medication under investigation. This could be reduced in the study by modeling with Cox regression incorporated with IPTW using propensity score.
The consistency of a beneficial effect of metformin on the prevention of BBT/cerebral meningioma and the dose-response pattern in different modelsstrengthened the robustness of the findings of the study.
There are several other strengths in the study. The exclusion of patients with a diagnosis of BBT within 6 months of diabetes diagnosis minimized reverse causality. With the use of existing medical records, self-reporting bias could be avoided. Because the drug cost-sharing is low in our NHI healthcare system, it is believed that detection bias as a result of discrepant socioeconomic status should be minimal. Furthermore, most of the healthcare expenses can be waived when the patients have low incomes, are veterans and receive prescription refills for chronic diseases.
Finally, we recognized that unmeasured confounders could never be adjusted for by statistical methods. Therefore, it is not known whether the results of the study could be biased by a lack of measurement data such as biochemistry, levels of insulin and some growth factors, immune profiles, hormone use, cell phone use, education levels, household conditions, nutritional status, dietary pattern, anthropometric factors, occupational exposure, physical activity, lifestyle, smoking, alcohol drinking and family history. However, a confounder needs to be correlated with the exposure (i.e., metformin use in the present study) and the disease (i.e., BBT/cerebral meningioma in the present study). Furthermore, it must not be a factor in the causal pathway in-between the exposure and the disease. Although the unmeasured variables may be risk factors of BBT/cerebral meningioma (disease), there is no evidence to support that they fit the other criteria to exert a confounding effect. Furthermore, the lack of histopathological data is another potential limitation associated with the study. It is worth pointing out that knowledge of absolute risk reduction and number needed to treat is important for decision making and clinical application. As the incidence of BBT was low, the absolute risk reduction calculated was too small (111/16,120 − 557/152,176 = 0.32%) and the number needed to treat (the reciprocal of absolute risk reduction) of 310 seemed to be too large as to be cost-effective to use metformin for the prevention of BBT, especially in people without diabetes mellitus.
In conclusion, this study supports a lower risk of BBT/cerebral meningioma in patients with T2DM who have used metformin, especially when metformin has been used for more than 2 years. However, additional studies are required to confirm the findings with more appropriate consideration of measured confounders and histopathological types. Because metformin is cheap and has a very safe profile with no risk of hypoglycemia, the usefulness of metformin in the prevention or treatment of BBT is worthy of in-depth investigation, in either diabetes patients or people without diabetes. Informed Consent Statement: Personal information was not available in the datasets and informed consent from the participants was not required according to local regulations.
# Data availability statement:
The datasets presented in this article are not readily available because public availability of the dataset is restricted by local regulations to protect privacy.
## Conflicts of interest:
The author declares no conflict of interest. |
Voluntary medical male circumcision (VMMC) for prevention of heterosexual transmission of HIV and risk compensation in adult males in Soweto: Findings from a programmatic setting
BackgroundClinical trials have clearly shown a reduction in HIV acquisition through voluntary medical male circumcision (VMMC). However, data assessing risk compensation under programmatic conditions is limited.. VMMC has been shown to decrease HIV infection under research conditions. However, data on risk compensation after circumcision in a programme setting is limited. Furthermore, evidence of risk compensation in males aged 24-40 years, the group with the highest HIV incidence in South Africa, also remains unknown.
There are concerns over the potential increase in high risk sexual behaviour as a result of risk compensation particularly in high HIV incidence age groups [bib_ref] Male circumcision for HIV prevention: current research and programmatic issues, Weiss [/bib_ref] [5] [bib_ref] Risk compensation in HIV prevention: Implications for vaccines, microbicides, and other biomedical..., Eaton [/bib_ref] which may mitigate the effectiveness of circumcision in preventing new HIV infections. Risk compensation is defined as modification of behaviour as a result of changes in the perceived risk of HIV infection [bib_ref] Risk compensation in HIV prevention: Implications for vaccines, microbicides, and other biomedical..., Eaton [/bib_ref]. With respect to HIV and circumcision, this would manifest itself in circumcised males knowing that they are less likely to contract HIV than before circumcision and in response, they might increase high-risk sexual behavior. In a 2013 study done in Uganda on male circumcision that assessed sexual behavior and HIV status, it was found that circumcised men are more likely to engage in risky behavior than their uncircumcised counterparts [bib_ref] Demographic Health Survey Working Papers: Male Circumcision, Sexual Behavior, and HIV Status..., Kibira [/bib_ref]. In contrast, another study done in Kenya found no evidence of risk compensation among males at least 24 months post-circumcision compared to the uncircumcised [bib_ref] Risk compensation following male circumcision: Results from a two-year prospective cohort study..., Westercamp [/bib_ref]. Also, a recent comparison of three cross-sectional studies showed an absence or minimal risk compensation [bib_ref] Changes in Male Circumcision Prevalence and Risk Compensation in the Kisumu, Kenya..., Westercamp [/bib_ref].
This study sought to further explore and elucidate the effect of VMMC on risky behavior post-circumcision in a programme setting [bib_ref] Risk compensation following male circumcision: results from a two-year prospective cohort study..., Westercamp [/bib_ref].
# Methods
The study was conducted at a high volume VMMC clinic based at Chris Hani-Baragwanath Academic Hospital in Soweto, a peri-urban township in South Africa. A prospective cohort of men requesting VMMC was established, inviting every third HIV-seronegative male, aged 18-40 years who reported living in Soweto from November 2012 to July 2014 to participate. Men were circumcised routinely and followed up prospectively thereafter. A post-circumcision visit was scheduled 12 months after VMMC.
An original target sample size of 904 was set to satisfy the primary objective of the study which was to determine a 50% difference in one-year HIV sero-incidence between before and after circumcision at a power of 80%, assuming a 15% non-informative loss to follow-up after VMMC in adult males. However, due to funding constraints and logistical challenges, HIV incidence was not reported on, and of the 496 participants enrolled, only follow-up data from 233 of these was utilised for the risk compensation analysis reported here. Selected VMMC-trained staff of the clinic were trained and conducted all study procedures. A structured questionnaire focused on participants' recall of HIV-risk behavior in the 12 months period pre-VMMC was conducted at enrolment. Surgical MMC was then performed on participants according to WHO guidelines . During the study follow-up period, participants were contacted monthly by telephone, to remind them of post-circumcision study visits. At a final study exit visit conducted 12 months post-VMMC, participants were tested for HIV using a rapid test and the structured questionnaire on risky behaviours was re-administered. Double data entry, was used to capture data into a REDCap database [bib_ref] Research Electronic Data Capture (REDCap)-a metadata driven methodology and workflow process for..., Harris [/bib_ref]. Questionnaires used were standardised and pilot tested prior to study commencement in October 2012.
The primary outcome of interest was HIV serostatus at the post-VMMC visit. Due to the rare occurrence of HIV incident infection over the study time frame~(3 sero-conversions in this case), HIV risk was represented by condom use at last sex act and concurrent sexual partnerships. These two variables were secondary outcomes of the study. Condom use at last sex act was defined as using a condom during the most recent sex act and concurrent sexual partnerships was defined as having more than one sexual partner at the time of the interview. We also measured perception of HIV risk as a possible confounder of concurrent sexual partnerships and condom use at last sex act. We defined perception of HIV risk as participant feeling that they were personally at risk of contracting HIV. Other confounders measured were having vaginal sexual intercourse in the last 12 months, a history of treatment of sexually transmitted infections (STIs) in the last 12 months, knowledge of sexual partners' HIV status at the last sex act and having taken alcohol at the time of last sexual act.
# Statistical analysis
Risk compensation was defined as a decrease in condom use at last sex act and or increase in concurrent sexual partners after circumcision. Selection of these two factors is based on the fact that concurrent sexual partnerships are main drivers behind high prevalence of HIV in sub-Saharan Africa among males. Furthermore, condom use is an effective mode of HIV prevention [bib_ref] Sexual risk factors for HIV infection in early and advanced HIV epidemics..., Chen [/bib_ref] [15] [bib_ref] Concurrent sexual partnerships help to explain Africa's high HIV prevalence: implications for..., Halperin [/bib_ref]. The association between categorical measures before and after circumcision was assessed using the McNemar's test.
# Ethical considerations
Study protocol was approved by the University of the Witwatersrand Human Research Ethics Committee (certificate #M120634). Chris Hani-Baragwanath Hospital Research Committee provided additional approval for the study. Strict confidentiality procedures were maintained and written informed consent was obtained from all participants. The study protocol can be accessed at: http://dx.doi.org/10.17504/protocols.io.seiebce.
# Results
## Demographics
The study enrolled 496 men, of whom 233 completed the 12-month follow-up visit [fig_ref] Fig 1: Screening, enrolment and follow-up of study participants [/fig_ref].
Distribution of baseline characteristics is shown in [fig_ref] Table 1: Baseline demographic characteristics by age-group [/fig_ref]. At sexual debut, higher condom use was reported for 18-24 years age group compared to 25-40 years age group (58% vs. 45%, p-value = 0.0469). More participants in 25-40 years age group had sexual intercourse in the previous twelve months compared to those in the 18-40 years age group (90% vs. 67%, p-value <0.0001). A higher proportion of 25-40 years olds compared to 18-24 years olds also reported knowledge of partner's HIV status (71% vs. 39%, p-value = <0.0001) and use of condoms at the last sexual act (57% vs. 21%, p-value = <0.0001).
Participants lost to follow-up or not yet completed 12 months visit did not differ from those retained in the study in relation to race, age, age at sexual debut, history of anal intercourse, condom use at last sex act, transactional sex, use of alcohol with sexual intercourse, risky sexual relationships, knowledge of sexual partners HIV status, and concurrent sexual partnerships (all p-values >0.05). However, those who completed follow-up were less likely to have used intravenous drugs (0% vs. 3%, p = 0.005) and less likely to have a history of an STI (2% vs 14%, p<0.001) [fig_ref] Table 2: Comparison of participants retained in the study and lost to follow-up [/fig_ref].
Risky behaviours pre and post-circumcision. The distribution of risky behaviours preand post VMMC is shown in . On comparing pre-to post-circumcision, there was a significant increase in the proportion of males engaging in sexual intercourse in the preceding twelve months from 184 (79%) to 204 (88%) (p-value = 0.0130), knowing the HIV status of sexual partner from 129 (55%) to 152 (65%) (p-value = 0.0063) and a perception of being at risk of contracting HIV from 162 (70%) to 195 (84%) (p-value = <0.0001). There was also a significant increase in males using a condom at last sex act from 92(40%) to 112 (48%) (pvalue = 0.0168). [fig_ref] Table 4: Sexual behaviours by age group-Before and 12 months after circumcision [/fig_ref]. There was an increase in proportion of men engaging in sexual intercourse for the 18-24 years age group. Similarly, knowledge of partners HIV status and condom use at last sex act was increased in the 18-24 age groups. In both age groups there was a significant increase in perception of being at risk of contracting HIV.
## Comparison of hiv risk behaviour before and after circumcision between age groups. a comparison of pre-and post-circumcision risky sexual behaviours stratified by age group (18-24 vs. 25-40 years) is shown in
# Discussion
Our study's findings corroborate evidence of absence of risk compensation after VMMC [17] [bib_ref] Risk compensation following male circumcision: results from a two-year prospective cohort study..., Westercamp [/bib_ref]. Our study also illuminated possible additional benefits of VMMC programs specifically the HIV counselling and VMMC education sessions. It was found that post-circumcision participants in the 18-24 years age group were more likely to know the HIV status of their sexual partner and more likely to use a condom at the last sex act . Absence of similar findings in the 25-40 years age will warrant further research considering that older men are at substantially higher HIV risk. Our study also found that participants in the 18-24 years age group were more likely to engage in sexual intercourse after VMMC than before. This is in line with findings in other parts of Africa [bib_ref] Risk compensation following male circumcision: results from a two-year prospective cohort study..., Westercamp [/bib_ref]. It could be explained by the fact that VMMC has been linked to sexuality and masculinity in some parts of South Africa which has increased uptake of the program but also that the cohort had aged by a year.
These findings show that there is little or no change in risky sexual behaviour after circumcision, particularly absence of evidence for an increase in concurrent sexual partnerships is reassuring, specifically in the context of prior research suggesting that concurrent sexual partnerships are a main driver behind the high prevalence of HIV among males in sub-Saharan Africa. [bib_ref] Concurrent sexual partnerships help to explain Africa's high HIV prevalence: implications for..., Halperin [/bib_ref]. The increase in the proportions having sexual intercourse post-circumcision are not large and are not by themselves risky as evidenced by increase in condom use, knowledge of partner's HIV status and lack of increase in alcohol use.
Limitations of this study include the high dropout rate as evidenced by less than half of participants completing the 12-month study visit. Data on risk factors was self-reported allowing for social desirability bias for which participants may have responded in a favourable manner. Having a single data point 12 months postcircumcision to collect data, may have introduced recall bias. Selection bias was introduced to the study by the inclusion criteria and having those who completed the 12 months follow-up visit being more likely to have used intravenous drugs and less likely to have a history of an STI. This limited generalizability of study findings as a representation of the general population with respect to age group, HIV status and those not seeking VMMC. Also, a lack of a control group which should have been uncircumcised males limited the generalizability of certain findings.
Effectiveness of VMMC must take into account changes in risk behaviours as well as possible effects of risk compensation. Further exploration must be undertaken to investigate why there was a significant increase or decrease in specific risk, and protective, behaviours post VMMC, and the relationship between risk behaviours and risk compensation.
# Conclusion
In this cohort we found in the 12 months post-circumcision compared to pre, participants in the 18-24 years age groups, were more likely to use a condom use at last sex act, more likely to know the HV status of sexual partner and more men (both 18-24-and 25-40-years age groups) considered themselves at risk of HIV infection. This suggests that there are other potential benefits conferred by VMMC in a programme setting as a biomedical and behavioural intervention for the prevention of HIV transmission. (7):920-31.
[fig] Fig 1: Screening, enrolment and follow-up of study participants. https://doi.org/10.1371/journal.pone.0213571.g001 Voluntary medical male for HIV prevention and risk compensation in Soweto PLOS ONE | https://doi.org/10.1371/journal.pone.0213571 March 7, 2019 [/fig]
[table] Table 1: Baseline demographic characteristics by age-group. [/table]
[table] Table 2: Comparison of participants retained in the study and lost to follow-up. [/table]
[table] Table 4: Sexual behaviours by age group-Before and 12 months after circumcision. [/table]
[table] 17: Mattson CL, Campbell R, Bailey RC, Agot K, Ndinya-Achola JO, Moses S. Risk compensation is not associated with male circumcision in Kisumu, Kenya: a multi-faceted assessment of men enrolled in a randomized controlled trial. PLoS One [Internet]. 2008 Jan [cited 2013 Mar 4]; 3(6):e2443. http://www. pubmedcentral.nih.gov/articlerender.fcgi?artid=2409966&tool=pmcentrez&rendertype=abstract PMID: 18560581 18. Chippindale S, French L. ABC of AIDS HIV counselling and the psychosocial management of patients with HIV or AIDS What is HIV counselling ? When is HIV counselling necessary ? Psychological responses to an HIV positive result. Bmj. 2011; 322:1533-5. 19. Humphries H, van Rooyen H, Knight L, Barnabas R, Celum C. 'If you are circumcised, you are the best': understandings and perceptions of voluntary medical male circumcision among men from KwaZulu-Natal, South Africa. Cult Heal Sex. 2015; 17 [/table]
|
Antiplatelet therapy for Staphylococcus aureus bacteremia: Will it stick?
AU : Pleaseconfirmthatallheadinglevelsarerepresentedcorrectly: Staphylococcus aureus bacteremia (SAB) AU : Theabstracthasbeenmadeasasingleparagraphasperthejou remains a clinically challenging infection despite extensive investigation. Repurposing medications approved for other indications is appealing as clinical safety profiles have already been established. Ticagrelor, a reversible adenosine diphosphate receptor antagonist that prevents platelet aggregation, is indicated for patients suffering from acute coronary syndrome (ACS). However, some clinical data suggest that patients treated with ticagrelor are less likely to have poor outcomes due to S. aureus infection. There are several potential mechanisms by which ticagrelor may affect S. aureus virulence. These include direct antibacterial activity, up-regulation of the innate immune system through boosting platelet-mediated S. aureus killing, and prevention of S. aureus adhesion to host tissues. In this Pearl, we review the clinical data surrounding ticagrelor and infection as well as explore the evidence surrounding these proposed mechanisms of action. While more evidence is needed before antiplatelet medications formally become part of the arsenal against S. aureus infection, these potential mechanisms represent exciting pathways to target in the host/pathogen interface.PLOS PATHOGENSPLOS Pathogens | https://doi.
## Author summary
Staphylococcus aureus remains a challenge to treat given its virulence and its ability to invade the bloodstream and spread to multiple sites in the body. Recently, it has been observed that patients taking the antiplatelet medication ticagrelor may have better infection outcomes. From this clinical observation, investigators have launched in vitro and animal studies to better understand by which mechanisms ticagrelor may affect S. aureus infection and clearance. In this Pearl, we review clinical data surrounding ticagrelor and infection as well as explore 3 different potential mechanisms of action that have been suggested by current studies. These mechanisms may involve boosting the host's plateletmediated innate immunity, representing an exciting direction for the treatment of S. aureus bacteremia. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111
# Introduction
Staphylococcus aureus bacteremia (SAB) remains a major clinical challenge with significant patient morbidity and mortality. To better address SAB, investigators seek antibacterial strategies that act in nontraditional ways, including those that augment the host immune response [bib_ref] Pharmacological Targeting of the Host-Pathogen Interaction: Alternatives to Classical Antibiotics to Combat..., Munguia [/bib_ref]. While platelets are well known for their role in thrombosis, they also participate in innate immunity. In vitro, platelets successfully kill S. aureus [bib_ref] Platelets Mediate Host Defense against Staphylococcus aureus through Direct Bactericidal Activity and..., Ali [/bib_ref]. Platelets can phagocytose S. aureus as well as secrete antibacterial peptides from alpha granules that kill S. aureus independent of antibodies [bib_ref] Activated platelets kill Staphylococcus aureus, but not Streptococcus pneumoniae-The role of FcγRIIa..., Wolff [/bib_ref] [bib_ref] Host defense role of platelets: Engulfment of HIV and Staphylococcus aureus occurs..., Youssefian [/bib_ref]. In addition to direct activity against S. aureus, platelets can also be activated by intravascular pathogens due to pattern recognition receptors, causing secretion of chemokines to recruit and enhance lymphocytes as well as communicate with endothelial cells [bib_ref] Platelets Mediate Host Defense against Staphylococcus aureus through Direct Bactericidal Activity and..., Ali [/bib_ref] , thereby augmenting the immune response. Clinically, thrombocytopenia in the setting of SAB has been associated with both a greater magnitude of bacteremia and patient mortality [bib_ref] Thrombocytopenia in Staphylococcus aureus bacteremia: Risk factors and prognostic importance, Gafter-Gvili [/bib_ref] , although it is not clear if this relationship is correlative or causative.
With advances in vascular medicine, platelet-modifying drugs such as ticagrelor, clopidogrel, and prasugrel are often prescribed for up to 1 year to patients suffering from acute coronary syndrome (ACS) [bib_ref] Ticagrelor versus Clopidogrel in Patients with Acute Coronary Syndromes, Wallentin [/bib_ref]. In the Study of Platelet Inhibition and Patient Outcomes (PLATO) randomized controlled trial [bib_ref] Ticagrelor versus Clopidogrel in Patients with Acute Coronary Syndromes, Wallentin [/bib_ref] , ticagrelor was found to be superior to clopidogrel in preventing death from myocardial infarct, stroke, and vascular causes in patients with ACS. Ticagrelor is a reversible inhibitor of the platelet adenosine diphosphate P2Y 12 receptor, whereas clopidogrel and prasugrel are irreversible inhibitors of the same receptor. It remains unclear whether platelet-modifying therapeutics influence the role of platelets in innate immunity, although there is preliminary in vitro, in vivo, and clinical evidence that ticagrelor may mitigate SAB.
Here, we review clinical evidence surrounding ticagrelor and infection as well as explore 3 potential pathways in which ticagrelor may inhibit S. aureus.
## Clinical data suggest that ticagrelor alters infection outcomes compared to patients taking other antiplatelet medications
In the PLATO trial, over 18,000 patients with ACS were treated with 1 year of either ticagrelor or clopidogrel [bib_ref] Ticagrelor versus Clopidogrel in Patients with Acute Coronary Syndromes, Wallentin [/bib_ref]. In a post hoc analysis of patients, adverse events were studied including rates of bacteremia/sepsis. Although the rates of these infections were similar in both groups, there were fewer deaths due to sepsis/bacteremia in the ticagrelor group (7 versus 23; p = 0.003).
The PLATO study renewed interest regarding infectious outcomes in patients following ACS. A total of 3 retrospective studies were published comparing patients on clopidogrel and ticagrelor. Among 9,518 patients treated with ticagrelor or clopidogrel (matched using propensity scoring), there were significantly fewer hospital readmissions due to infection with ticagrelor (6.11%) than with clopidogrel (10.53%) (HR 0.736, 95% CI 0.64 to 0.85; p < 0.001) [bib_ref] Risk of infectious events in acute myocardial infarction patients treated with ticagrelor..., Lee [/bib_ref]. In another propensity-matched retrospective study, 1.4% of 1,356 patients treated with ticagrelor compared to 3.6% of 1,356 patients treated with clopidogrel had gram-positive infections (HR 0.37; 95% CI 0.22 to 0.63; p < 0.001) [bib_ref] Effect of Ticagrelor on Reducing the Risk of Gram-Positive Infections in Patients..., Lupu [/bib_ref]. Last, a third retrospective study including over 26,000 patients measured the occurrence of SAB during the first year after initiation of either ticagrelor or clopidogrel [bib_ref] Ticagrelor and the risk of Staphylococcus aureus bacteraemia and other infections, Butt [/bib_ref]. PAU : PleasecheckwhethertheeditstothesentencePatientstreatedwithticagr atients treated with ticagrelor had significantly fewer episodes of SAB with absolute risk reduction of −0.19% (95% CI −0.32% to −0.05%; p = 0.006).
Notably, the findings from PLATO were a post hoc analysis, and these retrospective studies were correlative and not designed to determine cause and effect. However, these data in sum suggest that there may be mechanisms by which ticagrelor mitigates infection risk and, potentially, SAB.
## Ticagrelor has direct activity against s. aureus, albeit at supraphysiologic concentration
When evaluated with in vitro time-kill assays, ticagrelor was effective against methicillin-resistant S. aureus (MRSA), methicillin-susceptible S. aureus (MSSA), Staphylococcus epidermidis, Streptococcus agalactiae, and Enterococcus faecalis [bib_ref] Antibacterial Activity of Ticagrelor in Conventional Antiplatelet Dosages Against Antibiotic-Resistant Gram-Positive Bacteria, Lancellotti [/bib_ref]. It was not effective against 2 gram-negative pathogens, Escherichia coli and Pseudomonas aeruginosa. However, its antibacterial activity occurred at supraphysiologic concentrations. The minimum inhibitory concentration (MIC) of ticagrelor against MRSA was 20 μg/mL, whereas the physiologic concentration of ticagrelor at dosing for ACS in humans is between 0.8 and 1.2 μg/mL [bib_ref] Antibacterial Activity of Ticagrelor in Conventional Antiplatelet Dosages Against Antibiotic-Resistant Gram-Positive Bacteria, Lancellotti [/bib_ref]. Another study also found that ticagrelor inhibited a clinical isolate of MSSA but only at supraphysiologic concentrations (MIC 64 μg/mL) [bib_ref] Ticagrelor Increases Platelet-Mediated Staphylococcus aureus Killing, Resulting in Clearance of Bacteremia, Ulloa [/bib_ref]. Further, the combination of ticagrelor with the antimicrobials cefazolin or ertapenem was only additive rather than synergistic.
If concentrations of ticagrelor required for direct antistaphylococcal activity are not achievable clinically, how does ticagrelor exert an antibacterial effect at physiologic doses? In a murine model in which MRSA-inoculated polyurethane disks were implanted in the flanks of immunocompetent mice, those treated with ticagrelor at physiologic dosing had significantly decreased bacterial burden of their infected implant, suggesting that another mechanism in vivo may be driving the antistaphylococcal activity of ticagrelor [bib_ref] Antibacterial Activity of Ticagrelor in Conventional Antiplatelet Dosages Against Antibiotic-Resistant Gram-Positive Bacteria, Lancellotti [/bib_ref]. In sum, these results indicate that direct antibacterial activity of ticagrelor is unlikely to account for its apparent activity in vivo.
## Ticagrelor improves host platelet-mediated killing of s. aureus and decreases host thrombocytopenia
Platelets engage in the clearance of S. aureus by secretion of antimicrobial peptides and phagocytosis of bacteria as well as recruitment of other lymphocytes [bib_ref] Platelets Mediate Host Defense against Staphylococcus aureus through Direct Bactericidal Activity and..., Ali [/bib_ref] [bib_ref] Activated platelets kill Staphylococcus aureus, but not Streptococcus pneumoniae-The role of FcγRIIa..., Wolff [/bib_ref] [bib_ref] Host defense role of platelets: Engulfment of HIV and Staphylococcus aureus occurs..., Youssefian [/bib_ref]. In vitro, ticagrelor at physiologic concentration significantly enhanced the ability of human platelets to kill MRSA, whereas aspirin (another antiplatelet drug) did not [bib_ref] Repurposed drugs block toxin-driven platelet clearance by the hepatic Ashwell-Morell receptor to..., Sun [/bib_ref]. The same effect was reproducible against MSSA [bib_ref] Ticagrelor Increases Platelet-Mediated Staphylococcus aureus Killing, Resulting in Clearance of Bacteremia, Ulloa [/bib_ref]. Under microscopy, platelets incubated with S. aureus developed significant structural damage, although platelets treated with ticagrelor were relatively preserved, suggesting that ticagrelor may have a protective/stabilizing effect on platelets in the setting of S. aureus exposure [bib_ref] Repurposed drugs block toxin-driven platelet clearance by the hepatic Ashwell-Morell receptor to..., Sun [/bib_ref].
In an observational prospective study of 49 consecutive patients with SAB, thrombocytopenia correlated with increased mortality [bib_ref] Repurposed drugs block toxin-driven platelet clearance by the hepatic Ashwell-Morell receptor to..., Sun [/bib_ref]. Notably, isolates from SAB patients with more severe thrombocytopenia produced more α-toxin [bib_ref] Repurposed drugs block toxin-driven platelet clearance by the hepatic Ashwell-Morell receptor to..., Sun [/bib_ref] , an exotoxin that increases hepatic clearance of platelets through platelet desialylation. Mice infected with α-toxin-deficient S. aureus mutants had decreased thrombocytopenia and bacterial burden compared with mice infected with wild-type S. aureus. However, mice pretreated with physiologic concentrations of ticagrelor had decreased thrombocytopenia and improved survival during wild-type SAB [bib_ref] Repurposed drugs block toxin-driven platelet clearance by the hepatic Ashwell-Morell receptor to..., Sun [/bib_ref].
In a clinical case report, a 60-year-old man with refractory SAB and thrombocytopenia despite 5 days of antibiotic therapy was started on ticagrelor [bib_ref] Ticagrelor Increases Platelet-Mediated Staphylococcus aureus Killing, Resulting in Clearance of Bacteremia, Ulloa [/bib_ref]. Within 24 hours, his bacteremia resolved and platelet count improved. Discontinuation of ticagrelor led to recurrent thrombocytopenia, which then reversed with the resumption of ticagrelor. The patient was treated with 3 months of ticagrelor in addition to standard antibiotic therapy without further infection recurrence.
In sum, these in vitro and in vivo studies suggest that ticagrelor can enhance platelet-mediated killing of S. aureus as well as mitigate S. aureus-induced thrombocytopenia likely by preventing α-toxin-related desialylation. Given the role of platelets in innate immunity, maintaining platelet counts may contribute to improved outcomes in SAB as an additional benefit to ticagrelor therapy.
## Antiplatelet therapy inhibits s. aureus binding to host endothelial tissues
Among other adherence mechanisms, S. aureus binds platelets via interactions between its clumping factor A (clfA) and host platelet von Willebrand factor and fibrinogen [bib_ref] Clumping factor A mediates binding of Staphylococcus aureus to human platelets, Siboo [/bib_ref]. Activated platelets bind to the exposed extracellular matrix of damaged host endovascular tissue (such as heart valves). Therefore, preventing platelet aggregation on host endothelium by inhibiting platelet activation may mitigate SAB and its infectious complications. For example, S. aureus mutants lacking clfA were 50% less likely to cause endocarditis than wild-type strains in a SAB rat model [bib_ref] Role of Staphylococcus aureus coagulase and clumping factor in pathogenesis of experimental..., Moreillon [/bib_ref].
In ex vivo perfusion reactors, precoating bovine jugular veins with fibrinogen stimulated both human platelet and S. aureus surface binding [bib_ref] Antiplatelet therapy abrogates platelet-assisted Staphylococcus aureus infectivity of biological heart valve conduits, Ditkowski [/bib_ref]. However, the platelet α IIb β3 antagonist eptifibatide decreased S. aureus adhesion, likely due to inhibition of platelets. Likewise, the effect of antiplatelet therapy with aspirin and ticagrelor on S. aureus adhesion in the presence of human blood (including platelets) was tested under shear conditions [bib_ref] Antiplatelet therapy abrogates platelet-assisted Staphylococcus aureus infectivity of biological heart valve conduits, Ditkowski [/bib_ref]. Treatment with both aspirin and ticagrelor independently decreased S. aureus attachment to the lumen of bovine jugular veins; the combination of the 2 resulted in significantly less adhesion than aspirin alone. Dual antiplatelet therapy was also found in vivo to decrease endocarditis in a rat model of SAB due to inhibition of platelet binding [bib_ref] Prophylaxis of experimental endocarditis with antiplatelet and antithrombin agents: A role for..., Veloso [/bib_ref]. Preventing S. aureus binding to platelets and therefore minimizing contact with host tissues may be another mechanism by which ticagrelor and other antiplatelet drugs mitigate infection.
## Future directions
Clinical data from large patient cohorts suggest a protective effect of ticagrelor against infection. In vitro and rodent models have demonstrated that ticagrelor has direct antistaphylococcal activity at high concentrations and facilitates platelet-mediated killing of S. aureus, decreases SAB-induced thrombocytopenia, and mitigates binding of S. aureus to platelets and host tissue at physiologic concentrations [fig_ref] Fig 1: Schematic of proposed pathways in which ticagrelor may affect S [/fig_ref]. Therapeutic strategies that improve host immune function are appealing, as these are not prone to traditional bacterial resistance mechanisms [bib_ref] Pharmacological Targeting of the Host-Pathogen Interaction: Alternatives to Classical Antibiotics to Combat..., Munguia [/bib_ref]. In addition, repurposing existing licensed drugs is attractive, as the safety and adverse event profiles are well documented [bib_ref] Long-term tolerability of ticagrelor for the secondary prevention of major adverse cardiovascular..., Bonaca [/bib_ref].
Stronger evidence is needed to conclusively evaluate the clinical efficacy of ticagrelor in SAB. A prospective randomized controlled trial of patients receiving standard of care versus standard of care plus ticagrelor may bring further clarity. Anticipated risks and unanticipated consequences, including increased bleeding, would need to be carefully considered. In a prospective trial of over 20,000 patients, those randomized to take low-or high-dose ticagrelor did have significantly greater bleeding compared to placebo (6.2, 7.8, and 1.5%, respectively; p < 0.01), although 86% of bleeding events were nonmajor [bib_ref] Long-term tolerability of ticagrelor for the secondary prevention of major adverse cardiovascular..., Bonaca [/bib_ref]. Furthermore, there are conflicting reports that patients with endocarditis on anticoagulation may be more prone to cerebral hemorrhage due to emboli [bib_ref] Risk of stroke subsequent to infective endocarditis: A nationwide study, Østergaard [/bib_ref] , and this will further need to be weighed as a potential risk with use of ticagrelor in SAB. While promising, the potential of antiplatelet medication to treat staphylococcal infection remains uncertain.
[fig] Fig 1: Schematic of proposed pathways in which ticagrelor may affect S. aureus infection (not to scale): (1) enhancement of platelet-mediated S. aureus killing; (2) prevention of α-toxin-mediated platelet desialyation and subsequent removal from circulation via the hepatic Ashwell-Morell receptor; and (3) inhibition of S. aureus binding to host tissues via adhered platelets.https://doi.org/10.1371/journal.ppat.1010240.g001PLOS PATHOGENS [/fig]
|
Retroperitoneal bronchogenic cyst in suprarenal region treated by laparoscopic resection: A case report
BACKGROUNDBronchogenic cysts (BCs) are benign congenital foregut malformations that are mostly present in the mediastinum and pulmonary parenchyma but rarely seen in the retroperitoneum.CASE SUMMARYWe report the case of 17-year-old girl who complained of epigastric pain. A cystic lesion was found in the left suprarenal region on spectral computed tomography. The ovoid, well-defined, and homogeneous cystic lesion revealed slightly enhancement on conventional imaging but no enhancement on 40 KeV virtual mono-energetic images. The iodine density value of the lesion was 0.001 mg/mL and the Z-effective value was 7.25, which were close to those of fluid material in in vitro experiments. Magnetic resonance imaging revealed a cystic mass of intermediate signal intensity on T1-weighted imaging and high signal intensity on T2-weighted imaging. A laparoscopic surgery was carried out. Intraoperatively, a cystic lesion with a smooth surface was found in the left retroperitoneum. And the cystic wall was completely resected after intracystic fluid was suctioned. The histopathological examination findings of the lesion were compatible with BC. The patient recovered uneventfully without sighs of recurrence during a 10-mo follow-up period.CONCLUSIONRadiological examinations play a significant role in the diagnosis of suprarenal BCs and spectral images offer additional spectral parameters. Accurate preoperative diagnoses of retroperitoneal BCs based on thorough imaging examinations are beneficial to the operation of laparoscopic resection. Wu LD et al. Retroperitoneal bronchogenic cyst in suprarenal region WJCC https://www.wjgnet.com 7246Core Tip: Bronchogenic cysts (BCs) are benign congenital foregut malformations that are mostly present in the mediastinum and intrapulmonary but rarely seen in the retroperitoneum. We report a case of BC in the left suprarenal region which was diagnosed based on spectral computed tomography imaging and magnetic resonance imaging. After a laparoscopic surgery, the patient recovered uneventfully without sighs of recurrence during a 10-mo follow-up period.Citation: Wu LD, Wen K, Cheng ZR, Alwalid O, Han P. Retroperitoneal bronchogenic cyst in suprarenal region treated by laparoscopic resection: A case report. World J Clin Cases 2021; 9(24): 7245-7250
open-access article that was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution NonCommercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See
# Introduction
Bronchogenic cysts (BCs) are congenital foregut-derived aberrations of the respiratory tract in the process of embryonic development . They are usually located in the mediastinum and pulmonary parenchyma, and rarely occurred in the retroperitoneum. In most cases, retroperitoneal bronchogenic cysts (RBCs) occur near the left adrenal gland or pancreas.
In dual energy computed tomography (CT), material decomposition and quantification depend on photoelectric effect and Compton effect in different elemental compositions generated by X-ray. Dual layer detector spectral CT (also known as spectral CT) is a new detector-based dual energy CT for clinical use, in which the top and bottom layers of detector absorb low-and high-energy photons, respectively. Owing to the technology of novel dual layer detector, both conventional images and energy images reach perfect temporal and spatial registration. Spectral images are comprised of virtual mono-energetic images (VMI), iodine density map, Z-effective (Zeff) map, and so on. Spectral imaging has widely used in the study of abdominal diseases. It was reported that the optimal energy level of VMI could decrease pseudoenhancement effect of renal cysts in in vivo and in vitro experiments based on spectral imaging. Other spectral parameters such as iodine density value and Z-effective value were also used to differentiate cystic lesions from solid lesions . Currently, there is no report about imaging manifestations of suprarenl BCs based on spectral CT imaging. Herein, we report the first case of suprarenl BC found on the spectral CT and review cases of RBCs in English-language publications.
## Case presentation
## Chief complaints
A 17-year-old girl was referred to Union Hospital (Huazhong University of Science and Technology, Wuhan, Hubei Province, China) with epigastric pain.
## History of present illness
The patient had epigastric pain for a week without back pain or other symptoms.
## History of past illness
The patient had no previous history of hypertension or endocrine-related disease.
## Personal and family history
The patient was in good health in the past and had no family history. August 26, 2021 Volume 9 Issue 24
## Physical examination
Routine and specialized physical examination of the patient showed no obvious abnormalities.
## Laboratory examinations
Except erythrocyte count was slightly decreased, other laboratory results of the patient were within the normal limits.
## Imaging examinations
Spectral images showed a 29 mm × 17 mm × 28 mm ovoid, well-defined, and homogeneous cystic lesion in the left adrenal area. The lesion was slightly enhanced on conventional images while no enhancement on VMI 40KeV . In coronary plane of VMI 40KeV , the lesion margin was clear with significant contrast. The iodine density value of the lesion was 0.001 mg/mL and the Z-eff value was 7.25. Magnetic resonance imaging revealed a cystic mass with intermediate signal intensity on T1-weighted imaging (T1WI) and high signal intensity on T2-weighted imaging (T2WI). Chemical shift imaging revealed no fat component within the lesion.
## Final diagnosis
The histopathological characteristics of the lesion are compatible with BC. On gross examination, the cystic lesion was about 30 mm in size. Upon sectioning, the lesion contained gray yellowish fluid. The cystic wall with a smooth inner surface measured about 1-2 mm in thickness. The cyst contained pseudostratified ciliated columnar epithelium and mature hyaline cartilage.
## Treatment
To confirm the diagnosis, a retroperitoneal laparoscopic resection was carried out. During the surgery, a 2.0 cm skin incision was made in the mid-axillary line 1.5 cm above the iliac crest and a laparoscope was placed through a 10 mm trocar in this site. Two additional trocars (5 mm and 10 mm) were placed in the anterior and posterior axillary line below the inferior margin of the twelfth rib, respectively. The patient was placed in the left lateral position with the kidney rest elevated. The angle between the long axis of the patient's upper trunk and long axis of the table was 20º-30º. And surgeons were in the dorsal side of the patient close to buttocks. Meticulous dissection and fine operation technique were applied to avoid the injury to the peritoneum and vessels. Intraoperatively, a cystic lesion with a smooth surface was found in the left retroperitoneum. The cystic wall was completely resected after intracystic fluid was suctioned.
## Outcome and follow-up
The patient recovered uneventfully without sighs of recurrence during a 10-mo followup period.
# Discussion
BCs are benign malformation as a result of the abnormal budding of the foregut at the 3 rd -7 th weeks of the embryogenesis. When the connection between bronchial buds and tracheobronchial tree completely separates, the buds may migrate to unusual locations such as neck, intraspinal, pericardiac, and subdiaphragm. A search on PubMed database revealed 64 publications of RBCs published worldwide in the English literature. After screening the full texts and pathological results, 48 publications reporting on 50 cases of RBCs were eventually yielded. RBCs tend to be found on the left side of the abdomen. About half of them were located in the suprarenal area. According to Rud et al, the left pericardioperitoneal canal shuts later and is larger than the right one, which provides an explanation for the predominantly left side location of RBCs. The mean age of these patients was 39.1 ± 15.2 years. The mean size of cysts was 5.9 ± 3.2 cm. Slightly over half of patients had been identified incidentally without related symptoms. Of symptomatic patients, the majority complained of nonspecific epigastric pain and a small number complained of thoracic pain and back pain. To relieve clinical manifestations and reduce the risk of complications and malignant change, surgical excision is recommended. And laparoscopic resection has been widely used to lessen economic burden as well as postoperative pain of patients. More than half of patients with RBCs underwent laparoscopic resection and the majority were free of complications during the postoperative course.
BCs has ciliated, pseudo-stratified, columnar epithelium with parenchyma containing any one as follows: Seromucous glands, smooth muscle, or cartilage. Ultrasonography generally shows a cystic hypo/isoechoic lesion. On conventional images, RBCs are round or fusiform, well-circumscribed, and hypodense lesions with slight enhancement or without enhancement. RBCs usually manifest low or intermediate signal intensity on T1WI and high signal intensity on T2WI. In our case, increased attenuation on CT and increased signal intensity on T1WI suggested proteinaceous fluid. Some cases in the literature also reported this imaging appearance. Fat components were rarely seen in BCs. Most RBCs manifested as no signal loss on the opposed-phase imaging or no signal decrease on T1WI with fat suppression, which was similar to the appearance of our case .
We report the first case of RBC found on spectral CT. VMI are reconstructed by both high-energy and low-energy data sets of spectral imaging, which overcome pseudoenhancement of cystic lesion caused by beam-hardening artifacts . Also the image noise of whole spectrum is stable and relatively low. Therefore, the image quality of low-energy VMI is significantly improved. In our case, the lesion had slight enhancement on conventional images, while pseudo-enhancement of cystic lesion was eliminated and lesion conspicuity and margin delineation were more pronounced on VMI 40KeV . On iodine density map, iodine remain or uptake in specific blood vessels, organs, or lesions could be quantified using region of interest method. The extremely low iodine density of the lesion indicated a lack of cellular components, which excluded solid tumors with iodine uptake. Z-eff values depend on Compton effect and photoelectric effect of the material in the process of spectral decomposition. The Z-eff value of the lesion was 7.25, equal to that of fluid material reported on a phantom research of spectral imaging[8], which suggested that it was full of fluid contents.
The retroperitoneal laparoscopic resection was applied in our case. During the operation, the cystic wall was completely resected after intracystic fluid was suctioned. With thorough preoperative imaging and laboratory examinations, benign cystic lesions would be diagnosed confidently in the suprarenal region. Comparing to the laparoscopic surgery performed on other adrenal lesions, the surgery of RBCs has a clearer surgical field as well as shorter operative time.
# Conclusion
BCs are benign congenital foregut malformations caused by abnormal migration of bronchial buds. We report the first case of suprarenal BCs found on spectral CT. Radiological examinations play a significant role in diagnosis of suprarenal BCs and spectral images offer additional spectral parameters. Accurate preoperative diagnosis of RBCs based on thorough imaging examinations is beneficial to the operation of laparoscopic resection. |
Regulating Repression: Roles for the Sir4 N-Terminus in Linker DNA Protection and Stabilization of Epigenetic States
Silent information regulator proteins Sir2, Sir3, and Sir4 form a heterotrimeric complex that represses transcription at subtelomeric regions and homothallic mating type (HM) loci in budding yeast. We have performed a detailed biochemical and genetic analysis of the largest Sir protein, Sir4. The N-terminal half of Sir4 is dispensable for SIR-mediated repression of HM loci in vivo, except in strains that lack Yku70 or have weak silencer elements. For HM silencing in these cells, the Cterminal domain (Sir4C, residues 747-1,358) must be complemented with an N-terminal domain (Sir4N; residues 1-270), expressed either independently or as a fusion with Sir4C. Nonetheless, recombinant Sir4C can form a complex with Sir2 and Sir3 in vitro, is catalytically active, and has sedimentation properties similar to a full-length Sir4-containing SIR complex. Sir4C-containing SIR complexes bind nucleosomal arrays and protect linker DNA from nucleolytic digestion, but less effectively than wild-type SIR complexes. Consistently, full-length Sir4 is required for the complete repression of subtelomeric genes. Supporting the notion that the Sir4 N-terminus is a regulatory domain, we find it extensively phosphorylated on cyclin-dependent kinase consensus sites, some being hyperphosphorylated during mitosis. Mutation of two major phosphoacceptor sites (S63 and S84) derepresses natural subtelomeric genes when combined with a serendipitous mutation (P2A), which alone can enhance the stability of either the repressed or active state. The triple mutation confers resistance to rapamycin-induced stress and a loss of subtelomeric repression. We conclude that the Sir4 Nterminus plays two roles in SIR-mediated silencing: it contributes to epigenetic repression by stabilizing the SIR-mediated protection of linker DNA; and, as a target of phosphorylation, it can destabilize silencing in a regulated manner.
# Introduction
The eukaryotic genome is organized into euchromatic and heterochromatic domains that generally reflect their potential for gene expression. Chromatin repressed by the Silent information regulator (SIR) complex in the budding yeast Saccharomyces cerevisiae shares many key features with heterochromatin in higher eukaryotes. Notably, it has hypoacetylated nucleosomes [bib_ref] Transcriptional silencing in yeast is associated with reduced nucleosome acetylation, Braunstein [/bib_ref] [bib_ref] Highly specific antibodies determine histone acetylation site usage in yeast heterochromatin and..., Suka [/bib_ref] , is less accessible to DNA-binding enzymes than is euchromatin [bib_ref] Silencers and domains of generalized repression, Loo [/bib_ref] [bib_ref] Telomere-proximal DNA in Saccharomyces cerevisiae is refractory to methyltransferase activity in vivo, Gottschling [/bib_ref] [bib_ref] Active genes in budding yeast display enhanced in vivo accessibility to foreign..., Singh [/bib_ref] , it replicates late in S phase [bib_ref] Cell cycle-dependent establishment of a late replication program, Raghuraman [/bib_ref] and is spatially sequestered at the nuclear envelope or near the nucleolus [bib_ref] The clustering of telomeres and colocalization with Rap1, Sir3, and Sir4 proteins..., Gotta [/bib_ref]. The genes found within heterochromatin are generally silent, and in complex organisms this gene repression is crucial for the proper development of differentiated tissues and organs [bib_ref] Telomeric position effect: from the yeast paradigm to human pathologies?, Ottaviani [/bib_ref].
Unlike the situation in other eukaryotes, where histone H3 lysine 9 methylation and its specific ligands mediate repression, heritable transcriptional silencing in S. cerevisiae relies on the association of a trimeric SIR complex with unmodified histones (reviewed in [bib_ref] The establishment, inheritance, and function of silenced chromatin in Saccharomyces cerevisiae, Rusche [/bib_ref] [bib_ref] The molecular biology of the SIR proteins, Gasser [/bib_ref] [bib_ref] Common themes in mechanisms of gene silencing, Moazed [/bib_ref] [bib_ref] Silent chromatin at the middle and ends: lessons from yeasts, Buhler [/bib_ref]. This heterotrimeric complex contains equimolar amounts of Sir2, Sir3 and Sir4 [bib_ref] A homotrimerheterotrimer switch in Sir2 structure differentiates rDNA and telomeric silencing, Cubizolles [/bib_ref] , each of which is essential for the repression of promoters at the homothallic mating type loci, HMR and HML [bib_ref] Four genes responsible for a position effect on expression from HML and..., Rine [/bib_ref] and in subtelomeric domains [bib_ref] Modifiers of position effect are shared between telomeric and silent mating-type loci..., Aparicio [/bib_ref]. In analogy to centromeric position effect variegation in flies, repression at telomeres has been called telomere position effect, or TPE.
The SIR complex is targeted to the genes it represses by interacting with sequence-specific DNA-binding proteins that bind silencers or telomeric TG repeats. This binding initiates or ''nucleates'' the formation of silent chromatin on adjacent genes. Repressor activator protein 1 (Rap1; [bib_ref] Identification of silencer binding proteins from yeast: possible roles in SIR control..., Shore [/bib_ref] is a key factor for SIRmediated repression, because it has high affinity sites both at telomeres and in silencer elements [bib_ref] Identification of silencer binding proteins from yeast: possible roles in SIR control..., Shore [/bib_ref] [bib_ref] Connections between transcriptional activators, silencers, and telomeres as revealed by functional analysis..., Buchman [/bib_ref]. Furthermore, Rap1 interacts with both Sir3 and Sir4 [bib_ref] Evidence that a complex of SIR proteins interacts with the silencer and..., Moretti [/bib_ref]. HM silencer elements contain sites for two further sequence-specific factors, namely Abf1 (ARSbinding factor 1) and ORC (Origin recognition complex) [bib_ref] Separation of transcriptional activation and silencing functions of the RAP1-encoded repressor/activator protein..., Sussel [/bib_ref] [bib_ref] A yeast silencer contains sequences that can promote autonomous plasmid replication and..., Brand [/bib_ref]. Abf1 recruits the SIR complex by binding to Sir3 [bib_ref] The molecular biology of the SIR proteins, Gasser [/bib_ref] , and the largest subunit of ORC, Orc1, enhances SIR recruitment by binding Sir1, an intermediary protein that in turn binds Sir4 [bib_ref] Role of interactions between the origin recognition complex and SIR1 in transcriptional..., Triolo [/bib_ref].
The initial recruitment of Sir4 or Sir3 to telomeric TG-repeats or to silencers, brings in Sir2, a histone deacetylase [bib_ref] Silent information regulator 2 family of NAD-dependent histone/protein deacetylases generates a unique..., Tanner [/bib_ref] [bib_ref] Transcriptional silencing and longevity protein Sir2 is an NAD-dependent histone deacetylase, Imai [/bib_ref] [bib_ref] Mutations in Saccharomyces cerevisiae gene SIR2 can have differential effects on in..., Armstrong [/bib_ref] , which generates high-affinity binding sites for Sir3 by removing acetylation from the histone N-termini of nearby nucleosomes [bib_ref] Histone H3 and H4 N-termini interact with SIR3 and SIR4 proteins: a..., Hecht [/bib_ref] [bib_ref] Sir3-dependent assembly of supramolecular chromatin structures in vitro, Georgel [/bib_ref] [bib_ref] Reconstitution of heterochromatin-dependent transcriptional gene silencing, Johnson [/bib_ref]. Sir3 binds nucleosomes in a manner that is highly sensitive to histone H4 K16 acetylation [bib_ref] The dual role of H4K16 acetylation in the establishment of yeast silent..., Oppikofer [/bib_ref]. The sequential activation of this NAD-dependent histone deacetylase, its generation of high affinity binding sites for Sir3, and their occupancy by the trimeric SIR complex, allow a repressive chromatin structure to propagate along the chromatin fiber [bib_ref] SIR2 and SIR4 interactions differ in core and extended telomeric heterochromatin in..., Strahl-Bolsinger [/bib_ref] [bib_ref] Ordered nucleation and spreading of silenced chromatin in Saccharomyces cerevisiae, Rusche [/bib_ref]. Whereas Sir4 can be recruited to silencer elements independently of Sir2 and Sir3, the spreading of the SIR complex and formation of a silent domain require all three proteins [bib_ref] Ordered nucleation and spreading of silenced chromatin in Saccharomyces cerevisiae, Rusche [/bib_ref] [bib_ref] Targeting Sir proteins to sites of action: a general mechanism for regulated..., Cockell [/bib_ref]. Mutations that disrupt the interaction between Sir3 and Sir4 compromise repression of the HM loci and of genes at telomeres [bib_ref] A nonhistone proteinprotein interaction required for assembly of the SIR complex and..., Rudner [/bib_ref] [bib_ref] Structural basis for the role of the Sir3 AAA + domain in..., Ehrentraut [/bib_ref].
At 152 kDa, Sir4 is the largest and the least well conserved of the Sir proteins [bib_ref] Co-evolution of transcriptional silencing proteins and the DNA elements specifying their assembly, Zill [/bib_ref]. Its non-globular structure has rendered it refractory to biochemical analysis, except when expressed together with Sir2 [bib_ref] A homotrimerheterotrimer switch in Sir2 structure differentiates rDNA and telomeric silencing, Cubizolles [/bib_ref]. Sir2 and Sir4 form a stable heterodimer, which is mediated by residues 737-839 of Sir4 and a large pocket situated between Sir2's non-conserved N-terminus and its C-terminal catalytic domain (R. Sternglanz and R-M. Xu, personal communication). This tight interaction enhances the de-acetylation activity of Sir2 in vitro [bib_ref] A homotrimerheterotrimer switch in Sir2 structure differentiates rDNA and telomeric silencing, Cubizolles [/bib_ref] [bib_ref] Budding yeast silencing complexes and regulation of Sir2 activity by protein-protein interactions, Tanny [/bib_ref]. Sir4 also interacts with an array of additional factors that are required for efficient repression, leading to its designation as a scaffold for silent chromatin assembly [bib_ref] The molecular biology of the SIR proteins, Gasser [/bib_ref] [bib_ref] Common themes in mechanisms of gene silencing, Moazed [/bib_ref]. Importantly, the C-terminal coiled-coil of Sir4 (residues 1257-1358) dimerizes to generate Sir3-binding sites on its outer surface [bib_ref] Structure of the coiled-coil dimerization motif of Sir4 and its interaction with..., Chang [/bib_ref] [bib_ref] The Sir4 C-terminal coiled coil is required for telomeric and mating type..., Murphy [/bib_ref] , and this interphase is essential for SIRmediated repression [bib_ref] A nonhistone proteinprotein interaction required for assembly of the SIR complex and..., Rudner [/bib_ref]. This coiled-coil domain also binds Yku70 and Rap1 [bib_ref] Multiple interactions in Sir protein recruitment by Rap1p at silencers and telomeres..., Moretti [/bib_ref] [bib_ref] Silencing factors participate in DNA repair and recombination in Saccharomyces cerevisiae, Tsukamoto [/bib_ref] [bib_ref] Mutation of yeast Ku genes disrupts the subnuclear organization of telomeres, Laroche [/bib_ref] [bib_ref] Yeast Ku protein plays a direct role in telomeric silencing and counteracts..., Mishra [/bib_ref]. Yku70's interaction partner, Yku80, binds two sites within Sir4, one at the Sir4 N-terminus and one in the C-terminal 627 residues [bib_ref] Separation of silencing from perinuclear anchoring functions in yeast Ku80, Sir4 and..., Taddei [/bib_ref] [bib_ref] Separationof-function mutants of yeast Ku80 reveal a Yku80p-Sir4p interaction involved in telomeric..., Roy [/bib_ref]. The Ku heterodimer (Yku70/Yku80) not only facilitates SIR recruitment at telomeres, but helps anchor telomeres and silent chromatin at the nuclear envelope, which can enhance the efficiency of SIR-mediated repression [bib_ref] Mutation of yeast Ku genes disrupts the subnuclear organization of telomeres, Laroche [/bib_ref] [bib_ref] Perinuclear localization of chromatin facilitates transcriptional silencing, Andrulis [/bib_ref] [bib_ref] Evidence for silencing compartments within the yeast nucleus: a role for telomere..., Maillet [/bib_ref]. A second, more central domain of Sir4 called PAD (residues 950-1262; partitioning and anchoring domain) also mediates anchorage to the nuclear envelope [bib_ref] Separation of silencing from perinuclear anchoring functions in yeast Ku80, Sir4 and..., Taddei [/bib_ref] [bib_ref] Live imaging of telomeres: yKu and Sir proteins define redundant telomereanchoring pathways..., Hediger [/bib_ref] [bib_ref] Esc1, a nuclear periphery protein required for Sir4-based plasmid anchoring and partitioning, Andrulis [/bib_ref]. The PAD domain of Sir4 binds a nuclear envelopeassociated protein called Esc1 (Establishes silent chromatin 1) [bib_ref] Esc1, a nuclear periphery protein required for Sir4-based plasmid anchoring and partitioning, Andrulis [/bib_ref] [bib_ref] Multiple pathways for telomere tethering: functional implications of subnuclear position for heterochromatin..., Taddei [/bib_ref]. Disruption of ESC1 and YKU70 or YKU80 releases telomeres from the nuclear envelope, and selectively de-represses TPE, while repression at HM loci remains intact [bib_ref] Separation of silencing from perinuclear anchoring functions in yeast Ku80, Sir4 and..., Taddei [/bib_ref] [bib_ref] Sir-mediated repression can occur independently of chromosomal and subnuclear contexts, Gartenberg [/bib_ref] [bib_ref] The functional importance of telomere clustering: global changes in gene expression result..., Taddei [/bib_ref].
It is not surprising that the C-terminal half of Sir4 is crucial for silencing, given that it mediates protein-protein interactions with Rap1, Sir2, Sir3, Sir4, Yku70/Yku80 and Esc1. Although we know much less about the functions of the N-terminal part of Sir4, Marshall et al. [bib_ref] Functional domains of SIR4, a gene required for position effect regulation in..., Marshall [/bib_ref] reported that the N-terminus of Sir4 was required for silencing at the HM loci. They showed that expression in trans of an N-terminal fragment restored mating in the presence of a silencing-deficient C-terminal fragment of Sir4 (the final 45%, starting from about residue 744) [bib_ref] Functional domains of SIR4, a gene required for position effect regulation in..., Marshall [/bib_ref]. Since then, the first 270 residues of Sir4 (Sir4N) were shown to bind DNA in vitro [bib_ref] Reconstitution of yeast silent chromatin: multiple contact sites and O-AADPR binding load..., Martino [/bib_ref] and to interact with three proteins: Sir1 [bib_ref] Role of interactions between the origin recognition complex and SIR1 in transcriptional..., Triolo [/bib_ref] , Yku80 [bib_ref] Separationof-function mutants of yeast Ku80 reveal a Yku80p-Sir4p interaction involved in telomeric..., Roy [/bib_ref] and Sif2 [bib_ref] Sif2p interacts with Sir4p amino-terminal domain and antagonizes telomeric silencing in yeast, Cockell [/bib_ref] , a component of the SET3C deacetylase complex [bib_ref] Sif2p interacts with Sir4p amino-terminal domain and antagonizes telomeric silencing in yeast, Cockell [/bib_ref] [bib_ref] The S. cerevisiae SET3 complex includes two histone deacetylases, Hos2 and Hst1,..., Pijnappel [/bib_ref]. Although Sir4 binding to Sir1 or Yku80 facilitates SIR complex recruitment to HM loci and telomeres, neither interaction is essential for SIR-mediated silencing [bib_ref] Sir-mediated repression can occur independently of chromosomal and subnuclear contexts, Gartenberg [/bib_ref] [bib_ref] Epigenetic inheritance of transcriptional states in S. cerevisiae, Pillus [/bib_ref] [bib_ref] The Ku complex in silencing the cryptic matingtype loci of Saccharomyces cerevisiae, Patterson [/bib_ref]. Thus, it remained mysterious what function the Sir4 N-terminus might have.
Here we have explored the function of the N-and C-terminal domains of Sir4 in silencing at both the HM loci and yeast telomeres by means of biochemical and genetic assays. We reexamined the ability of the N-and C-termini to work together in trans and found, surprisingly, that a slightly shorter C-terminal fragment (Sir4C; residues 747-1358) than that used by Marshall et al. [bib_ref] Functional domains of SIR4, a gene required for position effect regulation in..., Marshall [/bib_ref] , is sufficient to silence HMR and HML in a sir4D background. Neither this C-terminal domain nor a fusion protein of Sir4C to the N-terminal 270 residues, however, was sufficient to complement fully a sir4 deletion for TPE. From this we conclude that the Sir4 N-terminus is dispensable for formation of a repressed chromatin structure, yet it is needed at telomeres or in situations in which SIR complex recruitment is compromised.
We confirmed by biochemical reconstitution assays that recombinant Sir4C is sufficient to form a complex with Sir2 and Sir3 that binds nucleosomal arrays in vitro and deacetylates histone H4 K16 ac . However, Sir4C-containing complexes bind with a four-fold lower affinity and confer less protection of linker DNA from micrococcal nuclease attack. Thus, the DNA binding affinity of Sir4N contributes substantially to the tight association of the SIR complex with chromatin, which becomes important when recruitment is compromised. To see if silencing is regulated through Sir4, we mapped phosphorylation sites within Sir4N in vivo and in vitro, and found that this domain is a major target for phosphorylation in living cells. Two key phosphoacceptor sites for the cyclin-dependent kinase, serine 63 and serine 84, influence the stability of repression at most telomeres showing TPE. We propose that Sir4N phosphorylation regulates the stability of subtelomeric repression during the cell cycle and possibly in response to environmental stress.
# Results
## Sir4c is sufficient for silencing at intact hml and hmr loci
To examine the function of the N-terminus of Sir4, we first repeated the assay of Marshall et al. [bib_ref] Functional domains of SIR4, a gene required for position effect regulation in..., Marshall [/bib_ref] in which N-and Cterminal fragments of Sir4 were expressed in trans and scored for the restoration of silencing at HML in a sir4D background. We created strains with either a full deletion of SIR4 (sir4D) or with a partial deletion of the endogenous SIR4 locus (sir4N), such that only its N-terminal 270 amino acids were expressed. We then expressed full-length Sir4 or various C-terminal fragments of the
## Author summary
Three Silent Information Regulator (SIR) proteins Sir2, Sir3, and Sir4 are involved in the epigenetic gene silencing of the homothallic mating (HM) loci and of telomere-proximal genes in budding yeast. They bind as a heterotrimeric complex to chromatin, repressing the underlying genes. Sir2 has an essential histone deacetylase activity, and Sir3 binds nucleosomes, with a high specificity for unmodified histones. We explored Sir4, whose role had largely remained a mystery. We report here that Sir4 N-and Cterminal domains have distinct functions: The Sir4 Cterminus binds all proteins essential for SIR-mediated silencing and is sufficient to repress HM loci, but surprisingly it is not sufficient to efficiently repress at telomeres. The Sir4 N-terminus binds DNA, which strengthens the SIR-chromatin interaction and helps target Sir4 to telomeric loci. In addition the Sir4 N-terminus binds sequence-specific factors that recruit Sir4 to sites of repression. We find that the Sir4 N-terminus is a target of mitotic phosphorylation. Mutation of the phosphoacceptor sites indicates that they help fine-tune subtelomeric repression. We propose therefore that phosphorylation of the Sir4 N-terminal domain modulates epigenetic repression at telomeres in response to cell cycle and/or stress situations.
protein from CEN-ARS plasmids (pRS) carrying the full SIR4 promoter and terminator [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref] , 1B). Because overexpression of either full-length protein or fragments of Sir4 derepress gene silencing [bib_ref] Sif2p interacts with Sir4p amino-terminal domain and antagonizes telomeric silencing in yeast, Cockell [/bib_ref] , we chose conditions that reproduced as closely as possible the endogenous Sir4 protein levels [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref] and data not shown). Quantitative mating assays can be used to determine the degree of repression at HML, because mating is compromised by coincident expression of a and a mating type information. In contrast to the findings of Marshall et al. [bib_ref] Functional domains of SIR4, a gene required for position effect regulation in..., Marshall [/bib_ref] , expression of a Cterminal fragment of Sir4 (Sir4C, residues 747-1358) alone was sufficient to repress HML, as indicated by the restoration of mating in a MATa sir4D strain [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref] , . Consistently, expression of Sir4C also repressed a TRP1 reporter inserted at HMR in both the sir4D and sir4N backgrounds [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref] , .
In trying to explain the discrepancy between our findings and those of Marshall and colleagues, we noticed that they had used a galactose-inducible Sir4 C-terminal fragment that was a few amino acids longer than ours, and co-expressed as well a slightly longer N-terminal fragment than we used [bib_ref] Functional domains of SIR4, a gene required for position effect regulation in..., Marshall [/bib_ref]. Intriguingly, our analysis of a longer C-terminal fragment (residues 731-1358; Sir4 731-1358 ), showed that it failed to repress an HMR::TRP1 reporter, either alone (in a sir4D background) or when expressed with Sir4N (in a sir4N background; [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref]. Immunoblotting showed that steady-state levels of the Sir4 731-1358 fragment were much lower than of those of the shorter Sir4C [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref]. The instability of the Sir4 731-1358 fragment would explain its inability to repress HMR; indeed, it is likely that the fragment used by Marshall and colleagues was also unstable, and therefore did not silence on its own. We tried also expressing a longer (330 residue) N-terminal fragment with both long and short Sir4C fragments, but observed no differences in the mating assay compared to the shorter Sir4N fragment (data not shown). Our results suggest that a stable 611-residue C-terminal fragment of Sir4 is sufficient to repress both HM loci.
The N-terminus of Sir4 contributes to repression at HM loci with incomplete silencers To date, the N-terminus of Sir4 was implicated in recruiting the SIR complex to silencers or to telomeres through its affinity for Sir1 or Yku80, respectively [bib_ref] Role of interactions between the origin recognition complex and SIR1 in transcriptional..., Triolo [/bib_ref] [bib_ref] Mutation of yeast Ku genes disrupts the subnuclear organization of telomeres, Laroche [/bib_ref] [bib_ref] Yeast Ku protein plays a direct role in telomeric silencing and counteracts..., Mishra [/bib_ref] [bib_ref] The Ku complex in silencing the cryptic matingtype loci of Saccharomyces cerevisiae, Patterson [/bib_ref] [bib_ref] The DNA end-binding protein Ku regulates silencing at the internal HML and..., Vandre [/bib_ref]. The interactions that recruit the SIR complexes to silencers are, however, redundant [bib_ref] A yeast silencer contains sequences that can promote autonomous plasmid replication and..., Brand [/bib_ref]. Therefore, we next tested the impact of Sir4N on silencing under conditions of compromised recruitment, that is, in strains lacking either Sir1 or Yku70 which eliminates Yku80 function as well [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref]. The expression of Sir4C in a sir1D strain could still restore silencing of HML, either in the absence (sir4D) or the presence of Sir4N (sir4N, [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref]. Consistent with it being a The N-terminal domain of Sir4 (Sir4N) is in red, the C-terminal domains in green (full Sir4C = 747-1358; light green PAD = 950-1262; dark green coiled coil domain = 1262-1358). B) Scheme of plasmids expressing Sir4 constructs. The plasmid's original promoter and terminator were replaced with a 1 kb sequence of the SIR4 59 region and 250 bp of the 39 region containing the endogenous promoter/terminator information. The same plasmid construct with different markers was used as needed. C) Silencing at HML of strains with various Sir4 domains was assayed by quantitative mating to a tester strain (GA858). The endogenous SIR4 copy was full length (SIR4; GA503), a C-terminal deletion (sir4N; GA5809) or a complete deletion of Sir4 (sir4D; GA5822). C-terminal or full length Sir4 was added back on a plasmid. Mating efficiency was normalized to the wild-type strain; data represent mean value 6 s.e.m, n.d. undetermined values. D) Plasmids similar to (C), but silencing at HMR was assayed using a TRP1 reporter (GA484, GA6072, GA5886). Serial dilutions of transformed strains were grown on control plates selecting for the plasmid only or on plates selecting for the plasmid and growth without tryptophan (monitoring repression of TRP1). doi:10.1371/journal.pgen.1002727.g001 direct binding partner of Sir1, Sir4N expression did not enhance silencing in the sir1D background [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref]. On the other hand, in the sir4D yku70D background, Sir4C only supported mating at 30% of wild-type levels. In this case, mating efficiency was indeed enhanced by co-expression of Sir4N (compare SIR4C in sir4D and sir4N, [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref]. This confirms that the N-and C-termini of Sir4 can complement in trans at the HML locus, as reported by [bib_ref] Functional domains of SIR4, a gene required for position effect regulation in..., Marshall [/bib_ref] , although in our hands, this is true only in yku70D cells.
To assess more directly whether Sir4N compensates for the absence of other recruitment sites at silencers, we used a HMR::TRP1 reporter strain that lacks either the A sequence (ORC-Sir1-binding site) or the B sequence (Abf1-binding site) within the E silencer [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] [bib_ref] Separation of transcriptional activation and silencing functions of the RAP1-encoded repressor/activator protein..., Sussel [/bib_ref] [bib_ref] A yeast silencer contains sequences that can promote autonomous plasmid replication and..., Brand [/bib_ref]. When we deleted the ORC-Sir1-binding element (HMR-EDA), Sir4C was no longer sufficient to repress the reporter gene at HMR [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] , . Similarly, in the absence of the Abf1-binding site (HMR-EDB, [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] , , Sir4C did not restore silencing, either with or without the Sir4N fragment. Thus, Sir4C is not sufficient for silencing at an HMR locus in which the silencers are weakened by deletion of a binding site for one of the recruitment factors.
The co-expression of Sir4C and Sir4N in trans did not enhance silencing at compromised silencers, as they did in the yku70 mutant [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref]. This may be explained if Sir4N interacts only weakly with the SIR complex. To test this possibility, we tethered the Sir4N and Sir4C domains with a short linker peptide, to form a stable fusion protein (Sir4N-C; [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref]. Importantly, when expressed in a sir4D background, Sir4N-C repressed the reporter gene as effectively as full-length Sir4 at the silencer-compromised HMR loci (HMR-EDA and HMR-EDB; [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] , . The strains expressing Sir4N-C were also competent for mating [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] , albeit with lower efficiency than cells expressing Sir4C alone, possibly due to an altered growth rate (see legend, [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref]. These data confirm a role for the N-terminus of Sir4 in silencing the HM loci when the binding of recruitment factors is compromised. Indeed, at HMR with weakened silencers, the expression of a Sir4 N-terminal fragment along with Sir4C allows repression, whereas Sir4C alone does not.
## Linking sir4n to sir4c increases but does not fully restore telomeric silencing
To test whether silencing at telomeres requires the N-terminus of Sir4, we monitored expression of a URA3 reporter gene at telomere 7L (Tel7L::URA3) [bib_ref] Telomere-proximal DNA in Saccharomyces cerevisiae is refractory to methyltransferase activity in vivo, Gottschling [/bib_ref] by assaying growth in the absence of uracil in a strain that lacks Ppr1, the transcription factor responsible for inducing URA3 in auxotrophic conditions [bib_ref] Silent domains are assembled continuously from the telomere and are defined by..., Renauld [/bib_ref]. In contrast to repression at the HM loci, telomeric silencing could not be established by expressing Sir4C [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] , nor by coexpressing Sir4C with Sir4N in trans [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] , . This was true not only for URA3 expression at Tel7L, but also for the ADE2 reporter gene expression at Tel5R [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] , . We also monitored Tel7L::URA3 repression by counter-selecting with the drug 5-FOA, with similar results [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref].
Given that repression at HMR with weakened silencers was enhanced by expression of a Sir4N-C fusion, we tested the effect of this hybrid on TPE. Surprisingly, expression of the Sir4N-C fusion in a sir4D strain failed to repress either Tel7L::URA3 or Tel5R::ADE2 reporters in the standard drop assay [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] , [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] , . We also assayed silencing by measuring mRNA levels of subtelomeric genes from telomeres 6R and 9R by quantitative PCR (QPCR; [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref]. Both genes were derepressed in cells expressing only Sir4C or Sir4N-C. For Tel9R we observed partial repression by Sir4N-C compared to sir4D [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] , . Intriguingly, the levels of the HMLa1 gene . Summary of silencing phenotypes. were only partial reduced when Sir4C was expressed, whereas expression of Sir4N-C conferred repression to near-background levels [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref]. A similar effect was observed at Tel7L::URA3 when URA3 was transcribed at basal levels (i.e. growth in the presence of uracil and in the absence of Ppr1; [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref]. In this case, Sir4N-C repressed transcription to background levels, unlike in the drop assay in the absence of uracil [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] , which strongly induces transcription from the URA3 promoter. Resistance to rapamycin is a sensitive means to monitor native telomeric silencing, as growth on the drug requires expression of multiple stress genes located near telomeres, which are normally silenced by the SIR complex [bib_ref] Regulation of subtelomeric silencing during stress response, Ai [/bib_ref]. We therefore monitored the level of stress gene expression by scoring for resistance to rapamycin in Sir4-, Sir4C-and Sir4N-C-expressing cells. Whereas SIR4 + cells fail to grow on rapamycin, intriguingly, both Sir4Cand Sir4N-C-expressing cells behaved like sir4D when grown in the presence of rapamycin [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref]. This argues that neither Sir4C nor Sir4N-C can prevent the induction of natural subtelomeric genes by stressful conditions [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] , suggesting that full-length Sir4 is needed for native subtelomeric repression, although not at HM loci.
Transcriptional repression generally correlates with the binding of Sir proteins throughout the silent domain, and we therefore tested the binding of Sir4 to HML and telomeres by chromatin immunoprecipitation (ChIP). We detected a clear enrichment of Sir4, Sir4C and Sir4N-C at HML-E and HML-a1 [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref]. Consistent with the silencing assays, on the other hand, only full length Sir4 was strongly enriched at telomeres [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref]. This confirms that full-length and truncated Sir4 proteins are bound at the sites that are silenced robustly, and shows again that Sir4C is not sufficient for binding in subtelomeric domains.
To see if Sir4C would be sufficient for silencing at telomeres if we enhanced SIR recruitment by Rap1, we monitored TPE in the absence of the Rap1-interacting factor 1 (Rif1), which competes [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref] with strains additionally carrying full deletions of YKU70 or SIR1 (GA6069, GA6070, GA6071, GA6062, GA6063, GA6064). Mating was normalized to wild-type cells and at least three independent experiments were quantified; data represent mean value 6 s.e.m. # indicates values below 10 23 , n.d. undetermined values. B, C, D) Testing silencing of compromised HMR: full sir4 deletion or endogenous sir4N were complemented with SIR4, SIR4C or a SIR4N-C fusion in strains carrying a TRP1 reporter at HMR. The HMR-E silencer carried a deletion of either the B (Abf1 binding; GA485, GA6888, GA6899) or A (Orc1-Sir1 binding; GA486, GA6890, GA6891) site. Dilution series for repression were performed as in [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref]. doi:10.1371/journal.pgen.1002727.g002
for Sir3 and Sir4 recruitment by the Rap1 C-terminus [bib_ref] A RAP1-interacting protein involved in transcriptional silencing and telomere length regulation, Hardy [/bib_ref]. Whereas deletion of RIF1 increased telomere length and SIR recruitment, leading to enhanced silencing [bib_ref] Evidence that a complex of SIR proteins interacts with the silencer and..., Moretti [/bib_ref] [bib_ref] Yeast Ku protein plays a direct role in telomeric silencing and counteracts..., Mishra [/bib_ref] , it did not increase Sir4C-or Sir4N-C-mediated repression at Tel7L or Tel5R::ADE2 [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] , S2E; compare to [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref]. Taken together, these data indicate that Sir4C is insufficient for TPE, and that Sir4N can contribute weakly to improve repression at native telomeric genes and reporters, yet only full-length Sir4 supports robust TPE.
## Sir4c can form a stable and active sir complex
Because Sir4C can silence HM loci, we asked whether Sir4C forms a stable complex with Sir2 and Sir3. To test this, we coexpressed Sir4C with Sir2 and Sir3 in baculovirus-infected insect cells. Using conditions identical to those used to purify the fulllength Sir2-Sir3-Sir4 complex, we were able to purify a SIR complex containing Sir4C , 4C; [bib_ref] A homotrimerheterotrimer switch in Sir2 structure differentiates rDNA and telomeric silencing, Cubizolles [/bib_ref] [bib_ref] Reconstitution of yeast silent chromatin: multiple contact sites and O-AADPR binding load..., Martino [/bib_ref]. Upon glycerol density gradient sedimentation the complex migrated in two distinct complexes: one containing Sir2, Sir3 and Sir4C and the other containing only Sir2 and Sir4C, exactly like the complex with full-length Sir4 (Figure 4C, 4D; [bib_ref] A homotrimerheterotrimer switch in Sir2 structure differentiates rDNA and telomeric silencing, Cubizolles [/bib_ref]. We conclude that Sir4C is sufficient to form a complex similar to the wild-type SIR complex, when expressed in insect cells.
To confirm that the Sir4 N-terminus is dispensable for the deacetylation activity of the SIR complex, we incubated recombinant protein complexes of either full-length Sir2-Sir3-Sir4 or truncated Sir2-Sir3-Sir4C with histone octamers that were fully acetylated on histone H4K16. We assayed H4K16 ac deacetylation over time by Western blotting [bib_ref] The dual role of H4K16 acetylation in the establishment of yeast silent..., Oppikofer [/bib_ref] , and found that the two complexes had similar deacetylation activities . We conclude that Sir4C forms a stable and active SIR complex, consistent with its ability to confer HM repression.
## Sir4n promotes high-affinity binding to chromatin and linker dna protection
To explain the contributions of the Sir4 N-terminus for repression in biochemical terms, we examined the contribution of Sir4N to SIR complex loading onto nucleosomal arrays in vitro. In a previous study, we showed that recombinant Sir4N has considerable non-specific affinity for DNA [bib_ref] Reconstitution of yeast silent chromatin: multiple contact sites and O-AADPR binding load..., Martino [/bib_ref]. To test whether this contributes to the affinity of the SIR complex for chromatin, we first compared the DNA-binding properties of the Sir2-Sir4 and Sir2-Sir4C complexes. Increasing amounts of each complex were titrated into a constant amount of a high-affinity histone octamer-binding sequence (Widom 601; [bib_ref] New DNA sequence rules for high affinity binding to histone octamer and..., Lowary [/bib_ref]. By using the binary Sir2-Sir4 complex rather than ternary complexes with Sir3, we could avoid contributions of Sir3 to DNA binding [bib_ref] Sir3-dependent assembly of supramolecular chromatin structures in vitro, Georgel [/bib_ref]. SIR complex association with DNA leads to the appearance of higher molecular weight species after native gel electrophoresis, and the disappearance of unbound DNA. We quantified the disappearance of the unbound DNA as a function of Sir2-Sir4 complex concentration. This showed that the truncated Sir2-Sir4C complex has about four-fold lower affinity than the full-length Sir2-Sir4 complex for naked DNA [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref].
We next examined the contribution of the Sir4 N-terminus to nucleosome binding, by titrating the complexes onto hexameric nucleosomal arrays assembled in vitro, as previously described [bib_ref] Reconstitution of yeast silent chromatin: multiple contact sites and O-AADPR binding load..., Martino [/bib_ref]. Again, by quantifying the disappearance of unbound nucleosomes, we found that the Sir2-Sir4C complex has roughly two-fold lower affinity for chromatin than the full-length Sir2-Sir4 complex [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref]. This effect was even more pronounced when we compared the binding of holo-SIR complex with that of the Sir2-Sir3-Sir4C complex. The complex carrying the truncated Sir4C had a much lower affinity for nucleosomal arrays than that containing full-length Sir4 [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref] , possibly because Sir3 sterically masks part of Sir4C's chromatin-binding surface [bib_ref] Reconstitution of heterochromatin-dependent transcriptional gene silencing, Johnson [/bib_ref] [bib_ref] Structural basis for the role of the Sir3 AAA + domain in..., Ehrentraut [/bib_ref].
Since the SIR complex is known to protect nucleosomal linker DNA from micrococcal nuclease (MNase) attack [bib_ref] The dual role of H4K16 acetylation in the establishment of yeast silent..., Oppikofer [/bib_ref] [bib_ref] Reconstitution of yeast silent chromatin: multiple contact sites and O-AADPR binding load..., Martino [/bib_ref] , we examined the contribution of Sir4N to linker DNA protection. Importantly, we used two-to four-fold more of the truncated Sir2-Sir4C complex than of wild-type Sir2-Sir4 complex, to ensure that equal amounts of chromatin-SIR complex were formed [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref]. The Sir2-Sir4C complex showed less linker DNA protection than the full-length Sir2-Sir4 complex [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref] , despite the fact that equal fractions of nucleosomes were bound in each reaction. These data suggest that the affinity of Sir4N for DNA promotes a tighter binding of SIR complexes to chromatin, thereby enhancing linker DNA protection. This attributes a function to the Sir4 N-terminus beyond recruitment by Sir1 or Yku80.
## Truncated sir4 mediates formation of sir3 foci independently of silencing
Silencing at telomeres is sensitive to the anchorage and clustering of the telomeres at the nuclear envelope [bib_ref] Perinuclear localization of chromatin facilitates transcriptional silencing, Andrulis [/bib_ref] [bib_ref] Evidence for silencing compartments within the yeast nucleus: a role for telomere..., Maillet [/bib_ref] [bib_ref] The functional importance of telomere clustering: global changes in gene expression result..., Taddei [/bib_ref]. Since Sir4C can restore silencing at HM loci but not at telomeres, we wondered whether Sir3 focus formation, as an indication of telomere clustering, might depend on Sir4N. To test this hypothesis, we expressed a Sir3-EGFP fusion protein in yeast cells in which the endogenous SIR4 gene was either deleted (sir4D) or truncated (sir4N), and either SIR4 or SIR4C was expressed from a CEN-ARS plasmid. Similar to a strain without tagged Sir3 [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] , S2B), both Sir4C and full-length Sir4 restored matingtype repression in the SIR3-EGFP background, but only full-length Sir4 was able to restore TPE fully [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref]. Although Sir4Cexpressing cells are competent to mate, they grow more slowly [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] and occasionally had larger and obviously distorted nuclei compared to wild-type cells. We imaged Sir3-EGFP in living cells and found that Sir3-EGFP foci formed when either Sir4 or the truncated Sir4C protein was expressed, as determined by counting the number of cells containing at least three Sir3-EGFP foci [fig_ref] Figure 6: Sir4C supports Sir3 focus formation in vivo [/fig_ref]. In about 20% of the Sir4C-expressing cells, we observed one to three very intense Sir3 foci in addition to the small telomere clusters, whether or not the endogenous SIR4 gene was present [fig_ref] Figure 6: Sir4C supports Sir3 focus formation in vivo [/fig_ref]. We conclude that the N-terminus of Sir4 is not necessary for the formation of Sir protein clusters, and thus that large Sir3-EGFP foci can form in the absence of TPE.
The dissociation of Sir3 foci from TPE confirms a recent report showing that non-perinuclear Sir3 clusters can form in cells unable to support SIR-mediated repression [bib_ref] Clustering heterochromatin: Sir3 promotes telomere clustering independently of silencing in yeast, Ruault [/bib_ref].
The Sir4 N-terminal domain is its major site of phosphorylation in G2/M-phase cells SIR complex binding at telomeres appears to be modulated in response to the physiological state of the cells. For instance, Sir proteins are released from telomeres both in mitotic cells [bib_ref] The dynamics of yeast telomeres and silencing proteins through the cell cycle, Laroche [/bib_ref] [bib_ref] Telomeric protein distributions and remodeling through the cell cycle in Saccharomyces cerevisiae, Smith [/bib_ref] , and in response to genotoxic stress [bib_ref] Relocalization of telomeric Ku and SIR proteins in response to DNA strand..., Martin [/bib_ref] [bib_ref] MEC1-dependent redistribution of the Sir3 silencing protein from telomeres to DNA double-strand..., Mills [/bib_ref]. Indeed, activation of the DNA damage checkpoint affects TPE, but not HM repression, much like the deletion of Sir4N. Moreover, subtelomeric domains contain a number of genes that are regulated in response to nutrient stress [bib_ref] Dynamics of Sir3 spreading in budding yeast: secondary recruitment sites and euchromatic..., Radman-Livaja [/bib_ref] [bib_ref] Sir3p phosphorylation by the Slt2p pathway effects redistribution of silencing function and..., Ray [/bib_ref] by a kinase cascade that targets, among other things, Sir3 [bib_ref] Regulation of subtelomeric silencing during stress response, Ai [/bib_ref]. Finally Sir4N harbors many potential phosphoacceptor sites, and whole phosphoproteome studies suggested that Sir4 is modified in a manner that fluctuates with the activity of the cellcycle regulated cyclin-dependent kinase (CDK; [bib_ref] Global analysis of Cdk1 substrate phosphorylation sites provides insights into evolution, Holt [/bib_ref] [bib_ref] Targets of the cyclin-dependent kinase Cdk1, Ubersax [/bib_ref]. Thus it was proposed that the N-terminal half of Sir4 might act as a phosphorylation-dependent regulatory domain [bib_ref] Co-evolution of transcriptional silencing proteins and the DNA elements specifying their assembly, Zill [/bib_ref].
To identify in vivo phosphoacceptor sites in Sir4, we first expressed a functional, epitope-tagged Sir4 from its endogenous locus. The Myc-tagged Sir4 protein was immunoprecipitated either from cycling cells or from cells that were arrested in G2/M phase by repressing CDC20, which encodes an essential anaphaseinitiating factor , S4B). We then used mass spectroscopy to identify the phosphorylated peptides. As predicted by Zill and colleagues [bib_ref] Co-evolution of transcriptional silencing proteins and the DNA elements specifying their assembly, Zill [/bib_ref] , most of the phosphorylated amino acid residues were in the N-terminal half of the protein [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref] and 7B, , [fig_ref] Table 2: Summary of Sir4 phosphopeptides identified [/fig_ref]. Moreover, ten of the twelve phosphorylated sites we identified contained the minimal consensus sequence for CDK, [S/T*]-P (in bold face in [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref] , [fig_ref] Table 2: Summary of Sir4 phosphopeptides identified [/fig_ref] , which can also be phosphorylated by the mitogenactivated protein kinase (MAPK). These sites were among those previously predicted to be targets for CDK and MAPK [bib_ref] Linear motif atlas for phosphorylation-dependent signaling, Miller [/bib_ref] [bib_ref] GPS 2.0, a tool to predict kinase-specific phosphorylation sites in hierarchy, Xue [/bib_ref].
Five of these [S/T*]-P sites showed at least 1.5-fold higher levels of phosphorylation in G2/M as compared to cycling cells [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref]. One of these was in the Sir4C domain (S1134), which, interestingly, is near a mapped SUMO-acceptor residue (K1128) within the Esc1-binding PAD domain [bib_ref] A proteomic strategy for gaining insights into protein sumoylation in yeast, Denison [/bib_ref]. However, mutation of this C-terminal phosphoacceptor site or the neighboring SUMO acceptor lysine yielded no detectable silencing-or anchoring-related phenotypes (data not shown). We therefore focused on the phosphoacceptor sites in the N-terminus of the protein. Within this domain, serine 63 (S63) was the site we detected most frequently over several experiments, while serine 84 (S84) showed a strict G2/M specificity.
To confirm the presence of CDK target sites within Sir4N, we exposed a recombinant Sir4N-GST fusion protein to a range of purified kinases in vitro. The recombinant protein was modified by CDK and protein kinase C, but not by yeast casein kinase II or the human MAP kinase, ERK [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref] , S5B). To examine whether the sites modified by CDK in vitro corresponded to the sites phosphorylated in vivo, we mutated CDK consensus sites within the Sir4N domain [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref]. We substituted consensus- . Sir4C can form a stable and active SIR complex in a recombinant system. A) SIR complexes as indicated were purified form coinfected insect cells. 1 mg of each complex was run on a SDS-PAGE and visualized by Coomassie staining. B) Purified Sir2-Sir3-Sir4 and Sir2-Sir3-Sir4C complex were incubated with histone octamers acetylated at H4K16 with or without the essential cofactor NAD. The deacetylation reaction was stopped after various time points by the addition of sample buffer and monitored by immuno blotting for H4K16 ac and H3, for equal loading. C, D) Sir2-Sir3-Sir4 or Sir2-Sir3-Sir4C complexes were analyzed by density gradient sedimentation. Fractions were run on 4-12% NuPAGEs Novex Bis-Tris Gels and stained with Sypro Ruby dye. Intensities of Sir2, Sir3 and Sir4 full length proteins were quantified (QuantityONE) and plotted in line graphs. The asterisk in D) indicates a Sir4 degradation band that runs very closely to Sir3. doi:10.1371/journal.pgen.1002727.g004 site threonines 7 and 13 by alanines (A7 and A13), and serines 63 and 84 by glycine residues (Sir4N GG ), alone and in combination [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref]. The mutant Sir4N domains were purified and used as substrates for phosphorylation by CDK in the presence of c 32 P-ATP. After trypsin digestion, the resulting phosphopeptides were resolved by high-resolution 1D gel electrophoresis [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref]. The identities of the cleavage products were determined both by co-migration with synthesized, digested peptides and by the absence of signal in the mutants in which serine or threonine had been replaced by non-phosphoaccepting amino acids [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref] and data not shown). Unlike the wild-type Sir4N protein, peptides carrying glycine substitutes at S63 and S84 lost almost all CDKmediated phosphorylation in vitro [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref] , [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref]. In contrast, alanine substitutions at T7 or T13 had only minor effects, alone or in combination. Given that S63 and S84 were phosphorylated by CDK both in vitro and in the endogenous protein recovered from mitotic cells, we propose that these two Sir4N residues are the major, physiological targets for CDK.
Mutation of phosphoacceptor sites does not alter Sir4N interaction with Yku80, Sir1, Sif2, or DNA To test the functional significance of Sir4N phosphorylation at S63 and S84 we analyzed the interactions of the non-phosphorylatable Sir4N mutant (Sir4N GG ) and the mutant carrying a mutation that mimics the phosphoserine residues (Sir4N DD ) with Sir1, Sif2 and Yku80. In yeast two-hybrid analyses, the interactions between Sir4N GG or Sir4N DD and Yku80, Sif2 or Sir1 were identical to the interactions between wild-type Sir4N and these binding partners [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref]. Thus, at least in this assay, substitution of S63 and S84 by either G or D does not perturb the binding of known ligands to Sir4N.
We next tested mutant and wild-type Sir4N fragments for their ability to bind DNA and protect linker DNA from MNase digestion. Intriguingly, the Sir4N GG mutant showed a higher affinity for DNA than the wild-type protein or the Sir4N DD mutant, suggesting that Sir4N phosphorylation might weaken its interaction with DNA [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref].The incubation of Sir4N with CDK in vitro prior to DNA binding, increased the affinities of both wild-type Sir4N and the non-phosphorylatable Sir4N GG mutant for DNA, due to nonspecific effects of the kinase [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref]. Indeed, by performing MNase digestion of nucleosomes bound by Sir4N GG , Sir4N DD , or the wild-type protein, we found equal protection of linker DNAs in all cases [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref]. We conclude that point mutations at these two major CDK phosphoacceptor sites in Sir4N do not substantially alter the affinity of the domain for either chromatin or DNA.
## Mutation of sir4n phosphorylation sites affects the stability of gene repression in vivo
Despite the absence of in vitro phenotypes for Sir4N bearing mutated S63 and S84 residues, we checked the effects of these two phosphoacceptor site mutations on silencing in vivo. To test this, we introduced the double mutations S63G-S84G (sir4 GG ) or S63D-S84D (sir4 DD ) into the endogenous SIR4 gene in a strain carrying both Tel5R::ADE2 and Tel7L::URA3 telomeric reporter genes. In the course of these experiments we serendipitously created an additional mutation at the Sir4 N-terminus, namely a proline to alanine substitution at residue 2 (sir4 P2A ). This substitution does not prevent cleavage of the initiator methionine or acetylation of the alanine at position 2 in contrast to the proline, and protein half life is predicted to be the same for either variant [bib_ref] N-terminal acetylation of cellular proteins creates specific degradation signals, Hwang [/bib_ref]. Indeed, we scored no significant effects on the half-life of Sir4 due to any of the mutations described above or below (data not shown).
We first tested the effects of the background P2A mutation alone on ADE2 silencing at Tel5R. We found that most sir4 P2A colonies were darker red than wild-type colonies , which indicates a more stable repression of ADE2. On the other hand, some colonies were completely white, indicating a low frequency of stable Tel5R::ADE2 reporter derepression . This suggested to us that the P2A mutation stabilizes either ''off'' or ''on'' epigenetic states at Tel5R. When the P2A mutation was combined with the non-phosphorylatable sir4 P2AGG mutation or the phospho-mimicking mutation (sir4 P2ADD ), we scored the same dark red color, but also noted that white colonies appeared at higher frequency . To quantify this phenomenon, we cultured single white or dark red colonies from each strain in liquid culture and plated them out after 24 h and 48 h to score the status of the ADE2 reporter by red vs white colony color. Whereas less than 1% of the sir4 P2A colonies switched from red (ADE2 repressed) to white (ADE2 derepressed), we found that 6-10% of the sir4 P2AGG or sir4 P2ADD colonies switched color, indicating that these mutations render the repressed state less stable .
Conversely, we found that 0.4-0.6% of sir4 P2A colonies switched from a derepressed to a repressed state (white to red), while the sir4 P2AGG or sir4 P2ADD strains remained completely derepressed, with no red colonies detected after 24 h of culturing of a white colony . This argues that mutation of S63 and S84 generally destabilizes silencing or impairs re-establishment of a repressed state. The effects are particularly noticeable in combination with the sir4 P2A mutation, which alone, for unknown reasons, stabilizes either state.
To test whether the effect of these mutations at Tel5R:ADE2 held true for another telomere, we spotted overnight cultures of single colonies onto uracil-deficient plates, to score for expression of the Tel7L::URA3 reporter gene . We observed loss of URA3 gene silencing in the sir4 P2AGG and sir4 P2ADD mutants. Moreover, the colonies growing on uracil-deficient plates were all white (ADE2 derepressed), indicating that Tel7L and Tel5R reporter genes were derepressed simultaneously . The loss of Tel7L::URA3 repression was far more pronounced for the white colonies of sir4 P2AGG and sir4 P2ADD strains than for the sir4 P2A mutation alone, arguing that alteration of the phosphorylation sites does indeed enhance derepression.
We next investigated whether the Sir4 phospho-site mutants caused general disruption of telomeric silencing when this is scored by growth on rapamycin [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref]. Indeed, the white (derepressed) colonies of the sir4 P2ADD and sir4 P2AGG strains showed more resistance to rapamycin than the red (repressed) colonies, and were almost as resistant as the sir4D strain . Quantification showed that this effect was far more pronounced in the sir4 P2ADD and sir4 P2AGG strains, than in the sir4 P2A strain (15% for sir4 P2A , 49% for sir4 P2AGG and 69% for sir4 P2ADD ). As in the switching assay, the phospho-mimicking mutation sir4 P2ADD produced the strongest derepressed state and strongest rapamycin resistance .
We generalized this observation by scoring mRNA levels at native subtelomeric genes. This was done both by mRNA analysis for genes at two natural telomeres, Tel6R and Tel9R [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] and by whole genome tiling arrays, that compared gene expression in wild-type SIR4 cells with either red or white colonies of the sir4 P2ADD mutant. The data in confirmed derepression at both natural telomeric genes and at Tel7L::URA3 in white sir4 P2ADD colonies, although not to the degree detected in the sir4D stain . We also observed derepression of HML in the white sir4 P2ADD cells. On the other hand, when we examine expression in the red sir4 P2ADD colonies, we find that subtelomeric genes are as stably repressed in the red sir4 P2ADD mutant cells as they are in a SIR4 wild-type strain . The microarray data confirmed this trend for subtelomeric genes . By overlaying the transcriptional effects on all genes as a function of their distance from the telomere (red line is lowess smoothed over all genes), we score an increase in expression from genes within the first 5 kb from the telomere in both the sir4D and the white sir4 P2ADD strains . We also observed a generally phosphorylated peptides of each digest was expected to be 1 and the corresponding correction factors were used for normalization of the phosphorylated peptides. For the five non-phosphorylated peptides the ratio average is displayed with error bars as standard deviation. C) In vitro phosphorylation followed by partial trypsin digestion of recombinant Sir4N and indicated Sir4N phosphosite mutants. The tryptic peptides were separated by high resolution SDS-PAGE and analyzed by radiography. Peptide sequences to the right indicate the migration pattern of trypsindigested in vitro phosphorylated standard peptides containing the indicated phospho-serine or -threonine residues D) Interaction of Sir4N with known interaction partners was analyzed by yeast two-hybrid analysis. Sir4N and Sir4N GG or Sir4N DD mutants were used as prey, for Yku80, Sir1 and Sif2 bait constructs that induce expression of the b-galactosidase gene upon interaction. At least three independent experiments were averaged for each value; data represent mean value 6 s.e.m. E) Sir4N fragment indicated in [fig_ref] Figure 7: Sir4N is the major site of phosphorylation [/fig_ref] and the respective mutants were expressed and purified from E. coli. DNA binding was performed and analyzed as in [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref] ; data represent mean value 6 s.e.m of three independent experiments. F) Sir4N fragments were bound to 6 mer arrays of nucleosomes and challenged with increasing amounts of MNase as in [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref]. Quantification of two independent experiments was used, data represent mean values. doi:10.1371/journal.pgen.1002727.g007 . Sir4N phosphoacceptor site mutants show increased accumulation of active states and overall derepression of TPE. A, B) Single colony streaks for ADE2 color assay monitor silencing of the indicated mutant strains (GA6018, GA5887, GA5888, GA5822, GA503). C,D) Quantification of cells swapping from a silent red to a de-repressed white state or vice versa. Single colonies were grown for the indicated time and dilutions plated on YPAD to monitor colony color. E) Single white or red colonies as in /8D were grown overnight and then spotted in stronger repressed state in the sir4 P2ADD red colony for the genes that lie closest to telomeres, as expected. Thus the effects of the mutations on the ADE2 reporter can be confirmed and extended to native subtelomeric genes.
[formula] URA3 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - [/formula]
In summary, thanks to the stabilizing effect of the sir4 P2A background, we are able to demonstrate a defect in the sir4 P2ADD and the sir4 P2AGG strains in silencing. The derepressed state suggested by white colony color coincided with general derepression of TPE and with resistance to rapamycin, as reported for the sir4D strain. We note that the phenotypes were somewhat stronger for the phospho-mimicking sir4 P2ADD mutant than for sir4 P2AGG . Taken together, these results indicate that the Sir4 N-terminus helps modulate stable gene repression at telomeres by being phosphorylated on the target sites we have identified.
# Discussion
SIR-mediated transcriptional repression in budding yeast has been studied genetically and biochemically for over 20 years, yet we still do not fully understand the functions of its core components nor how it is regulated either during the cell cycle or in response to stress. In this study we addressed the molecular role and regulation of Sir4, the largest and least conserved Sir protein. On the basis of our new findings and a large body of earlier work, we can assign four roles to different domains of Sir4 and describe their functions in SIR-mediated repression. First, as described previously, the C-terminal half of the Sir4 protein has a scaffolding function that is essential for assembling Sir2 and Sir3 into the SIR complex and delivering it to chromatin (by binding to Rap1 and Yku). Second, the N-terminal 270 residues of Sir4 have a recruitment function by binding Sir1 and Yku80. Third, the Sir4N contributes to the tight association of the SIR complex with DNA in vitro and enhances nucleosomal linker protection. This domain is essential in vivo for repression of HM loci under suboptimal conditions (i.e. when the silencers in HMR are compromised or when Yku70 is absent) and contributes significantly to TPE. Finally, we find that both the extreme N-terminus and the adjacent central domain of Sir4 are heavily phosphorylated in vivo and that the mutation of two phosphoacceptor sites in the N-terminus affects the stability of subtelomeric repression. Given that Sir4 residues S63 and S84 are phosphorylated in mitotic cells, we speculate that the phosphorylation of Sir4N regulates the stability of TPE through the cell cycle.
The C-terminus of Sir4 is sufficient to establish a silent chromatin structure that both the Nand C-terminal domains of Sir4 were required for silencing of HML. By using a slightly shorter and significantly more stable Cterminal domain of Sir4 than that used by Marshall and colleagues (residues 747-1358), we show that the Sir4 N-terminus is dispensable for repression of intact HML and HMR loci, although not for the repression of subtelomeric reporter genes. The ability of Sir4C to silence HM reflects, in part, the strong redundancy in Sir factor recruitment pathways at HM loci, i.e. several silencer factors redundantly recruit Sir3 and Sir4 [bib_ref] A yeast silencer contains sequences that can promote autonomous plasmid replication and..., Brand [/bib_ref].
Consistent with our finding that Sir4C is sufficient to silence intact HMR in the presence of Sir2 and Sir3, we find that recombinant complexes containing Sir2, Sir3, and Sir4C are stable upon isolation, retain full histone H4K16 deacetylation activity and bind nucleosomal arrays in vitro. H4K16 deacetylation by Sir2 has been proposed to provoke a conformational change in the SIR complex that increases its binding to chromatin [bib_ref] The dual role of H4K16 acetylation in the establishment of yeast silent..., Oppikofer [/bib_ref] [bib_ref] Reconstitution of yeast silent chromatin: multiple contact sites and O-AADPR binding load..., Martino [/bib_ref] [bib_ref] Assembly of the SIR complex and its regulation by O-acetyl-ADP-ribose, a product..., Liou [/bib_ref]. We do not know whether in the SIR complex the Nterminal domain of Sir4 contributes to this change in conformation, although we note that Sir4C-containing complexes retain deacetylation activity.
We explain the inability of a longer Sir4C fragment to repress (as described in [bib_ref] Functional domains of SIR4, a gene required for position effect regulation in..., Marshall [/bib_ref] by its instability [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref]. Indeed, this is consistent with an earlier study of the san1-1 mutant, which partially restores mating in a strain expressing only a Sir4 C-terminal domain [bib_ref] Genetic and molecular characterization of suppressors of SIR4 mutations in Saccharomyces cerevisiae, Schnell [/bib_ref]. San1 is a ubiquitin ligase that targets misfolded proteins for degradation by the proteasome, and one of its targets is Sir4 [bib_ref] Sir Antagonist 1 (San1) is a ubiquitin ligase, Dasgupta [/bib_ref]. Our work and this study suggest that a sufficiently stable Sir4C fragment provides all the essential interactions necessary for formation of silent chromatin at HM loci -most crucially a tight association with Sir2 and Sir3 to create an active, heterotrimeric complex that can interact stably with chromatin [bib_ref] A homotrimerheterotrimer switch in Sir2 structure differentiates rDNA and telomeric silencing, Cubizolles [/bib_ref] [bib_ref] Reconstitution of yeast silent chromatin: multiple contact sites and O-AADPR binding load..., Martino [/bib_ref]. It is not, however, sufficient for TPE.
## The sir4 n-terminus enhances sir4 binding to hm loci via protein-and dna-interactions
Although the Sir4 C-terminus is sufficient for HM repression under certain conditions, we also found conditions that render the N-terminus essential for efficient HM silencing, namely when one of the recruiting elements was deleted at the HMR-E silencer (DA Orc1-Sir1 site or DB Abf1 site), or in the absence of YKU70. This weakens the recruitment of Sir3 or Sir4 to the HM silencers. The requirement for Sir4N under these conditions is consistent with its ability to bind Sir1 and Yku80. Sir1 recruits Sir4 to HM silencers by direct interaction with Orc1 [bib_ref] Role of interactions between the origin recognition complex and SIR1 in transcriptional..., Triolo [/bib_ref]. The Yku70/80 complex can stabilizes silent chromatin by two means, first by recruiting the HM loci to Sir-clusters at the periphery [bib_ref] Live imaging of telomeres: yKu and Sir proteins define redundant telomereanchoring pathways..., Hediger [/bib_ref] [bib_ref] Sir-mediated repression can occur independently of chromosomal and subnuclear contexts, Gartenberg [/bib_ref] and second by helping form a promoter-silencer interaction at HM loci through a looping mechanism [bib_ref] Long-range communication between the silencers of HMR, Valenzuela [/bib_ref] [bib_ref] RAP-1 factor is necessary for DNA loop formation in vitro at the..., Hofmann [/bib_ref] [bib_ref] High-resolution structural analysis of chromatin at specific loci: Saccharomyces cerevisiae silent mating..., Weiss [/bib_ref]. Nevertheless, Yku70/80 is only essential for mating in the absence of Sir1 and vice versa [bib_ref] The Ku complex in silencing the cryptic matingtype loci of Saccharomyces cerevisiae, Patterson [/bib_ref] [bib_ref] The DNA end-binding protein Ku regulates silencing at the internal HML and..., Vandre [/bib_ref]. Consistent with this, we observe that Sir4N enhances mating efficiency in the absence of Yku70.
These two interactions, however, are probably not the only functions of the Sir4 N-terminus in HM silencing. First, there are two other sites of contact between Sir4 and the Yku70/80 complex [bib_ref] Silencing factors participate in DNA repair and recombination in Saccharomyces cerevisiae, Tsukamoto [/bib_ref] [bib_ref] Separation of silencing from perinuclear anchoring functions in yeast Ku80, Sir4 and..., Taddei [/bib_ref] , and second, expression of a Sir4N-C fusion protein represses an HMR locus that lacks one of its binding sites for ORC (HMR-EDA; [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref]. This indicates a function for the Sir4 Nterminus that is independent of its interaction with Sir1. We propose that the additional function is its strong non-specific affinity for DNA, which contributes to a tighter interaction of the dilution series onto YPAD plates for ADE2 color development, or on plates lacking uracil. To test rapamycin sensitivity, cells were additionally spotted onto SC plates containing 2.5 nM rapamycin. Colony growth was quantified as % of survivors growing on plates lacking uracil or containing 2.5 nM rapamycin. Results are plotted in the bar graph (two independent isogenic strains, each scored in 4-8 experiments; combined data are represented as mean value 6 s.e.m). We note that the sir4 P2A effect is stronger at Tel5R than at Tel7L, possible because the silencing of reporters at Tel5R is much weaker to begin with [bib_ref] Position effect at S. cerevisiae telomeres: reversible repression of Pol II transcription, Gottschling [/bib_ref] [bib_ref] Subtelomeric elements influence but do not determine silencing levels at Saccharomyces cerevisiae..., Mondoux [/bib_ref]. F) Relative quantification of mRNA of HML-APLHA1 and the three indicated subtelomeric genes (see [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] in white or red colonies recovered from SIR4, sir4D and sir4 P2ADD red and white colonies (GA503, GA5822, GA5887). Data from three biological replicates are represented as mean values 6 sem G) Microarray analysis of SIR4, sir4D and sir4 P2ADD red and white colonies (GA503, GA5822, GA5887). Plotted are the zero centered fold changes of log2 expression values of genes as a function of their distance from the telomere in relation to data for an isogenic SIR4 + strain. Black spots represent single genes, the red line is lowess smoothed over all genes. doi:10.1371/journal.pgen.1002727.g008 SIR complex to chromatin and to enhanced linker DNA protection in vitro.
## Full-length sir4 is necessary for full subtelomeric repression
At telomeres, the Sir4 N-terminus is required for TPE. Moreover, while the expression of a Sir4N-C fusion slightly reduces mRNA levels compared to Sir4C, it could not suppress reporter genes when their promoters were induced. Since Sir1 is not required for subtelomeric repression, these effects are independent of Sir1 [bib_ref] Modifiers of position effect are shared between telomeric and silent mating-type loci..., Aparicio [/bib_ref]. Rather, we suggest that the DNA-binding affinity of Sir4N increases binding of the SIR complex to telomeres to enhance the stability of repression. Full TPE, however, appears to require not only the first 270 amino acids of Sir4, but also the unstructured region between 270 and 744. Thus we suggest that another, yet unidentified function may be attributed to this domain.
Besides promoting tight association of the SIR holocomplex with DNA, the N-terminus of Sir4 may also regulate the strength or character of the Sir3-Sir4 interaction. We find that expression of Sir4C enhances the formation Sir3-EGFP foci even when there is no TPE. Similarly, Sir3-EGFP foci in the absence of TPE were observed in a strain overexpressing a non-acetylatable form of Sir3 [bib_ref] Clustering heterochromatin: Sir3 promotes telomere clustering independently of silencing in yeast, Ruault [/bib_ref] , indicating that Sir protein clustering does not always lead to gene repression. It is possible that the Sir4C protein has a stronger affinity for Sir3 than does full-length Sir4. This is consistent with an earlier hypothesis that the Sir4 N-terminus interferes the binding of Sir3 to Sir4 [bib_ref] Separationof-function mutants of yeast Ku80 reveal a Yku80p-Sir4p interaction involved in telomeric..., Roy [/bib_ref] [bib_ref] Silent information regulator protein complexes in Saccharomyces cerevisiae: a SIR2/SIR4 complex and..., Moazed [/bib_ref]. If true, the expression of Sir4C alone may lead to the sequestration of Sir3 into foci that antagonize repression. Consistent with this, we note that telomeres are highly sensitive to changes in Sir protein levels [bib_ref] Targeting Sir proteins to sites of action: a general mechanism for regulated..., Cockell [/bib_ref] , and that Sir4C expression is somewhat toxic to cells, while that of the Sir4N-C fusion is not [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] and data not shown).
The N-terminus is the major site of Sir4 phosphorylation and fine-tunes subtelomeric stress-response genes In this study we characterize Sir4 as a phosphoprotein and map key phosphoacceptor sites in the N-terminal domain of the protein. We show that two sites that are phosphorylated by CDK in vitro are also phosphorylated in mitotic cells in vivo. Zill and colleagues [bib_ref] Co-evolution of transcriptional silencing proteins and the DNA elements specifying their assembly, Zill [/bib_ref] speculated that the N-terminus of Sir4 may be specialized for fine-tuning or regulating silencing in response to environmental factors. Indeed, we find that the N-terminus of Sir4 is its major site of phosphorylation in vivo.
What functions of Sir4 might be affected by its phosphorylation? As Sir4N is dispensable for HM silencing, but is essential for TPE, we reasoned that the state of Sir4 phosphorylation might affect the repression of subtelomeric genes, many of which are activated only under conditions of nutrient stress [bib_ref] Regulation of subtelomeric silencing during stress response, Ai [/bib_ref]. Phosphorylation of the Sir4N terminus by CDK may also destabilize silent chromatin in mitosis. Consistent with this, previous work has shown that SIR complexes are partially released from telomeres in mitotic yeast nuclei [bib_ref] The dynamics of yeast telomeres and silencing proteins through the cell cycle, Laroche [/bib_ref] [bib_ref] Telomeric protein distributions and remodeling through the cell cycle in Saccharomyces cerevisiae, Smith [/bib_ref] , and yeast heterochromatin is most accessible to transcription factors and gene activation during G2/M phase [bib_ref] Overcoming telomeric silencing: a transactivator competes to establish gene expression in a..., Aparicio [/bib_ref]. Moreover, passage through mitosis, which is accompanied by CDK-dependent protein phosphorylation and dephosphorylation, is important for the establishment of silencing [bib_ref] Cell cycle requirements in assembling silent chromatin in Saccharomyces cerevisiae, Kirchmaier [/bib_ref] [bib_ref] Role of DNA replication in the repression of silent mating type loci..., Miller [/bib_ref] [bib_ref] Heterochromatin spreading at yeast telomeres occurs in M phase, Martins-Taylor [/bib_ref]. Although it is not clear why, we note that the partial release of factors from chromatin during mitosis is a common feature of eukaryotes. Heterochromatin protein 1 (HP1; [bib_ref] Heterochromatin protein 1 distribution during development and during the cell cycle in..., Kellum [/bib_ref] and the Polycomb complex (PcG; [bib_ref] The distribution of polycomb-group proteins during cell division and development in Drosophila..., Buchenau [/bib_ref] are both partially released during mitosis in Drosophila. The Polycomb protein EZH2 is a direct target of CDK [bib_ref] CDK1-dependent phosphorylation of EZH2 suppresses methylation of H3K27 and promotes osteogenic differentiation..., Wei [/bib_ref] [bib_ref] Cyclin-dependent kinases regulate epigenetic gene silencing through phosphorylation of EZH2, Chen [/bib_ref] , and HP1 release appears to be due to phosphorylation of histone H3 on Ser10 by Aurora B kinase [bib_ref] Regulation of HP1-chromatin binding by histone H3 methylation and phosphorylation, Fischle [/bib_ref] [bib_ref] Histone H3 serine 10 phosphorylation by Aurora B causes HP1 dissociation from..., Hirota [/bib_ref].
In yeast, we show that the substitution of the Sir4 phosphoacceptor sites S63 and S84 by acidic amino acids, or by nonphospho-accepting glycines, had only minor effects on the TPE [fig_ref] Figure 6: Sir4C supports Sir3 focus formation in vivo [/fig_ref]. However, by combining these mutations with a fortuitous mutation of residue 2 (proline to alanine or P2A), we could observe that they indeed tend to derepress TPE. The sir4 P2A mutation alone had a strong stabilizing effect on either silent or open epigenetic states, in that ADE2 repression at Tel5R was seen to be either enhanced or abolished. The phospho-acceptor site mutants (sir4 P2ADD or sir4 P2AGG ) has a tendency to derepress TPE and thereafter to retain the derepressed state, which is readily visible thanks to the stabilizing effect of sir4 P2A . Indeed, reestablishing a repressed state occurred less frequently in the sir4 P2ADD or sir4 P2AGG mutant cells. Importantly, we then showed by microarray analysis that this was accompanied by a general derepression of subtelomeric genes, similar -albeit less pronounced -to that observed in sir4D cells. Together our analysis suggests that the modification of Sir4N S63 and S84 CDK target sites either directly derepresses TPE or interferes in the reestablishment of a repressed state. We suggest that the Sir4 Nterminal domain regulates repression both during the cell cycle and in response to environmental stress, which is most likely mediated by a MAP kinase cascade. Given the abundance of confirmed CDK phosphoacceptor sites in this domain, and putative MAP kinase sites, this may be one of the main functions of Sir4 N-terminus.
# Materials and methods
## Plasmids, strains, and yeast methods
All strains and plasmids are described in Text S1. Standard techniques were used for cloning, yeast strain generation and growth. To obtain Sir4 expression similar to the endogenous levels, the Sir4 locus (1 kb 59 of start and 250 bp 39) was cloned into a CEN-ARS plasmid. Introduction of a NcoI site at the Sir4 start codon allowed subcloning of shorter Sir4 fragments and introduced the P2A mutation. To introduce the Sir4 phosphosite mutations into the genome, the respective plasmids were digested with SalI and SacI, purified fragments were integrated into a sir4::KanMX6 (GA5822) strain. Positive clones were selected for growth on SC medium+0.1% 5-FOA and checked by sequencing of the genomic locus. At least two independent transformants were analyzed.
To check Sir4 expression levels, the CEN-ARS plasmids were transformed into a protease-deleted strain (GA73) and cells were grown to OD 600 ,1 prior to lysis by bead-beating. Standard techniques were used for SDS-PAGE and immunoblotting. Mcm2 (yN-19) antibody is from Santa Cruz, anti Myc antibody 9E10. Sir2 and Sir4 antibody have been described previously [bib_ref] The clustering of telomeres and colocalization with Rap1, Sir3, and Sir4 proteins..., Gotta [/bib_ref] [bib_ref] A cytosolic NAD-dependent deacetylase, Hst2p, can modulate nucleolar and telomeric silencing in..., Perrod [/bib_ref]. Yeast two-hybrid analysis was carried out as described previously but using the GA181 strain [bib_ref] Interaction trap/two-hybrid system to identify interacting proteins, Golemis [/bib_ref] [bib_ref] Mechanistically distinct roles for Sgs1p in checkpoint activation and replication fork maintenance, Bjergbaek [/bib_ref].
## Silencing assays
For quantitative mating assays, plasmid transformed strains and the tester strain were grown overnight in SC medium. 10 7 cells of the mating-tester strain (GA858) were mixed in 1 ml YPAD medium containing 2610 6 cells transformed with a given Sir4 plasmids and grown for 5 h at 30uC (see also [bib_ref] Functional domains of SIR4, a gene required for position effect regulation in..., Marshall [/bib_ref]. Cells were then grown 3 days at 30uC on SD medium to select for diploids and SC medium -tryptophan/-methionine to normalize cell numbers.
Silencing of indicated reporter genes was performed as described [bib_ref] Functional characterization of the N terminus of Sir3p, Gotta [/bib_ref] , after growth overnight in selective media. Tenfold dilution series starting at 10 7 cells/ml were performed in triplicates on appropriate media. Silencing of the ADE2 reporter was scored after 3 days growth on YPAD 30uC and subsequent maintenance at 4uC for 4 days.
## Recombinant protein purification
Sir4C-expressing baculovirus was generated using the Baculo-Gold linearized cDNA (BD Bioscience) and Cellfectin reagent (Invitrogen) according to manufacturer's instructions. SIR complexes were co-expressed in insect sf21 cells and purified as described previously [bib_ref] The dual role of H4K16 acetylation in the establishment of yeast silent..., Oppikofer [/bib_ref] [bib_ref] Reconstitution of yeast silent chromatin: multiple contact sites and O-AADPR binding load..., Martino [/bib_ref]. Buffers contained 10 mM TEA pH8 when the proteins were purified for gradient sedimentation. For gradient sedimentation, 200 ml Calmodulin-column eluate was layered on a 4 ml 5-25% glycerol gradient (10 mM TEA pH8, 150 mM sodium chloride, 0.01% Tween-20) prepared using a Gradient Master (BioComp) in Beckmann 11661 polyallomer tubes. Gradients were centrifuged for 18 h at 4uC and 30'000 g and fractionated into 100 ul aliquots Samples of 20 ml were analyzed by SDS-PAGE and SyproRuby staining. Sir4N fragments were expressed in E. coli and purified from inclusion bodies using standard Ni-NTA techniques as described previously [bib_ref] Reconstitution of yeast silent chromatin: multiple contact sites and O-AADPR binding load..., Martino [/bib_ref].
## In vitro phosphorylation and phosphopeptide mapping
Recombinant Sir4N-GST was purified from E. coli using standard procedures in PBS buffer. For phosphorylation assays, the proteins were incubated with the indicated kinases for 1 h at 37uC (a kind gift of E. Nigg; CDK2 (NM_001798 Proqinase)) and P 32 -c-ATP. To analyze phosphorylation, Sir4N was run on a SDS-PAGE and proteins were detected by Coomassie staining and radiography. For phospho-peptide detection, Sir4N-GST was first digested by partial tryptic digest after in vitro phosphorylation. Peptides were lyophilized and analyzed on alkaline peptide gels and radiography according to the method of West and Bonner [bib_ref] Phosphorylation of histones in cells treated with hypertonic and acidic media, Pantazis [/bib_ref]. Briefly, samples were resuspended in loading buffer containing 0.125 M Tris-HCl pH 6.8 and 6 M urea. Peptides were then separated on 0.5 mm thick gels containing, 40% acrylamide, 0.037% bis-acrylamide, 0.75 M Tris-HCl pH 8.8. Gels were run at 10 mA for approximately 4 h. Peptide size was determined by co-migration with synthesized peptides that were phosphorylated and loaded alongside.
## Chromatin and dna binding, mnase digestion
Chromatin and DNA binding as well as MNase digestions were performed as described previously [bib_ref] The dual role of H4K16 acetylation in the establishment of yeast silent..., Oppikofer [/bib_ref] [bib_ref] Reconstitution of yeast silent chromatin: multiple contact sites and O-AADPR binding load..., Martino [/bib_ref]. Briefly, 6 mer of nucleosomes diluted to 2.5610 28 M were incubated with increasing amounts of indicated proteins in 10 mM TEA pH7.4, 25 mM sodium chloride for 20 min on ice and analyzed on 0.7% native agarose gels (0.26 TB) run at 4uC. DNA binding assays were performed using a 167 bp 601-Widom sequence DNA fragment Cy5-labeled by Klenow enzyme at an AvaI site. DNA bindings were performed in 150 mM sodium chloride at a DNA concentration of 2.5610 29 M, using the same conditions as for chromatin binding. Linker DNA protection assays were performed by adding 0.25-1 U MNase and 1 mM calcium chloride to 1 pmol chromatin that had been pre-incubated with indicated amounts of Sir proteins. After 10 min on ice 5 mM EGTA was added and samples were deproteinized by incubation with proteinase K for 30 min at 30uC. Deacetylation, expression and purification of homogeneously H4K16 reactions were carried out as described previously [bib_ref] The dual role of H4K16 acetylation in the establishment of yeast silent..., Oppikofer [/bib_ref]. 10 nM of purified Sir2-Sir3-Sir4 and Sir2-Sir3-Sir4C complex were incubated with 70 nM of H4K16 ac histone octamers with or without 150 mM of NAD. The reaction was carried out at 30uC in 50 mM Tris pH 8, 50 mM sodium chloride, 2.7 mM potassium chloride, 1 mM magnesium chloride and 0.005% Tween-20. The reaction was stopped at the indicated time points by addition of Laemmli sample buffer and samples were analyzed by 4-12% SDS PAGE by immuno blotting using H4K16 ac antibody (Millipore 07-329) and H3 antibody (Abcam ab1791).
## Immunoprecipitation and phosphosite mapping
Cells (GA5691, GA5589, GA1275) were grown overnight to OD 600 = 0.6 in YPA-galactose media, then shifted to YPA-glucose for 2 h at 30uC for mitotic arrest (G2/M cells). Cycling cells were grown in the same carbon source, but did not contain the GALp:CDC20 allele that leads to G2/M arrest in glucose. Cells were harvested, washed once in ice-cold PBS and resuspended in one pellet volume of lysis buffer without detergent (50 mM HEPES pH7.5, 500 mM sodium acetate, 5 mM magnesium acetate, 0.1 mM EDTA, 5% glycerol [bib_ref] A nonhistone proteinprotein interaction required for assembly of the SIR complex and..., Rudner [/bib_ref]. The resuspended cells were snap frozen in liquid nitrogen and broken using a ball-mill (363 min at 30 1/s; Retsch MM4000); the cell powder was stored at 280uC. For immunoprecipitations, the cell powder was mixed with an equal volume of lysis buffer containing protease and phosphatase inhibitors and 1% Triton X-100. After thawing on ice for 5 min, cell extract was cleared by centrifugation and 5 mg of proteins were incubated at 4uC with 50 ml of Affi-prep protein A beads (BioRad) crosslinked to 9E10 antibodies [bib_ref] Large-scale purification of the vertebrate anaphase-promoting complex/cyclosome, Herzog [/bib_ref]. Beads were washed with lysis buffer and stably bound proteins were eluted twice with 1.56 bead volume of 2 M glycine pH2 which was neutralized afterwards by Tris pH 8.0. For mass spectroscopy analysis, the eluates were processed by reduction and alkylation of the cysteines followed by sequential digestion with AspN and chymotrypsin or with trypsin only. The peptides were separated by nano-HPLC (Agilent 1100 nanoLC system, Agilent Technologies) coupled to an LTQ Orbitrap Velos hybrid mass spectrometer (Thermo Scientific) operated in positive mode using a top 5 DDA method. Inclusion lists were partially added to the method to search for expected peptides and to confirm already identified phosphorylated Sir4 peptides. Phosphorylated peptides and phosphosites were determined searching SwissProt data base restricted to S. cerevisiae using Mascot 2.3 (Matrix Science). Resulting sequences were inspected manually. Relative quantification was performed by integration of LC-MS extracted ion chromatograms. The peak areas of the corresponding phosphorylated peptides were normalized to the average of the peak areas of five non-phosphorylated Sir4 peptides. For prediction of CDK and MAPK sites, GPS2.1 software was used [bib_ref] GPS 2.0, a tool to predict kinase-specific phosphorylation sites in hierarchy, Xue [/bib_ref].
Microscopy C-terminally EGFP-tagged Sir3 (GA3128, GA6287, GA6288) was monitored in live cells grown to mid-log phase in SC medium and then embedded in an agarose pad as described [bib_ref] Visualizing yeast chromosomes and nuclear architecture, Meister [/bib_ref]. For quantification of Sir3-EGFP foci, all images were taken the same day and treated with the same threshold to quantify foci versus intense foci (above that threshold).
## Mrna purification, qpcr, and microarray
Cells of indicated strains/colony color were grown to OD 600 .0.6 and mRNA was purified using Qiagen mini RNeasy Kit. Reverse Transcriptase (RT) reaction was performed using ProtoScript AMV Kit (NEB#E6550). For QPCR, 0.5 ml of the RT reaction was used in a total volume of 10 ml using the GoTaq qPCR Master Mix (Promega, A6002), sybr green method and the ONE STEP fast cycler (ABI). For primers see Text S1. Values were normalized to ACT1 to account for samples differences and then to Sir4.
For microarrays, 100 ng of total RNA were amplified with the GeneChip WT Double-Stranded Target Assay (Affymetrix) and hybridized to GeneChip S. cerevisiae Tiling 1.0R Arrays following the ''GeneChip Whole Transcript (WT) Double-Stranded Target Labeling Assay Manual'' (Affymetrix) with a hybridization time of 16 h. The Affymetrix Fluidics protocol FS450_0001 was used for washing. Scanning was performed with Affymetrix GCC Scan Control v. 3.0.0.1214 on a GeneChip Scanner 3000 with autoloader. Raw data CEL files were read into R (version 2.14.1) using the Bioconductor (version 2.9) package Affy and a custom CDF package (available upon request). Probe sets were summarized and probe set-level values normalized with the RMA function. Gene coordinates for S. cerevisiae genes (EF3) were downloaded from Biomart (central.biomart.org) and chromosome length information was retrieved from the chromInfo table of the UCSC genome browser (genom.ucsc.edu) for SacCer_Apr2011/ sacCer3. Fold changes were calculated using the lmFit and eBayes functions as implemented in the limma package. Fold changes for telomeric genes were centered around zero and plotted against the distance to the closest chromosome end. Smoothing was performed with the lowess (locally weighted scatterplot smoothing) function and fold changes for each contrast were scaled and centered using the function scale. [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref] An additional 16amino acids renders the Sir4C fragment labile and unable to silence at HMR. A) Plasmids expressing Sir4C (residues 747-1358), a slightly longer Sir4 Cterminal fragment (residues 731-1358) or full length Sir4, under the Sir4 promoter and terminator, were transformed into yeast strains carrying a TRP1 reporter at HMR and the indicated genotype at the SIR4 locus (GA5886, sir4D; GA6072, sir4N (1-270)). Ten-fold dilution series were grown on plates selective for the plasmid (control) with or without tryptophan, to score for HMR repression. B) The plasmids indicated in A) were transformed into a strain lacking major vacuolar proteases and the SIR4 gene (GA73). Extracts of logarithmically growing cells were analyzed by immunoblotting using a Sir4 and Mcm2 antibody. Asterisks indicate Sir4 bands. (EPS) [fig_ref] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci [/fig_ref] Sir4C and Sir4N-C mediate mating type but not telomeric repression even in the absence of Rif1. A) Mating of cells expressing the Sir4N-C fusion. A Sir4 deletion strain (GA5822) was transformed with the indicated Sir4 plasmids and a ten-fold dilution series thereof was mixed in YPAD with a mating tester strain GA858. After incubation at 30uC, cells were spotted on plates lacking tryptophan for growth and on SD plates to score for mating. All Sir4 constructs support mating in the sir4D strain. Using quantitative mating assays as in [fig_ref] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR [/fig_ref] , we note that Sir4N-C cells mate less efficiently than Sir4C-expressing cells (efficiency is 0.73%60.11% compared to wild type Sir4), although they repress the HMR::TRP1 reporter more efficiently. We think that our mating assay overestimates the efficiency of mating of Sir4C cells, because the growth rate of these cells is much slower than wild-type or Sir4N-C-expressing cells (i.e. see [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] for slower growth and [fig_ref] Figure 6: Sir4C supports Sir3 focus formation in vivo [/fig_ref] for distorted nuclei). This gives the resulting diploids (now SIR4/SIR4C) a growth advantage over single colony haploids (SIR4C), which are used to normalize mating efficiency. B) Silencing of URA3 reporter at Tel7L as in [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] , just that a strain background expressing the first 270 residues of Sir4 (sir4N; GA5809) was used. C) To monitor the ADE2 reporter at Tel5R, the transformed cells (as in A) were spotted in ten-fold dilution series onto YPAD plates, and color developed after cell growth at 4uC. The reddish color indicates repression, which is manifest only in the wild-type strain (SIR4) or in sir4D complemented with a plasmid expressing full-length SIR4. D) Mating assay as in A) using strains as in C) to test silencing at HML of rif1D strains. All constructs except the vector control support HM repression and allow efficient mating. E) Telomeric silencing was tested in rif1D strains carrying both URA3 and ADE2 reporters as indicated, by ten-fold dilution series on plates lacking uracil or on YPAD media, respectively (GA503, GA7137, GA7144). F) Sir4 ChIP. Sir4, Sir4C and Sir4N-C were expressed in a sir4D (GA5822) strain and their binding to HML and telomeres analyzed by ChIP/QPCR using an antibody raised against Sir4C. Primers used for QPCR are in Supplementary Information S1. Bars represent averages of biological duplicates/ QPCR triplicates, data represent mean value 6 s.e.m. To the left are schemes of the HML loci and telomeres analyzed indicating the location of QPCR primer pairs (short black lines). (EPS) [fig_ref] Figure 3: Sir4C is not sufficient for silencing at telomeres [/fig_ref] Sir4C repression is not altered in sir4D cells containing Sir3-EGFP however Sir4C slightly de-represses in cells SIR4 expressing Sir3-EGFP. The indicated Sir4 plasmids were transformed into a yeast strain carrying Tel7L::URA3 and an EGFP tag on the endogenous Sir3 and the indicated genotype at the endogenous SIR4 locus (SIR4, GA3128; sir4N, GA6287; sir4D, GA6288). Ten-fold dilution series were grown on selective plates for the indicated plasmids (control), and the expression of the URA3 reporter was scored on plates lacking uracil or containing 0.1% 5-FOA. Sir4C expression derepresses slightly in a wild-type SIR4 background. For mating, the cells were mixed with a 106 excess of mater tester strain (GA858) on YPAD plates, grown overnight at 30uC and then replica plated on plates selecting for diploids. (EPS) Analysis of immunopurified Sir4 and MS spectrum of pS63. A, B) Immunopurified Sir4-Myc was separated on 4-12% NuPAGEs Novex Bis-Tris Gels and analyzed by silver staining or immunoblotting using antibodies against Myc and Sir2. A strain with untagged Sir4 was used as a specificity control (GA5589, GA5691). The full-length Sir4 band in A) is indicated with an asterisk. IN: cell extract, input of IP; P: pellet of whole cell extract; SN: supernatant after IP; IP: glycine eluates from beads. C) pS63 can be detected. CID spectrum of the Sir4 peptide SRPSTAIHTpSPHQPS (m/z 841.88) derived from a combined AspN and chymotrypsin digest. The neutral loss of phosphoric acid is indicated by '-98', the loss of water by '-18'. (EPS) [fig_ref] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA... [/fig_ref] In vitro phosphorylation of Sir4N by CDK does not influence DNA binding. A) In vitro phosphorylation of recombinant Sir4N with human Cyclin dependent kinase (CDK), Protein Kinase C (PKC), Casein kinase II (CKII) and Ets regulated kinase (ERK). For Sir4N, radiography and Coomassie staining are shown. For the control substrates H1, Casein and MBP (Myelin Basic Protein) only the radiography data is shown below the respective Sir4N lane. B) In vitro phosphorylation of Sir4N, Sir4N DD or H1 by CDK2/cycA kinase, shows that much CDK phosphorylation is lost when S63/S84 are mutated. C) DNA binding assay of Sir4N and Sir4N GG using the same conditions as in B) but without radioactive labeling. Three independent experiments were quantified, data represent mean values 6 sem. (EPS) [fig_ref] Figure 6: Sir4C supports Sir3 focus formation in vivo [/fig_ref] sir4 DD or sir4 GG mutations silence efficiently at telomeres. Sir4 phosphoacceptor site mutants as in / 8B but without the P2A mutation were integrated at the endogenous SIR4 locus into a strain carrying both the Tel5-R::ADE2 reporter gene and Tel7L::URA3 reporter gene. A colony of each strain was grown overnight and spotted onto YPAD for color visualization of the ADE2 reporter in ten-fold dilution series. Strains used are GA503, GA5822, GA6362, and GA6363.
## Supporting information
## (eps)
Text S1 The Text S1 file contains supplementary methods (ChIP) and tables for yeasts strains, plasmids and QPCR primers used in this study.
[fig] Figure 1: A truncated Sir4C is sufficient for silencing at HML and HMR. A) Scheme of Sir4 indicating important domains and their interactions. [/fig]
[fig] Figure 2: Sir4C is not sufficient for silencing at compromised HM loci. A) Quantitative mating assays were performed as in [/fig]
[fig] Figure 3: Sir4C is not sufficient for silencing at telomeres. A) Telomeric silencing was monitored by a Tel7L::URA3 reporter gene (GA503, GA5809, GA5822) expressing the indicated proteins from pRS, including the SIR4N-C fusion. Growth on plates containing 5 nM rapamycin (rapa) was also monitored. B) Relative mRNA levels of three different subtelomeric genes and HML-ALPHA1 were measured using QPCR. Bars represent averages of biological triplicates, data represent mean value 6 s.e.m. C) Scheme of the HM loci and telomeres analyzed, indicating additional recruiting elements and distances of promoters from nucleating elements. doi:10.1371/journal.pgen.1002727.g003 [/fig]
[fig] Figure 5: Sir4C has reduced affinity for DNA and chromatin and protects linker DNA less from MNase attack. A) Increasing amounts of Sir2-Sir4 or Sir2-Sir4C complexes were titrated into a fixed amount of 167 bp 601-Widom Cy5-labeled DNA. Samples were separated by native agarose gel electrophoresis and visualized by Cy5 fluorescence. Binding in three independent experiments was quantified by measuring the disappearance of the unbound DNA and normalized to input; data represent mean value 6 s.e.m. B) Sir2-Sir4 and Sir2-Sir4C complexes were titrated into constant amounts of 6 mer arrays of unmodified nucleosomes. Samples were analyzed as (A); chromatin was visualized by SybrSafe staining. C) Sir2-Sir3-Sir4 and Sir2-Sir3-Sir4C complexes were titrated into constant amounts of 6 mer arrays of unmodified nucleosomes as in (B), with only one experiment analyzed. D) Indicated concentrations of Sir2-Sir4 or Sir2-Sir4C were bound to chromatin as in (B) and incubated with increasing amounts of MNase for 10 min on ice prior to deproteinization. DNA was analyzed by agarose gel electrophoresis and SybrSafe staining. The amount of full length 6 mer DNA was quantified to monitor degree of digestion. Data from at least three experiments are represented as mean value 6 s.e.m. doi:10.1371/journal.pgen.1002727.g005 [/fig]
[fig] Figure 6: Sir4C supports Sir3 focus formation in vivo. A, B) Sir3-EGFP foci were monitored in logarithmically growing cultures using live microscopy. Sir3-EGFP was tagged at its endogenous locus and the strains carried the indicated forms of SIR4 (GA3128, GA6287, GA6288). Full-length Sir4 or Sir4C were added back on plasmids. Images were quantified by counting cells having .3 Sir3 foci at a low signal threshold or .1 Sir3 focus at a high threshold of equally treated images (n.240 cells/sample; .2 independent experiments; data represent mean value 6 s.e.m). C) Single focal planes of deconvolved images of Sir3-EGFP as above; size bar 1 mm. doi:10.1371/journal.pgen.1002727.g006 [/fig]
[fig] Figure 7: Sir4N is the major site of phosphorylation. A) Scheme of Sir4 as in Figure 1A, indicating identified phosphoacceptor sites of fulllength Sir4. Serine 63 and 84 mutated in subsequent experiments are indicated in red. Sites having the [S/T*]-P consensus are in bold. B) Relative quantification of the enrichment of Sir4 phosphopeptides in G2/M over cycling cells by LC-MS of a trypsin and of a combined AspN/chymotrypsin digest of a Sir4-IP experiment. For each digest the extracted ion chromatograms were integrated and the ratios of peptides detected in G2/M versus log growing cells were calculated for five phosphorylated as well as five non-phosphorylated Sir4 peptides. The ratio average of the five non- [/fig]
[table] Table 2: Summary of Sir4 phosphopeptides identified. [/table]
|
Accelerated whole breast irradiation in early breast cancer patients with adverse prognostic features
Purpose: Accelerated whole breast irradiation (AWBI) and conventional whole breast irradiation (CWBI) were compared to determine whether AWBI is as effective as CWBI in patients with early breast cancer and adverse prognostic features. Patients and methods: We included 330 patients who underwent breastconserving surgery (BCS) and post-operative radiation therapy (RT) using AWBI for pT1-2 and pN0-1a breast cancer from 2007 to 2010. These patients were matched with 330 patients who received CWBI according to stage, age (±3 years), and the year of BCS. AWBI of 39 Gy and CWBI of 50.4 Gy were given in 13 and 28 fractions, respectively.Results: Median follow-up time was 81.9 months. There were no statistically significant differences between the AWBI and CWBI groups in terms of age, stage, tumor grade, or molecular subtype. More patients with Ki-67 index ≥ 14% were present in the AWBI group (AWBI 47.0% vs. CWBI 10.3%; P<0.01). The 5-year ipsilateral breast tumor relapse (IBTR) rates for the AWBI and CWBI groups were 0.8% and 1.8%, respectively (P=0.54). High tumor grade was a statistically significant risk factor for IBTR (5-year IBTR rate: 2.9%; P=0.01). Ki-67 ≥ 14% was marginally related to IBTR (5-year IBTR rate: 2.2%; P=0.07). There were no statistically significant differences in the hazard ratios between the AWBI and CWBI groups according to any of the risk factors. There were no acute grade 3 toxicities in the AWBI group. There were no late grade 3 toxicities in either group.Conclusions: AWBI is comparable to CWBI in early breast cancer with adverse prognostic features.
# Introduction
Tumors arising from different tissues manifest diverse patterns of growth, disease progression, and response to radiation. The variation in biology among tumors derived from various tissues provides a rationale for using different radiation dose fractions among tumor types. The alpha/beta ratio indicates the radiosensitivity of a specific tissue. Rapidly growing tumors with high alpha/ beta ratios (6 to 14 Gy) are responsive to lower doses and thus are suitable for hyperfractionation. In contrast, slowly growing tumors with low alpha/beta ratios (1.5 to 5 Gy) can be controlled more effectively by a higher dose per fraction [bib_ref] A review of alpha/beta ratios for experimental tumors: implications for clinical studies..., Williams [/bib_ref]. Tumors of the breast tend to exhibit slow proliferation.
This understanding of breast radiobiology has justified the use of hypofractionation with fractional doses > 2 Gy in breast cancer, and precipitated several large randomized trials conducted on patients with early breast cancer [bib_ref] Long-term results of hypofractionated radiation therapy for breast cancer, Whelan [/bib_ref] [bib_ref] The UK Standardisation of Breast Radiotherapy (START) trials of radiotherapy hypofractionation for..., Haviland [/bib_ref] [bib_ref] Effect of radiotherapy fraction size on tumour control in patients with early-stage..., Owen [/bib_ref]. Among these trials, the Royal Marsden
## Clinical research paper
Oncotarget 81889 www.impactjournals.com/oncotarget Hospital/Sutton and Gloucestershire Oncology Center (RMH/SGOC) and Standardisation of Breast Radiotherapy (START) A trials compared hypofractionation using 3 Gy fractions with that using conventional 2 Gy fractions and reported comparable oncologic outcomes and satisfactory toxicity profiles. While hypofractionated whole breast irradiation in the RMH/SGOC trial was administered every other day over 5 weeks to balance the entire treatment period with that of conventional fractionation, in the START A trial, accelerated whole breast irradiation was accelerated by daily administration of 3 Gy fractions [bib_ref] The UK Standardisation of Breast Radiotherapy (START) trials of radiotherapy hypofractionation for..., Haviland [/bib_ref] [bib_ref] Effect of radiotherapy fraction size on tumour control in patients with early-stage..., Owen [/bib_ref]. The alpha/beta ratios for tumor control and for late effects were 4.6 Gy and 3.4 Gy, respectively, according to dose-response curves for breast cancer patients in both trials [bib_ref] Hypofractionated whole-breast radiotherapy for women with early breast cancer: myths and realities...., Yarnold [/bib_ref]. These data again advocate the use of hypofractionated regimens [bib_ref] Hypofractionated whole breast irradiation: new standard in early breast cancer after breast-conserving..., Kim [/bib_ref].
However, breast cancer as an entity is histologically heterogeneous [bib_ref] The pathology of invasive breast cancer. A syllabus derived from findings of..., Fisher [/bib_ref]. A worse prognosis for breast cancers with a high histologic grade has long been recognized [bib_ref] Histological grading and prognosis in breast cancer; a study of 1409 cases..., Bloom [/bib_ref] [bib_ref] Pathological prognostic factors in breast cancer. I. The value of histological grade..., Elston [/bib_ref]. More recently, with advancement of techniques in molecular biology, breast cancer has been further classified into luminal A, luminal B, HER2-positive, and triplenegative subtypes [bib_ref] Personalizing the treatment of women with early breast cancer: highlights of the..., Goldhirsch [/bib_ref]. Triple-negative breast cancer, especially exhibits rapid growth and is associated with shorter survival compared with other molecular subtypes [bib_ref] Triple-negative breast cancer, Foulkes [/bib_ref]. Triple-negative breast cancers are associated with higher Ki-67 indices. The Ki-67 index indicates the proliferative potential of tumor cells, and high Ki-67 index is a well-known factor for poor prognosis [bib_ref] Ki67 in breast cancer: prognostic and predictive potential, Yerushalmi [/bib_ref]. Nonetheless, subgroups exhibiting poor prognostic factors are also candidates for postoperative irradiation if staged as early breast cancer amenable to breast conservation therapy. The biological heterogeneity of breast cancer poses the question of whether hypofractionation is also applicable to tumors with adverse prognostic factors. To answer this question, accelerated whole breast irradiation (AWBI) and conventional whole breast irradiation (CWBI) were compared to determine whether AWBI is as effective as CWBI in patients with early breast cancer and adverse prognostic features. Although AWBI in early breast cancer has been evaluated in randomized trials, the efficacy of AWBI in biologically more aggressive subsets of breast cancer has seldom been explored. This is a report on the efficacy of AWBI in early breast cancers with adverse prognostic features.
# Materials and methods
## Patients
Patients with early breast cancer (pT1-2 and pN0-1a) who received breast-conserving surgery (BCS) and post-operative radiation therapy (RT) at the National Cancer Center (Goyang, Korea) from January 2007 to December 2010 were included. RT was administered as AWBI [bib_ref] Phase 2 trial of accelerated, hypofractionated whole-breast irradiation of 39 Gy in..., Kim [/bib_ref]. Out of the 343 eligible patients, 13 with secondary malignancies other than ductal carcinoma in situ, cervical carcinoma in situ, and thyroid cancer were excluded. Thus 330 total patients were included. Either sentinel lymph node biopsy or axillary lymph node dissection was performed in all patients for surgical axillary staging. Re-excision was permitted for tumors with involved or close margins. RT was started after completion of adjuvant chemotherapy. Hormonal therapy was commenced at the same time as RT. The tumor grade was assessed using the Nottingham grading system [bib_ref] Prognostic significance of Nottingham histologic grade in invasive breast carcinoma, Rakha [/bib_ref]. The molecular subtype was classified according to the St. Gallen consensus as follows: luminal A, luminal B, HER2 positive, and triple negative [bib_ref] Personalizing the treatment of women with early breast cancer: highlights of the..., Goldhirsch [/bib_ref]. Trastuzumab therapy was indicated for T2 or N1 tumors with HER2 receptor positivity. We matched these patients one to one with patients who received RT using CWBI after BCS in Seoul National University Hospital (Seoul, Korea). The patients were matched according to stage, year in which BCS was performed, and age (±3 years). A total of 330 CWBI patients were matched and compared. The Ki-67 index determined by immunohistochemistry was assessed manually by two pathologists in both institutions. The central review boards of both institutions approved the entire course of this study.
## Radiation therapy
The patients were placed on a breast board in the supine position for simulation. Both institutions used computed tomography (CT)-based simulation and 3D conformal planning for all patients. The whole breast with a margin of 1 cm was the planning target volume for whole breast irradiation in the AWBI group. The boost volume was the tumor bed, indicated with surgical clips, plus a 2 cm margin. The initial whole breast irradiation for the CWBI group was superiorly bordered by the sternoclavicular junction and inferiorly bordered by a parallel line 2 cm below the inframammary fold. The midline bisecting the anterior chest wall was the medial border. The mid-axillary line comprised the lateral border. The primary boost for the CWBI group was given to the same volume as that in the AWBI group. Neither group received irradiation to the supraclavicular or internal mammary nodal regions. Whole-breast irradiation was administered to all patients using 6 MV tangential fields, including axillary levels I to II, except to women with a large thorax who required 15 MV with wedges to achieve adequate dose distribution. An electron beam with energies ranging from 6 to 12 MeV was used for the tumor bed boosts. The electron energy was selected according to the depth of the tumor bed.
RT was given daily (Monday to Friday) at both centers. The AWBI group underwent whole breast irradiation of 39Gy in 13 fractions and consecutive www.impactjournals.com/oncotarget : Patient, tumor, and treatment-related characteristics
## : survival analyses of relapse in conventional whole breast irradiation (cwbi) and accelerated whole breast irradiation (awbi)
Oncotarget 81891 www.impactjournals.com/oncotarget boost on the surgical cavity of 9 Gy in three fractions. An extra fraction of 3 Gy up to 12 Gy in four fractions was administered to cases with a close margin. Whole breast irradiation in the CWBI group was delivered up to 50.4 Gy in 28 fractions followed by a boost of 9 Gy in five fractions. A boost of 14 Gy in seven fractions was delivered to those patients with a close margin.
## Follow-up
The first follow-up was at 2 to 3 months after the completion of RT. The patients were followed up again at 6 months and then yearly thereafter. At each follow-up, an interview and physical examination were performed, and clinical photographs were obtained. Edema, erythema/ hyperpigmentation, and wet desquamation, as radiationinduced skin toxicities, were evaluated. A four-point scale was used to grade these toxicities: 0 = none, 1 = mild, 2 = moderate, and 3 = severe [bib_ref] Clinical outcome and cosmesis in African-American patients treated with conservative surgery and..., Tuamokumo [/bib_ref]. The grading was performed at baseline, on the last day of RT, and at each follow-up. The skin toxicities were evaluated by the same clinician at each center.
## Endpoints and statistics
The tumor-, treatment-, and toxicity-related variables were compared between the AWBI and CWBI groups using the chi-squared test, t-test, and Fisher's exact test. Survival time was calculated from the date of BCS. Ipsilateral breast tumor relapse (IBTR) was defined as relapse in the treated breast. Loco-regional relapse (LRR) was defined as any relapse in the treated breast and/or regional lymphatic area. Distant metastasis (DM) was defined as relapse in distant organs outside the regional area. Relapse-free survival (RFS) was defined as the time to any first relapse including IBTR, LRR, or DM (whichever occurred first). Overall survival (OS) was defined as the time to death for the deceased patients or the time to the last follow up for the surviving patients. Kaplan-Meier survival analysis was used to calculate the survival rates, and the survival differences were assessed by the log-rank test. To identify risk factors for IBTR in the entire cohort, the rate of IBTR was compared between each adverse risk factor (e.g. Ki-67 index ≥ 14%) and its favorable counterpart (e.g. Ki-67 index < 14%) for all patients in the AWBI and CWBI groups combined. Then, the risk factors for IBTR were analyzed for the AWBI and CWBI groups separately. The hazard ratios (HRs) for IBTR incidence in the entire cohort, according to each risk factor, were estimated using the Cox proportional hazard model. The average of points for skin toxicity grades were calculated and compared between the AWBI and CWBI groups. An α level of 0.05 was used to define statistical significance. SPSS for Windows software (ver. 22; SPSS Inc., Chicago, IL, USA) was used for the statistical analyses.
# Results
## Characteristics of patients, tumors, and treatments
The median duration of follow up was 81.9 months (range: 3.8-119.7 months). summarizes the patient and treatment characteristics. The distributions of age, tumor and nodal stages, resection margin status, histologic grade, and molecular subtype were similar between the AWBI and CWBI groups. In both groups, approximately 90% of patients were diagnosed with a ductal histology. There was no statistically significant difference in the proportion of patients receiving hormonal therapy and trastuzumab therapy. However, in the AWBI group compared with the CWBI group, a significantly higher proportion of patients with a Ki-67 index ≥14% (AWBI 47.0% vs. CWBI 10.3%; P < 0.01) and a higher frequency of chemotherapy administration was observed (n = 249 [75.5%] vs. n = 193 [58.5%], P < 0.01; respectively).
## Survival analyses
The 5-year survival rates of the CWBI and AWBI groups are summarized in . At the time of analysis, there were four patients with IBTR in the AWBI group, with a 5-year IBTR rate of 0.8%. There were seven patients with IBTR in the CWBI group, with a 5-year IBTR rate of 1.8%. There was no statistically significant difference in the IBTR rate between the groups. The HR for IBTR in the AWBI group compared with the CWBI group was 0.68 (95% CI: 0.20-2.33; P = 0.54), as shown in .
Six patients (5-year event rate: 1.7%) in the AWBI group and eight patients (5-year event rate: 2.4%) in the CWBI group developed LRR (P = 0.78; . There were 6 and 10 cases of DM, with 5-year event rates of 1.9% and 3.0% in the AWBI and CWBI groups, respectively (P = 0.37). There were 17 patients with a first relapse at any site (5-year event rate: 4.5%) in the CWBI group and 10 patients (5-year event rate: 2.8%) in the AWBI group (P = 0.32). The HR for RFS in the AWBI group compared with the CWBI group was 0.67 (95% CI: 0.31-1.47; P = 0.32), as shown in . There was one death due to breast cancer in the AWBI group compared with three cases in the CWBI group (P = 0.35).
## Factors associated with ibtr
The factors that showed an association with IBTR are listed in . The 5-year IBTR rate for all patients Oncotarget 81892 www.impactjournals.com/oncotarget in both the AWBI and CWBI groups was 1.2%. Among the several risk factors analyzed in the entire cohort, high tumor grade was significantly related to a higher IBTR rate compared with low and intermediate grades (P = 0.01; . The 5-year IBTR rate for high-grade tumors was 2.9%, which was greater than that of low-to intermediategrade tumors by a factor of 10. A Ki-67 index ≥ 14% (5year IBTR rate: 2.2%) showed a greater association with IBTR, with marginal significance, compared with a Ki-67 index < 14% (5-year IBTR rate: 0.9%; P = 0.07). Although higher rates of IBTR were observed in the T2 (vs. T1), N1 (vs. N0), and non-luminal (vs. luminal) subgroups of the entire cohort, the differences were statistically insignificant.
In the AWBI group alone, the 5-year IBTR rate was 0.8%, and that for high-grade tumors was 1.9% compared with 0% for low-to intermediate-grade tumors (P = 0.08). Ki-67 index ≥ 14% showed a trend toward an association with a higher IBTR rate (5-year IBTR rate: 1.4%) compared with Ki-67 index < 14% (5-year IBTR rate: 0%) in this group (P = 0.08). Pathologic N1 stage showed a stronger association with IBTR compared with N0 stage in the AWBI patients, with marginal significance (P = 0.06).
In the CWBI group, the 5-year IBTR rate was 1.8%. In this group, no statistically significant risk factors for IBTR were identified. CWBI patients with a Ki-67 ≥ 14% had a 5-year IBTR rate of 6.0%, which was higher than the rate (1.4%) in patients with a Ki-67 < 14% by a factor of four, with marginal significance (P = 0.06).
## Figure 1: cumulative incidence of ipsilateral breast tumor relapse (ibtr) (a) and relapse-free survival (b) for patients after accelerated hypofractionation (awbi) or conventional fractionation (cwbi) radiation therapy www.impactjournals.com/oncotarget
## Comparison of hrs for ibtr in the risk factor subgroups
The HRs for IBTR according to IBTR risk factor are illustrated in . HRs in AWBI were compared with CWBI for each risk factor, and no statistically significant differences were detected. The HR for IBTR in patients with high-grade tumors, a significant risk factor for IBTR in all patients, was 0.77 (95% CI: 0.18-3.25; P = 0.72). The HR for IBTR in patients with a Ki-67 ≥ 14%, a marginally significant risk factor for IBTR in all patients, was 0.50 (95% CI: 0.08-3.11; P = 0.45). The HR for IBTR in patients with N1 stage, a marginally significant risk factor for IBTR in AWBI patients, was 2.18 (95% CI: 0.20-24.33; P = 0.53).
All of the risk factor subgroups had HRs < 1, except for the T2 (1.06; 95% CI: 0.15-7.63) and N1 (2.18; 95% CI: 0.20-24.33) subgroups, although none were statistically significant. AWBI was not inferior to CWBI with respect to the risk of IBTR in both the patient subgroups, i.e., those with favorable as well as adverse factors, such as age < 50 years (HR 0.60, 95% CI: 0.11-3.30), high tumor grade (HR = 0.77, 95% CI: 0.18-3.25), non-luminal subtype (HR = 0.90, 95% CI: 0.15-5.42), and Ki-67 index ≥ 14% (HR 0.50, 95% CI: 0.08-3.11).
# Treatment-related skin toxicities
No acute grade 3 toxicities were observed in the AWBI group, while 1 (0.3%) patient with grade : Factors associated with ipsilateral breast tumor relapse (IBTR) among patients who received accelerated whole breast irradiation (AWBI) and conventional whole breast irradiation (CWBI) www.impactjournals.com/oncotarget Oncotarget 81895 www.impactjournals.com/oncotarget 3 edema, 56 (17%) patients with grade 3 erythema/ hyperpigmentation, and 4 (1.2%) patients with grade 3 wet desquamation immediately after RT. There were no late grade 3 toxicities in either the AWBI or CWBI group.
The average points for toxicity grades were compared between the AWBI and CWBI groups . Immediately after RT completion, the average points for breast edema were significantly higher in the CWBI group (CWBI, 0.61 vs. AWBI, 0.38; P < 0.01). However, the AWBI group showed significantly higher points for breast edema at 6 months (AWBI, 0.58 vs. CWBI, 0.12; P < 0.01) and 1 year after RT (AWBI, 0.22 vs. CWBI, 0.07; P < 0.01). A statistically significant difference was no longer apparent by 2 years . Immediately after RT completion, the average points for erythema/ hyperpigmentation were significantly higher in the CWBI group compared with the AWBI group (CWBI, 1.68 vs. AWBI, 0.57; P < 0.01) and remained higher for up to 3 years; the difference was statistically significant .
# Discussion
Breast cancer, being a slow-growing tumor, is effectively controlled by hypofractionated RT, as demonstrated in several randomized trials [bib_ref] Long-term results of hypofractionated radiation therapy for breast cancer, Whelan [/bib_ref] [bib_ref] The UK Standardisation of Breast Radiotherapy (START) trials of radiotherapy hypofractionation for..., Haviland [/bib_ref] [bib_ref] Effect of radiotherapy fraction size on tumour control in patients with early-stage..., Owen [/bib_ref]. However, breast cancer also includes biologically more aggressive subtypes with a high proliferative potential and greater risk of recurrence. Historically, treatment acceleration using hypofractionation has shown clinical success in head and neck cancer, which is infamous for its rapid growth [bib_ref] Hyperfractionated or accelerated radiotherapy in head and neck cancer: a meta-analysis, Bourhis [/bib_ref]. To date, many studies have reported remarkable outcomes with fraction sizes > 2 Gy delivered to head and neck cancers [bib_ref] Multi-institutional trial of accelerated hypofractionated intensity-modulated radiation therapy for early-stage oropharyngeal cancer..., Eisbruch [/bib_ref] [bib_ref] Dose intensified hypofractionated intensity-modulated radiotherapy with synchronous cetuximab for intermediate stage head..., Thomson [/bib_ref] [bib_ref] Hypofractionated accelerated radiotherapy with concurrent carboplatin for locally advanced squamous cell carcinoma..., Chan [/bib_ref] [bib_ref] Head and neck squamous cell carcinoma in patients aged > or =..., Italiano [/bib_ref]. A recent study also demonstrated success and survival benefits with hypofractionation in non-small cell lung cancer [bib_ref] Long-term follow-up of patients with locally advanced non-small cell lung cancer receiving..., Walraven [/bib_ref]. The effective control of rapidly growing tumors, such as squamous cell carcinoma of the head and neck or non-small cell lung cancer, by accelerated hypofractionation suggests that accelerated whole breast irradiation may also be suitable for biologically more aggressive subsets of breast cancer with poor prognostic features.
Although randomization of poor prognostic factors has not been performed in hypofractionation trials to date, unplanned subgroup analyses in randomized trials reported comparable results for the hypofractionated arm among patients with adverse prognostic features. In a meta-analysis of the START trials, comparison of LRR demonstrated an HR of 0.86 (95% CI: 0.59-1.25) for hypofractionated regimens compared with conventional regimens in the high-grade tumor subgroup, although this was statistically insignificant [bib_ref] The UK Standardisation of Breast Radiotherapy (START) trials of radiotherapy hypofractionation for..., Haviland [/bib_ref]. A populationbased cohort study also demonstrated equivalent hypofractionation outocmes in high-grade tumors [bib_ref] The impact of hypofractionated whole breast radiotherapy on local relapse in patients..., Herbert [/bib_ref]. A subgroup analysis in a Canadian trial suggested a lower efficacy of hypofractionation in high-grade tumors. This finding may be due to the absence of a radiation boost [bib_ref] Long-term results of hypofractionated radiation therapy for breast cancer, Whelan [/bib_ref] ; although the patients included in this trial comprised a carefully selected population with T1-2N0 stage disease, the efficacy of a boost is well-established, even in those with early-stage cancer [bib_ref] Role of a 10-Gy boost in the conservative treatment of early breast..., Romestaing [/bib_ref] [bib_ref] Wholebreast irradiation with or without a boost for patients treated with breast-conserving..., Bartelink [/bib_ref]. Another explanation may be that an outdated tumor grading system, the Scharff Bloom Richardson grading system, which has now been replaced with the Nottingham grading system, was employed in the Canadian trial [bib_ref] Long-term results of hypofractionated radiation therapy for breast cancer, Whelan [/bib_ref]. The Nottingham grading system is more quantitative with higher reproducibility, and its prognostic value in IBTR after breast conservation therapy has been well-demonstrated [bib_ref] Prognostic significance of Nottingham histologic grade in invasive breast carcinoma, Rakha [/bib_ref] [bib_ref] Impact of pathological characteristics on local relapse after breastconserving therapy: a subgroup..., Jones [/bib_ref]. There was no difference in the risk of LRR in node-positive patients treated with hypofractionation (HR = 0.80, 95% CI: 0.57-1.11) in the START (Standardisation of Breast Radiotherapy) A and B trials [bib_ref] The UK Standardisation of Breast Radiotherapy (START) trials of radiotherapy hypofractionation for..., Haviland [/bib_ref].
Until today, no subgroup analysis of the Ki-67 index has been performed in a hypofractionation setting. However, Ki-67 is a widely investigated marker of tumor proliferation, and its prognostic and predictive roles in breast cancer have been evaluated [bib_ref] Ki67 in breast cancer: prognostic and predictive potential, Yerushalmi [/bib_ref]. In a meta-analysis of Ki-67 involving 12,155 early breast cancer patients, a high Ki-67 index was significantly related to worse prognosis, with an HR of 1.93 (95% CI: 1.74-2.14; P < 0.001) for disease-free survival (DFS) and an HR of 1.95 (95% CI: 1.7-2.24; P < 0.001) for OS [bib_ref] Ki-67 as prognostic marker in early breast cancer: a meta-analysis of published..., De Azambuja [/bib_ref]. A high Ki-67 index has been defined by researchers according to cut-off levels ranging from 5-30% [bib_ref] Ki-67 as prognostic marker in early breast cancer: a meta-analysis of published..., De Azambuja [/bib_ref]. Its predictive role in the adjuvant setting also has been shown in randomized trials on systemic therapy. The HR for relapse was 1.6 (95% CI: 1.2-2.3; P < 0.01) in the Ki-67 index > 20% subgroup in a randomized trial comparing different chemotherapy regimens [bib_ref] Ki67 expression and docetaxel efficacy in patients with estrogen receptor-positive breast cancer, Penault-Llorca [/bib_ref]. DFS was significantly lower in the Ki-67 index > 11% subgroup, with an HR of 1.8 (95% CI: 1.4-2.3; P = 0.0001), in a randomized trial comparing different hormonal agents [bib_ref] Prognostic and predictive value of centrally reviewed Ki-67 labeling index in postmenopausal..., Viale [/bib_ref].
In this study, high tumor grade was significantly related to IBTR in the entire cohort (both the AWBI and CWBI groups). Another risk factor associated with a higher risk of IBTR, with marginal significance, in all patients was a Ki-67 index ≥ 14%. N1 stage was related to IBTR, with marginal significance, in the AWBI group alone. There was no difference in the risk of IBTR between the AWBI and CWBI arms with respect to the poor prognostic factors evaluated, including high tumor grade, Ki-67 ≥ 14%, and nodal positivity. As shown previously in randomized trials, the non-inferiority of AWBI was also demonstrated by our data. This study further supports the use of AWBI for breast cancers with adverse prognostic features.
However, we must consider the different rates of Ki-67 ≥ 14% in the AWBI and CWBI groups, owing to the cut-off value of 14% within the gray zone of 11-30%, which has only fair inter-observer reproducibility [bib_ref] An interobserver reproducibility analysis of Ki67 visual assessment in breast cancer, Shui [/bib_ref]. This is because assessment of Ki-67 expression by manually counting stained cells is labor intensive, www.impactjournals.com/oncotarget and estimating the percentage of stained cells has low reproducibility [bib_ref] Ki67 in breast cancer: prognostic and predictive potential, Yerushalmi [/bib_ref]. There is a clear consensus on the modest reproducibility of Ki-67 among pathologists, and thus assessments by two pathologists or automated analysis has been recommended. Despite the improvement in reproducibility with automated analysis, a higher risk of counting normal cells with higher Ki-67 expression is associated with automated analysis, whereas pathologists are able to differentiate tumor cells from normal cells [bib_ref] Ki67 in breast cancer: prognostic and predictive potential, Yerushalmi [/bib_ref]. Therefore, the risk of counting normal cells in an automated analysis may be overcome if the reproducibility of manual counting is improved by employing two pathologists. Both centers performed two assessments by two pathologists. In an international study on the reproducibility of Ki-67 expression analysis, in contrast to high intra-laboratory reproducibility, the inter-laboratory reproducibility was only moderate, and the mean Ki-67 values varied up to 30% [bib_ref] An international Ki67 reproducibility study, Polley [/bib_ref]. This suggests that the difference in the rate of Ki-67 index ≥ 14% in our study was due to its multi-institutional design. Nonetheless, more patients with a Ki-67 index ≥ 14% and fewer patients with IBTR were observed in the AWBI group. These results suggest that AWBI is not inferior to CWBI with respect to local control of more aggressive tumors.
With regard to skin toxicity, the longer treatment time of CWBI induced a higher rate of acute toxicities with more severe conditions, such as grade 3 toxicities. The incidence of erythema/hyperpigmentation was higher in the CWBI group immediately after RT and remained so for up to 3 years. Breast edema was observed more frequently in the AWBI group at 6 months and 1 year after RT, but the statistically significant difference was no longer apparent after 2 years. Thus, in terms of hyperpigmentation, AWBI was favorable over CWBI. However, these data should be interpreted with caution, because different physicians from each center graded the skin toxicities. In the RMH/ SGOC trial, which used the same fraction size as that in our study, AWBI was superior to CWBI in terms of late toxicities in all clinical assessments, including cosmesis, breast edema, shrinkage, distortion, and induration up to 10 years. Nevertheless, the estimated rate of any change in breast appearance by photographic assessment at 10 years was 2.7% higher in the AWBI than CWBI arm [bib_ref] Fractionation sensitivity and dose response of late adverse effects in the breast..., Yarnold [/bib_ref]. Therefore, further development of the data is needed in our study.
The major limitation of this study was the low IBTR rate, which was insufficient for statistical differentiation of significant risk factors. The only statistically significant risk factor was high tumor grade. A Ki-67 index ≥ 14% was only marginally related to IBTR. Thus, a longer follow-up and additional patients are needed for greater statistical validity [bib_ref] Influence of different boost techniques on radiation dose to the left anterior..., Park [/bib_ref] [bib_ref] Estimation of the risk of secondary malignancy arising from whole-breast irradiation: comparison..., Han [/bib_ref] [bib_ref] Early cardiac toxicity following adjuvant radiotherapy of left-sided breast cancer with or..., Cao [/bib_ref]. However, the IBTR rates in this study were low compared with those in previous trials. At 5 years, the CWBI group in this study achieved an IBTR rate of 1.8%, compared with 6.7% in the START A trial [bib_ref] The UK Standardisation of Breast Radiotherapy (START) trials of radiotherapy hypofractionation for..., Haviland [/bib_ref]. Similarly, the AWBI group in this study had an IBTR rate of 0.8%, compared with 8.7% in the START A trial [bib_ref] The UK Standardisation of Breast Radiotherapy (START) trials of radiotherapy hypofractionation for..., Haviland [/bib_ref]. Our low IBTR rate may demonstrate the efficacy of AWBI in early breast cancer, including more aggressive subtypes. In addition to its retrospective design, another limitation of this study was that the AWBI and CWBI arms originated from different centers. Chemotherapy was given more frequently in the AWBI group, which may have affected not only the rate of DM but also those of IBTR, LRR, and RFS.
To the best of our knowledge, this is the first report addressing the efficacy of hypofractionation in patients with early breast cancer with high proliferative potential. Our study demonstrates comparable efficacy between AWBI and CWBI for histologically heterogeneous breast malignancy with adverse prognostic features. Moreover, AWBI was gentler in terms of skin hyperpigmentation. Clinicians should not be discouraged from selecting AWBI for the treatment of biologically more aggressive tumors with poor prognostic factors in breast conservation therapy of early breast cancer.
[fig] Figure 3, Figure 2: Comparison of average points for toxicity grades between accelerated whole breast irradiation (AWBI, solid line) and conventional whole breast irradiation (CWBI, dotted line) for breast edema (A) and erythema/ hyperpigmentation (B) Hazard ratios for ipsilateral breast tumor relapse in subgroup of patients with related risk factors. *Pindicates the significance of difference in hazard ratio between the accelerated hypofractionation and the conventional fractionation for each risk factor subgroup. [/fig]
|
Therapeutic Progress in Treating Vertebral Dissecting Aneurysms Involving the Posterior Inferior Cerebellar Artery
Among the variations of vertebral artery dissecting aneurysms (VDAs), VDAs involving the posterior inferior cerebellar artery (PICA), especially ruptured and high-risk unruptured aneurysms, are the most difficult to treat. Because the PICA is an important structure, serious symptoms may occur after its occlusion. Retained PICAs are prone to re-bleeding because VDAs are difficult to completely occlude. There is therefore confusion regarding the appropriate treatment for VDAs involving the PICA. Here, we used the PubMed database to review recent research concerning VDAs that involve the PICA, and we found that treatments for VDAs involving the PICA include (i) endovascular treatment involving the reconstruction of blood vessels and blood flow, (ii) occluding the aneurysm using an internal coil trapping or an assisted bypass, (iii) inducing reversed blood flow by occluding the proximal VDA or forming an assisted bypass, or (iv) the reconstruction of blood flow via a craniotomy. Although the above methods effectively treat VDAs involving the PICA, each method is associated with both a high degree of risk and specific advantages and disadvantages. The core problem when treating VDAs involving the PICA is to retain the PICA while occluding the aneurysm. Therefore, the method is generally selected on a case-by-case basis according to the characteristics of the aneurysm. In this study, we summarize the various current methods that are used to treat VDAs involving the PICA and provide schematic diagrams as our conclusion. Because there is no special field of research concerning VDAs involving the PICA, these cases are hidden within many multiple-cases studies. Therefore, this study does not review all relevant documents and may have some limitations. Thus, we have focused on the mainstream treatments for VDAs that involve the PICA.
# 1.introduction
Intracranial vertebral dissecting aneurysms (VDA) involving the posterior inferior cerebellar artery (PICA) are not rare. For instance, Hernandez-Duran et al. conducted a meta-analysis in 2014 that included a total of 637 patients from 39 studies. PICA involvement was observed in 14.62% of these patients [bib_ref] Clinical outcomes of patients with vertebral artery dissection treated endovascularly: a meta-analysis, Hernandez-Duran [/bib_ref]. It is difficult to treat intracranial VDAs, and the therapeutic strategy for treating a VDA depends on the anatomical relationship between the VDA and the PICA. VDAs can be divided into four types: proximal to the PICA, distal to the PICA, involving the PICA, or no PICA involvement [bib_ref] Dissecting aneurysms of the vertebral artery: a management strategy, Iihara [/bib_ref]. Among these types, VDAs involving the PICA are the most difficult to treat [bib_ref] Microsurgical anatomy of the extracranial-extradural origin of the posterior inferior cerebellar artery, Fine [/bib_ref].
For ruptured and high-risk unruptured VDAs involving the PICA, the treatment approach should be carefully considered. Furthermore, VDAs involving the PICA should perhaps be more carefully followed Ivyspring International Publisher post-operatively than those without such involvement because this type of aneurysms is more likely to grow and recur [bib_ref] Basilar extension and posterior inferior cerebellar artery involvement as risk factors for..., Matsukawa [/bib_ref]. Current therapies for VDAs involving the PICA include: revascularization with conventional stents, revascularization with coils and stents, and revascularization with flow-diverting stents. Additional treatments include aneurysm trapping or aneurysm resection with craniotomy, which is subsequently supported with bypass and other methods [bib_ref] Classification and treatment of vertebral dissecting aneurysm, Sano [/bib_ref].
All treatments for VDAs involving PICA should be based upon a full assessment of the blood supply to the contralateral vertebral artery. Following this assessment, if it was observed that the blood supply to the contralateral VA was sufficient, only the reconstruction of the PICA must be considered when choosing the treatment option, and there are many options to choose from. Conversely, contralateral VA dysplasia and insufficient blood supply may introduce difficulties during treatment [bib_ref] Multimodal treatment of ruptured dissecting aneurysms of the vertebral artery during the..., Hamada [/bib_ref].
Although the methods for treating VDAs involving the PICA have been described in previous articles that have reviewed various cases, to the best of our knowledge, there is currently no single independent report of each specific method. Therefore, we used the PubMed database to review the current research concerning VDA involving the PICA, and we extracted data from various case studies to assess each method. This study summarizes the various methods that are currently used to treat VDAs involving the PICA and provides schematic diagrams in the conclusions.
## Indications for treatment
There is not much controversy surrounding therapeutic approaches for ruptured VDAs, and without aggressive treatment, the prognosis in these patients is bad [bib_ref] Early Magnesium Treatment After Aneurysmal Subarachnoid Hemorrhage: Individual Patient Data Meta-Analysis, Mees [/bib_ref]. For instance, a study by Yamada et al. in 2004 described two patients who presented with VDA involving the PICA that was accompanied by subarachnoid hemorrhage. However, these two patients were treated conservatively, and both died as a result of the recurrence of a ruptured hemorrhage at 1 month and 1 day after onset [bib_ref] Intracranial vertebral artery dissection with subarachnoid hemorrhage: clinical characteristics and outcomes in..., Yamada [/bib_ref]. However, in rare examples, a VDA can also develop into an occlusion [bib_ref] Long-term follow-up study of unruptured vertebral artery dissection: clinical outcomes and serial..., Nakagawa [/bib_ref]. For instance, in 2002, Wakamoto et al. reported one case of PICA-involved VDA in which a dissecting aneurysm of the PICA spontaneously reduced during a conservative 7-month course of therapy. Similar results were reported by Arai et al. in 2012 [bib_ref] Spontaneous occlusion of PICA-involved dissecting aneurysm with development of a collateral channel..., Arai [/bib_ref]. Although a VDA can spontaneously occlude, spontaneous recanalization as well as enlargement have been observed in most cases, and these ruptured VDAs require treatment [bib_ref] Uncommon course for a vertebral artery dissection: rupture, occlusion and recanalization, Matsumoto [/bib_ref].
The natural course of a VDA is very different between ruptured and unruptured VDAs. Many neuro-interventionists do not agree that unruptured incidental VDAs should be treated. For instance, Kai et al., in a 2011 report, suggested that the nature of an unruptured VDA is not highly aggressive. However, if the dissection site enlarges without manifesting new symptoms, it should be occluded. In patients with recurrent ischemic attacks, antiplatelet therapy should be considered [bib_ref] Strategy for treating unruptured vertebral artery dissecting aneurysms, Kai [/bib_ref]. In unruptured VDAs, dilation appears to be a useful predictive factor for hemorrhagic intracranial VDAs in which the adventitia has maximally extended [bib_ref] Pathomorphometry of ruptured intracranial vertebral arterial dissection: adventitial rupture, dilated lesion, intimal..., Ro [/bib_ref] [bib_ref] Endovascular interventional therapy and classification of vertebral artery dissecting aneurysms, Wang [/bib_ref] [bib_ref] Prediction of rebleeding from angiographic features in vertebral artery dissecting aneurysms, Takagi [/bib_ref].
## Endovascular treatment for the reconstruction of blood vessels and blood flow
This method reconstructs the vascular structures of the VA and retains the PICA. In involves the simple application of stents in combination with a coil or a flow-diverting stent [bib_ref] Dissection of the V4 segment of the vertebral artery: clinicoradiologic manifestations and..., Yoon [/bib_ref]. The simple application of stents for VDAs involving the PICA was instituted during the early developmental stage of endovascular treatment continues to be occasionally used, especially when reconstructing a parent artery using LVIS stents [bib_ref] Endovascular treatment of PICA aneurysms with a Low-profile Visualized Intraluminal Support (LVIS..., Samaniego [/bib_ref]. However, this type of treatment is controversial because it achieves good effects in some aneurysms but poor effects in others. During the early years of treatments for this condition, coronary stents were usually used. Mehta et al. reported two cases of ruptured VDAs involving the PICA in 2003. In these cases, a stent was initially placed, but a follow-up observation revealed that the aneurysms were not completely resolved. Thus, a stent-within-stent arrangement was used, and the aneurysms were resolved [bib_ref] Stent-within-a-stent technique for the treatment of dissecting vertebral artery aneurysms, Mehta [/bib_ref]. Zenteno et al. reported two cases of ruptured VDAs involving the PICA that were treated using coronary stents in 2003. The patients were separately followed up at 3 months and 6 months. DSA showed a total thrombosis in the aneurysm, with patency in both the VA and PICA [bib_ref] Sole stenting bypass for the treatment of vertebral artery aneurysms: technical case..., Zenteno [/bib_ref].
Stent treatment might yield better effects when applied to unruptured aneurysms than ruptured aneurysms. Ahn et al. reported two cases of unruptured VDAs involving the PICA in 2006 that involved the occurrence of thrombosis after stent placement, and follow-up revealed a stable embolism and good effects [bib_ref] Endovascular treatment of intracranial vertebral artery dissections with stent placement or stent-assisted..., Ahn [/bib_ref]. Koh et al. reported a case of an unruptured bilateral VDA with unilateral PICA involvement. A double-stent implantation was performed to treat the VDA that involved the PICA, and at a six-month follow-up, it was demonstrated that the VDA involving the PICA had been remodeled and completely reconstituted. The PICA was preserved and displayed normal blood flow [bib_ref] Bilateral vertebral-artery-dissecting aneurysm causing subarachnoid hemorrhage cured by staged endovascular reconstruction after..., Koh [/bib_ref]. Lv et al. published a multiple-case study of VDAs that involved the PICA. Among the reviewed cases, seven were treated with a simple stent, and the prognoses in these cases were good [bib_ref] Clinical outcomes of ruptured and unruptured vertebral artery-posterior inferior cerebellar artery complex..., Lv [/bib_ref]. Similar results were reported by .
Although the above studies all reported that treatment with simple stents can yield in good results in VDAs involving the PICA, after stent placement, the poor ability of stents to resist blood flow might result in poor curative effects in some aneurysms. Lee et al. reported three cases of VDA involving the PICA in 2010. In this study, two patients with PICA lesions were treated using a stent procedure, and one of these patients died of re-hemorrhagia [bib_ref] Endovascular treatment of ruptured dissecting vertebral artery aneurysms--long-term follow-up results, benefits of..., Lee [/bib_ref]. Shin et al. reported five cases of VDA involving the PICA in 2012. In this study, 4 cases were treated with a double-stent implantation. Among these four cases, the aneurysms were substantially resolved in two cases and partially solved in two cases, the latter of which required further treatment [bib_ref] Endovascular treatment of bilateral intracranial vertebral artery dissecting aneurysms presenting with subarachnoid..., Shin [/bib_ref]. As a result, the simple application of a single stent or multiple stents to reconstruct arteries within an aneurysm is an effective treatment for some cases of VDAs that involve the PICA, but this treatment can fail in other cases as a result of recurrence or rupture and hemorrhage. Indeed, the simple application of stents as a form of treatment is rarely mentioned in recent studies. This type of treatment appears to have been primarily used only before 2012. Furthermore, in cases in which the homolateral artery of the VDA is the main blood supply artery for the posterior circulation, when the contralateral VA experiences dysplasia, treatment with a simple stent is generally adopted to prevent the dysplasia from affecting the basal arterial blood supply. These data are summarized in [fig_ref] Table 1: Data for conventional stent implantations or stent-assisted coil reconstructions [/fig_ref]. [bib_ref] Single-stage endovascular treatment of subarachnoid hemorrhage related to bilateral vertebral artery dissecting..., Zhao [/bib_ref] 58/F 1 stent in coil Partial obliteration of aneurysm, followed up for seven months, good recovery. (2) Revascularization using stent-assisted coils [fig_ref] Figure 2: Stent-assisted coil occlusion for vascular reconstruction [/fig_ref] This method was implemented as a result of the unreliability of simple stent treatment for VDAs involving the PICA. As an alternative to treatment with a simple stent, this method involves the use of a stent that has been combined with a coil. Coils are more effective at introducing thrombosis. For instance, in 2006, Ahn et al. reported two cases of patients with VDAs involving the PICA who were subject to coils and stents to induce thrombosis. Both cases achieved satisfactory embolism and a good prognosis [bib_ref] Endovascular treatment of intracranial vertebral artery dissections with stent placement or stent-assisted..., Ahn [/bib_ref]. Shin et al. reported eight cases of PICA-involved VDA in 2015. Of these, stent-assisted coil embolization was considered in cases with unruptured aneurysms, and these cases resulted in a satisfactory prognosis [bib_ref] Endovascular Treatment of Intracranial Vertebral Artery Dissecting Aneurysms: Follow up Angiographic and..., Shin [/bib_ref]. Similar results were reported by Sadato et al. in 2010 [bib_ref] Endovascular treatment of vertebral artery dissection using stents and coils: its pitfall..., Sadato [/bib_ref].
Although stent-assisted coil embolization yielded good results, as reported in the literature, the recurrence of the aneurysm and re-hemorrhagia were the main problems associated with treatment. For instance, Suzuki et al. reported four cases of VDAs involving the PICA in 2008. These patients were treated with stent-assisted coils, and no surgery-related complications were noted. However, one patient died as a result of re-hemorrhagia, one experienced severe disability, one experienced moderate disability, and one exhibited good effects [bib_ref] Endovascular surgery using stents for vertebral artery dissecting aneurysms and a review..., Suzuki [/bib_ref]. Similar results were reported by , Nam et al. in 2015 [bib_ref] Endovascular treatment of acute intracranial vertebral artery dissection: long-term follow-up results of..., Nam [/bib_ref] , and Lim et al. in 2015 [bib_ref] Endovascular Treatment of Vertebral Artery Dissecting Aneurysms That Cause Subarachnoid Hemorrhage :..., Lim [/bib_ref].
Therefore, although stent-assisted coil embolization can achieve satisfactory results when used to treat VDAs involving the PICA, the stent-assisted coil method has limitations. Because the PICA is derived from the aneurysm, the aneurysm cannot be densely filled, and regrowth and rupture hemorrhage occasionally occur after treatment, and some VDAs can recanalize and bleed after dense embolization. VDAs involving the PICA are therefore difficult to treat [bib_ref] Rebleeding from a vertebral artery dissecting aneurysm after endovascular internal trapping: adverse..., Sugiu [/bib_ref] [bib_ref] Long-term follow up of antegrade recanalization of vertebral artery dissecting aneurysm after..., Kojima [/bib_ref].
Patients with bilateral VDAs involving the lateral or bilateral PICA are harder to treat. Occluding the lateral VA might result in rapid growth and the rupture of the contralateral VDA. Therefore, treatment with a simple stent or a stent-assisted coil should be used to prevent the VA from being blocked [bib_ref] Stent-assisted coil embolization and computational fluid dynamics simulations of bilateral vertebral artery..., Kono [/bib_ref]. In 2015, Zhao et al. reported three cases of bilateral VDAs that included two cases with unilateral PICA involvement and one case with bilateral PICA involvement. These patients were treated with stents and coils, and the VA remained unobstructed. The effects of treatment were satisfactory [bib_ref] Single-stage endovascular treatment of subarachnoid hemorrhage related to bilateral vertebral artery dissecting..., Zhao [/bib_ref]. The data described in this section are summarized in [fig_ref] Table 1: Data for conventional stent implantations or stent-assisted coil reconstructions [/fig_ref].
## (3) revascularization with flow-diverting stents (figure 3)
Incomplete occlusion is a major problem in aneurysms that are treated using a simple stent or stent-assisted coil embolization, and this complication results in uncertainty in the treatment. Flow-diverting stents, which have increased in popularity in recent years, may overcome the disadvantages of the two previously described methods. Flow-diverting stents can be used to treat VA aneurysms. When appropriately sized to the vessel wall and positioned in the VA, the device should cover the origin of the PICA without impairing flow through the branching artery [bib_ref] Reconstructive endovascular treatment of a ruptured vertebral artery dissecting aneurysm using the..., Ducruet [/bib_ref].
Mazur et al. reported four cases of VDAs involving the PICA in 2005 in which a good prognosis was achieved after the application of the pipeline. In 3 cases, the aneurysm was observed to be occluded at the six-month follow-up, and the PICA was patently covered without in-stent stenosis or endoleak. The follow-up results for one of these cases are pending. The procedure was complicated by in-stent thrombosis and PICA occlusion, and the patients were treated with abciximab and recanalization [bib_ref] Pipeline embolization device for the treatment of vertebral artery aneurysms: the fate..., Mazur [/bib_ref]. In 2016, Fang et al. reported two cases of VDAs involving the PICA that were treated using a Tubridge flow diverting stent. In one of these patients, three Tubridge overlapping stents were used. At the postoperative six-month and 36-month follow-ups, both patients exhibited good recovery that included VDA occlusion and unblocked PICAs [bib_ref] Long-Term Outcome of Tubridge Flow Diverter(S) in Treating Large Vertebral Artery Dissecting..., Fang [/bib_ref]. As a result, the flow-diverting stent could potentially be broadly applied to treat VDAs involving the PICA.
## Internal coil trapping for aneurysm occlusion or assisted with bypass
Theoretically, the most effective method for treating VDAs involving the PICA is the direct occlusion of the aneurysm and the sacrificing of the VA and PICA. Studies have shown that ischemic complications are experienced in 21.7% cases in which the PICA is sacrificed. However, regarding the location of the lesion and the mode of the procedure, in the Japanese Registry of Neuroendovascular Therapy 1, the results showed that occluding the PICA did not affect the ratio of favorable outcomes despite the increased incidence of ischemic complications [bib_ref] Endovascular treatment for ruptured vertebral artery dissecting aneurysms: results from Japanese Registry..., Satow [/bib_ref].
(1) Internal coil trapping for aneurysm occlusion with the sacrifice of the PICA [fig_ref] Figure 4: Internal coil trapping of aneurysms involving the sacrificed of the PICA [/fig_ref] A retrospective analysis reported that in some studies, the VA and PICA were sacrificed to treat VDAs. Sometimes, thin PICAs were sacrificed because thin PICAs supply only a small portion of the lower vermis [bib_ref] Medullary infarction as a poor prognostic factor after internal coil trapping of..., Endo [/bib_ref]. For instance, Peluso et al. applied internal coil trapping that included the PICA origin in two cases of VDAs involving the PICA. No clinical symptoms or PICA infarctions were not noted at follow-up in computed tomography or magnetic resonance imaging in 2008. No complications related to the treatment were reported [bib_ref] Endovascular treatment of symptomatic intradural vertebral dissecting aneurysms, Peluso [/bib_ref]. Similar results were also reported by . However, most patients who are treated with internal coil trapping display symptoms. For instance, Yasui et al. reported one case in 2000, in which the patient was treated with internal coil trapping that sacrificed the PICA. Though the PICA showed good compensatory circulation, the patient experienced transient mild hoarseness and dysphasia that alter completely resolved [bib_ref] Subarachnoid hemorrhage from vertebral artery dissecting aneurysms involving the origin of the..., Yasui [/bib_ref]. Iihara et al. reported two cases of VDA involving the PICA in 2002. One case displayed deficits, dysarthria and dysphagia of the oculomotor nerve and facial nerve after internal trapping of the aneurysm was applied. The patient's symptoms gradually recovered for the following 44 months, but a left-side mild disability remained. The other patient also experienced a left-side mild disability [bib_ref] Dissecting aneurysms of the vertebral artery: a management strategy, Iihara [/bib_ref]. Similar results were reported by [bib_ref] Coil embolization for the treatment of ruptured dissecting vertebral aneurysms, Kurata [/bib_ref]. Therefore, using internal coil trapping to occlude the aneurysm and PICA is associated with a large degree of risk when used in patients with VDAs involving the PICA. This technique is therefore not accepted as a standard treatment.
(2) Internal coil trapping for aneurysm occlusion with the sacrifice of the PICA and bypass assistThe main risk when using internal coil trapping to occlude an aneurysm in VDAs involving the PICA is related to the occlusion of the PICA. Therefore, recanalizing the PICA after the occlusion of the aneurysm can provide satisfactory therapeutic effects. For instance, Park et al. reported 1 case of a VDA involving the PICA in 2014. The patient was subjected to occipital artery-PICA bypass followed by endovascular trapping of the VDA and the origin of the PICA two days later, and positive effects were noted [bib_ref] Occipital artery-posterior inferior cerebellar artery bypass for the treatment of aneurysms arising..., Park [/bib_ref]. A similar result was reported by Lee et al. in 2010 [bib_ref] Endovascular treatment of ruptured dissecting vertebral artery aneurysms--long-term follow-up results, benefits of..., Lee [/bib_ref]. Occipital artery-PICA bypass-assisted internal coil trapping was performed to treat an aneurysm. The procedure included the sacrifice of the PICA, as shown in. In addition to the occipital artery, PICA-PICA bypass is also used. Chung et al. treated one case of VDA involving the PICA in 2014 in which endovascular segmental coil occlusion and side to side anastomosis of the PICA was applied, and the prognosis in this patient was good [bib_ref] A case of endovascular treatment for followed by side to side bypass..., Chung [/bib_ref]. More complex operations are also available. For instance, Chandela et al. treated one case of VDA involving the PICA in 2008 by applying a coiling embolism to the aneurysm. Recurrence was noted, and the patient was therefore subjected to occipital artery-PICA bypass. PICA-PICA bypass and internal coil trapping VDA were applied when the blood flow was found to be insufficient, and the effects were positive [bib_ref] Treatment of a complex posterior fossa aneurysm in a child using side-to-side..., Chandela [/bib_ref]. PICA-PICA bypass after internal coil trapping of an aneurysm, including PICA sacrifice, was described, as shown in.
Poor prognoses have also been reported. Yuki et al. reported one case of VDA involving PICA in 2005. The patient was treated with internal trapping followed by complete occipital artery-PICA bypass. The patient died after the operation [bib_ref] Endovascular management of dissecting vertebrobasilar artery aneurysms in patients presenting with acute..., Yuki [/bib_ref]. Endo studied five cases in which VDAs involved the PICA in 2013. In these patients, internal coil trapping was completed within 24 hours, and an occipital artery-PICA anastomosis followed by internal coil trapping was subsequently performed. Three patients experienced a medulla infarction after the operation. The study reported that coil occlusion of the long segment of the VA led to medullary infarction and that occipital artery-PICA bypass did not prevent medullary infarction [bib_ref] Medullary infarction as a poor prognostic factor after internal coil trapping of..., Endo [/bib_ref]. In rare cases, a stent can be placed into the PICA from the VA through the aneurysm using coil embolization of the aneurysm and the distal VA. In 2010, Chung et al. reported 1 case of VDA that involved the PICA in which VA-to-PICA stenting was applied to preserve the PICA while treating a ruptured VAD. The dissected segment of the VA was completely occluded by coil embolization, and the prognosis was good [bib_ref] Vertebral artery occlusion with vertebral artery-to-posterior inferior cerebellar artery stenting for preservation..., Chung [/bib_ref]. Moreover, Fukuda et al. reported one case of VDA involving the PICA in 2015. Endovascular internal trapping of the enlarged distal VAD was initially performed. After one month, an enterprise stent placement from the PICA to the proximal VA that included coil embolization for the proximal VAD was performed. The dissected VA segment was occluded by the coil embolization, and the PICA was preserved [bib_ref] Stent-assisted embolization for ruptured vertebral artery dissection involving the origin of the..., Fukuda [/bib_ref]. Similar results were reported by [bib_ref] Management of hemorrhagic dissecting vertebral artery aneurysms involving posterior inferior artery, Guo [/bib_ref]. It is occasionally difficult to perform a VA occlusion, and such aneurysms are prone to recanalization. Chen et al. reported seven cases of VDAs involving the PICA in 2013. Although aneurysm trapping was successfully performed after a stent was placed from the distal VA to the PICA, and although the prognoses in these patients was good, three cases exhibited aneurysm recanalization during follow-up, and these patient were subsequently monitored during follow-up [bib_ref] Stent placement to treat ruptured vertebral dissecting aneurysms, Chen [/bib_ref]. This method can be used to effectively treat VDAs involving the PICA. However, when the PICA is healthy and the VA is enlarged, there is a large difference in the diameter of the vessels. Thus, stent placement is difficult in these patients. Given the angle, it is also very difficult to introduce a catheter and guide wire from the VA to the PICA. Thus, this technique is not currently used as a routine treatment. The length of the VA to be occluded should be minimized as much as possible to reduce the occurrence of a medullary infarction. These data are summarized in [fig_ref] Table 2: Data for occlusions of aneurysms by internal coil trapping or bypass assistance [/fig_ref].
## Occlusion of the proximal vda resulting in reverse blood flow or assisted bypass
In VDAs, if the blood flow can be reversed so that it flows into the aneurysm to supply the PICA, the directional changes in blood flow both reduce the pressure of the blood flow and also alter the impact point of the blood flow, thereby reducing the risk of aneurysm rupture. As a result, the occlusion of the proximal VDA can be used as a treatment option. If blood flow reversal is insufficient, it can be combined with a bypass of the PICA. To implement this method, a full assessment of the compensatory blood flow in the contralateral vertebral artery must be performed, and this technique can be used only when the blood supply of the contralateral vertebral artery is sufficient [bib_ref] Treatment of dissecting vertebral aneurysm, Kai [/bib_ref].
(1) Proximal endovascular occlusion or clipping of VDAsThe occlusion of the proximal VA in VDAs leads to directional changes in blood flow, which reduces the risk of a ruptured aneurysm. In 2004, Kobayashi et al. reported 2 cases of VDAs involving a PICA that was treated using the endovascular occlusion of a proximal VDA, with good effects [bib_ref] Endovascular Treatment for Ruptured VA Dissecting Aneurysm Involving the Origin of PICA, Kobayashi [/bib_ref]. Moreover, Lv et al. reported one multiple-case study of a VDA that involved the PICA in 2010. In this study, the VA was proximally occluded to treat 8 cases. The prognoses in these patients were good. However, treatment for the aneurysm was incomplete in 7 of these cases and sub-complete in one case [bib_ref] Clinical outcomes of ruptured and unruptured vertebral artery-posterior inferior cerebellar artery complex..., Lv [/bib_ref]. The proximal endovascular occlusion of the VDA is described in. In addition to endovascular occlusion, surgical clipping is also used as an alternative treatment. For instance, in 2005, Sugiu et al. reported 1 case of VDA involving the PICA. Craniotomy was used to clip the proximal VA and reverse the blood flow, and the prognosis in this patient was good [bib_ref] Emergent endovascular treatment of ruptured vertebral artery dissecting aneurysms, Sugiu [/bib_ref]. Proximal clipping of the VDA is described in. Because this method does not effectively eradicate the aneurysm, the aneurysm may not disappear. For instance, in 2004, Shin et al. described a 71-year-old man with progressive myelopathy and a VDA involving the PICA. The patient was treated with a coil to proximally occlude the VA to reverse the flow. The aneurysm did not change in size during follow-up, but the patient's symptoms improved [bib_ref] Treatment of vertebral artery dissecting aneurysms presenting with progressive myelopathy, Shin [/bib_ref]. Although this method is effective, the proximal occlusion technique can result in retrograde flow from the contralateral VA, which can cause rebleeding. The increased length of the proximal occlusion area, and specifically the segment that is most proximal to the dilated portion, is associated with an increased incidence of medullary infarction following internal trapping. These results indicate that this complication may be avoidable [bib_ref] Effect of coil packing proximal to the dilated segment on postoperative medullary..., Ikeda [/bib_ref]. In 1993, Takai reported a case of a VA-dissected aneurysm that showed rebleeding after a proximal occlusion [bib_ref] Vertebral artery dissecting aneurysm rebleeding after proximal occlusion--case report, Takai [/bib_ref]. Similar results were reported by and . Moreover, Nakatomi et al. reported 1 case of VDA involving the PICA in 1999. The patient was followed up after proximal clipping was performed. Excessive retrograde flow from the distal vertebral artery into both the PICA and the pseudolumen were noted, and the latter persisted after proximal clipping was performed. The patient was assumed to have developed moderate retrograde flow that was sufficient to maintain a patent pseudolumen during the chronic stage [bib_ref] Persistent patent pseudolumen of ruptured dissecting aneurysm involving the posterior inferior cerebellar..., Nakatomi [/bib_ref].
Positive effects may be better in unruptured cases. Nam et al. reported one case of unruptured VDA involving PICA in 2015. The patient was in stable condition and experience a good prognosis after proximal occlusion was performed [bib_ref] Endovascular treatment of acute intracranial vertebral artery dissection: long-term follow-up results of..., Nam [/bib_ref]. The mentioned cases were both unilateral VDA involving the PICA, and there are also some cases in which bilateral VDAs involving the PICA have been treated by inducing the retrograde flow of the basilar artery via the occlusion of the bilateral VAs [bib_ref] Bilateral vertebral artery dissecting aneurysm with subarachnoid hemorrhage treated with staged bilateral..., Inoue [/bib_ref].
(2) Proximal occlusion or clipping of VDAs with simultaneous assistance by bypass [fig_ref] Figure 8: Proximal occlusion or clipping of the VDA or bypass assistance [/fig_ref] Proximal VDA occlusion caused blood flow to be reversed, resulting in a reduction in the risk of a ruptured aneurysm. If blood flow to the PICA is insufficient, the revascularization of the PICA can be used to assist it. Iihara et al. reported one case of VDA involving the PICA in 2002. The patient was subjected to the proximal occlusion of the parent artery and an occipital artery-PICA bypass, and the patient experienced a good recovery [bib_ref] Dissecting aneurysms of the vertebral artery: a management strategy, Iihara [/bib_ref]. A similar result was reported by . When the VDA is proximally occluded or clipped, the spinal artery should be considered. For instance, Carlson et al. reported 1 case of VDA involving the PICA in 2015. The patient was initially subject to an occipital artery to PICA anastomosis. Then, the VA was clipped immediately proximal to the dilated segment. No distal clip was applied to allow the retrograde filling of the anterior spinal artery from the contralateral vertebral artery, and a follow-up angiography revealed VDA remodeling [bib_ref] Tailored PICA Revascularization for Unusual Ruptured Fusiform Vertebro-PICA Origin Aneurysms: Rationale and..., Carlson [/bib_ref]. Moreover, staging treatments have also been reported. Hamasaki et al. reported one case of VDA involving the PICA in 2014. The patient was subject to proximal endovascular occlusion in the acute stage. However, as a result of insufficient blood flow, occipital artery-PICA anastomosis and surgical trapping were subsequently applied in the chronic stage. The patient was followed up for 11 months and experienced a good prognosis [bib_ref] Treatment of ruptured vertebral artery dissecting aneurysms. A short report, Hamasaki [/bib_ref]. These data are summarized in [fig_ref] Table 3: Data for reversed blood flow caused by the occlusion of a proximal... [/fig_ref].
## Surgical reconstruction of blood flow
Some cases of VDAs involving PICA cannot be resolved using endovascular surgery when craniotomy clipping or resection of VDA is adopted and blood flow is simultaneously re-established [bib_ref] Multimodal treatment of ruptured dissecting aneurysms of the vertebral artery during the..., Hamada [/bib_ref]. Many factors should be considered when reconstructing blood flow. In addition to the PICA on the aneurysm side, the blood supply of the posterior circulation of the contralateral VA should also be considered. If the blood supply is sufficient, only the PICA on the aneurysm side should be reconstructed. If the blood supply is insufficient, the construction of the VA may also be required after the resection of the aneurysm or posterior cerebral artery bypass is performed to guarantee sufficient blood supply to the posterior circulation. (1) PICA-PICA bypass or PICA re-implantation to VA after VDA trapping [fig_ref] Figure 1: Conventional stent implantation for vascular reconstruction [/fig_ref] The advantage of this approach is that foreign vascular grafting is not required [bib_ref] Surgical treatment for vertebral artery-posterior inferior cerebellar artery aneurysms: special reference to..., Kawashima [/bib_ref]. This method was reported as early as 1991 by Takikawa [bib_ref] Vertebral dissecting aneurysm treated with trapping and bilateral posterior inferior cerebellar artery..., Takikawa [/bib_ref]. Durward also reported on the technique in 1995 [bib_ref] Treatment of vertebral artery dissecting aneurysm by aneurysm trapping and posterior inferior..., Durward [/bib_ref]. The PICA and VA proximal to the aneurysm can be anastomosed in an end-to-end fashion. For instance, Ogasawara et al. reported one case of a 40-year-old man with VDA involving the PICA in 2006 in which this method was adopted [bib_ref] Treatment of vertebral artery aneurysms with transposition of the posterior inferior cerebellar..., Ogasawara [/bib_ref]. This method is described in [fig_ref] Figure 9: PICA reimplantation to the VA after VDA trapping [/fig_ref]. Moreover, PICA-PICA bypass is also an alternative. For instance, Abla et al. reported 10 cases of VDAs involving the PICA in 2015 in which aneurysms trapped at the PICA origin were revascularized using a PICA-PICA bypass, and PICA reimplantation served as an alternative treatment. Positive results were observed after treatment. Guo et al. published a similar report in 2014 [bib_ref] Management of hemorrhagic dissecting vertebral artery aneurysms involving posterior inferior artery, Guo [/bib_ref] , and Endo et al. published a similar report in 2005 [bib_ref] Ruptured vertebral artery dissecting aneurysm followed by spontaneous acute occlusion and early..., Endo [/bib_ref]. This method is described in [fig_ref] Figure 1: Conventional stent implantation for vascular reconstruction [/fig_ref]. Because craniotomy can cause a large amount of trauma, complications can appear in some cases. In 2004, Shin et al. reported 1 case of VDA involving the PICA that displayed in which the onset of progressive myelopathy was observed. After aneurysm trapping, the PICA was then sutured to the arteriotomy. The prognosis revealed incomplete Wallenberg syndrome [bib_ref] Treatment of vertebral artery dissecting aneurysms presenting with progressive myelopathy, Shin [/bib_ref]. Kakino et al. studied 6 cases of VDAs involving PICA in 2004 in which anastomoses were performed in a side-to-side fashion at the posterior medullary segment of the PICA. The VA was subsequently occluded by clipping both proximal and distal to the aneurysm, and the PICA was occluded by clipping distal to the aneurysm. One of these cases died after the operation, one case displayed moderate disability, and four cases experienced a good recovery [bib_ref] Treatment of vertebral artery aneurysms with posterior inferior cerebellar artery-posterior inferior cerebellar..., Kakino [/bib_ref]. When vasospasm occurs in the VA, a surgical treatment involving PICA-PICA bypass or PICA re-implantation into the VA after VDA trapping can become more difficult [bib_ref] Ruptured vertebral artery dissecting aneurysm followed by spontaneous acute occlusion and early..., Endo [/bib_ref].
(2) Reconstruction of the PICA with a vascular graft after VDA trapping [fig_ref] Figure 1: Conventional stent implantation for vascular reconstruction [/fig_ref] When the aneurysm is larger and the length of the PICA after aneurysm trapping has been performed is insufficient, foreign vascular grafts are often applied as a treatment option. In 2002, Hamada et al. treated 4 cases of VDAs involving the PICA using VA-PICA bypass with a superficial temporal artery graft and achieved satisfactory effects [bib_ref] Vertebral artery-posterior inferior cerebellar artery bypass with a superficial temporal artery graft..., Hamada [/bib_ref]. In addition to the superficial temporoparietal artery, the radial artery is also a good option in these cases. Czabanka et al. reported one case of a VDA involving the PICA in 2011. After trapping the aneurysm, the patient was submitted to VA-PICA bypass with a radial artery graft, and the resulting prognosis was good [bib_ref] Vertebral artery-posterior inferior cerebellar artery bypass using a radial artery graft for..., Czabanka [/bib_ref]. (3) Reconstruction of VA and PICA after VDA trapping or removal [fig_ref] Figure 1: Conventional stent implantation for vascular reconstruction [/fig_ref] The choice of surgical treatment is very important in occipital artery VDAs involving the PICA. In 2014, Wang et al. performed occipital artery-PICA bypass and aneurysm trapping using surgical clips to treat these patients and achieved good effects [bib_ref] Long-term outcomes of trapping vertebral artery-posterior inferior cerebellar artery dissecting aneurysms after..., Wang [/bib_ref]. In addition, Lim et al. treated two cases of VDAs involving the PICA in 2015. The cases showed partial trapping and surgical trapping after occipital artery-PICA bypass, and they each recovered after the operation [bib_ref] Endovascular Treatment of Vertebral Artery Dissecting Aneurysms That Cause Subarachnoid Hemorrhage :..., Lim [/bib_ref]. Similar results were reported by , and . In the presence of an aneurysm space-occupying effect or contralateral VA dysplasia, the blood supply to the posterior circulation is occasionally insufficient, and aneurysm resection and reconstruction of the VA are therefore needed. Kubota et al. reported one case of VDA involving the PICA in 2014. After an aneurysm resection was performed, the radial artery was used to reconstruct the VA, and an OA-PICA bypass was performed. The effects were positive [bib_ref] Vertebral artery-to-vertebral artery bypass with interposed radial artery or occipital artery grafts:..., Kubota [/bib_ref]. With regard to trapping treatments that can be used for VDAs involving the PICA, cases with bilateral VA dissecting aneurysms are problematic and should be noted. Using a surgical intervention to treat a dissecting aneurysm of the bilateral VAs on only one side carries the risk of rupture of the contralateral lesion. Increased hemodynamic stress may be important to the development and rupture of dissections in the contralateral VA [bib_ref] Development of a dissecting aneurysm on the vertebral artery immediately after occlusion..., Kubo [/bib_ref] [bib_ref] Bilateral dissecting aneurysms of the vertebral arteries resulting in subarachnoid hemorrhage: case..., Yasui [/bib_ref]. Otawara et al. reported two cases of bilateral VDAs involving the PICA in 2002. The PICA was submitted to vascular reconstruction after aneurysm trapping was performed. Once case died as a result of a rupture after the operation, whereas the other case had a good prognosis [bib_ref] Dissecting aneurysms of the bilateral vertebral arteries with subarachnoid hemorrhage: report of..., Otawara [/bib_ref]. In 2009, Katsuno et al. reported one case in which a rupture of a VA dissecting aneurysm developed immediately after a dissecting aneurysm was trapped in the contralateral vertebral artery [bib_ref] Rupture of a vertebral artery dissecting aneurysm developing immediately after trapping of..., Katsuno [/bib_ref].
In summary, in cases with VDAs involving the PICA, craniotomy and aneurysm trapping or resection are also a good choice. However, blood flow to the PICA must be reconstructed. These data are summarized in .
## Other treatments
In addition to the above-described methods, less common treatments, such as wrapping aneurysms, are also available. In 2007, Lee et al. reported three cases of VDAs involving the PICA in which a wrapping operation was performed, and the patients recovered postoperatively [bib_ref] Spontaneous dissecting aneurysm of the intracranial vertebral artery: management strategies, Lee [/bib_ref]. Although positive effects were reported in this study, this technique is viewed as an early treatment option, and its safety is questionable. The technique is currently rarely used.
In 2014, Kobayashi reported 2 cases of VDAs involving the PICA that were treated using the following methods. The affected VA was occluded near its root to introduce collateral blood flow from the deep cervical artery into the VA trunk. The controlled antegrade VA flow and retrograde flow from the contralateral VA created a watershed at the dissecting aneurysm that promoted thrombosis of the pseudolumen and preserved the antegrade blood flow to the PICA [bib_ref] Endovascular Treatment for Ruptured VA Dissecting Aneurysm Involving the Origin of PICA, Kobayashi [/bib_ref]. This method is effective, but it cannot be used as a standard treatment.
Other methods can also be used in some VDAs with good anatomical structure and conditions conducive to aneurysm reconstruction using aneurysm clips. In 2010, Li et al. reported two case of VDA involving the PICA that was reconstructed using 2 artery clips. The proximal occlusion of the aneurysm was performed using fenestrated clips in one patient, whereas the reconstruction of the PICA was performed using fenestrated clips in the other patient. Satisfactory effects were obtained [bib_ref] Vertebral artery dissecting aneurysm treated by proximal occlusion and posterior inferior cerebellar..., Li [/bib_ref]. . Data for surgical reconstruction of blood flow.
## No. type author/year
Treatments Prognosis 1 [fig_ref] Figure 9: PICA reimplantation to the VA after VDA trapping [/fig_ref] Durward/1995 [bib_ref] Treatment of vertebral artery dissecting aneurysm by aneurysm trapping and posterior inferior..., Durward [/bib_ref] An aneurysm was treated by trapping. The PICA was anastomosed to the vertebral artery proximal to the dissection.
Good recovery.
2 [fig_ref] Figure 9: PICA reimplantation to the VA after VDA trapping [/fig_ref] Yasui/2000 [bib_ref] Subarachnoid hemorrhage from vertebral artery dissecting aneurysms involving the origin of the..., Yasui [/bib_ref] An aneurysm was treated with coil occlusion of a proximal aneurysm, but reverse blood flow allowed the aneurysm to grow. Then, aneurysm trapping was adopted prior to OA-PICA reconstruction.
Good recovery.
3 [fig_ref] Figure 9: PICA reimplantation to the VA after VDA trapping [/fig_ref] Otawara/2002 [bib_ref] Dissecting aneurysms of the bilateral vertebral arteries with subarachnoid hemorrhage: report of..., Otawara [/bib_ref] Two cases of bilateral VDAs involving the PICA were subjected to aneurysm trapping and then reconstruction of the PICA with a PICA-PICA bypass.
One case experienced postoperative rupture and died, whereas the other case exhibited good prognosis. 4 [fig_ref] Figure 9: PICA reimplantation to the VA after VDA trapping [/fig_ref] Kakino/2004 [bib_ref] Treatment of vertebral artery aneurysms with posterior inferior cerebellar artery-posterior inferior cerebellar..., Kakino [/bib_ref] In six cases, anastomoses were performed in a side-to-side fashion at the posterior medullary segment of the PICA. The VA was subsequently occluded by clipping it proximal and distal to the aneurysm, and the PICA was occluded by clipping it distal to the aneurysm.
One case died, one case had medium disability, and 4 cases recovered. Czabanka/2011 [bib_ref] Vertebral artery-posterior inferior cerebellar artery bypass using a radial artery graft for..., Czabanka [/bib_ref] After aneurysm trapping, VA-PICA bypass was performed using a radial artery graft. Good recovery.
12 [fig_ref] Figure 1: Conventional stent implantation for vascular reconstruction [/fig_ref] Takemoto/2010 [bib_ref] Surgical treatment of intracranial VA dissecting aneurysm, Takemoto [/bib_ref] Two cases of VDA involving the PICA were subjected to aneurysm trapping and OA-PICA bypass to avoid serial infraction of the PICA area.
Mild disability.
13 [fig_ref] Figure 1: Conventional stent implantation for vascular reconstruction [/fig_ref] Park/2014 [bib_ref] Occipital artery-posterior inferior cerebellar artery bypass for the treatment of aneurysms arising..., Park [/bib_ref] Two cases of VDA involving the PICA were subjected to OA-PICA bypass and then aneurysm trapping using surgical clips.
Good recovery.
14 [fig_ref] Figure 1: Conventional stent implantation for vascular reconstruction [/fig_ref] Lim/2015 [bib_ref] Endovascular Treatment of Vertebral Artery Dissecting Aneurysms That Cause Subarachnoid Hemorrhage :..., Lim [/bib_ref] Two cases were subjected to partial trapping and surgical trapping after OA-PICA bypass. Good recovery. 15 [fig_ref] Figure 1: Conventional stent implantation for vascular reconstruction [/fig_ref] One case of resection of an aneurysm, VA-to-VA bypass with an interposed radial artery, reconstruction of the PICA using the OA.
# Conclusions
Although many methods can be used to treat VDAs involving the PICA, our literature review demonstrates that each treatment is associated with a very high degree of risk. The period from earlier years, when techniques involved blood flow reconstruction via craniotomy, to more recent years and the current practice of reconstructing the vascular cavity using blood flow-diverting stents, represents the therapeutic history of VDAs that involve the PICA. The core problem when treating VDAs involving the PICA involves the retention of the PICA and the occlusion of the aneurysm cavity. Therefore, a variety of methods have emerged, including simple stent implantations, stent-assisted coil embolization, and blood flow-diverting stent implantations. Furthermore, some researchers also have attempted to reverse the blood flow to supply to the PICA by occluding the proximal aneurysm. Occlusion and aneurysm trapping or resections are available as methods to assist in the reconstruction of the PICA and VA. Trapping is the most reliable method for preventing rebleeding. Proximal clipping and occlusion are optional procedures, but these techniques do not always prevent rebleeding given the fact that circulation is continuous [bib_ref] Surgical treatment for ruptured vertebral artery dissecting aneurysms, Fukasawa [/bib_ref].
In addition, the compensatory situation in the contralateral VA should also be considered. A contralateral VA with dysplasia cannot provide sufficient blood flow to the posterior circulation, and the aneurysm cavity must therefore be reconstructed during the treatment of such VDAs. This includes the use of stents or stent-assisted coil embolization or the reconstruction of the bilateral vertebral artery of the aneurysm without the direct occlusion of the VA on the aneurysm side. Some methods cannot completely occlude an aneurysm, and in these cases, the aneurysm is prone to recurrence. In 2014, Zhao et al. reported 10 cases of VDAs involving the PICA that were associated with incomplete embolism. Furthermore, the degree of immediate occlusion was closely associated with recurrence after reconstructive treatments were performed [bib_ref] The interaction between stent(s) implantation, PICA involvement, and immediate occlusion degree affect..., Zhao [/bib_ref]. However, the complications that are associated with treatments for VDAs that involve the PICA cannot be ignored. In a study by Endo et al. that was published in 2013, 10 cases in which patients were treated for VDAs exhibited medullary infarction, and these included four cases of VDAs that involved the PICA [bib_ref] Medullary infarction as a poor prognostic factor after internal coil trapping of..., Endo [/bib_ref].
[fig] Figure 1: Conventional stent implantation for vascular reconstruction. Pre: preoperation, Post: postoperation, VA: vertebral artery, PICA: posterior inferior cerebral artery. [/fig]
[fig] 1: Revascularization with a conventional stent(Figure 1) [/fig]
[fig] Figure 2: Stent-assisted coil occlusion for vascular reconstruction. Pre: preoperation, Post: postoperation, VA: vertebral artery, PICA: posterior inferior cerebral artery. [/fig]
[fig] Figure 3: Blood flow-diverting stent for vascular reconstruction. Pre: preoperation, Post: postoperation, VA: vertebral artery, PICA: posterior inferior cerebral artery. [/fig]
[fig] Figure 4: Internal coil trapping of aneurysms involving the sacrificed of the PICA. Pre: preoperation, Post: postoperation, VA: vertebral artery, PICA: posterior inferior cerebral artery. [/fig]
[fig] Figure 5 A: : OA-PICA bypass-assisted internal coil trapping of an aneurysm in which the PICA is sacrificed; B: PICA-PICA bypass after internal coil trapping of the aneurysm with PICA sacrifice. Pre: preoperation, Post: postoperation, VA: vertebral artery, PICA: posterior inferior cerebral artery, OA: occipital artery. [/fig]
[fig] 3: Internal coil trapping for aneurysm occlusion with stent implantation from the VA to the PICA (Figure 6) [/fig]
[fig] Figure 6: Internal coil trapping for aneurysm occlusion with stent implantation from the VA to the PICA. Pre: preoperation, Post: postoperation, VA: vertebral artery, PICA: posterior inferior cerebral artery. [/fig]
[fig] Figure 7 A: : Proximal endovascular occlusion of the VDA; B: Proximal surgical clipping of the VDA. Pre: preoperation, Post: postoperation, VA: vertebral artery, PICA: posterior inferior cerebral artery. [/fig]
[fig] Figure 8: Proximal occlusion or clipping of the VDA or bypass assistance. Pre: preoperation, Post: postoperation, VA: vertebral artery, PICA: posterior inferior cerebral artery. [/fig]
[fig] M: : male, F: female, VA: vertebral artery. VDA: vertebral dissecting aneurysm. OA: occipital artery. PICA: posterior inferior cerebellar artery. [/fig]
[fig] Figure 9: PICA reimplantation to the VA after VDA trapping. Pre: preoperation, Post: postoperation, VA: vertebral artery, PICA: posterior inferior cerebral artery. [/fig]
[table] Table 2: Data for occlusions of aneurysms by internal coil trapping or bypass assistance. [/table]
|
QSMRim-Net: Imbalance-aware learning for identification of chronic active multiple sclerosis lesions on quantitative susceptibility maps
Background and Purpose: Chronic active multiple sclerosis (MS) lesions are characterized by a paramagnetic rim at the edge of the lesion and are associated with increased disability in patients. Quantitative susceptibility mapping (QSM) is an MRI technique that is sensitive to chronic active lesions, termed rim + lesions on the QSM. We present QSMRim-Net, a data imbalance-aware deep neural network that fuses lesion-level radiomic and convolutional image features for automated identification of rim + lesions on QSM. Methods: QSM and T2-weighted-Fluid-Attenuated Inversion Recovery (T2-FLAIR) MRI of the brain were collected at 3 T for 172 MS patients. Rim + lesions were manually annotated by two human experts, followed by consensus from a third expert, for a total of 177 rim + and 3986 rim negative (rim− ) lesions. Our automated rim + detection algorithm, QSMRim-Net, consists of a two-branch feature extraction network and a synthetic minority oversampling network to classify rim + lesions. The first network branch is for image feature extraction from the QSM and T2-FLAIR, and the second network branch is a fully connected network for QSM lesion-level radiomic feature extraction. The oversampling network is designed to increase classification performance with imbalanced data. Results: On a lesion-level, in a five-fold cross validation framework, the proposed QSMRim-Net detected rim + lesions with a partial area under the receiver operating characteristic curve (pROC AUC) of 0.760, where clinically relevant false positive rates of less than 0.1 were considered. The method attained an area under the precision recall curve (PR AUC) of 0.704. QSMRim-Net out-performed other state-of-the-art methods applied to the QSM on both pROC AUC and PR AUC. On a subject-level, comparing the predicted rim + lesion count and the human expert annotated count, QSMRim-Net achieved the lowest mean square error of 0.98 and the highest correlation of 0.89 (95% CI: 0.86, 0.92). Conclusion: This study develops a novel automated deep neural network for rim + MS lesion identification using T2-FLAIR and QSM images.
# Introduction
Multiple sclerosis (MS) is an inflammatory disease of the central nervous system, characterized by lesions in the brain and spinal cord [bib_ref] Multiple Sclerosis and Brain Atrophy, Sahraian [/bib_ref]. A particular type of MS lesion, called a chronic active lesion, is characterized by an iron-enriched rim of activated macrophages and microglia in histopathology studies [bib_ref] Slow expansion of multiple sclerosis iron rim lesions: pathology and 7 T..., Dal-Bianco [/bib_ref] [bib_ref] Persistent 7-tesla phase rim predicts poor outcome in new multiple sclerosis patient..., Absinta [/bib_ref] [bib_ref] Quantitative susceptibility mapping identifies inflammation in a subset of chronic multiple sclerosis..., Kaunzner [/bib_ref] [bib_ref] QSM is an imaging biomarker for chronic glial activation in multiple sclerosis..., Gillen [/bib_ref]. Chronic active lesions are visible with in-vivo susceptibility magnetic resonance imaging (MRI) techniques, where these lesions show a paramagnetic rim [bib_ref] Slow expansion of multiple sclerosis iron rim lesions: pathology and 7 T..., Dal-Bianco [/bib_ref] [bib_ref] Quantitative susceptibility mapping identifies inflammation in a subset of chronic multiple sclerosis..., Kaunzner [/bib_ref] [bib_ref] Seven-tesla phase imaging of acute multiple sclerosis lesions: a new window into..., Absinta [/bib_ref] [bib_ref] Imaging cortical lesions in multiple sclerosis with ultra-high-field magnetic resonance imaging, Pitt [/bib_ref] [bib_ref] Quantitative in vivo magnetic resonance imaging of multiple sclerosis at 7 Tesla..., Hammond [/bib_ref] [bib_ref] Identification of chronic active multiple sclerosis lesions on 3T MRI, Absinta [/bib_ref] [bib_ref] Tracking iron in multiple sclerosis: a combined imaging and histopathological study at..., Bagnato [/bib_ref] [bib_ref] Iron deposition in multiple sclerosis lesions measured by susceptibility-weighted imaging filtered phase:..., Hagemeier [/bib_ref] [bib_ref] Chronic multiple sclerosis lesions: characterization with high-fieldstrength MR imaging, Yao [/bib_ref] [bib_ref] Multiple sclerosis: validation of MR imaging for quantification and detection of iron, Walsh [/bib_ref] [bib_ref] Lesion heterogeneity on high-field susceptibility MRI is associated with multiple sclerosis severity, Harrison [/bib_ref] [bib_ref] A serial in vivo 7T magnetic resonance phase imaging study of white..., Bian [/bib_ref] on the edge. The presence of chronic active MS lesions is associated with a more severe disease course [bib_ref] Lesion heterogeneity on high-field susceptibility MRI is associated with multiple sclerosis severity, Harrison [/bib_ref] [bib_ref] Association of chronic active multiple sclerosis lesions with disability in vivo, Absinta [/bib_ref] [bib_ref] Clinical and pathological insights into the dynamic nature of the white matter..., Frischer [/bib_ref] [bib_ref] Progressive multiple sclerosis patients show substantial lesion activity that correlates with clinical..., Luchetti [/bib_ref] and there is currently much interest in using these lesions as an imaging biomarker.
In studies of chronic active MS lesions on MRI, lesions are typically identified on the T2-weighted-Fluid-Attenuated Inversion Recovery (T2-FLAIR) image and then are determined to be chronic active through visual inspection on susceptibility imaging. This process is time consuming, as all lesions must be reviewed for a rim, and prone to interand intra-rater variability [bib_ref] Longitudinal multiple sclerosis lesion segmentation: resource and challenge, Carass [/bib_ref] [bib_ref] ALL-Net: Anatomical information lesion-wise loss function integrated into neural network for multiple..., Zhang [/bib_ref]. For these lesions to be further studied at a large scale and translated into clinical practice, there is a great need for automated methods to identify chronic active MS lesions.
Quantitative susceptibility mapping (QSM) is an MRI technique that provides in vivo quantification of magnetic susceptibility changes related to iron deposition [bib_ref] Quantitative susceptibility mapping (QSM): decoding MRI data for a tissue magnetic biomarker, Wang [/bib_ref] [bib_ref] Quantitative susceptibility map reconstruction from MR phase data using Bayesian regularization: validation..., De Rochefort [/bib_ref]. QSM identifies chronic active MS lesions [bib_ref] Quantitative susceptibility mapping in multiple sclerosis, Langkammer [/bib_ref] [bib_ref] Iron in multiple sclerosis and its noninvasive imaging with quantitative susceptibility mapping, Stüber [/bib_ref] [bib_ref] Quantitative susceptibility mapping (QSM) of white matter multiple sclerosis lesions: interpreting positive..., Wisnieff [/bib_ref] , which are termed QSM rim positive (rim+) lesions. We propose an automated method to identify QSM rim + lesions, QSMRim-Net, using QSM and T2-FLAIR images of the brain. Our method is a deep convolutional neural network which consists of a two-branch network that fuses QSM and T2-FLAIR imaging features derived from a deep residual network [bib_ref] Deep residual learning for image recognition, He [/bib_ref] with lesion-level radiomic features from the QSM [bib_ref] Radiomics: Extracting more information from medical images using advanced feature analysis, Lambin [/bib_ref] , In addition, a Synthetic Minority Oversampling TEchnique (SMOTE)-based oversampling network (Deep-SMOTE) is developed to alleviate data imbalance issue caused by the small number of rim + lesions. This is the first deep learning method to fuse convolutional imaging features with radiomic features. Furthermore, QSMRim-Net with DeepSMOTE is the first end-to-end deep neural network that can be trained with online minority oversampling for rim + lesion classification.
Two previous methods have been developed to identify chronic active MS lesions on phase imaging [bib_ref] RimNet: A deep 3D multimodal MRI architecture for paramagnetic rim lesion assessment..., Barquero [/bib_ref]. RimNet [bib_ref] RimNet: A deep 3D multimodal MRI architecture for paramagnetic rim lesion assessment..., Barquero [/bib_ref] uses convolutional features, while Automatic Paramagnetic Rim Lesions (APRL) [bib_ref] Fully automated detection of paramagnetic rims in multiple sclerosis lesions on 3T..., Lou [/bib_ref] uses radiomic features. To put the performance of QSMRim-Net into context, we compare it to these two methods applied to the QSM using both lesion-level and patient-level performance metrics.
# Materials and methods
## Dataset
## Mri image acquisition and preprocessing
QSMRim-Net was evaluated on an MS imaging dataset collected at Weill Cornell [fig_ref] Table 1: Demographics information for the study cohort [/fig_ref]. The dataset consists of 172 MS patients enrolled in an ongoing prospective database for MS research. The database was approved by the local Institutional Review Board and written informed consent was obtained from all patients prior to their entry into the database.
The imaging was performed on a 3 T Magnetom Skyra scanner (Siemens Medical Solutions, Malvern, PA, USA). The Siemens scanning protocol consisted of the following sequences: 1) 3D sagittal T1weighted (T1w) MPRAGE: Repetition Time (TR)/Echo Time (TE)/ Inversion Time (TI) = 2300/2.3/900 ms, flip angle (FA) = 8 - , GRAPPA parallel imaging factor (R) = 2, voxel size = 1.0 × 1.0 × 1.0 mm3; 2) 2D axial T2-weighted (T2w) turbo spin echo: TR/TE = 5840/93 ms, FA = 90 - , turbo factor = 18, R = 2, number of signal averages (NSA) = 2, voxel size = 0.5 × 0.5 × 3 mm3; 3) 3D sagittal fat-saturated T2w fluid attenuated inversion recovery (T2-FLAIR) SPACE: TR/TE/TI = 8500/ 391/2500 ms, FA = 90 - , turbo factor = 278, R = 4, voxel size = 1.0 × 1.0 × 1.0 mm3. For axial 3D multi-echo GRE sequence for QSM: axial field of view (FOV) = 24 cm, TR/TE1/ΔTE = 48.0/6.3/4.1 ms, number of TEs = 10, FA = 15 - , R = 2, voxel size = 0.75 × 0.93 × 3 mm3, scan time = 4.2 min. QSM images were reconstructed by MEDI-0 [bib_ref] MEDI+ 0: morphology enabled dipole inversion with automatic uniform cerebrospinal fluid zero..., Liu [/bib_ref] algorithm from multi-echo GRE data. T2-FLAIR images were then preprocessed using the FSL toolbox [bib_ref] Advances in functional and structural MR image analysis and implementation as FSL, Smith [/bib_ref]. We applied N4 inhomogeneity correction algorithm to the acquired images and linearly co-registered T2-FLAIR images to the magnitude space of QSM.
## Lesion segmentation and rim + lesion annotation
T2-FLAIR lesion masks were created for all patients in the dataset. These masks were obtained by segmenting the T2-FLAIR image using the LST-LPA algorithm in the LST toolbox version 3.0.0 (www.statisticalmo delling.de/lst.html) [bib_ref] An automated tool for detection of FLAIR-hyperintense whitematter lesions in multiple sclerosis, Schmidt [/bib_ref] , followed by manual editing, and finalized by the consensus of two expert raters. Confluent lesions may occur when pathologically distinct lesions grow close to each other and form a large spatially connected lesion. These confluent lesions in the dataset were identified, then broken up and labeled by a human expert. After lesion segmentation and confluent lesion separation, a total of 4163 individual lesions were identified. Masks were further edited on the QSM image to ensure that these masks matched the lesion on QSM. Examples of a rim + and rim− lesion on QSM and T2-FLAIR are shown in [fig_ref] Figure 1: Example of MS lesions on an axial slice of the QSM [/fig_ref]. and an overview of the annotation process is shown in .
Rim + and rim− lesions were manually annotated by two human experts, who reviewed each of the 4,163 T2-FLAIR lesions for rim status on the QSM. The two reviewers had moderate agreement, with a kappa of 0.59. For lesions with disagreement, a final consensus was obtained from a third human expert. After the rim lesion annotation, 177 lesions were identified as rim + lesions and 3,986 lesions were identified as rim− lesions. An overview of the annotation process is shown in . and the distribution of rim + lesions per patient is shown in [fig_ref] Figure 3: Distribution of the number of paramagnetic rim lesions [/fig_ref].
# Methodology
## Network architecture
QSMRim-Net is a two-branch network consisting of four parts: a convolutional network for image feature extraction, a fully connected network for radiomic feature extraction, a SMOTE-based oversampling network for synthesizing rim + features in the latent feature space, and a final classifier that outputs the probability that a lesion is rim+ (see [fig_ref] Figure 4: Schematic of the proposed QSMRim-Net for paramagnetic rim lesion identification [/fig_ref]. QSMRim-Net takes as input the QSM, T2-FLAIR, and a lesion segmentation mask (described in the section above). For image feature extraction over the lesion segmentation mask, we use a deep residual network (ResNet) [bib_ref] Deep residual learning for image recognition, He [/bib_ref] with 18 layers as our backbone network. We modified the convolutional kernels from 2D to 3D, used two input channels to accommodate the QSM and T2-FLAIR images, and used two categories (rim + and rim− ) for the last linear layer. For radiomic feature extraction over the lesion segmentation mask, radiomic features [bib_ref] Cardiac computed tomography radiomics, Kolossváry [/bib_ref] [bib_ref] Radiomic features are superior to conventional quantitative computed tomographic metrics to identify..., Kolossváry [/bib_ref] were calculated on the QSM (described in detail in the section below). The multilayer perceptron (MLP) for radiomic feature extraction consists of two fully connected layers. The first layer is a linear layer followed by a onedimensional batch normalization (Ioffe and Szegedy, 2015), a Swish activation function, and a dropout layer. The second layer has the same structure as the first layer, except that it does not include a dropout layer. To fuse the convolutional and radiomic features, we performed vector concatenation on the second dimension for feature vectors from both the output of the residual network and the MLP and processed the new feature vector with another fully connected layer (see . To alleviate the data class imbalance issue, we further applied the DeepSMOTE network (described in detail in the Section 2.2.3) to oversample these latent features of rim + lesions during the training phase. The result of the model is a probability that a lesion is a rim + lesion. This probability can be thresholded to create a binary indicator of a lesion being rim+.
# Radiomic feature analysis
Radiomic features have been shown to be effective in many applications of medical image analysis [bib_ref] Radiomic phenotype features predict pathological response in non-small cell lung cancer, Coroller [/bib_ref] [bib_ref] Radiomic features are associated with EGFR mutation status in lung adenocarcinomas, Liu [/bib_ref] [bib_ref] Advancing the cancer genome atlas glioma MRI collections with expert segmentation labels..., Bakas [/bib_ref] [bib_ref] Estimation of Multiple Sclerosis lesion age on magnetic resonance imaging, Sweeney [/bib_ref]. For QSMRim-Net, radiomic features were calculated over each lesion using the pyradiomics package [bib_ref] Computational radiomics system to decode the radiographic phenotype, Van Griethuysen [/bib_ref]. Specifically, we calculated four different types of radiomic features on the input image: 1) first-order statistics such as harmonic mean and geometric mean, 2) gray-level co-occurrence matrix (GLCM) statistics such as interquartile range and energy sum, 3) gray-level run-length matrix (GLRLM) statistics such as runpercentage, and 4) geometric-based parameters such as ratio of lesion surface to volume. In addition, Coiflet wavelet filters were applied to yield the 8 decompositions of the input image and radiomic features were calculated on the wavelet images. Wavelet filters were implemented with the PyWavelet package [bib_ref] PyWavelets: a Python package for wavelet analysis, Lee [/bib_ref]. In total, 527 radiomic features were calculated over each lesion on the QSM for our model.
## Smote-based oversampling network
Rim + lesions are rare, with a prevalence of 4.25% in our dataset. This poses a great challenge to training any learning model. To overcome this challenge, we propose DeepSMOTE, a novel oversampling network that leverages the latent features extracted from deep neural network. Intuitively, DeepSMOTE can be thought of as finding the two nearest neighbors of each rim + lesion and taking a linear combination
## Fig. 2.
Schematic of the rim + and rim-lesion annotation process. First, we use LST [bib_ref] An automated tool for detection of FLAIR-hyperintense whitematter lesions in multiple sclerosis, Schmidt [/bib_ref] to obtain an initial lesion segmentation mask. Second, a human expert performs manual correction and confluent lesion separation, followed by mask edits based on QSM. Third, rim + lesions are manually annotated by two human experts, followed by consensus from a third expert.
of the lesion's features with each of the neighbors' features to produce synthetic observations. DeepSMOTE consists of a multi-layer perceptron (MLP) for feature transformation followed by the synthetic sample generation. The MLP is designed in a similar style as the network branch for radiomic feature extraction, where there are two consecutive fully connected networks, each having a linear layer, a 1-dimensional batch normalization, and a Swish activation function. The MLP is used to fuse features from the two network branches and reduces the feature dimension from 1024 to 512 for efficient computation (Figs. 4 and 5). Next, the two nearest neighbors for each rim + lesion in the mini-batch are determined using Euclidean distance. To train QSMrim-Net on a modern GPU, a small portion (mini-batch) of size N (64 samples in a mini-batch in our implementation) from the entire training dataset was randomly sampled during one forward-backward pass. Suppose there are n rim + lesions in a particular mini-batch. Let x p (p ∈ {1,2,3⋯n}) be the latent feature vector of a rim + lesion in the mini-batch. Let x p1 and x p2 be the two nearest rim + lesions to this lesion in the mini-batch, with respect to Euclidean distance. We generate two synthetic samples:
[formula] x p1 = α p1 x p + ( 1 − α p1 ) x p1 [/formula]
(1)
[formula] x p2 = α p2 x p + ( 1 − α p2 ) x p2(2) [/formula]
where α p1 and α p2 are randomly generated numbers in (0,1) to form a linear combination of the observation x p with x p1 and x p2 respectively. The result, x p1 and x p2 , are then the linear combinations of the rim + lesion and its nearest neighbors. This results in N +2n observations on which to train the network, the N samples in the mini-batch and the and the 2n synthetic samples. DeepSMOTE differs from the original SMOTE The QSM image and the lesion mask are used to extract radiomic features, followed by feature extraction of an MLP. A tensor concatenation operation is performed to fuse convolutional and radiomic features, and a DeepSMOTE layer is used to perform synthetic minority feature over-sampling during the training phase. algorithm, which oversamples the minority class by the reciprocal of the percentage of the minority class present in the dataset. DeepSMOTE instead samples 2n synthetic samples for each mini-batch during training, as each forward-backward pass of the deep neural network is done in a mini-batch of the entire dataset and oversampling too many rim + lesions in a single mini-batch may result in overfitting of the network.
## Training details
We applied a stratified five-fold cross-validation procedure to train and validate the performance of QSMRim-Net and the other methods. The stratified procedure was performed to balance the number of rim + lesions in each of the five folds. As seen in [fig_ref] Figure 3: Distribution of the number of paramagnetic rim lesions [/fig_ref] , we first grouped subjects into four groups, where the first group contained subjects with no rim + lesion, the second subjects with 1-3 rim + lesions, the third subjects with 4-6 rim + lesions, and the fourth subjects with more than 6 rim + lesions. The data was then randomly split into the five folds within each of these groups. All experiments were conducted with this stratified five-fold cross validation setting.
Input images were cropped into image patches with a fixed size 32 × 32 × 16 voxels, followed by a masking out of non-lesion area. The largest rim + lesion had a size of 15 × 20 × 5 voxels and only 7 rim− lesions were above the size of 32 × 32 × 16. As many lesions are confluent and may overlap in the patch, we only left the target lesion area in the patch. We then performed data augmentation to improve the performance of the model, providing better model generalizability. For augmentation in the training set, lesions were moved to align their center of mass to the geometric center of the image patch. We then used random flipping, random affine transformations, and random blurring to augment our data. Flipping was performed on an orthogonal direction randomly chosen from the axial, coronal, or sagittal direction. Affine transformations were performed with a random scale ranging from 0.95 to 1.05 and a random rotation degree between − 5 and 5 - . The final transformed patch was obtained after a trilinear interpolation. The blurring was performed using a random-sized Gaussian filter where the kernel radius was determined by4σ + 0.5. The voxel size of our QSM image was 0.75 × 0.75 × 3, thus for the coronal and sagittal direction, we randomly sampled σ ∼ U (0.1, 0.95), and for the axial direction we randomly sampled σ ∼ U (0.03, 0.3).
Version 3.7 of Python was used for all analysis. We implemented our network using the PyTorch library on a computer equipped with a single Nvidia 1080Ti GPU. The Adam algorithm; with an initial learning rate of 0.0001 and multistep learning rate scheduler with milestones at 50%; 70%, and 90% of the total epochs, was used to train the network weights. A mini-batch size of 64 was used for training, and training was stopped after 40 epochs. We used three random seeds to train three models for each fold and the final prediction result was determined by majority voting. We performed a sensitivity analysis and found that three random seeds performed well in terms of computational cost versus performance increase as a balance of all performance metrics.
# Comparator methods
Two automated methods have been developed to identify chronic active lesions on phase images [bib_ref] RimNet: A deep 3D multimodal MRI architecture for paramagnetic rim lesion assessment..., Barquero [/bib_ref]. RimNet [bib_ref] RimNet: A deep 3D multimodal MRI architecture for paramagnetic rim lesion assessment..., Barquero [/bib_ref] develops a multi-modal VGGNetto extract rim information from image patches of the phase and T2-FLAIR images. APRL [bib_ref] Fully automated detection of paramagnetic rims in multiple sclerosis lesions on 3T..., Lou [/bib_ref] applies a SMOTE and a random forest model to first-order radiomic features derived from individual lesions on the phase images. To evaluate QSMRim-Net, we compared the performance of the proposed algorithm with these two methods. Both methods were originally implemented on the phase, therefore we adapted these methods to a QSM implementation for use with our data. For RimNet, we used the QSM image along with the T2-FLAIR image as the network inputs. For APRL, we used the QSM image to extract the first order radiomic features as done in the original implementation. We applied SMOTE as done in the original APRL method to oversample the rim + lesion features by the reciprocal of its percentage present in our dataset. In addition to APRL, we also evaluated a neural network with the radiomic features, which is denoted as APRL (NN). The APRL (NN) uses the same network architecture as the radiomic branch of our QSMRim-Net and uses all 527 radiomic features instead of only the first order radiomic features. In the remainder of the manuscript, we refer to the implementation of APRL with the random forest as APRL (RF).
## Statistical evaluation
To evaluate the performance of each method, partial receiver operating characteristic (pROC) curves with false positive rates up to 0.1 and precision-recall (PR) curves of the different validation folds were interpolated using piecewise constant interpolation and averaged to show the overall performance at the lesion-level. For each curve, the area under the curve (AUC) was computed directly from the interpolated and averaged curves. As rim + lesions are rare and a small subset of the total number of lesions (4.25% of the lesions), allowing a high false positive rate threshold would produce results that are not clinically relevant. We therefore examine the pROC for false positive rates between 0 and 0.1. In addition, to create binary indicators of rim + versus rim− lesions, we thresholded the model probabilities to maximize the F1 score. The F1 score is the harmonic mean of precision (positive predictive value) and sensitivity, where F 1 = 2⋅ Precision- Sensitivity Precision+Sensitivity . We selected the F1 score as it a balance of two metrics (positive predictive value and sensitivity) as opposed to optimizing on a single metric. The thresholds for different folds of the cross-validation were chosen separately, and all results were obtained by a concatenation of the results for the five folds. For the lesion-wise analysis, accuracy, F1 score, sensitivity, specificity, and positive predicted value (PPV) were used to characterize the performance of each automated method. The results for accuracy, specificity, and PPV were non-parametrically bootstrapped 500 times on subject to test for statistically significant differences between the methods.
We also assessed performance at the subject-level. We used the F1 score criteria for thresholding and compared the number of predicted rim + lesions and the expert count number of rim + lesions for each subject. Pearson's correlation coefficient was used to measure the correlation between the two values. Mean Squared Error (MSE) was also used to measure the averaged accuracy for the model predicted count.
## Ablation study
We conducted an ablation study to evaluate the effects of changing components of the QSMRim-Net network using ResNet as the backbone network. First, we examined two different fusion methods for the imagelevel features. In RimNet, a multi-modal architecture was used to fuse image-level features. In QSMRim-Net we concatenated images in the channel dimension of the tensor for input into the network. Second, we investigated the effects of incorporating the radiomic feature as a separate network branch. Third, we studied how the DeepSMOTE layer affects the network performance. The following four schemes were evaluated based on the method for fusing the QSM and T2-FLAIR imaging features and whether we adopt radiomic feature or DeepSMOTE layer: 1) images were fused as in QSMRim-Net with no radiomic features, 2) images were fused as in RimNet with no radiomic features, 3) images were fused as in QSMRim-Net with radiomic features, 4) images were fused as in QSMRim-Net with radiomic feature and DeepSMOTE network was adopted. shows the lesion-wise performance metrics of the proposed QSMRim-Net and the other methods, using the F1-score as a threshold. QSMRim-Net outperformed the competitors in all metrics used for evaluation. With a slightly higher overall accuracy and specificity with other methods, QSMRim-Net resulted in a 9.8% and 23.3% improvement in F1 score, 3.5% and 14.3% improvement in sensitivity and 16.8% and 33.1% improvement in PPV compared to Rim-Net [bib_ref] RimNet: A deep 3D multimodal MRI architecture for paramagnetic rim lesion assessment..., Barquero [/bib_ref] and APRL (RF) [bib_ref] Fully automated detection of paramagnetic rims in multiple sclerosis lesions on 3T..., Lou [/bib_ref] , respectively. The increase in accuracy and specificity were very moderate for QSMRim-Net compared to the other methods. This is due to the fact that there are many rim− lesions, so the accuracy and specificity for all methods is very high. From the non-parametric bootstrap, no significant differences were found between the methods for accuracy, sensitivity, specificity, and positive predictive value. shows the pROC curves and the PR curves for the different methods. The proposed QSMRim-Net obtained 4.68% and 21.01% higher pROC AUC (0.760) than Rim-Net (0.726) and APRL (RF) (0.628), meaning that for more clinically relevant false positive rates of less than 0.1, QSMRim-Net has higher performance than the other methods. The proposed QSMRim-Net out-performed both Rim-Net and APRL (RF) by 9.8% and 60.0% respectively in PR AUC, indicating the effectiveness of the proposed DeepSMOTE network and fusion of information from both convolutional image and radiomic features. Interestingly, APRL (NN) that uses our neural network architecture outperformed APRL (RF) significantly in both pROC AUC and PR AUC, indicating the potential for the neural network to exploit high-dimensional non-linear relationships from the radiomic features.
# Results
# Lesion-wise analysis
# Subject-wise analysis
We calculated the predicted count of rim + lesions from each of the models and compared this to consensus expert count for each subject. The consensus expert count of rim + lesions ranged from 0 to 17 among the 172 subjects, with a median of 2 rim + lesions among subjects with at least one rim [fig_ref] Figure 1: Example of MS lesions on an axial slice of the QSM [/fig_ref]. The predicted count of rim + lesions from QSMRim-Net ranged from 0 to 14, with a median of 1 rim + lesion among the subjects with at least one rim (IQR 1-4).
The Pearson's correlation between the predicted count and the gold standard count was 0.89 (95% CI: 0.86, 0.92). [fig_ref] Figure 7: The predicted count of rim + lesions from QSMRim-Net versus the expert... [/fig_ref] shows the scatterplot for the predicted count versus the gold standard count, along with the identity line. The Pearson's correlations for the other methods were found to be lower than QSMRim-Net: 0.88 (95% CI: 0.85, 0.91) for APRL (NN), 0.77 (95% CI: 0.70, 0.82) for APRL (RF), and 0.75 (95% CI: 0.67, 0.81) for Rim-Net. The performance of QSMRim-Net on the patient-level is statistically significantly higher than that of APRL (RF) and Rim-Net, but not APRL (NN). The MSE for the predicted count of the QSMRim-Net was 0.98. The MSE for the other methods were found to be higher: 1.02 for APRL (NN), 2.26 for APRL (RF), and 2.47 for Rim-Net. shows the results from the ablation study. For image fusion, the fusion technique in QSMRim-Net outperformed the fusion technique for RimNet. For radiomic feature fusion, we can see that the network with radiomic feature fusion performs better than the networks without radiomic feature fusion. The network with Deep SMOTE oversampling outperforms the other variants in all metrics.
## Ablation study
## Prediction time and software
After fitting the QSMRim-Net model, creating rim + lesion probabilities for a subject took on average 4.6 s (sd = 4.1 s). A software implementation of the fitted QSMRim-Net model can be found on GitHub at https://github.com/tinymilky/QSMRim-Net
# Discussion
In this paper, we propose QSMRim-Net, a deep convolutional neural network for identifying rim + lesions on QSM MRI. This is the first study to introduce an end-to-end two-branch network enabled with the DeepSMOTE oversampling technique that can effectively fuse both convolutional image and radiomic features.
Our QSMRim-Net achieved better performance than two previously developed methods when applied to the QSM, APRL (RF) [bib_ref] Fully automated detection of paramagnetic rims in multiple sclerosis lesions on 3T..., Lou [/bib_ref] and Rim-Net [bib_ref] RimNet: A deep 3D multimodal MRI architecture for paramagnetic rim lesion assessment..., Barquero [/bib_ref]. The results on a lesion-level were not found to be statistically significant, and this can be attributed to the small number of rim + lesions in the dataset. It is also important to interpret these results considering that APRL (RF) and Rim-Net were originally designed for phase imaging and were applied to the QSM for this study. The accuracy of QSMRim-Net also outperformed the accuracy of the original APRL and Rim-Net on phase imaging, with accuracies of 0.8 and 0.946, respectively, compared to QSMRim-Net's accuracy of 0.976. The increase in performance can be attributed to our carefully designed convolutional neural network architecture [fig_ref] Figure 4: Schematic of the proposed QSMRim-Net for paramagnetic rim lesion identification [/fig_ref]. Rim-Net used VGG-Netand APRL (RF) used a random forest. Our QSMRim-Net adopted a ResNet [bib_ref] Deep residual learning for image recognition, He [/bib_ref] architecture that uses identity shortcut connections to prevent gradient vanishing, which reduces computational complexity and allows for the training of deeper networks than VGG-Net. We also observed that a neural network with MLP (APRL (NN)) achieved better performance than a random forest model (APRL (RF)) on radiomic features. This shows that a properly designed neural network can extract discriminative information from highly non-linear radiomic feature data. Another contributor to QSMRim-Net's performance is that it effectively fuses the complementary information from the convolutional image and radiomic features. We also showed in the ablation study that the neural network architecture design choices for fusing features from different sources is important for improving rim + lesion identification performance.
In addition to the deep neural network model, our result may also benefit from the utilization of QSM. Compared to the phase images used in the original implementation of Rim-Net and APRL (RF), QSM can measure the underlying tissue apparent magnetic susceptibility, enabling the quantification of specific biomarkers, such as iron, that are independent of imaging parameters. Rim + lesions are characterized by Lesion-wise results of the QSMRim-Net and other methods using a stratified five-fold cross-validation scheme. PPV denotes positive predictive value, TP# denotes the number of true positives, FP# denotes the number of false positives, FN# denotes the number of false negatives, and TN# denotes the number of true negatives. For TP#, FP#, FN#, and TN# the percentage of lesions in each designation is also provided. The best performing method for each of metrics is bolded. a paramagnetic rim with iron deposited at the edge of the lesion. QSM is sensitive to such magnetic susceptibility changes and provides consistent measurements of the susceptibility value of the rim across patients and scanners, which is beneficial for a machine learning model such as deep neural network to learn patterns of rim + and rim− lesions. While our QSMRim-Net is inspired by Rim-Net and APRL (RF), we found that implementing these two methods on our dataset using QSM resulted in 10.2% and 25.0% improvement of PPV, 7.8% and 26.5% reduction in sensitivity respectively, compared to its original implementation of Rim-Net and APRL (RF) on phase images. The APRL algorithm used only first-order radiomic features for identifying chronic active MS lesions. For QSMRim-Net, we expanded upon this set of features, using the radiomic features of first-order statistics, GLCM statistics, GLRLM statistics and geometric-based parameters combined with wavelet filters [bib_ref] Radiomics: the facts and the challenges of image analysis, Rizzo [/bib_ref]. There are other radiomic features, such as gray level size zone matrix (GLSZM) statistics, neighboring gray tone difference matrix (NGTDM) statistics, and gray level dependence matrix (GLDM) statistics that would be interesting to explore in the QSMRim-Net model. Other filters, such as a Laplacian filter, could also be used with the radiomic features. Due to the limited sample size available for this analysis, we limited the number of radiomic features and filter combinations. Future work exploring if there are performance gains from the addition of other radiomic features and filters is warranted.
The high imbalance of rim + and rim− lesions is a challenge for machine learning models. We found that APRL (RF) with SMOTE oversampling outperforms its counterpart without SMOTE by 2.9% in F1 score, indicating the importance of oversampling of the minority class. While it is feasible to synthesize radiomic features by linear interpolation using SMOTE, it is not possible to synthesize meaningful images by pixel-level linear interpolation. It has been shown empirically that deep neural networks can linearize the manifold of images into Euclidean subspace [bib_ref] Better mixing via deep representations, Bengio [/bib_ref]. The partial receiver operating characteristic (pROC) curve and precision-recall (PR) curves for the proposed (QSMRim-Net) and comparator methods. AUC denotes the area under the curve. We use clinically relevant false positive rates of less than 0.1 to compute the pROC AUC, in order to account for the rare nature of rim + lesions. Our QSMRim-Net algorithm outperformed all other algorithms on pROC AUC (FPR < 0.1) and PR AUC. Ablation study on QSMRim-Net and its variants. PPV denotes positive predictive value, TP# denotes the number of true positives, FP# denotes the number of false positives, FN# denotes the number of false negatives, and TN# denotes the number of true negatives. Image Fusion indicates whether the model performed image-level feature fusion, Radiomic Fusion indicates whether the model performed feature fusion between image and radiomic features, and Deep SMOTE indicates whether the model applied the DeepSMOTE network for rim + feature oversampling. The best performing method for each of metrics is bolded. 2017), enabling the possibility of linear interpolation using latent features from deep layers from the network. Inspired by the SMOTE results from APRL (RF) and the deep feature interpolation, we propose Deep-SMOTE network to alleviate the data imbalance issue, and the results in shows the effectiveness of applying DeepSMOTE for data oversampling.
For the QSMRim-Net performance results, we thresholded the output probabilities from the algorithm using the optimized F1 score, resulting in a sensitivity of 0.678 for detecting rim + lesions. However, in a research scenario or in clinical practice, missing any rim + lesions may not be acceptable. Thus, to demonstrate the performance of QSMRim-Net in these settings, we also applied a high sensitivity threshold, using the largest sensitivity below 0.95. In practice, experts can use this high sensitivity threshold to reduce the number of lesions that need to be manually reviewed for rim + status. QSMRim-Net performed with a false positive number of 538 lesions, a reduction of 14.2% and 19.1% compared to Rim-Net and APRL (RF). With QSMRim-Net, in this dataset, only 715 lesions would need to be reviewed by an expert instead of all 4163 lesions, saving 82.8% of review time.
We also obtained results for all rim + lesion identification algorithms on a patient-level, showing that QSMRim-NET outperformed the other methods. In a previous study, as the overall total lesion burden increased, patients with at least one rim + lesion on QSM performed worse on both physical disability and cognitive assessments. Having four or more chronic active lesions on phase imaging has also been shown to correlate with disability [bib_ref] Association of chronic active multiple sclerosis lesions with disability in vivo, Absinta [/bib_ref]. In addition, these lesions have been used diagnostically to differentiate patients with MS from other neurological conditions [bib_ref] Paramagnetic rim lesions are specific to multiple sclerosis: an international multicenter 3T..., Maggi [/bib_ref]. If rim + lesions are to be used prognostically or diagnostically, then the patient-level results may be more important than identifying individual rim + lesions for clinical translation.
To further understand the limitations of the QSMRim-Net algorithm's performance, we also examined the false positive and false negative results. The false positive and negative results tended to be lesions that the two human experts did not agree upon. Using the F1 score threshold, 40.9% of the false positive lesions and 35.1% of false negatives lesions were lesions with human expert disagreement. This contrasts with 22.5% of the true positives and 2.4% of the true negatives. Visual examination of the lesions showed that veins were challenging for the algorithm, resulting in false positives. On QSM the vein is hyperintense. In cases where the vein formed a rim-like shape in a rim-lesion, this often resulted in a false positive . When a rim + lesion was found heterogeneously hyperintense, this often resulted in a false negative . Rim-lesions with a higher intensity value on the QSM tended to cause false positives, while rim + lesions with a lower intensity value tended to cause false negatives. Our future work involves further understanding of these patterns to reduce FPs and FNs from the algorithm in order to improve research and clinical translation.
One limitation of this work is that it relies on manual lesion segmentations that have been edited further on the QSM . Future work involves pairing QSMRim-Net with an automated T2-FLAIR lesion segmentation algorithm, such as All-Net [bib_ref] ALL-Net: Anatomical information lesion-wise loss function integrated into neural network for multiple..., Zhang [/bib_ref] with geometric loss [bib_ref] Geometric Loss for Deep Multiple Sclerosis lesion Segmentation, Zhang [/bib_ref] and attention-based approaches [bib_ref] RsaNet: Recurrent slice-wise attention network for multiple sclerosis lesion segmentation, Zhang [/bib_ref] [bib_ref] Efficient Folded Attention for Medical Image Reconstruction and Segmentation, Zhang [/bib_ref] , followed by an automated method to separate confluents lesions [bib_ref] Memory U-Net: Memorizing Where to Vote for Lesion Instance Segmentation, Zhang [/bib_ref]. It is important to acknowledge that pairing the algorithm with an automated lesion segmentation algorithm may increase the error of QSMRim-Net, as there may be false positive and false negative lesions in the automated segmentation. Future work will examine a full pipeline for automated segmentation followed by rim + lesion identification. We plan to adapt and train the algorithm to work directly on T2-FLAIR lesion segmentations.
A limitation of our study was the number of rim + lesions available to train and validate the model. We had a total of 177 rim + lesions available for this analysis. Future work will involve larger datasets to validate the QSMRim-Net model. An additional challenge for the algorithm is the rare nature of rim + lesions. Only 4.25% of the lesions in this study were identified as rim + lesions, posing a great challenge to learning based methods. We proposed DeepSMOTE for data oversampling to alleviate the data class imbalance, but as future work we . Visual examples of a true positive, a false positive, a false negative and a true negative produced by QSMRim-Net. The QSM is shown on the left and the T2-FLAIR on the right. The lesion of interest indicated with a red arrow. (A) A rim + lesion that is correctly identified. (B) A rim-lesion with a vein forming a rim-like shape that was falsely identified as rim + by QSMRim-Net. (C) A rim + lesion with that was missed by QSMRim-Net. (D) A rim-lesion that is correctly identified. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) plan to develop techniques on imbalance-aware loss functions, such as geometric loss [bib_ref] Geometric Loss for Deep Multiple Sclerosis lesion Segmentation, Zhang [/bib_ref].
Another limitation of this study is the lack of an independent cohort to support the generalizability of the method. Future work will be to validate the findings of this study to additional data acquired on different scanners and at other imaging centers. One challenge of generalizing this work to different scanners and centers is that there may be site and scanner specific effects in the data. While QSM has been shown to be reproducible across scanners [bib_ref] Reproducibility of quantitative susceptibility mapping in the brain at two field strengths..., Deh [/bib_ref] , there may still be some effects. Future work will use intensity normalization [bib_ref] Statistical normalization techniques for magnetic resonance imaging, Shinohara [/bib_ref] and image harmonization methods [bib_ref] Harmonization of multi-site diffusion tensor imaging data, Fortin [/bib_ref] [bib_ref] Harmonization of cortical thickness measurements across scanners and sites, Fortin [/bib_ref] [bib_ref] Adjusting batch effects in microarray expression data using empirical Bayes methods, Johnson [/bib_ref] directly on the images themselves as well as on the radiomic features.
A further limitation is inter-rater variability in identifying rim + lesions. To reduce the impact of this, we had two raters evaluate lesions for rim + status and any disagreements were adjudicated by a third reviewer. In this work, we used a binary classification of whether a lesion had a rim. As discussed in [bib_ref] Fully automated detection of paramagnetic rims in multiple sclerosis lesions on 3T..., Lou [/bib_ref] there are many factors that influence the strength of the rim + lesion signature on QSM and a more nuanced approach to classify these lesions may be beneficial.
In conclusion, QSMRim-Net is the first deep learning-based method that integrates DeepSMOTE for data oversampling and fuses modern convolutional imaging features with traditional radiomic features to automatically identify rim + MS lesions on QSM. QSMRim-Net outperformed other state of the art methods on rim + lesion identification on QSM and has the potential to aid in the clinical translation for the rim + lesion biomarker.
## Credit authorship contribution statement
[fig] Figure 1: Example of MS lesions on an axial slice of the QSM (left) and corresponding axial slice of the T2-FLAIR (right). The digit 1 marked with red indicates a rim + lesion and the digit 2 marked with green indicates a rim-lesion. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) [/fig]
[fig] Figure 3: Distribution of the number of paramagnetic rim lesions (rim + lesions) per subject in the Weill Cornell dataset. The plot is colored by the groups used for the stratified five-fold cross validation. [/fig]
[fig] Figure 4: Schematic of the proposed QSMRim-Net for paramagnetic rim lesion identification. (Top Row) The deep residual network takes in both QSM and FLAIR images to extract convolutional features. (Bottom Row) [/fig]
[fig] Figure 5: Schematic of the DeepSMOTE network layer. N is the number of samples in a training mini-batch, and n is the number of rim + samples in the mini-batch. The input features go through an MLP for feature transformation, followed by selecting rim + samples from the mini-batch. The transformed rim + features are used to generate a Euclidean distance-based similarity followed by latent feature interpolation. The original features and the oversampled features are concatenated, resulting in a total of N +2n samples in the output of DeepSMOTE. [/fig]
[fig] Figure 7: The predicted count of rim + lesions from QSMRim-Net versus the expert human count (ρ = 0.89(95%CI : 0.86, 0.92), MSE = 0.98). Points in the plot have been jittered for better visualization. The linear regression line for the predicted count versus the gold standard count with 95% CI is also shown (solid blue) along with the identity line (dashed blue). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) [/fig]
[table] Table 1: Demographics information for the study cohort. [/table]
|
A fixed-dose combination tablet of gemigliptin and metformin sustained release has comparable pharmacodynamic, pharmacokinetic, and tolerability profiles to separate tablets in healthy subjects
Background:In type 2 diabetes mellitus, fixed-dose combination (FDC) can provide the complementary benefits of correction of multiple pathophysiologic defects such as dysfunctions in glycemic or metabolic control while improving compliance compared with separate tablets taken together. The objective of the study reported here was to compare the pharmacodynamic (PD), pharmacokinetic (PK), and tolerability profiles of gemigliptin and extended-release metformin (metformin XR) between FDC and separate tablets.Methods:A randomized, open-label, single-dose, two-way, two-period, crossover study was conducted in 28 healthy male volunteers. Two FDC tablets of gemigliptin/metformin 25/500 mg or separate tablets of gemigliptin (50 mg ×1) and metformin XR (500 mg ×2) were orally administered in each period. Serial blood samples were collected up to 48 hours post-dose to determine dipeptidyl peptidase 4 (DPP-4) activity using spectrophotometric assay and concentrations of gemigliptin and metformin using tandem mass spectrometry. Geometric mean ratios (GMRs) of FDC to separate tablet formulations and their 90% confidence intervals (CIs) were calculated to compare the PD and PK parameters between the two formulations. Tolerability was assessed throughout the study. Results: The plasma DPP-4 activity-time curves of the FDC and the separate tablets almost overlapped, leading to a GMR (90% CI) of the FDC to separate tablets for the plasma DPP-4 activity and its maximum inhibition of 1.00 (0.97-1.04) and 0.92 (0.82-1.05), respectively. Likewise, all of the GMRs (90% CIs) of FDC to separate tablets for the area under the plasma concentration-time curve and maximum plasma concentration of gemigliptin and metformin fell entirely within the conventional bioequivalence range of 0.80-1.25. Both the FDC and separate tablets were well tolerated. Conclusion: The PD, PK, and tolerability profiles of gemigliptin and metformin XR in FDC and separate tablets were found to be comparable. The FDC tablet of gemigliptin and metformin sustained release can be a convenient therapeutic option in patients with type 2 diabetes mellitus requiring a combination approach.
# Introduction
Type 2 diabetes mellitus (T2DM) is a multifactorial disease threatening the health of all nations. Major pathophysiology of T2DM is characterized by insulin resistance and pancreatic β-cell failure. However, other mechanisms such as incretin deficiency submit your manuscript | www.dovepress.com
## Dovepress
## Dovepress
## 730
Park et al or resistance to incretin in the gastrointestinal tract also play an important role in the development of glucose intolerance in T2DM patients. Therefore, effective treatment of T2DM may require an approach using two or more drugs in combination to correct those multiple pathophysiological defects. [bib_ref] From the triumvirate to the ominous octet: a new paradigm for the..., Defronzo [/bib_ref] Although metformin has been recommended as a first-line oral antidiabetic drug, combination therapy with other antidiabetic agents may offer additional glycemic control. [bib_ref] Clinical Guidelines Committee of the American College of Physicians. Oral pharmacologic treatment..., Qaseem [/bib_ref] For this purpose, a dipeptidyl peptidase 4 (DPP-4) inhibitor has advantages in reducing the side effects of conventional antidiabetic agents such as weight gain (experienced with sulfonylureas, meglitinides, insulin, and thiazolidinediones), and hypoglycemia (with sulfonylureas, meglitinides, and insulin). [bib_ref] Clinical management strategies for type 2 diabetes, Cefalu [/bib_ref] Furthermore, the additive antidiabetic effects of metformin and DPP-4 inhibitors may lead to better metabolic control. For example, in patients with T2DM who had been inadequately controlled by metformin alone, the addition of a DPP-4 inhibitor not only lowered glycated hemoglobin (HbA 1c ), and fasting and 2-hour postprandial plasma glucose, but also improved β-cell function (homeostatic model assessment [HOMA]-β) and the proinsulin/insulin ratio. [bib_ref] Sitagliptin Study 020 Group. Efficacy and safety of the dipeptidyl peptidase-4 inhibitor..., Charbonnel [/bib_ref] Gemigliptin (Zemiglo ® , LG Life Sciences, Seoul, South Korea) is a potent, selective, competitive, and long-acting DPP-4 inhibitor that was approved for use in patients with T2DM in June 2012 in Korea. [bib_ref] Gemigliptin, a novel dipeptidyl peptidase 4 inhibitor: first new anti-diabetic drug in..., Kim [/bib_ref] In a single ascending dose study, gemigliptin was rapidly absorbed after oral administration at 25-600 mg with a median time to peak concentration (T max ) of 2.0 hours (0.5-5.1 hour) post-dose and the mean terminal elimination half-life (t ½ ) ranged from 16.7 to 21.3 hours. Gemigliptin exhibited linear pharmacokinetic (PK) characteristics over the range of 50 to 400 mg in that study. The rate, but not the extent of absorption, appeared to be influenced by food intake. [bib_ref] Pharmacokinetics, pharmacodynamics, and tolerability of the dipeptidyl peptidase IV inhibitor LC15-0444 in..., Lim [/bib_ref] Following a single oral dose of extended-release metformin (metformin XR) (Glucophage ® , Bristol-Myers Squibb Company, Princeton, NJ, USA), a median value of T max was 7 hours (4-8 hours) post-dose and approximately 90% of the absorbed drug was eliminated via the renal route within the first 24 hours, with a t ½ of approximately 6.2 hours. Although the extent of absorption from the metformin XR tablet increased by approximately 50% when given with food, there was no effect of food on the maximum plasma concentration (C max ) and T max of metformin.In the dose-ranging Phase II clinical trial with gemigliptin, 50 mg was found to be the optimal dose based on a significant improvement in glycemic control and the overall safety profile. [bib_ref] A multicenter, randomized, placebocontrolled, double-blind phase II trial evaluating the optimal dose,..., Rhee [/bib_ref] Furthermore, in a previous drug-drug interaction study performed in healthy volunteers, multiple coadministration of gemigliptin (50 mg once-daily) and metformin (1,000 mg twice-daily) resulted in beneficial pharmacodynamic (PD) interactions such as an additive increase in the plasma concentration of active glucagon-like petide-1 (GLP-1) without untoward PK interactions at steady state. [bib_ref] Pharmacokinetic and pharmacodynamic interaction between gemigliptin and metformin in healthy subjects, Shin [/bib_ref] In this study, the tolerability profile was also comparable between gemigliptin alone and gemigliptin in combination with metformin. [bib_ref] Pharmacokinetic and pharmacodynamic interaction between gemigliptin and metformin in healthy subjects, Shin [/bib_ref] Therefore, 50 mg and 1,000 mg are likely to be the most preferred doses for gemigliptin and metformin, respectively, in combination therapy.
A novel fixed-dose combination (FDC) of gemigliptin/ metformin sustained release (SR) 25/500 mg (Zemimet ® SR 25/500 mg, LG Life Sciences) has been developed, which may lead to better compliance than two separate tablets taken together. Two tablets of the FDC of gemigliptin/metformin SR, which slowly release metformin over the dosing interval can be a convenient treatment option with a once-daily regimen. Like other FDC formulations, however, the efficacy, tolerability, and PK profiles of gemigliptin and metformin given as FDC should be comparable to those of the individual drugs given as separate tablets. Based on this understanding, the objective of the present study was to compare the PD, PK, and tolerability profiles of gemigliptin/metformin SR between FDC and separate tablets. To this end, a single-dose, two-way crossover, comparative PD and PK study was conducted in healthy volunteers.
# Methods subjects
Healthy male volunteers aged 20-45 years with a body mass index of 18.0-27.0 kg/m 2 and a fasting blood glucose level of 70-125 mg/dL were enrolled into the study if they presented no abnormalities based on medical history, physical examination, twelve-lead electrocardiogram, clinical laboratory tests, or urine drug screening (amphetamine, barbiturates, benzodiazepines, cannabinoids, cocaine, opiates, and cotinine). Written informed consent was obtained before any study-related procedure was performed. The study was approved by the Institutional Review Board at Seoul National University Hospital, Seoul, South Korea, and was conducted in accordance with the principles of the Declaration of Hel sinki and Good Clinical Practice standard (clinicaltrials.gov registration number: NCT01662674).
## Study design
This was a randomized, open-label, single-dose, two-way, two-period, crossover study. Eligible subjects randomly received two tablets of FDC of gemigliptin/metformin SR 25/500 mg (Zemimet SR 25/500 mg) in one period, and a gemigliptin 50 mg tablet (Zemiglo) plus two metformin XR Drug Design, Development and Therapy 2015:9 submit your manuscript | www.dovepress.com
## Dovepress
## Dovepress
## 731
FDC gemigliptin and metformin SR in healthy subjects 500 mg tablets (Glucophage) in the other period. There was a 7-day washout between the two periods, which was deemed appropriate given that this was longer than five-times the t ½ of both gemigliptin and metformin XR and no significant PK interactions were observed between the two drugs. [bib_ref] Pharmacokinetic and pharmacodynamic interaction between gemigliptin and metformin in healthy subjects, Shin [/bib_ref] In each period, subjects were hospitalized at the Clinical Trial Center at Seoul National University Hospital a day before administration of the study drug. After a high-fat meal (900 kcal; fat content of 50% or more), subjects orally took the study drug with 240 mL of water, and the study procedures were performed for 48 hours until discharge.
Within a week of completing Period II as planned, subjects reported to the Clinical Trials Center again for an end-of-study visit and laboratory tests.
## Sample-size calculation
The within-subject coefficient of variation for the PK para meters (C max , area under the time-concentration curve [AUC]) was assumed to be 20% for gemigliptin and metformin based on previous studies. [bib_ref] Pharmacokinetics, pharmacodynamics, and tolerability of the dipeptidyl peptidase IV inhibitor LC15-0444 in..., Lim [/bib_ref] [bib_ref] Pharmacokinetics and dose proportionality of extended-release metformin following administration of 1,000, 1,500,..., Cullen [/bib_ref] The number of subjects required to detect a difference of 20% or more in the PK parameters between the treatments (ie, FDC vs separate tablets) at a significance level of 0.05 and a power of 90% was 24. Assuming a drop-out rate of ~20%, a total of 28 subjects were to be enrolled.
## Blood collection
Serial blood samples were collected at time zero (ie, pre-dose); 30 minutes; 1 hour; and 1.5, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 10.0, 12.0, 16.0, 24.0, 36.0. and 48.0 hours post-dose using ethylenediaminetetraacetic acid and heparinized tubes for PD (DPP-4 activity) and PK (plasma concentrations of gemigliptin and metformin) analyses, respectively. Blood samples were centrifuged for 10 minutes at 4°C approximately at 1,550 and 1,910 g for PD and PK, respectively. Then 0.3 mL aliquots of plasma for gemigliptin analysis were placed into polypropylene tubes, which contained 0.3 mL of 5% formic acid in deionized water (high-performance liquid chromatography grade). Another 0.5 mL aliquot of plasma for metformin analysis was placed into a polypropylene tube (three tubes were prepared for each drug analysis). All plasma samples were stored at -20°C in the freezer for less than 24 hours. Then they were transferred to a deep freezer and stored at less than -70°C in the dark until required for sample analysis.
Plasma dipeptidyl peptidase 4 activity DPP-4 activity in plasma was determined using a continuous spectrophotometric assay with the substrate Gly-Pro-pNA (Bachem, Bubendorf, Switzerland), as has been reported previously. [bib_ref] Pharmacokinetic and pharmacodynamic interaction between gemigliptin and metformin in healthy subjects, Shin [/bib_ref] The assay was based on the ability of DPP-4 to cleave the substrate Gly-Pro-pNA into pNA, resulting in an increase in absorbance at 390 nm, where the absorbance was expressed as milli optical density (mOD). Enzymatic activity was calculated as the slope (in mOD/min) from 4 to 14 minutes. Intra-and inter-assay precisions were 2.28%-7.57% and 3.64%-12.75%, respectively. The lower limit of quantification was 1.43 mOD/min.
## Plasma concentrations of gemigliptin and metformin
Plasma concentrations of gemigliptin and metformin were determined using tandem mass spectrometry (MS/MS) (API™ 4000 LC/MS/MS system, AB Sciex, Framingham, MA, USA) with electrospray in positive ionization mode, coupled with high-performance liquid chromatography (Agilent 1100 series HPLC system, Agilent Technologies, Wilmington, DE, USA). Chromatographic separation was performed under gradient conditions using a Luna C8 column (30.0×2.0 mm, 3 μm; Phenomenex, Torrance, CA, USA) for gemigliptin and using a PC HILIC (hydrophilic interaction liquid chromato graphy column; 2.0 mm ID ×150 mm, Shiseido, Yokohama, Japan) for metformin, as reported previously. [bib_ref] Pharmacokinetic and pharmacodynamic interaction between gemigliptin and metformin in healthy subjects, Shin [/bib_ref] The calibration curves were linear over the range of 1-2,000 ng/mL for gemigliptin and 2-2,000 ng/mL for metformin (r0.9975 and r0.9983, respectively). The between-run coefficients of variation (CV) of gemigliptin and metformin were all 14.2% and the accuracy ranged between 96% and 104%.
## Pharmacodynamic and pharmacokinetic data analyses
## 732
Park et al 0 to infinity (AUC inf ). The terminal elimination rate constant (λ z ) was estimated from a regression line of log-transformed plasma concentrations versus time over the terminal log-linear portion. The t ½ was calculated as natural logarithm of 2 divided by λ z . AUC inf was calculated as AUC 0-48 h + C 48 /λ z , where C 48 is the plasma concentration at 48 hours post-dose.
## Tolerability assessment
The occurrence of adverse events (AEs) was recorded during the entire study period including at the end-of-study visit. Data from all the study participants who were administered the study drug at least once were included for tolerability assessment, which also considered changes from baseline in physical examination, vital signs, twelve-lead electrocardiograms, and clinical laboratory tests.
# Statistical analysis
The PD and PK parameters were summarized using descriptive statistics. Using log-transformed data, a general linear mixed effects model was developed to compare the PD (AUEC 0-48 h and I max ) and PK (AUC 0-48 h and C max ) parameters between treatments, where period, sequence, and treatment were fixed effects, while subjects nested in sequence was a random effect. Using the model, the geometric mean ratio (GMR) and its 90% confidence interval (CI) of FDC to individual drug given as separate tablets were estimated for the PD and PK parameters. Comparability was concluded if the 90% CI of the GMR for the PD or PK parameters was entirely contained within the conventional bioequivalence range of 0.80-1.25. P-values 0.05 were considered statistically significant. All statistical analyses were performed using SPSS software (v 18.0, IBM Corporation, Armonk, NY, USA).
# Results
## Demographic characteristics
A total of 28 subjects were enrolled into the present study, one of whom withdrew consent during the washout period due to a personal reason. Therefore, the 27 subjects who completed the study as planned were included in PD and PK analyses, while all of the enrolled subjects (n=28) were included in the tolerability analysis population. The mean ± standard deviation of age, height, weight, and body mass index was 27.4±4.5 years, 174.3±6.4 cm, 69.4±8.8 kg, and 22.8±2.4 kg/m 2 , respectively, none of which significantly differed between sequence groups (all P-values 0.05, Student's t-test).
## Pharmacodynamics
After a single oral administration of gemigliptin and metformin at 50 and 1,000 mg, respectively, the plasma DPP-4 activity was comparable between FDC and separate tablets as assessed by AUEC 0-48 h and I max . As a result, the GMR (90% CI) of FDC to separate tablets for AUEC 0-48 h and I max was 1.00 (0.97-1.04) and 0.92 (0.82-1.05), respectively. Furthermore, the mean plasma DPP-4 activity profiles of FDC and the separate tablets over time almost overlapped [fig_ref] Figure 1: Mean plasma dipeptidyl peptidase 4 [/fig_ref] , and no trend or systematic deviation was found in individual comparison for AUEC 0-48 h [fig_ref] Figure 1: Mean plasma dipeptidyl peptidase 4 [/fig_ref] and I max [fig_ref] Figure 1: Mean plasma dipeptidyl peptidase 4 [/fig_ref].
## Pharmacokinetics
The systemic exposures to gemigliptin and metformin after a single oral administration of 50 and 1,000 mg, respectively, given as FDC were similar to those when they were taken together as separate tablets. The mean plasma concentrationtime profiles of gemigliptin and metformin were superimposable [fig_ref] Figure 2: Mean plasma concentration-time profiles of [/fig_ref]. Furthermore, all of the GMRs (90% CI) of FDC to separate tablets for the C max and AUC 0-48 h of gemigliptin and metformin fell entirely within the conventional bioequivalence range of 0.80-1.25 . Other PK parameters, including apparent clearance and t ½ , were also comparable between FDC and separate tablets for gemigliptin and metformin.
## Tolerability
Only one drug-related AE (diarrhea) occurred after FDC administration, whereas three drug-related AEs (one Plasma dipeptidyl peptidase 4 (DPP-4) activity after a single oral administration of gemigliptin at 50 mg and metformin at 1,000 mg given as fixed-dose combination or separate tablets
## 733
FDC gemigliptin and metformin SR in healthy subjects Pharmacokinetic parameters of gemigliptin and metformin after a single oral administration at 50 mg and 1,000 mg, respectively, given as fixed-dose combination or separate tablets incident each of diarrhea, nausea, and blurred vision) occurred after separate tablets taken together. All AEs were mild in intensity and resolved spontaneously without any intervention. Likewise, no clinically significant changes were noted on clinical laboratory tests, during vital-sign monitoring, on twelve-lead electrocardiograms, or physical examinations throughout the entire study period.
# Discussion
This study indicates that an FDC tablet of gemigliptin and metformin SR has comparable PD, PK, and tolerability profiles to separate tablets taken together in healthy males. The comparability between the FDC of gemigliptin and metformin SR and their separate tablets was demonstrated in three ways in the present study. First, the GMRs and their 90% CIs of FDC and separate tablets for AUEC 0-48 h and I max were contained entirely within the conventional bioequivalence range of 0.80-1.25 . Likewise, the mean plasma DPP-4 activity of FDC and the separate tables was superimposable over time after dose [fig_ref] Figure 1: Mean plasma dipeptidyl peptidase 4 [/fig_ref] , and individual comparison of AUEC 0-48 h and I max did not reveal any systematic deviation [fig_ref] Figure 1: Mean plasma dipeptidyl peptidase 4 [/fig_ref]. Second, all of the PK parameters for gemigliptin and metformin were comparable between FDC and the separate tablets, again leading to the GMR and its 90% CI for FDC and separate tablets of each individual drug falling entirely within the conventional bioequivalence range of 0.80-1.25 for both C max and AUC 0-48 h . Lastly, both FDC and separate tablets were well tolerated. Collectively, our results suggest that the FDC tablet of gemigliptin and metformin SR at 50 and 1,000 mg, respectively, can be a beneficial therapeutic option for patients with T2DM who require both gemigliptin and metformin for additional glycemic control. Therefore, the FDC of gemigliptin and metformin SR may act against multiple pathophysiological defects in T2DM such as pancreatic islet dysfunction and insulin resistance while keeping the convenience of a single-time administration, thereby improving the patient's compliance. [bib_ref] Linagliptin/Metformin fixed-dose combination treatment: a dual attack to type 2 diabetes pathophysiology, Koliaki [/bib_ref] [bib_ref] Bioequivalence of saxagliptin/metformin extended-release (XR) fixed-dose combination tablets and single-component saxagliptin and..., Boulton [/bib_ref] Whereas two different kinds of separate tablets should be concomitantly administered (eg, one tablet of gemigliptin 50 mg and two tablets of metformin XR 500 mg), only one tablet is needed in the case of FDC (eg, gemigliptin and metformin SR 25 mg/500 mg). Therefore, FDC can reduce pill burden to T2DM patients, which leads to increased convenience and higher compliance, while not affecting the PD, PK, or tolerability profile of individual drugs, as shown in our results.
# Conclusion
The PD, PK, and tolerability profiles of gemigliptin and metformin XR after a single oral administration in healthy males were comparable between FDC and separate tablets. The FDC of gemigliptin and metformin SR can be a convenient therapeutic option in patients with T2DM who require a combined approach.
In the present study, plasma DPP-4 activity was evaluated as a surrogate marker of PD parameters including fasting levels of glucose, HbA 1c , insulin, and glucagon. Inhibition of DPP-4 activity results in the increase of the concentrations of both active GLP-1 and gastric inhibitory polypeptide (GIP) by stabilizing and preventing the degradation of these enzymes. In normoglycemic healthy subjects, increase in GLP-1 and GIP levels does not have clinically meaningful effects on glucose levels. However, in T2DM patients, elevated GLP-1 and GIP levels may lead to insulin release and lowering of glucose concentrations. [bib_ref] Accelerating drug development using biomarkers: a case study with sitagliptin, a novel..., Krishna [/bib_ref] The comparable degree of reduction in DPP-4 activity observed after administration of FDC of gemigliptin and metformin SR and separate tablets taken together in the present study (Table 1, [fig_ref] Figure 1: Mean plasma dipeptidyl peptidase 4 [/fig_ref] is very much likely to give rise to similar efficacy profiles between the two formulations when administered in T2DM patients.
The present study was conducted on patients in the fed state -that is, after a high-fat meal of 900 kcal with a fat content 50% -because metformin is recommended to be taken with a meal to minimize its common gastrointestinal side effects such as diarrhea, 7,14 whereas gemigliptin can be administered with or without food. The high-fat meal did not affect the PK profiles of gemigliptin and metformin XR when they were administered as FDC. 14 Therefore, the comparable PK profiles of gemigliptin and metformin XR in FDC and separate tablets shown in the present study under the fed condition are likely to be reproduced in the fasting state as well.
Both the FDC and separate tablets of gemigliptin and metformin XR were well tolerated in this study. Diarrhea and nausea, two of the three drug-related AEs seen in the present study, are frequently associated with metformin.Another drug-related AE -blurred vision after separate tablets -was mild in intensity, and it recovered spontaneously.
# Limitations
This study has several limitations. First, the PD and PK characteristics of FDC in healthy, relatively young, male subjects may not represent those in patients with T2DM. However, it is less likely that the PKs of either formulation would be affected more in a different subject population although further studies are warranted in populations with various disease statuses and demographic characteristics. Second, the comparability in the duration and extent of DPP-4 inhibition, PKs, and tolerability of the FDC and separate tablets of gemigliptin and metformin XR after a single administration seen in the present study may not necessarily hold after repeated administration. In this
## Drug design, development and therapy
## Publish your work in this journal
Submit your manuscript here: http://www.dovepress.com/drug-design-development-and-therapy-journal Drug Design, Development and Therapy is an international, peerreviewed open-access journal that spans the spectrum of drug design and development through to clinical applications. Clinical outcomes, patient safety, and programs for the development and effective, safe, and sustained use of medicines are a feature of the journal, which has also been accepted for indexing on PubMed Central. The manuscript management system is completely online and includes a very quick and fair peer-review system, which is all easy to use. Visit http://www.dovepress.com/testimonials.php to read real quotes from published authors.
[fig] Figure 1: Mean plasma dipeptidyl peptidase 4 (DPP-4) activity after a single oral administration of gemigliptin at 50 mg and metformin at 1,000 mg given as fixed-dose combination or separate tablets. (A) Mean percent inhibition-time profiles. (B) Area under the plasma DPP-4 activity-time curve from 0 to 48 hours post-dose (AUEC 0-48 h ); and (C) Maximum inhibition of DPP-4 activity (I max ). Note: The error bars in (A) denote the standard deviations. (B and C) The different symbols and lines represent different subjects assessed in the study. Abbreviations: h, hours; min, minutes; mOD, milli optical density. [/fig]
[fig] Figure 2: Mean plasma concentration-time profiles of (A) gemigliptin and (B) metformin after a single oral administration at 50 mg and 1,000 mg, respectively, given as fixed-dose combination or separate tablets. Notes: The error bars denote the standard deviations. Insets show the linear scale. Abbreviation: h, hours. [/fig]
|
Concurrent development of facial identity and expression discrimination
Facial identity and facial expression processing both appear to follow a protracted developmental trajectory, yet these trajectories have been studied independently and have not been directly compared. Here we investigated whether these processes develop at the same or different rates using matched identity and expression discrimination tasks. The Identity task begins with a target face that is a morph between two identities (Identity A/Identity B). After a brief delay, the target face is replaced by two choice faces: 100% Identity A and 100% Identity B. Children 5-12-years-old were asked to pick the choice face that is most similar to the target identity. The Expression task is matched in format and difficulty to the Identity task, except the targets are morphs between two expressions (Angry/Happy, or Disgust/ Surprise). The same children were asked to pick the choice face with the expression that is most similar to the target expression. There were significant effects of age, with performance improving (becoming more accurate and faster) on both tasks with increasing age. Accuracy and reaction times were not significantly different across tasks and there was no significant Age x Task interaction. Thus, facial identity and facial expression discrimination appear to develop at a similar rate, with comparable improvement on both tasks from age five to twelve. Because our tasks are so closely matched in format and difficulty, they may prove useful for testing face identity and face expression processing in special populations, such as autism or prosopagnosia, where one of these abilities might be impaired.
# Introduction
Face processing is a complex cognitive ability that we rely on to process important information about others, such as identity, emotional expression, direction of attention, sex, and age. Early theoretical models of face processing have suggested that some of these abilities operate independently [bib_ref] Understanding face recognition, Bruce [/bib_ref] [bib_ref] The distributed human neural system for face perception, Haxby [/bib_ref]. For example, one particularly influential model of face processing [bib_ref] Understanding face recognition, Bruce [/bib_ref] proposed a clear separation between what was referred to as 'expression analysis' and components dedicated to identity recognition called 'face recognition units' and 'person identity nodes'. Complementing this model, Haxby and colleagues [bib_ref] The distributed human neural system for face perception, Haxby [/bib_ref] [bib_ref] Human neural systems for face recognition and social communication, Haxby [/bib_ref] proposed a distributed human neural a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 system for face processing that assigned the processing of changeable aspects of faces (such as for facial expression) to the Superior Temporal Sulcus (STS), and the processing of invariant aspects of faces (such as for facial identity) to the Lateral Fusiform Gyrus. These models propose that the separation between the processing of facial identity and expression occurs early, and that these processes remain separate. Electrophysiological and neuropsychological evidence supports these models. For example, repetitive transcranial magnetic stimulation to the right occipital face area disrupts both identity and expression processing whereas stimulation of the right posterior superior temporal sulcus and somatosensory cortex selectively disrupts expression processing and not identity processing [bib_ref] Transcranial magnetic stimulation disrupts the perception and embodiment of facial expressions, Pitcher [/bib_ref] [bib_ref] Facial expression recognition takes longer in the posterior superior temporal sulcus than..., Pitcher [/bib_ref] [bib_ref] Triple dissociation of faces, bodies, and objects in extrastriate cortex, Pitcher [/bib_ref]. In addition, some brain damaged individuals with impaired expression recognition have normal identity recognition [bib_ref] Expression is computed separately from facial identity, and it is computed separately..., Humphreys [/bib_ref] [bib_ref] Prosopo-affective agnosia as a symptom of cerebral organic disease, Kurucz [/bib_ref] [bib_ref] Facial expression processing after amygdalotomy, Young [/bib_ref] , and vice versa [bib_ref] Perceptual and anatomic patterns of selective deficits in facial identity and expression..., Fox [/bib_ref] [bib_ref] Intact recognition of facial expression, gender, and age in patients with impaired..., Tranel [/bib_ref]. Cases of developmental prosopagnosia, characterized by severe face identity recognition deficits, can also have normal facial expression recognition, suggesting that these abilities can develop separately [bib_ref] Normal recognition of emotion in a prosopagnosic, Duchaine [/bib_ref]. There is evidence that these abilities may separate at a young age: a 5-year-old boy with severe prosopagnosia nevertheless demonstrated normal facial expression processing [bib_ref] Severe developmental prosopagnosia in a child with superior intellect, Jones [/bib_ref].
In contrast to the models that suggest independent processing of identity and expression information, other models suggest a complex interaction between the two, positing that rather than operating completely autonomously, there may be a single representation system that processes both identity and expression [bib_ref] Understanding the recognition of facial identity and facial expression, Calder [/bib_ref] [bib_ref] Understanding person perception, Young [/bib_ref]. Supporting this interactive view, there is evidence of an asymmetric relationship between these processes such that identity information interferes with expression processing, but not the other way around (i.e. participants can ignore task-irrelevant expression information and selectively attend to identity information in a speeded identity sorting task, but they cannot ignore task-irrelevant identity information during a speeded expression sorting task) [bib_ref] The relationships between processing facial identity, emotional expression, facial speech, and gaze..., Spangler [/bib_ref] , though see 20 for findings showing that expression can interfere with identity judgments]. Functional Magnetic Resonance Imaging (fMRI) findings also show that the Fusiform Face Area, traditionally associated with the processing of facial identity [bib_ref] The distributed human neural system for face perception, Haxby [/bib_ref] , is also activated when processing facial expression, and responds to irrelevant changes to expression when attention is directed to facial identity [bib_ref] The involvement of the "fusiform face area" in processing facial expression, Ganel [/bib_ref]. Additionally, some groups question the neuropsychological dissociation between identity processing and expression processing in individuals with face recognition deficits [bib_ref] Understanding the recognition of facial identity and facial expression, Calder [/bib_ref]. They argue that although individuals with prosopagnosia often report intact expression recognition, careful testing can reveal deficits in expression processing, albeit often less severe than the identity processing difficulties that define prosopagnosia. Ultimately, the theoretical debate about the independence of identity and expression processing continues.
Normal face processing appears to follow a protracted developmental course. Improvements in facial identity recognition have been reported from early childhood (e.g. as young as 5 years) through adolescence (e.g. 16 years) [bib_ref] Changing abilities in recognition of unfamiliar face photographs through childhood and adolescence...., Lawrence [/bib_ref] [bib_ref] Encoding and storage effects in 7-year-olds' and 10-year-olds' memory for faces, Ellis [/bib_ref] [bib_ref] Age effects in children's memory for unfamiliar faces, Flin [/bib_ref] [bib_ref] From piecemeal to configurational representation of faces, Carey [/bib_ref] [bib_ref] Domain-specific development of face memory but not face perception, Weigelt [/bib_ref] , even when controlling for general cognitive factors, such as IQ [bib_ref] Changing abilities in recognition of unfamiliar face photographs through childhood and adolescence...., Lawrence [/bib_ref]. Online testing of more than 60,000 people showed peak face recognition performance well into adulthood, at age 32 [bib_ref] Where cognitive development and aging meet: Face learning ability peaks after age..., Germine [/bib_ref]. It has been argued that some of these age-related improvements in face processing can be explained by general cognitive development (e.g. improved attentional skills) rather than the development of face specific processes [bib_ref] Early maturity of face recognition: No childhood development of holistic processing, novel..., Crookes [/bib_ref]. However, this view did not take into account the possibility that different aspects of face identity processing may develop at different rates. In fact, it was recently discovered that the ability to discriminate the identity of faces that are presented simultaneously develops at a different rate than the ability to remember the identity of a face [bib_ref] Domain-specific development of face memory but not face perception, Weigelt [/bib_ref]. This suggests that, within identity processing, face perception and face memory are dissociable and follow different developmental trajectories.
With regards to the development of expression processing, studies in children from preschool to adolescence show development of facial expression processing from 3.5 years to 15 years [bib_ref] The development of emotional face processing during childhood, Batty [/bib_ref] [bib_ref] The effect of type of task on children's identification of facial expressions, Markham [/bib_ref] [bib_ref] Preschool children's decoding of facial emotions, Boyatzis [/bib_ref] [bib_ref] Recognition of facial expressions of emotions in school-age children: The intersection of..., Vicari [/bib_ref]. More recent studies indicate that children can recognize happy expressions at very subtle intensities early on (i.e. by 5 years), but that sensitivity to other expressions continues to develop until 10 years, and beyond [bib_ref] Developmental changes in children's sensitivity to facial expressions of varying intensities, Gao [/bib_ref] [bib_ref] Influence of intensity on children's sensitivity to happy, sad, and fearful facial..., Gao [/bib_ref]. Despite the converging evidence suggesting an extended developmental timeline, each of these studies have examined expression recognition in a different age range and it is difficult to compare findings across studies because of methodological differences, such as different stimuli and tasks [bib_ref] Annotation: Development of facial expression recognition from childhood to adolescence: Behavioural and..., Herba [/bib_ref].
While it appears that both identity recognition and expression recognition follow a lengthy developmental trajectory, these trajectories have not been directly compared. Recent MRI studies provide some insight into what such a comparison might reveal. Face selective areas within the fusiform gyrus, which is implicated in face identity recognition, increase in size [bib_ref] Differential development of high level visual cortex correlates with category-specific recognition memory, Golarai [/bib_ref] [bib_ref] Grill-Spector K. Differential development of the ventral visual cortex extends through adolescence, Golarai [/bib_ref] (and density, see [bib_ref] Microstructural proliferation in human cortex is coupled with the development of face..., Gomez [/bib_ref] from childhood to adulthood. In contrast, face selective areas in the Superior Temporal Sulcus, which is implicated in the processing of gaze information and social communication cues, do not appear to change in size during this time [bib_ref] Differential development of high level visual cortex correlates with category-specific recognition memory, Golarai [/bib_ref]. This differential development of these brain areas predicts that in matched tasks of identity and expression processing, an interaction between age and task could emerge. Specifically, slow growth of face-selective fusiform areas may reflect slower behavioral improvement in face identity processing behavior, while an adult-sized STS in childhood may reflect a more rapid development of functions associated with the STS, such as expression processing.
In the present study we aimed to test this prediction by directly comparing the development of identity and expression processing. We created tasks of facial identity and expression discrimination that are matched in format and difficulty to assess the development of these abilities in children 5-12-years-old. Both tasks present an ambiguous target face that is a morph between two faces that vary on the dimension of interest. After a brief delay, the target face is replaced by two choice faces, and the child is asked to pick the choice face that is most similar to the target on the task dimension. If we find an interaction between task and age, such that performance improves more quickly with age for one task than the other, this would support theories that suggest that these abilities operate independently. In contrast, if we find similar improvement with age for the two tasks, this could support theories that posit that identity and expression processes may rely on shared mechanisms. These matched tests of face identity and expression discrimination provide the first direct comparison of the developmental trajectories of these abilities. Sensitive tests of these related but dissociable face processing functions can provide indices of deviations from normal developmental trajectories in children with disorders such as developmental prosopagnosia or autism.
# Method participants
Participants (n = 136, 63 = females, 130 = right handed) between the ages of 5-12-years (mean = 8.3; SD = 2.3) were recruited by email or over the phone through the research participant registry at the Institute of Child Development at the University of Minnesota. We chose this age range because we wanted to examine pre-adolescent school-age children. The data from nine participants were not included in the analysis because they did not complete one or both tasks, resulting in a sample of 127 children [fig_ref] Table 1: Participants by age and gender [/fig_ref]. Upon arrival, the experimenter explained that the purpose of the study was to assess face processing in typically developing children. After the study was explained in detail, parents signed permission forms, and children who were 8-years-old or older signed assent forms to confirm their willingness to volunteer in the study. Children completed a series of tasks for this and other studies. Breaks were given between tasks when requested. Testing took less than 1 hour. Children were compensated for their participation. This study was approved by the Institutional Review Board at the University of Minnesota.
## Tasks
We developed matched tasks of facial identity and expression discrimination [fig_ref] Fig 1: Experimental paradigm [/fig_ref]. The Identity task begins with the presentation of a target face (2s) that is a morph between two identities (Identity A/Identity B). After a delay (400ms), the target face is replaced by two choice faces: 100% Identity A and 100% Identity B. A total of 4 male and 4 female identities were used to create 4 morph continuums (2 male continuums, 2 female continuums, [fig_ref] Fig 1: Experimental paradigm [/fig_ref]. One male and one female morph continuum had happy facial expressions and the others had neutral facial expressions. The child was asked to pick the choice face that is most similar to the target identity by pressing an arrow key (left arrow = face on the left, right arrow = face on the right). The experimenter emphasized that the child should do his or her best to choose the "correct" face and not be concerned about speed. The children were not told that there was no correct answer in the 50% morph trials. The Expression task is matched in format and difficulty to the Identity task, except the targets are morphs between two expressions (Angry/ Happy, or Disgust/Surprise). In each trial the target and choice faces were of the same identity. Two male and two female identities were used (i.e. one male and one female Angry/Happy, one male and one female Disgust/Surprise). The child was asked to indicate by key press which choice face has the expression that is most similar to the target expression, again emphasizing that the child should try to choose the "correct" face, and not be concerned about speed.
Before beginning, children were given eight practice trials for each task. The experiment itself was broken down into eight randomly administered blocks of trials: 4 blocks Identity, 4
## Stimuli
Stimuli were adult male and female faces from the Karolinska Directed Emotional Faces database. Identities were paired for morphing based on subjective similarity of basic characteristics such as hair color, face shape, etc. to help ensure that morphs appeared natural. We chose angry, happy, disgusted, surprised, and neutral emotional expressions from the set of basic universal emotions [bib_ref] Constants across cultures in the face and emotion, Ekman [/bib_ref]. Faces were 8.7˚x 8.7˚when viewed at 60cm. They appeared against a solid gray background and were revealed through 7.6˚x 8.7˚oval windows that covered ears and most of the hair. Morph faces were presented in the center of the screen. Choice faces were presented side by side, separated by a distance of 2.45˚.
# Analysis
We first removed trials with response time that exceeded 2.5 SD from the within-subject average response time. Thus, for each subject we removed trials with response times that were too fast, possibly indicating anticipatory responding, or too slow, possibly due to inattention. [fig_ref] Table 2: Percentage of trials removed for each task by age [/fig_ref] shows the percentage of trials removed for each Age and Task. To ensure that an equal number of trials were removed across ages and tasks, we ran a logit model on the status of the trial (rejected/not rejected), entering Age and Task (Expression and Identity), with their interaction, and testing their significance using a Type 3 Analysis of Deviance (as implemented in the R package car. None of the effects were significant, indicating that the same number of trials were removed across Ages and Tasks: Age X 2 (1) = 0.37, p = .54; Task X 2 (1) = 0.002, p = .97; and Age x Task X 2 (1) = 0.004, p = .95. Accuracy. We characterized performance in the two tasks by computing a measure of accuracy. For this analysis, we first removed all trials at 50% morphing (because there was no correct answer for these trials). Next, for each face pair continuum we classified a trial as correct if the subject responded "A" when the presence of "face A" was more than 50%, or pressed "B" when the presence of "face A" was less than 50%. We fitted a single logit mixed model across subjects entering Age and Task (Expression and Identity) with their interaction as fixed effects, and subjects and type of continuum as random effects (random intercepts). Statistical significance of the individual terms was tested using a Wald's test (type 3) as implemented in the package car in R.
Reaction time. Reaction time data were calculated using only correct trials as defined above. We fitted a single linear mixed effect model across all the subjects using Age and Task (Expression and Identity) as fixed effects with their interaction, and subjects and type of continuum as random effects (random intercepts). Statistical significance of the individual terms was tested using a Wald's test (type 3) as implemented by the package car in R.
Psychometric curves. To characterize discrimination performance for each age group, we estimated a psychometric curve at the group level by fitting logit mixed effect models on the subjects' responses (face A or B) for each of the eight morph continua and each of the eight age groups. We entered subjects as a random effect (with random intercepts) and the percentage of morphing (0-100% in 10% steps, rescaled to -1 and 1 to allow convergence) as a fixed effect. We analyzed the steepness of the psychometric function by using the parameter estimate (slope) of the percentage morphing term, since this estimate is proportional to the steepness of the logistic function at the point of maximal inflection (i.e., where the first derivative is maximal). The steepness of the psychometric function is a proxy of how categorical the judgment is and larger slope indicates better performance (i.e. perfect observers would show a step function, always selecting face A if the morph contained more than 50% face A, and always selecting face B if the morph contained less than 50% face A).
We obtained 64 such slopes (8 age groups x 8 face pairs), divided the face pairs into expression and identity tasks, and ran a two-way ANOVA (Age x Task) on the slopes to estimate statistical significance. Each cell contained four samples, one for each face pair continuum. We tested significance for the main and interaction effects. We also ran a two-way ANOVA (Age x Expression type) to compare slopes for the two types of expression morphs (Angry/Happy vs. Disgusted/Surprised). Because of the small number of data points per cell, we consider this analysis exploratory (S1 All analyses were performed in R (version 3.2.2) using the Our analyses reveal that the slopes of the curves increase linearly with age (indicating better performance), and that overall, participants were able to discriminate expressions slightly better than identities. We quantified the effect of Age by averaging the slopes over the two tasks and computing the ratio between the slope at age 12 and the slope at age 5. We discovered that 12-year-olds showed slopes that were 2.63 steeper than 5-year-olds [fig_ref] Fig 7: percentage of morphing from one extreme to the other [/fig_ref]. We then quantified the effect of Task by computing the ratio between the average slope for expression and the slope for identity across all ages. This showed that the effect size of Task was only slight: slopes in the expression task were only 1.15 times steeper than in the identity task.
# Results
## Accuracy
# Discussion
Whether face identity and expression are processed separately or by shared mechanisms is an ongoing topic of debate . The present study was designed to directly compare the development of these two face processing abilities, in tasks that were matched in format and difficulty. Children between the ages of 5-12 years performed categorization tasks that targeted their ability to discriminate facial identity and facial expression. For both accuracy and reaction time, there was a main effect of age, and the improvement in performance on both tasks appeared steady from age five to twelve. There was no main effect of task and there was no interaction between age and task, indicating that both tasks showed approximately the same rate of improvement with age. These results indicate that these tasks are relatively well matched in terms of their sensitivity to developmental differences, suggesting that Group-level psychometric curves for each age group and morph continuum, divided according to the task: Expression (left column) and Identity (right column). The x-axis maps the they may be useful for investigating individual differences in identity and expression processing (e.g. large dissociations in performance in individuals may indicate clinically significant impairments, such as prosopagnosia or autism). Importantly for our primary research question, these results appear to support the idea that face identity and expression processing share common mechanisms. However, they do not rule out the possibility that identity and expression processing are separable: it is still possible that two independent systems could develop at the same rate. While an interaction between age and task would have provided compelling support for separable identity and expression processing systems, the lack of interaction in this study could be related to certain characteristics of our experimental design. First, given the finding that discrimination of face identity and memory for face identity develop at different rates [bib_ref] Domain-specific development of face memory but not face perception, Weigelt [/bib_ref] , it is possible that tasks with a more demanding memory load could reveal different developmental trajectories for identity and expression processing. Our tasks are well-matched discrimination tasks, and the memory load is minimal: participants need only remember a single face, and the delay between the target and choice faces is brief (400 ms). Our study represents a first step towards directly comparing the development of identity and expression processing, but future experiments could manipulate memory load to further investigate the relative development of these two abilities. Second, these results could be a function of the age range tested. We tested pre-adolescent children 5-12-years-old, but others have found two separate stages in the development of facial expression recognition: one from 5-12 years, and another from 13 years through adulthood [bib_ref] Mapping the development of facial expression recognition, Rodger [/bib_ref]. Testing children at intermediate ages (e.g. 9-16 years) with our matched tests may uncover differences in development that were not revealed here. Third, it is possible that rather than tapping identity and expression processing, our tasks in fact measure basic perceptual mechanisms that are not specific to either ability. We think this is unlikely since both tasks have a delay between the target face and the choice faces, preventing participants from engaging direct comparison or feature matching.
One related question about the present findings is whether the age-related improvements in identity and expression processing that we documented are face-specific (i.e. reflect the development of face processing mechanisms), or due to general cognitive development (i.e. reflect improved attention and other test taking skills). Unfortunately our study does not allow us to differentiate between the two alternatives.
This study does have important strengths: we tested a large number of children (over 100) from 5-12-years-of-age and our tasks are well matched in terms of format and difficulty, positioning them to reveal individual differences in identity and expression processing. This study represents a first step towards investigating the relative development of identity and expression processing, but these results do not preclude the possibility that the development of these abilities is indeed separable. More work needs to be done to address this issue. In the mean time, applying the present tasks to work with special populations (e.g. prosopagnosia, autism), may be useful for assessing whether, and at what developmental stage, one ability is disproportionately affected relative to the other in some individuals.
## Supporting information
[fig] Fig 1: Experimental paradigm. a) Format of identity and expression discrimination tasks. b) Example morph continuums for identity and expression tasks. https://doi.org/10.1371/journal.pone.0179458.g001 [/fig]
[fig] Fig 2: shows the mean accuracies for each Age and Task (see S2 Fig for data plotted by individual). We found that the overall accuracy increased with age, Χ 2 (1) = 57.83,p < .001, and that the two tasks were matched in difficulty, as revealed by the non-significant main effect of Task, Χ 2 (1) = 0.44, p = 0.51 and non-significant interaction between Age and Task, Χ 2 (1) = 0.12, p = 0.73. Participant performance for one task was highly correlated with performance in the other task when controlling for Age, r p = 0.51 partial correlation, n = 127, p < .001,Fig 3. See S3 Fig for correlations for each age. [/fig]
[fig] Fig 3: Correlation between identity and expression: Accuracy. Scatterplot illustrating correlation between participant accuracy on identity and expression tasks, in %. Data include all trials except those with 50% morphs, which do not have a correct response. Grey shading indicates 95% confidence interval.https://doi.org/10.1371/journal.pone.0179458.g003 Development of facial identity and expression discrimination PLOS ONE | https://doi.org/10.1371/journal.pone.0179458 June 15, 2017 We found a significant main effect of Age, F(1, 60) = 137.10, p < .001, η p 2 = 0.70, and a small, but significant main effect of Task, F(1, 60) = 7.69, p < .05, η p 2 = 0.11, but no Age x Task interaction, F(1, 60) = 002, p = 0.89, η p 2 ffi 0. [/fig]
[fig] Fig 5: Correlation between identity and expression: Reaction time. Scatterplot illustrating correlation between participant reaction times on identity and expression tasks (in seconds). Data include only trials that were answered correctly (trials with 50% morphs are not included because they not have a correct response). Grey shading indicates 95% confidence interval.https://doi.org/10.1371/journal.pone.0179458.g005Development of facial identity and expression discrimination [/fig]
[fig] Fig 6: Psychometric curves for identity and expression tasks. [/fig]
[fig] Fig 7: percentage of morphing from one extreme to the other (e.g. 30% MAH = Male Angry face with 30% Male Happy face); the y-axis maps the proportion of responses to the second extreme (for MAH, proportion of responses "Male Happy Face"). MAH = Male Angry to Happy; FAH = Female Angry to Happy; MDS = Male Disgust to Surprise; FDS = Female Disgust to Surprise; Slope estimates. Boxplots of the slope estimates for each age group and task. Individual points represent outliers (defined as those points exceeding 1.5 times the inter-quartile range).https://doi.org/10.1371/journal.pone.0179458.g007Development of facial identity and expression discrimination [/fig]
[fig] S1: Fig. Comparison of slopes for the two types of expression morph continuums across age. We compared the mean slope for Angry/Happy morphs to the mean slope for Disgusted/ Surprised morphs using a 2-way ANOVA (Age x Expression type). We found a significant main effect of Age, F(1,28) = 90.4, p<0.001, and a significant main effect of Expression Type, F(1,28) = 16.5, p<0.001, but no Age x Expression interaction F(1,28) = 0.7, p = 0.40. (TIF) S2 Fig. Individual accuracy data. Individual data representing accuracy for the Identity and Expression tasks, plotted by participant age. (TIF) S3 Fig. Correlations: Accuracy. Correlation between accuracies on Identity task and Expression task plotted by participant age. Error bars represent 95% confidence intervals. (TIF) S4 Fig. Correlations: Reaction time. Correlation between reactions times on Identity task and Expression task plotted by participant age. Error bars represent 95% confidence intervals. (TIF) File. Data. Raw data in.csv format. (CSV) S2 File. Data description. Text file describing the data set. (TXT) [/fig]
[table] Table 1: Participants by age and gender. *Participants were excluded for not completing one or both tasks https://doi.org/10.1371/journal.pone.0179458.t001 [/table]
[table] Table 2: Percentage of trials removed for each task by age. [/table]
|
Cellular Therapeutics for Heart Failure: Focus on Mesenchymal Stem Cells
Resulting from a various etiologies, the most notable remains ischemia; heart failure (HF) manifests as the common end pathway of many cardiovascular processes and remains among the top causes for hospitalization and a major cause of morbidity and mortality worldwide. Current pharmacologic treatment for HF utilizes pharmacologic agents to control symptoms and slow further deterioration; however, on a cellular level, in a patient with progressive disease, fibrosis and cardiac remodeling can continue leading to end-stage heart failure. Cellular therapeutics have risen as the new hope for an improvement in the treatment of HF. Mesenchymal stem cells (MSCs) have gained popularity given their propensity of promoting endogenous cellular repair of a myriad of disease processes via paracrine signaling through expression of various cytokines, chemokines, and adhesion molecules resulting in activation of signal transduction pathways. While the exact mechanism remains to be completely elucidated, this remains the primary mechanism identified to date. Recently, MSCs have been incorporated as the central focus in clinical trials investigating the role how MSCs can play in the treatment of HF. In this review, we focus on the characteristics of MSCs that give them a distinct edge as cellular therapeutics and present results of clinical trials investigating MSCs in the setting of ischemic HF.
# Introduction
Heart failure (HF) has become a major epidemic throughout the world. Resulting as the common end pathway for a myriad of cardiovascular disease processes, HF is the most common cause of hospital admission in patients over 65 years old, with the number of individuals having HF reaching 8 million and expected costs in the United States exceeding 40-70 billion dollars [bib_ref] Forecasting the impact of heart failure in the United States: a policy..., Heidenreich [/bib_ref]. The foundation of current therapy for HF is pharmaceutical interventions. Certain subsets of patients with HF may benefit from advanced therapies including cardiac resynchronization therapy (CRT), mechanical circulatory support devices, and even transplant, which is reserved to the sickest patients. However, these measures are not without pitfalls; pharmaceutical therapies have side effects, and CRT, while advantageous, is only available to some patients [bib_ref] Refining success of cardiac resynchronization therapy using a simple score predicting the..., Maass [/bib_ref]. Recently, there has been a push to investigate more innovative treatments for HF that aim at not only improving clinical symptoms but also improving cardiovascular pathophysiology. While pharmaceutical and device therapy can improve the pathophysiology of ischemic HF, nonischemic HF still has limited options currently. One of the leading treatments under investigation for HF is the use of mesenchymal stem cells (MSCs). Mesenchymal stem cells are multipotent adult stem cells that have been at the forefront of regenerative medicine research. Mesenchymal stem cells are unique cells that can be cultured ex vivo and utilized as cellular therapies in a variety of disease states. Currently, MSCs are being entertained as treatment modalities in cardiovascular disease states such as acute myocardial infarction, fibrosis, and heart failure. There are other cell therapies that have been explored in translational projects including those of induced pluripotent stem cells (iPSCs) and vector-based gene therapy. Here, we focus on MSCs and their desirable properties as cellular therapeutics in heart failure and implicate their potential use in clinical practice.
## Mesenchymal stem cells in cell-based therapies
Mesenchymal stem cells are a type of adult stem cells that are multipotent cells [bib_ref] MicroRNA profiling reveals age-dependent differential expression of nuclear factor κB and mitogen-activated..., Pandey [/bib_ref] [bib_ref] Repair of tissues by adult stem/progenitor cells (MSCs): controversies, myths, and changing..., Prockop [/bib_ref]. Mesenchymal stem cells maintain the ability to give rise to a number of different end-cell lineages including bone cells, adipose cells, stromal cells, muscle cells, tendon cells, and other mesenchymal cells [bib_ref] MicroRNA profiling reveals age-dependent differential expression of nuclear factor κB and mitogen-activated..., Pandey [/bib_ref] [bib_ref] Repair of tissues by adult stem/progenitor cells (MSCs): controversies, myths, and changing..., Prockop [/bib_ref] [bib_ref] Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal..., Ortiz [/bib_ref]. Mesenchymal stem cells are utilized for endogenous cell-tocell communication and paracrine signaling and also employ these properties for cellular repair when utilized in cellular therapeutics [bib_ref] Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal..., Ortiz [/bib_ref]. Although not conclusively proven, mesenchymal stem cells are postulated to achieve these processes via expression of a wide spectrum of secreted factors and to a lesser extent direct end-cell differentiation for replacement of damaged cells [bib_ref] Repair of tissues by adult stem/progenitor cells (MSCs): controversies, myths, and changing..., Prockop [/bib_ref] [bib_ref] Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal..., Ortiz [/bib_ref]. Factors that are expressed by MSCs include cytokines, chemokines, and adhesion molecules, which then regulate the activation and/or inhibition of molecular signaling pathways for endogenous cellular repair [bib_ref] MicroRNA profiling reveals age-dependent differential expression of nuclear factor κB and mitogen-activated..., Pandey [/bib_ref]. Additionally, MSCs are immunoprivileged cells given the lack of expression of major histocompatibility complex II (MHC II) complexes in their multipotent state [bib_ref] Immunosuppressive properties of mesenchymal stem cells: advances and applications, De Miguel [/bib_ref]. Furthermore, MSCs have been shown to decrease inflammation and inflammatory cues as well as to promote angiogenesis [bib_ref] MicroRNA profiling reveals age-dependent differential expression of nuclear factor κB and mitogen-activated..., Pandey [/bib_ref] [bib_ref] Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal..., Ortiz [/bib_ref] [bib_ref] Cell-based therapies for myocardial repair: emerging role for bone marrow-derived mesenchymal stem..., Silva [/bib_ref]. Mesenchymal stem cells represent an ideal candidate in the emerging field of regenerative medicine [bib_ref] Current understanding and challenges in bioprocessing of stem cell-based therapies for regenerative..., Ratcliffe [/bib_ref]. These properties combined with the accessibility of MSCs from the bone marrow or adipose tissues make MSCs ideal candidates for cell-and gene-based therapies. Since their initiation, isolation, and description by Friedenstein and colleagues from bone marrow, MSCs have been considered potential candidates for cell-based therapies [bib_ref] Fibroblast precursors in normal and irradiated mouse hematopoietic organs, Friedenstein [/bib_ref]. Mesenchymal stem cells have been isolated from a wide variety of tissues including bone marrow tissue, adipose tissue, cardiac tissue, umbilical cord tissue, as well as those of other sites [bib_ref] Multipotent mesenchymal stem cells from adult human synovial membrane, Bari [/bib_ref] [bib_ref] Mesenchymal progenitor cells in human umbilical cord blood, Erices [/bib_ref] [bib_ref] Human heart, spleen, and perirenal fat-derived mesenchymal stem cells have immunomodulatory capacities, Hoogduijn [/bib_ref] [bib_ref] Amniotic fluid as a novel source of mesenchymal stem cells for therapeutic..., Anker [/bib_ref] [bib_ref] Culture expanded canine mesenchymal stem cells possess osteochondrogenic potential in vivo and..., Kadiyala [/bib_ref] [bib_ref] Human adipose tissue is a source of multipotent stem cells, Zuk [/bib_ref]. The therapeutic potential of MSCs makes them the ideal candidates for cell-and gene-based therapies for a number of reasons including their applicability for "off the shelf use" potential in cellular therapeutics; their immunoprivileged state; their ability to express cytokines, chemokines, and adhesion molecules; and their ability to be expanded to sufficient therapeutic quantities ex vivo [bib_ref] MicroRNA profiling reveals age-dependent differential expression of nuclear factor κB and mitogen-activated..., Pandey [/bib_ref] [bib_ref] Cell-based therapies for myocardial repair: emerging role for bone marrow-derived mesenchymal stem..., Silva [/bib_ref] [bib_ref] Concise review: mesenchymal stem/multipotent stromal cells: the state of transdifferentiation and modes..., Phinney [/bib_ref]. Previous studies have shown that MSCs maintain their therapeutic potential even after cryopreservation, further adding to the list of desirable characteristics of MSCs [bib_ref] Mesenchymal stem cells as a biological drug for heart disease: where are..., Sanina [/bib_ref] [bib_ref] Suppression of allogeneic T-cell proliferation by human marrow stromal cells: implications in..., Tse [/bib_ref]. Additionally, given their immunoprivileged state, allogeneic treatment modalities could be a legitimate possibility as well for the desired therapy in heart failure patients.
## Mscs and paracrine signaling
While MSCs can differentiate along mesenchymal lineages to produce end-cell types needed for repopulation of damaged tissues, the primary mechanism postulated by which MSCs are able to direct and facilitate endogenous cellular repair is via paracrine signaling [bib_ref] Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal..., Ortiz [/bib_ref] [bib_ref] Concise review: mesenchymal stem/multipotent stromal cells: the state of transdifferentiation and modes..., Phinney [/bib_ref]. It is postulated that utilizing paracrine signaling, MSCs direct and modulate the cellular microenvironment by promoting survival and proliferation : Multipotent capacity of mesenchymal stem cells. MSCs are derived from numerous tissue sources including bone marrow and adipose tissue. They are able to differentiate into various end-cell types including osteoblasts, adipocytes, chondrocytes, and myoblasts. Additionally, they are immunoprivileged, therefore allowing autologous as well as allogeneic therapeutic potential. They can also be cryopreserved, while maintaining their multipotent properties, thus allowing them to be ideal candidates for "off the shelf" cell-based therapies. of endogenous cells, induce angiogenesis, quell inflammation, inhibit apoptosis, and recruit endogenous progenitor cells to endpoint differentiation, with an end result of functional improvement [bib_ref] Angiogenic effects of human multipotent stromal cell conditioned medium activate the PI3K-Akt..., Hung [/bib_ref] [bib_ref] Introduction to stem cell therapy, Biehl [/bib_ref] [bib_ref] Stemness" does not explain the repair of many tissues by mesenchymal stem/multipotent..., Prockop [/bib_ref] [bib_ref] Mesenchymal stem cells are superior to angiogenic growth factor genes for improving..., Shyu [/bib_ref]. Many studies have attempted to characterize the mechanisms by which MSCs produce a more favorable microenvironment conducive to endogenous cellular repair. It appears among the key components involved in modulating cellular signaling cascades including anti-inflammatory cues, proangiogenic signaling, and avoidance by immune surveillance [bib_ref] MicroRNA profiling reveals age-dependent differential expression of nuclear factor κB and mitogen-activated..., Pandey [/bib_ref] [bib_ref] Repair of tissues by adult stem/progenitor cells (MSCs): controversies, myths, and changing..., Prockop [/bib_ref] [bib_ref] Angiogenic effects of human multipotent stromal cell conditioned medium activate the PI3K-Akt..., Hung [/bib_ref] [bib_ref] Mesenchymal stem cells are superior to angiogenic growth factor genes for improving..., Shyu [/bib_ref]. Many of the cytokines expressed by MSCs work by quelling the inflammatory signaling present in disease states including HF [bib_ref] Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal..., Ortiz [/bib_ref] [bib_ref] The role of IL-1 in the pathogenesis of heart disease, Bujak [/bib_ref] [bib_ref] Enhanced interleukin-1 activity contributes to exercise intolerance in patients with systolic heart..., Van Tassell [/bib_ref]. Furthermore, recent studies have shown that IL-1β may play a critical role in instigating the onset of HF [bib_ref] Enhanced interleukin-1 activity contributes to exercise intolerance in patients with systolic heart..., Van Tassell [/bib_ref]. Mesenchymal stem cells are able to secrete potent levels of interleukin 1 receptor antagonist (IL-1rn), which has been thought to be a key regulator of MSC-based therapy [bib_ref] Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal..., Ortiz [/bib_ref].
With MSCs available from a variety of tissues, one question that frequently arises with the proposal of MSCs as cellbased therapies is "Are MSCs taken from the varying tissues equal in directing paracrine-mediated endogenous cellular repair?" Many studies have investigated the differences in MSCs isolated from different tissue sources, and the potential of these cells remains the same despite the location of isolation [bib_ref] MicroRNA profiling reveals age-dependent differential expression of nuclear factor κB and mitogen-activated..., Pandey [/bib_ref] [bib_ref] The advancing field of cell-based therapy: insights and lessons from clinical trials, Telukuntla [/bib_ref]. While MSCs possess multiple attributes desirable for the ideal cell-based therapy for HF, pitfalls do currently exist as well. Limitations as to method, timing, and dose of cells remain unknown for MSC administration in the setting of HF.
Despite the fact that MSCs can be scaled to large quantities ex vivo, limitations still exist as to delivery of an adequate number of cells given the progression and individual state of the disease for a given patient [bib_ref] Cell-based therapies for myocardial repair: emerging role for bone marrow-derived mesenchymal stem..., Silva [/bib_ref]. This problem seems to have been overcome with the use of allogeneic MSCs from healthy donors. Other studies have reported concerns regarding retention of injected cells over time in the heart given issues of low engraftment and limited retention of MSCs or other cell-based therapies [bib_ref] Intra-coronary arterial injection of mesenchymal stromal cells and microinfarction in dogs, Vulliet [/bib_ref]. However, given that the primary function of MSCs seems to be promotion of endogenous repair via paracrine signaling rather than direct end-cell lineage differentiation, absolute cell numbers can be deceptive. Additionally, local engraftment of cells is not entirely necessary given that paracrine signaling largely contributes to the overall function of MSCs in directing endogenous cellular repair [bib_ref] Intravenous hMSCs improve myocardial infarction in mice because cells embolized in lung..., Lee [/bib_ref] [bib_ref] Paracrine action mediate the antifibrotic effect of transplanted mesenchymal stem cells in..., Li [/bib_ref]. Thus, while several mechanisms have been investigated to identify the ideal route of delivery of MSCs, further work is necessary regarding optimization of delivery to the area of injury given poor understanding of how MSCs would be best utilized [bib_ref] A quantitative, randomized study evaluating three methods of mesenchymal stem cell delivery..., Freyman [/bib_ref] [bib_ref] Impact of timing and dose of mesenchymal stromal cell therapy in a..., Richardson [/bib_ref] [bib_ref] Bone marrow mesenchymal stem cells stimulate cardiac stem cell proliferation and differentiation, Hatzistergos [/bib_ref]. Nonetheless, advancements have been made with recent clinical trials demonstrating safety of allogeneic MSCs [bib_ref] Comparison of allogeneic vs autologous bone marrow-derived mesenchymal stem cells delivered by..., Hare [/bib_ref]. Additionally, many of these studies were conducted in the context of acute infarction, with significant work still needing to be initiated in the setting of HF.
## Heart failure
Heart failure is a major cause of morbidity and mortality worldwide, with greater than 5.6 million individuals afflicted in the US alone [bib_ref] Heart disease and stroke statistics-2012 update: a report from the American Heart..., Roger [/bib_ref]. Heart failure is among the most common diagnoses for hospital admission with estimates of approximately one percent of the western world afflicted and constitutes approximately 400,000 new admissions annually [bib_ref] Epidemiology of heart failure, Miller [/bib_ref]. It is now appreciated that the underlying cellular processes in HF are an interplay of myocardial factors, systemic factors, and local inflammation [bib_ref] The "modern" view of heart failure: how did we get here?, Katz [/bib_ref]. Collectively, the disarray of molecular pathways, downstream signaling, and subsequent gene expression culminates in the debilitating clinical disease state of HF. Recent studies have demonstrated that inflammatory cues play a critical role in instigating the onset of HF at the molecular level with cytokines such as interleukin 1 beta (IL-1β) and nuclear factor kappa B (NF-κB) contributing critically to left ventricular (LV) deterioration [bib_ref] The role of IL-1 in the pathogenesis of heart disease, Bujak [/bib_ref] [bib_ref] Enhanced interleukin-1 activity contributes to exercise intolerance in patients with systolic heart..., Van Tassell [/bib_ref] [bib_ref] Interleukin-1β modulation using a genetically engineered antibody prevents adverse cardiac remodelling following..., Abbate [/bib_ref] [bib_ref] Interleukin-1 receptor type I signaling critically regulates infarct healing and cardiac remodeling, Bujak [/bib_ref] [bib_ref] Requirement of nuclear factor-κB in angiotensin II-and isoproterenol-induced cardiac hypertrophy in vivo, Freund [/bib_ref] [bib_ref] Regulation of cardiac hypertrophy by intracellular signalling pathways, Heineke [/bib_ref].
## Current treatment of heart failure
Current treatment strategies in HF focus on minimizing disease morbidity, reducing hospitalizations, and prevention of mortality [bib_ref] Heart disease and stroke statistics-2012 update: a report from the American Heart..., Roger [/bib_ref]. Despite the significant economic burden as well as morbidity and mortality stemming from HF, no promising treatment modalities to reverse the disease process currently exist. In particular, end-stage HF results in a common final pathway initiated by several signaling mechanisms that are ultimately characterized by myocardial dysfunction and cardiac remodeling. Current treatment options for HF include drug therapy, cardiac resynchronization therapy, mechanical circulatory support, and/or cardiac transplantation [bib_ref] Advanced heart failure in critical patients (INTERMACS 1 and 2 levels): ventricular..., Attisani [/bib_ref] [bib_ref] Heart failure with a normal left ventricular ejection fraction: epidemiology, pathophysiology, diagnosis..., Garg [/bib_ref] [bib_ref] Postcardiac transplant survival in the current era in patients receiving continuous-flow left..., Kamdar [/bib_ref] [bib_ref] Cardiac resynchronization therapy with or without defibrillator: experience from a high-volume Belgian..., Verbrugge [/bib_ref]. While cardiac transplantation does improve mortality and quality of life, it remains a limited therapy given the epidemiologic restriction of donor hearts available. Furthermore, an ever-increasing number of HF patients have no remaining therapies available given these restrictions [bib_ref] Under-utilization of evidence-based drug treatment in patients with heart failure is only..., Lenzen [/bib_ref]. Potential new therapies for HF will likely require targeted molecular therapy thus integrating local and systemic inflammation, promoting neoangiogenesis, and developing a methodology by which LV dysfunction can ideally be restored. Cellular therapeutics could allow a greater number of patients afflicted with HF to benefit from therapy than is possible via current advanced heart failure therapies, especially, given consideration that the only true treatment currently available for HF patients remains cardiac transplantation, which is prohibitive given the associated costs and limited donor hearts [bib_ref] Cell-based therapies for myocardial repair: emerging role for bone marrow-derived mesenchymal stem..., Silva [/bib_ref]. Indeed, cell-based therapies and regenerative medicine-directed therapies for HF would significantly change the course of disease progression and patient outcomes. Mesenchymal stem cells represent an ideal candidate in the emerging field of regenerative medicine that could be at the forefront of cell-based therapies for heart failure.
## Clinical trials and msc-based therapies for heart failure
Recently, several clinical trials are ongoing in order to determine the safety and efficacy of MSC-based treatment for acute myocardial infarction (MI), with far fewer trials investigating the use of MSC-based therapies in the setting of HF [fig_ref] Table 1: Clinical trials investigating mesenchymal stem cells in heart failure [/fig_ref] [bib_ref] Translational development of mesenchymal stem cell therapy for cardiovascular diseases, Hare [/bib_ref] [bib_ref] Rationale and design of the transendocardial injection of autologous human cells (bone..., Trachtenberg [/bib_ref]. Most of the trials utilizing MSCs as a therapeutic option are in ischemic heart failure, and there are little data to date on the treatment in nonischemic cardiomyopathies. Indeed, MSCs have been used in clinical trials to treat both ischemic and nonischemic heart failure with both approaches showing promising results [bib_ref] Randomized comparison of allogeneic versus autologous mesenchymal stem cells for nonischemic dilated..., Hare [/bib_ref] [bib_ref] Ixmyelocel-T for patients with ischaemic heart failure: a prospective randomised double-blind trial, Patel [/bib_ref]. While these diseases differ in terms of presence or absence of coronary artery disease, the MSCs are directed at generating new myocardium. One could speculate that MSCs would potentially be more effective in nonischemic disease because the damaged myocardium still has adequate blood supply. In the ischemic HF trials, studies have already started to show improvements in regional and global systolic and diastolic function, reversal of LV remodeling, and enhanced myocardial collateralization and coronary perfusion using the regenerative potential of MSCs [bib_ref] Cardiac repair with intramyocardial injection of allogeneic mesenchymal stem cells after myocardial..., Amado [/bib_ref] [bib_ref] A randomized, double-blind, placebo-controlled, dose-escalation study of intravenous adult human mesenchymal stem..., Hare [/bib_ref] [bib_ref] Allogeneic mesenchymal stem cells restore cardiac function in chronic ischemic cardiomyopathy via..., Quevedo [/bib_ref] [bib_ref] Early improvement in cardiac tissue perfusion due to mesenchymal stem cells, Schuleri [/bib_ref] [bib_ref] Mesenchymal stem cells differentiate into an endothelial phenotype, enhance vascular density, and..., Silva [/bib_ref]. Of note, the benefits of MSC therapy appear to be seen in the relative short term, and there remains a question as to the long-term benefits of MSC therapy as in other types of cell-based therapy. This suggests that the paracrine-mediated effects of cell-based therapy may be directly related to cell survival. Mechanistically, much remains to be elucidated as to the exact means by which MSCs achieve reversal of LV remodeling. It has been suggested that limiting inflammation coupled with deposition of extracellular matrix components deposited by MSCs may help limit the total scar size thereby decreasing LV dimensions and possibly improving diastolic function [bib_ref] Intravenous hMSCs improve myocardial infarction in mice because cells embolized in lung..., Lee [/bib_ref] [bib_ref] Paracrine action mediate the antifibrotic effect of transplanted mesenchymal stem cells in..., Li [/bib_ref] [bib_ref] Mesenchymal cell transplantation and myocardial remodeling after myocardial infarction, Dixon [/bib_ref] [bib_ref] Immuno-inflammatory regulation effect of mesenchymal stem cell transplantation in a rat model..., Du [/bib_ref] [bib_ref] Allogeneic administration of fetal membrane-derived mesenchymal stem cells attenuates acute myocarditis in..., Ishikane [/bib_ref] [bib_ref] Transplantation of bone marrow-derived stem cells improves myocardial diastolic function: strain rate..., Schneider [/bib_ref]. Despite these advances, further investigation into the mechanisms by which MSCs are able to facilitate these actions is warranted. Some studies suggest critical roles played by cytokines, expressed by MSCs or progenitor cells recruited by MSCs, that may be integral to cardiac recovery including insulin like growth factor 1 (IGF-1), hepatocyte growth factor (HGF), and vascular endothelial growth factor (VEGF) [bib_ref] Mesenchymal stem cells as trophic mediators, Caplan [/bib_ref] [bib_ref] Myocardial protection from ischemia/ reperfusion injury by endogenous and exogenous HGF, Nakamura [/bib_ref] [bib_ref] Angiogenesis therapies for cardiovascular disease, Pedrotty [/bib_ref] [bib_ref] The cardioprotective effect of mesenchymal stem cells is mediated by IGF-I and..., Sadat [/bib_ref] [bib_ref] Cardiac stem cells possess growth factor-receptor systems that after activation regenerate the..., Urbanek [/bib_ref].
With the ever-increasing popularity of MSCs as potential cellular-based therapeutics for HF, a number of clinical trials have recently been completed and several are underway investigating the perspective roles how MSCs could play in the setting of HF events [fig_ref] Table 1: Clinical trials investigating mesenchymal stem cells in heart failure [/fig_ref]. The majority of these trials have looked at MSCs, primarily for the reasons outlined previously [bib_ref] MicroRNA profiling reveals age-dependent differential expression of nuclear factor κB and mitogen-activated..., Pandey [/bib_ref] [bib_ref] Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal..., Ortiz [/bib_ref] [bib_ref] Immunosuppressive properties of mesenchymal stem cells: advances and applications, De Miguel [/bib_ref] [bib_ref] Enhanced interleukin-1 activity contributes to exercise intolerance in patients with systolic heart..., Van Tassell [/bib_ref] [bib_ref] Paracrine action mediate the antifibrotic effect of transplanted mesenchymal stem cells in..., Li [/bib_ref] [bib_ref] Impact of timing and dose of mesenchymal stromal cell therapy in a..., Richardson [/bib_ref]. Several trials have attempted to investigate the use of MSCs in the setting of acute and chronic HF, with a variety of strategies on cell delivery. Initial trials focused on events closer to the time of the infarct, with cell administrations occurring in the setting of acute myocardial infarction, with MSCs being investigated in an attempt to prevent ischemic cardiomyopathy [bib_ref] A randomized, double-blind, placebo-controlled, dose-escalation study of intravenous adult human mesenchymal stem..., Hare [/bib_ref]. Still, other studies have looked at the therapeutic potential of MSCs in the setting of nonischemic cardiomyopathies including the setting of chemotherapy and dilated cardiomyopathies [bib_ref] Randomized comparison of allogeneic versus autologous mesenchymal stem cells for nonischemic dilated..., Hare [/bib_ref] [bib_ref] Mesenchymal stem cell therapy for doxorubicin cardiomyopathy: hopes and fears, Ezquer [/bib_ref]. However, given the healthcare burden manifested by ischemic chronic HF, later trials have looked at the treatment potential of MSCs outside the time of acute MI [bib_ref] Adult bone marrow cell therapy improves survival and induces long-term improvement in..., Jeevanantham [/bib_ref]. The addition of studies investigating the population of patients afflicted with HF remains paramount given that the disease state remains the leading reason for morbidity and mortality. Furthermore, trials investigating nonischemic HF are needed as well. On a physiologic and biochemical level, chronic ischemic HF changes the microenvironment and biochemical milieu of signaling that occurs, thus altering cardiovascular physiology [bib_ref] Comparison of allogeneic vs autologous bone marrow-derived mesenchymal stem cells delivered by..., Hare [/bib_ref] [bib_ref] Intracoronary transplantation of autologous bone marrow mesenchymal stem cells for ischemic cardiomyopathy..., Chen [/bib_ref] [bib_ref] Mesenchymal stromal cells for cardiovascular repair: current status and future challenges, Mathiasen [/bib_ref]. In addition to MSCs, other cellbased therapies are also under consideration in clinical trials as potential alternatives for cell-based therapies; however, the majority of trials remain focused on MSCs given their desirable characteristics, ease of use, and accessibility [bib_ref] Postcardiac transplant survival in the current era in patients receiving continuous-flow left..., Kamdar [/bib_ref] [bib_ref] Cardiopoietic stem cell therapy in heart failure: the C-CURE (cardiopoietic stem cell..., Bartunek [/bib_ref] [bib_ref] Rationale and design of the first randomized, double-blind, placebo-controlled trial of intramyocardial..., Mathiasen [/bib_ref] [bib_ref] A randomized study of transendocardial injection of autologous bone marrow mononuclear cells..., Perin [/bib_ref]. Indeed, no other cell-based therapy for HF continues to hold as much potential as MSCs for a true "off the shelf" approach that can be utilized in autologous or allogeneic modalities.
## Results of clinical trials with mscs
Predominantly, clinical trials have demonstrated that MSCs are safe for administration without increased risk for adverse events [bib_ref] Suppression of allogeneic T-cell proliferation by human marrow stromal cells: implications in..., Tse [/bib_ref] [bib_ref] Postcardiac transplant survival in the current era in patients receiving continuous-flow left..., Kamdar [/bib_ref] [bib_ref] Cardiac resynchronization therapy with or without defibrillator: experience from a high-volume Belgian..., Verbrugge [/bib_ref] [bib_ref] Under-utilization of evidence-based drug treatment in patients with heart failure is only..., Lenzen [/bib_ref]. Furthermore, results of the previous trials suggest that treatment with MSCs does not increase risk of posttreatment arrhythmias [bib_ref] Postcardiac transplant survival in the current era in patients receiving continuous-flow left..., Kamdar [/bib_ref]. Additionally, studies have shown significant improvement in patient exercise tolerance [bib_ref] Postcardiac transplant survival in the current era in patients receiving continuous-flow left..., Kamdar [/bib_ref] [bib_ref] Under-utilization of evidence-based drug treatment in patients with heart failure is only..., Lenzen [/bib_ref]. Investigations have also looked to see if MSCs utilized as concurrent or adjuvant therapies with existing treatment modalities for HF such as in the setting of mechanical circulatory support devices can provide benefits [bib_ref] Mesenchymal precursor cells as adjunctive therapy in recipients of contemporary left ventricular..., Ascheim [/bib_ref]. While many of these clinical trials have found trends towards improvement in New York Heart Association (NYHA) class, the results have not always been statistically significant or with dramatic improvements in treatment versus nontreatment groups. However, this trend seems to be changing. Recent trials have started to demonstrate findings that are statistically significant, likely due to several factors, among which is increasing sample size [bib_ref] Postcardiac transplant survival in the current era in patients receiving continuous-flow left..., Kamdar [/bib_ref] [bib_ref] Under-utilization of evidence-based drug treatment in patients with heart failure is only..., Lenzen [/bib_ref]. Another factor playing a role in observed results likely lies with MSCs themselves. The route of administration has been a focal point of investigation of MSCs in HF. Initial studies were in the setting of acute myocardial infarction, and intracoronary delivery was the standard practice [bib_ref] Advances in the treatment of ischemic diseases by mesenchymal stem cells, Li [/bib_ref]. The limitations to delivering MSCs by the intracoronary route is the observation that the cells are rapidly washed out. Investigators have tried to increase the dwelling time in the coronary artery by delivering the cells and then occluding the coronary artery. To date, it is not clear that this approach increases the number of cells remaining in the heart. Attempts have been made to investigate alternative approaches as well.
Intravenous allogeneic MSCs have been tested in a small pilot study in patients with nonischemic cardiomyopathy [bib_ref] Intravenous allogeneic mesenchymal stem cells for nonischemic cardiomyopathy: safety and efficacy results..., Butler [/bib_ref]. The authors speculate that the anti-inflammatory effects of the MSCs may be the mechanism of action because the cells were delivered intravenously. Other studies have investigated the role of endomyocardial delivery of MSCs and coronary sinus approaches as well, albeit the latter was with bone marrow aspirate and not purely MSCs [bib_ref] Comparison of allogeneic vs autologous bone marrow-derived mesenchymal stem cells delivered by..., Hare [/bib_ref] [bib_ref] Randomized comparison of allogeneic versus autologous mesenchymal stem cells for nonischemic dilated..., Hare [/bib_ref] [bib_ref] Transendocardial mesenchymal stem cells and mononuclear bone marrow cells for ischemic cardiomyopathy:..., Heldman [/bib_ref] [bib_ref] REVIVE trial: retrograde delivery of autologous bone marrow in patients with heart..., Patel [/bib_ref]. However, the route of administration remains an area that requires more investigation.
Studies have quelled previous concerns regarding safety in administration of MSCs [bib_ref] Cell-based therapies for myocardial repair: emerging role for bone marrow-derived mesenchymal stem..., Silva [/bib_ref] [bib_ref] Adult bone marrow cell therapy improves survival and induces long-term improvement in..., Jeevanantham [/bib_ref] [bib_ref] Intracoronary transplantation of autologous bone marrow mesenchymal stem cells for ischemic cardiomyopathy..., Chen [/bib_ref] [bib_ref] A randomized study of transendocardial injection of autologous bone marrow mononuclear cells..., Perin [/bib_ref]. This has even been investigated in the setting of dilated cardiomyopathies [bib_ref] Randomized comparison of allogeneic versus autologous mesenchymal stem cells for nonischemic dilated..., Hare [/bib_ref]. The POSEIDON-DCM trial is a randomized comparison of allogeneic versus autologous MSCs for nonischemic dilated cardiomyopathy delivered transendocardially [bib_ref] Comparison of allogeneic vs autologous bone marrow-derived mesenchymal stem cells delivered by..., Hare [/bib_ref]. Although in a small trial, the early results are encouraging showing that the allogeneic cells increased ejection fraction and functional activity with no significant serious adverse events. The CONCERT-HF trial is an ongoing study investigating the combination of MSCs and c-kit + cardiac stem cells in ischemic cardiomyopathy (NCT02501811), and is, interestingly, a trial that employs autologous bone marrowderived MSCs. While the original approach for cell-based therapy in heart failure focused on using autologous cells, investigators are now using allogeneic cells because of the potential to deliver larger numbers of cells without harvesting from the patient. Allogeneic cells can be obtained from younger subjects where the cells may have more regenerative capacity based on protein expression and bioinformatics [bib_ref] MicroRNA profiling reveals age-dependent differential expression of nuclear factor κB and mitogen-activated..., Pandey [/bib_ref]. Furthermore, it has been suggested that HF, at its central process, is an inflammatory process [bib_ref] Interleukin-1β modulation using a genetically engineered antibody prevents adverse cardiac remodelling following..., Abbate [/bib_ref]. In addition, it has been shown that MSCs are potently antiinflammatory [bib_ref] Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal..., Ortiz [/bib_ref]. Knowing this, it follows that MSCs could slow and potentially reverse the ill effects of HF, if not too far progressed. Lastly, using allogeneic cells would be less costly and be able to deliver a true "off the shelf" approach. Allogeneic MSC therapy has been shown to be as safe and efficacious as autologous MSC therapy [bib_ref] Comparison of allogeneic vs autologous bone marrow-derived mesenchymal stem cells delivered by..., Hare [/bib_ref] [bib_ref] Randomized comparison of allogeneic versus autologous mesenchymal stem cells for nonischemic dilated..., Hare [/bib_ref] [bib_ref] Cardiac repair with intramyocardial injection of allogeneic mesenchymal stem cells after myocardial..., Amado [/bib_ref] [bib_ref] Allogeneic administration of fetal membrane-derived mesenchymal stem cells attenuates acute myocarditis in..., Ishikane [/bib_ref] [bib_ref] A randomized study of transendocardial injection of autologous bone marrow mononuclear cells..., Perin [/bib_ref] [bib_ref] Transendocardial mesenchymal stem cells and mononuclear bone marrow cells for ischemic cardiomyopathy:..., Heldman [/bib_ref] [bib_ref] Effect on left ventricular function of intracoronary transplantation of autologous bone marrow..., Chen [/bib_ref] [bib_ref] Bone marrow-derived mesenchymal stromal cell treatment in patients with severe ischaemic heart..., Mathiasen [/bib_ref] [bib_ref] Effect of transendocardial delivery of autologous bone marrow mononuclear cells on functional..., Perin [/bib_ref] [bib_ref] Autologous mesenchymal stem cells produce reverse remodelling in chronic ischaemic cardiomyopathy, Schuleri [/bib_ref]. The primary mechanism by which MSCs are able to promote endogenous repair appears to be via paracrine signaling; however, the cells still need to hone to the area of interest and be retained, at least for some duration, to exert their therapeutic effects. Furthermore, the area(s) where MSCs hone to must have at least some vascularization given that MSCs act via paracrine signaling. Current administration of MSCs remains entrenched in traditional approaches including intravenous (IV) and intracardiac (IC) modes of delivery [bib_ref] Epidemiology of heart failure, Miller [/bib_ref] [bib_ref] Postcardiac transplant survival in the current era in patients receiving continuous-flow left..., Kamdar [/bib_ref] [bib_ref] Cardiac resynchronization therapy with or without defibrillator: experience from a high-volume Belgian..., Verbrugge [/bib_ref] [bib_ref] Under-utilization of evidence-based drug treatment in patients with heart failure is only..., Lenzen [/bib_ref] [bib_ref] Stem cells for myocardial repair with use of a transarterial catheter, Wang [/bib_ref]. Recent trials have investigated the use of allogeneic MSCs, which has been safe and effective in a comparable capacity to autologous MSC therapeutic strategies [bib_ref] Postcardiac transplant survival in the current era in patients receiving continuous-flow left..., Kamdar [/bib_ref]. Furthermore, preclinical studies as well as data from clinical trials have suggested that young donor MSCs have different signaling pathways activated when compared with older donor MSCs, given the changing cellular dynamics of aged MSCs [bib_ref] MicroRNA profiling reveals age-dependent differential expression of nuclear factor κB and mitogen-activated..., Pandey [/bib_ref] [bib_ref] Under-utilization of evidence-based drug treatment in patients with heart failure is only..., Lenzen [/bib_ref]. Thus, use of allogeneic young donor MSCs for treatment in an "off the shelf" therapeutic option make MSCs even more favorable as a cell-based therapy modality. This approach is being used in the Dream-HF trial with allogeneic mesenchymal precursor cells (NCT02032004). Indeed, an innovative delivery of MSCs could provide the missing component to propel MSCs as the long-sought-after treatment option for HF.
## Delivery of mscs with biomaterials
Delivery with tissue-engineered biomaterials could provide an innovative delivery system to enable MSCs to further develop as a treatment option for HF and other cardiovascular disorders. Administration of MSCs via a venous approach has the risk of cells honing to an area of damage that is outside the heart. While IC administration of MSCs will confirm that MSCs will be present in the heart, often the cells are injected into the scar, which given its lack of adequate vascular supply, would not be hospitable to MSCs as therapeutics.
Biomaterials are gaining increasing interest, especially with the advent of new technologies that allow for innovative treatment modalities. Administration of MSCs with the aid of biomaterials such as a scaffold could potentially resolve some of the perceived issues with MSC delivery in the setting of HF. Mesenchymal stem cells have been used in combination with biomaterials in preclinical studies with some promising results. Investigators injected a self-setting salinized hydroxypropyl methylcellulose seeded with MSCs and showed improvements in LV remodeling and infarct expansion in a rat model of myocardial infarction [bib_ref] Intramyocardial delivery of mesenchymal stem cell-seeded hydrogel preserves cardiac function and attenuates..., Mathieu [/bib_ref]. Another approach to improve retention of transplanted cells in the diseased heart is to inject the cells in an in situ cross-linked alginate hydrogel [bib_ref] Human adipose-derived stromal cells in a clinically applicable injectable alginate hydrogel: phenotypic..., Follin [/bib_ref]. Adipose-derived MSCs embedded in alginate retain their viability, maintain their paracrine potential, and are not immunogenic suggesting that using alginate hydrogels may be a method to enhance delivery of MSCs in the clinical arena. A similar approach encapsulating MSCs in an alginate hydrogel patch has shown potential clinical benefit in a rat infarct model with evidence of improved cardiac function, decreased scar size, and increased peri-infarct vasculature [bib_ref] Cellular encapsulation enhances cardiac repair, Levit [/bib_ref].
The ideal scaffold would allow delivery of MSCs in such a way that would allow the cells to exert their paracrine signaling and not to impede the release of these secreted factors. Furthermore, the scaffold used for cell delivery would create not only a vector for delivery of the cells but also a more hospitable microenvironment from which MSCs could direct endogenous repair. It would also be advantageous if the scaffold could help generate its own new blood supply. This would overcome the current dilemmas of how to target MSCs to specific areas of the heart as well as concerns of engraftment in a potentially hostile environment, the postinfarction myocardium. Additionally, a biodegradable scaffold would exist only transiently; once the cells have established, the scaffold would no longer be needed. Among the current hypotheses of MSCs used as therapeutics, one suggested that MSCs typically initiate and direct the early phases of endogenous repair, and once the process is sufficiently underway, these cells are not required and are not retained for longer periods. Ideally, the scaffold would result in minimal inflammation, remain only transiently, not require removal, and promote angiogenesis.
## Cellular scaffolds
Several approaches have been suggested or are under investigation as cellular delivery alternatives to IV or IC administration of cells. Among these methods are cell-or tissue-based scaffold, electroshock-assisted cell delivery, and polymers for transport [bib_ref] Transcoronary transplantation of functionally competent BMCs is associated with a decrease in..., Assmus [/bib_ref] [bib_ref] Effect of shock wave-facilitated intracoronary cell therapy on LVEF in patients with..., Assmus [/bib_ref] [bib_ref] Granulocytecolony stimulating factor or granulocyte-colony stimulating factor associated to stem cell intracoronary..., Bocchi [/bib_ref] [bib_ref] Use of stem cells in heart failure treatment: where we stand and..., Sanchez [/bib_ref]. Criticisms of cell-based therapies point towards poor retention of cells, insufficient number of cells utilized for therapy, poor engraftment in the hostile conditions of HF, disruption of molecular honing signals, and excess extracellular matrix secondary to fibrosis as the pitfalls of cell administration-based therapies as options for HF. The potential pitfalls described in cell-based therapies can all be overcome by the use of cellular/tissue scaffolds. Previously, we have described a cellular scaffold that has been developed and demonstrated hemodynamic improvement as well as promotion of angiogenesis [bib_ref] Viable fibroblast matrix patch induces angiogenesis and increases myocardial blood flow in..., Lancaster [/bib_ref] [bib_ref] Implantation of a three-dimensional fibroblast matrix improves left ventricular function and blood..., Thai [/bib_ref]. Such a cellular scaffold would not only allow for adequate delivery in regard to cell number but also location as it could be placed over the region of the scar. Furthermore, the scaffold would provide a more hospitable setting than the surrounding infarct, which would otherwise be much more hostile for MSC function. The scaffold would thus work in conjunction with MSCs to modulate the cellular microenvironment to make it more favorable and promote cellular repairs. MSCs administered with the scaffold would provide the necessary components for increased angiogenesis to occur, while reducing the need for honing and engraftment of MSCs. Utilizing a cellular scaffold, with known cardiovascular improvements in the setting of HF, with MSCs, given their potential as cellular therapeutics, could provide the elusive component necessary to progress cell-based therapies in HF.
# Conclusion
MSCs were once touted as the ideal candidate for cell-and gene-based therapies. They were identified as the cells that would change regenerative medicine via their ability to differentiate into end-cell lines, allowing the shortage of donor organs to become a nonfactor in treatment of many endstage disease states. As MSCs have been further investigated, it is their paracrine signaling that has come to the forefront and become the characteristic that makes them ideal candidates as cellular therapeutics. The ability to modulate the cellular microenvironment through expression of various cytokines and regulation of signal transduction pathways to direct and promote endogenous cellular repair is considered the hallmark function of MSCs. Furthermore, MSCs via their paracrine signaling are able to recruit dormant progenitor cells to aid in the regenerative process. Clinical trials have demonstrated that MSCs are safe and can play as a mainstay of treatment of HF. Novel delivery of MSCs as therapeutics in HF can overcome many of the current pitfalls such as hostile environment of HF for regenerative medicine and retention of cells. Cellular scaffolds in particular can assure that critical numbers of MSCs are able to reach the target area, whereby MSCs can then direct endogenous cellular repair.
## Current and future perspectives
MSCs have been of interests for their potential as cellular therapeutics since their first description 37 years ago. Their ability to differentiate into end-cell lineages enticed investigators to believe that the dream of creating organs in the laboratory had become a reality. It was not until recently that it was determined that the therapeutic potential of MSCs was primarily in their mechanism of paracrine signaling, and not with the differentiation potentials. MSCs have been investigated in a myriad of disease states. Among the most devastating and costly disease of which currently remains is chronic HF. The scope of MSCs as potential therapies in HF is still in the very early stages. While significant progress is currently being made with ischemic HF clinical trials revealing that MSCs are not only safe for administration but may also provide the much anticipated therapeutic benefit. However, future research is needed to elucidate the ideal delivery of MSCs in the setting of ischemic cardiomyopathies, and research is needed in nonischemic cardiomyopathies as well. The mechanisms at play by which MSCs function to improve molecular and clinical state of HF need to be identified. Furthermore, clinically relevant endpoints of MSC therapy such as exercise time and functional capacity are important metrics to assess as we strive to improve the quality of life in patients with heart failure.
## Conflicts of interest
The authors have nothing to disclose. The authors declare that there is no conflict of interest regarding the publication of this article. : Cellular scaffold (arrows) placed over a scar in a rat heart in vivo 3 weeks after infarction. Three weeks after patch implantation, 6 weeks after infraction, evidence of angiogenesis from the patch to native myocardium was observed.
[table] Table 1: Clinical trials investigating mesenchymal stem cells in heart failure. [/table]
|
MaizeGDB's new data types, resources and activities
MaizeGDB is the Maize Genetics and GenomicsDatabase. Available at MaizeGDB are diverse data that support maize research including maps, gene product information, loci and their various alleles, phenotypes (both naturally occurring and as a result of directed mutagenesis), stocks, sequences, molecular markers, references and contact information for maize researchers worldwide. Also available through MaizeGDB are various community support service bulletin boards including the Editorial Board's list of high-impact papers, information about the Annual Maize Genetics Conference and the Jobs board where employment opportunities are posted. Reported here are data updates, improvements to interfaces and changes to standard operating procedures that have been made during the past 2 years. MaizeGDB is freely available and can be accessed online at
# Introduction
Maize (Zea mays ssp. mays) has long been the number one production crop in the United States, and in 2001 it also became number one in the world. As maize is economically important and also serves as a model organism for genetics research, it is one of the most highly researched organisms in existence. The maize genome has an especially high level of DNA sequence polymorphism and extended regions of non-homology between inbred lines [bib_ref] Intraspecific violation of genetic colinearity and its implications in maize, Fu [/bib_ref] , hence the diversity represented by the maize gene pool is unparalleled in both a phenotypic and molecular sense. This provides a unique vehicle to explore questions in evolution, domestication, development, trait expression, functional allelic diversity and the interrelated processes that shape such events and their outcomes.
The application of new technologies and bioinformatic tools coupled with thorough phenotypic evaluation for useful traits and molecular characterization of diverse maize germplasm offers the potential for significant discovery via translational genomics. The goal of turning the identification and evaluation of functional and evolutionarily important allelic variation into a comprehensive genomics activity is dependent on being able to associate diverse information in a seamless manner. Maize Genetics and Genomics Database (MaizeGDB) is the community resource charged with developing informatic solutions for storing, displaying and linking comprehensive maize data, so that they are made easily accessible to researchers worldwide. Described here are updates and improvements to MaizeGDB that have been made over the course of the past 2 years.
Throughout this paper direct links are listed parenthetically to enable direct access to relevant data. Those data also could be accessed via search and browse mechanisms available on the MaizeGDB front page. For example, to find a map that is listed parenthetically, one could search either 'all records' or 'maps' using the search bar at the top (and bottom) of any MaizeGDB pageusing the map name. That same data also could be accessed from the front page by navigating to the Maps Data Center using the link shown in. On the Maps Data Center Search page, the Simple or Advanced Search mechanisms could be used to search maps by name, type, or using other limiters. Other data types also can be accessed from the front page using similar tactics. Recombination Nodule maps are uncommon: at present, they are only available for maize, mouse, and tomato [bib_ref] Male mouse recombination maps for each autosome identified by chromosome painting, Froenicke [/bib_ref] [bib_ref] Two-dimensional spreads of synaptonemal complexes from solanaceous plants. VI. High-resolution recombination nodule..., Sherman [/bib_ref]. As many researchers are unfamiliar with this unique data type and to demonstrate the utility of the newly available maize Recombination Nodule maps, an example of how they can be used to speed up experiments that utilize the maize translocation stocks follows.
## New data types
A researcher wants to determine whether the gene product of her new mutant gof1, which maps to 3L within 1 cM of tub6, acts cell autonomously using a mosaic analysis. Since there are no suitable cell autonomous markers known to be proximal to gof1, she decides to use an A-A translocation to bring gof1+ distal to a cell autonomous marker on another chromosome. To do this, she needs to find a translocation stock with a breakpoint on 3L proximal to her gene of interest, and a breakpoint on another chromosome that is distal to a gene that can be used as a cell autonomous marker. To find a list of available translocations, she uses the link on the front page of MaizeGDB to get to the Stocks Data Center, scrolls down the page (past the green 'Advanced Stock Query' box) to Maize Genetics Cooperation Stock Center Resources [bib_ref] Maintaining collections of mutants for plant functional genomics, Scholl [/bib_ref] , and clicks the link for the 'Stock Center Catalog.' She clicks the link for 'Reciprocal Translocation (comprehensive list)' to arrive at http://www.maizegdb.org/ cgi-bin/stockcatalog.cgi?id¼3, and decides to try 1049B T1-3(5242) (3L.65; 1L.90). In order to determine the approximate positions of these breakpoints relative to genetically mapped loci, she goes to the Maps Data Center. Important information, data centers, tools and news items are accessible from the MaizeGDB home page. Data and bulletin boards as well as links to noteworthy or high profile projects are accessible directly on the front page. All MaizeGDB pages have the same header, which is loaded with functionality enabling, e.g. searches of all data from any page (a) and access to tools including the Community Curation Tools. The 'Maps' Data Center (b) can limit results by contained loci, chromosome, source or mapping panel and also allows direct access to unique map types including the Recombination Nodule maps through its 'Map Reports and Tools' section. Likewise, the 'Stocks' Data Center (c) enables queries by focus linkage group, genotypic variation, karyotypic variation and other limitors. Bulletin boards that keep researchers connected with the community include the Editorial Board (d), the Maize Genetics Executive Committee pages (e), and the Annual Maize Genetics Conference, (which is referred to simply as 'The Maize Meeting') site (f). Dates of database updates are available directly on the front page (g), as are important major efforts of interest to all maize researchers like the Maize Genome Sequencing project.
shows that 3L.65 lies about 5 cM proximal to tub6. This means that the breakpoint is proximal to her gene of interest! Using the Translator again for chromosome 1, she finds the position of 1L.90 relative to the 'Genetic 2005 1' map (accessible at http://www.maizegdb.org/cgi-bin/displaymaprecord. cgi?id¼940880). The output table shows that lw1, a suitable cell autonomous marker [bib_ref] Analysis of genetic mosaics shows that the extra epidermal cell divisions in..., Hake [/bib_ref] , lies between cent1 and the breakpoint. Using the stock for T1-3(5242), she can set up a stock heterozygous for gof1, lw1, and the translocation where gof1+ and lw1+ reside on the translocation, and the recessive mutant alleles reside on the normal (non-translocation) chromosomes. Without access to the maize Recombination Nodule map data and the Morgan2 McClintock Translator, cytological and genetic maps cannot be integrated directly. Setting up such an experiment would have required many additional tests using various translocation stocks.
## Data updates
All types of data (references, loci, maps, etc.) are updated regularly as time and human resources allow. Listed here are a few of the major recent updates to content.
MaizeGDB's sequence set is made up of all public maize sequences including expressed sequence tag (EST), cDNA, genome survey sequence (GSS), sequence tagged site (STS), HTC and genomic DNA sequences from GenBank (12) as well as the Uniprot (13) protein sequence set. This dataset is updated monthly utilizing a custom pipeline set up by workers at PlantGDB [bib_ref] Comparative plant genomics resources at PlantGDB, Dong [/bib_ref]. Whereas in the past only the Z.mays ssp. mays sequence set was included in the update, a recent change was made to the sequence set update pipeline to include sequences for all subspecies of Z.mays. Also stored at MaizeGDB is contig membership information for the PlantGDB GSS and PUT (putative unique transcript) contigs as well as the TIGR TC (tentative consensus) EST contig set ].
The IBM2 FPC0507 maps (http://www.maizegdb.org/cgibin/displaymapresults.cgi?term¼fpc0507) represent the anchored BAC finger print contigs which are currently being used to guide the B73 Maize Genome Sequencing Project. These maps were derived using anchor information for 414 contigs as assigned in the July 2005 manually edited FPC product [(16); http://www.genome.Arizona.edu/fpc/maize]. Represented on the map are loci where at least 2 BACs in the contig were empirically associated with a molecular marker. Positions for loci were derived using a hybrid coordinate comprised of an integer representing the nearest genetic anchor point followed by a decimal and the FPC consensus band (CB) coordinate. Loci are associated with the defining BACs, markers, and marker sequences, and are linked to the current contig representation at Arizona. This map adds over 25 000 new loci to MaizeGDB, most of which are associated with overgo probes designed to detect full-length cDNAs [bib_ref] Anchoring 9371 maize expressed sequence tagged unigenes to the bacterial artificial chromosome..., Gardiner [/bib_ref]. This map serves to integrate the genetic and physical maps and improves high resolution mapping capabilities for researchers working to localize a trait or phenotype of interest to a small region of the genome, thus facilitating, e.g. candidate gene discovery by chromosome walking.
Initiated to assist in anchoring BAC contigs and continued to support maize genetics research, the IBM Neighbors product approximates the genetic map orders of all loci mapped to better than 5 cM and most recently includes the loci which have only been placed onto anchored FPC contigs. Like the anchored BAC FPC contigs, the IBM Neighbors representation utilizes the IBM2 map as the framework. Each version is maintained, and the primary map source for a locus coordinate is displayed. The most recent version (http://www.maizegdb.org/cgi-bin/displaymapresults. cgi?term¼ibm2*2005*neighbors) includes well-ordered mutants based on the Genetic 2005 compilation (18), which incorporated the UMC 98 maps [bib_ref] A consensus genetic map, intermated B73 x Mo17 (IBM) neighbors, Polacco [/bib_ref].
Over 5300 newly mapped loci, based on sequenced probes where most were derived from a cDNA, were provided to MaizeGDB from mapping projects that are using highresolution intermated recombinant inbred panels of stocks. These came from: the Schnable lab (3391 insertion-deletion loci on 2 map versions, IBM IDP +MMP versions 4 and 5; http://maize-mapping.plantgenomics.iastate.edu), Matthieu Falque [1680 RFLP loci on IBM GNP and GNP LHRF maps; [bib_ref] Linkage mapping of 1454 new maize candidate gene loci, Falque [/bib_ref] ], and the Community Mapping Service (289 loci; described below).
## New functionalities
Along with additions of new data, MaizeGDB personnel remain committed to maintaining and improving upon the interface to the database. New map displays have 'sequence view' (which shows the sequences associated with each locus on the map) and 'primer view', (which shows the primers and probes for each locus on the map) options, which are accessible toward the top of all default map views (e.g.
http://www.maizegdb.org/cgi-bin/displaymaprecord. cgi?id¼143431). Search algorithms have been refined to allow researchers to simply enter a term in the search field toward the top of any page at MaizeGDB (seeand quickly find relevant results, and summary information is now available on search results pages so that the most relevant records can easily be identified. For example, if a researcher were to search all records with the string 'r1', over 170 loci would be found. To help with locating the locus for an exact match, i.e. the locus r1 colored1, the exact match is shown at the top of the list of results. In addition, though the appearance of many data displays has not changed, the underlying code has been rewritten to improve load times and to optimize computational efficiency.
In an effort to improve access to diverse maize data, the MaizeGDB interface has been modified to include an abundance of linkages to other databases including the Plant Ontology Consortium [(21); http://www.plantontology.org], Gramene [(22); http://www.gramene.org] and the Maize Sequencing Project's genome browser. Data displays provide abundant context-sensitive linkages to records in other databases, enabling users to visit, e.g. a gene record at MaizeGDB and with just a click quickly find related sequences, annotated maps and similar sequences at other websites.
Recently the Community Curation Tools (accessible through the 'tools' link at the top of any page at MaizeGDB;were updated to enable the entry of QTL data.
Insights gained from experience with QTL data entry into the legacy MaizeDB [bib_ref] Reporting and accessing QTL information in USDA's Maize Genome Database, Byrne [/bib_ref] were leveraged in planning the functionality of this module, and linking of trait and map location to germplasm containing the superior allele is enforced. New automatic nomenclature features ensure consistency and minimize data entry effort. General functionality of the QTL module is consistent with that of the other Community Curation Tools described previously [bib_ref] The maize genetics and genomics database. The community resource for access to..., Lawrence [/bib_ref].
## Community support activities
MaizeGDB hosts an Editorial Boardwhose members communicate with MaizeGDB personnel monthly to report their selections of current and historic literature germane to maize research. The list of Editorial Board selections is appropriate to guide journal clubs or for use by individuals keen to remain abreast of the advances relevant to maize research. It can be accessed at http://www.maizegdb.org/ editorial_board.php.
MaizeGDB personnel support the Maize Genetics Executive Committee (MGEC;by providing a venue for them to contact and interact with the community of maize researchers. MaizeGDB personnel create and administer customized community surveys and handle and monitor elections for the MGEC, (which involves the creation of methods for anonymous balloting and key-based restrictions to preclude stuffing the ballot box). These services help the MGEC to understand the needs of and communicate clearly with the community of maize researchers. Note also that author M. L. Schaeffer is a member of the MGEC.
The Annual Maize Genetics Conference (seeis growing by leaps and bounds, with a nearly 20% increase in attendance within the last 3 years and a near doubling in the number of abstracts submitted. Workers at MaizeGDB created a set of tools to accept abstract submissions and to manage review of the abstracts by Conference Steering Committee members. MaizeGDB personnel also maintain the mailing list for the Conference Steering Committee, and authors T. E. Seigfried and M. L. Schaeffer serve on the conference steering committee in an ex officio capacity and also assemble and print the conference program.
MaizeGDB hosts the Maize Newsletter (MNL), and Maize GDB Curator M. L. Schaeffer is a co-editor of that publication. The main MNL site is at MaizeGDB (http://www.ma izegdb.org/mnl.php), and new volumes are staged at the University of Missouri-Columbia (http://www.agron.missouri. edu). Contributions from collaborators are posted as received, and, with minor editing, redacted for printing once a year. The MNL also includes annual reports from the Maize Genetics Cooperation-Stock Center, the MaizeGDB, and any new map syntheses developed by MaizeGDB and collaborators. [bib_ref] Linkage mapping of 1454 new maize candidate gene loci, Falque [/bib_ref] and data submitted for inclusion in the public IBM Neighbors map (described above). The data sources are attributed, and sequence accessions related to mapped loci as well as sequence details, such as primers required to reproduce the mappings, are actively solicited and annotated.
## Database and availability
Standard operating procedures, accessibility and machine architecture are reviewed in detail elsewhere [bib_ref] The maize genetics and genomics database. The community resource for access to..., Lawrence [/bib_ref]. The following is a brief description of how the project databases are administered and explains availability and methods of access. The MaizeGDB schema is accessible online at http://www.maizegdb.org/MaizeGDBSchema.pdf. Presently there are three copies of the database and interface, which exist on three identical servers. The interface on each server interacts with data on the local copy of the database, thereby allowing the maintenance of a production environment (i.e. the copy accessed through http://www.maizegdb.org), a curation or staging environment, and an isolated testing and development environment. The development environment functions as a playground where data manipulation and interface development are tested. The curation database stores the most current data. As data are entered into the curation database (by researchers using the Community Curation Tools and by professional curators), they are initially listed as non-public and can only be viewed by MaizeGDB staff members. Once the data are reviewed, a curation level tag is changed so that the new records will become publicly accessible. Updates to production are carried out by replacing the existing production copy of the database with a duplicate of the curation database and the latest sequence files from PlantGDB. This update generally occurs on the first Tuesday of each month (see. The curation database is backed up on a daily basis and is available for download (http://goblin1.zool.iastate.edu/~oracle/) for those who have Oracle RDBMS installed locally. Requests to gain readonly SQL access to the database should be directed by email to [email protected]. Data housed at MaizeGDB are in the public domain, hence they are freely available for use without a license.
## Future plans
The genome of maize inbred line B73 is being sequenced, and the creation and public availability of the official site (called the Maize Genome Browser; http://www.maize sequence.org) is in the offing. The Maize Genome Browser has embedded links to MaizeGDB throughout, and contextsensitive links from MaizeGDB into the Maize Genome Browser are planned. By creating links to the Maize Genome Browser, MaizeGDB can connect researchers with up-to-date views of the maize genome as it is sequenced without bearing the responsibility of supporting an independent genome browser for maize. To learn more about the Maize Genome Sequencing Consortium's plans and to find updates on their progress, visit http://www.maizegdb.org/sequencing_project. php.
In most cases, a model organism's official gene models are housed at the model organism database (MOD). As the maize genome is being sequenced and it is anticipated that MaizeGDB (the MOD for maize) will be charged with storing and making available the official gene models, plans are in the works to put together an infrastructure for supporting the storage of this new data type and to create a version control system to allow for the storage of each major gene model release.
Not all datatypes are currently integrated with sequence data. Breeders find it difficult to locate all genomic and phenotypic data for plant germplasm collections because breeding and sequence data are housed in separate, disconnected databases. Although MaizeGDB stores maize data related to genetics and genomics, most historical, geographic origin, characterization and evaluation data associated with the National Plant Germplasm System's Plant Genetic Resources collections are housed in the Germplasm Resource Information Network (GRIN; http://www.ars-grin.gov/npgs/). Work to integrate MaizeGDB with GRIN is a high-priority item for development in the coming year so that breeders are enabled to more easily associate genetic and genomic data with traditional crop improvement information.
[fig] Figure 1: Important information, data centers, tools and news items are accessible from the MaizeGDB home page. Data and bulletin boards as well as links to noteworthy or high profile projects are accessible directly on the front page. All MaizeGDB pages have the same header, which is loaded with functionality enabling, e.g. searches of all data from any page (a) and access to tools including the Community Curation Tools. The 'Maps' Data Center (b) can limit results by contained loci, chromosome, source or mapping panel and also allows direct access to unique map types including the Recombination Nodule maps through its 'Map Reports and Tools' section. Likewise, the 'Stocks' Data Center (c) enables queries by focus linkage group, genotypic variation, karyotypic variation and other limitors. Bulletin boards that keep researchers connected with the community include the Editorial Board (d), the Maize Genetics Executive Committee pages (e), and the Annual Maize Genetics Conference, (which is referred to simply as 'The Maize Meeting') site (f). Dates of database updates are available directly on the front page (g), as are important major efforts of interest to all maize researchers like the Maize Genome Sequencing project. [/fig]
|
Radiation therapy is an established modality in the treatment of head and neck cancer patients. Compromised wound healing in irradiated tissues is a common and challenging clinical problem. The pathophysiology and underlying cellular mechanisms including the complex interaction of cytokines and growth factors are still not understood completely. In this review, the current state of research regarding the pathomechanisms of compromised wound healing in irradiated tissues is presented. Current and possible future treatment strategies are critically reviewed.
# Introduction
Radiation is employed as neoadjuvant, primary and adjuvant therapy for head and neck cancer. Complications after radiation therapy occur in up to 60 percent of surgical patients. Clinical sequelae include skin atrophy, soft tissue fibrosis, desquamation, epithelial ulceration, fistula formation and major vessel rupture. Impaired periand postoperative wound healing and the complications associated with it can be observedfrequently and may require extensive reconstructive efforts .
## Physiological wound healing
Adequate wound healing involves interactions of cells. Cell biologic mechanisms relevant to the process include interaction of keratinocytes, fibroblasts and endothelial cells. Epithelial closure of a wound is an important aspect of this complex biological process and relies primarily on the concerted action of activated keratinocytes and dermal fibroblasts. Three phases of wound healing with distinctive biochemical profiles have been described.
Hemostasis and inflammation (phase 1, day 0 to 4), are followed by proliferation (phase 2, day 3 to week 3) and maturation (phase 3, week 3 to 2 years).
These three phases are regulated by a complex network of interacting cytokines, growth factors and their cellular receptors.
## Effects of radiation therapy on wound healing
Wound healing occurs in an ordered sequence of cellular interactions. Repetitive radiation injury disrupts this highly organized sequence of events, resulting in repetitive inflammatory responses and ongoing cellular regeneration.
There is an important distinction to be made between the early and the late side-effects of radiation therapy: Early side effects include erythema, dry desquamation, hyperpigmentation and hair loss. Late effects include skin atrophy, dryness, telangiectasia, dyschromia, dyspigmentation, fibrosis, and ulcers.
The inflammatory and proliferative phases may be disrupted by the early effects of radiation. Affected factors during the inflammatory phase include transforming growth factor beta (TGFβ), vascular endothelial growth factor (VEGF), tumor necrosis factor-α (TNF-α), interferon-γ (IFN-γ) and proinflammatory cytokines such as interleukin-1 and interleukin-8. These cytokines are overexpressed after the radiation injury leading to uncontrolled matrix accumulation and fibrosis.
The proliferative phase is characterized by granulation tissue formation, re-epitheliaziation and neovascularization. This phase is mainly regulated by TGFβ, VEGF, epidermal growth factor (EGF), fibroblast growth factor (FGF) and platelet-derived growth factor (PDGF). Nitric oxide (NO) promotes wound healing by an induction of collagen deposition. NO levels have been reduced in irradiated wounds of experimental animals. This finding may explain the impaired strength of irradiated wounds.
During the remodeling phase, matrix metalloproteinases (MMP) and their tissue inhibitors are central to the process. MMP-1 is decreased after radiation therapy, which may contribute to inadequate soft tissue reconstitution.
Keratinocytes represent a crucial cell type in the repair of late epithelial wounds and ulcers. Multiple molecular biological changes are observed in this cell after radiation when compared to radiation-naïve skin. In human radiogenic wounds, these cells show a shift in the expression from high molecular keratins 1 and 10 to the low molecular keratins 5 and 14. In non-healing ulcers, keratinocytes display a decreased expression of transforming growth factor-alpha and -beta(1), fibroblast growth factor 1 and 2, keratinocyte growth factor, vascular endothelial growth factor, and hepatocyte growth factor. Expression of the matrix metalloproteinases 2, 12 and 13 has been shown to be elevated in irradiated human keratinocytes and fibroblasts.
Fibroblasts play the central role in wound healing through deposition and remodeling of collagen fibers. In irradiated tissue, fibroblasts have been shown to generate a disorganized deposition of collagen bundles. One likely mechanism resulting in disorganized collagen deposition is dysregulation of MMP and TIMP. These enzymes regulate extracellular matrix synthesis. As TGFbeta in turn regulates MMPs and TIMPs, this cytokine may be of particular relevance to radiogenic ulcers.
## Current strategies in treating irradiated wounds
Established strategies in treating radiogenic ulcers with delayed and inadequate healing include standard wound care, vacuum-assisted devices, substitution of nutritional deficiencies, and measures to optimize blood and oxygen supply.
Hyperbaric oxygen seems to optimize the partial pressure of tissue oxygen. It is frequently used for the treatment of osteoradionecrosis. The postulated effects include an increased capillary density and more complete neovascularization. hyperbaric oxygen in a cell culture model and found that its application resulted in the downregulation of 9 genes involved in adhesion, angiogenesis, inflammation and oxidative stress. Interleukin-8 mRNA levels were also suppressed after daily exposure of hyperbaric oxygen to endothelial cells in this study. IL-8 is one key Clinical case of a 55-year-old male, six months after primary radiochemotherapy due to an advanced squamous cell carcinoma of the hypopharynx. Skin atrophy and soft tissue necrosis were observed 8 weeks after the completion of therapy. factor in the inflammatory phase of wound healing. This suggests that the benefit of hyperbaric oxygen may not be limited to osteoradionecrosis and further investigations regarding additional therapeutic effects may be valuable.
Currently, hyperbaric oxygen therapy is clinically used in chronic diabetic ulcers and wound healing complications after radiotherapy. Its efficacy has been proven by several randomized trials, but it is it is important to emphasize that for both diabetic wounds and radiation injuries, this therapy is used in conjunction with standard wound care management techniques. A possible new approach for the hyperbaric oxygen therapy is the treatment of dermal wounds after flap surgery. Multiple case reports and animal studies have been written on this issue. Prospective clinical trials are necessary to support the application of hyperbaric oxygen to improve flap survival.
## Future aspects in the treatment of irradiated wounds
Research into new therapeutic approaches to treat radiogenic ulcers includes: Special dressings, injection of (multipotent) cells, topical administration of active substances and the use of growth factors.
## Special dressings
Different types of membranes and other covers of radiogenic skin injuries have been developed in the past.
## Hydrogel membranes
Hydrogel membranes have a stable, flexible and transparent structure. Data of Lu et al. indicate that the hydrogel membranes act as a semiocclusive dressing and maintain a moist environment over the wound bed. Through this mechanism, re-epithelialization is thought to be enhanced because of the accumulation of cytokines and growth factors that support wound healing. The use of hydrogels for partial thickness wounds after burns or laser resurfacing is supported by case reports. Prospective, controlled data on their treatment effect in radiogenic wounds are not available.
## Silver membranes
Bacterial cellulose impregnated membranes and poviargol, an antiseptic substance containing silver, was shown to accelerate the healing of irradiated wounds in an animal model. A recent animal study has revealed that silver nanoparticles have the potential to promote wound healing through accelerated re-epithelialization and enhanced differentiation of fibroblasts. The exact effect of silver-coated membranes on skin repair is only partially understood. Hypothetic mechanisms include the fact that collagen bundle organization and the tensile properties of the skinare found to be improved.
## Skin allografts
A human deceased-donor skin allograft provides dermoprotection and promotion of re-epithelialization. It is utilized until autografting is possible or re-harvesting of donor sites becomes available. This treatment is currently limited to burn injuries. It may be hypothesized that similar effects may be achieved in radiogenic ulcers, but clinical data and trials with sufficient numbers are not available.
## Injection of cells
Wound healing is a complex mechanism that requires active cellular interactions. These interactions are only possible if a sufficient number of intact and healthy cells are present. Radiation leads to impaired cellular activity and cell death. Injection of cells has the potential of enhancing wound healing in irradiated tissues. Nevertheless, there is a significant risk of infection and possible tumor induction by injecting viable cells in vivo. All of the studies discussed below have to be seen in this context. All interventions are experimental and presently not applied in routine clinical practice.
## Autologous fibroblasts
Autologous, unirradiated fibroblasts were shown in a rat model by Ferguson et al. to significantly improve healing of the irradiated surgical wound. This was evidenced by injections of culture medium, irradiated fibroblasts and non-irradiated fibroblasts in previously irradiated surgical wounds. The autologous non-irradiated fibroblasts caused greater increases of breaking load and ultimate tensile strength of the wounds than the irradiated control cells. Other authors also report that implanting isolated dermal fibroblasts leads to a significant increase of wound strength in irradiated mice skin. This effect may rescue wounds from the otherwise irreversible effect of prior irradiation. Both studies show promise for the treatment of irradiated wounds in patients. The injection of autologous fibroblasts would be possible at the end of a surgical procedure in irradiated patients. But this method has not yet led to the conduction of human clinical trials. One problem in this context is the risk of inducing neoplastic lesions by injecting viable cells.
## Multipotent stem cells
Dermis-derived multipotent stem cells seem to be easily harvested from animal skin. Chunmeng reports that both topical and systemic transplantation of dermis-derived multipotent cells accelerates healing after a radiation-induced injury. The suggested mechanism is the synthesis of important wound healing factors such as VEGF, PDGF and TGFbeta by these cells. An interesting finding of this animal study on rats was also the analysis of the cell supernatants. A medium previously incubated with dermisderived stem cells seems to contain relevant amounts of wound-healing promoting factors such as VEGF, PDGF and TGF-beta. This presents a valuable approach for further clinical studies.
Human mesenchymal stem cells have also been discussed in the therapy of radiogenic ulcers. Adiposederived stem cells have been described for the therapy of limited local injuries and seem to improve angiogenesis and the reconstitution of dermal architecture.
Akita et al. describe the use of adipose-derived stem cells (ADSC) in a female suffering from a radiogenic wound 40 years after irradiation due to a uterine carcinoma. They used an artificial dermis impregnated with ADSC to cover the wound bed and injected ADSC into the margins. This method was combined with the local administration of a basic fibroblast growth factor to improve angiogenesis. The authors documented excellent results of wound healing and a durable regenerated tissue after 1.5 years. The results of this study may be open to various interpretations. The observed effect may be due to the adipose-derived stem cells, the artificial dermis or the application of growth factors. Combined or additive effects cannot be excluded to explain the success of this approach. Clinical studies investigating the effect of adipose-derived stem cell injections into radiogenic wounds are not available. Hadad et al. developed a wound-healing model to study such effects in pigs. They found no effect of the ASCs injections alone, but they could document improved wound healing by a combination therapy of ASC and platelet rich plasma injections into irradiated wounds. The most important effects were an accelerated wound closure and an increased microvessel density after the combined treatment.
A proangiogenetic effect of stem cell injections has been suggested by the authors, but other mechanisms may also serve to explain the findings in this study. New vessel formation and the optimizing of tissue microcirculation present a valuable effect of stem cell injections with respect to cutaneous wound healing. But the benefit on wound healing might end up providing a favorable environment for tumor recurrence after radiotherapy. High risk constellations or the application of stem cell injections in elderly people who have a diminished life time may put these patients at risk for developing tumor recurrence, which would justify further clinical trials. The clinical observations of Akita et al. are supported by another porcine study. Porcine skin seems to react similarly to human skin as far as radiogenic injuries are concerned. Also wound-healing mechanisms are similar in humans and pigs. That is why porcine animal studies are of major value in this field of research. Forcheron et al. injected autologous adipose-derived stem cells into the skin of pigs irradiated with 50Gy. This study focused on the cutaneous radiation syndrome. In this context the authors observed an improved clinical wound healing and an enhanced re-epitheliasation in animals injected with adipose-derived stem cells.
The antioxidant effects of adipose-derived stem cellsare reported by Kim et al.. These effects seem to be mainly mediated through the activation of dermal fibroblasts and keratinocytes via the paracrine mechanism.
Another advantage of adipose-derived stem cells over other stem cell sources is that they are easily obtained in large quantities by liposuction. Also their potency to synthesize growth factors and cytokines shows promise for the use in skin repair and regeneration. A cell culture study by Lee et al. supports the stimulatory effects of adipose-derived stem cells on cutaneous wound healing. In their study, the proliferation of fibroblasts and their collagen synthesis were increased by a conditioned medium of adipose-derived stem cells in vitro.
The studies mentioned above reveal that adiposederived stem cells seem to interact directly and via paracrine mechanisms with the key cells of wound healing. Clinical trials analyzing the role of these cells and their combination with other wound-healing promoting factors would contribute to a better understanding of the potential value of this treatment regimen.
## Experimental application of active agents
The topical application of active agents to reduce the side effects of radiation therapy was analyzed in several cell culture and animal studies.
## Bioflavonoids
Biologically active polyphenolic bioflavonoids have been shown in cell culture studies with keratinocytes to have positive effects on angiogenesis. The suggested mechanism is an increased expression of the vascular endothelial growth factor via the TNF (tumor necrosis factor)-alphapathway. Angiogenesis plays a central role in wound healing. VEGF is believed to be the most prevalent, efficacious and long-term signal that is known to stimulate angiogenesis in wounds.
## Histone deacetylase inhibitors
Histone deacetylase inhibitors (phenylbutyrate, trichostatin A and valproic acid) were shown to suppress the cutaneous radiation syndrome in rat skin. Chung et al. analyzed different histone deacetylase inhibitors with respect to acute skin reaction and dermal fibrosis. They used vaseline as a negative control and performed immunohistological analyses. The underlying mechanism seems to be the suppression of the aberrant expression of radiation-induced transforming growth factor-beta 1 and 2 and tumor necrosis factor alpha.
This study presents the value of antitumor histone deacetylase inhibitors which may suppress cutaneous radiation syndrome and are possible new agents for increasing therapeutic gain in cancer radiotherapy.
## Ascorbic acid
Ascorbic acid has been shown to result in the significant acceleration of healing of radiogenic ulcers. This was shown in a mouse model with doses of 10,16 and 20 Gy. A full thickness skin wound was created after radiation. Histological evaluations were performed at various times after wounding. Pretreatment with ascorbic acid augmented the synthesis of collagen significantly as revealed by an increase in hydroxyproline content. In the ascorbic acid group, an earlier wound closure was observed compared to the control group. Optimizing collagen synthesis in a poorly healing wound by the application of ascorbic acid seems to be reasonable, but clinical trials are still lacking.
## Copper tripeptide
Copper tripeptide accelerates the growth of normal and irradiated fibroblasts, which was shown in a cell culture study by Primary human dermal fibroblasts were explanted from intraoperative specimens obtained from patients after radiation therapy for head and neck cancer. Normal unirradiated fibroblasts served as a control. Irradiated fibroblasts treated with copper tripeptide showed a similar proliferation rate as the untreated controls, and produced significantly more basic fibroblast growth factor and vascular endothelial growth factor than untreated controls. An early increase in growth factors and cell proliferation by irradiated fibroblasts treated with copper tripeptide may improve wound healing.
## Thrombin receptor-activating peptide
Thrombin receptor-activating peptide (P517-30) has been shown to increase wound-breaking strength in irradiated tissues. This substance is a synthesized highaffinity thrombin receptor binding peptide. In a rat model, Cromack et al. revealed that P517-30 directly stimulates resident endothelial cells and fibroblasts to overcome dermal and circulating monocytic deficits. These results suggest a mechanism to accelerate wound healing with a potential clinical application and emphasize the activity of thrombin as a growth factor. Special hydrogels could be possible applications for these growth factors.
## Application of growth factors
Different growth factors have been analyzed for their potential role in wound healing, albeit their carcinogenic potential must be taken into consideration.
Important factors that may be suitable for this therapy include recombinant human granulocyte colony-stimulating factor (rhG-CSF), recombinant human macrophage colonystimulating factor (rhM-CSF), basic fibroblast growth factor, TGF-beta and an inhibitor of transforming growth factor (TGF)-beta(1) receptor kinase.
## Tgf-beta and tgf-beta antisense oligonucleotides
The exogenous application of TGF-beta (1) and (3) has the potential to act as a radioprotective agent, especially in adjuvant therapeutic regimens. In a rat model, reduced TGF-beta(3) expression was observed in the irradiated graft bed and induction of collagen synthesis was observed after application of TGF-beta (1).
Whether these findings are relevant in the clinical setting after radiation therapy remains to be seen, because TGF-beta suppression seems to have favorable effects as well:
TGF-β is a strong stimulator of extracellular matrix deposition. A new pharmacological approach was the development of the TGF-beta antisense oligonucleotide technology. This intervention shows promise as a therapeutic option for the inhibition of proteolytic tissue destruction, which is one key approach to optimize wound healing. Irradiated wounds often fail to heal because of high amounts of MMPs. TGF-beta antisense oligonucleotides seem to affect the major cell types of dermal wound healing: Fibroblasts, keratinocytes and endothelial cells were influenced in their gene expression of MMPs. MMP-1 and MMP-9 were significantly decreased after the treatment of fibroblasts and keratinocytes with TGF-beta antisense oligonucleotides in vitro. Animal models of wound healing and scarring after eye surgery documented an anti-scarring effect of the TGF-beta antisense oligonucleotides in vivo.
TGF-beta antisense oligonucleotide technology shows promise. VEGF up-regulation was observed in vitro and a pro-angiogenetic effect of TGF-beta antisense oligonucleotides in radiation-induced dermal wounds was suggested. Another study showed TGF-beta antisense oligonucleotides result in an increased expression of MMP protein and mRNA in tissue samples from radiation-induced chronic dermal wounds when compared to normal human skin. These effects were observed by immunohistochemistry and microarray analysis. Antisense TGF-beta oligonucleotide treatment also significantly down-regulates MMP secretion in vitro. Through these mechanisms, proteolytic tissue destruction may be inhibited in radiogenic ulcers. Clinical trials would be helpful to better understand the effects of TGF-beta oligonucleotides in vivo. The interesting issue in this context is balancing the positive effects of matrix formation by TGF-beta and the negative side effects of a TGF-beta overexpression after radiotherapy.
## Platelet-derived growth factor
Platelet-derived growth factor (PDGF) may be useful as a topical agent in post-irradiation surgical incisions. PDFG is already in clinical use and shows favorable results. In previously irradiated tissue, rhPDGF (recombinant human platelet-derived growth factor) has been shown to enhance wound healing through the induction of granulation tissue formation. PDGF is a cytokine that is only activated in the presence of bone marrowderived cells like wound macrophages. The transformation of wounds from a chronic to a short-term healing state after rhPDGF treatment was documented by serial histological examinations. The risk of tumor induction is difficult to quantify and must be considered prior to clinical application.
## Granulocyte macrophage-colony stimulating factor
Granulocyte macrophage-colony stimulating factor (GM-CSF) has been shown to modulate lipid peroxidation and glutathione content in skin wounds.
It has been demonstrated that GM-CSF increases the number of neutrophils, eosinophils and monocytes with corresponding bone marrow changes and might be suitable for different approaches in cancer therapy. Irradiation decreases incisional healing and produces oxygen radicals that damage cells. Because of the lipid component in the membrane, the cellular membranes are particularly susceptible to radiation damage due to peroxidation. Glutathione acts as a co-substrate in the enzymatic repair of radiation damage. Suppressed levels of glutathione have been shown to increase after the administration of granulocyte macrophage-colony stimulating factor in irradiated rats. The reduction of complications associated with radiochemotherapy is one possible goal of GM-CSF application.
## Recombinant human epidermal growth factor (rhegf)
The use of recombinant human epidermal growth factor (rhEGF) was studied in mice after radiation with 45 Gy. Histological examinations showed that treatment with EGF accelerated normal wound healing when compared to no treatment. Collagen distribution was significantly increased in the group treated with EGF. Dermal and epidermal structure was also more stable in the treatment group.
## Combined treatments
Pollard et al. investigated cytokine expression in a cell culture model of irradiated human skin isolates. Basic fibroblast growth factor and vascular endothelial growth factor were found to be less expressed in the radiationinduced dermal wounds. These authors concluded that an early combined increase in basic fibroblast growth factor and vascular endothelial growth factor production in irradiated fibroblasts may improve wound healing. Follow-up clinical studies have not yet been conducted.
There is some evidence to support the combination of different growth factors to escalate their therapeutic effect. Important factors that may be suitable for combination therapy include recombinant human granulocyte colony-stimulating factor (rhG-CSF), recombinant human macrophage colony-stimulating factor (rhM-CSF), basic fibroblast growth factor and an inhibitor of transforming growth factor (TGF)-beta(1) receptor kinase.
Sugiyama et al. studied combined treatment in rats after local irradiation with 30 Gy. Recombinant human granulocyte colony-stimulating factor (rhG-CSF), recombinant human macrophage colony-stimulating factor (rhM-CSF) and an inhibitor of transforming growth factor (TGF)-beta1 receptor kinase, were injected into a full-thickness incisional wound site in the dorsal skin. Following combined treatment with the above three compounds the breaking strength of the irradiated skin increased to approximately one-half of that in the nonirradiated skin. Histological analysis of the wounded skin revealed an increase in the formation of collagen fibers. Moreover, the increased breaking strength was associated with an increase in a subpopulation of fibrocytes.
# Conclusion
Radiotherapy is an integral modality of head and neck cancer therapy. Compromised wound healing is an important side effect of radiation therapy. The current body of literature comprises a large number of studies investigating the molecular, cellular and clinical effects of compromised wound healing as well as current and possible future therapeutic strategies. Many of the mechanisms leading to cell injuries are still not completely understood. A better understanding of tolerance doses, improved timing of the radiation regimen, and radiation sources would allow a more focused tumor treatment. These advances have led to an improved, but not yet complete protection of healthy tissues. The clinical challenge to optimize wound healing in irradiated patients remains. The present paper critically reviews and summarizes the literature concerning the biology and possibly therapeutic strategies of radiation-induced compromise in wound healing, including stem cell injections and application of growth factors. |
Upregulation of Inflammatory Genes and Downregulation of Sclerostin Gene Expression Are Key Elements in the Early Phase of Fragility Fracture Healing
Background: Fracture healing is orchestrated by a specific set of events that culminates in the repair of bone and reachievement of its biomechanical properties. The aim of our work was to study the sequence of gene expression events involved in inflammation and bone remodeling occurring in the early phases of callus formation in osteoporotic patients.Methodology/Principal Findings: Fifty-six patients submitted to hip replacement surgery after a low-energy hip fracture were enrolled in this study. The patients were grouped according to the time interval between fracture and surgery: bone collected within 3 days after fracture (n = 13); between the 4 th and 7 th day (n = 33); and after one week from the fracture (n = 10). Inflammation-and bone metabolism-related genes were assessed at the fracture site. The expression of proinflammatory cytokines was increased in the first days after fracture. The genes responsible for bone formation and resorption were upregulated one week after fracture. The increase in RANKL expression occurred just before that, between the 4 th -7 th days after fracture. Sclerostin expression diminished during the first days after fracture.Conclusions:The expression of inflammation-related genes, especially IL-6, is highest at the very first days after fracture but from day 4 onwards there is a shift towards bone remodeling genes, suggesting that the inflammatory phase triggers bone healing. We propose that an initial inflammatory stimulus and a decrease in sclerostin-related effects are the key components in fracture healing. In osteoporotic patients, cellular machinery seems to adequately react to the inflammatory stimulus, therefore local promotion of these events might constitute a promising medical intervention to accelerate fracture healing.
# Introduction
The management of fragility fractures associated with osteoporosis is difficult due to several factors including inadequate fixation strength of implants used to stabilize the fracture until union of bone occurs. In particular, the fragility fractures affecting the metaphyseal region of long bones are associated with an increased rate of complications. Several studies report nonunion in 2-10%, malalignment after surgery in 4-40%, metal work failure in 1-10%, and reoperation in 3-23% [bib_ref] Fracture healing in osteoporotic fractures: is it really different? A basic science..., Giannoudis [/bib_ref]. Experimental studies have shown that the decline in the capacity for fracture repair is age related. Disturbance of the full redevelopment of mechanical strength within fracture calluses in elderly animals has been shown in experimental rat models. In the human being it is possible that fracture healing is affected by aging [bib_ref] Cellular basis for agerelated changes in fracture repair, Lu [/bib_ref] , particularly in the elderly osteoporotic patients [bib_ref] The influence of osteoporosis in femoral fracture healing time, Nikolaou [/bib_ref]. In another study it was described that in 6, 26 and 52 weeks old rats, there is upon aging a delay in radiographic progression of fracture healing but the expression of the key genes involved in this process is not age-dependent [bib_ref] Fracture healing in osteoporotic fractures: is it really different? A basic science..., Giannoudis [/bib_ref] [bib_ref] The effect of age on gene expression in adult and juvenile rats..., Desai [/bib_ref] [bib_ref] Young, adult, and old rats have similar changes in mRNA expression of..., Meyer [/bib_ref]. The fracture healing response and its temporal gene expression in elderly patients with osteoporosis has not been adequately investigated at the cellular and molecular level. Identification of the mechanisms that lead to fracture healing disturbances in patients with osteoporosis is of outstanding importance because they could allow prevention and better management of these healing complications. In addition, the biological processes behind fracture healing in osteoporosis might hold the key for future medical interventions.
Fracture healing recapitulates certain aspects of skeletal development and growth, involving interplay of cells, growth factors and extracellular matrix. Following injury, a blood clot is formed in the fracture site [bib_ref] Fracture healing: the diamond concept, Giannoudis [/bib_ref] [bib_ref] Play and players in bone fracture healing match, Marzona [/bib_ref]. This hematoma is the source of several signaling molecules that induce an inflammatory cascade of events that initiate healing [bib_ref] The expression of cytokine activity by fracture callus, Einhorn [/bib_ref] [bib_ref] Expression of osteoprotegerin, receptor activator of NF-kappaB ligand (osteoprotegerin ligand) and related..., Kon [/bib_ref]. Based on histological observations of healing fractures, bone repair was defined in animal models by an initial inflammatory phase (lasting for about three days), a catabolic phase where damaged tissues are removed, and an anabolic phase where new bone is rebuilt. Within several days of the initial inflammatory response there is a sequence of events that results in the formation of new bone through the development of a structure named callus. Experimental studies have related temporal gene expression with bone healing. In a study with Sprague-Dawley rats, gene expression was evaluated on days 3 and 11 post-fracture. The authors showed that different molecular pathways of gene expression regulate different phases of bone healing [bib_ref] Microarray analysis of gene expression during the inflammation and endochondral bone formation..., Rundle [/bib_ref].
This work aims to study the profile of genes involved in inflammation and bone remodeling during the 3 major steps of the early phase of callus formation in human bone after a hip fragility fracture.
# Results
## Study population
Fifty-six patients 8067 years of age, 75% of female gender, which suffered a hip fragility fracture, were enrolled in this study. There were no statistical significant differences in age and gender between the 3 study groups : those who had surgery less than three days after fracture (group 1), between four and seven days (group 2) and eight or more days post-fracture (group 3).
## Inflammation and growth factors genes
The local expression of nine genes related to the inflammatory phase of bone healing (IL-1b, IL-6, TNF, BMP2, BMP4, TGF-b1, IGF-I, FGF-2 and PDGF-b) was analyzed [fig_ref] Figure 1: Relative expression of inflammation and growth factors genes grouped according to the... [/fig_ref] , [fig_ref] Table 2: Comparison between the relative gene expression levels of patients submitted to hip... [/fig_ref]. IL-1b, IL-6 and TNF are cytokines that have an important role in potentiating the inflammatory cascade. Concordantly, the expression of these genes was highest during the first 3 post-fracture days and decreases thereafter. Specifically, IL6 had a higher expression in group 1 than in group 2 (p-value = 0.021) and IL1B, although expressed at low levels, remained stable, decreasing slightly after the 4 th -7 th day post-fracture (p-value = 0.087). On the contrary, TNF expression was stable, showing only a slight tendency to decrease over time (p-value = 0.208).
BMPs are a set of growth factors and cytokines belonging to the TGF-b superfamily and are involved in the creation of bone tissue architecture. In fracture healing, BMP-2 and BMP-4 play important roles in osteoblast differentiation. Accordingly, it was observed that the expression levels of BMP2 were highest until 3 days post-fracture and decreased thereafter (p-value = 0.023), while BMP4 expression remained fairly constant in all groups (pvalue = 0.852). TGFB exhibited a constant negative slope between the three groups (p-value = 0.051).
IGF-I is a hormone involved in bone matrix synthesis and there were no differences in its expression levels in the three groups analyzed (p-value = 0.817). The growth factors FGF-2 and PDGFb are involved in the formation of new blood vessels. Their expression tended to decrease slightly from group 1 to group 2 and was clearly decreased after 8 days post-fracture (FGF2: pvalue = 0.091 and PDGFb: p-value = 0.043).
Overall, these findings suggest that the expression levels of inflammatory genes and growth factors are particularly high during the three first days post-fracture and decrease from the day 4 onwards.
## Osteoprotegerin, rank and rankl
OPG is a negative regulator of bone resorption and, as expected, its expression was slightly lower in group 3 than in group 1 (p-value = 0.168) [fig_ref] Figure 2: Relative expression of bone metabolism-related genes divided according to the time between... [/fig_ref] , [fig_ref] Table 2: Comparison between the relative gene expression levels of patients submitted to hip... [/fig_ref]. On the other hand, RANK produced by osteoclast precursors showed a tendency to increase over time (p-value = 0.072). Finally, RANKL expressed by osteoblasts, stromal cells and immune system cells had its highest level at days 4 to 7 post-fracture (group 2) and decreased thereafter (p-value = 0.267).
The ratio RANKL/OPG regulates the balance between remodeling and formation. In this study, the ratio RANKL/OPG mRNA peaked in group 2 and tended to decrease later on (pvalue = 0.078).
# Steoblast-related genes
The expression of three genes that play important functions in the osteoblast and its activity was studied [fig_ref] Figure 2: Relative expression of bone metabolism-related genes divided according to the time between... [/fig_ref] , [fig_ref] Table 2: Comparison between the relative gene expression levels of patients submitted to hip... [/fig_ref]. Core-binding factor, alfa subunit 1/runt-related transcription factor 2 (CBFA1/RUNX2) and osterix (OSX) are transcription factors that play a crucial role in osteoblast differentiation, and ALP is an enzyme expressed at a later stage being involved in bone matrix maintenance. In our results, CBFA1/RUNX2 expression was slightly higher in group 2 than in group 1 and then remained constant and was similar in groups 2 to 3 (p-value = 0.521). On the other hand, OSX expression levels were similar in groups 1 and 2 and marginally higher in group 3 (p-value = 0.149). Regarding ALP, levels were slightly lower in group 2 as compared to group 1 but were similar between groups 2 and 3 (p-value = 0.726).
# Osteocyte-related genes
Sclerostin (SOST) is produced by the osteocyte and regulates negatively osteoblast differentiation by inhibiting Wnt/b-catenin . Description of the study population divided between the event of fracture and surgery. signaling. In this work we found a highly significant reduction in SOST levels overtime (p-value = 0.001) [fig_ref] Figure 2: Relative expression of bone metabolism-related genes divided according to the time between... [/fig_ref] , [fig_ref] Table 2: Comparison between the relative gene expression levels of patients submitted to hip... [/fig_ref].
# Osteoclast-related genes
Four genes that regulate osteoclast differentiation and function were studied [fig_ref] Figure 2: Relative expression of bone metabolism-related genes divided according to the time between... [/fig_ref] , [fig_ref] Table 2: Comparison between the relative gene expression levels of patients submitted to hip... [/fig_ref]. Tartrate-resistant acid phosphatase (TRAP) is an enzyme expressed by pre-osteoclasts and active osteoclasts; on the other hand, cathepsin K (CTSK), b3 subunit of the avb3 integrin (ITGB3) and ATPase H + transporter (ATP6V0D2) are proteins involved in the process of bone resorption.
We observed a marked increase in TRAP and a more modest change in CTSK and ATP6V0D2 expression over the post fracture period (TRAP: p-value = 0.009; CTSK: p-value = 0.027 and ATP6V0D2: p-value = 0.076). On the other hand, the expression of ITGB3 remained constant (p-value = 0.658).
# Discussion
Although some studies have addressed the sequence of events in fracture healing, the research was mainly based on histological examination of healthy individual's tissue and on molecular studies in animal models. Based on these previous histological and molecular studies of healing fractures the initial stages of these process have been proposed to include an early inflammatory and unspecific anabolic phase (first 24 hours up to third day), immediately followed by a non specific catabolic phase (up to the end of the first week) that sets the conditions for a more bone specific anabolic phase (first week and thereafter) [bib_ref] Microarray analysis of gene expression during the inflammation and endochondral bone formation..., Rundle [/bib_ref] [bib_ref] Bone remodeling during fracture repair: The cellular picture, Schindeler [/bib_ref] [bib_ref] Transcriptional analysis of fracture healing and the induction of embryonic stem cell-related..., Bais [/bib_ref] [bib_ref] Vascular endothelial growth factor stimulates bone repair by promoting angiogenesis and bone..., Street [/bib_ref] [bib_ref] Identification of novel gene expression in healing fracture callus tissue by DNA..., Khan [/bib_ref]. Thus, in our study, patients were grouped into these three phases according to the time between fracture and surgery and the main objective was to address the gene expression variations during early callus formation in patients that have suffered a hip fragility fracture. We showed that many inflammation-related genes have higher levels of expression until 3 days post-trauma while genes related to osteoblast and osteoclast activity increase at day 4 and thereafter.
Cytokine gene expression (IL1B, IL6 and TNF) was more pronounced during the first days after fracture, as described in younger individuals. Specifically, the decrease in IL6 expression levels was far more pronounced than what was observed with the other pro-inflammatory cytokines evaluated. Of interest, reports state that IL1B is upregulated in response to the fracture event but in IL1B knockout mice there was no change in callus formation and bone and cartilage matrix production [bib_ref] Action of IL-1beta during fracture healing, Lange [/bib_ref]. On the other hand, in the absence of TNF signaling there was a 2-4 days delay in chondrogenic differentiation and a 2-3 weeks delay in endochondral tissue resorption [bib_ref] Impaired fracture healing in the absence of TNF-alpha signaling: the role of..., Gerstenfeld [/bib_ref]. Regarding IL-6, studies in knockout mice have shown that there was a delay in callus formation and lower osteoclast density [bib_ref] Callus mineralization and maturation are delayed during fracture healing in interleukin-6 knockout..., Yang [/bib_ref]. Therefore, in accordance with our results, IL-6 appears to have a pivotal role in the early phase of fracture healing, probably through the increase in the pro-osteoclastogenic stimuli. Moreover, expression levels of TGFB, BMP2, BMP4, PDGFB and FGF2 was highest during the first 3 days post-fracture. The variation encountered in these elderly fragility fracture patients is similar to the findings obtained in animal models studies [bib_ref] The expression of cytokine activity by fracture callus, Einhorn [/bib_ref] and in healthy younger subjects [bib_ref] The values of some plasma components during the early phases of fracture..., Lyritis [/bib_ref] where the inflammatory phase occurs before day 3 post-fracture, being IL-6 a crucial player in this early phase of fracture healing.
Regarding the RANK-RANKL-OPG system, OPG expression diminished gradually after fracture, releasing the inhibitory signal for osteoclast differentiation. Concordantly, RANKL peaks at 4 th -7 th day after trauma, creating a stimulus for osteoclast differen- tiation from its precursors. Therefore, the ratio RANKL/OPG was high during days 4-7 post-fracture, not only due to an increase in RANKL but also to a decrease in OPG expression switching the balance to a pro-resorptive status, as described in a young mouse model [bib_ref] Expression of osteoprotegerin, receptor activator of NF-kappaB ligand (osteoprotegerin ligand) and related..., Kon [/bib_ref].
Concerning osteoblast differentiation and activity, it was observed that CBFA1/RUNX2 and OSX, two regulatory factors essential for its differentiation [bib_ref] Osteoblasts and bone formation, Caetano-Lopes [/bib_ref] , had a weak increase indicating the beginning of an initial osteogenic phase. CBFA1/RUNX2 increases from the initial phase of bone healing whereas OSX increases after 4 days post-fracture sustaining the evidence that OSX acts later than CBFA1/RUNX2 in osteoblast lineage commitment [bib_ref] Transcriptional regulation of bone formation by the osteoblastspecific transcription factor Osx, Zhang [/bib_ref]. On the other hand, ALP, a marker for osteoblast activity, decreases from early stages of bone healing. IL1B -interleukin-1b; IL6 -interleukin-6; TNF -tumor necrosis factor; TGFB1 -transforming growth factor b1; BMP -Bone morphogenetic protein; IGF1 -insulin growth factor-1; FGF2 -fibroblast growth factor 2; PDGFB -platelet derived growth factor b; OPG -osteoprotegerin; RANK -receptor activator of nuclear factor kB; RANKL -RANK ligand; CBFA1/RUNX2 -core binding factor a1/runt-related transcription factor 2; OSX -osterix; ALP -alkaline phosphatase; SOST -sclerostin; TRAP -tartrateresistant acid phosphatase; CTSK -cathepsin K; ITGB3 -subunit b3 of the integrin avb3; ATP -ATPase H + transporter. doi:10.1371/journal.pone.0016947.t002
This gene expression profile had been already observed in other studies [bib_ref] The values of some plasma components during the early phases of fracture..., Lyritis [/bib_ref] [bib_ref] Effect of fracture on bone turnover markers: a longitudinal study comparing marker..., Ivaska [/bib_ref] [bib_ref] Comparison in bone turnover markers during early healing of femoral neck fracture..., Ikegami [/bib_ref] and it is not entirely surprising, since this enzyme is involved in bone matrix production and our study is focused on the early changes related to fracture, in a stage where the formation of new bone matrix is still not occurring. Sclerostin is a protein produced by the osteocyte that inhibits canonical Wnt/b-catenin signaling, thus blocking osteoblast proliferation and differentiation. The contribution of this pathway to fracture healing depends on the function of b-catenin in different stages of fracture repair, namely in the commitment and regulation of osteoblasts [bib_ref] Beta-catenin signaling plays a disparate role in different phases of fracture repair:..., Chen [/bib_ref]. Only one study in young mice has addressed the levels of expression of sclerostin during fracture repair and they found that this protein was downregulated during the process [bib_ref] Distinct functionalities of bone morphogenetic protein antagonists during fracture healing in mice, Dean [/bib_ref]. In fact, our results, the first obtained in fragility fracture patients, showed that SOST expression decreases significantly from the beginning of the healing cascade, suggesting that there is an initial blockage of osteoblast proliferation and differentiation that is subsequently released over the period of fracture healing.
The role of the osteoclast in bone healing is somewhat controversial. Bone formation overcomes the loss of continuity and osteoclasts seem to play a role at a later phase, in the remodeling stage. Moreover, in a longitudinal study where the serum levels of biochemical markers associated with bone metabolism were assessed, the authors showed that the markers for bone resorption remained elevated up to four months after fracture [bib_ref] Effect of fracture on bone turnover markers: a longitudinal study comparing marker..., Ivaska [/bib_ref]. At gene expression level, we found that the osteoclast-specific genes TRAP, CTSK and ATP6V0D2 were significantly increased from day 8 onward after fracture, pointing to an activation of osteoclast function. In fact, the RANKL/OPG ratio is highest in group 2, whereas the CTSK values are increased in group 3, indicating that during 4-7 days after fracture, osteoclastogenesis stimulus was ongoing intensively whereas at day 8 and later osteoclasts containing cathepsin K had already been formed in relatively high numbers. The active role of osteoclast during the early phase of fracture healing was already described in sheep where it was proposed that these cells not only resorb bone but adjust the system, together with osteoblasts, in order to improve bone strength [bib_ref] Osteoclastic activity begins early and increases over the course of bone healing, Schell [/bib_ref].
Due to the fact that we are dealing with human subjects, the study had to have a cross-sectional design. Thus, it is not possible to rule out the intrinsic variability of different individuals. However, the statistical significance for many of the changes described seems to refute this. Besides, the RNA used was extracted from the site of fracture (trabecular bone) that not only has the bone cells that we are interested in, but also other cell types, such as adipocytes, bone marrow cells and cells infiltrating the tissues during the initial healing phase. However, the bone remodeling genes studied are relatively specific for bone cells and it is unlikely that this technical aspect represent a relevant confounding factor in our study.
Taken together, our results indicate that in patients with hip fragility fractures, the expression of inflammation-related genes is highest during the first days after fracture but from day 4 onwards there is a shift towards bone cell remodeling genes, suggesting that the machinery of bone healing is conserved in osteoporotic bone. In addition, the changes observed in IL-6 expression profile suggest that this pro-inflammatory cytokine plays a pivotal role in triggering the healing cascade. Moreover, sclerostin expression is quickly reduced after fracture and we hypothesize that this allows osteoblasts to escape from its inhibitory effect, thus promoting the expression of bone formation genes. Interestingly, RANKL expression is subsequently increased, generating the stimulus for osteoclast activity, as confirmed also by the later rise in the expression of the bone resorption-related genes. Our findings bring new insights for clarifying bone fracture healing process in osteoporotic patients. We propose that an initial inflammatory stimulus and a decrease in sclerostin-related effects are key events for an adequate fracture healing. Thus, in osteoporotic patients, locally promoting these events might provide promising medical interventions for accelerating fracture healing and reduce the rate of complications.
# Methods
## Sample collection
Patients that suffered a low-energy hip fracture and underwent total hip replacement surgery at the Orthopedic Department of Hospital de Santa Maria were consecutively recruited for this study from 2007 until 2009. Epidemiological and clinical data such as age, gender and days between the fracture and surgery were collected. Patients with other metabolic bone diseases and with bone metastases were excluded.
Written informed consent was obtained from all patients and the study was conducted in accordance with the ethical principles for medical research involving human subjects expressed in the Declaration of Helsinki, as amended in Edinburgh (2000), and was approved by Santa Maia Hospital Ethics Committee.
According to the time between fracture and surgery, patients were divided in three groups: those who had hip replacement surgery between zero and three days after fracture (group 1), four and seven days after fracture (group 2) or eight or more days after fracture (group 3).
After the medical procedure, the femoral epiphyses were snapfrozen at 280uC.
## Rna extraction
Without defrosting the bone specimen, small trabecular bone pieces were collected from the site of fracture and pulverized using a mortar and pestle. RNA was then extracted using TRIzol reagent (Invitrogen, UK) with proteinase K (Bioline, UK) digestion [bib_ref] Proteinase K added to the extraction procedure markedly increases RNA yield from..., Egyhazi [/bib_ref] to better dissolve the dense extracellular matrix.
The procedure used was a modified version of the protocol described by Ireland [bib_ref] Analysis of Gene Expression in Bone by Quantitative RT-PCR, Ireland [/bib_ref]. Briefly, 80mg of bone powder was placed in TRIzol reagent and homogenized. Lipids were solubilized with 0.2 volumes of chloroform and the fraction containing RNA was preserved. Proteinase K digestion (3.3mg proteinase K/mg bone) was performed at 55uC. RNA was precipitated with 1 volume of icecold isopropyl alcohol. RNA pellet was dissolved in RNase/DNasefree water. As this method leaves residual chemical contaminants, RNA was cleaned using a commercial kit (RNeasy mini kit, Qiagen, Germany) and genomic DNA contaminants were removed by Quantitative reverse transcription-polymerase chain reaction (PCR)
Reverse transcription cDNA synthesis was performed on 60ng of RNA from each sample using the DyNAmo cDNA synthesis kit (Finnzymes, Finland) and 300ng of random hexamer primers according to the manufacturer's instructions.
Each cDNA template (3ng/ml) was amplified in duplicate with DyNAmo Flash SYBR green qPCR kit (Finnzymes, Finland) on a Rotor-Gene thermocycler (Qiagen, Germany) according to the manufacturer's instructions. Reactions were incubated at 50uC for 2 minutes and at 95uC for 7 minutes, followed by denaturation at 95uC for 10 seconds and annealing/extension at 60uC for 10 seconds. The reactions were validated by the presence of a single peak in the melt curve analysis.
Primers for the housekeeping and target genes [fig_ref] Table 3: Real time PCR primer sequences of the genes studied [/fig_ref] were designed using the software Probefinder (http://qpcr.probefinder. com, Roche, Switzerland) in order to anneal in separate exons preventing amplification of contaminating genomic DNA.
Real time PCR results were analyzed using the standard curve analysis. The cycle threshold (C T ) is defined as the number of cycles required for the fluorescent signal to cross the threshold and exceed the background level. The efficiency of the PCR should be 100%, meaning that for each cycle the amount of product doubles. A good reaction should have an efficiency of 90-100%, which corresponds to a slope between 23.58 and 23.10. The conversion of the C T value in relative expression levels was performed with the slope and the Y intersect extracted from the standard curve and applying the equation 10 (Y intersect-CT/slope) [bib_ref] Real-time PCR for mRNA quantitation, Wong [/bib_ref] [bib_ref] Real time quantitative PCR, Heid [/bib_ref]. The values obtained were normalized with the housekeeping genes b-2microglobulin (B2M) and phosphomannomutase 1 (PMM1).
# Statistical analysis
To define the exposure variable, patients were divided, according to the number of days between the event of fracture and the surgery, in three groups. The distributions of continuous variables were compared between groups using either analysis of variance (ANOVA), for normally-distributed characteristics, or Kruskal-Wallis H test, for distributions with significant deviation from normality (according to Shapiro-Wilk test). For nominal variables, chi-squared test was used. Significance level was set at 0.05.
Statistical analysis was performed using the Statistical Package for Social Sciences manager software, version 17.0 (SPSS, Inc, Chicago, IL, USA).
[fig] Figure 1: Relative expression of inflammation and growth factors genes grouped according to the time between the event of fracture and the surgery. Each gene was normalized to the expression of the housekeeping genes B2M and PMM1. *p-value,0.05 for comparisons between the three groups. (Points represent median values). IL1B -interleukin-1b; IL6 -interleukin-6; TNF -tumor necrosis factor; IGF1 -insulin growth factor-1 ; FGF2 -fibroblast growth factor 2 ; PDGFB -platelet derived growth factor b; BMP -Bone morphogenetic protein; TGFB1transforming growth factor b1. doi:10.1371/journal.pone.0016947.g001 [/fig]
[fig] Figure 2: Relative expression of bone metabolism-related genes divided according to the time between the event of fracture and the surgery. RANK, RANKL and OPG (A), osteoblast (B), osteocyte (C) and osteoclast-specific genes (D) were studied in the three study groups. Each gene was normalized to the expression of the housekeeping genes B2M and PMM1. *p-value,0.05 for comparisons between the three groups. (Points represent median values). RANK -receptor activator of nuclear factor kB; RANKL -RANK ligand; OPG -osteoprotegerin; CBFA1/RUNX2 -core binding factor a1/runt-related transcription factor 2; OSX -osterix; ALP -alkaline phosphatase; SOST -sclerostin; ITGB3 -subunit b3 of the integrin avb3; TRAP -tartrate-resistant acid phosphatase; ATP -ATPase H + transporter; CTSK -cathepsin K. doi:10.1371/journal.pone.0016947.g002 [/fig]
[table] Table 2: Comparison between the relative gene expression levels of patients submitted to hip replacement surgery due to lowenergy fracture in relation to the days between fracture and surgery. [/table]
[table] Table 3: Real time PCR primer sequences of the genes studied.B2M -b2-microglobulin; PMM1 -phosphomannomutase 1; IL1B -interleukin-1b; IL6 -interleukin-6; TNF -tumor necrosis factor; TGFB1 -transforming growth factor b1; BMP -Bone morphogenetic protein; IGF1 -insulin growth factor-1; FGF2 -fibroblast growth factor 2; PDGFB -platelet derived growth factor b; OPG -osteoprotegerin; RANK -receptor activator of nuclear factor kB; RANKL -RANK ligand; CBFA1/RUNX2 -core binding factor a1/runt-related transcription factor 2; OSX -osterix; ALP -alkaline phosphatase; SOSTsclerostin; TRAP -tartrate-resistant acid phosphatase; CTSK -cathepsin K; ITGB3 -subunit b3 of the integrin avb3; ATP -ATPase H + transporter. doi:10.1371/journal.pone.0016947.t003 [/table]
|
Developing and evaluating multimedia information resources to improve engagement of children, adolescents, and their parents with trials (TRECA study): Study protocol for a series of linked randomised controlled trials
Background: Randomised controlled trials are widely established as the best method for testing health interventions whilst minimising bias. However, recruitment and subsequent retention of children and adolescents in healthcare trials is challenging. Participant information sheets are often lengthy and difficult to read and understand. Presenting key information using multimedia may help to overcome these limitations and better support young people and their parents in deciding whether to participate in a clinical trial. Methods: The TRECA (TRials Engagement in Children and Adolescents) study has two phases. The first phase involves a qualitative study with children and adolescents and their parents to inform the development of multimedia information resources and iterative user testing to refine the resources. The second phase will embed the use of the multimedia information resources into six host trials in the United Kingdom. Patients and parents approached to participate in the host trials will be randomly allocated to either use the multimedia information resource in conjunction with standard participant information sheets, the multimedia information resource alone, or the standard participant information sheets alone. The primary outcome will be the effect of the multimedia information resources on recruitment into trials. Other outcomes measured include the effect of multimedia information resources on retention of participants into the host trials and the impact on family members' decision-making processes, when compared to standard participant information sheets alone.Discussion: This study will inform whether multimedia information resources, when developed using participatory design principles, are able to increase recruitment and retention of children and adolescents into trials. There is also the potential for patients to make better informed decisions through the use of multimedia information resources. The multimedia information resources also have the potential to assist with providing information on other healthcare decisions outside of clinical trials.
# Background
The effectiveness and safety of healthcare interventions is best determined through randomised controlled trials [bib_ref] Methods to improve recruitment to randomised controlled trials: cochrane systematic review and..., Treweek [/bib_ref] [bib_ref] Barriers to participation in randomised controlled trials: a systematic review, Ross [/bib_ref]. However, major barriers to the successful conduct and outcome of clinical trials are levels of recruitment and retention. In the UK, only a small proportion of trials actually recruit successfully to time and target [bib_ref] What influences recruitment to randomised controlled trials? A review of trials funded..., Mcdonald [/bib_ref] [bib_ref] Short report: how often do UK primary care trials face recruitment delays?, Bower [/bib_ref] [bib_ref] Practical considerations for conducting dental clinical trials in primary care, Martin-Kerry [/bib_ref] [bib_ref] Patterns of patient enrollment in randomized controlled trials, Haidich [/bib_ref]. Furthermore, in practice, it remains relatively uncommon for a patient to participate in a clinical trial despite the development of National Health Service organisational structures that facilitate the integration of clinical research and patient care [bib_ref] A healthy nation: strengthening child health research in the UK, Modi [/bib_ref] , for example, the Clinical Research Networks. Inadequate recruitment or retention have implications not only for conclusive results but also external validity and generalisability of the trial findings [bib_ref] Threats to applicability of randomised trials: exclusions and selective participation, Britton [/bib_ref].
There is now international recognition of the importance of paediatric clinical trials to inform healthcare decisions for children and adolescents [bib_ref] Clinical trials in children, Joseph [/bib_ref] [bib_ref] Children's views on their involvement in clinical research, Chappuy [/bib_ref] [bib_ref] The development of medicines for children. Part of a series on Pediatric..., Rocchi [/bib_ref] [bib_ref] Children are not just small adults: the urgent need for high-quality trial..., Klassen [/bib_ref]. High quality trials involving children are essential to ensure that medication and treatments used in children are effective and safe [bib_ref] Clinical trials in children, Joseph [/bib_ref] [bib_ref] The development of medicines for children. Part of a series on Pediatric..., Rocchi [/bib_ref] [bib_ref] Children are not just small adults: the urgent need for high-quality trial..., Klassen [/bib_ref]. The lack of successful clinical trials leads to many healthcare decisions for children and adolescents being made with inadequate evidence, including evidence extrapolated from trials involving adults [bib_ref] Children are not just small adults: the urgent need for high-quality trial..., Klassen [/bib_ref]. Only 6% of recently registered clinical trials in the UK involved children [bib_ref] A healthy nation: strengthening child health research in the UK, Modi [/bib_ref]. The publication rate of trials in adults has almost doubled over a 20-year period, a rate increase that is around six times higher than for paediatric trials over the same period [bib_ref] Child vs adult randomized controlled trials in specialist journals: a citation analysis..., Cohen [/bib_ref]. In the past, this low rate of paediatric trials was thought to be mainly due to a concern for the vulnerability of children leading to a reluctance by clinicians to undertake clinical trials with young children [bib_ref] A healthy nation: strengthening child health research in the UK, Modi [/bib_ref] [bib_ref] Pediatricians' attitudes toward randomized controlled trials involving children, Caldwell [/bib_ref]. Nevertheless, high rates of patient or parent refusal have also been identified as a key barrier for successful completion of these trials [bib_ref] Recruitment to pharmacokinetic research in children: what are the strategies that improve..., Menzies [/bib_ref] , although a recent study has shown lower refusal rates for paediatric trials involving therapeutic drugs [bib_ref] Recruitment in pediatric clinical research was influenced by study characteristics and pediatricians'..., Kaguelidou [/bib_ref].
A potential barrier to recruitment and retention is the information provided to potential trial participants [bib_ref] Barriers to participation in randomised controlled trials: a systematic review, Ross [/bib_ref] [bib_ref] Parents' agendas in paediatric clinical trial recruitment are different from researchers' and..., Woolfall [/bib_ref] [bib_ref] Factors that influence parental decisions to participate in clinical research: consenters vs..., Hoberman [/bib_ref]. How children and parents make decisions regarding participation in research and what information is important to them remain areas of uncertainty [bib_ref] Parents' agendas in paediatric clinical trial recruitment are different from researchers' and..., Woolfall [/bib_ref] [bib_ref] How do children and parents make decisions about pediatric clinical research?, Varma [/bib_ref] [bib_ref] Why do parents enroll their children in research: a narrative synthesis, Fisher [/bib_ref]. Conventionally, participant information about a trial is provided in printed form. These documents should be understandable to potential trial participants and assist their decision-making [bib_ref] What do our patients understand about their trial participation? Assessing patients' understanding..., Behrendt [/bib_ref]. However, the format of this information has received recurrent criticism, notably for being too long, difficult and technical [bib_ref] Standard 1: consent and recruitment, Caldwell [/bib_ref] [bib_ref] Readability of pediatric biomedical research informed consent forms, Tarnowski [/bib_ref] [bib_ref] Are research participants truly informed? Readability of informed consent forms used in..., Ogloff [/bib_ref] [bib_ref] Improving informed consent: suggestions from parents of children with leukemia, Eder [/bib_ref]. Furthermore, the content of trial sheets is mostly guided by regulatory agencies and can be inconsistent with what patients want to know [bib_ref] Parents' agendas in paediatric clinical trial recruitment are different from researchers' and..., Woolfall [/bib_ref] [bib_ref] Standard 1: consent and recruitment, Caldwell [/bib_ref] [bib_ref] A goodness-of-fit approach to informed consent for pediatric intervention research, Masty [/bib_ref]. A number of studies report that trial participants do not understand information contained within participant information sheets (PISs) [bib_ref] Hello, hello-it's English I speak!": a qualitative exploration of patients' understanding of..., Stead [/bib_ref] [bib_ref] Using digital multimedia to improve parents' and children's understanding of clinical trials, Tait [/bib_ref] and that the information can be very wordy and overwhelming [bib_ref] Standard 1: consent and recruitment, Caldwell [/bib_ref] [bib_ref] Communication about children's clinical trials as observed and experienced: qualitative study of..., Shilling [/bib_ref]. Potential participants who have lower levels of literacy are most likely to be affected by this [bib_ref] Digital multimedia: a new approach for informed consent?, Tait [/bib_ref]. Furthermore, good graphic design, such as a structure that aids navigation of the information and visual appeal to invite and engage the reader, is often lacking in PISs. For example, written information should inform a decision about participation, but may act more as a prospectus for the trial and as a contract between researchers and the participant [bib_ref] Do informed consent documents for cancer trials do what they should? A..., Armstrong [/bib_ref]. Re-writing, redesigning and user-testing of trial information can produce an understandable and preferred resource [bib_ref] Performance-based readability testing of participant information for a Phase 3 IVF trial, Knapp [/bib_ref] [bib_ref] Performance-based readability testing of participant materials for a phase I trial: TGN1412, Knapp [/bib_ref] [bib_ref] Can user testing of a clinical trial patient information sheet make it..., Knapp [/bib_ref].
An alternative for providing information to potential trial participants is through the use of a multimedia information (MMI) resource [bib_ref] Digital multimedia: a new approach for informed consent?, Tait [/bib_ref] [bib_ref] A randomised controlled study of an audiovisual patient information intervention on informed..., Hutchison [/bib_ref] [bib_ref] Innovating information-delivery for potential clinical trials participants. What do patients want from..., Shneerson [/bib_ref] , which presents key information using a combination of video, animation, text and audio through a website. However, research is needed to identify and evaluate different ways of presenting MMI about research to children and parents [bib_ref] Standard 1: consent and recruitment, Caldwell [/bib_ref]. Multimedia presentation can be understood through reading, listening and watching, and allows people with different preferences to use the resource effectively [bib_ref] Digital multimedia: a new approach for informed consent?, Tait [/bib_ref]. MMI resources can contain all key information that would be found in a written participant information about the trial but focusing on information deemed important for children, adolescents and their parents [bib_ref] Parents' agendas in paediatric clinical trial recruitment are different from researchers' and..., Woolfall [/bib_ref] [bib_ref] A randomised controlled study of an audiovisual patient information intervention on informed..., Hutchison [/bib_ref] when deciding whether to participate in a trial. Furthermore, MMI resources can enable the patient to select the order in which they access the information and allows people with different preferences to use the resource more effectively. Finally people's familiarity with websites and the frequency of their use means that MMI presented on a computer (or smartphone) may now be used intuitively and easily by most people.
In educational settings, it has been estimated that individuals will remember approximately 10% of what they read, 20% of what is heard, 30% if they can visualise and hear the information, and 50% if they observe someone doing something with an explanation [bib_ref] Instructional animation versus static pictures: a metaanalysis, Höffler [/bib_ref]. Multimedia, which involves using more than one medium of expression or communication, has been shown to be at least as effective as printed information [bib_ref] Providing information on metered dose inhaler technique: is multimedia as effective as..., Savage [/bib_ref] and often more effective in informing people [bib_ref] Interactive method of informing patients of the risks of intravenous contrast media, Hopper [/bib_ref] [bib_ref] Internetenabled interactive multimedia asthma education program: a randomized trial, Krishna [/bib_ref] [bib_ref] Comparative analysis of print and multimedia health materials: a review of the..., Wilson [/bib_ref]. MMI about medical procedures can improve patient knowledge [bib_ref] Interventions to improve patient comprehension in informed consent for medical and surgical..., Schenker [/bib_ref] [bib_ref] Interventions to promote informed consent for patients undergoing surgical and other invasive..., Kinnersley [/bib_ref] , but some have had variable impact on participant understanding [bib_ref] Improving understanding in the research informed consent process: a systematic review of..., Nishimura [/bib_ref]. However, a recent trial of children and adolescents undergoing endoscopy showed that presentation of information in electronic format produced more certain consent decisions, compared to printed information [bib_ref] A novel method to enhance informed consent: a prospective and randomised trial..., Friedlander [/bib_ref]. There is limited information about the effect of MMI for patients on trial recruitment rates [bib_ref] A randomised controlled study of an audiovisual patient information intervention on informed..., Hutchison [/bib_ref] [bib_ref] Innovating information-delivery for potential clinical trials participants. What do patients want from..., Shneerson [/bib_ref] , although a relevant study is underway examining the use of MMI in trials recruiting adults in the UK [bib_ref] A multimedia intervention to enhance recruitment to clinical trials in primary care..., Bower [/bib_ref] [bib_ref] Systematic techniques for assisting recruitment to trials (START): study protocol for embedded,..., Rick [/bib_ref]. Furthermore, studies using multimedia to improve children's understanding of clinical trials have demonstrated improved understanding when compared with those using traditional paper-based information [bib_ref] Using digital multimedia to improve parents' and children's understanding of clinical trials, Tait [/bib_ref]. There are a number of paediatric studies, particularly in the United States, using video for informed consent within a website format; however, the structure of this information presentation is restrictive, requiring users to view the video from start to finish [bib_ref] Use of mobile devices and the internet for multimedia informed consent delivery..., Blake [/bib_ref]. MMI resources have the potential to inform and engage potential trial participants in ways that printed information can struggle to do (see [fig_ref] Figure 1: Possible pathway of effectiveness of multimedia information [/fig_ref] for possible effectiveness pathway).
The TRECA (TRials Engagement in Children and Adolescents) study will develop two MMI resources through the use of participatory design involving individual and focus group interviews with children and adolescents with long-term health conditions and their parents . This will ensure that the MMI resources are developed to meet the needs and preferences of potential end users. The study will examine whether providing key information about clinical trials through the use of MMI resources increases recruitment and retention and enables better decision-making of children and adolescents participating in trials in the UK.
# Methods/design
Aims The aims of TRECA are to evaluate the potential for MMI resources to improve the quality of decision-making about participation in healthcare trials involving children and adolescents with long-term health conditions, and to assess the impact of MMI resources on trial recruitment and retention and the quality of decision-making.
The objectives of the TRECA study are: To involve children and adolescents with long-term health conditions, their parents and trial researchers and clinicians, in the development of two MMI resources for use when children and adolescents are being asked to consider participation in a healthcare trial.
To obtain and analyse qualitative data from focus groups with members of key stakeholder groups (i.e. children and adolescents with long-term health conditions, parents, clinicians, trial managers) to ensure that the content and format of the MMI resources reflect their needs and preferences.
To user-test the MMI resources with children and adolescents (and parents) to test their ability to inform potential users.
To evaluate the MMI resources in a series of trials embedded within host trials in the UK, testing their effect on recruitment and retention rates and decision-making by comparing the effect of providing standard written participant information with provision of the MMI resource either in addition to the standard written participant information or the provision of the MMI resource alone.
## Design overview
The study is divided into two phases, namely Phase one (development) and Phase two (evaluation). The study design is shown in [fig_ref] Figure 2: TRECA study design [/fig_ref]. The development phase (Phase one) involves qualitative methods followed by user testing, aiming to produce two MMI resources (refer to [fig_ref] Figure 3: Development of the multimedia information [/fig_ref] for development stages of the MMI resources), with generic elements relevant to any trial involving children and adolescents and a template for the addition of specific content for individual host healthcare trials. In the evaluation phase (Phase two), the two MMI resources will be tested in a series of embedded trials hosted within healthcare trials (refer to [fig_ref] Figure 4: Phase two study design [/fig_ref] for the Phase two study design), following the addition of a small amount of host trialspecific content to the MMI resource. The MMI resources will be tested for their impact on decisions about trial participation taken by children and adolescents and/or parents and behaviours (rates of recruitment to, and retention in, the host trials). The SPIRIT checklist describing the protocol is available as Additional file 1.
Phase one: development MMI resource development
Two MMI resources will be developed [fig_ref] Figure 3: Development of the multimedia information [/fig_ref] to be tested as an adjunct to, or replacement for, printed written PISs for potential child and adolescent trial participants, and their parents. The overall goal is that young potential participants and their parents will use the MMI resources to inform their decisions about entry and ongoing participation in the host trials. Both MMI resources will have generic trial information (e.g. on randomisation, study withdrawal, confidentiality, altruism and personal benefit) and a section for trial-specific information (e.g. trial purpose, intervention, number of appointments and length). The MMI resources are designed to be used by children, adolescents and their parents to assist them with making an informed decision about whether to participate in a trial.
The MMI resources will be commissioned from a specialist commercial supplier, so that their appearance and functions are professional, sophisticated and contemporary. One of the MMI resources will be intended for use by children and their parents, and the other by adolescents and their parents. Learning from a previous study looking at the development of MMI resources for adults deciding whether to participate in healthcare trials will be incorporated into the design of the current MMI resources [bib_ref] Systematic techniques for assisting recruitment to trials (START): study protocol for embedded,..., Rick [/bib_ref] [bib_ref] Interventions to improve recruitment and retention in clinical trials: a survey and..., Bower [/bib_ref]. The distinct content and format of the two MMI resources will be informed by the qualitative study. The qualitative and user-testing data will also inform the topics included in each MMI resource, and the mixture of media (text, animation, video and infographics). We will also explore the potential for the MMI resources to be interactive, for example, allowing children and adolescents to post questions to the host trial research team and take mini-quizzes.
Participatory design: qualitative study to identify the information needs and preferences of potential users of MMI resources Individual and focus group interviews will be undertaken with (1) children and adolescents (aged 9-11, 12-14 and 15-17 years) with long-term health conditions ; (2) parents of children and adolescents with long-term health conditions; and (3) researchers and clinicians who have experience with working in children's clinical trials. Sampling will be purposive and will aim to achieve Long-term health conditions Long-term health conditions cover a wide range of conditions, that can be life-long, slowly deteriorating, potentially curable and with a variable course. These conditions include (list not exhaustive):
[formula] - Diabetes - Asthma - Juvenile arthritis - Cancer - Cystic fibrosis - Muscular dystrophy [/formula]
- Early manifestation of a condition that may become chronic (e.g. acne)
Conditions which would not be included in this definition of long-term health conditions are vaccinations and those which are treated as emergency or acute conditions variation with regard to age, sex, long-term health condition, trial experience, ethnicity, and sociodemographics of participants. We plan to interview 6-10 people from each participant group, although we will increase the number as required until new data cease contributing to the analysis. For the clinician and researcher focus groups we will sample to ensure a range of roles are represented. The age split for children and adolescents is based approximately on children's likely capacity for a role in assent/consent decisions, using an established three-part categorisation.
Participants will be recruited through a number of mechanisms, including Alder Hey Children's Hospital, Generation R Young People Advisory Groups (groups with membership of children and adolescents who advise on the design of research that involves children and adolescents in the National Health Service) and three patient interest groups: PORT (Paediatric Oncology Reference Team), the Invisible Illness group and the UK Juvenileonset Systemic Lupus Erythematosus Study Group.
Data will be collected during two rounds of individual and focus group interviews held approximately 2-3 months apart. Focus groups are preferred because they will enable discussion of ideas amongst participants, although the wishes of those who would opt for individual interviews will be accommodated. The first round will take place before the MMI resources have been designed in order to inform their content, style and delivery. The second round will take place after draft MMI resources have been produced in order to explore participants' views on these and their suggestions for amendment.
Semi-structured individual and focus group interviews will be topic guided and focus on (1) preferences for information about research and (2) preferences for content, style and delivery of MMI resources. Participants will be prompted to discuss items of information adapted from a study [bib_ref] What potential research participants want to know about research: a systematic review, Kirkby [/bib_ref] of adult trial participants' information needs, and items of information identified from a study of children and adolescents' trial [bib_ref] Young people's experiences of participation in clinical trials: reasons for taking part, Luchtenberg [/bib_ref]. Items will include why the trial is being done, whether the child or adolescent has to take part or not, whether the participant will benefit personally from taking part, and who will know the child or adolescent is taking part.
Participants will also be prompted about their preferences for how the MMI resource looks and functions as informed by the website industry Webby Awards criteria, including general needs and preferences for MMI content, potential for interactivity (e.g. question posting, quizzes) and how the information should be presented (e.g. whether people should receive the printed information sheet before or after the MMI resource).
A pilot focus group will comprise younger children (6-8 years old) and involve showing them the developed MMI resource to explore their perspectives on whether the MMI resource is easy for them to use and understand the information provided.
Where possible, the two rounds of data collection will include the same participants to facilitate respondent validation of the ongoing analysis, with some replacement for the second round, when required. Each focus group will have between four and ten participants and be approximately 90 minutes in duration. Where participants prefer, they will be offered an individual interview either face-to-face or via Skype.
All individual and focus group interviews will be audio-recorded and transcribed verbatim.
## Qualitative data analysis to inform the development of the mmi resources
Analysis of qualitative data will be thematic and focus on identifying what information is important for children, adolescents and their parents when deciding to participate in a trial and their thoughts on the design aspects of the MMI resource. Line by line coding will be undertaken to identify key themes within each transcript. Analysis will be led by one researcher with guidance provided by a qualitative research expert (BY) with regular meetings to discuss data interpretation. The codes will be grouped into themes and organised using NVivo version 10 software. Following the principles of participatory design, the findings on the needs and preferences of potential users will inform the development of two prototype MMI resources. We will work with patient and public involvement (PPI) members, seeking their thoughts on how to apply the findings into the development and the design of the MMI resources. The MMI resources will be informed directly based on the data obtained in the individual and focus group interviews. Where the views of the different participant groups (children/adolescents, parents and clinicians) diverge markedly, we will focus on the needs and preferences of children and adolescents.
## User testing
Once the two MMI resource prototypes have been developed, the MMI resources will undergo user testing. We will adapt the method of user testing employed in the development of printed information, including PISs for trials [bib_ref] Can user testing of a clinical trial patient information sheet make it..., Knapp [/bib_ref]. This involves an iterative process with changes being made to the MMI resource in response to the data received. We anticipate having two rounds of user testing, with changes made to the MMI resources, as required, after the first round. User testing involves small participant samples to generate quantitative data, in which data patterns are interpreted to identify any problems with a piece of information that may be responsive to change. Participants will be observed using the MMI resources and then participate in brief structured interviews about the MMI resources. This will involve testing participants' knowledge and understanding of the information in the resources and asking them to indicate where in the resource this information is located. The generated quantitative data are indicative, not definitive, and not analysed statistically. The emphasis is on identifying aspects of the information resource that might hinder understanding.
## User testing sampling
Participants for the user testing will be recruited via a number of primary and secondary schools in Yorkshire, UK, and aim for a diverse sample of participants in terms of ethnicity, socioeconomic status and levels of academic ability, including some participants who have English as a second language. Individuals who participated in the qualitative study will not be eligible for the user testing, as user testing produces its most valid results with participants who are not already familiar with the intervention being tested [bib_ref] Performance-based readability testing of participant information for a Phase 3 IVF trial, Knapp [/bib_ref].
We will use the conventional sampling method in user testing, namely rounds of 20 participants [bib_ref] Cognitive and usability engineering methods for the evaluation of clinical information systems, Kushniruk [/bib_ref]. In the testing of the MMI resource intended for older children and adolescents, the samples of 20 will comprise a mix of parents, children and adolescents (aiming to achieve a spread of ages across the 12-17 years range). For the less complex MMI resource, rounds will comprise 10 parents and 10 younger children, aiming to achieve a spread across ages 6-11 years, with parents and children using the MMI resource together.
For each MMI resource, the two rounds of user testing will use different participants, to remove any effect of prior learning. We will ensure that the samples in the two rounds have similar profiles in terms of age and sex, to better indicate problems in the MMI resources requiring change.
## User testing data analysis
The data derived from user testing interviews will be analysed quantitatively, although data are indicative (e.g. 80% cannot find a particular piece of information). Each item on the questionnaire will derive the following scoring criteria: finding (found, found with difficulty (i.e. found but only after a set time, usually more than 3 minutes) or not found); understanding (understood, understood with difficulty (i.e. understood but only after question rewording or repetition) or not understood).
## Phase two: evaluation of mmi resources embedded trials of mmi resources within host clinical trialsdesign
The effectiveness of the two MMI resources will be evaluated in a series of trials embedded within six host trials in the UK (see [fig_ref] Figure 4: Phase two study design [/fig_ref] for Phase two design) that are recruiting children and adolescents with long-term health conditions, using methods we have developed previously [bib_ref] Systematic techniques for assisting recruitment to trials (START): study protocol for embedded,..., Rick [/bib_ref]. That is, participants in each host trial will be allocated randomly to one of two or more different intervention groups. The embedded trial (aka nested trial or 'trial within a trial') will be run with potential participants in each host trialthese people will be allocated randomly to receive either the MMI resource plus the printed information, the MMI resource alone, or the printed information alone, to evaluate their relative effects on recruitment rates (and the secondary outcomes).
The objective is to test the effects of the MMI resources on cognition and behaviour. The primary outcome will be whether rates of recruitment to the host trials are increased. We will also test (1) whether individuals who see the MMI resource(s) make a more informed decision about trial participation (or not); (2) whether rates of retention in the host clinical trials are increased; and (3) whether individuals are more satisfied with the process of consent or assent.
Recruitment of host trials will occur through the National Institute for Health Research Clinical Research Network, funding bodies and through investigator networks. In previous qualitative work [bib_ref] Trials within trials? Researcher, funder and ethical perspectives on the practicality and..., Graffy [/bib_ref] , we have identified that, while trialists welcome the idea of embedding trials of recruitment interventions in their studies, these need to be compatible with the host trial design and not impose an extra workload. In the embedded trials, allocation to groups will be achieved by random number generator. Particularly, in trials in which we use individual randomisation to the MMI resources, it will be practicably very difficult to achieve concealment of randomisation. Trials will use individual or cluster randomisation. Masking of the allocation at outcome measurement will not be possible since patients cannot be masked to the information format they will receive but, as they will be unaware of the embedded information trial, a lack of masking will not bias their responses or decisions.
## Trial eligibility criteria
Trials will be eligible for inclusion in TRECA if they are recruiting within the UK and involve testing an intervention with children and adolescents who have a long-term health condition. Within each embedded trial, participants will be children and adolescents being asked to participate in the host healthcare trial and/or their parents. This is critical, as it means that the host trial and the embedded trial have different sample sizes. For the host trial the sample comprises those children and adolescents/parents agreeing to participate; for the embedded trial of the MMI resources the sample comprises those asked to participate. In some trials, the number asked to participate is much larger, often more than double the host trial sample size.
Eligible trials should ideally have a sufficient sample size to detect a difference between groups in the embedded trial; trials will be using only printed or video participant information materials as standard (i.e. not already including an MMI resource), and will be recruiting at least some children and adolescents who have the potential to contribute to a decision about consent or assent to participation in the trial. Trials will not be included if they are only recruiting children too young to understand an MMI resource (e.g. children aged under 5 years), or only children with intellectual impairment such that understanding or use of the MMI resource is not possible.
We will aim to ensure that eligible trials vary with respect to the long-term health condition (using PRISM study criteria), the age of children and adolescents being recruited into the trial, host trials unit, the type of intervention (e.g. pharmaceutical, physical therapy, psychological), and the way that the MMI resource will be presented to patients (e.g. parent and child viewing together versus adolescent viewing separately to parent).
## Method of embedding mmi into trials
Each of the embedded trials will use a three-arm design, in which individuals will receive the standard written trial PIS alone, the standard PIS in addition to the MMI resource, or the MMI resource alone [fig_ref] Figure 2: TRECA study design [/fig_ref]. We will consider making the written PIS available via the MMI resource, for example, by a link within the MMI resource to read or print the PIS document. This would have the advantage of being more efficient, allowing people to access both the MMI resource and the PIS via the computer. However, we will consider important practical concerns such as text readability on screens and participant preferences, and so will seek the opinions of participants in the focus groups during the development study phase. We will also measure the number of page views and the frequency which individual elements of the MMI resource are viewed.
Patients allocated to the control arm of the embedded trial will be given the printed PIS only (as is usual). Those allocated to one of the intervention arms will receive either the MMI resource alone or the printed PIS and the MMI resource(s). We will not determine the order in which participants access the PIS and MMI resource (for those participants who receive both) and will leave this for the host trial to determine, to suit the practical demands of patient recruitment. However, we will ask the host trial to record the order in which participants are given and access the PIS/MMI resource, and report this observation in the report of each embedded trial. The MMI resources will be presented in the clinic on a computer or dedicated tablet computer. Participants will also be able to access the MMI resources at home (via smartphone or a tablet or PC) via a link that is emailed to them. In some circumstances, home viewing will take place before the patient's decision on clinical trial participation has been taken. Some patients will also want to be given access to the MMI resource after they have decided to take part in the host healthcare trial, just as they would if they had been given standard printed information only.
## Outcome measures
The primary outcome will be the rates of recruitment to each host trial. We will calculate the proportion of patients who agree to participate from the total number approached, for each arm of the embedded trial. This study is investigating whether MMI resources improve the quality of decision-making, related to individuals being more informed about the trial. Improved decisionmaking may have no impact on trial recruitment rates, although it is also possible that it could either increase or decrease rates as a result of individuals being more informed. It is therefore important that we also measure secondary outcomes of retention in the trials, and quality of decision-making. We will measure retention by obtaining data on the number and timing of drop outs from each host trial. The quality of decision-making by potential host trial participants will be measured through the completion by children, adolescents and parents (as relevant) of a decisional scale, adapted from one used within the REFORM trial [bib_ref] Preliminary analysis of a new measure of quality of patient decision making..., Knapp [/bib_ref] and drawing conceptually on the SURE [bib_ref] Validation of SURE, a four-item clinical checklist for detecting decisional conflict in..., Ferron Parayre [/bib_ref] and DelibeRATE scales [bib_ref] Making a decision about trial participation: the feasibility of measuring deliberation during..., Gillies [/bib_ref] [bib_ref] Increasing readiness to decide and strengthening behavioral intentions: Evaluating the impact of..., Sivell [/bib_ref]. We will report study results in line with published guidelines [bib_ref] Guidelines for reporting embedded recruitment trials, Madurasinghe [/bib_ref].
## Sampling considerations
The projected effect of the MMI resources will vary in size according to the setting of the host trial, the background recruitment rate, and the intervention being tested in the host trial. This makes it difficult to establish a sample size calculation for each of the six embedded trials. The effectiveness of the MMI resources is being assessed against three outcome measures, namely host trial recruitment rate, quality of decision-making on participation (or not) and host trial retention rate. Results from each embedded trial will be combined in a prospective meta-analysis for decision scores and recruitment rate data from all six host trials participating in the TRECA study.
## Ppi
The Investigators and research team have a strong commitment to PPI in the TRECA Study. TRECA has a Patient and Parent Advisory Group which will play a key role in reviewing and providing input into documentation used in the various stages of the study, including topic guides and prototype MMI resources. The Patient and Parent Advisory Group will also participate in the piloting of user testing questionnaires to ensure that the question wording and length are appropriate. Two members of the Patient and Parent Advisory Group are also members of the TRECA Study Advisory Group and one or two PPI members will be involved in co-facilitation of the second set of focus groups.
# Discussion
Whilst trials have been undertaken for many decades in medicine, limited data is available about the decisionmaking processes for trial participants and which information is important for them when considering whether to participate [bib_ref] Barriers to participation in randomised controlled trials: a systematic review, Ross [/bib_ref] [bib_ref] What do our patients understand about their trial participation? Assessing patients' understanding..., Behrendt [/bib_ref] [bib_ref] What potential research participants want to know about research: a systematic review, Kirkby [/bib_ref]. Studies that have examined the information needed for potential participants to make such decisions often do not include the level of detail that participants wanted to receive [bib_ref] What potential research participants want to know about research: a systematic review, Kirkby [/bib_ref]. The numbers of studies looking at this issue in relation to children, adolescents and their parents, is limited [bib_ref] How do children and parents make decisions about pediatric clinical research?, Varma [/bib_ref] [bib_ref] Why do parents enroll their children in research: a narrative synthesis, Fisher [/bib_ref] [bib_ref] Standard 1: consent and recruitment, Caldwell [/bib_ref] [bib_ref] Young people's experiences of participation in clinical trials: reasons for taking part, Luchtenberg [/bib_ref].
The TRECA study will identify the information that children and adolescents consider to be most important when deciding whether or not to participate in clinical trials. Informed by the principles of participatory design, we will produce MMI resources that aim to meet children, adolescents and their parents' needs and preferences for information content and presentation. In particular, this should lead to improved patient information resources that offer relevant information in a way that is accessible to all potential participants, whilst also increasing their understanding of clinical trials in general. Ultimately, it is anticipated that improving participant information resources will increase participation in clinical trials. Specifically, this study will lead to the development of two MMI resources that are suitable for children and adolescents invited to future clinical trials. The findings may also be of use to researchers in different settings, such as education or mental health, in order to better inform and recruit participants to studies. It may also identify principles that are transferable to other medical settings where providing accessible information is crucial, for example, regarding patient decisions about hospital procedures or treatment options.
Results from the TRECA study will be published in peer-reviewed journals and disseminated widely through conferences and events. Where possible, we will also use social media to publicise the findings. In particular, we will aim to engage with health charities, self-help groups and lobbyists, with a view to informing children and adolescents with a variety of long-term health conditions who may be involved in research.
## Trial status
Phase one of the study has commenced. Recruitment for Phase two has not yet commenced.
## Additional file
[fig] Figure 1: Possible pathway of effectiveness of multimedia information (MMI) resources [/fig]
[fig] Figure 2: TRECA study design. MMI multimedia information experiences [/fig]
[fig] Figure 3: Development of the multimedia information (MMI) resources [/fig]
[fig] Figure 4: Phase two study design. MMI multimedia information; CRN Clinical Research Network [/fig]
[fig] Additional file 1: Standard Protocol Items Recommendations for Interventional Trials (SPIRIT) 2013 checklist. (DOC 122 kb) Abbreviations MMI: multimedia information; PIS: participant information sheet; PPI: patient and public involvement; TRECA: TRials Engagement in Children and Adolescents [/fig]
|
Erector spinae plane block with catheter for management of percutaneous nephrolithotomy
Introduction: Percutaneous nephrolithotomy is a procedure used for management of refractory renal calculi. Oral and parenteral opioids, along with local anesthetic infiltration, neuraxial anesthesia, and paravertebral blocks are the most common methods of managing intra-operative and post-operative pain for these patients. The erector spinae plane block with catheter (ESPC) is a newer interfascial regional anesthetic technique that can be used to manage peri-operative pain in these patients.Clinical findings: Three patients complained of significant flank pain were scheduled for percutaneous nephrolithotomy under general anesthesia in the prone position.Diagnoses: Patients were diagnosed with large renal calculi.Therapeutic interventions: Patients received ESPC in the pre-operative holding area at the level of the T7 transverse process. The ESPCS were bolused with a solution of 30 mL 0.25% bupivacaine with 4 mg dexamethasone prior to surgery. Patients also received oral tramadol 50 mg and acetaminophen 1 g as part of the multimodal pain protocol prior to surgery. After the procedure, the patients were bolused with 0.25% bupivacaine or started on an infusion of 0.25% bupivacaine to manage their pain.Outcomes: No opioid or other pain medications, other than the local anesthetic solution given in the ESPCs, were used during the intra-operative or post-operative period for management of pain in these patients. Visual analogue scale (VAS) scores were below 4 for all patients in the post-operative period, and patients did not report any issues with post-operative nausea or vomiting.Conclusion: These patients were compared to 3 prior patients who had undergone percutaneous nephrolithotomy without ESPC. The 3 patients without ESPC placement reported increased VAS scores, had increased opioid/pain medication consumption intraoperatively and postoperatively, and had increased incidence of perioperative nausea when compared to our ESPC patients. Our report shows that ESPC, in combination with a multimodal pain protocol, can be a good option for management of patients undergoing percutaneous nephrolithotomy. Abbreviations: ESPB = erector spinae plane block, ESPC = erector spinae plane block with catheter, MAR = medication administration record, PACU = Post-Anesthesia Care Unit, PCNL = percutaneous nephrolithotomy, PONV = postoperative nausea and vomiting, VAS = visual analogue scale.Keywords: erector spinae plane block, visceral pain, somatic pain, peri-operative pain management, case report All data generated or analyzed during this study are included in this published article (and its supplementary information files).
# Introduction
Percutaneous nephrolithotomy (PCNL) is a minimally invasive surgical procedure often used for definitive treatment of refractory renal calculi. A nephrostomy tube is usually inserted after PCNL to provide adequate urinary drainage, hemostasis, and access for possible complications. However, nephrostomy tubes can cause significant postoperative pain and prolonged hospitalization which can lead to patient dissatisfaction.The main sources of acute postoperative pain after PCNL are visceral pain originating from the kidney and ureter, and somatic pain from the incision site. Visceral pain is primarily transmitted through the T10 to L2 spinal nerves and incisional pain is conducted via T8 to T12 due to the incision site often at the 10th to 11th intercostal space.Current modalities used to manage pain for patients undergoing PCNL include oral and parenteral opioid administration, local anesthetic infiltration, and certain neuraxial and regional anesthesia procedures including thoracic epidurals and paravertebral blocks.These techniques have proven to provide appropriate analgesic coverage but are not without risks and potential side effects.In recent years, the erector spinae plane block (ESPB) has been gaining popularity as a regional acute pain technique for an Editor: Maya Saranathan.
No grants, no sponsors, and no funding sources provided direct or indirect financial support to the work contained in the manuscript. assortment of procedures/pathologies.ESPB is an ultrasoundguided injection of local anesthetic into the paraspinal interfascial plane. The injectate is placed deep to the erector spinae muscle causing it to separate from the posterior surface of the transverse process.Benefits of using this technique include its lower risk profile and fewer contraindications when compared to neuraxial techniques.There has been significant recent effort to elucidate the exact mechanism of analgesic action for ESPB; early data has shown the mechanism of action to be local anesthetic action at the ventral and dorsal rami of the spinal nerves, but this is still being explored and not yet fully understood.ESPB can be performed as a single shot injection or via catheter placement. Pain after single shot ESPB injections has been reported to recur after 2 to 3 h despite using long-acting local anesthetics.It has been hypothesized that the decreased duration of action for single shot injections is due to local anesthetic absorption. Thus, many practitioners are transitioning to the use of continuous catheters to provide longer lasting analgesia.In 2018, Kim et al published a single-patient case report of a patient who underwent placement of an erector spinae plane catheter (ESPC) for postoperative pain management after PCNL.Intraoperative opioid use or the incidence of postoperative nausea and vomiting (PONV) was not discussed. As the use of opioids intraoperatively have been shown to potentially lead to hyperalgesia and increased use of opioids postoperatively,we sought to discover whether ESPC could be used to effectively manage intraoperative pain as well as postoperative pain while decreasing the side effects associated with opioid use.
In this case series, we report three cases of ESPC used for intraoperative and postoperative pain management of PCNL procedures. Each patient received an ESPC that helped achieve satisfactory analgesia while decreasing opioid consumption and PONV.
## Case 1
## Patient information
A 59-year-old female (97 kg, 160 cm) was scheduled for left PCNL due to a 9 mm stone in the left renal pelvis.
## Therapeutic intervention
The ESPC was performed using ultrasound at the level of T7. The T7 transverse process was located using the inferior angle of the scapula as a landmark. A curvilinear 5-2 MHz ultrasound probe was placed on the patient's back in a cephalad-to-caudal fashion at the level of T7. The trapezius and erector spinae muscle were visualized on the left side, and a 19G Tuohy needle was advanced (in-plane to the ultrasound probe) in a caudal direction toward the left-sided transverse process of T7. After confirming that the tip of the needle was below the erector spinae muscle with a small bolus of our local anesthetic solution (30 mL of 0.25% bupivacaine with dexamethasone 4 mg), we injected 20 mL of the solution into the space, noting the caudal spread of the injectate on ultrasound. Then, a 20-gauge epidural catheter was inserted through the Tuohy needle and advanced 5 cm into the space. Placement of the catheter under the erector spinae muscle was confirmed by injecting the final 10 mL of the local anesthetic through the catheter while visualizing the spread of the solution on ultrasound. The catheter was then secured using the LOCKIT Plus catheter securement device, which allows for mechanical securement of the catheter while maintaining its patency.
Additionally, it provides easy access for visual inspection of the insertion site via a transparent window.
## Follow-up and outcomes
The PCNL was performed in the prone position using general anesthesia. The patient received acetaminophen 1 g in the preoperative area as part of a multimodal pain protocol (she refused the tramadol 50 mg which is also part of multimodal pain protocol). She underwent general anesthesia and received no opioid or other pain medications during the 118-min procedure; she was extubated without complication at the end of the case. Pain scores in the Post-Anesthesia Care Unit (PACU) were assessed by nursing using visual analogue scale (VAS) scores and recorded in the medication administration record (MAR). All medications given for pain and PONV were recorded in the MAR as well.
The patient remained in the PACU for 231 min and reported VAS scores of 0/0/0/0/0. She received ondansetron 4 mg intraoperatively and 4 mg prophylactically in PACU due to a patientreported history of severe PONV. The patient was given an additional 10 mL bolus of 0.25% bupivacaine 1-h prior to discharge, and the catheter was removed with the tip intact. The patient was discharged home from the PACU without complications and reported no pain at all.
## Patient perspective
The patient returned to the hospital for a left-sided ureteral stent exchange 15 days later and reported that she had no residual pain in the days following her previous procedure.
## Case 2
## Patient information
A 66-year-old female (66 kg, 160 cm) presented for left PCNL due to a 1 cm left renal stone.
## Therapeutic intervention
In the preoperative holding area, the patient underwent ESPC with catheter placement at T7 using a technique identical to Case 1. She was given 30 mL of the 0.25% bupivacaine and dexamethasone 4 mg solution.
## Follow-up and outcomes
The patient received acetaminophen 1 g and tramadol 50 mg prior to surgery as part of the multimodal pain protocol. She underwent general anesthesia for her 85-min procedure; intraoperatively, the patient received no opioid or other pain medications. She received ondansetron 4 mg prophylactically to prevent PONV.
Upon arrival in PACU, the patient admitted to having mild pain and reported a VAS score of 4. The ESPC was bolused with 5 mL of 0.25% bupivacaine, after which she reported VAS scores of 0/0. The patient was admitted overnight for observation due to concern for bleeding at the surgical site. The ESPC was connected to a continuous infusion in PACU; the settings for the infusion were 6 mL an hour of 0.25% bupivacaine with a patientcontrolled bolus of 4 mL every 20 min as needed. The patient did not use the patient-controlled bolus function and reported having
## Case 3
## Patient information
A 71-year-old female (132 kg, 157 cm) underwent right PCNL for a 7 mm renal stone.
## Therapeutic intervention
As in Case 1 and Case 2, the patient underwent preoperative placement of ESPC. She was given 30 mL of the local anesthetic solution prior to surgery as in Case 1 and Case 2.
## Follow-up and outcomes
Patient received acetaminophen 1 g and tramadol 50 mg as part of the multimodal pain protocol. The 116-min procedure was performed under general anesthesia; she received no intraoperative opioid or other pain medication. She did receive a prophylactic dose of ondansetron 4 mg to prevent PONV. In the PACU, the patient reported VAS scores of 0/0/0/0/0/0/0 and denied PONV. One-hour prior to discharge, the ESPC was bolused with 10 mL of 0.25% bupivacaine and the catheter was removed with the tip intact. She was discharged after a 237-min stay in PACU without pain.
# Discussion
Neuraxial and regional blockade have commonly been used for relieving pain from a myriad of surgical procedures and pain syndromes. ESPC placement has emerged as an effective regional technique, providing excellent analgesia with reduced opioid requirements.Additionally, these blocks are relatively simple to perform with a favorable safety profile. They can be used for a variety of procedures in which multiple dermatomes are targeted for the purpose of providing adequate sensory blockade. Jain et al reported excellent sensory blockade using this regional technique for breast, abdominal, and thoracic procedures, as well as for patients suffering from chronic neuropathic pain.ESPC mitigates potential procedural complications including hypotension from epidural analgesia and vascular puncture from paravertebral block. Additionally, the risk of pneumothorax associated with intercostal nerve block and interpleural block is significantly reduced.Its continuous nature, extensive craniocaudal spread, distance from the surgical field, and excellent sensory blockade could make ESPC superior to many of our current first line analgesic treatments.
For patients undergoing PCNL procedures, much of the pain produced is visceral in nature secondary to the large size of the renal calculus with subsequent stretching of capsular fibers involving the T10 to L2 dermatomes.The local anesthetic administered during the ESPC has a wide area of distribution and likely spreads via action on the ventral and dorsal roots affecting various levels from T1 to T11, based on various studies.Hence, patients undergoing PCNL have the potential to greatly benefit from this type of peripheral nerve block.
In order to further assess the potential benefit of the ESPC, we retrospectively examined the medical records of the three previous patients at our institution who underwent PCNL without ESPC. The first patient we examined received acetaminophen 1 g and tramadol 50 mg as part of the preoperative multimodal pain protocol. She received fentanyl 200 mcg and ketamine 100 mg intraoperatively, and, in PACU, she received hydromorphone 0.5 mg IV and reported VAS scores of 5/0/0/0. She also required ondansetron 4 mg and haloperidol 1 mg for PONV in PACU. The second patient received acetaminophen 1 g and tramadol 50 mg preoperatively, fentanyl 200 mcg intraoperatively, and ketorolac 30 mg and one hydrocodone/acetaminophen 5 mg/325 mg pill in PACU. Her VAS scores in PACU were 3/3/2/2. The third patient received acetaminophen 1 g and tramadol 50 mg preoperatively, fentanyl 100 mcg intraoperatively, and two hydromorphone 0.5 mg IV pushes in PACU. She also required two hydrocodone/acetaminophen 10 mg/325 mg pills for pain during her 1-day admission as well as ondansetron 4 mg on two separate occasions for PONV. Her VAS scores in PACU and on the medical floor were 5/6/2/0/0/6/3/6/3/7/3/3. The patients we examined without ESPC placement had increased VAS scores post-operatively, increased opioid/pain medication consumption intraoperatively and postoperatively, and increased incidence of perioperative nausea when compared to our ESPC patients (see. In our experience of patients who underwent ESPC for PCNL at our institution, intraoperative and postoperative opiate usage, PONV requiring treatment, and VAS scores were all effectively decreased. Our case report shows that ESPC for PCNL can be an effective method to manage intraoperative and postoperative pain in combination with a multimodal pain protocol.
# Author contributions
|
Chinese expert consensus on programming deep brain stimulation for patients with Parkinson’s disease
Background: Deep Brain Stimulation (DBS) therapy for the treatment of Parkinson's Disease (PD) is now a well-established option for some patients. Postoperative standardized programming processes can improve the level of postoperative management and programming, relieve symptoms and improve quality of life.Main body: In order to improve the quality of the programming, the experts on DBS and PD in neurology and neurosurgery in China reviewed the relevant literatures and combined their own experiences and developed this expert consensus on the programming of deep brain stimulation in patients with PD in China.Conclusion: This Chinese expert consensus on postoperative programming can standardize and improve postoperative management and programming of DBS for PD.
# Background
Deep Brain Stimulation (DBS) therapy for the treatment of Parkinson's Disease (PD) is now a well-established option for some patients. An expert consensus in China on DBS in PD was released in 2012, which standardized the indications and processes of DBS therapy, and strengthened the close coordination/cooperation between Neurosurgery and Neurology. In recent years, diagnosis and treatment of PD has made great progress both locally and internationally. The International Parkinson and Movement Disorder Society (MDS) developed a new standard for clinical diagnosis in PD in 2015 [bib_ref] MDS clinical diagnostic criteria for Parkinson's disease, Postuma [/bib_ref]. China also developed an diagnostic criteria for PD, which standardized diagnosis and treatmentand Chinese PD treatment guidelines (third edition)that regulated the diagnosis and treatment of PD in China.
Postoperative programming is an important part of DBS for PD. In combination with drug therapy, standardized programming processes can identify the best postoperative stimulation parameters, relieve symptoms and improve quality of life. Based on the understanding of the importance of DBS management and programming in PD patients, in order to improve the level of postoperative management and programming, the Chinese Medical Association neurosurgery branch of functional neurosurgery group, the Chinese Medical Association neurology branch of Parkinson's disease and Movement disorders group, the Chinese Physician Association neurosurgeon branch of functional neurosurgery expert committee and the Chinese Physician Association neurologist branch of Parkinson's disease and Movement disorders professional committee of national experts, in full review of relevant literature on the basis of the combination of recent years PD diagnosis and DBS therapy of the latest research results, combined with the clinical experience of the expert group to discuss the development of the "Chinese expert consensus on postoperative programming Deep Brain Stimulation for Parkinson's disease".
## Initial programming
The time at which the first programming is initiated remains controversial
Most hospitals choose to start programming at 2 to 4 weeks postoperative. However, few other hospitals begin programming during the hospitalization period. Although early programming stimulation can help facilitate treatment of patients, we have to consider the microlesion effect and relatively significant impedance variation following the DBS electrode placement, which usually last weeks or even months postoperatively. Thus, a later programming allows for the microlesion effects to disappear and electrode impedance to become relatively stable. Current-constant stimulation (CCS) can dynamically adjust the voltage to the change of electrode impedance to tissue contact surface, providing more stability in stimulation strength. Although first programming tested under CCS is recommend in some paper [bib_ref] Programming deep brain stimulation for Parkinson's disease: the Toronto western hospital algorithms, Picillo [/bib_ref] [bib_ref] Current or voltage? Another Shakespearean dilemma, Beaulieu-Boire [/bib_ref] , whether CCS is better than the traditional voltage-constant stimulation (VCS) is not sure, as the literature directly comparing these two is sparse, with no clear conclusion at least on motor symptoms so far. Before programming, doctors should inform patients that dyskinesia, dizziness, numbness and other stimuli related complications could be experienced during the programming to obtain their understanding and cooperation.
## Initial parameters setting
DBS programming is a difficult and time-consuming process, which is best conducted by a highly trained clinician who understands not only the technical aspects of DBS but also PD-related issues and the management of pharmacological treatments. Importantly, postoperative MRI or CT scans are recommended to identify the electrode location. Prior to the initiation of stimulation, the recording of each electrode's contact impedance is recommended to troubleshoot hardware problems [bib_ref] Deep brain stimulation for Parkinson disease: an expert consensus and review of..., Bronstein [/bib_ref] [bib_ref] Deep brain stimulation: postoperative issues, Deuschl [/bib_ref]. These can also be used in future programming as a reference. Generally, the side expressing most severe symptoms is first selected, before proceeding to the contralateral side. Then, one determines for each electrode contact the amplitude threshold for inducing a clinical response and side effects using monopolar stimulation and gradually increase in amplitude (0.2-0.5 V) so as not to cause discomfort. We recommend programmer to participate in or refer to the DBS intraoperative electrophysiology and anatomic target in order to understand the placement of the electrode electrode contacts for postoperative programming.
The initial programming process is as follows: Connect the programmer to the implantable pulse generator. When the connection is complete, input the patient's basic information and stimulator-related information.
To test the impedance of each electrode contact, so as to make sure the connections are intact and then predict the selection of contacts. Test the corresponding contacts one by one to observe the effect on the patient's symptoms and also note side effects. While monopolar stimulation mode is preferred, bipolar stimulation modes are also considered according to the patient's response to adjusted stimulation parameters (frequencies, pulse widths, voltage/current). The bipolar setting is often used to limit side effects identified under monopolar stimulation setting where good response is elicited but limited by side effects. An optimal contact is then selected based on the response after a thorough screening programming on each contact, based on the response and side effects profiles.
The first programming is usually performed off medication to prevent any interference on the effects of DBS. The goal is to determine the amplitude threshold for clinical benefits and determine side effects for each of the four electrode contacts [bib_ref] Deep brain stimulation: postoperative issues, Deuschl [/bib_ref] [bib_ref] Basic algorithms for the programming of deep brain stimulation in Parkinson's disease, Volkmann [/bib_ref]. Some centers also exam the patient again after the medication is on during the same visit to make sure that the programmed setting at medication off status does not produce dyskinesia at medication on status.
## Stimulus parameters
There are many published cases and clinical trials that report the most commonly used DBS frequencies, pulse widths and electrode configurations. However, in the clinic there is wide variability that exists between individual patients, meaning they cannot simply receive the given parameters. As a result, the most commonly used parameters should be considered as starting points rather than set end points. Generally, the initial parameters are set to a monopolar mode, with a pulse width of 60 μs and a frequency of 130 Hz, and a stepwise increase in amplitude according to the patient's response [bib_ref] Deep brain stimulation: postoperative issues, Deuschl [/bib_ref] [bib_ref] Basic algorithms for the programming of deep brain stimulation in Parkinson's disease, Volkmann [/bib_ref] [bib_ref] Pilot study on a fast postoperative programming approach to subthalamic nucleus stimulation..., Li [/bib_ref].
## Dbs chronic stimulation
## General principles
Before programming, it is necessary to ask the patient about changes in symptoms since last visit and exam the patients to understand the clinical problem and also get a new baseline to compare with before you re-program them. Also you need check the DBS settings (contact, voltage, pulse width and frequency) and impedance and current before you reprogram them and compare them to the parameters you put on at his/her last visit and make sure that the DBS still works as should be before your reprogramming. Optimization of DBS parameters is usually attained within 3 to 6 months over 4 to 5 programming sessions [bib_ref] Deep brain stimulation for Parkinson disease: an expert consensus and review of..., Bronstein [/bib_ref]. The overall goal is to alleviate symptoms and prevent side effects by using minimal stimulation intensity and a minimum dose to obtain the maximal therapeutic efficacy.
## Chronic parameters setting
According to the patient's condition and response, monopolar stimulation, bipolar stimulation or double monopolar stimulation mode can be chosen. The proportion of monopolar stimulation and double monopolar stimulation settings are increased slightly as time goes on.
The voltage is generally not more than 3.6 V when using original battery, and the voltage of rechargeable stimulator is not subject to this restriction.
The change in voltage for the subthalamic nucleus (STN)-DBS is relatively large but should be less than 3.6 V, while the frequency change should be less than 190 Hz. Using a greater pulse width means that the stimulated nucleus and surrounding structure are affected, which easily induces adverse effects and increases energy consumption. If the voltage is increased to 3.6 V, the narrowest pulse width of 60 μs can be then increased to 90 μs, allowing for a subsequent reduction in voltage to avoid side effects. 60 Hz will be tried in freezing of gait, dysphagia and axial symptoms in patients with freezing of gait at 130 Hz or higher frequency. Generally, if the electrode position is correct, the narrowest pulse width is sufficient. The combination of the highest voltage with the narrowest pulse width is most effective [bib_ref] The impact on Parkinson's disease of electrical parameter settings in STN stimulation, Moro [/bib_ref]. When patients demonstrate freezing of gait, dysarthria and other midline symptoms, interleaved stimulation, low frequency stimulation(LFS), multiple simulation settings or variable frequency stimulation (VFS) mode are recommended [bib_ref] Variable frequency stimulation of subthalamic nucleus for freezing of gait in Parkinson's..., Jia [/bib_ref] [bib_ref] Low-frequency stimulation of STN-DBS reduces aspiration and freezing of gait in patients..., Xie [/bib_ref] [bib_ref] Low-frequency subthalamic stimulation in Parkinson's disease: long-term outcome and predictors, Zibetti [/bib_ref].
## Important concepts of deep brain stimulation
## Vcs and ccs
At present, the Implanted pulse generator (IPG) has two modes: VCS and CCS. VCS refers to the output voltage that is fixed, whereby the current varies with impedance. In contrast, CCS refers to the output current that is fixed, which provides a specific electrical current that automatically adjusts the voltage depending on the impedance. Current studies have shown that CCS is safe for PD patients [bib_ref] Constant current versus constant voltage subthalamic nucleus deep brain stimulation in Parkinson's..., De Noriega [/bib_ref] [bib_ref] Subthalamic deep brain stimulation with a constant-current device in Parkinson's disease: an..., Okun [/bib_ref]. Longer follow-up studies are necessary to better clarify the impact of CCS on clinical outcome [bib_ref] Switching from constant voltage to constant current in deep brain stimulation: a..., Preda [/bib_ref]. VCS is commonly used in China and abroad. However, there is less experience with the use of CCS, meaning uncertainty lies in the relationship between the magnitude of the current value and efficacy of treatment in patients. Generally, brain tissue and electrode interface impedance changes over a short period postoperatively, meaning a CCS is recommended in some paper but still not clear accepted so far [bib_ref] The rationale driving the evolution of deep brain stimulation to constant-current devices, Bronstein [/bib_ref].
## Interleaved stimulation
With interleaved stimulation, two programs can be interleaved in an alternating fashion on the same lead. Each program drives the same stimulation frequency but different combination of active electrodes, pulse width, and amplitude can be applied. Interleaved stimulation emerges as an effective programming strategy for maximizing symptom control in PD, while regular programming settings do not achieve such favorable results [bib_ref] Interleaved programming of subthalamic deep brain stimulation to avoid adverse effects and..., Ramirez-Zamora [/bib_ref]. For different symptoms, two different contacts of interleaved stimulation may have a better effect, but requiring different voltage for the treatment of complex symptoms to find the best balance between the efficacy and adverse reactions [bib_ref] Interleaved programming of subthalamic deep brain stimulation to avoid adverse effects and..., Ramirez-Zamora [/bib_ref] [bib_ref] Interleaving programming of subthalamic deep brain stimulation to reduce side effects with..., Wojtecki [/bib_ref]. Limited evidence so far, mainly on case report showed that interleaved stimulation may be helpful for gait disorders [bib_ref] Reversible improvement in severe freezing of gait from Parkinson's disease with unilateral..., Brosius [/bib_ref] and dysarthria [bib_ref] Individualized current-shaping reduces DBS-induced dysarthria in patients with essential tremor, Barbe [/bib_ref].
## Multiple simulation settings
Physicians can preset multiple simulation settings that can benefit the PD patients on different occasions. Firstly, to meet patient's needs in different situations. For example, if they feel unsatisfied with their stimulation settings, they could alternate between different settings under the physicians' instruction. Notably, physicians should always determine the upper limit of the stimulation variable that the patient can use safely and restrict the ability of the patient to exceed that limit. Secondly, the programmer can find the most appropriate stimulation parameters through the "titration" approach. The patient or his caregiver properly operate the patient controller is a prerequisite for using the multiple simulation settings. After a period of time the application program is not effective, the patient can switch to another program group.
## Lfs
The DBS stimulation settings that are regularly used are high frequency (> 100 Hz). High frequency stimulation (HFS) improves the cardinal symptoms in patients with Parkinson disease (PD). However, it is less effective at improving the axial symptoms, which include postural instability, gait disorders, speech and swallowing dysfunction. In recent years, LFS has drawn much interest for the improvement of axial symptoms, in comparison with the effects of high frequency stimulation [bib_ref] Low-frequency stimulation of STN-DBS reduces aspiration and freezing of gait in patients..., Xie [/bib_ref] [bib_ref] Early use of 60 Hz frequency subthalamic stimulation in Parkinson's disease: a..., Ramdhani [/bib_ref] , although there are still some studies argued that there was no significant difference between HFS and LFS for controlling symptoms of gait disorders and stability [bib_ref] Low-frequency versus high-frequency subthalamic nucleus deep brain stimulation on postural control and..., Vallabhajosula [/bib_ref]. Thus, some authors still think that further studies are needed for clearly demonstrating the treatment of axial symptoms with LFS in PD patients [bib_ref] Fasano a. Low-frequency deep brain stimulation for Parkinson's disease: great expectation or..., Di [/bib_ref]. Overall, LFS is recommended in patients with freezing of gait at HFS of STN DBS, although LFS could worsen the tremor in small proportion of patients.
## Vfs
Tradition stimulation settings drive a constant frequency stimulation, which gives an electrical pulse with constant frequency to the target brain nucleus. High frequency stimulation (HFS) of STN provides long-term improvement of the cardinal signs of PD. However, freezing of gait (FOG) and many other axial symptoms respond poorly to HFS, with symptoms increasing in severity over time. Overall evidence suggests that LFS improves axial symptoms in patients with freezing of gait at HFS of STN DBS, although LFS could worsen the tremor in small proportion of patients. Limited evidence from case report that VFS alternates between low and high frequencies, which may significantly improve freezing of gait and other axial symptoms in PD patients, while maintaining beneficial effects on cardinal symptoms [bib_ref] Variable frequency stimulation of subthalamic nucleus for freezing of gait in Parkinson's..., Jia [/bib_ref]. Further studies, especially large scale randomized controlled clinical trials are needed. The programmer can try to use VFS for gait disorders besides LFS.
## Patient control device
IPG can be adjusted and programmed. Physicians will set the IPG with patient's basic information and use stimulation settings from the professional programming device. Physicians can allow patients to change certain DBS stimulation variables, such as voltage, and also check the IPG battery or on/off states with their own control device.
## Rechargeable ipg
Commercially available rechargeable IPGs provide a significant advantage in reducing the number of surgeries for IPG replacement. However, there are concerns regarding the safety of using rechargeable IPGs. Patients should strictly follow the instructions. While patients should charge their IPGs in vitro, they should notice the temperature warning to avoid electrical skin burns.
## Common problems and troubleshooting
It must be clear that DBS is an effective supplementary means of PD drug treatment. As a treatment for PD, DBS does not alleviate all symptoms. DBS could significantly improve the cardinal symptoms of PD, including tremor, rigidity and akinesia. However, in the PD patients with axial symptoms, such as postural abnormal gait, balance disorders, dysarthria, dysphagia, the DBS therapy is still challenging. DBS has no therapeutic effect on cognitive impairment and other non-motor symptoms. So taking good care and rehabilitation training may be a better choice. Based on the current level of medical technology development, doctors and patients should have a reasonable expectation on DBS therapy.
## Dyskinesia
Several types of dyskinesia have been identified, which include peak-dose, end-dose and biphasic dyskinesia [bib_ref] Clinical spectrum of levodopa-induced complications, Aquino [/bib_ref]. Peak-dose dyskinesia manifests after patients take levodopa treatment. End-dose dyskinesia means wearing-off motor fluctuations. Biphasic dyskinesia includes the two types of dyskinesia above.
STN-DBS can induce or aggravate dyskinesia. DBSinduced dyskinesia indicates the accurate location of the implanted electrode and a good prognosis. In most cases, DBS-induced dyskinesia will gradually disappear over several days to several months after turning on the IPG. Physicians can also try different combinations of parameters, choose a bipolar stimulation setting and adjust the dose of levodopa for reducing dyskinesia. Globus pallidus internal (GPi)-DBS has a direct antidyskinetic effect, while STN-DBS depends on the stimulation of dorsal-lateral contact or postoperative reduction of dopaminergic medications [bib_ref] Surgical treatment of dyskinesia in Parkinson's disease, Munhoz [/bib_ref]. After the control of motor symptoms, PD patients should slowly reduce the dose of levodopa and dopamine receptor agonists to reduce the incidence of dyskinesia. During long-term follow-up, it's better to maintain stimulation parameters stable, adjustment of anti PD drugs should comply with the "dose titration" principle. If dyskinesia persists, amantadine can be considered [bib_ref] Collective physician perspectives on non-oral medication approaches for the management of clinically..., Odin [/bib_ref].
## Dizziness
Dizziness accounts for many reasons and is not limited to symptoms of PD, drug side effects, and side effects of stimulation or other diseases (such as hypotension). It is recommended that there is a reduction in voltage or pulse width, change in stimulation contact, or switch to bipolar stimulation mode (to adjust the electric field range) for alleviating dizziness. If the symptoms still exist after turning off the IPG, the likely cause of dizziness is the adverse drug reactions or effect of other diseases. Therefore, it is recommended to manage drug treatments to reduce dizziness.
## Speech impairment
During disease progression, speech impairment of PD patients laguage will aggravate. It is recommended to attempt to take the following measures to reduce the speech impairments in DBS therapy: 1) reduce the voltage and pulse width of stimulation; 2) use bipolar stimulation; 3) change active electrode contacts; 4) reduce the frequency of stimulation; 5) apply VFS; 6) apply interleaved Stimulation; 7) adjust the dosage of levodopa; 8) undergo speech rehabilitation training. However, these measures have the potential to reduce the effect on the motor symptom, so programmer should weigh the advantages and disadvantages.
## Dysphagia
Dysphagia is an axial symptom, which is inevitable during the natural progression of PD. Dysphagia can induce respiratory problems and is the main cause of death in PD patients. Although STN-DBS is considered to be more likely to induce speech impairments than GPi-DBS, there are no clinical trials that have directly compared the effects of both targets on swallowing function. Due to the progression of PD, it is hard to improve the dysphagia function via adjustment of drugs or DBS programming, which leaves the high quality of nursing care the only choice for symptom control. Nursing care includes eating soft food and methods to assess swallowing function to ensure that there are no breathing problems. Swallowing function can be improved through adjusting the medicine or rehabilitation training. A high quality of nursing care can prevent aspiration of content and induced asphyxia [bib_ref] Swallowing and deep brain stimulation in Parkinson's disease: a systematic review, Troche [/bib_ref]. For these patients with FOG and dysphagia at usual HFS, LFS of 60 Hz could also be tried [bib_ref] Low-frequency stimulation of STN-DBS reduces aspiration and freezing of gait in patients..., Xie [/bib_ref].
## Gait initiation difficulty
Gait disorders caused by FOG respond poorly to HFS. In addition, the severity of gait disorders becomes more prominent as PD progresses. It is found that not all the patients benefit from adjusted medication but could benefit from the LFS. Gait disorders that respond well to levodopa pre-operation usually helps predict the improvement of gait disorders with STN or GPi-DBS surgery. Interleaved stimulation and VFS have been shown to be effective for the improvement of gait disorders [bib_ref] Variable frequency stimulation of subthalamic nucleus for freezing of gait in Parkinson's..., Jia [/bib_ref]. Patients with gait disorders can also benefit from physical therapy and rehabilitation training.
## Balance disorders and gait disorders
Patients with gait disorders need to increase the drug dose during chronic stimulation period. If the dopaminergic drugs are ineffective, doctors may consider amantadine or other drugs [bib_ref] Motor and cognitive outcome in patients with Parkinson's disease 8 years after..., Fasano [/bib_ref]. STN-DBS postoperative, doctors should try interleaved stimulation, LFS, VFS or increase the dose of levodopa when patients with FOG [bib_ref] Freezing and hypokinesia of gait induced by stimulation of the subthalamic region, Tommasi [/bib_ref]. For the patients with balance disorders, the programmer should pay attention to maintaining the bilateral limb muscle tension symmetry as well as DBS parameters and drugs. The dyskinesia, orthostatic hypotension and other factors affecting the balance of patients with other factors should also be considered [bib_ref] Modulation of gait coordination by subthalamic stimulation improves freezing of gait, Fasano [/bib_ref]. The following options can be recommended. 1) reduction of voltage; 2) change in active electrode contacts (i.e. use of a more ventral contact); 3) reduction in frequency; 4) application of VFS; 5) application of interleaved Stimulation.
## Pain
These symptoms should be assessed to ensure pain is caused by PD rather than any other disease the patient may have. There are several subtypes of pain including musculoskeletal pain, dystonia pain, central pain, radicular pain and akathitic pain. These subtypes respond differently to levodopa treatment, DBS surgery, rehabilitation training and other therapeutic methods. PD patients that suffer from pain could benefit from LFS [bib_ref] Effect of low-frequency deep brain stimulation on sensory thresholds in Parkinson's disease, Belasen [/bib_ref]. Some evidence has shown that anti-parkinson drugs relieve musculoskeletal pain, while anti-depressant drugs alleviate central pain. It should also be noted that pain may also be a side effect of anti-parkinson drugs and withdrawal or adjustment of regimens can improve pain relief [bib_ref] Effects of deep brain stimulation on pain and other nonmotor symptoms in..., Cury [/bib_ref] [bib_ref] An 8-year follow-up on the effect of subthalamic nucleus deep brain stimulation..., Jung [/bib_ref].
## Cognitive impairment
Postoperative the incidence of cognitive impairment is about 0.8-5.1% [bib_ref] Collective physician perspectives on non-oral medication approaches for the management of clinically..., Odin [/bib_ref]. The effects of DBS surgery on cognitive function include the effects of surgical microlesiom and DBS stimulation related effects. Due to damage to the basal ganglia-dorsal prefrontal loop, DBS lead trajectory may lead to cognitive decline [bib_ref] Influence of deep brain stimulation of the subthalamic nucleus on cognitive function..., Wu [/bib_ref]. Damage to the caudate nucleus in the electrode path can lead to decreased working memory and overall cognitive function decline [bib_ref] Relation of lead trajectory and electrode position to neuropsychological outcomes of subthalamic..., Witt [/bib_ref]. Cognitive function impairment was seen in both GPi and STN DBS, but less impaired in GPi than STN [bib_ref] Cognitive and psychiatric effects of STN versus GPi deep brain stimulation in..., Wang [/bib_ref]. Patient's speech and semantic fluency may decline after STN-DBS [bib_ref] Clinical outcome of deep brain stimulation for Parkinson's disease, Deuschl [/bib_ref]. A systematic cognitive assessment of patients after DBS should include a cognitive assessment in both DBS on and off [bib_ref] Collective physician perspectives on non-oral medication approaches for the management of clinically..., Odin [/bib_ref].
For PD patients with cognitive impairment after DBS surgery, the contacts and stimulation parameters should be adjusted [bib_ref] Frequency-dependent reciprocal modulation of verbal fluency and motor functions in subthalamic deep..., Wojtecki [/bib_ref]. For STN-DBS, contacts should be adjusted to avoid STN non-sensory-motor function subregions [bib_ref] Progress on anatomy of subthalamic nucleus, Zhao [/bib_ref]. Compared with low-frequency stimulation, HFS leads to a higher risk of language fluency decline [bib_ref] Frequency-dependent reciprocal modulation of verbal fluency and motor functions in subthalamic deep..., Wojtecki [/bib_ref] , so trying to reduce the frequency is a good option. The drugs, including cholinesterase inhibitors (rivastigmine and donepezil) and memantine [bib_ref] Pharmacological treatment of Parkinson disease: a review, Connolly [/bib_ref] [bib_ref] Efficacy and safety of cholinesterase inhibitors and memantine in cognitive impairment in..., Wang [/bib_ref] , for cognitive impairment after DBS can refer to the treatment principles of PD cognitive impairment [bib_ref] Collective physician perspectives on non-oral medication approaches for the management of clinically..., Odin [/bib_ref] [bib_ref] Pharmacological treatment of Parkinson disease: a review, Connolly [/bib_ref]. The caregivers should take good care of PD patients with cognitive impairment [bib_ref] Long-term cognitive outcome of bilateral subthalamic deep brain stimulation in Parkinson's disease, Kim [/bib_ref] [bib_ref] Cognition and depression following deep brain stimulation of the subthalamic nucleus and..., Combs [/bib_ref].
## Postoperative depression, anxiety and apathy
The risk factors for postoperative depression including a history of preoperative depression, rapid or excessive reduction of dopaminergic drugs, and having difficulty to adapt to life changes [bib_ref] Neuropsychiatric effects of subthalamic neurostimulation in Parkinson disease, Volkmann [/bib_ref]. Therefore, for DBS patients with postoperative depression, doctors can consider increasing the equivalent dose of levodopa or using dopamine receptor agonist (Pramipexole). When these medical measures prove ineffective, doctors should consider the serotonin reuptake inhibitors and psychotherapy [bib_ref] Summary of the recommendations of the EFNS/MDS-ES review on therapeutic management of..., Ferreira [/bib_ref] [bib_ref] Deep brain stimulation: neuropsychological and neuropsychiatric issues, Voon [/bib_ref]. DBS high frequency stimulation may lead to acute depressive state, which need to be reprogramming [bib_ref] Neuropsychiatric effects of subthalamic neurostimulation in Parkinson disease, Volkmann [/bib_ref].
Treatment of postoperative anxiety should be based on the principles for PD with anxiety [bib_ref] The psychiatric and neuropsychiatric symptoms after subthalamic stimulation for Parkinson's disease, Mosley [/bib_ref]. Benzodiazepine drugs are commonly used anti-anxiety drugs, which may increase cognitive impairment or fall risk. Selective serotonin reuptake inhibitors can also be used in anxiolytic treatment. The incidence of apathy after DBS was related to preoperative non-motor symptom fluctuations, anxiety, young patients and impulse control disorders [bib_ref] Non-motor dopamine withdrawal syndrome after surgery for Parkinson's disease: predictors and underlying..., Thobois [/bib_ref] [bib_ref] Medical Management of Parkinson's disease after initiation of deep brain stimulation, Fasano [/bib_ref]. Apathy after DBS can be improved by the use of selective D2/D3 receptor agonists [bib_ref] Neuropsychiatric issues in Parkinson's disease, Cooney [/bib_ref].
## Dbs postoperative oral drug management
The principle of drug application for PD is consistent, regardless of before or after DBS. Doctors should follow the Chinese PD treatment guidelines, refer to the International Movement Disorders Association, the American Academy of Neurology, the European Federation of Neurological Association and other international guidelines and recommended consensus. Meanwhile, the efficacy of DBS and potential adverse reaction risk should also be considered to comprehensively adjust drugs.
DBS postoperative include early postoperative period and long-term follow-up. The DBS perioperative period include the first day to 1 week postoperative. PD patients should take preoperative drugs after anesthesia awareness. It is suggested to maintain the dose of preoperative levedopa in patients with significant preoperative motor symptoms. The dose of levedopa can be reduced appropriately due to the microlesion effect after DBS electrode implantation. The patients with preoperative dyskinesia should reduce the dose of levedopa. Other anti-PD drugs can be suspended to avoid the corresponding adverse reactions. In the early postoperative period (1 week to 4 weeks after operation), the microlesion effect gradually faded, and anti-PD drugs could be gradually adjust to preoperative medication. Levedopa is still recommended to the elderly over the age of 70 for PD patients with preoperative mild cognitive impairment or neuropsychiatric symptoms. For patients with preoperative or postoperative dyskinesia, the equivalent dose of levodopa may be considered to reduce. For PD patients with severe preoperative or postoperative end-of-dose dyskinesia, it's better to increase the levodopa equivalent daily dose (LEDD) or other anti-PD drugs.
During DBS chronic stimulation period, doctors can determine the type and dose of drugs according to patient's motor symptoms and non-motor symptoms after the application of conventional DBS parameters. Usually STN-DBS postoperative, LEDD can be reduced by 30-50% [bib_ref] Subthalamic deep brain stimulation with a constant-current device in Parkinson's disease: an..., Okun [/bib_ref] [bib_ref] Clinical outcome of deep brain stimulation for Parkinson's disease, Deuschl [/bib_ref] [bib_ref] A randomized trial of deep-brain stimulation for Parkinson's disease, Deuschl [/bib_ref] [bib_ref] Deep brain stimulation plus best medical therapy versus best medical therapy alone..., Williams [/bib_ref] (Class 1 Evidence, Class I recommendation). Drugs can be progressively reduced under the guidance of a physician if patient's symptoms can be alleviated by DBS without severe complications. If the voltage reaches the limit, continued increase will significantly increase energy consumption, except for rechargeable DBS. It should be aware that overly reduced dopaminergic medication postoperatively can cause anxiety and depression. Also increase in voltage could cause side effect, not just because of the energy consumption, regardless if whether rechargeable battery or "not" is added. Due to DBS-related adverse reactions, patients can gradually adjust drugs after reaching a steady state. The selection of drug and dose should refer to patient's age, occupation and life needs and postoperative symptoms.
# Conclusion
Postoperative management and DBS programming for PD patients are important aspects of DBS therapy, which must be paid full attention. While DBS has been in use for nearly 30 years, there is still no systematic programming guidelines. This has led to inconsistent and inefficient adjustments in stimulation parameters, as well as numerous and unnecessary patient visits. These issues have compelled us to find ways to improve the efficiency of our programming sessions that are aimed at quality improvements, thereby enhancing the patient's quality of care. Patients should be regularly referred for evaluation and parameters adjustment. We advocate that in the future, establishment of programming database replace the current paper-based programming record for ease of review and sharing experience. In recent years, there are some new progress in DBS programming, such as the directional electrode and advanced DBS software. These new technologies play a huge role in the development of DBS programming, which can be applied under the guidance of doctors. It must be clear that none of the above-mentioned IPGs has all the features to date. Both the doctors, patients and their caregivers should read the DBS equipment instructions in detail to give full play to the role of DBS. |
Towards patient-centred cancer care: cross-cultural validity and responsiveness of the Turkish Integrated Palliative care Outcome Scale
Background: A valid measure to describe the most important needs and concerns of people with life-threatening illnesses is missing in Cyprus. Our aim was to adapt and test the cross-cultural validity and responsiveness of the Integrated Palliative care Outcome Scale (IPOS) in a cohort of Turkish speaking cancer patients.Methods:The IPOS (English) patient-reported measure was translated into Turkish following published guidelines including, 2 independent forward, 2 independent blind backward translations, expert panel review by 7 members and field testing with 11 cognitive interviews (5 patients and 6 specialists) and final approval of the copyright holder. Consecutive cancer patients (n = 234) seen by the community palliative care services were recruited from Help Those with Cancer Society (KHYD); of those 82 were followed-up. The instrument was administered by personal interview. Confirmatory Factor Analysis was used to validate the factor structure of Turkish IPOS. Internal consistency reliability of the subscales was evaluated by Cronbach's alpha and Intraclass Correlation Coefficient respectively. Validity was assessed by calculating Pearson's correlation coefficient (r) between Turkish IPOS scores and Turkish version of EQ-5D-3L -a validated generic measure of health status developed by the EuroQol Group.Results: Turkish IPOS is conceptually and semantically equivalent to the English version and linguistically valid. The CFA was inconclusive for the three factor structure due to low sample size, as the SRMR and CFI tests only approached the defined minimums warranting further investigation. There were low levels of missing values, and no ceiling or floor effects. The Physical (α = 0.91) and the Social and Quality of Care Issues (α = 0.75) sub-scales showed good internal consistencies, however Emotional sub-scale showed poor internal consistency (α = 0.64). The reliability of the Physical (ICC = 0.51, 0.45-0.56 95% CI) and Social Quality of Care Issues (ICC = 0.50, 0.42-0.57 95% CI) were moderate. Poor internal consistency (α =0.64) and reliability (ICC = 0.31, 0.24-0.39, 95% CI) was obtained for Emotional Subscale. Construct validity was evidenced through significant correlations in the predicted directions (Continued on next page)
(Continued from previous page) and strength with EQ-5D. Turkish IPOS showed higher needs and concerns in participants at more advanced stages than those at earlier stages of cancer. The standardized response mean (SRM) of − 0.94 suggested large internal responsiveness to clinical change.
Conclusion: Turkish IPOS is a clear, relevant, acceptable measure and responsive to the needs and concerns of cancer patients, observing regional differences, it may have implications for use in other Turkish speaking communities. Future studies are needed to clarify the factor structure, assess its external responsiveness and to improve the properties of its Emotional subscale.
Keywords: Patient reported outcome measures , Palliative care, Psychometrics , Validation study, Cancer Background A comparative review of palliative care development in the Middle Eastern Cancer Consortium (MECC) member countries identified 'lack of awareness and understanding of palliative care needs at public, government and professional levels' as barriers to provision of palliative care to cancer patients in this region. These observations maintain their relevance today for cancer patients living in MECC member countries including the setting for this study, Cyprus. Palliative care as ' … an approach that improves the quality of life of patients and their families … through the prevention and relief of suffering by means of early identification and impeccable assessment and treatment of pain and other problems, physical, psychosocial and spiritual'is rarely recognized. As a concept, palliative care is only equated with management of pain in cancer and available only in oncology inpatient and intensive care units.
Within Cyprus, resources for health services including community home-based services are scarce and receive no or little government support. The health expenditure is the lowest compared to other European Union member countries and 25% of the population has no access to public health services. Early palliative care as cost-effective intervention in cancer patients has already been demonstrated. Evidence of the effectiveness and costeffectiveness of palliative care services may promote their integration for persons living with all serious illnesses in Cyprus. After years of lobbying, local cancer charities have been formally authorized to deliver community homebased palliative care services in Cyprus. This recognition may serve as a 'catalytic action'that may increase the coverage of the palliative care services beyond inpatient and intensive care unit settings to the community. The charities must support home services through fundraising and ensure their resources address the most important concerns early. This has necessitated availability of a brief and valid patient-reported outcome measure (PROM) to describe the most important concerns of people with living with cancer.
Routine use of PROMs can improve the quality and relevance of servicesand also can help build awareness in healthcare providers about the most important needs of patients and families. The Integrated Palliative care Outcome Scale (IPOS) is a valid brief PROM that addresses the most important concerns such as symptoms, information needs, practical concerns, anxiety and low mood, family anxieties and overall feelings of being at peace of persons living with serious illnesses, such as cancer. IPOS has been used with cancer patientsand adapted to many cultures.
In this study, we first translated and culturally adapted IPOS into Turkish, and evaluated the cross-cultural validity and responsiveness of IPOS in Turkish speaking community of cancer patients in Cyprus. The findings of the study could have implications for millions of patients living with serious illnesses around the world who have Turkish as their native language.
# Methods
The study commenced after attaining permissions from the POS Development Team as the copyright holder of IPOS, the EuroQol Group for use of Turkish version of EQ-5D-3L. Ethical approval was obtained from the Eastern Mediterranean University Publication and Research Ethics Committee (ETK00-2017-103). Written informed consent was obtained from all participants included in the study.
## Participants and procedures
Patients seen by the community palliative care team including new or old referrals, of Help Those with Cancer Society in Cyprus (KHYD), aged 18 and above were eligible to participate in the study. Patients who were not referred to the community palliative care services, or were younger than 18, were excluded from the study. Invitation and recruitment was consecutive and stopped once the target size of at least 10 cases per itemwere reached. Two hundred and thirty-four consenting patients were asked to complete the Turkish IPOS alongside the EQ-5D-3L. For follow-up, 82 participants were asked to complete the Turkish IPOS a second time during their next routine visit. Due to unavailability of resources, the follow-up visit and assessment could only carried out in the next scheduled routine visit by the community palliative care team rather a specified time period. Data were collected by the 4 community nurses during routine home visits.
Patients were also asked to complete questions about their age, gender, education, marital status, occupation, number of children, how they were meeting treatment and care expenditures, their sources of support and co-morbidities. Clinical information on the primary tumour site including stage was extracted by the Community Palliative Care Team (CPCT) from their membership registration files. The testing and reporting of the measurement properties of the Turkish IPOS followed the Consensus-based Standards for the selection of health Measurement Instruments (COSMIN) recommendations.
## Measures/questionnaires integrated palliative care outcome scale (ipos)
IPOS is a 10 question, 17-item brief PROM addressing symptoms, information needs, practical concerns, anxiety, low mood, family anxieties and overall feeling of being at peace of persons living with life-threatening illnesses. It is scored on a 5-point Likert scale (0-4) with higher scores indicating an overwhelming presence of symptoms and needs not addressed. Patients may also list their main problems and concerns and any additional symptoms. The 7 day patient version recommended for use in community-based services was used in this study. An earlier study identified Physical Symptoms, Emotional Issues and Support (Social issues and Quality of Care) as three sub-scales of IPOS.
EuroQol Group's Generic Health Status Preference based Measure -EQ-5D-3L EQ-5D-3L evaluates a person's health status based on 5 dimensions: mobility; self-care; usual activities; pain/discomfort; and anxiety/depression. It is scored on a scale from 1 to 3 where 1 represents 'no problems', 2 denotes 'some problems', and 3 'extreme problems.' It is accompanied by a Visual Analogue Scale (EQ VAS) which records the patient's self-rated health on a vertical VAS with the endpoints 'Best imaginable health state' and 'Worst imaginable health state'. Turkish version of EQ-5D-3L has been validated in cancer patientsand therefore was chosen as comparative measure.
## Translation and cross-cultural adaptation
The 'Manual for cross-cultural adaptation and psychometric validation of the IPOS development group'was rigorously followed for the translation and crosscultural adaptation. Literature review and consultations with patient, their families and clinicians was used to establishing IPOS' face and content validity. Face and content validity were reconfirmed in the validation phase of the study by the free text entries of respondents.
Four translators and two mediators followed the independent forward-blind backward translation method to translate IPOS (English) to Turkish. Two translators who carried out the two independent forward translations, were native in Turkish and fluent in English. One of the translators were an English teacher, the second one was a general practitioner. Two translators, who carried out the blinded backward translations were native English speakers, who have moved to Cyprus from the UK, and were fluent in Turkish. Both were English teachers. The mediators were two cancer patients. The expert review panel included all four translators and two mediators, and the first author. The expert review panel reviewed and synthesized the translations, and prepared the pre-final Turkish IPOS for piloting (cognitive testing).
A total of 5 cancer patients and 6 clinicians including nursing staff and counsellors, completed the pre-final Turkish IPOS in order to test the alternative wording and to check understandability, interpretation, and cultural relevance of the translation. Probes such as 'What does the term practical problems mean to you?', 'In your own words can you repeat what this question is asking?', 'Was that question easy or hard to answer?' and 'I can see that you have chosen moderately or 2 as your response or score, why is that?'were used. Participants found the questionnaire easy to understand. Completed reports on the general information on participants, translators and expert review panel members, summary of all phases of the process and on the synthesis of the two forward translations and the two backward translations together was collated in a final report. These documents with the Turkish IPOS Pre-Approval Version as agreed by the research team and the copyright holder, were sent to POS Development Team at King's College London for review and final approval before proceeding to validation phase.
# Statistical analysis
Descriptive and multivariate analyses were carried out to present participant characteristics and to assess data completeness, distribution, internal consistency, reliability, validity and internal responsiveness of Turkish IPOS using IBM SPSS Statistics 25.0 and Amos 25.0. Pairwise deletion was used in multivariate statistical analyses. Participant characteristics were summarized as percentages of the total participants. Mean or median with standard deviation or range of values were presented where relevant. Data completeness was assessed by Missing Value Analysis (MVA). For acceptability, means, standard deviation, range of values, skewness, kurtosis as well as the ceiling and floor effects were assessed for the total scores as well as for the three sub-scales of IPOS. Content validity were re-confirmed through patient responses to two free-text parts of the Turkish IPOS, one about main problems and concerns, and the second on any other symptoms not listed.
A first order Confirmatory Factor Analysis (CFA) was used to assess how well the hypothesized 3-factor model fits the observed data. The model parameters were estimated using Full Information Maximum Likelihood (FIML) method to make maximal use of all data. Post-hoc modifications were undertaken to improve model fit. As the chi-square statistic has been shown to fail to discriminate between good fitting models and poor fitting models in small sample sizes (n ≤ 250), combinational rules and cut-off criterion for fit statistics that minimize Type I and Type II errors in small samples were considered. Comparative Fit Index (CFI) and Standardized Root Mean Square Residual (SRMR) were used to evaluate the model fit with CFI ≥ 0.95 and SRMR< 0.09 suggested good fit.
Internal consistency of each sub-scale was evaluated with Cronbach's α coefficientwhere α values in the range of 0.70-0.90 is indicative of good internal consistency. In order to estimate the reliability, Interclass Correlation Coefficients (ICCs) using twoway mixed effect model was obtained and ICCs for single measurement are reported. The choice of the model was informed by the fact that multiple administrators from a selected pool carried out the assessments and consistency of measurement was critical. ICCs less than 0.5 are indicate poor, between 0.5 and 0.75 indicate moderate, 0.75 and 0.9 are indicate of good, and above 0.90 indicate excellent reliability.
Construct validity where conceptual convergence and divergence between Turkish IPOS and EQ-5D-3L items were evaluated with bivariate correlation analysis, where Pearson's correlation coefficient (r) between 0 and 0.2 indicated weak, 0.2-0.4 low, 0.4-0.6 moderate and 0.6-0.8 strong relationship given statistical significance. Known groups (discriminant) validity was assessed by independent samples t-test comparing total and sub-scale scores in participants with early (Stages I-II) versus advanced cancer (Stages III-IV).
Standardized Response Mean (SRM) was calculated to estimate change in scores standardized relative to variability between the participants to assess internal responsiveness. Baseline and follow-up scores were compared using paired t-test. Bonferroni corrected p ≤ 0.01 was used for significance to account for multiple comparisons.
# Results
Literature and consultations confirmed that IPOS covered all of the relevant concepts in relation needs and concerns and had face and content validity. Pain and emotional difficulties emerged as the most prevalent issues. In the 2 forward translations there were differences in the grammatical tense used; translators used present continuous tense, but the mediators used past tense. The item on practical problems resulting from illness was particularly challenging, therefore the word 'practical' was removed and the question revised to ask about 'personal and financial problems' caused by illness. The 2 blind backward translations produced from final reconciled forward translation of Turkish IPOS were both similar with no significant differences in meaning. The expert panel reviewed all versions to improve clarity and produced the pre-final Turkish IPOS.
Following cognitive testing, instructions were reworded and further shortened. The response options were generally clear and well understood; the only difficulty was with 'occasionally' which was suggested to be changed to 'zaman zaman' rather than 'ara sıra'. Patients struggled with the information item which was revised to emphasize that the question was not about the knowledge the patients possessed but the information given to them. Turkish IPOS Pre-approval Version was sent to the Cicely Saunders Institute POS Team for independent review. Following this and minor amendments, Turkish IPOS was produced and validation study was initiated.
For the validation study, data collection took place between July 2017 and October 2018. Participants were mostly women (73.5%), almost all were struggling with treatment expenses (91.5%) and most received support from family members (83.3%). Of the 43.6% who reported having comorbidities, 50% had hypertension and 28.4% had diabetes. Fifty three percent of the participants had advanced cancer and breast cancer was the most prevalent (48.3%).
MVA identified four items with higher number of missing responses in baseline assessment (9.4-7.7%) . As the data were not missing completely at random (Little's MCAR test χ 2 = 291.512, DF = 184, p < 0.0001), listwise deletion or imputation of missing values was not undertaken. Patients at moderate stages of their illness (Stage II and III) had more than 5% of responses missing for these items. Only two items were missing 1 response each in the follow-up assessment.
CFA fit indices were CFI = 0.756 and SRMR = 0.0002 indicated poor fit of the model to the data (χ 2 = 496.369, df = 87, χ 2 /df = 5.705, p < 0.0001) and post-hoc modifications were made. Post-hoc modification involved adding in covariances of error terms based on theoretical and relevant associations of items measured. The model fitand . The Emotional sub-scale showed poor internal consistency (α = 0.64), three of the items were worthy of retention but deletion the item on anxiety of family and friends would increase α to 0.75. The reliability of this subscale was also poor (ICC = 0.31, 0.24-0.39, 95% CI). The Physical subscale showed good internal consistency (α = 0.91) but moderate reliability (ICC = 0.51, 0.45-0.56 95% CI). Social and Quality of Care Issues subscale showed good internal consistency (α = 0.75) and marginally moderate reliability (ICC = 0.50, 0.42-0.57 95% CI).
The participants listed feeling worried, financial difficulties, lack of support, impact of cancer on the family on activities of everyday life as their main concerns in response to the open-ended first question in IPOS. As these topics were all covered by IPOS, its content validity was reconfirmed.
Correlations with EQ-5D-3L domains were significant and, mostly in the directions and strength predicted. For example, IPOS Emotional subscale, had low correlations with EQ-5D Mobility, Self-care, Usual Activity, slightly higher but still low correlations with Pain/Discomfort and Anxiety/Depression and moderate negative correlation with self-rated health. The IPOS total scores were normally distributed . There were no ceiling or floor effects for the total score or for any of the sub-scale scoresThe mean time between the two administrations were 4.6 weeks (Range = 2-7, SD = 1.5). SRM of − 0.94 suggests large responsiveness. Total scores at follow-up were significantly lower compared to baseline with a mean difference of 9.87 (7.18-12.58, 95% CI), t (57) = 7.33, p < 0.0001. A similar trend was observed for the physical (x 1 -x 2 = 7.02, t (72) =10.3, p < 0.0001) and quality of care/social support subscales (x 1 -x 2 = 2.11, t (61) = 6.35, p < 0.0001). However, emotional concerns significantly worsened compared to baseline (x 1 -x 2 = − 1.74, t (61) = − 4.2, p < 0.0001).maps change scores for 58 cancer patients with complete data across all four scales and assessments.
# Discussion
In this study we adapted IPOS to Turkish and evaluated its cross-cultural validity and responsiveness in cancer patients. Most of the items were translated easily and perceived to be culturally relevant. Contrary to experiences of teams working on other language translations, "at peace" item was particularly easy to adapt to Turkish as this is a very culturally relevant concept. Items on "practical problems addressed" and "information" were the most challenging to translate and adapt because the word "practical" is commonly used interchangeably to refer to "application", and "information" can translate to "knowledge possessed". Both items were clarified during cognitive testing.
Studies with cancer patients may be missing dataand higher rates of responses may be missing in palliative care studies compared to other studies. Data for the baseline and follow-up assessment were impressively complete for most of the items. The baseline assessment of items on depression, information, anxiety of friends and family, and sharing feelings with friends and family generated the most non-responses. It is possible that patients may be struggling to come to terms with their cancer diagnosis or struggling to express the emotional impact of cancer on themselves, friends and family. Also, participants may have felt uncomfortable expressing their true responses to the item on "information" as not to offend the community palliative care staff caring for them, as was observed in the Japanese IPOS validation study.
Evaluation of SRMR and CFI values of the final CFA model could not confirm or negate that the threedomain structure proposed in the original IPOS was an acceptable fit to our data, in the original validation study 2-factor solution was implicated to be better fitting. Even though CFI and SMRM parameters approached the minimums, they were not within the required defined parameters recommended for small samples. Future larger studies may be needed to confirm the three-domain structure. Emotional subscale, specifically had poorly loading items. Lack of consensus on cut off criteria for factor loadings in studies with small samples, necessities future larger studies before further actions are taken in relation to removal of items.
Turkish IPOS is acceptable, with no floor or ceiling effects. The mean and standard distribution of scores are consistent with findings from a recent validation study in cancer patients. Turkish IPOS illustrated moderate to marginally moderate internal consistency and good to moderate reliability with its Physical and Social and Quality of Care subscales. However, the Emotional sub-scale showed poor internal consistency and reliability. These results agree with the findings in the original validation study, which reported low internal consistencies for Emotional and Social and Quality of Care subscales. In line with observations from the CFA, the removal of the item 'Have any of your family or friends been anxious or worried about you?' is indicated for improving internal consistency.
Content and face validity, ascertained through the translation and cross-cultural adaptation phase as well as at the validation phase, was also good as all of the main concerns such as feeling worried about treatment, the future, loss of independence and dying, the impact of the illness on the family, financial impact of illness, issues with getting support and pain were the main concerns brought up in the open-ended questions were covered in IPOS. Construct validity was demonstrated with significant but moderate to low correlations with EQ-5D-3L items and VAS scale in the predicted direction. These findings were consistent with the findings of the Japanese IPOS. Turkish IPOS also demonstrated known-groups validity as cancer patients at advanced stages had significantly higher total IPOS scores.
Turkish IPOS was able to detect statistically significant changes in concerns and symptoms between baseline and follow-up assessments. The effect size statistics were large. These findings supported internal responsiveness of the Turkish IPOS. The improvement in the total score was mostly due to significant improvement in the physical and quality of care/social support aspects. However, small but significant deterioration was reported with the emotional subscale. The general improvement or deterioration in the scores could be the effect of home palliative care services, but further studies are required to understand these changes in the context of clinical and subjective meaning.
There are further limitations of this study which need to be acknowledged. Patients were recruited through convenience sampling and may not be representative of all cancer patients. Additionally, patients receiving community palliative care services from the cancer charity may be more financially disadvantaged compared to the overall population of cancer patients. Patients with breast cancer were over-represented in the study as the charity was established by a breast cancer survivor and attracts more memberships among women living with breast cancer. Also, more women participated in the study as men may be less willing to share their experiences of having cancer. These limitations may affect the generalizability of the findings. The potential regional differences in Turkish language must also be taken into consideration. For this reason, minor amendments might be needed before Turkish IPOS is used in other Turkish speaking communities.
A team of researchers in Turkey have recently completed the cognitive testing of Turkish IPOS and have confirmed initiation of validation work with no revisions or modifications. Further research is needed to establish the range of Turkish IPOS scores that indicate clinical Correlation is significant at the 0.01 level change meaningful to the patient and to establish its reliability and validity in non-cancer patient populations.
# Conclusion
Turkish IPOS is simple to complete and clearly understood. It measures the important aspects of needs and concerns that are relevant to Turkish speaking cancer patients and it is acceptable. Physical and Social and Quality of Care subscales are internally consistent and reliable, can be used to evaluate needs and concerns in these areas.
There is good evidence that Turkish IPOS is a valid measure of concerns and needs, and responds to the severity of illness and is able to measure change over a specified time frame. Future studies are necessary to investigate the performance of the Emotional subscale, to clarify the factor structure and to evaluate the extent to which change in a IPOS relates to corresponding change in a reference measure of clinical or health status. Observing regional differences, Turkish IPOS may have implications for use in other Turkish speaking communities. |
Hyperlipasemia and Potential Pancreatic Injury Patterns in COVID-19: A Marker of Severity or Innocent Bystander?
## C oronavirus disease 2019 (covid-19) is caused by
severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Gastrointestinal symptoms in COVID-19 are being reported increasingly.Data on the involvement of the pancreas in COVID-19 have been emerging, and multiple case reports of SARS-CoV-2-induced acute pancreatitis have been published in the literature.Angiotensin-converting enzyme-2 is the target receptor of SARS-CoV-2 and is expressed abundantly by both the exocrine and endocrine pancreatic tissues.The presence of angiotensin-converting enzyme-2 in the pancreas could make it susceptible to SARS-CoV-2, resulting in interstitial leakage of pancreatic lipase, adipose tissue lipolysis, and potentially toxic fatty acid-induced damage. These changes can at least contribute to cytokine storm, multiorgan dysfunction, and COVID-19 morbidity.Due to its nonspecificity, lipase could be elevated in a myriad of conditions, such as infections, renal dysfunction, medication-related, gastrointestinal, and hepatobiliary disease.Given this, it is critical to evaluate the prevalence of hyperlipasemia in COVID-19 and predict clinical outcomes.
# Methods
We conducted a systematic search using PubMed, Embase, Ovid, and Google Scholar databases from December 1, 2019 to October 9, 2020, to evaluate hyperlipasemia in patients with COVID-19. The following search terms were used: COVID-19, SARS-CoV-2, lipase, pancreatic injury, and pancreas. The articles with relevant data on the prevalence of hyperlipasemia and its effect on COVID-19 severity were examined. All adult patients with nasopharyngeal reverse transcription polymerase chain reaction positive for SARS-CoV-2 were included in the analysis. Severe COVID-19 was defined as clinical deterioration resulting in adverse clinical outcomes, such as admission to the intensive care unit (ICU), need for mechanical ventilation, or death. Hyperlipasemia was defined as any elevation in lipase levels above the upper limit of the normal reference level. The definition of hyperlipasemia varied among the studies because of differences in the range of lipase levels, as did the definition of severe COVID-19.
OpenMeta[Analyst] software was used to estimate the pooled prevalence of lipase elevations among patients with COVID-19 and the pooled odds ratio for severe COVID-19 among this subset of patients. Results were reported with a 95% confidence interval, and a P value of <.05 was considered statistically significant. Heterogeneity was assessed using the I 2 test, and I 2 >50% was taken as a measure of moderate inter-study variation.
# Results
The initial search yielded 52 articles. After excluding duplicates and review articles, 7 studies 3,6-10 (6 retrospective observational studies and 1 prospective observational studywere included in the pooled analysis. A flow chart depicting the study screening and selection process is presented in.
Data on the point prevalence of hyperlipasemia were available in all 7 studies, whereas only 4 reported clinical outcomes, ICU admission status, and need for mechanical ventilation or death. The normal range of serum lipase levels differed among the studies. Although 4 studies had an average upper limit of 50-60 U/L, 2 had a higher cutoff (>300 U/L) for lipase levels. Data for 756 patients with COVID-19 were reported in these studies, of which 92 had hyperlipasemia. All of the studies were single-center experiences except two. Two studies defined severe COVID-19 as patients needing ICU admission, 1 study as a need for mechanical ventilation, and only 1 study used a combination of all of the above three to define severe COVID-19. Supplementary outlines the baseline characteristics of the included studies.
The results of this pooled analysis are shown inand B. Among 756 patients with COVID-19 with available lipase levels, the pooled prevalence of hyperlipasemia was 11.7% (95% confidence interval, 0.094-0.140; P ¼ .001; I 2 ¼ 0%). The pooled odds ratio for severe COVID-19 in these patients was 3.143 (95% confidence interval, 1.543-6.400; P ¼ .003); mild inter-study heterogeneity was observed (I 2 ¼ 27%).
# Discussion
Based on the results of our pooled analysis, hyperlipasemia was found in 11.7% of patients affected by SARS-CoV-2. Patients with COVID-19 with hyperlipasemia are at an approximately 3-fold higher risk of poor clinical outcomes, including need for ICU admission, mechanical ventilation, or death.
Although multiple mechanisms have been proposed for pancreatic injury in COVID-19, the exact etiology remains unclear. Some of the mechanisms include direct pancreatic tissue damage by SARS-CoV-2 and intense inflammatory response (interleukin 1b, interleukin 6, and tumor necrosis factors) with cytokine storm-mediated tissue injury. Furthermore, studies found that patients with COVID-19 with pancreatic injury had a higher prevalence of severe illness on admission; lower levels of CD3 þ and CD4 þ T cells; and higher levels of aspartate aminotransferase, g-glutamyl transferase, creatinine, lactate dehydrogenase, and erythrocyte sedimentation rate.Limitations of this study include a modest sample size of 756 patients with COVID-19, of which 92 patients had elevated serum lipase levels. The degree of hyperlipasemia was not uniform across all studies. Potential confounders, such as age, comorbidities, and medication use, could alter the results of this study. Furthermore, a lack of highquality randomized controlled trials with adjustment for potential confounders is a notable limitation. Given the inclusion of observational studies, selection bias, information bias, and confounding bias are possible. The prevalence of hyperlipasemia could be underestimated due to a lack of testing and nonreporting of the data for many patients.
Severe pancreatic injury resulting in acute pancreatitis might not be a common event in COVID-19. As evidenced by lipase elevation, mild to moderate pancreatic injury is a clinically significant finding in these patients. Future prospective studies are warranted to ascertain the exact impact of lipase elevation in COVID-19 and guide management strategies for these patients.
# Supplementary material
Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at http://doi.org/10.1053/ j.gastro.2020.10.037.
[fig] Figure 1: (A) Pooled incidence rate of hyperlipasemia in patients with COVID-19. (B) Pooled odds ratio of severe COVID-19 in patients with hyperlipasemia. [/fig]
|
Characterisation of the Subaquatic Groundwater Discharge That Maintains the Permanent Stratification within Lake Kivu; East Africa
Warm and cold subaquatic groundwater discharge into Lake Kivu forms the large-scale density gradients presently observed in the lake. This structure is pertinent to maintaining the stratification that locks the high volume of gases in the deepwater. Our research presents the first characterisation of these inflows. Temperature and conductivity profiling was conducted from January 2010 to March 2013 to map the locations of groundwater discharge. Water samples were obtained within the lake at the locations of the greatest temperature anomalies observed from the background lake-profile. The isotopic and chemical signatures of the groundwater were applied to assess how these inflows contribute to the overall stratification. It is inferred that Lake Kivu's deepwater has not been completely recharged by the groundwater inflows since its turnover that is speculated to have occurred within the last *1000 yrs. Given a recent salinity increase in the lake constrained to within months of seismic activity measured beneath the basin, it is plausible that increased hydrothermal-groundwater inflows into the deep basin are correlated with episodic geologic events. These results invalidate the simple two-component end-member mixing regime that has been postulated up to now, and indicate the importance of monitoring this potentially explosive lake. 3 / 21 the depths specified. The temperatures at the point of maximum difference is visually represented by the colour scale in the upper right hand corner of each image. The scale is increasing for cold SGD, and decreasing for warm SGD. The profile points are plotted with the coldest temperature on top for A-B, and warmest temperatures on top for C-H. The high variability observed in the temperatures in some locations, particularly for source B, is due to the temporal and spatial changes in the depth of the gradient (G3).
# Introduction
Lake Kivu formed its current structure following its * 400 m lake-level rise and initiation of hydrothermal activity * 10 ka [bib_ref] The late Pleistocene and Holocene stratigraphy and paleolimnology of Lakes Kivu and..., Haberyan [/bib_ref] [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref]. The lake consists of a permanently stratified anoxic monimolimnion that reaches 486 m depth, and a seasonally stratified mixolimnion within the top 65 m. Several sources of subaquatic groundwater discharge (SGD) into the northern part of the Main Basin of Lake Kivu are responsible for maintaining this form. A disruption of this stratification from a mixing event could result in the expulsion of a detrimental amount of CO [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref] and CH 4 that have been accumulating within the monimolimnion. These gases have been accruing for over hundreds of years with the current partial pressure of CH 4 equivalent to 60 km 3 at standard temperature and pressure, and CO 2 equivalent to 300 km 3 at standard temperature and pressure. The last lake-mixing event is hypothesized to have occurred as recently as * 750-1000 years ago [bib_ref] The history of subaquatic volcanism recorded in the sediments of Lake Kivu,..., Ross [/bib_ref]. Currently, the possibility of a lake-turnover, consequent of gas oversaturation within the monimolimnion, is particularly limited by the steep density gradient observed at * 260 m depth and the hydrostatic pressure in the deepwater.
The persistence of this strong density stratification in the monimolimnion requires (1) sustained hydrothermal inputs (warm SGD) that are maintained at specific depths by freshwater inputs (cold SGD), and (2) a dilute mixolimnion maintained by cold SGD and surface inflows [bib_ref] Stratification, mixing and transport processes in Lake Kivu, Schmid [/bib_ref]. Three major density gradients (pycnoclines) are observed below the pycnocline that separates the mixolimnion from the monimolimnion. Increasingly warm SGD has been lifting the lower boundary of these pycnoclines since the onset of measurements in 1974 [bib_ref] Comment on An additional challenge of Lake Kivu in Central Africa-upward movement..., Schmid [/bib_ref]. A recent warming of the deepwater of the lake has been observed, which is probably attributed to increased temperature or discharge of the SGD [bib_ref] Recent warming of Lake Kivu, Katsev [/bib_ref]. The warm groundwater consists of salt and CO 2 that sustains the density stratification, whereas CH 4 produced by microbial activity in the anoxic layer weakens the stability of this stratification [bib_ref] Weak mixing in Lake Kivu: New insights indicate increasing risk of uncontrolled..., Schmid [/bib_ref]. A similar stratification scheme is observed for the separate basin Kabuno Bay. However, the deepwater of this basin has a higher conductivity and dissolved CO 2 concentration relative to even the deepest waters in the Main Basin of Lake Kivu [bib_ref] Water and gas chemistry at Lake Kivu (DRC): Geochemical evidence of vertical..., Tassi [/bib_ref] [bib_ref] Variability of carbon dioxide and methane in the epilimnion of Lake Kivu, Borges [/bib_ref].
The sources of SGD into the monimolimnion account for * 20% of the total inflows [bib_ref] Stratification, mixing and transport processes in Lake Kivu, Schmid [/bib_ref]. The absolute discharge of these sources was estimated via a 1D advective-diffusive model applied to reproduce the CH 4 profile and the salinity gradients that define the major pycnoclines [bib_ref] Weak mixing in Lake Kivu: New insights indicate increasing risk of uncontrolled..., Schmid [/bib_ref]. The model required steady-state conditions within the density stratified layers of the lake, and is therefore overparameterised to a certain extent. Upward advection as a result of SGD yields an estimated residence time of 800-1000 years for the gases within the monimolimnion [bib_ref] Weak mixing in Lake Kivu: New insights indicate increasing risk of uncontrolled..., Schmid [/bib_ref]. In addition to advection, double-diffusive convection in Lake Kivu modifies the gas, salinity, and temperature structure produced by these sources. Double-diffusion works to level out the steep density gradients, and removes heat from the deepwater faster than the salts and gases [bib_ref] Double-diffusive convection in Lake Kivu, Schmid [/bib_ref]. The differences in diffusion between heat, and gas/salt further stabilise the stratification in the deepwater, thereby lowering the possibility of degassing from the increased hydrothermal inflows [bib_ref] Double-diffusive convection in Lake Kivu, Schmid [/bib_ref] [bib_ref] Double-diffusive interfaces in Lake Kivu reproduced by direct numerical simulations, Sommer [/bib_ref].
Although SGD plays a vital role in maintaining the structure of Lake Kivu, the subsurface hydrology in the Kivu Basin has not been studied. The water table geometry of the groundwater system is probably very complex, as is observed in lake-groundwater studies in other tectonically active regions [bib_ref] Escape of volcanic gas into shallow groundwater systems at Unzen Volcano (Japan):..., Ohsawa [/bib_ref] [bib_ref] Exploration and discovery in Yellowstone Lake: results from high-resolution sonar imaging, seismic..., Morgan [/bib_ref] [bib_ref] Strontium isotopes and rare earth elements as tracers of groundwater-lake water interactions, Bwire Ojiambo [/bib_ref] [bib_ref] Comparative assessment of the water balance and hydrology of selected Ethiopian and..., Ayenew [/bib_ref]. Major fractures to the north of Lake Kivu align from the summit crater of Nyiragongo into the Main Basin, where similarly lineated faults are observed in the bathymetric map and high resolution seismic profiles of the sublacustrine setting [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref]. The inhomogeneity of the geologic-surrounding makes classical hydrogeology work using piezometers not applicable in the Lake Kivu Basin. However, understanding the geochemical properties of groundwater coupled to the lake is a crucial factor in determining the future structure of the lake.
The aim of the present study was to determine the locations and chemical properties of SGD in Lake Kivu as a base for assessing the relevance of the different SGD sources for driving physical and biogeochemical processes in the lake. Our investigations were based on CTD-profiles and chemical/isotope analyses conducted on groundwater/lake water from within the lake, and assuming for the interpretation a horizontal homogeneity of the lake water [bib_ref] Stratification, mixing and transport processes in Lake Kivu, Schmid [/bib_ref]. The application of geochemical tracers assisted in determining the groundwater-lake water system dynamics and validate the presence of SGD. The isotopic ratios helped to decipher different water masses in addition to determining the subsurface environment containing the groundwater, and where the groundwater is recharging. Eight sources of SGD were located using conductivity and temperature profiling over the course of two years. The aquatic chemistry, including the δ 18 O and δ 2 H isotopic values of the groundwater, were used to characterise the discharge of these sources. A complex mixing-regime is hypothesized to exist within Lake Kivu, which reacts to episodic geologic events that increase the heat, salt, and CO 2 content of the deepwater. This has implications for the present and future socio-economics of the region; including the possibility of a lake overturn event and the current extraction of methane from the deepwater for power production.
## Regional and hydrological setting
Lake Kivu is situated at an elevation of 1463 masl. It has approximately a 1:2 ratio of lake-water (2385 km 2 ) to watershed (4940 km 2 ) area, and a lake volume of 549 km [bib_ref] The history of subaquatic volcanism recorded in the sediments of Lake Kivu,..., Ross [/bib_ref]. There have been no rivers in the relatively flat, 685 km 2 , area to north of the lake ('Virunga lava field' in [fig_ref] Fig 1: Fig 1 [/fig_ref] observed within at least the last decade. Recent hydroclimate data suggests drought-like conditions within the last few decades in this region [bib_ref] Multidecadal variability in East African hydroclimate controlled by the Indian Ocean, Tierney [/bib_ref] that is also recorded in the Kivu Basin, albeit with considerable interannual fluctuations [bib_ref] Modelling Lake Kivu water level variations over the last seven decades, Muvundja [/bib_ref]. Subaerial and subaquatic volcanic A total of 290 profiles were taken in the Main Basin (orange and blue), and 11 profiles in Kabuno Bay (green). The background profiles (blue) were used to calculate an average profile for determining the temperature differences at the locations of SGD. A total of 7 sediment samples were analyzed to determine the geochemical differences at the locations of cold (KS7, 8) and warm (KS17, [bib_ref] Modelling Lake Kivu water level variations over the last seven decades, Muvundja [/bib_ref] [bib_ref] Balancing nutrient inputs to Lake Kivu, Muvundja [/bib_ref] SGD relative to the background (KS4, 5, and 21). The pink outlined box represents the location analyzed for temperature differences of SGD in [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref]. The Virunga Lava field north of the basin, which is void of surface flows, could represent an area of high aquifer recharge. (Inset: Location of the Main Basin in Lake Kivu; Albertine Rift Valley in eastern central Africa) activity in the Lake Kivu Basin has defined the current structure of the lake, and the presence of subaquatic volcanic features has been outlined by previous research [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref]. Lake Kivu is recharged predominately by direct precipitation (1404 mm/yr), followed by input of more than 100 streams and rivers [bib_ref] Balancing nutrient inputs to Lake Kivu, Muvundja [/bib_ref]. The outflow of the lake is maintained by the Ruzizi River discharge that drains into Lake Tanganyika, and by evaporation [bib_ref] Balancing nutrient inputs to Lake Kivu, Muvundja [/bib_ref]. The total hydrological output of 6.3 km 3 /yr; 2.7 Ruzizi + 3.6 evaporation, and the inputs (1.6 rivers + 1.3 SGD + 3.5 precipitation) are given by Muvundja et al. [bib_ref] Modelling Lake Kivu water level variations over the last seven decades, Muvundja [/bib_ref]. The overall uncertainty in the water budget is approximately 15 to 20% and the seasonal and annual variability in the lake's volume is within this range of 1 km 3 /yr [bib_ref] Modelling Lake Kivu water level variations over the last seven decades, Muvundja [/bib_ref]. The 1.3 km 3 /yr input by SGD calculated by Schmid et al. [bib_ref] Weak mixing in Lake Kivu: New insights indicate increasing risk of uncontrolled..., Schmid [/bib_ref] is * 40% of the Ruzizi outflow. This value is relatively well constrained by the salt budget, as the SGD is the major source of salt to the lake. The calculated inputs at 180 and 250 m depth are one order of magnitude greater than those below these depths and contribute up to 90% of the total SGD into the lake. The SGD drives an average upwelling rate ranging between * 0.15 m/yr at the main pycnocline and * 0.7 m/yr in the upper monimolimnion. This upwelling is the main driver for the transport of nutrients [bib_ref] Physical and bio-geochemical limits to internal nutrient loading of meromictic Lake Kivu, Pasche [/bib_ref] and dissolved gases [bib_ref] Methane sources and sinks in Lake Kivu, Pasche [/bib_ref] [bib_ref] Carbon cycling of Lake Kivu (East Africa): net autotrophy in the epilimnion..., Borges [/bib_ref] from the deepwater to the mixolimnion. With regard to subsurface drainage, so far no work has been conducted in the Kivu watershed.
# Materials and methods
## Ctd profiling and temperature mapping
A total of 301 CTD profiles were collected from January 2010 to March 2013 using a RBR XR-620 probe [fig_ref] Fig 1: Fig 1 [/fig_ref]. Locations and sampling dates are given in the supplementary S1 No special permissions for field work on Lake Kivu were required, because the study was part of a joint scientific project involving the University of Rwanda (former Kigali Institute of Science and Technology) and the Institut Supérieur Pédagogique de Bukavu (DR Congo). Initial sampling was conducted at different depths along the northern shore of the lake, based on previous observations of temperature peaks in CTD profiles [bib_ref] Response of Lake Kivu stratification to lava inflow and warming, Lorke [/bib_ref] and temperature microstructure profiles [bib_ref] Double-diffusive convection in Lake Kivu, Schmid [/bib_ref] taken in this region. Further profiling was done in the central part of the Main Basin as a background reference, and southward in the eastern part of the lake to confirm the initial assumption that no important SGD can be observed in this region far from the main submerged volcanic structures.
Subsequently, profiles were sampled more densely around locations where significant signals from SGD were found in the initial campaign. Furthermore, a spatial hydrological analysis was calculated using the ARC-GIS tools Flow Direction and Flow Accumulation, on the bathymetric map produced in a previous study [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref]. The resulting stream network was used to trace dense groundwater that was suspected to be flowing along the basin floor via CTD profiling. A subset of initially 12 and subsequently 16 profiles was selected to be sampled every few months in order to evaluate the temporal variability of the signals from the SGD. Most CTD profiles covered the full vertical extent from the water surface to the local bottom depth.
Temperature signals caused by the SGD were detected in the profiles using an algorithm implemented in MATLAB version R-2013a. Each of the profiles were linearly interpolated using a 0.5 m median-average window. A background profile was averaged using a median filter applied to 25 profiles taken in the deep basin [fig_ref] Fig 1: Fig 1 [/fig_ref]. The temperatures and conductivities, corrected to a temperature of 25°C, of all profiles and the background profile are given in the supplementary S2 and S3 Tables, respectively. Within each depth range, the maximum difference in temperature was extracted, where the differences were positive for warm sources of SGD, and negative for cold sources of SGD. The depth of maximum difference was then used to extract the temperature and conductivity of the profile at this depth. The conductivity was not used to locate SGD because of spikes produced in the profile when the probe entered the sediment. These spikes are not filtered from the interpolating due to dragging along the sediment floor during profiling. The resulting temperatures were plotted for analysis using the free software GMT 4.5.8, and the functions described within [bib_ref] Free software helps map and display data, Wessel [/bib_ref] [bib_ref] Free software helps map and display data, Wessel [/bib_ref]. On occasion the small temperature variability observed in is not a direct consequence of the inflows, but a result of the algorithm used for determining the temperature differences in combination with a slight difference in the depth of the gradient at different locations. It is suspected that these minor depth differences are an effect of internal waves observed within the lake.
## Chemical and isotopic analysis
Sampling for chemical analysis was conducted over the course of three field campaigns in January 2010, October 2010, and January 2012. The results of the chemical analysis of all individual water samples are given in the supplementary S4 Water samples were collected with 5-liter Niskin bottles. The exact depths of the water samples at the sources of SGD were determined by consulting the RBR probe data prior to analysis, which was attached to the bottom of the Niskin bottle. For the areas of warm groundwater discharge, and for depths below 200 m, the open valve at the top of the Niskin bottle was capped with a balloon to prevent sample loss from vigorous outgassing upon ascent. Samples were filtered immediately with a 0.4 micron Whatman TM disc filter for analysis of major cations and anions. Sample portions were acidified with ultra pure HNO 3 for metal analysis, and portions were also oxidized with ultra pure H 2 O 2 and shaken in order to convert H 2 S for total sulphide (∑S) measurements. The latter samples were measured for SO2− in the lab. Major ions were measured at Eawag in Switzerland by ICP-MS (Perkin-Elmer ELAN 5000), and ICP-OES (SPECTRO-CIROS), and on a Metrohm ion chromatograph (733 IC Separator Center, 732 IC Detector, 762 IC Interface). Averages were taken from multiple methods of analysis for most of the major cations and anions. The δ 18 O and δ 2 H isotopic values were determined using a Picarro L1102-i laser analyzer, and the δ values are reported in ‰ relative to the standard SMOW. Chemical analysis of poor quality and major outliers were discarded prior to any interpretation of the data. Alkalinity was titrated in the field with the use of a 716 DMS Titrino TM (Metrohm) and 0.1 mol/L HCl. Vertical profiles for pH were recorded in situ in October 2010 for the lake background profile, and at the locations of groundwater inflows, with a Seabird SBE-19 conductivity, temperature and depth (CTD) probe. The pH was calibrated onboard the research vessel using a 3-point calibration curve that spanned the lake-water pH values. The CO 2 was calculated from the pH and alkalinity using the free software PHREEQC provided by the USGS. Furthermore, the consistent negative charge balance, which was suspected to be a result of MgCO 3 precipitation prior to the analysis in the lab, was offset by calculating Mg 2+ using PHREEQC. The calculated values result in no difference to the interpretation of the data. Water density was calculated as a function of temperature, salinity, CO 2 , and CH 4 for the background samples, using the Chen-Millero equation [bib_ref] Precise thermodynamic properties for natural waters covering only the limnological range, Chen [/bib_ref]. Here within, salinity is defined as the total of dissolved solids and gases in the water mass. The geochemical analysis for TOC, HCO 3 − , and S of the sediment samples is given by Bhattarai et al. [bib_ref] Local conditions structure unique Archaeal communities in the anoxic sediments of meromictic..., Bhattarai [/bib_ref]. An additional analysis was conducted here on sample KS4 in order to determine the Fe concentration using an Axios, PANalytical wave-length dispersive X-ray fluorescence spectrometer (WD-XRF). The resulting Fe concentration was then compared to the XRF counts within the top 2 cm of the core KV-04 (taken at the same time as KS4), reported in a previous study [bib_ref] The history of subaquatic volcanism recorded in the sediments of Lake Kivu,..., Ross [/bib_ref]. A transfer function was then applied to the counts in the top 2 cm of core KV-21 (KS21), in order to calculate the concentration of Fe here. A different XRF scanner was used for all other sediment samples, and therefore the transfer function was not applicable.
# Results
# Background profile
CTD profiling of the background locations in [fig_ref] Fig 1: Fig 1 [/fig_ref] reveals the first pycnocline that separates the mixolimnion from the monimolimnion from 65 m up to 150 m depth (G1), and the major pycnoclines in the deepwater centred at * 190 (G2), 260 (G3), and 315 (G4) m water depth, with a minor pycnocline centred at * 385 (G5) m depth [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref]. A background profile is also delineated for Kabuno Bay ('KB' in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref] , which is located to the northwest of the Main Basin [fig_ref] Fig 1: Fig 1 [/fig_ref] and separated by a saddle at * 11 m depth. The Kabuno Bay profile is thermally stratified to * 11 m depth at which point the salinity and temperature increase dramatically to 140 m, with two main pycnoclines centred at * 15 and 120 m depth.
In agreement with previous observations [bib_ref] Stratification, mixing and transport processes in Lake Kivu, Schmid [/bib_ref] , the background profiles sampled sufficiently far from the influence of SGD showed high horizontal homogeneity and little temporal variability. The standard deviation of temperature below 80 m depth in the 25 profiles forming the background profile generally ranged between 0.01 and 0.02°C and reached a maximum of 0.06°C at 80 m depth. The latter was probably largely induced by small vertical changes of the main gradient layer (G3), as a temperature difference of 0.06°C corresponds to a vertical displacement of less than 0.5 m at this depth. Similarly, for specific conductivity the standard deviations were typically around 0.005 mS/cm outside the gradient layers, and between 0.01-0.02 mS/cm inside the gradient layers, with a maximum of 0.08 mS/cm observed at 250 m depth corresponding to a vertical displacement of less than 0.5 m.
## Spatial distribution of sgd
Cold SGD. Two point sources of cold SGD were located in the northeastern part of the Main Basin, directly west of the submerged flank of the phreatomagmatic cone Mount Goma. . The black line is the average background profile (BG). The hatching at the centre represents the depths where water samples were taken in the Main Basin. The shading in the background marks the depth range where the chemical and isotopic values for the samples were averaged; grey shading for the Main Basin, and beige shading for Kabuno Bay (only one sample was analyzed for each of the Kabuno Bay layers). The codes given on the righthand side are the layers analyzed for the parameters listed in [fig_ref] Table 1: Aquatic chemistry [/fig_ref] The location of sources A and B in [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref] are represented by the coldest temperatures observed within the respective frames. Point source A in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref] , at * 190 m depth, deviates from the background temperature profile by 1.52°C and conductivity profile by 0.26 mS/cm. The next three coldest temperatures at the same location and depth as A are on average 0.8°C warmer than the observed point source. Directly to the west of A, point source B is observed in [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref] at a depth just above the main density gradient (G3 in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref]. This source is 2.41°C cooler and has a 1.33 mS/cm lower conductivity than the background profile. Most profiles within 1-2 km of this location present cooler temperatures than the background profile at a similar depth. Furthermore, the coldest temperatures at this depth are typically observed to the southeast of this source, and particularly along the eastern flank of Mount Goma. In addition to the strong signals from the point sources A and B, small negative temperature peaks were observed in numerous profiles in this region in the depth range between G1 and G2 (e.g. negative offset in profiles A, B, and F in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref]. An additional source of cold SGD was expected at the top of the lower pycnocline, G4. Although the point source was not located, a profile * 500 m southeast of source A and B yields a maximum temperature difference that is 0.4°C cooler than the background profile. The sources of cold SGD all appear to be located along a possible fracture where Mount Goma abuts the steep sloping wall to the north [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref].
Warm SGD. A total of six sources of warm SGD (C-H in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref] are located almost entirely in the northwestern part of the basin at the volcanic platform delineated by previous research [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref]. Source C appears as a warm plume in the conductivity and temperature profile at * 50 m depth in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref]. A cluster of profiles revealing this plume are located at the channel that separates Kabuno Bay and the Main Basin by a saddle at * 11 m water depth in [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref]. The warmest signal is 0.74°C warmer and 0.21 mS/cm saltier than the background profile. This plume is observed up to 10 km away from the channel. Since the plume enters the lake in the mixolimnion, where the background profile temperature varies seasonally due to meteorological forcing, the signal of the plume may be influenced by the temperature variation of the mixolimnion water entrained into the plume. The calculated differences in the plume versus the background profile may therefore be affected by seasonal changes unrelated to the effective plume signal. Source D fills a small basin up to 120 m water depth, which is located southwest of the channel that connects Kabuno Bay to the Main Basin in [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref]. There is a sharp temperature and salinity gradient (profile D in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref] that separates the water below this depth from that above it. The water below the gradient is 1.08°C warmer, and 1.98 mS/cm saltier than the background profile. There are two profiles taken directly outside the basin to the south that present similar temperatures, however, additional profiling in the area did not yield the observation of a warm plume from this source. Point source E was observed along the slope of the shoreline to the southeast of source D in [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref] E. This source is located at the bottom of the pycnocline G2 and has a temperature that is 3.56°C warmer, and is 2.99 mS/cm saltier than the background profile (profile E in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref]. However, no warm plume was observed in the profiles within the vicinity of this point source. Directly south of this source, there is a warm plume centred at * 285 m depth that is observed in a cluster of profiles bordering a magmatic dome feature in [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref] , which is described by Ross et al. [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref]. This plume is a maximum of 0.14°C warmer and only 0.02 mS/cm saltier than the background profile. Additionally, a point source just below the depth of this plume (profile F in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref] was observed directly west of the cold source B [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref] , which is 0.42°C warmer, and 0.12 mS/cm saltier than the background profile. The profiles between these two locations are all colder than either the plume or point source itself [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref]. Two other point sources are observed in a * 1 km wide channel just southeast of the same dome feature in [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref] and H, directly below G4 and G5 in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref]. These sources are 0.58 and 2.3°C warmer, and 0.78 and 3.03 mS/cm saltier, respectively, than the background profile. Although the implied plumes from the point sources of warm SGD are
## Aquatic chemistry
The aquatic chemistry of the background profile is presented within layers 1-7 in [fig_ref] Table 1: Aquatic chemistry [/fig_ref] , and 3, which represent the mixed layers of the lake above and below its pycncolines (G1-5 in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref]. A similar representation is given for Kabuno Bay, where 3 layers are presented (KB1-3 in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref].
The density is calculated from the state variables in [fig_ref] Table 1: Aquatic chemistry [/fig_ref] and compared to the density calculated for the sampled sources of SGD A-H. From this, we can determine in which layer we expect the measured sources of groundwater to stratify if disregarding plume formations. The cold sources A and B have densities of 0.9995 and 1.0002 g/cm 3 , respectively. We would therefore expect source A to stratify within the mixed layer 3 and source B to stratify at the bottom of layer 4. In regards to the warm sources of SGD, we expect that source C with a density of [bib_ref] The history of subaquatic volcanism recorded in the sediments of Lake Kivu,..., Ross [/bib_ref] would stratify at the bottom of layer 2. The sampled warm source F will stratify between layer 5 and 6 according to its density of 1.0018 g/cm 3 , however, this warm source was sampled at the location of its plume near the magmatic dome. The sampled sources D, E, and H should all stratify at the bottom of the lake according to their densities, which are greater than the density of 1.0022 g/cm 3 determined in layer 7. Additionally, theses sources are within the density range observed in layers KB2 and KB3 of Kabuno Bay [fig_ref] Table 1: Aquatic chemistry [/fig_ref] Cations and anions measured in the background samples 1-7 increase with depth in the lake at a relatively constant proportionality for Mg, K, Na, and Cl. Furthermore, Ca concentrations within the mixolimnion decreased by a factor of * 6 relative to the decrease in these elements [fig_ref] Table 2: Aquatic chemistry [/fig_ref] , as a consequence of its control by aragonite precipitation. Similar observations were also recorded by Pasche et al. [bib_ref] Physical and bio-geochemical limits to internal nutrient loading of meromictic Lake Kivu, Pasche [/bib_ref] , and Degens et al. [bib_ref] Lake Kivu: structure, chemistry and biology of an East African rift lake, Degens [/bib_ref] , and a single mixing line was inferred. A two component end-member phenomenon was originally observed for the background profile, and samples taken from tributaries and terrestrial hydrothermal springs by Degens et al. [bib_ref] Lake Kivu: structure, chemistry and biology of an East African rift lake, Degens [/bib_ref]. The authors concluded that a single mixing line is a consequence of mixing two water masses; one being the evaporated-surface lake water system enriched in δ 18 O (3.5 ‰), and the other the meteoric-hydrothermal reservoir that is δ [bib_ref] Modelling Lake Kivu water level variations over the last seven decades, Muvundja [/bib_ref] [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] , where the most enriched sample is the warm source E followed by a sample taken within layer 7 of the background. Another significant observation is that Mg:Na and Mg:Ca increase with sampling depth in Kabuno Bay [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref]. Furthermore, the warm source H has a lower Mg:Na ratio but a higher Mg: Ca ratio than the warm source D. The warm source F, sampled at its plume location, plots with the background samples at the same depth. In regards to the cold sources, A has a similar δ 18 O, Mg:Na [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] , and Cl:Mg [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] relative to the warm source C. One sample of source B is relatively enriched in Mg:Na and slightly less enriched in Cl:Mg, similar to the deeper sources [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref].
From the ternary plots in [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] , we observe that all the samples and sources cluster closely together for the Cl-K-Na plot in [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] , with the highest relative Na and Cl and lowest K values observed in the warm source E. More trending is observed in [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] that is similar to that described for [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] , where the deep Kabuno Bay samples cluster separately from the background samples and sources, but more similarly to the deep sources D and H. Furthermore, as the proportion of Mg decreases, there is a concomitant increase in K, and Ca trending towards the background samples and sources in the monimolimnion [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref]. The effect of the relatively high proportion of Cl in source E is observed in [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] , but absent when replacing the Cl with Ca in in [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref].
The redox sensitive elements S, Fe, and Mn are presented in [fig_ref] Table 3: Aquatic chemistry [/fig_ref]. The concentration of S remains constant in layers 1-5 at * 0.17 mmol/L and then increases to * 0.29 mmol/L below G4. Pasche et al. [bib_ref] Physical and bio-geochemical limits to internal nutrient loading of meromictic Lake Kivu, Pasche [/bib_ref] measured a relatively constant concentration of S 2− below 160 m depth (* 0.23 mmol/L), concluding that the precipitation of pyrite is responsible for Fe-depletion from the water column. We observe a molar ratio of Fe:S in the background sediment samples KS4 and KS21 [fig_ref] Table 4: Geochemistry of surface sediment samples [/fig_ref] of * 1:2. This ratio implies the presence of FeS 2 in the sediment, which was also inferred from the downcore measurements of S and Fe conducted in a previous study [bib_ref] The history of subaquatic volcanism recorded in the sediments of Lake Kivu,..., Ross [/bib_ref]. If we assume that all S is present as FeS 2 in the sediment, then we can roughly estimate an increase in S and Fe in the sediment by a factor of * 5 and 6 at the locations of the cold and warm SGD in the Main Basin, respectively. A concentration of 0.001 mmol/L of Fe was measured in the water column below the main density gradient, whereas 1-2 orders of magnitude higher concentrations are observed in the warm sources (D and H in [fig_ref] Table 3: Aquatic chemistry [/fig_ref] , and in one sample from the cold source B. Conversely, Mn concentrations below the mixolimnion remain constant throughout the measured background samples and within the sources of SGD at * 0.006 mmol/L.
# Discussion
## Cold sgd
Cold SGD appears to flow from a fracture along the northern shoreline of the lake. Two sources were identified at this location (A and B) that emerge as point sources. SGD similar to the characteristics of A and B might also be discharging along the eastern border of Mount Goma, where additional point sources of cold SGD were observed. The large variability observed in the profiles at the location of source A up to the depth of the mixolimnion could imply [fig_ref] Table 1: Aquatic chemistry [/fig_ref].
additional sources of groundwater here, including diffusive flow. Source A appears similar in its isotopic and chemical composition to source C that is located at the channel connecting Kabuno Bay [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref] , which suggests a similar end-member. It is apparent from the chemical analysis in [fig_ref] Table 3: Aquatic chemistry [/fig_ref] that source B has a lower δ 18 O isotopic value (1.9 ‰) than the water sampled above it in layer 4 (2.6 ‰), and a higher value than that sampled below in layer 5 (1.1 ‰).
From the combination of its cooler temperature, its density, and the lower cation and anion content relative to that of the background δ 18 O, we infer that this source is stratifying at its depth of discharge and diluting the warm and salty water that is pushed upwards by advection from below. Since the water above this source has a higher δ 18 O isotopic value instead of an expected value between the range of 1.1-1.9 ‰, this would imply that source B is one component in the mixing regime and the lake water a second component. The measured salinities of 2.54 ‰ for source A, and 3.15 ‰ for source B from the samples taken at the location of SGD [fig_ref] Table 1: Aquatic chemistry [/fig_ref] are lower and higher, respectively, than the salinities of 2.71 and 2.46 ‰ estimated from the conductivity profiles in [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref]. This could imply that our samples are contaminated with lake water. Schmid et al. [bib_ref] Weak mixing in Lake Kivu: New insights indicate increasing risk of uncontrolled..., Schmid [/bib_ref] calculated discharge rates of 0.69 and 0.47 km 3 /yr, and salinities of 2.1 and 2.7 ‰ at the same depth as these two sources, respectively. From the estimated salinities given here, we propose that the actual discharge for source A is higher, and for source B is lower, in order to produce the salinities currently observed in the lake under steady-state conditions. This would affect the rate of upward advection estimated in the lake at these depths.
The S content in source B (0.21 mmol/L, [fig_ref] Table 3: Aquatic chemistry [/fig_ref] is nearly equivalent to that found above and below this source in layers 4 and 5. Furthermore, Fe was detected at concentrations 10x higher in one sample at the location of source B (0.015 mmol/L), than that found below G3 (0.001 mmol/L, [fig_ref] Table 3: Aquatic chemistry [/fig_ref]. This implies that Fe manifests from the groundwater discharge at source B. We have no information on the oxidation state of S within the SGD. However, it is well possible that it is discharged as SO2− from the phreatic aquifer likely supplying the cold source here. Further evidence of pyrite precipitation at the location of the cold SGD is contained within the sediment, which has 5x higher pyrite at the depths just below sources A and B [fig_ref] Table 4: Geochemistry of surface sediment samples [/fig_ref].
## Warm sgd
The six sources of warm SGD were located predominately at the volcanic platform to the northwest [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref]. Volcanic gas/fluid-groundwater interactions have often been cited as one of the major factors controlling hydrothermal discharge, and are suspected to be the primary reason for the onset of the density stratification in Lake Kivu [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref] [bib_ref] Hydrothermal origin of metals in some East African rift lakes, Degens [/bib_ref] [bib_ref] Modern and Ancient Lake Sediments, The International Assciation of Sedimentologists, Stoffers [/bib_ref]. From the enrichment of cations and alkalinity, it is clear that the sampled warm sources C, D, E, F, and H, and samples KB2 and KB3 all originate from hydrothermal sources. Mineral leaching by CO 2 -rich geothermal induced waters increases the composition of alkalinity and cations in the system, while maintaining an equilibrium with the pCO 2 [bib_ref] Geochemistry of hydrothermal fluids from Axial Seamount hydrothermal emissions study vent field,..., Butterfield [/bib_ref]. Based on the densities calculated for sources E, F, G, and H, we assume that these sources all contribute to SGD into the deepwater of the Main Basin [fig_ref] Table 1: Aquatic chemistry [/fig_ref]. Furthermore, we assume that source D is filling the basin observed in [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref] , but is likely not overflowing into the Main Basin. Therefore, we have located four sources of warm SGD equivalent to that suggested by Schmid et al. [bib_ref] Weak mixing in Lake Kivu: New insights indicate increasing risk of uncontrolled..., Schmid [/bib_ref] to be discharging below the main pycnocline (G3). However, the high salinities and temperatures of the sources located in our study [fig_ref] Fig 2: The conductivity and temperature profiles observed at the location with the maximum... [/fig_ref] and [fig_ref] Table 1: Aquatic chemistry [/fig_ref] would suggest a significantly lower discharge of these sources than the cumulative discharge of 0.15 km 3 /yr calculated by [bib_ref] Weak mixing in Lake Kivu: New insights indicate increasing risk of uncontrolled..., Schmid [/bib_ref] [8] that contained lower salinities and temperatures.
Source C is measured at 50 m depth directly at the channel that connects the Main Basin to Kabuno Bay, where a dominant plume is observed here [fig_ref] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the... [/fig_ref]. The warm and relatively salty water of this source presents a similar chemistry to that found in the background profile in layer 3 and in source A [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] , which could indicate geothermally warmed groundwater that initially feeds source A. The density of this source indicates stratification within layer 2, however, the dominant plume observed at this location implies that it mixes within the mixolimnion in layer 1. We infer that this source has little effect on the salinity of the mixolimnion compared to river inflows, upwelling from the monimolimnion, evaporation and outflow via the Ruzizi River.
The high proportion of Mg relative to Na, Ca, and K in samples KB2 and KB3 and next in sources D and H, relative to the background samples, indicates that Mg is readily being leached in the hydrothermal system. Source E has the lowest Mg:Na and highest Cl:Mg content suggesting a different hydrothermal source here. These results are coherent with the ratio of [HCO 3 − ]: ] that determines the pH of the system, which is the same for samples KB2 and KB3 (1.2), slightly lower for source D (0.9), and 3x lower for source E (0.4) [fig_ref] Table 1: Aquatic chemistry [/fig_ref]. This suggests a higher geothermal pCO 2 at source E exists compared to that of the system that feeds into Kabuno Bay and sources D and H. Overall, the similarities in the chemistry and isotopic composition of KB2 and KB3, and that of sources D and H, are remarkable [fig_ref] Table 1: Aquatic chemistry [/fig_ref] , 2, and 3). We therefore suggest that the same hydrothermal source contributes to the SGD in Kabuno Bay and the Main Basin, where there is likely a hydrogeologic connection between the two. The decreased Mg:Na, Mg:Ca, and Mg:K (data not shown) in sources D and H relative to samples KB2 and KB3 [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] , could simply be from hydrothermal percolation through the different lacustrine sediment in the Main Basin that is often aragonite rich [bib_ref] The history of subaquatic volcanism recorded in the sediments of Lake Kivu,..., Ross [/bib_ref] [bib_ref] Abrupt onset of carbonate deposition in Lake Kivu during the 1960s: response..., Pasche [/bib_ref]. The encompassing hydrothermal system has a δ 18 O isotopic composition of * -2.7 ‰, where the slightly higher isotopic composition determined in source D (-2.3 ‰) is likely due to dilution by lake water within the basin here. Comparatively, the hydrothermal system of source E has a significantly higher δ 18 O isotopic value (0.7 ‰).
The concentration of S in the sources D, E and H, and Kabuno Bay samples KB2 and KB3, is the same as that found in the background samples in layers 6-7 [fig_ref] Table 3: Aquatic chemistry [/fig_ref]. We assume that S is precipitated immediately as FeS 2 upon discharge, and that the remaining Fe measured forms a complexation with HCO 3 − . With regard to the spatial distribution of the S in the sediment at the location of the warm sources, the highest S and lowest CO concentrations are given at KS18 near source E, which could support the theory that more FeS 2 is precipitated from the sources with a lower HCO 3 − concentration [fig_ref] Table 1: Aquatic chemistry [/fig_ref].
## Groundwater and lake-water mixing dynamics
Conventional hydrochemical techniques are not always easy to apply in a rift environment where subsurface hydrothermal systems commonly exist in proximity to active volcanoes [bib_ref] Escape of volcanic gas into shallow groundwater systems at Unzen Volcano (Japan):..., Ohsawa [/bib_ref] [bib_ref] Exploration and discovery in Yellowstone Lake: results from high-resolution sonar imaging, seismic..., Morgan [/bib_ref] [bib_ref] Aqueous sulfur speciation possibly linked to sublimnic volcanic gas-water interaction during a..., Takano [/bib_ref] ]. Although it is not possible to quantify the rate of SGD into the Main Basin from locating and characterizing the sources alone, we are able to postulate how the sources of SGD contribute to the stratification in the Main Basin based on certain principles. Firstly, groundwater flow systems can be schematized into two major vertical zones: a shallow zone of active, fast water flow and a deeper zone of relatively slower flow and longer residence times. Secondly, the dominance of upward vertical advection, the horizontal homogeneity within the background profiles, and the lack of turbulent diffusion in Lake Kivu allows us to estimate hydrothermal discharge based on a simple box-model. To begin, we hypothesize that within the Lake Kivu Basin there is a zone containing a major unconfined aquifer, with a deeper underlying, volcanic fractured aquifer existing in the rift floor . Within the Lake Kivu catchment, the Virunga lava field [fig_ref] Fig 1: Fig 1 [/fig_ref] is devoid of Conceptual model of the interaction of the SGD with the volcanic/geothermal system present beneath Lake Kivu and north of the lake towards Nyiragongo. The illustrated deep magma reservoir, deep dike, and shallow dike, including depths and extent, were modelled after Wauthier et al.. The warm SGD C, D, E, G, and H trend along a N-W direction towards Kabuno Bay. The cold SGD A and B, and warm source F are located just west of the phreatomagmatic cone Mount Goma [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref]. The blue and red arrows depict the convective groundwater flow paths within the fractured system. The blue arrows indicate shallow groundwater and the red arrows indicate deeper groundwater, which interacts with the geothermal system. surface flows, and precipitation in this area probably contributes significantly to aquifer recharge [bib_ref] Modelling Lake Kivu water level variations over the last seven decades, Muvundja [/bib_ref]. Furthermore, faults connecting this aquifer to the Main Basin have a northeastsouthwest alignment with the Nyiragongo volcano chain [bib_ref] Lake-level rise in the late Pleistocene and active subaquatic volcanism since the..., Ross [/bib_ref]. In general, the hydrologic properties of fault zones are thought to be highly anisotropic. Vertical faults can serve as conduits for horizontal flow along the fault, or barriers to horizontal flow across the fault, or both. Additionally, faults are conduits for vertical flow and can connect aquifers over different depths [bib_ref] Groundwater flow through anisotropic fault zones in multiaquifer systems, Anderson [/bib_ref]. A study conducted in Yellowstone revealed individual thermal systems aligned at depth to form a large, contiguous geothermal reservoir [bib_ref] Exploration and discovery in Yellowstone Lake: results from high-resolution sonar imaging, seismic..., Morgan [/bib_ref]. In Lake Kivu, the cold sources of SGD are probably vertically isolated from the geothermal source , and therefore their temperatures should represent the average ambient temperatures in the region. However, it is possible that the cold SGD along the northern shoreline of Lake Kivu is vertically and horizontally connected to the hydrothermal system. This is indicated by the proximity of the point source F to that of A and B, the one sample of source B that appears similar to that of the warm sources D and H, and the similarity of the warm source C with that of A. Furthermore, the hydrothermal source F may be connected to the plume observed at a similar depth at the volcanic platform. All of which would imply that the hydrothermal and cold groundwater systems are vertically and horizontally connected. The warm sources D and H, and the sources that contribute to the deepwater profile in Kabuno Bay, appear to be discharged from a deep contiguous geothermal reservoir that is trending northwest, and likely includes source G based on its location .
Mixing within a regional groundwater flow system has the effect of averaging the isotopic composition of various groundwaters from different recharge environments. The isotopic composition of groundwater in such cases converges on the mean weighted value of all recharge contributions. If the recharge of regional groundwater occurs at higher altitudes where the average temperatures are lower, precipitation will be isotopically depleted compared to the local aquifers recharged at lower elevations. [bib_ref] Modelling Lake Kivu water level variations over the last seven decades, Muvundja [/bib_ref] O/ [bib_ref] Comparative assessment of the water balance and hydrology of selected Ethiopian and..., Ayenew [/bib_ref] O ratios are further fractionated through rockwater interactions in the aquifer, where the rate of equilibrium reactions will proceed faster in geothermally heated waters [bib_ref] Temperature-dependent self-diffusion coefficients of water and six selected molecular liquids for calibration..., Holz [/bib_ref]. The δ 18 O isotopic values of KB2 and KB3 are the same as those of source H, and therefore preclude isotopic changes from rock-water interactions. The warm source E has a higher δ 18 O isotopic value of 0.7 ‰ that is attributed to a different hydrothermal source, which is likely recharged at significantly lower elevations than H. This is also assumed for the cold source B with a δ 18 O isotopic value of 1.9 ‰. We suggest that the cold sources A and B, and the warm sources C and E are discharging from a local unconfined aquifer. However, source E is connected to a geothermal reservoir that enriches the NaCl and pCO 2 in this source. Furthermore, sources D and H contributing to the lake, and KB2 and KB3 below the main gradient in Kabuno Bay, are probably associated with local deep thermal sources that are isolated from the unconfined aquifer, such as that depicted in .
The contribution of these sources to the stratification of the lake appears to go beyond a simple linear-mixing system. Here we postulate at least a four-component end-member mixing regime: (1) the lake profile following a lake mixing event, (2) SGD sources A, B and C, (3) SGD sources G and H, and (4) SGD source E. It is apparent from the δ 18 O versus δ 2 H isotopic values [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref] that the background profile is trending towards an evaporative source in the seasonally stratified mixolimnion, and a negative source in the deepwater recharged by SGD. The dominance of upward advection in the lake [bib_ref] Weak mixing in Lake Kivu: New insights indicate increasing risk of uncontrolled..., Schmid [/bib_ref] , and the low self-diffusion coefficient of water, preclude significant downward fluxes by turbulent diffusive transport under present conditions. Given the estimated discharge of the SGD [bib_ref] Stratification, mixing and transport processes in Lake Kivu, Schmid [/bib_ref] , the isotopic composition of the deepwater in the lake should have approached that of the SGD since the likely onset of stratification with the lake-level rise some 10 ka BP. A coherent explanation for these contrasting observations is that a mixing event has been effectively transporting evaporated surface water to the deepwater in the recent past. This could also explain why the cold SGD B has isotopically depleted δ 18 O values relative to the layers above this source. If the lake were to mix under the present conditions, the resulting profile would have a δ 18 O value of 2.8 ‰. The contribution of the SGD would then begin to produce the large gradients observed in the lake. We propose that both sources H and E are contributing to the lake stratification, with a higher contribution from source H than E [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref]. Specifically, the layers 5 and 6 appear to receive a higher contribution from source H, while layer 7 receives a higher contribution from source E [fig_ref] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu [/fig_ref].
There has been an * 1% increase in conductivity below 320 m depth observed in CTD profiles taken in the deepwater (data not shown), which is constrained to have occurred between May 2007 and July 2008. Additionally, a higher Cl concentration and alkalinity is measured in the lake below 320 m in this study relative to that conducted by Pasche et al. [bib_ref] Physical and bio-geochemical limits to internal nutrient loading of meromictic Lake Kivu, Pasche [/bib_ref] in 2007. Given the M w 5.9 earthquake beneath the lake on February, 2008, it is possible that the conductivity increase is attributed to this event. Lake Albano has recently increased in CO 2 as an effect of seismic events in the area [bib_ref] Combined effects of dissolved solids and temperature on the density stratification of..., Wüest [/bib_ref] , and a comparative study in Yugama Lake observed an increase in Cl − and SO2− concentrations in correlation with volcanic tremors. In the latter system, several hot springs and cold springs existed in different parts of the catchment, where the hot springs were aligned along major regional faults [bib_ref] Aqueous sulfur speciation possibly linked to sublimnic volcanic gas-water interaction during a..., Takano [/bib_ref]. The complexity of this aquifer system may be similar in dynamics to that found at Lake Kivu, which is illustrated in .
The increased Cl concentrations observed in Lake Kivu will have a greater effect on conductivity relative to increases in HCO 3 − [bib_ref] Magma sources involved in the 2002 Nyiragongo eruption, as inferred from an..., Wauthier [/bib_ref]. However, if the conductivity increase were a result of the sources G and H observed in this study below 320 m depth, then a discharge volume of 1.08 km 3 from source G down to 370 m depth, and a volume of 0.24 km 3 from source H below 370 m depth can be calculated. This cumulative SGD volume of 1.32 km 3 is much larger than the volume of 0.12 km 3 /yr estimated by Schmid et al. [bib_ref] Weak mixing in Lake Kivu: New insights indicate increasing risk of uncontrolled..., Schmid [/bib_ref]. Given that the last lake mixing event was expected to have occurred * 750-1000 years ago [bib_ref] The history of subaquatic volcanism recorded in the sediments of Lake Kivu,..., Ross [/bib_ref] , total SGD over this time period from episodic events would effectively produce the lake salinity-stratification observed today.
# Conclusions
Lake Kivu presents a dynamic system in which its stratification is highly influenced by groundwater inflows. The characterisation of these groundwaters is pertinent to predicting the future stratification of the lake. Net groundwater flow into lakes is typically calculated as the residual in the hydrologic budget, providing no information on the magnitude of inflow or outflow. Determining the location of these groundwater sources and identifying how their inflows contribute to the stratification in Lake Kivu is the first step to quantifying the SGD into the lake. Within the groundwater system in the Lake Kivu catchment, multiple locations of cold and warm SGD likely co-exist with varying heat and solute transport processes. The outcome of this research indicate that the lake conductivity and temperature profiles are a product of episodic geologic events within the lake. Such events implicate the importance of lake monitoring at Kivu. Additional studies should therefore be applied to confirm the temporal and spatial resolution of SGD into Lake Kivu. These results therefore serve as a substantial base for developing a transient model of the past and future density stratification in the lake.
Supporting Information S1 Metadata of the CTD (Conductivity, Temperature, and Depth) profiles used in the present study, including profile codes, and sampling times and locations. The profiles used for calculating the background profile are marked with (BG). (TXT) S2 Temperatures observed in the CTD profiles listed in S1 The profiles were interpolated to 0.5 m intervals. (TXT) S3 Conductivities, corrected to 25°C, observed in the CTD profiles listed in S1 The profiles were interpolated to 0.5 m intervals. (TXT)
S4 Results of the chemical analyses on water samples from Lake Kivu included in the present study. The sample codes correspond to the sampling locations in [fig_ref] Table 3: Aquatic chemistry [/fig_ref] ; BG: background; KB: Kabuno Bay.
(TXT)
[fig] Fig 1: Fig 1. A total of 290 profiles were taken in the Main Basin (orange and blue), and 11 profiles in Kabuno Bay (green). The background profiles (blue) were used to calculate an average profile for determining the temperature differences at the locations of SGD. A total of 7 sediment samples were analyzed to determine the geochemical differences at the locations of cold (KS7, 8) and warm (KS17, 18, 19) SGD relative to the background (KS4, 5, and 21). The pink outlined box represents the location analyzed for temperature differences of SGD in Fig. 3. The Virunga Lava field north of the basin, which is void of surface flows, could represent an area of high aquifer recharge. (Inset: Location of the Main Basin in Lake Kivu; Albertine Rift Valley in eastern central Africa) [/fig]
[fig] Fig 2: The conductivity and temperature profiles observed at the location with the maximum temperature difference for each source inFig. 3 (A-H) [/fig]
[fig] Fig 3: The 290 profiles, represented by the circles, were analyzed to locate the maximum difference of temperature from the background profile at difficult to trace beyond their location of discharge, many of the profiles taken within the deep basin (> 400 m water depth) express small peaks in temperature of up to 0.1°C at the bottom of the pycnoclines G3, G4, and G5. [/fig]
[fig] Figure 2 b: calculated to offset the charge balance error as explained in the methods section c average based on two samples from 2012. A third sample from 258 m depth taken in 2010 was not included in the averaging, see supplementary S4 [/fig]
[fig] Fig 4: Geochemical plots for a complete characterisation of the SGD into Lake Kivu. (a) The δ 18 O v. δ 2 H values represent the end-member mixing within Lake Kivu and are compared to the Global Meteoric Water Line (GMWL). ii) Plot of Mg 2+ /Na + v. δ 18 O, ii) Cl − /Mg 2+ v. δ 18 O, and iv) Mg 2+ /Ca 2+ v. δ 18 O. The gray arrows below each graphic indicate increasing concentrations of the specific elements. (b) Ternary plots of i) Cl − -K + -Na + , and ii) K + -Mg 2+ -Cl − and iii) Mg 2+ -Ca 2+ -K + . The percentage of each element represented by the sample is given by the axis that increases in the direction of the arrow. A legend is indicated at the top and describes the grouping illustrated in Fig. 2 and listed in [/fig]
[table] Table 1: Aquatic chemistry: state variables for density calculations. [/table]
[table] Table 2: Aquatic chemistry: major cations and anions. [/table]
[table] Table -: indicates that the sample was not analyzed, or was discarded, for this parameter ± is the standard deviation calculated from layers or sources where more then one sample was analyzed doi:10.1371/journal.pone.0121217.t002 [/table]
[table] Table 3: Aquatic chemistry: redox sensitive elements and stable isotopes. [/table]
[table] Table c: measured in the sample from 258 m depth from 2010, see supplementary S4Table −indicates that the sample was not analyzed, or was discarded, for this parameter ± is the standard deviation calculated from layers or sources where more then one sample was analyzed doi:10.1371/journal.pone.0121217.t003 [/table]
[table] Table 4: Geochemistry of surface sediment samples. [/table]
|
An Exploration of U.S. Southern Faith Leaders’ Perspectives of HIV Prevention, Sexuality, and Sexual Health Teachings
Reducing human immunodeficiency viruses (
# Introduction
Despite making up only 2% of the U.S. population, Black men who have sex with men (BMSM) account for 75% of all new HIV infections. While just over a third of the U.S. population lives in the southern United States, more than 50% of HIV incidence is reported from this region. Tennessee (TN) requiring collaborative and coordinated partnerships, including the faith community, to address core pillars. Taken together, both national strategies highlight opportunities to explore threats to HIV resurgence. This paper discusses faith leaders' teaching perspectives of HIV prevention, sexuality, and sexual health in their predominantly Black congregations. The goal is to identify potential barriers to inform the development of effective interventions in churches and local Ending the Epidemic implementation plans.
# Materials and methods
## Design, recruitment, and sample
This study employed a descriptive, qualitative design to gain a better understanding of teaching perspectives of HIV prevention, sexuality, and sexual health among a convenience sample of faith leaders pastoring predominantly Black congregations in Memphis, Tennessee. Faith leaders are defined as any member of the ministerial staff appointed by the senior pastor, such as the senior pastors themselves, youth minister, and first lady. A participatory research approach was employed by involving all community-academic partners to address the study objectives, and the project was conceptualized by the community. This partnership consisted of a non-profit, faith-based community organization, an HIV prevention coalition, and a university. Data collection and recruitment involved both community and academic partners. Finally, data were interpreted and prepared for publication using collaborative processes with two community partners included as co-authors.
Multiple strategies to recruit faith leaders were employed, including personal and professional networks. A publicly available e-directory compiling addresses and telephone numbers of approximately 300 Memphis area traditional Black churches was reviewed, and data were extracted. Study announcements were made at local public events. Distribution of study advertisements occurred at churches, community events, health centers, and the local ministerial alliance.
Faith leaders involved in ministry work at a predominantly Black church who self-reported being at least 18 years of age were eligible to join the discussion group. These included pastors, ministry leaders, and other leaders (i.e., youth advisor, assistant health and church clerk, Aides for HIV/AIDS Networking to Guide and Educate Lives (A.N.G.E.L.)) appointed by the pastor. All participants had to have the authority to make decisions about health ministry activities for the church. The University of Memphis Institutional Review Board (PRO-FY 2359) approved the research.
## Data collection
Four focus groups (FG) were facilitated by our trained community collaborator (fourth author), who serves as a Health and Wellness Minister and has engaged over 40 churches in HIV prevention. The size of the groups ranged from 5 participants (FG1, FG2, FG3) to 11 participants (FG4). Two groups were stratified by gender, but not faith leader role. All leaders actively participated, and the facilitator ensured each participant had an opportunity to speak with consideration to the potential variations in hierarchy levels of faith leaders. The first author (study principal investigator) trained her to facilitate the discussions and to conduct the informed consent process. Additionally, the first author took copious notes and audiotaped the discussion to ensure that all verbal and non-verbal forms of communication were captured. Each focus group lasted approximately two hours and was held at a faith-based community health center.
We used a semi-structured focus group interview, in which participants were asked about how sexuality and sexual intercourse were discussed in their church. The interview guide was informed, vetted, and approved by the community-academic partnership. Discussion questions included: "How, if at all, does your church address issues of sex? sexuality? homosexuality?", "What types of messages do you share with members of your congregation about sex, sexuality?" All participants completed a brief survey following the informed consent process and before the focus group discussions began. The survey included demographic questions (e.g., gender, age, education, marital status), denominational affiliation (e.g., Baptist, Methodist, non-denominational), their leadership role in the church they represented, and comfort level discussing sexual health content areas with adults and youth congregational members. To assess comfort level, participants were asked about each content area using the stem, "How comfortable are you discussing the following with the youth and/or adults in your church/faith-based organization?" The topics included sexual behaviors, such as anal, oral, and vaginal sex, sexuality, and condom use. Response options ranged from 1 to 4, where 1 represented not comfortable, and 4 signified very comfortable. Faith leaders were compensated with a $40 gift card for their time.
# Analysis
Focus group discussions were transcribed verbatim. Using the research question as a guide, the community and academic partners took a deductive thematic approach to develop the codebook. There were 39 codes in the final codebook. To begin the coding process, each transcript was coded independently by two members of the research team. Next, the two researchers discussed the categories each had independently identified through open coding during their regularly scheduled weekly meetings. The weekly meetings were convened to ensure the trustworthiness and dependability of the research findings. Resolution of any coding discrepancies between the 2 researchers involved a consensus process. This allowed the researchers to review the transcript text together, jointly focus on the words, and to best identify concepts raised by the faith leaders.
# Results
The final sample included 26 faith leaders across 23 churches-9 Pastors, 11 Ministry Leaders, and 6 others (i.e., youth advisor, church/health clerk, A.N.G.E.L.). The mean age was 54, and 65% of the sample was female. Christian denominations represented in the sample, included African Methodist Episcopal (3.8%), Baptist (65.3%), Church of God in Christ (23.1%), and Non-denominational (7.7%) faith traditions. Tables 1 and 2 display the distribution of comfort level discussing sexual behaviors (e.g., anal, oral, and vaginal sex), sexuality, and condom use. Overall, faith leaders felt "comfortable" or "very comfortable" discussing key sexual health behaviors as they relate to HIV risk. Three themes emerged to describe the faith leaders' perspectives of HIV prevention, sex, and sexuality teachings after a thorough evaluation of focus group discussions. Each theme is described below in detail. Interestingly, participants often self-selected to discuss issues of sexuality akin to homosexuality; hence the terms were sometimes used synonymously.
## Restricted by scripture: "not a lot of room in our teachings for homosexuality"
Scripture is the inspired guidance that orchestrates the worship process and experience and guides the teachings. Faith leaders in our focus groups believed scripture should guide how sex is discussed in church. One participant said:
"We go back in biblical teaching concerning relationships. We feel that you should be married couples before you indulge in sexual activities". Across all four focus groups, this proved to be the general consensus, that premarital sex was strongly discouraged. However, almost all acknowledged the reality of unmarried church members still engaging in sex. Faith leaders described a need to hold the congregation accountable and make them aware of the consequences of sex outside of marriage. Faith leaders discussed nurturing relationships by actively seeking out certain groups, such as the youth, to provide sexual health information and support. Examples of how youth can confidentially call on them for condoms and other advice off church premises were given. Nearly all faith leaders were aware of these premarital sexual relationships. As a result, they believed it was their responsibility to communicate to youth the costs of sex outside of marriage, such as sexually transmitted diseases (STDs).
Participants made a firm distinction about what constitutes a married couple. One participant summarizes their views which were typical of other participants:
"He (God) made a man for a woman and a woman for a man".
Participants explained that different interpretations of biblical guidance and church doctrines make sexual health teachings difficult, and that church bylaws do not permit conversations about same sex relationships. For context, bylaws (framework) support operations and structure. Some religious denominations have a book of discipline that sets the rules that have been decided at the annual general conference meeting. The churches in this denomination must abide by those rules. These rules have specific instructions on topics such as marriage. One faith leader explained: "There is not a lot of room in our teaching for homosexuality. Homosexuality in our bylaws is not received".
Hence, if the topic is discussed, it is limited within the purview of immoral biblical teachings. As illustrated by these quotes, denominational bylaws can perpetuate stigma towards homosexuality, which may limit the ability of a church to address HIV fully.
One participant quoted Mathew 28:20, which reads, "Teaching them to observe (obey) all things whatsoever I have commanded you" as the role of the faith leader in serving congregations. This implies you act like Jesus, especially in how you treat people, and you live your life by the scripture. However, it was universally clear across participants that there were limitations to what is discussed as it relates to same sex relationships and HIV prevention. As one participant noted:
"We have had members who were living with HIV and they have been embraced . . . but I'm not saying anything about condoms".
Teaching compassion like Jesus, caring for and providing support to those who are living with HIV, is consistent with biblical scriptures. The faith leader did not hesitate to embrace someone that was ill with compassion but was not willing to say anything about the promotion of condoms because it would exploit perversion, which may inadvertently increase poorer health outcomes. These limitations have a direct impact on church engagement in HIV prevention messaging.
## Problematic silence: "they don't talk"
Participants hinted that the silence and secrecy in congregations around sexual activity in congregations make it difficult to provide HIV prevention and sexual health information. The most notable dangerous depiction of silence was the intersection of poverty and sexual risk behaviors. One pastor explained:
"Many of the people we deal with were brought up in poverty, and it has its own culture. If you're in a culture where you don't have a lot of money what do you promote-your, sexuality. I'm a man, you know, I can make babies. That becomes a dominant thing. The more babies you got, the more man you are, you know, in their own thinking".
Participants were certain that limited employment opportunities and education in this community also affected their ability to openly address HIV and other health problems. Several participants also pointed to young, unwed mothers, as further evidence of the damaging effects of being silent about sexual health teachings. They believed that not discussing sex and sexuality left young people uninformed and less equipped to protect themselves in sexual relationships.
Participants also acknowledged the danger of silence and secrecy among individuals in some of the more rural parts of the Mid-South region (Northern Mississippi, Memphis, TN, North Eastern Arkansas). They believed that adults living with HIV were not always willing to ask for help, as their preference was to keep their HIV status private. As one participant stated:
"But in the rural areas people are very secretive. They don't talk so that needs to be broken. If you can break the shell of that you can get more education to them. But they're very secretive and they will not let you into what's going on with them so that's one of the things that I picked up on".
As noted above, enhancing HIV educational efforts is a critical, yet sometimes challenging strategy to encourage more conversations about HIV within closed networks. Another suggested strategy to disrupt the silence surrounding HIV was to personalize HIV when increasing congregational awareness about HIV risk and prevention. The personal relevance of HIV was described as a determinant of support for those infected or affected by HIV. In general, the congregation's concern about HIV is low if they are not affected personally. One pastor explains:
"Until it hits home you really don't pay attention until, one of your friends have it, your cousin, somebody. That's when you want to talk about it".
As a teaching tool to break the silence, several participants discussed the impact of inviting community members to share their personal testimony and how their message influenced church members to get educated and know their status.
## Tackling multiple stigmas: "the church has to stop judging and start ministering"
There was consensus across focus groups that stigma towards homosexuality was a clear obstacle to fully implement HIV prevention activities. Many endorsed a 'one man and one woman doctrine'. However, participants recognized the limitations and dangers of this perspective in their efforts to address HIV fully. One participant shared:
"I think that's part of the reason why the church has been sort of negligent in addressing the issue around HIV because of the church's stance on homosexuality".
Homosexuality was considered a sin by all participants. Faith leaders discussed how sins are sometimes ranked. Almost unanimously, participants stated that homosexuality is believed to be the worst sin in the eyes of many churches and many congregations. Faith leaders felt like people needed to resist the temptations of homosexuality, as if sexuality was a personal choice. However, there still remained a level of tolerance and compassion for the individual despite their sexuality or sexual behaviors. One pastor discussed how his congregation had cultivated an atmosphere of tolerance for those members who are different.
Other faith leaders discussed ministerial counseling resources on their church grounds to address some of the hurt, pain, and life issues congregations experience. There was concern that congregants who feel judged for their sexuality may experience threats to their mental wellbeing. One participant explained:
"The highest rate of suicide now is among lesbians and gays and, also veterans. The alienation that one gets, even with the lifestyle, but they're human".
Dealing with physical and mental wellbeing facilitates spiritual wellbeing according to faith leaders. Pastors believed it was important as leaders to be cognizant of overall wellbeing and be prepared to minister to the whole person, including the spirit and mind, through worship experiences, small group ministries, and health fairs.
While the issue of stigma toward homosexuality was raised, the impact of stigma on HIV prevention and behavioral risk factors also was discussed in each group. Almost all participants had been involved in or held a church health fair. One woman described common behaviors of congregations as a by-product of the impact of HIV stigma: "I say the stigma's there. We will have brochures on sexuality, HIV/AIDS, risk factors, and everybody that will pass by, like, they don't see it . . . But if you leave those out and after everything is over you go back they're all gone because people do want to know, but it's just the stigma. They don't want to be seen there".
It is clear that congregants are eager to receive HIV related information but are reluctant to openly express their desires for more information because of the stigma associated with HIV. Seeking information about HIV may raise unwarranted judgments about congregants' sexual behaviors. Another participant perpetuated a common belief held by faith leaders and one that is often disseminated to congregations and communities. She explained to the group: "The guys are not going to come out and talk openly about it because of the stigma that's attached to the homosexual behavior that's going on in the jails and it is a lot of it, believe me. I mean our young men-African American males are participating in this behavior big time in the jails. Then they come out of the jails, get back with their girlfriends and that's why you see the numbers".
These salient examples shared by faith leaders illustrate the influence of stigma associated with HIV in faith contexts. Although faith leaders are acknowledging the impact of stigma on health-seeking behaviors, they were unaware of interventions or messages that would decrease stigma. Thus, faith leaders are in agreement that leaders in the church would benefit greatly from education to combat stigma.
# Discussion
There is a growing body of literature in faith-based HIV research that suggests the field has moved far beyond simplistic conversations of abstinence only, no sex before marriage, and homosexuality as a sin. However, the vast majority of this work has been conducted in more progressive cities outside of the Deep South, such as Baltimore, Los Angeles, Kansas City, New York City, and Philadelphia.
Church leadership faces its own unique concerns in the South, which is traditionally thought to be more conservative than other regions of the country. The reality remains that there are major challenges with addressing the topics of sex, sexuality, and HIV in some Southern churches that might not be as prevalent outside of the South. Taking a more culturally congruent approach to prevention is certainly suggested to engage churches; however, conversations about sex and sexuality cannot be ignored, especially in the South, where such topics can be a stigmatizing force. While promoting abstinence as the ideal, faith leaders recognized the reality of premarital sexual activity and same sex relationships occurring among congregational members. Likewise, church leaders in our study had compassion for and embraced those living with HIV, despite their sexuality and previous known sexual risk behaviors. Acknowledgment of the importance of educating congregations about HIV and no longer remaining silent has implications for how the conversation about HIV prevention moves forward.
Homosexuality was used synonymously with sexuality among participants, which implies little distinction between sexual orientation and the ability to simply experience sexual feelings. Faith leaders self-selected the term homosexuality over sexuality. This is congruent with other studies with faith leaders framing sexuality to a very limited definition. Nonetheless, faith leaders were in conflict with reconciling biblical teachings of homosexuality. Faith leaders practice compassion and tolerance but adhere to and abide by denominational bylaws condemning homosexuality. This finding reflects the existence and persistence of opposing responses that some faith leaders have. This persistent struggle of faith leaders is not uncommon to the feelings that some gay, lesbian, and bisexual individuals feel. Particularly among black men who have sex with men, negative religious experiences create internalized distress.
There are competing concerns of how best to integrate sensitive topics. The nature and direction of what is said to church members suggest that faith leaders are comfortable saying homosexuality is a sin, yet there is still a place for lesbian, gay, bisexual, transgender and queer (LGBTQ+) persons in the church. Understanding how this dual reality is perceived by and impacts Black gay men needs to be explored.uncovers dimensions of Black gay men's connectivity to spirituality and departure from religion as one possible explanation. He describes how hyper-masculinity is constructed and may be a by-product of homophobia in the church, placing men at risk for HIV. On the other hand, researchers have examined the range of health activities in churches and have found HIV-related stigma reduction is couched or incorporated as part of awareness and advocacy efforts. The utility of using similar approaches with sexuality also warrants further investigation.
Our findings demonstrate that interpretations of biblical scripture and selective use of scripture affects how faith leaders engage in specific conversations about sexuality and sex altogether. Faith leaders use scripture to guide their messaging, which may unintentionally lead to discussions about sexuality that have severe implications for perpetuating stigma. Prior research has shown that when stigma around their sexuality is present in churches, Black gay and bisexual men might suffer from social isolation and an unwillingness to seek guidance from their church leadership. Stigma around sexuality has also been known to ignite fears of losing congregants among some pastors. A church case study in the Midwest, assessing barriers that clergy experienced in starting and sustaining an HIV ministry, found that the primary barrier was views regarding sexuality. Likewise, discussions around the challenges of addressing sexuality in its entirety did occur in our focus groups. Moving forward, how do we help faith leaders effectively address sexuality and simultaneously respect their doctrine?
One strategy to increase readiness to address same sex relationships is for clergy to initiate discussions about how people who do not feel accepted and judged has implications on their mental wellbeing. This was a salient finding in one of our discussion groups that should not be ignored. The minority stress modelexplains how the intersection of discrimination, prejudice, and stigma among lesbian, gay, and bisexual individuals, may create a hostile and stressful environment that leads to worse mental health. Given their salient role in counseling mind, body, and spirit, clergy can gradually incorporate HIV-related prevention messages along this vein. Moreover, research suggests Black men who have sex with men use church as a means to cope and reconcile sexuality through worship and prayer for the forgiveness of sins. Tailoring interventions, with the help of church leadership and members of the congregations, could be one way to help alleviate stigma and help to decrease social isolation among Black gay and bisexual men. Facilitating readiness to acknowledge the benefits of coping strategies in spiritual counseling and ministerial work also should be considered in clergy development and theological training.
Another strategy is to take a risk reduction approach to sexual health within churches. Some faith leaders acknowledged premarital sex among their congregants and discussed efforts to reduce their risk of contracting HIV or having an unplanned pregnancy. The effectiveness of risk reduction approaches has been well documented among several at-risk sub-populations. Faith leaders were not as rigid of how their interpretations of the bible were communicated, as many assisted with STD prevention and supplied sexual health resources for subgroups of the church. They discussed the importance of holding the congregation accountable.
Considering the entire continuum of HIV prevention and care, the roles of faith leaders are not only critical among HIV prevention messaging, but also for people living with HIV (PLWH). Given the high participation of church attendance in the southern region, it is exceedingly probable that PLWH attend worship services regularly without disclosing their HIV status to fellow church members. Clergy could reinforce the importance of medication adherence and retention in medical care during worship services. Health ministries could facilitate linkage to medical care or HIV testing events. Therefore, resources to increase compassion and church support for those infected and affected could benefit many congregations and reduce HIV-related stigma in churches.
# Study limitations
We chose to use focus groups as opposed to individual interviews because of the inter-faith connections this community exudes and to examine group norms. However, the topic of sexuality and sexual health may have been too sensitive to discuss in focus groups. We diversified the groups to minimize any discomfort and gave participants the opportunity to refrain from answering certain questions. This allowed for a thought-provoking group discussion across denominations and leadership levels. Faith leaders self-selected to participate, thus may view issues of sexuality differently than non-participants given their familiarity with HIV risk behaviors. The possible impact of HIV framing should be considered within the interpretation of the qualitative findings. Declines in physical attendance and church participation over the years may have limited the findings as faith leaders discussed variations in church-based activities but not those online. Sampling is another limitation as faith leaders were mostly recruited from a convenience professional and personal network of Christian churches from one city in a very conservative U.S. Southern state. Finally, as with all qualitative studies, the small sample limits generalizability. Yet Memphis has been compared to other urban southern cities like Atlanta and New Orleans with similar new HIV infection rates and religious influence.
# Conclusions
This study described how HIV prevention, sexuality, and sexual health teachings are addressed from the perspective of Black faith leaders ministering in the South. Given the differing perceptions of clergy and faith leaders in the conservative South, it is often assumed that inclusiveness and compassion for people living with HIV and BMSM are absent in churches. Our findings demonstrate that this is not the case, but rather there are nuanced approaches in the delivery of teachings across faith leaders. Future studies might consider the ways in which biblical scripture is interpreted and selected to promote messaging and implications of denominational bylaws in both virtual and traditional worship spaces. Working in tandem with all LGBTQ+ communities and faith leaders to dissect and craft messages will reach those most at risk for acquiring HIV and those living with HIV to provide a truly inclusive and supportive environment. This becomes particularly important as nationwide jurisdictions develop local Ending the HIV Epidemic Plans integrating strategies to address stigma and homophobia within the four pillars.
To situate examples of our findings in the context of relevant EtHE Initiative pillars, we propose partnerships engaging faith communities to consider the following recommendations:
## 1.
Diagnose Pillar-acknowledge premarital sex among congregational members and offer regular HIV testing on church premises to normalize testing and educate members on knowing their HIV status.
## 2.
Prevent Pillar-recognize the possibility of congregational members engaging in condomless sex and substance use and offer HIV prevention information on pre-exposure prophylaxis and syringe services programs. 3.
Treat Pillar-deal with the whole person and promote physical, mental, and spiritual health by encouraging congregational members to access wellness services to improve health outcomes effectively.
## 4.
Respond Pillar-maintain transparency with congregations and the surrounding church community of potential outbreaks and become familiar with resources to get needed treatment and prevention services to those affected.
Disentangling our primary findings regarding the stringency of bylaws, suffering in silence, and stigmatization as potential threats to HIV prevention and applying them directly to the EtHE pillars, we may inform the development of effective interventions in churches and local Ending the HIV Epidemic implementation plans. |
The role of governance in implementing sustainable global health interventions: review of health system integration for integrated community case management (iCCM) of childhood illnesses
Improving health outcomes in countries with the greatest burden of under-5 child mortality requires implementing innovative approaches like integrated community case management (iCCM) to improve coverage and access for hard-to-reach populations. ICCM improves access for hard-to-reach populations by deploying community health workers to manage malaria, diarrhoea and pneumonia. Despite documented impact, challenges remain in programme implementation and sustainability. An analytical review was conducted using evidence from published and grey literature from 2010 to 2019. The goal was to understand the link between governance, policy development and programme sustainability for iCCM. A Governance Analytical Framework revealed thematic challenges and successes for iCCM adaptation to national health systems. Governance in iCCM included the collective problems, actors in coordination and policy-setting, contextual norms and programmatic interactions. Key challenges were country leadership, contextual evidence and information-sharing, dependence on external funding, and disease-specific stovepipes that impede funding and coordination. Countries that tailor and adapt programmes to suit their governance processes and meet their specific needs and capacities are better able to achieve sustainability and impact in iCCM.
# Introduction
Global health interventions should align with population needs and the health issues that resonate from resource constraints in health systems, poor access to health services in the population and governance of programmes that address the burden of childhood illnesses. Programmes that manage resource constraints, integrate lessons learnt and adapt to changing infectious disease conditions can offer resiliency and extended capabilities when facing emerging threats such as novel coronavirus , or other challenges that strain health systems and potentially diminish progress in combating persistent disease threats like malaria, pneumonia and diarrhoea. These diseases are leading causes of mortality in children under-5 years of age (U5) and can be better addressed through effective governance of health programmes that facilitate sustainable progress in reducing mortality by improving access to essential health services and implementation of evidence-based interventions with dedicated investment schemes.
## Key questions
What is already known? ► Governance, country leadership and management in global health strategies for child health have notable impact on programme success. Despite countries participating in the integrated community case management (iCCM) approach and developing policies to implement programmes, the resultant iCCM programmes were not always effective or sustainable.
What are the new findings? ► There are four key challenges that have impacted iCCM governance processes, programme effectiveness and sustainability. Specifically, lack of country leadership, the need for local contextual evidence on iCCM programmes to tailor country-specific iCCM programmes and information-sharing between implementers, dependence on external funding and disease-specific 'stovepipes' or silos that impede funding and coordination of programme activities.
What do the new findings imply? ► A strategic approach targeting each of these key challenges will improve governance of iCCM programmes and increase the likelihood for effectiveness and sustainability over time. Improved programmes may then contribute to reaching country goals for addressing burden of childhood diseases.
## Bmj global health
Persistent inequities in maternal and child health are enabled by barriers to health coverage, including access to care in urban centres and rural environments. In addition, on access to care, many are not afforded good quality, or face direct and indirect costs for health services that render treatment for preventable diseases improbable. Child health encompasses the nurturing care needed for a child to both 'survive and thrive' to their greatest potential and well-being. The 'good health' that ensues must be facilitated by equitable access to care. Many communities face persistent poverty due to socioeconomic disparities that require interference through high-level policy development and governmental influence. In these instances, limitations to health intervention impact and low uptake of available health services, then contributes to increased risk for illness, disease-associated morbidity and mortality.Given the leading causes of mortality U5 can be attributed to three infectious diseases, programmes targeting these diseases can have considerable impact in reducing morbidity and mortality, provided the programmes are governed effectively and offer reach to the underserved communities with the highest burden of disease. As a primary indicator for concern, reductions in mortality would denote progress for goals in child health. The United Nations' Children's Fund (UNICEF) global child survival call to action asked countries to strive for 20 or fewer deaths per 1000 live births by 2035,and their Strategy to Health 2016-2030 emphasises the necessary shift towards a health system strengthening approach that places a focus on equitable access through integrated, and communitybased care.Providing equitable access however requires that the primary health system has the infrastructure and resources to drive successful programme implementation. Moreover, governance, including the coordination, partnerships and management of programmes that sit within broader health systems' strengthening and global health strategies is intimately linked with sustainability, and anticipated success. From this stance, this project sought to understand governance attributes linked to success of integrated community case management (iCCM) programmes to identify thematic challenges in programme adaptation to national health system (NHS) structures.
Health systems' strengthening and governance in iCCM A strong health system connotes multisectoral engagement and hosts a community-based system for accessing health services at a local level.ICCM is evidence-based and focuses on improving access to effective case management for malaria, diarrhoea and pneumonia through deployment of community health workers (CHW) to increase reach to underserved populations. The concept of CHW programmes in public health systems is not novel; 7-11 however, the community-based government-led strategies that utilise them are varied by country with differing levels of success. While evidence has mounted showing the potential impact on child mortality through increased coverage of quality treatment services, challenges remain in achieving the greatest impact from iCCM. Since 2010, the iCCM strategy has complemented the WHO Integrated Management of Childhood Illnesses (IMCI) strategy that was initiated in 1999 to focus on delivery of treatment interventions through an integrated case management approach at the health facility level. A joint statement on iCCM by the WHO and the UNICEF, in 2012, stated that delivery of health services is often weak with low coverage for populations that have the greatest need.While IMCI had many successes, there were clear inefficiencies in reach to the most vulnerable populations, and needed strategic adaptation to extend its reach to address insufficient coverage and capture underserved populations in rural communities with high burden of illness and low utilisation and access to care. In estimates of the potential impact for community case management, significant reductions in morbidity and mortality made an integrated approach plausible. Prompt and effective community management of pneumonia, malaria and diarrhoea has been found to reduce mortality by 70%, 60% and 70%-90%, respectively.The iCCM programmes use CHWs based in their respective communities to deliver diagnostic and treatment services for multiple childhood illnesses. This includes training, equipping and supporting CHWs to assess, classify and (1) treat uncomplicated diarrhoea, pneumonia and malaria using oral rehydration salts (ORS)/zinc, oral antibiotics and artemisinin-based combination therapy respectively; and, (2) refer children with signs of severe illness and acute malnutrition to an appropriate referral facility. In the context of health system strengthening, iCCM fits as a programmatic contribution to overall goals set forth by WHO and UNICEF. The WHO framework for action towards strengthening health systems to improve health outcomes, addresses challenges to ensuring essential public health functions exist in an effective system that meets population needs.Similarly, UNICEF includes community health, national investments and governance as core aspects of their solutions for challenges to improving health systems.ICCM as an intervention begins to fill key gaps in reach and coverage identified in the implementation of the IMCI programme. In that regard, it is imperative to understand what is needed for successful implementation and sustainability of iCCM, as well as lessons learnt from the past implementation failures or scale ups that did not achieve maximum impact.
Governance, country leadership and management in global health strategies for child health have notable impact on programme success. Commitment to corresponding policies has also contributed to the greatest reductions in child mortality. [bib_ref] Where's the leadership? Future commitments of Unicef and WHO for global child..., Costello [/bib_ref] Approximately, one-third of countries participating in global strategies have developed multisectoral policies related to social determinants of child health; 6 however, the strategic approaches to meet national goals related to child health needs are BMJ Global Health often not normalised, lack leadership and the prioritisation that is required to achieve effectiveness and sustainability. For example, in a review of IMCI, 72 of 92 participating countries had an IMCI community health component where CHWs provided treatment for children. Of those 72 programmes, only 52 countries had provisions for iCCM. [bib_ref] Global implementation survey of integrated management of childhood illness (IMCI): 20 years..., Boschi-Pinto [/bib_ref] Governance is not prescriptive or normative; rather, it is relative to the society, culture, politics and systems at play. Decision-making processes, alongside political systems and social structures exist and influence adoption of global strategies, such as iCCM. Despite global consensus and awareness for the need to apply iCCM for achievement of national child health goals, 5 14 21 establishment of country-level policy varies. [bib_ref] Policy challenges facing integrated community case management in sub-Saharan Africa, Bennett [/bib_ref] Where supportive policies do exist, challenges remain, extending from policy to programme implementation and expansion. Understanding the process of policy development, strategic management through programme initiation and implementation is necessary to gauge programme potential for success and sustainability in a country.
# Methods
The purpose of this analysis was to understand governance attributes of iCCM programme success, using iCCM benchmark components 13 for (a) coordination and policy setting, (b) costing and financing and (c) monitoring and evaluation (M&E) and health information systems (HIS), to identify thematic challenges in iCCM programme adaptation to NHS structures. This was achieved by determining and assessing the technical and financial inputs required for government-led community health systems to provide curative services to their most vulnerable populations and identifying entrypoints in the governance process where solutions could be targeted. While much is known about the utility and effectiveness of CHW programmes and community-based case management for childhood illnesses, less is understood about impact of governance on child health initiatives, such as iCCM, as it is adapted in NHS structures.
## Search strategy and selection criteria
A review of indexed and grey literature, including academic publications, organisational reports, government documents, funding and technical support agency evaluations was performed. A search was conducted in the Cochrane review database using the terms 'integrated community case management' for identification of registered trials, and systematic reviews. A search was also conducted in the Pubmed central database using the terms 'integrated community case management', for which there is no MeSH subheading, ['integrated community case management' AND governance], and ['integrated community case management' AND policy]. The first tier of selection criteria included articles related to childhood illnesses and/or malaria, and exclusion of studies on case management in the elderly or other special populations (eg, homeless or mental health) or generalised integrated healthcare. References of selected articles were also reviewed for relevance and inclusion. Additional documents were identified on programme websites, specific journal supplements on global health policy, ministry of health websites and funding and technical support organisations resource databases.
## Analytical approach
To investigate iCCM governance, in a broad sense, to improve programme success, the Governance Analytical Framework (GAF) [bib_ref] In: Research for sustainable development: foundations, experiences and preferences, Hufty [/bib_ref] is applied to iCCM policy development and programme implementation processes. The two basic assumptions of the GAF are that governance processes are found in any society and those processes exist as a set of observable phenomena. As a result, processes can be analysed from a non-normative perspective and governance may be converted into a methodology, for the study of systems of social norms and interactions that determine how public decisions are made. [bib_ref] In: Research for sustainable development: foundations, experiences and preferences, Hufty [/bib_ref] The lens of key programme attributes for governance; specifically, policy, management and coordination, and financing were used to exemplify systems structures in different countries. In the WHO/UNICEF guidance for implementing iCCM, it was recommended that countries examine policy options, build on existing programmes and initiatives, ensure quality of care, supply-chain management and logistics and monitor and assess data to identify gaps in coverage, patterns in care-seeking behaviour and other key indicators that could be applied to improving programme effectiveness.As a basis for future programme evaluation and documenting measurable impact, benchmarks for implementation were also developed to facilitate country planning, implementation, monitoring and assessment of iCCM activities. The benchmarks for implementation included: (i) coordination and policy-making; (ii) costing and financing; (iii) human resources; (iv) supply chain management; (v) service delivery and referral; (vi) communication and social mobilisation; (vii) supervision and performance quality assurance and (viii) M&E and HIS. Of these benchmarks for implementation, i, ii and viii served as the proxy measure and contextual focus for development of thematic trends in a GAF for iCCM.
## Patient and public involvement
There was no patient or public involvement in this study. However, dissemination of the study results to relevant actors within iCCM policy development and partnering organisations will impact the wider patient communities reliant on iCCM services. Improved governance of global health programmes will aid in improved access and coordination of services needed to combat childhood illnesses.
# Results
A total of 47 countries were included with varying levels of available information on policy and programme uptake.
## Bmj global health
The countries selected were included based on the available evidence identified through the search strategy for countries that have implemented iCCM programmes. The map in depicts the percent quantity of evidence by country that was used to understand governance for iCCM. Total evidence by governance attribute was stratified by (1) policy, (2) coordination and management, (3) costing and financing and (4) neutral. Articles that discussed cost or finance in the context of or with policy and sustainability were included in the policy category. Similarly, studies that conducted evaluations were included in the coordination and management category.
## Governance analytical framework for iccm
The iCCM governance structure was described generically, within the GAF 25 as (a) the collective problems impacting iCCM success; (b) actors involved in the coordination and policy-setting of iCCM; (c) contextual norms for health systems in a given country and (d) the nodal points that serve as the intersection for programmatic interactions. [fig_ref] Figure 2: Governance analytical framework [/fig_ref] reveals the iCCM governance structure and illustrates the process for introducing interventions that address nodal point problems to achieve success in iCCM health system integration and sustainability. Using this iCCM governance structure, resultant key themes exemplify significant challenges to governance processes, using the empirical evidence from country-specific examples of iCCM. Each component of iCCM governance is further explained followed by a thematic summary of the key challenges and considerations for addressing them.
## Problems impacting iccm success
Countries are often receptive to global health strategies that target key issues relevant to the health of their population. Despite receptivity, the anticipated impact to addressing these health problems is not observed. The analysis revealed that one limiting factor is the hesitancy in policy development and limitations in implementation of corresponding programmes. [bib_ref] A health systems approach to integrated community case management of childhood illness:..., Mcgorman [/bib_ref] Over the last decade, the number of countries developing iCCM policy has steadily increased. However, despite written policy advancing iCCM as a supported approach to child health, [bib_ref] Global implementation survey of integrated management of childhood illness (IMCI): 20 years..., Boschi-Pinto [/bib_ref] programmes were not always implemented to scale. In some cases, policy was written and codified, but implementation never occurred. As previously mentioned, 52 countries of 100 countries surveyed in the review of IMCI, had plans, policy or CHW programme components and infrastructure for iCCM. [bib_ref] Global implementation survey of integrated management of childhood illness (IMCI): 20 years..., Boschi-Pinto [/bib_ref] This is a significant increase from previous surveys which reported [bib_ref] Community case management of childhood illness in sub-Saharan Africa -findings from a..., Rasanathan [/bib_ref] countries implementing iCCM in 2013, [bib_ref] Community case management of childhood illness in sub-Saharan Africa -findings from a..., Rasanathan [/bib_ref] and 18 countries in 2010.Another problem impeding iCCM success is the challenge of developing policy based on adaptation of broadbased global health strategies to meet country needs while confined by country capacity. Evidence from Country-specific evidence for iCCM governance. Countries 47 with available evidence for iCCM programme implementation. The map and corresponding list of countries depicts the percent quantity or proportion of evidence used to understand governance for iCCM and reflects the potential bias and range of documentation on programme implementation. iCCM, integrated community case management, BMJ Global Health case studies investigating iCCM policy development and programme implementation revealed that policy may be established in an ad hoc fashion without an informed and formal process. [bib_ref] Policy challenges facing integrated community case management in sub-Saharan Africa, Bennett [/bib_ref] For success, translating policy to action can be achieved through 'championing' and dedicated leadership by the country, 6 14 18 invested collaborators that can offer both technical and financial support and community empowerment with a clear demand for use of services by the community. [bib_ref] The governance of national community health worker programmes in low-and middle-income countries:..., Schneider [/bib_ref] A key concern identified in the analysis relates to insufficient attention for evidence gathering, synthesis and assessment to ensure gaps do not exist in the integration of new evidence to policy and programmes, as was perceived with the IMCI strategy. With this awareness, lessons learnt from key IMCI programme challenges should be used to inform governance practices for iCCM. In many ways, iCCM partially fills the gaps in the community component of IMCI that were lost to the focus on training and skill enhancements. Provisions should ensure systematic processes for evidence generation and capture by conducting substantive large-scale country evaluations with funding and implementing partners, and using validated tools for measuring impact. In assessing early implementation of iCCM, there were few evaluations available to understand the key challenges and concerns in programme implementation or impediments to success. [bib_ref] Current scientific evidence for integrated community case management (iCCM) in Africa: findings..., Diaz [/bib_ref] This has improved recently with substantial increases in evaluations done to generate country specific, and sometimes district or village specific clarifications for programme successes and failures. [bib_ref] Evidence of impact: iCCM as a strategy to save lives of children..., Prosnitz [/bib_ref] M&E of iCCM, and broader strategies influencing child health outcomes, improves adaptation of programme structure and service delivery for populations to achieve the greatest impact from programme potential. Evaluations can provide revelations in programme effectiveness or the lack thereof. For example, evaluations in Burkina Faso, Ethiopia and Malawi revealed programmatic implementation issues related to coverage, demand and utilisation of iCCM services that led to less than impressive gains in child health targets that could be directly attributed to programme implementation. [bib_ref] Independent evaluation of the integrated community case management of childhood illness strategy..., Amouzou [/bib_ref] The results also show that M&E of programmes can reveal key issues that present opportunities for improving programme management. For example, in recent evaluations of the Rapid Access Expansion (RaCE) iCCM programme, launched in five countries, Democratic Republic of Congo, Malawi, Mozambique, Niger and Nigeria, in 2013 by the WHO, there was an average of 10% reduction in child mortality using the Lives Saved Tool for estimated impact to child mortality across four RaCE project sites. In contrast, the evaluation of the RaCE programme in Mozambique estimated that there were no under-5 lives saved; likely due to broader issues in supplychain and procurement causing stock-outs of critical medications needed for iCCM success. [bib_ref] Evidence of impact: iCCM as a strategy to save lives of children..., Prosnitz [/bib_ref] Understanding the broader systems needs can aid in programme implementation and eventual impact.Overall, improved
## Bmj global health
M&E of iCCM programmes from inception can reveal programmatic implementation issues, address needs for coverage, demand and utilisation of services and provide an opportunity to improve on programme management and sustainability.
Finally, a key problem revealed in programme success is that measurable indicators of CHW impact on diseases and child mortality are not always captured or translated to national HISs. The lack of efficient data collection makes key data for measuring programme effectiveness missing in programme evaluations. When useful data is collected, it is often of poor quality and incompatible for comparisons with overarching child health data and targets. Efficient and coordinated data generation and surveillance at the local level is critical to inform policy-makers on programme effectiveness, whose support is needed to maintain funding and sustain the programme. [bib_ref] Integrated community case management for childhood illnesses: explaining policy resistance in Kenya, Juma [/bib_ref] Global actors in coordination and policy-setting of iCCM The analysis revealed that global actors play a significant role in the governance of iCCM. Each actor brings differences in economic, social, cultural and symbolic capital that influences the mobilisation and support of resources to contribute to policy development, programme implementation and community engagement in a country. The strategic interaction of these actors is inextricably linked with the governance process. [bib_ref] In: Research for sustainable development: foundations, experiences and preferences, Hufty [/bib_ref] Our results show that in the early assessments of iCCM, success was linked to partnerships in funding, technical support and national governments. There was also evidence that an advantageous approach to adopting iCCM programming was to capitalise on opportunities for health systems strengthening using existing CHW cadres, and once-siloed single disease initiatives. Actors exist at different levels of real or perceived influence and power. Their roles can be classified as strategic, relevant and/or secondary to the use and flow of resources to achieve goals and affect decision-making. [bib_ref] In: Research for sustainable development: foundations, experiences and preferences, Hufty [/bib_ref] For iCCM, policy transfer and adoption from a global health strategy to a national policy involved different mechanisms. One mechanism included high-level organisations increasing situational awareness by introducing the strategy to countries and transferring policy recommendations. Another mechanism was for funding agencies to present opportunities for financial support to adopt and adapt a particular global health strategy. Organisations and collaborative partners would socialise the strategy in order to set global norms about its effectiveness and utility, and generate interest at a country level. [bib_ref] Altogether now… understanding the role of international organizations in iCCM policy transfer, Bennett [/bib_ref] While this can be construed as positive plays by actors to mobilise others in implementing a global health strategy, in the case of iCCM, there were also alternative dynamics at the country level that could inhibit policy uptake nationally. This puts the top-down and bottom-up approaches to influencing policy adoption at odds. Actors, such as policy entrepreneurs, can influence the policy-making process and help advance or inhibit progress in policy development based on their social or authoritative networks. [bib_ref] Policy entrepreneurs and structural influence in integrated community case management policymaking in..., Shearer [/bib_ref] These individuals or organisations can become integral in successful advancement of global health strategies. [bib_ref] Integrated community case management for childhood illnesses: explaining policy resistance in Kenya, Juma [/bib_ref] It was also revealed that pre-existing communities specific to disease, organisation and roles also influenced iCCM adoption and uptake at the national level. Silos for funding streams created dissonance in programme prioritisation for specific diseases and encouraged hesitancy in adoption of the iCCM policy with the awareness that dependence on funding would drive imbalanced programme development. Some actors, such as The Global Fund, have restructured their funding model to encourage a holistic approach to programme development.Similarly, the WHO RaCE programme ensured inclusion of sustainability roadmaps and strategies to facilitate increased capacity to manage iCCM programmes. [bib_ref] Integrated community case management: planning for sustainability in five African countries, Yourkavitch [/bib_ref] While it is possible for actors involved in funding mechanisms, or strategy development to alter their line of work to suit necessary changes to drive more comprehensive programming, country-specific political structures may not be capable of restructuring budgetary lines domestically or reorganising programme hierarchical structures that can adequately support generalised comprehensive strategies and funding streams. From this stance, the country context and the actors within are integral in maintaining political will, prioritisation and improving internal collaborations so programmes can be effectively implemented. [bib_ref] iCCM policy analysis: strategic contributions to understanding its character, design and scale..., George [/bib_ref] Country-specific contextual norms The analysis determined that contextual norms exist at all levels of the governance process, impacting decisionmaking for conceptual acceptance and policy development to support iCCM. Contextual norms relate to the social norms that exist within the culture and social environment of the country, organisations and communities that play a role in iCCM programme implementation, prioritisation and acceptance. For eventual adaptation to an introduced concept, there lies a process of rejection, resistance and internalisation. [bib_ref] In: Research for sustainable development: foundations, experiences and preferences, Hufty [/bib_ref] This was evident for early adapters to iCCM policy compared with those countries that exhibited some initial hesitation to implement all iCCM components. As was stated by George and colleagues, 'Much of the policy resistance to scaling up iCCM is not an aversion to what the intervention promises, but an acknowledgement that the health system effects of iCCM are broad ranging, requiring strategic analysis and resourceful management; skill sets that are underrepresented in resource constrained health systems'. [bib_ref] iCCM policy analysis: strategic contributions to understanding its character, design and scale..., George [/bib_ref] Contextual awareness by national policy-makers on their capacity to implement iCCM impacts policy development and uptake. Additionally, local health system structures need evidence of expansion experiences with iCCM scale-up that reflect their local circumstance. Despite evidence of programme efficacy conceptually, without pilot studies and evaluations in the country context, it was difficult to discern if the same successes and impact BMJ Global Health to reducing child mortality could be achieved. By highlighting the benefit of these programmes in the local context, actors influence the level of political will, backing and eventual budgeting for implementation and scale-up.
Other necessary considerations of contextual norms became evident in challenges for estimating costeffectiveness of iCCM implementation based on demand and use by community members. Health and careseeking behaviours, as well as understanding the needs and expectations of the populations influenced the use of services and community acceptance, which impacts the effectiveness of the programme. [bib_ref] The way forward for integrated community case management programmes: a summary of..., Young [/bib_ref] Additionally, global health security is a growing and persistent concern in many areas. Consideration for how to maintain services and advance strategies in environments at risk of local or regional instability, natural disasters, disease epidemics and other emerging threats that affect access and utility of health services is exceedingly important.Nodal points for iCCM interactions Nodal points are where challenges that impact programme success emerge; key actors are excluded or key populations are not reached through the intervention because of poor considerations for contextual norms. Our results showed that the overall management of iCCM programmes was dependent on adoption of policy into national health strategies. The interactions between policy adoption and eventual programme implementation have a trickle-down effect that influences programme success. For example, stock-outs of critical medications occur as a result of procurement strategies that did not consider iCCM programme needs. At a higher level, technical and bureaucratic considerations lead to concerns in cost and financing for scale-up, or dissonance between political stance and iCCM policy expansion. [bib_ref] Integrated community case management for childhood illnesses: explaining policy resistance in Kenya, Juma [/bib_ref] Reluctance to scale-up and expand policy depending on politics can be alleviated by local evidence generation and addressing specific concerns for strengthening key programme elements. Relieving tension at this nodal point facilitates policy development that is compatible with national goals. However, issues can still remain in harmonising programme management and coordination with contextual norms and key actors. Additionally, it was revealed that dependence on external funding, and uncertain outcomes in policy negotiations also impede programme sustainability. [bib_ref] Relaunch of the official community health worker programme in Mozambique: is there..., Chilundo [/bib_ref]
# Discussion
Using the GAF for iCCM, overall problems in iCCM governance can be described within the constructs of the various actors, contextual norms and nodal points that influence policy-making and programme implementation processes. The qualitative analysis revealed thematic challenges that exist within the GAF for iCCM which highlights key issues to address in the coordination and development of iCCM programmes. There are four main themes, described as key challenges to iCCM governance processes; (1) country leadership and integration into NHS's policy and infrastructure; (2) need for information-sharing and contextual evidence; (3) dependence on external funding impacts sustainability and (4) programme funding and coordination can be limited by disease-specific 'stovepipes' or silos. These challenges and relevant considerations for programme planning and implementation are summarised in table 1.
## Country leadership and health systems' integration
Ownership of iCCM at the country level has been described as a key indicator for programme success across reviews of iCCM programmes. The strategy should be integrated as a component to the primary healthcare (PHC) system with clear expectations on objectives and scope of the programme. Notably, iCCM cannot replace the PHC system; however, iCCM can play a significant role in extending reach and effectiveness of the PHC by addressing the needs of the population. 7 14 ICCM can be strategically implemented within a national plan in collaboration with partners that have clear and predefined roles should be tailored to community structure and needs. In addition, data integration with national HIS is ideal. Integration would ensure compatibility with health facility Addressing thematic challenges in iCCM governance processes
## Key challenges considerations for programme planning and implementation
Country leadership and health systems' integration ► Support from country leadership and ownership of iCCM concepts to facilitate integration into national health system's policy and infrastructure Information-sharing ► Ensure information-sharing between country programmes and partnering organisations ► Increase generation of a contextual pool of iCCM evidence for countries to use Dependency ► Increase national investments to cost and financing to improve sustainability of programmes ► Reduce dependence on external funding Stovepipes ► Continue efforts to integrate disease specific stovepipes ► Improve coordination across programme initiatives that encourage a strategic approach to meeting child health goals iCCM, integrated community case management.
BMJ Global Health data and allow for adequate M&E of programme effectiveness for child health targets. Improved management of data generation could also reduce burden on CHW responsibilities that have little value for programme management or goals. While programme expansion may incur additional responsibilities on CHWs, it is important to ensure scale up does not increase burden for workers that will correlate to poor data quality and decreased programme impact. [bib_ref] Policy challenges facing integrated community case management in sub-Saharan Africa, Bennett [/bib_ref] For example, in a review of six countries with iCCM policy, CHWs had responsibilities beyond iCCM priority diseases to provide additional child health and even some adult services. [bib_ref] Policy challenges facing integrated community case management in sub-Saharan Africa, Bennett [/bib_ref] There was evidence of subjective hesitancy to iCCM implementation due to awareness that broader health system needs and capacity for long-term effective management of an iCCM programme were lacking. Enhancing or adapting existing systems aided coordination and contributed to programme success versus development of entirely new programmes. In particular, adaptations that are implemented within existing health paradigms ensures that contextual norms specific to that country are not lost. Information-sharing Information-sharing and the generation of evidence that supports iCCM scale-up plays a significant role in policy uptake and advancement. The availability of data emphasising the utility, effectiveness and success of iCCM contributes to policy development and inclusion in NHS. [bib_ref] Policy challenges facing integrated community case management in sub-Saharan Africa, Bennett [/bib_ref] The shared and collective experiences on approaches for maintenance and sustainability of iCCM is needed so programmes can adapt to changing needs. Local evidence alongside evidence from other countries and collaborating partners would offer a myriad of relevant scenarios to understand factors that impact programme success. This should accompany improved M&E schemes, supported by reliable and quality data, shared in accessible platforms among partners. Local evidence is a significant factor in gaining and maintaining iCCM support; however, data and information-sharing with collaborating partners and other countries offers insight into successful strategies for scale-up, noting potential contextual limitations. Dependency hinders sustainability One of the greatest challenges to iCCM is that sustainability of programmes is relative to cost and financing. Programmes are better suited when key programme attributes are funded in full or at least in part by national governments to minimise dependence on external funding mechanisms. Situating iCCM policy within the NHS facilitates longevity and reliable management of iCCM components, but this is not always possible depending on the capacity and structure of the health system. Reliance on external funding in some cases may be needed for programme maintenance, though it may hinder the governance process and can lead to key issues in programme management and long-term sustainability. For example, in the RaCE Mozambique programme, shortages of supplies and 'widespread stockouts due to weaknesses in the health system' limited delivery of supplies and subsequently treatments that resulted in low impacts to child mortality despite a mature iCCM programme and corresponding policy. [bib_ref] Evidence of impact: iCCM as a strategy to save lives of children..., Prosnitz [/bib_ref] Long-term sustainability requires active engagement between the political leaders, organisational partners and other key actors that play a role in programme maintenance. Sustainability planning with external funding and implementing partners should address programme maintenance costs, community-level service delivery platforms, reliable drug supplies and CHW programme funding. Disease-specific stovepipes There have been substantial increases in empirical evidence on community-based programmes and CHWs, including iCCM; however, a disease or programmespecific orientation was also evident which hinders the effectiveness of an overall integrated strategy when single disease initiatives hold the foundation of the programme. This raises concerns for the design and sustainability of integrated national programmes. [bib_ref] The global pendulum swing towards community health workers in low-and middle-income countries:..., Schneider [/bib_ref] Dissolution of the disease specific lens can eliminate barriers to comprehensive programming linked to funding requirements. The integrated approach of iCCM uses improved coordination of efforts from service delivery to managerial government ministries and implementing partners at all levels of programming and funding, so there is a greater chance for comprehensive governance practices that support iCCM advancement.
# Limitations
This review was conducted to understand iCCM governance, so there exists some bias in scope based on inclusion and exclusion criteria and use of publicly available information. There are likely publications and reports not included within the pool of evidence used. Moreover, it is evident that bias exists in the literature for country-specific programmes based on funding and implementing partners that could support published documentation of programme governance attributes. As such, there may be countryspecific data not included that could have added value to the themes of governance processes identified.
The key challenges discussed are specific to iCCM and are not generalisable to governance of global health programmes in a broad context. There are however some parallels to existing programmes and efforts towards solutions. For example, the high burden, high impact approach is a targeted malaria response in the highest burden countries to drive success in meeting reduction goals.The approach has succeeded in exemplifying programme progress through high-level political BMJ Global Health engagement and support. In Uganda, a country-led process of political and multisectoral engagement, and community mobilisation has been established, including increased domestic funding, partnerships within and across government and community programmes and means for M&E. Despite the current success, challenges remain in ensuring sustainability with continued domestic funding, accountability and operationalisation of initiative components to the programme.
# Conclusions
Governance processes for iCCM are influenced by the contextual country norms for health system structure, utilisation and capacity. Moreover, iCCM success is dependent on factors of sustainability, national ownership and evidence-based strategic approaches to implementation and scale-up. A deep understanding of the governance process as it exists within a country facilitates the appropriate adaptation of the iCCM strategy that will suit country's needs, expectations and capacity.
[fig] Figure 2: Governance analytical framework (GAF) for integrated community case management (iCCM). Adapted GAF 17 for analytical interpretation of iCCM governance processes. *Intervention includes efforts to address problems at all nodal point interactions, or key challenges to the iCCM governance structure that can influence a positive outcome and solution for health systems integration and programme sustainability. [/fig]
|
Miscellaneous Topics in Computer-Aided Drug Design: Synthetic Accessibility and GPU Computing, and Other Topics
Background: Computer-aided drug design is still a state-of-the-art process in medicinal chemistry, and the main topics in this field have been extensively studied and well reviewed. These topics include compound databases, ligand-binding pocket prediction, protein-compound docking, virtual screening, target/off-target prediction, physical property prediction, molecular simulation and pharmacokinetics/pharmacodynamics (PK/PD) prediction. Message and Conclusion: However, there are also a number of secondary or miscellaneous topics that have been less well covered. For example, methods for synthesizing and predicting the synthetic accessibility (SA) of designed compounds are important in practical drug development, and hardware/software resources for performing the computations in computer-aided drug design are crucial. Cloud computing and general purpose graphics processing unit (GPGPU) computing have been used in virtual screening and molecular dynamics simulations. Not surprisingly, there is a growing demand for computer systems which combine these resources. In the present review, we summarize and discuss these various topics of drug design.These methods and resources in CADD are supported by various computational technologies such as PCs, cluster machines, cloud computing, general purpose graphics processing unit (GPU) computing and ad-hoc specialized computers.Fig. (2)shows these resources used in CADD. Setting up a CADD system on a GPU or in the cloud is still a difficult task for users, while ordinary PCs and workstations are quite user-friendly. We also review how to set up a cloud and GPU computer environment.
# Introduction
We review miscellaneous topics in computer-aided drug design (CADD), including non-scientific, technical, old and forgotten topics, since there have been a number of good reviews published already on the major topics in CADD. Computer-aided drug design is an assemblage of various computational methods and resources. These include compound databases, molecular dynamics simulations, ligand-binding pocket predictions, protein-compound dockings, structure-based drug screenings, ligand-based drug screenings, similarity searches, de-novo drug design, property predictions like LogS (aqueous solubility) and LogP ow (water-octanol partitioning coefficient) prediction, target/off target predictions, and predictions of synthetic accessibility.shows the relationship among these methods. They are based on the chemical compound structures, while the pharmacokinetics and pharmacodynamics studies are mainly based on the experimental data.
There have been many reviews reporting on compound databases, molecular dynamics simulation, ligand-binding pocket prediction, protein-compound docking, structure-based drug screening, ligand-based drug screening, similarity searches, and de-novo drug design. Thus, in the present review, we focus on a number of important but less-studied topics. We review the synthetic accessibility (SA) prediction SA is an important aspect of drug design, since in some cases computer-designed compounds cannot be synthesized. In addition, we briefly consider the correlation between the sales price of approved drugs and the SA values. drug design. However, the predicted reaction paths can be unrealistic in some cases, since steric effects (such as atom collisions and other inter-atomic interactions) are not taken into consideration in these approaches and the reaction databases include incorrect entries.shows one example of an SA prediction based on reaction data [bib_ref] RASA: A rapid retrosynthesis-based scoring method for the assessment of synthetic accessibility..., Huang [/bib_ref] [bib_ref] Computer-assisted design of complex organic syntheses, Corey [/bib_ref] [bib_ref] Computer-assisted mechanistic evaluation of organic reactions. 1. overview, Timothy [/bib_ref] [bib_ref] Automatic problem solving applied to synthetic chemistry, Bersohn [/bib_ref]. The compound in question is decomposed into fragments by breaking the relatively unstable bonds. This process is repeated until these fragments reach the starting chemicals included in the database. The bond breaking is performed based on the reaction database. This decomposition process generates a number of reaction paths. Then the difficulty of each reaction is estimated and the total SA of the compound in question is calculated. There are several ways to estimate the SA. One method evaluates the depth of the shortest path among all the predicted paths, since the SA depends on the reaction steps from the starting materials to the final product [bib_ref] RASA: A rapid retrosynthesis-based scoring method for the assessment of synthetic accessibility..., Huang [/bib_ref]. Generally speaking, if the compound is synthesized within 5 steps, the synthesis is easy. The other method evaluates the SA by considering the number of possible reaction paths, the difficulty of each step in the predicted path, and LogP on which the difficulty of the purification process depends [bib_ref] RASA: A rapid retrosynthesis-based scoring method for the assessment of synthetic accessibility..., Huang [/bib_ref]. The difficulty of synthesis will be decreased if the number of possible reaction paths increases. The difficulties of the reactions are estimated by chemists a priori. LogP prediction is more precise than LogS prediction and the descriptor-based LogP prediction is fast enough for practical use.
These retro-synthesis approaches depend on the reaction database. Satoh et al. examined the 329 data entries of the ChemInform (ISIS-ChemInform MDL-Information systems Inc.) database [bib_ref] Verification of a chemical reaction database-is it sufficient for practical use in..., Satoh [/bib_ref]. They reported that 46% of the data contained some error. Namely, they found wrong numbers of reaction steps in 25% of the data, and wrong reactions in 21%, wrong reaction sites in 19%, wrong chemical structures in 15%, wrong reaction conditions in 9% and wrong yield constants in 5% of the data. Most of the errors were human input errors. The SA and reaction path prediction theory should be improved in the future, as same as the theory, the quality check of various databases should be important.
## Synthetic accessibility prediction without a reaction database
A quantitative SA prediction method was reported by [bib_ref] Development of a method for evaluating drug-likeness and ease of synthesis using..., Takaoka [/bib_ref] [bib_ref] Development of a method for evaluating drug-likeness and ease of synthesis using..., Takaoka [/bib_ref]. This method had two important features. First, it is able to predict SA values based on only the chemical structure of the compound in question and thus does not require a reaction database. Second, the method has already been assessed in a study in which expert manual assessment was used to evaluate the SA (in the original paper, "synthetic easiness") of compounds in a teaching data set. The compounds in the teaching data set were described by molecular descriptors, and the weights of the descriptors were determined to reproduce the SA. This approach worked well, suggesting that SA is useful.
Of course, SA is not a well-defined concept but a fuzzy idea [bib_ref] Assessment of the consistency of medicinal chemists in reviewing sets of compounds, Lajiness [/bib_ref]. Many years ago, people thought that organic compounds could not be synthesized by humans but only by God. In the 19 th century, organic compounds were made from charcoal and organic polymers could not be synthesized. In the middle of the 20 th century, many reactions were developed based on oil. Thus, the SA values of compounds would have changed in each epoch along with the available organic chemistry.
The latest methods are based on a commercially available compound database and molecular descriptors. In these methods [bib_ref] Development of a method for evaluating drug-likeness and ease of synthesis using..., Takaoka [/bib_ref] [bib_ref] Estimation of synthetic accessibility score of drug-like molecules based on molecular complexity..., Ertl [/bib_ref] , reaction databases and retro-synthesis analyses are not necessary (see . Instead, SA is estimated from the probability of the existence of substructures of the compound calculated based on a compound library, the number of symmetry atoms, and the number of chiral centers in the compound. A steric effect, such as that of atomic collision, could be partially taken into consideration by the probability of the existence of substructures of the compound. It has recently become easier to access free compound databases, such as PUBChem [bib_ref] PubChem: A public information system for analyzing bioactivities of small molecules, Wang [/bib_ref] , ChEMBL [bib_ref] ChEMBL: A large-scale bioactivity database for drug discovery, Gaulton [/bib_ref] , ZINC [bib_ref] ZINC -a free database of commercially available compounds for virtual screening, Irwin [/bib_ref] [bib_ref] ZINC -a free tool to discover chemistry for biology, Irwin [/bib_ref] , and LigandBOX [bib_ref] Advanced in-silico drug screening to achieve high hit ratio -Development of 3D-compound..., Fukunishi [/bib_ref] [bib_ref] LigandBoxa Database for 3D structures of chemical compounds, Kawabata [/bib_ref] , and many vendors have made the catalogs of their compounds (see [fig_ref] Table 2: List of major chemical vendors and number of compounds supplied by them [/fig_ref]. In general, the databases of compound structures are more reliable than the reaction databases. [fig_ref] Table 2: List of major chemical vendors and number of compounds supplied by them [/fig_ref] shows trends in the number of compounds provided by some of the major compound vendors available on the LigandBOX database (URL http://ligandbox.protein.osaka-u.ac.jp/ligandbox// cgi-bin/index.cgi?LANG=en). As shown in this table, the total number of commercially available compounds (stocks) has not changed for several years. Every year, several hundred-thousand newly designed compounds appear and almost the same numbers of compounds are sold out. Thus the total number of compounds in stock has not changed substantially. Also, the number of natural compounds has been increasing gradually. On the other hand, the number of fragments has been increasing recently, since fragmentbased drug discovery became a trend in medicinal chemistry.shows one example of an SA prediction algorithm included in the MolDesk software package [bib_ref] Prediction of synthetic accessibility based on commercially available compound databases, Fukunishi [/bib_ref] (URL: http://www.moldesk.com/). This routine performs the SA prediction within about 0.1 seconds for a given compound with one click manipulation.
In the MolDesk software package, SA is calculated as follows: where c0, c1, c2, c3, and c4 are fitting parameters. Parameters c0-c4 are optimized to reproduce the SA determined by expert manual assessment. In this review, the SA determined by expert manual assessment is designated the "human SA" and the SA calculated by Eq. 1 is called the "calculated SA." SA is estimated from the probability of existence (Sprob) calculated based on a compound library, the number of symmetry atoms (Nsym), the total number of atoms (Ntot) of the compound, the number of chiral centers of the compound (Nchiral), and the graph complexity (Sgraphcomplexity) [bib_ref] The first general index of molecular complexity, Bertz [/bib_ref]. The Nsym is the number of chemically (topologically) equivalent atoms. If a compound consists of fragments frequently found in an available-compound database, it should be easy to synthesize. On the other hand, if a compound consists of fragments rarely or never found in an available-compound database, it would be difficult to synthesize. The probability of existence (Sprob) was calculated on the basis of the decomposition of the compound into fragments, and the probability of existence of each fragment was estimated according to the compound library. Any kind of substructure descriptor (the extended connectivity fingerprint (ECFP) descriptors developed by SciTegic [bib_ref] Estimation of synthetic accessibility score of drug-like molecules based on molecular complexity..., Ertl [/bib_ref] , MACCS key, Dragon, etc.) could be used for the SA prediction.
All compounds in the library were decomposed into small fragments. Let N and N(i) be the total number of fragments found in the library and the total number of fragments found in the library that were exactly the same as the i-th fragment of the compound in question, respectively. The probability of existence of the i-th fragment in library (P(i)) is given by
[formula] P(i)=N(i)/N. Eq. 3 [/formula]
The total probability of existence of the compound in question is then given as .
[formula] = i total i P P ) ( Eq. 4 and = = i total prob i P P S )) ( ( log ) ( log [/formula]
Eq. 5
As shown in , the compound library should consist of already-synthesized available compounds. . Substructure-based synthetic accessibility (SA) prediction. SA is estimated from the probability of existence (Sprob) calculated based on a compound library, the number of symmetry atoms (Nsym), the total number of atoms (Ntot) of the compound, the number of chiral centers of the compound (Nchiral), and the graph complexity (Sgraph-complexity).
The correlation coefficients between the predicted SA values and the human SAs are about 0.5-0.8 for these prediction methods. The R value (0.56) and the average error (1.2) obtained by MolDesk are similar to those between the human SAs (R=0.59 with a standard deviation of 0.22 and average error=1.1).
## Correlation among human sas
Takaoka et al. reported that the human SA values (synthetic accessibility (or synthetic easiness in the original text) values of visual inspection by individuals) were slightly dependent on the individual chemists, even within the same company. Nonetheless, while the concept of SA remains somewhat fuzzy and poorly defined in the manner of concepts like "good" or "bad," such abstractions are often quite useful. [fig_ref] Table 3: Correlation coefficient between human SAs evaluated by two chemists [/fig_ref] show the correlations among human SAs reported in previous articles [bib_ref] RASA: A rapid retrosynthesis-based scoring method for the assessment of synthetic accessibility..., Huang [/bib_ref] [bib_ref] Structure and reaction based evaluation of synthetic accessibility, Boda [/bib_ref] [bib_ref] Prediction of synthetic accessibility based on commercially available compound databases, Fukunishi [/bib_ref]. The human SAs were strongly dependent on the skill and experience of the individual chemist. The SA values estimated by the 2 chemists who belonged to the same company were still similar to each other. The SA values estimated by the chemists from different companies were different from each other and showed almost no correlation. SA would be expected to depend on the equipment available at the individual company, as well as on the training and experience of the company chemists.
## Relationship between the sales price of ap-proved drugs and sa
In Japan, about 10% of approved drugs are not sold in the market, since the drug prices do not meet the drug development costs. Thus the drug price is an important factor in drug development. We examined the relationship between the sales price and SA of approved drugs in the United States and Japan (US) [bib_ref] Prediction of synthetic accessibility based on commercially available compound databases, Fukunishi [/bib_ref]. Of course, the price of a drug is critical to its practical adoption by patients. For this reason, the ability to predict the price of a new drug in the drug-design process would be highly useful.
The prices of new drugs in the United States and Japan showed a weak correlation to the calculated SA values. Their correlation coefficients were 0.32 and 0.29, respectively (see. The prices of generic drugs showed the same trends as the new drugs. The reason for this should be that the price of a generic drug follows the price of the original drug.
The price of a drug may depend on many kinds of costs, including the costs of development, phase trials, and patents, as described in the next section (see. In addition, the market size and efficacy of the drug should be considered to decide the price. This means that a drug that is difficult to synthesize is not always expensive. The average SA of a drug is about 6, meaning that the average difficulty of drug synthesis is not extremely high.
## Drug prices in the us and japan
Because different countries have different methods of setting the prices of drugs prices, these prices vary widely around the world. In the US, the pharmaceutical companies determine drug prices based on a capitalist paradigm. On the other hand, in Japan the government sets the price of drugs based on their estimated importance to the healthcare system. These differences also reflect the different health insurance systems in each country. Thus a comparison between the drug prices in the US and Japan should reveal differences between a free market-based and government-based healthcare system.
There is almost no correlation between the sales prices of drugs in the US and Japan [bib_ref] Prediction of synthetic accessibility based on commercially available compound databases, Fukunishi [/bib_ref]. On the other hand,shows the correlation between the logarithm of the price of drugs in the US and the price in Japan. The correlation is very high (R = about 0.8). This means that the prices of drugs are roughly estimated and the values of drugs could change depending on each society.
## Cloud computing in cadd
Cloud computing has become a quite popular technology. Usually, the cloud computer is a large-scale computer and computer resources such as CPU and disk space are served on demand (see. Two of the major cloud computer services are the Amazon Web Service (AWS: URL https://aws.amazon.com/?nc1=h_ls) . The cloud computer is very similar to the conventional server computer. The difference between the cloud and conventional server is that the cloud is a virtual machine on which the OS, middleware, programs and computational environment must be prepared for application computing by users in general, and the virtual machine disappears when the users deallocates the virtual machine. Although this procedure (allocation and deallocation of the virtual machine) enables computers to be used flexibly, it is also complicated and time-consuming. The other important aspect of cloud computing is the requirement of high security or privacy. On a conventional server, we can see the status of other users' jobs. Conversely, we cannot see any information on the other users of a virtual machine. The scalability and high security have enabled many pharmaceutical companies to perform their virtual screenings, similarity searches and MD simulations on cloud computers. Cloud computing also enables us to access specialized hardware. The general purpose graphics processing unit (GPU) computational resource is somewhat troublesome for many users. Since the evolution of the GPU hardware is very fast, the GPU software is strongly dependent on the hardware, and the CUDA software (URL: https://developer.nvidia.com/cuda-zone) depends on the GPU hardware. Every year, new GPU hardware has appeared and the CUDA version has been updated. The GPU program should be tuned up for each GPU, since the performance of GPU programs depends on the balance of the number of GPU cores and memoryband width. Also the application of GPU is quite limited. This means that the GPU is used only when the GPU programs are available. In contrast, CPUs are always used for all application programs.
One of the problems of cloud computing is that the cloud machine is a virtual machine that basically begins as an empty box in cases where we allocate our own resources. We must prepare our own computational environment before starting target computations such as virtual screenings or MD simulations. This is almost equivalent to the procedure of setting up a brand new machine, and it must be done each time a new connection is made to the cloud computer. The other problem is that we must pay for the number of computations and in many cases it is difficult to predict the computational cost before the computations have actually been run. Finally, it is necessary to retrieve the data from the cloud computer before the close of service on each use. AWS provides more than 40 services (see, including CPU, disk, database, and security resources. Most of the charges for these services are proportional to the amount of usage of the cloud services. To realize a scientific calculation, the users must integrate and combine these cloud services adequately. Because there are so many cloud services, selecting the best combination for each scientific calculation can be hard work for the users. In addition, the users need resources on demand. Instead of asking the system providers, in some cases the users can manage their cloud computing using the available cloud services.
The cloud controller software assists in the process of preparing the computational environment and retrieval of data. There have been several reports on this software. Since the explanation of the mechanism of the cloud controller software is beyond the scope of this review, we will simply explain how to use such software. The controller software depends on the kind of computations to be made. Thus, we must select the best controller software for the computer programs that we want to use. AceCloud is a commandline cloud controller [bib_ref] AceCloud: Molecular dynamics simulations in the Cloud, Harvey [/bib_ref]. The users can submit their jobs, check the job status, abort them, and retrieve the results through AceCloud (see.
We show one example of the commercial cloud controller software "MolGate" (BY-HEX LLP, Tokyo Japan. URL http://byhex.com/ 2015) in. MolGate is a web-based program with SSH communication and the users can use both the CPU and GPU resources. When input files are prepared for MD simulation a pri-. Service menu of Amazon cloud computing.. Command list for cloud computing.
## Directory service
Host and Manage Active Directory
## Trusted advisor
Optimize Performance and Security
Analytics EMR
## Managed hadoop framework
## Data pipeline
Orchestration for Data-Driven Workflows
## Kinesis
Real-time Processing of Streaming Big Data
## Machine learning
Build Smart Applications Quickly and Easily
## Mobile services cognito
## User ldentity and app data synchronization
## Device farm
Test Android, Fire OS, and iOS apps on real devices in the Cloud
## Mobile analytics
Collect, View and Export App Analytics
## Sns
## Workmail preview
Secure Email and Calendaring Service ori, MolGate can prepare the environment on the cloud and start the calculation within 10 minutes. The example shows how to perform MD simulation on AWS.shows the procedures followed by the user. The user must allocate and deallocate the computational resources manually in addition to performing the login process. The procedure consists of the following four steps.
Step Step 4. Users retrieve the resulting data after the job is finished. On AWS, users can see only their virtual machine. Once the CPU resources are allocated, users will not be bothered by other users. Most of the costs of cloud computing are due to the CPU time, while the disk space usage is very inexpensive or even free in many cases.
To use the cloud computer, we should estimate the cost of the job a priori [bib_ref] Scaling predictive modeling in drug development with cloud computing, Moghadam [/bib_ref]. The computational cost depends on the size of the simulation system, the duration of the simulation, what kind of program will be used, and which service (type of server machine) of the cloud will be used. BY-HEX LLP provides a service for predicting the cost (URL: http://by-hex.com/). The total number of atoms is entered into the simulation system, and the site estimates the price per one MD simulation step. This prediction is still primitive, but the service is useful.
## Machine setup: gpu and compound database
GPU computation has also become very popular [bib_ref] The impact of molecular dynamics on drug design: applications for the characterization..., Mortier [/bib_ref] [bib_ref] Accelerating molecular modeling applications with graphics processors, Stone [/bib_ref] [bib_ref] A survey of general-purpose computation on graphics hardware, Owens [/bib_ref] [bib_ref] General purpose molecular dynamics simulations fully implemented on graphics processing units, Anderson [/bib_ref] [bib_ref] Routine microsecond molecular dynamics simulations with AMBER on GPUs. 1. Generalized Born, Goetz [/bib_ref] [bib_ref] GPU-SD and DPD parallelization for Gromacs tools for molecular dynamics simulations, Goga [/bib_ref] [bib_ref] Adapting a Message-Driven Parallel Application to GPU-Accelerated Clusters, Phillips [/bib_ref] [bib_ref] Molecular dynamics simulations accelerated by GPU for biological macromolecules with a non-Ewald..., Mashimo [/bib_ref]. GPU computing is particularly suitable for performing molecular dynamics simulation programs like AMBER [bib_ref] Routine microsecond molecular dynamics simulations with AMBER on GPUs. 1. Generalized Born, Goetz [/bib_ref] , Gromacs [bib_ref] GPU-SD and DPD parallelization for Gromacs tools for molecular dynamics simulations, Goga [/bib_ref] , NAMD [bib_ref] Adapting a Message-Driven Parallel Application to GPU-Accelerated Clusters, Phillips [/bib_ref] and psygene-G/myPresto [bib_ref] Molecular dynamics simulations accelerated by GPU for biological macromolecules with a non-Ewald..., Mashimo [/bib_ref]. The computational details of these programs are described in detail elsewhere. Some of these GPU programs are freely available. One of the most serious problems for end users is how to set up the GPU machine for these MD programs. The other problem is that the system size for the GPU computation must be larger than the minimum size that is determined by the program. Since most GPU programs adopt a spacedecomposition method for parallel computing, the system must be decomposed into sub systems. This means that the MD of a small system (like a single molecule) is not suitable for GPU computing.shows how to set up the MD program for the GPU. Many computers have GPU boards that are used mainly for graphics. To utilize the GPU for an application program, we must uninstall the GPU graphics driver software and install CUDA, a computer language for use with GPU. Most of the GPU-MD programs adopt CUDA for GPU computations. The slot number of each GPU board in the computer must be explicitly indicated in CUDA. Suppose our machine has three GPU boards. In some cases, a poor GPU is allocated to slot 1 for graphics. The additional two expensive GPU boards have slots 0 and 2 for the MD simulation. In some cases, a poor GPU is allocated to slot 0 for graphics. The additional two expensive GPU boards have slots 1 and 2 for the. Software and hardware GPU framework.
MD simulation. This allocation depends on the machine. If the GPU programs use a slow and a fast GPU board, the computation speed depends on the speed of the slow GPU board. This setup job is somewhat time-consuming work. The user can request that the computer vendor set up the machine, or in some cases, the vendor will provide a pre-installed machine such as MolSpace (LEVEL FIVE Co., Ltd. Tokyo; URL: http://www.level-five.jp/).
In this section, we explain how to start using the GPU computation, since the set up the environment for the GPU computation is a much more time-consuming and complex process than the conventional CPU. Most of the GPU programs run on the CUDA language. This means that the GPU of the Intel Core-i series is not suitable for GPU computing, since the GPU of core-i is not designed for CUDA. Many computers utilize the GPU card for graphics. To use the GPU card for a scientific calculation program, we must stop using the GPU for graphics and change the status of the GPU card for application only. Our example is designed for CUDA7.0 running on Linux x86_64 (RedHat6 (RHEL6) and Cen-tOS6). The RPM package of CUDA for RHEL6/CentOS6 is available from the NVIDIA CUDA download website (https://developer.nvidia.com/cuda-downloads).
In order to use GPUs for running CUDA application programs, we must check the versions of the software and hardware. For example, CUDA for CentOS6 runs on the combination of kernel 2.6.32, GCC version 4.4.7, and GLIBC version 2.12. Each CUDA version runs on each GPU card. We must check what kind of GPU card is available on the CUDA GPUs site (https://developer.nvidia.com/cuda-gpus). These points can be checked using the following 5 steps.
Step 1. Check the version of the OS (obtained by "$ cat /etc/*release").
Step 2. Check the version of the kernel (obtained by "$ uname -a").
Step 3. Check the version of GCC (obtained by "$ gcc --version").
Step 4. Check the version of GLIBC (obtained by "$ rpm -q glibc").
Step 5. Check the GPU card (obtained by "$ lspci | grep -i nvidia").
In the present review, we explain how to set up the GPU for MD simulation mainly for an NVIDIA CUDA environment for RedHat6/CentOS6, since CUDA is now popular for GPU simulations. In order to use the NVIDIA driver PRM package, an EPEL package must be installed. This process should be done by following 7 steps as follows.
Step 1. Download a suitable EPEL, such as # yum install (http://dl.fedoraproject.org/pub/epel/epel-release-latest-6.noarch.rpm).
Step 2. Install the CUDA on the command mode (run level 3), then set the run level using the command "# /sbin/init 3."
Step 3. Install the CUDA repository ("# rpm --install cuda-repo-rhel6-7-0-local-7.0-28.x86_64.rpm").
Step 4. Clear the cache of yum ("# yum clean expire-cache").
Step 5. Install CUDA by yum ("# yum install cuda"). Answer yes ("y") to all the questions.
Step 6. Restart the computer system by typing the command "# /sbin/shutdown -r now."
Step 7. Change the default parameter of EPEL. Change "enable=1" to "enable=0" for [epel] in the file "/etc/yum.repos.d/epel.repo".
In order to compile our GPU application on our computer, we must set the path for the CUDA library. We add the following two lines in ~/.cshrc for c-shell or ~/.bashrc for bash. export PATH=/usr/local/cuda/bin:$PATH export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PA TH Before starting our own application calculations, we check the software and hardware environment.
We download the sample program for CUDA and compile it using the following two steps.
Step 1. Put the CUDA sample on an suitable directory. For example, to put the sample on the current directory (.), we use the command "$ cuda-install-samples-7.0.sh .".
Step 2. Compile the sample using the command ("$ cuda-installsamples-7.0.sh .").
In the above section, we have shown how to prepare the CUDA environment for GPU applications. Now we can start the GPU applications. There are several GPU programs available on the Internet, and most of these are MD simulation programs. We will explain how to use psygene-G of the myPresto program suit developed by our group, since most of the other application programs should follow a similar procedure. Psygene-G is available on the myPresto download site (http://presto.protein.osakau.ac.jp/myPresto4/); after downloading, unzip the file using any suitable directory.
Psygene-G is an MPI parallel GPU program [bib_ref] Molecular dynamics simulations accelerated by GPU for biological macromolecules with a non-Ewald..., Mashimo [/bib_ref]. To compile this program, we must modify Makefile for our software environment. There are two Makefiles. One is for the MPI parallel program part (src/Makefile) and the other is for the CUDA program part (src/cuda/Makefile). Users should modify the Makefile of each application program in a manner similar to the following example.
Modification of Makefile for the MPI program (src/Makefile) to GPU in this example: (1) The CUDA version is indicated as "CUDA=cuda.x.y" for CUDA version x.y. (2) Activate the "#GPU Lib settings" block.
(3) Psygene-G is written in FORTRAN90, and the MPI library for FORTRAN90 is indicated by "FC=mpif90". (4) This is psygene-G specific indication. The memory size for a million atoms is indicated by "OPTIONS = -D_LARGE_SYSTEM"
[fig] Figure 1: Schematic representation of methods in computer-aided drug design. [/fig]
[fig] Figure 2: Schematic representation of resources in computer-aided drug design.Miscellaneous Topics in Computer-Aided Drug DesignCurrent Pharmaceutical Design, 2016, Vol. 22, No. 23 3557 Fig. (3). Two major synthetic accessibility (SA) prediction procedures. [/fig]
[fig] Figure 4: Reaction path-based synthetic accessibility prediction. [/fig]
[fig] Figure 6: An example of SA prediction by a commercial program (MolDesk GUI software. Information and Mathematical Science Bio Inc., Tokyo Japan). [/fig]
[fig] Figure 8: Correlation between SA and the price of newly approved drugs in the US. [/fig]
[fig] Figure 9: Setting of drug prices in the US and Japan. The setting of drug prices depends on the country, type of disease and type of drug. [/fig]
[table] Table 1: A selection of software programs for synthetic accessibility. [/table]
[table] Table 2: List of major chemical vendors and number of compounds supplied by them. [/table]
[table] Table 3: Correlation coefficient between human SAs evaluated by two chemists. [/table]
[table] Table 4: Correlation coefficient between human SAs evaluated by five chemists reported in RASA software. [/table]
[table] Table 5: Correlation coefficient between human SAs evaluated by five chemists reported in SYLVIA software. [/table]
[table] Table 6: Correlation coefficient between human SAs evaluated by five chemists reported by Taisho Pharmaceutical Co., Ltd. [/table]
|
A dermatologic perspective on multisystem inflammatory syndrome in children
As of May 2020, an emerging immune-mediated syndrome primarily affecting children has been detected primarily in Europe and the United States. The incidence of this syndrome appears to mirror the initial infectious assault with a delay of several weeks. This syndrome has been termed "multisystem inflammatory syndrome in children" (MIS-C) and is observed in association with the coronavirus disease 2019. The phenotypes of presentation include several characteristic features, including prolonged fever, skin eruptions, neck stiffness, and gastrointestinal manifestations with pronounced abdominal pain. Shock and organ dysfunction on presentation are frequent but inconsistent, whereas respiratory distress is typically, and notably, absent. We have reviewed the recent published data aiming to better understand MIS-C, with a focus on its mucocutaneous manifestations.
# Introduction
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a novel coronavirus that originated in Wuhan, China, in December of 2019. [bib_ref] First case of 2019 novel coronavirus in the United States, Holshue [/bib_ref] This virus causes a severe respiratory distress syndrome among several other severe manifestations. [bib_ref] First case of 2019 novel coronavirus in the United States, Holshue [/bib_ref] Starting in early 2020, the virus spread rapidly worldwide, compelling the World Health Organization (WHO) to declare a global pandemic on . To date, there are more than 15 million cases with more than half a million deaths due to coronavirus disease 2019 (COVID-19), globally. Children make up a small percentage of these cases. According to the Centers for Dis-
## Defining mis-c
Multisystem inflammatory syndrome in children describes an emergent childhood inflammatory disorder, with similarities to Kawasaki disease (KD), KD shock syndrome (KDSS), toxic shock syndrome (TSS), and macrophage activating syndrome (MAS). Due to its recent nascence, the constellations of manifestations defining MIS-C can be found under a number of names, such as pediatric inflammatory multisystem syndrome temporally associated with SARS-CoV-2 (PIMS-TS), Kawasaki-like disease, or toxic shocklike syndrome. Altogether, there are now approximately 1000 cases of MIS-C documented worldwide. [bib_ref] Childhood multisystem inflammatory syndrome -a new challenge in the pandemic, Levin [/bib_ref] As the spotlight shines on this new inflammatory disease, we are beginning to have more clarification of its clinical presentation and pathogenesis.
Both the CDC and the WHO have released their own MIS-C diagnostic criteria to help clinicians in making such diagnosis. Tables 1 and 2 review these criteria in detail. The main clinical manifestations that both groups focus on include fever for more than 24 hours, laboratory evidence of inflammation, tw o or more organ involvement (commonly gastrointestinal, followed by cardiac and renal), mucocutaneous findings, and either a positive test or exposure within 4 weeks of clinical manifestations. A negative COVID polymerase chain reaction (PCR) does not rule out this diagnosis. Because MIS-C is thought to be an immune-mediated secondary response to the virus, COVID PCRs are usually negative at the time of the illness, and antibodies are positive in the majority of cases. 11 , 14-21 Importantly, diagnostic criteria for this condition are generally very broad, and thus there is significant challenge in identifying which patients falling under these diagnostic criteria are true MIS-C. Demographic data of children with MIS-C from 13 recent large case series are summarized in [fig_ref] Table 3: Recently published descriptive studies on multisystem inflammatory syndrome in children from the... [/fig_ref]. The size of these studies ranges from 8 to 186 children. We recap patient mean/median age, sex, cutaneous signs/clinical manifestations, SARS-CoV-2 testing method, treatments, and outcome. Fever and GI clinical manifestations were the top two most common systemic signs/clinical manifestations seen in children who met criteria for MIS-C. The majority of patients were previously healthy. The two most common comorbidities were asthma and obesity. The median age of children with MIS-C was between 8 and 12 years. Boys were equal or more prevalent in all but two publications. In the U.S. studies, the majority of the children affected were
## Mucocutaneous manifestations of mis-c
Although mucocutaneous manifestations are not very common among children with COVID-19 at large, they are among the top clinical manifestations in children with MIS-C. Most of the clinical information to date comes from small descriptive studies, such as case series and case reports. Documented cutaneous findings reported in children with COVID-19 include a nonspecific maculopapular eruption, followed by chilblain-like or pernio-like acral lesions, urticarial lesions, livedo reticularis, papulovesicular or varicella-like lesions, petechiae or dengue-like lesions, and erythema multiforme-like lesions. [bib_ref] Clinical characteristics of coronavirus disease 2019 in China, Guan [/bib_ref] [bib_ref] Classification of the cutaneous manifestations of COVID -19: A rapid prospective nationwide..., Galván-Casas [/bib_ref] [bib_ref] Varicella-like exanthem associated with COVID-19 in an 8-year-old girl: A diagnostic clue?, Genovese [/bib_ref] [bib_ref] Reply to COVID-19 can present with a rash and be mistaken for..., Jimenez-Cauhe [/bib_ref] [bib_ref] Chilblain-like lesions in children following suspected COVID-19 infection, Colonna [/bib_ref] [bib_ref] Clustered cases of acral perniosis: Clinical features, histopathology, and relationship to COVID-19, Cordoro [/bib_ref] [bib_ref] Chilblains in children in the setting of COVID-19 pandemic, Andina [/bib_ref] [bib_ref] Comment on "Chilblains-like lesions in children following suspected COVID-19 infection, Joob [/bib_ref] [bib_ref] Erythema multiforme-like lesions in children and COVID-19, Torrelo [/bib_ref] Multiple larger case series published in the recent months 10-12 , 14-21 , 23 , 24 focus on characterizing and understanding MIS-C. Each case series identified children who met criteria for MIS-C and described the cutaneous clinical manifestations that they presented. The following descriptive terminology was used to describe the skin findings: "conjunctivitis," "rash," "red/cracked lips," "lips/oral cavity changes," "cheilitis," "extremity changes," and "hand/feet edema." Most of the studies did not provide photographs of the eruption and did not attempt to describe it. Others that did used the following nondescriptive terminology: "polymorphous," "general," "variable," "skin desquamation," "diffuse," "non-specific," or simply "rash." Refer to [fig_ref] Table 3: Recently published descriptive studies on multisystem inflammatory syndrome in children from the... [/fig_ref]. [bib_ref] First case of 2019 novel coronavirus in the United States, Holshue [/bib_ref] A targeted surveillance for MIS-C at multiple pediatric health centers across the United States was conducted and identified 126 children who met criteria for MIS-C. Mucocutaneous findings were identified in 74% of children who met criteria for MIS-C. Of these, 59% had nonspecific eruption, 55% bilateral conjunctivitis, 42% oral mucosal changes, abnormalities such as aneurysms more common Renal involvement rare Pro-inflammatory state typically less common and less severe Lymphopenia not common Thrombocytosis and 37% peripheral extremity changes. [bib_ref] Multisystem inflammatory syndrome in U.S. children and adolescents, Feldstein [/bib_ref] A review of more than 191 potential MIS-C cases of hospitalized children reported to the New York State Department of Health indicated that 95 met criteria for MIS-C. Of the children who met criteria for MIS-C, 60% had a diffuse nonspecific eruption, whereas 56% had conjunctivitis and 27% oral mucosal changes . [bib_ref] Multisystem inflammatory syndrome in children in New York State, Dufort [/bib_ref] Looking at all the 13 case series presented in [fig_ref] Table 3: Recently published descriptive studies on multisystem inflammatory syndrome in children from the... [/fig_ref] , the percentage of children diagnosed with MIS-C who developed mucocutaneous findings were conjunctivitis 27%-93%, oral mucosal changes 25%-87%, rash 47%-81%, and hand/feed erythema and edema 27%-68%.summarizes the top mucocutaneous manifestations of children with MIS-C. The skin findings associated with MIS-C tend to be more common in younger children and decrease with age. Of the children between 0 and 5 years old, 87% had mucocutaneous findings, compared with only 61.5% of those 13-20 years old. 19
## Similarities and differences with kawasaki disease
As we discuss the various mucocutaneous manifestations associated with MIS-C, many sound similar to other diseases, specifically KD (both typical and atypical). The differential diagnosis of a child presenting with a mucocutaneous eruption includes the following: Stevens Johnson syndrome/toxic epidermal necrolysis (SJS/TEN), drug reaction with eosinophilia and systemic symptoms (DRESS), mycoplasma-induced rash and mucositis / reactive infectious mucocutaneous eruption (MIRM/RIME), and toxic shock syndrome, as well as other viral or postviral etiologies. Identifying the similarities and differences between these entities is crucial in distinguishing them and making the correct diagnosis.
From a clinical perspective, a subset of patients with MIS-C show features that are similar to KD, including mucocutaneous findings. The question arises as to whether these two syndromes represent a single entity, rather than completely separate conditions. Kawasaki shock syndrome, which is a known variant of KD, as well as severe KD both recapitulate many of the more common features found in MIS-C, such as shock, pro-inflammatory state and pathology, and some of the hematologic disturbances, such as thrombocytopenia, which at first glance appear unique to MIS-C. A reasonable argument can be made that MIS-C represents a severe Kawasaki spectrum.
Many of the demographic features of the disease differ, such as age (mean age 8-12 years in MIS-C versus the rarity of Kawasaki cases in children older than 5 years). The absence of predilection for Asian populations being affected in MIS-C is also contrary to what one might see in KD. [bib_ref] Pediatric coronavirus disease-2019 -Associated multisystem inflammatory syndrome, Shulman [/bib_ref] For instance, only 1 of 17 children diagnosed with MIS-C in New York City was of Asian descent. [bib_ref] Multisystem inflammatory syndrome related to COVID-19 in previously healthy children and adolescents..., Cheung [/bib_ref] The most common cardiac abnormality associated with MIS-C is left ventricular dysfunction/myocarditis, unlike coronary artery abnormalities in KD. [bib_ref] Acute heart failure in multisystem inflammatory syndrome in children (MIS-C) in the..., Belhadjer [/bib_ref] Of the children who met criteria for MIS-C at Columbia University Irving Medical Center/New York-Presbyterian Morgan Stanley Children's Hospital in New York City, 35% had moderate left ventricular dysfunction, whereas only one child developed a medium-sized aneurysm. [bib_ref] Multisystem inflammatory syndrome related to COVID-19 in previously healthy children and adolescents..., Cheung [/bib_ref] Coronary aneurysms are observed in other vasculitis syndromes, such as Takayasu arteritis and polyarteritis nodosa, [bib_ref] Aneurysm of the left main coronary artery in Takayasu arteritis, Sajeev [/bib_ref] but rarely are they seen in association with other vi-ral and infectious conditions. [bib_ref] Acute myocardial infarction due to polyarteritis nodosa in a young female patient, Wi [/bib_ref] The presence of an aneurysm itself is not a defining feature that would necessitate relating the two syndromes. [fig_ref] Table 5: Comparing and contrasting multisystem inflammatory syndrome in children with Kawasaki disease [/fig_ref] further compares and contrasts clinical manifestations of MIS-C and KD.
# Conclusions
MIS-C is a recently recognized childhood inflammatory disorder seen in patients with confirmed or suspected COVID-19. The increasing incidence 3 to 4 weeks after COVID-19 infection suggests a postinfectious phenomenon occurring in susceptible individuals. We have reviewed recently published large case series on MIS-C in an effort to identify and characterize the mucocutaneous manifestations of MIS-C and conclude that the vast majority of MIS-C cases present with mucocutaneous manifestations, including conjunctivitis, diffuse eruptions, and oral mucosal changes, represent MIS-C. The eruption of MIS-C is typically diffuse and nonspecific, whereas mucocutaneous manifestations of MIS-C are more common in younger children with the prevalence decreasing with age. Although the mucocutaneous findings of MIS-C may resemble those of KD, the two conditions differ widely for the mean age of onset, race predilection, and the associated findings [fig_ref] Figure 1: A 7-year-old girl diagnosed with multisystem inflammatory syndrome in children, presented with... [/fig_ref].
## Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
[fig] Figure 1: A 7-year-old girl diagnosed with multisystem inflammatory syndrome in children, presented with erythematous urticarial-appearing plaques on the neck overlying lymphadenopathy. [/fig]
[fig] Figure 2: A 12-year-old boy diagnosed with multisystem inflammatory syndrome in children, presented with erythematous macules and papules coalescing into patches and plaques, respectively, on the trunk and extremities. This patient also presented with palmar erythema(Figure 3). [/fig]
[fig] Figure 3: The same patient as inFigure 2. Patient presented with bilateral palmar erythema. [/fig]
[table] Table 1: The Centers for Disease Control and Prevention case definition for multisystem inflammatory syndrome in children All four diagnostic criteria must be met: 1. Age < 21 years 2. Clinical presentation should include all of the following: a. Fever ≥38.0 °C for ≥24 hours or report of subjective fever lasting ≥24 hours. b. Laboratory evidence of inflammation, including, but not limited to, one or more of the following: No alternative plausible diagnoses. 4. Positive for current or recent SARS-CoV-2 infection by RT-PCR, serology, or antigen test; or COVID-19 exposure within the 4 weeks before the onset of clinical manifestations. [/table]
[table] Table 2: The WHO case definition of multisystem inflammatory syndrome in children [/table]
[table] Table 3: Recently published descriptive studies on multisystem inflammatory syndrome in children from the United States and Europe [/table]
[table] Table 5: Comparing and contrasting multisystem inflammatory syndrome in children with Kawasaki disease [/table]
|
Oxidized Phosphatidylserine: Production and Bioactivities
Recent development of analytical methods for lipid hydroperoxides and preparation of highly pure lipid hydroperoxides have revealed the important new pathophysiological roles of oxidized phospholipids. Generation of reactive oxygen species and subsequent oxidative stress leads to random oxidation of membrane phospholipids. However, recent studies have reported that anionic phospholipid molecules such as phosphatidylserine (PS) and cardiolipin are preferentially oxidized during apoptosis, resulting in efficient apoptosis execution and apoptotic cell clearance by phagocytes. This review is exclusively focused on selective production of oxidized PS (oxPS) during apoptosis as well as the novel roles of oxPS under pathophysiological conditions.
From the 80s to the 90s, a number of studies on roles of lipid peroxidation in a variety of diseases and/or animal disease models were carried out. [bib_ref] A new direction in the study of carbon tetrachloride hepatotoxicity, Reckenagel [/bib_ref] [bib_ref] Lipid peroxidation and membrane damage of the heart, Ytrehus [/bib_ref] [bib_ref] The free radical pathology and the microcirculation in the major central nervous..., Demopoulos [/bib_ref] However, whether generation of lipid hydroperoxides resulted in or from diseases was usually argued. It is assumed to be due to the absence of precise methods for analyzing lipid hydroperoxides at that time. Recent development of analytical methods using mass spectrometry 4 as well as preparation of highly pure lipid hydroperoxides [bib_ref] Preparation of pure lipid hydroperoxides, Ibusuki [/bib_ref] have revealed the following novel pathophysiological roles of oxidized phospholipids: specific epitopes of innate immunity receptors, modification of intracellular signal transduction, pro-and anti-inflammatory activities, enhancement of reactive oxygen species (ROS) generation, angiogenesis, calcification of atherosclerotic plaques, inhibition of acquired immunity and enhancement of blood coagulation. [bib_ref] Generation and biological activities of oxidized phospholipids, Bochkov [/bib_ref] [bib_ref] Lipid peroxidation generates biologically active phospholipids including oxidatively N-modified phospholipids, Devies [/bib_ref] These roles of oxidized phospholipids manifested in the above studies were predominantly related to those of oxidatively modified products of phosphstidylcholine (PC), the most abundant phospholipid in the membrane. Phosphatidylserine (PS) is also known to be a preferential target of in vivo oxidation. [bib_ref] A role for oxidative stress in apoptosis: oxidation and externalization of phosphatidylserine..., Kagan [/bib_ref] The aim of this review is to exclusively focus on the production of oxidized phosphatidylserine (oxPS) and its biological activities.
## Production of oxps
It is well known that non-enzymatic oxidation of phospholipids including PS in membranes can be initiated by free radicals or non-radical ROS under many pathological conditions. [bib_ref] Generation and biological activities of oxidized phospholipids, Bochkov [/bib_ref] Here, selective, but not random oxidation of negatively charged phospholipids such as PS and cardiolipin (CL) is described in detail.
Asymmetric distribution of phospholipids in plasma membranes has first postulated the concept of "phospholipid asymmetry" suggesting that phospholipids in plasma membrane are distributed asymmetrically between two leaflets. Indeed, in normal cells the cholinecontaining phospholipids, PC and sphingomyelin, reside mainly in the outer leaflet of plasma membrane, whereas the aminophospholipids, PS and, to a lesser extent, phosphatidylethanolamine (PE), are confined to the inner leaflet. [bib_ref] Membrane phospholipids asymmetry and signal transduction, Verkleij [/bib_ref] It seems that at least three lipid translocators are related to maintain this asymmetry. ATP-dependent flippases, members of type 4 P-type ATPases, also re-ferred to as aminophospholipid translocases, are specific for aminophospholipids, with a preference for PS over PE, and catalyze the inward translocation of aminophospholipids. [bib_ref] P4 ATPases -lipid flippases and their role in disease, Folmer [/bib_ref] Another ATP-dependent translocator floppases, members of ATP-binding cassette transporters, catalyze the efflux of phospholipids from the inner to the outer leaflet with little selectivity for polar head group of the phospholipids. [bib_ref] Transbilayer (flip-flop) lipid motion and lipid scrambling in membranes, Contreras [/bib_ref] Energy-independent scramblases are non-selective and catalyze bidirectional transbilayer movement of phospholipids Ca 2+ -dependently but ATPindependently. [bib_ref] Phospholipid scramblase: an update, Bevers [/bib_ref] [bib_ref] Calcium-dependent phospholipid scramblase activity of TMEM16 protein family members, Suzuki [/bib_ref] PS externalization to the cell surface commonly found in apoptotic cells is likely to be associated with the inhibition of flippases as well as the activation of scramblases in the plasma membrane. [bib_ref] Programmed cell clearance: molecular regulation of the elimination of apoptotic cell corpses..., Fadeel [/bib_ref]
## Oxps generation in apoptotic cells
There is accumulating evidence showing that apoptosis induced by various stimuli in a number of different cell types is accompanied by preferential oxidation of PS. [bib_ref] A role for oxidative stress in apoptosis: oxidation and externalization of phosphatidylserine..., Kagan [/bib_ref] [bib_ref] Random versus selective membrane phospholipid oxidation in apoptosis: role of phosphatidylserine, Fabisiak [/bib_ref] [bib_ref] Bcl-2 inhibits selective oxidation and externalization of phosphatidylserine during paraquat-induced apoptosis, Fabisiak [/bib_ref] [bib_ref] NADPH oxidase-dependent oxidation and externalization of phosphatidylserine during apoptosis in DM-SO-differentiated HL-60..., Arroyo [/bib_ref] [bib_ref] Selective peroxidation and externalization of phosphatidylserine in normal human epidermal keratinocytes during..., Shvedova [/bib_ref] [bib_ref] Depletion of Bcl-2 by an antisense oligonucleotide induces apoptosis accompanied by oxidation..., Koty [/bib_ref] [bib_ref] Phosphatidylserine peroxidation/externalization during staurosporine-induced apoptosis in HL-60 cells, Matsura [/bib_ref] Furthermore, it has been reported that oxidation and externalization of PS almost simultaneously occur during apoptosis. [bib_ref] Fine-tuning phagocytic clearance of apoptotic cells by phosphatidylserine oxidation, Matsura [/bib_ref] [bib_ref] Oxidative lipidomics of apoptosis: redox catalytic interactions of cytochrome c with cardiolipin..., Kagan [/bib_ref] As the source of oxidizing equivalents required for redox catalysis of PS oxidation was not identified, we determined which kind of ROS oxidized PS in the cells undergoing apoptosis. [bib_ref] Endogenously generated hydrogen peroxide is required for execution of melphalan-induced apoptosis as..., Matsura [/bib_ref] We used H 2 O 2 -resistant HP100 cells derived from HL-60 cells, which expressed catalase (CAT) 2.5 times more than HL-60 cells but contained the same levels of primary antioxidant enzymes (glutathione peroxidase and superoxide dismutase) and apoptosis-related proteins (Bcl-2 and Bax) as HL-60 cells. HP100 cells but not HL-60 cells exerted a higher resistance to apoptosis induced by anti-cancer reagent melphalan (Mel). This resistance to Mel-induced apoptosis in HP100 cells was abolished by pretreatment with a CAT inhibitor 3-amino-1,2,4-triazole (3-AT), suggesting that overexpression of CAT in HP100 cells was mainly responsible for their resistance to Mel-induced apoptosis. Treatment of HL-60 cells with Mel induced ROS production in the cells. However, no increase in ROS generation in HL-60 cells co-treated with exogenous CAT following Mel treatment. On the other hands, there was no ROS generation in naïve HP100 cells both before and after Mel treatment although Mel caused ROS production in HP100 cells pretreated with 3-AT as well. Thus, it was suggested that ROS generated in the cells treated with Mel is predominantly H 2 O 2 . Mel induced also PS oxidation and externalization as well as cytochrome c (cyt c) from mitochondria into cytosol in HL-60 cells but not HP100 cells. In addition, CAT inhibition by 3-AT restored the sensitivity of HP100 cells to PS oxidation and externalization after Mel exposure. This suggested that Mel-induced H 2 O 2 indeed plays a pivotal role in implementation of apoptosis (PS oxidation, PS externalization and cyt c release) as a required messenger.
## The presence of oxps on the apoptotic cell surface
We demonstrated for the first time that oxPS exists both within and on the surface of apoptotic cells. [bib_ref] The presence of oxidized phosphatidylserine on Fasmediated apoptotic cell surface, Matsura [/bib_ref] To determine PS on the cell surface, Jurkat cells following anti-Fas antibody treatment were labeled with membrane-impermeable fluorescamine, a probe for visualizing lipids that contain primary amino groups. Their total lipids were extracted and subjected to twodimensional high-performance thin-layer chromatography (HPTLC). Thereafter the HPTLC plate was first sprayed with N,N,N',N'-tetramethyl-p-phenylenediamine dihydrochloride (TMPD) to detect the oxidation of PS and then exposed to UV lights to determine externalized PS. Since fluorescamine and TMPD interact with a primary amine in the polar head group and hydroperoxide in acyl residue of oxPS, respectively, we can detect both externalization and oxidation of PS on the same HPTLC plate, i.e., prove the presence of oxPS on the surface of the cells. Additionally, to determine PS unbound to fluorescamine, i.e., PS within the cells, the same HPTLC plate was sprayed with ninhydrin, another probe that reacts with primary amino groups. The above methods enabled direct detection of oxPS both within and on the surface of cells on the same HPTLC plate. The experiments using the above methods suggested that treatment of Jurkat cells with anti-Fas antibody increased oxPS within the cells and caused oxPS to appear on the cell surface.
## The mechanisms underlying oxidation of ps during apoptosis
It is considered that a potential mechanism underlying selective oxidation of CL and PS by H 2 O 2 during apoptosis relies on the specific interactions between positively charged cyt c (net charge is +8e at neutral pH) and negatively charged phospholipids such as PS in the cytosolic leaflet of the plasma membrane and CL in the mitochondria. [bib_ref] Oxidative lipidomics of apoptosis: redox catalytic interactions of cytochrome c with cardiolipin..., Kagan [/bib_ref] [bib_ref] Cytochrome c/cardiolipin relations in mitochondria: a kiss of death, Kagan [/bib_ref] Cyt c is a globular protein containing heme and its heme iron has 6 coordination bonds. In cyt c, heme iron has two axial bonds: one with His 17 on the proximal side of heme and one with Met 80 on the distal side. [bib_ref] Oxidative lipidomics of apoptosis: redox catalytic interactions of cytochrome c with cardiolipin..., Kagan [/bib_ref] When cyt c binds to membrane PS or CL with electrostatic forces, hydrophobic interactions and hydrogen bonding, the protein globule is partially unfolded, likely leading to disruption of Fe-Met 80 coordination bond followed by enhancement of reactivity of cyt c to ROS such as H 2 O 2 . [bib_ref] Oxidative lipidomics of apoptosis: redox catalytic interactions of cytochrome c with cardiolipin..., Kagan [/bib_ref] [bib_ref] Cytochrome c/cardiolipin relations in mitochondria: a kiss of death, Kagan [/bib_ref] Cyt c seems to get peroxidase activity by such conformational changes. Presumably, highly oxidized heme (compound I or II) and proteinbased tyrosyl radical formed in the presence of H 2 O 2 in PS-or CL-activated peroxidase forms of cyt c can subtract an electron from unsaturated acyl chains of PS or CL followed by the addition of oxygen and formation of oxPS (PSOOH) or oxCL (CLOOH), similar to peroxidase reaction to cyclooxygenase (COX). [bib_ref] Oxidative lipidomics of apoptosis: redox catalytic interactions of cytochrome c with cardiolipin..., Kagan [/bib_ref] Newly formed oxPS (PSOOH) or oxCL (CLOOH) may function as substrates for cyt c peroxidase activity and further propagate lipid peroxidation, even in the absence of H 2 O 2 .
In cell-free model experiments, 27 cyt c-derived tyrosyl radical was measured by low-temperature electron paramagnetic resonance spectroscopy in the presence of cyt c, phospholipid-containing liposomes and H 2 O 2 . The production of tyrosyl radical depended strongly on the presence of PS-containing liposomes, supporting the above hypothesis indirectly.
It has been suggested that peroxidase activity of cyt c is closely related to PS oxidation and externalization as mentioned above. Therefore, we determined whether departure of cyt c from mitochondria is required for production of oxPS as well as its externalization during Fasmediated apoptosis in Jurkat cells using two selective inhibitors of caspase-8 (Casp-8) and caspase-3 (Casp-3) functioning upstream and downstream from mitochondria, respectively. [bib_ref] Mechanisms underlying production and externalization of oxidized phosphatidylserine in apoptosis: involvement of..., Yamashita [/bib_ref] As shown in [fig_ref] Figure 1: Effect of caspase inhibitors on cyt c release into the cytosol [/fig_ref] , inhibition of Casp-8 reduced mitochondrial cyt c release into the cytosol [fig_ref] Figure 1: Effect of caspase inhibitors on cyt c release into the cytosol [/fig_ref] , the amount of oxPS not only within but also on the surface of Jurkat cells [fig_ref] Figure 1: Effect of caspase inhibitors on cyt c release into the cytosol [/fig_ref] , Casp-3 activation, and apoptotic cell number after treatment with anti-Fas antibody. In contrast, selective inhibition of Casp-3 was unable to suppress cyt c release [fig_ref] Figure 1: Effect of caspase inhibitors on cyt c release into the cytosol [/fig_ref] , and the amount of oxPS both within and on the surface of the cells after anti-Fas antibody [fig_ref] Figure 1: Effect of caspase inhibitors on cyt c release into the cytosol [/fig_ref] as expected, although it inhibited activation of Casp-3 and apoptosis. These results strongly suggested that mitochondrial event, especially cyt c departure from mitochondria plays a critical role in production of oxPS within the cells and subsequent its appearance on the cell surface during apoptosis.
## Production of oxps under pathological conditions (in vivo)
Pulmonary phospholipid peroxidation after inhalation exposure of mice to single-walled carbon nanotubes was identified in three relatively minor classes of anionic phospholipids, PS, CL and phosphatidylinositol (PI). [bib_ref] Global phospholipidomics analysis reveals selective pulmonary peroxidation profiles upon inhalation of single-walled..., Tyurina [/bib_ref] This non-random peroxidation of phospholipids coincided with the accumulation of apoptotic cells in the lung. In another animal models of lung injury (hyperoxic acute lung injury 30 and gamma-radiation-induced lung injury 31 ), the preferential formation of oxPS and oxCL in the lung was also detected. These lung injuries were all accompanied by apoptotic cell accumulation in the lung.
In mouse model of Alzheimer disease (AD), which showed aberrant expression of human myeloperoxidase in astrocytes, there was selective accumulation of two anionic phospholipid hydroperoxides, oxPS (PSOOH) and oxPI (PIOOH) in the brains. [bib_ref] Aberrant expression of myeloperoxidase in astrocytes promotes phospholipid oxidation and memory deficits..., Maki [/bib_ref] Furthermore, in postmortem brain samples from AD patients, accumulation of the same individual molecular species of oxPS and oxPI was observed. [bib_ref] Aberrant expression of myeloperoxidase in astrocytes promotes phospholipid oxidation and memory deficits..., Maki [/bib_ref] In addition to the above brain diseases, it has been demonstrated that selective oxidation of CL and PS in rat cortical neurons is triggered during staurosporine-induced apoptosis. 33 Total body irradiation to mice induced intestinal injury accompanied by apoptosis as well as selective accumulation of oxCL and oxPS in small intestines. [bib_ref] Oxidative lipidomics of gamma-irradiation-induced intestinal injury, Tyurina [/bib_ref] It has been reported that anti-phospholipid antibodies in sera from patients with alcoholic liver disease (ALD) target apoptotic cells by specifically recognizing oxPS, suggesting the production of oxPS in ALD. [bib_ref] Anti-phospholipid antibodies associated with alcoholic liver disease target oxidized phosphatidylserine on apoptotic..., Vay [/bib_ref]
## Bioactivities of oxps
Role in apoptotic cell clearance It has been established that PS serves as a recognizable "eat-me" signal for phagocytes through its translocation from the inner to the outer leaflet of the plasma membrane during apoptosis. [bib_ref] Mechanisms of phosphatidylserine exposure, a phagocyte recognition signal, on apoptotic T lymphocytes, Verhoven [/bib_ref] However, oxPS also appears on the surface of apoptotic cells during apoptosis. [bib_ref] The presence of oxidized phosphatidylserine on Fasmediated apoptotic cell surface, Matsura [/bib_ref] Therefore, it is important to determine whether oxPS on the surface of apoptotic cells enhances the recognition and engulfment of apoptotic cells by phagocytes. Several receptors for apoptotic cell uptake have been identified on the surface of the phagocytes. They include the lectins that bind altered sugars on apoptotic cells, CD36 (in conjunction with integrins alphaVbeta3 and alphaVbeta5) that binds thrombospondin, LRP1/CD91 (in conjunction with calreticulin) that binds complement C1q, CD14 that binds intercellular adhesion molecule 3 (ICAM3), and the scavenger receptors that bind oxidized LDL. [bib_ref] Clearing the dead: apoptotic cell sensing, recognition, engulfment, and digestion, Hochreiter-Hufford [/bib_ref] Furthermore, PS is recognized either directly via receptors such as brain angiogenesis inhibitor 1 (BAI1), T-cell immunoglobulin-and mucin-domain-containing 4 (TIM-4), TIM-1, and Stabilin-2 or indirectly via bridging molecules such as milk fat globule-EGF factor 8 (MFG-E8), growth-arrest-specific 6 (Gas6), or protein S. [bib_ref] Clearing the dead: apoptotic cell sensing, recognition, engulfment, and digestion, Hochreiter-Hufford [/bib_ref] MFG-E8 is expressed and secreted by professional phagocytes, associates with integrins alphaVbeta3/ alphaVbeta5 on phagocytes, and binds PS on apoptotic cells. [bib_ref] Identification of a factor that links apoptotic cells to phagocytes, Hanayama [/bib_ref] As for oxPS, it has been reported that the oxPS, but not non-oxidized PS, serves as a preferred ligand for class B scavenger receptor CD36-mediated phagocytosis by macrophages. [bib_ref] Oxidized phosphatidylserine-CD36 interactions play an essential role in macrophage-dependent phagocytosis of apoptotic..., Greenberg [/bib_ref] Moreover, it has been shown that MFG-E8 preferentially interacts with oxPS, and to a lesser extent, with non-oxidized PS. [bib_ref] Milk fat globule epidermal growth factor 8 (MFG-E8) binds to oxidized phosphatidylserine:..., Borisenko [/bib_ref] In another study, [bib_ref] A role for oxidative stress in apoptosis: oxidation and externalization of phosphatidylserine..., Kagan [/bib_ref] liposomes containing oxPS inhibited phagocytosis of apoptotic cells more potently than non-oxidized PS. Furthermore, non-apoptotic cells treated with liposomes containing both oxPS and non-oxidized PS were more efficiently phagocytosed than cells treated with nonoxidized PS alone. [bib_ref] A role for oxidative stress in apoptosis: oxidation and externalization of phosphatidylserine..., Kagan [/bib_ref] Taken together, these findings indicate that oxPS may act in combination with naïve PS as an important signal on the cell surface to facilitate the recognition of apoptotic cells. In other words, oxPS is likely to en-hance the clearance of apoptotic cells by phagocytes. In addition, it has been reported that oxPS acts as a "nonenzymatic scramblase" to facilitate translocation of both PS and oxPS molecules into the cell surface. [bib_ref] Oxidation of phosphatidylserine: a mechanism for plasma membrane phospholipid scrambling during apoptosis?, Tyurina [/bib_ref]
## Anti-inflammatory activities
## Protection against endothelial barrier dysfunction and acute lung injury
The tight intercellular barrier of endothelial cell (EC) maintaining low permeability is adequately regulated by a counterbalance of barrier-protective and barrier-disruptive bioactive molecules in the circulation. [bib_ref] Polar head groups are important for barrier-protective effects of oxidized phospholipids on..., Birukova [/bib_ref] Birukova et al. showed that oxidized 1-palmitoyl-2-arachidonoylsn-glycero-3-phosphoserine (oxPAPS) potently protected pulmonary EC barrier function and induced the remodeling of pulmonary EC actin cytoskeleton. [bib_ref] Polar head groups are important for barrier-protective effects of oxidized phospholipids on..., Birukova [/bib_ref] In that study, oxPAPS inhibited the increase in permeability in human pulmonary arterial EC induced by lipopolysaccharide (LPS), a major component of the outer membrane of Gram-negative bacteria. oxPAPS also attenuated neutrophil accumulation and myeloperoxidase activation in bronchoalveolar lavage fluid in mice treated with LPS (in vivo model of acute lung injury), and suppressed lung barrier dysfunction in the mice. Furthermore, ox-PAPS diminished pulmonary EC barrier dysfunction induced by interleukin (IL)-6, an inflammatory mediator or thrombin, an edemagenic mediator. The protection by oxPAPS against thrombin-induced pulmonary EC barrier dysfunction was attributed to both attenuation of Rho-dependent pathway of endothelial contraction leading to hyperpermeability and stimulation of Racmediated EC barrier recovery. These results suggest that oxPAPS potently suppresses endothelial barrier dysfunction induced by inflammatory and edemagenic agents in vitro and in vivo potentially due to attenuation of Rho signaling as well as stimulation of Rac signaling. Since the protective effect of oxPAPS is much more than that of oxidized 1-palmitoyl-2-arachidonoyl-sn-glycero-3-phosphocholine (oxPAPC), it seems that negatively charged polar head group but not oxidized sn-2 residue of oxPAPS may play some important role in its protective ability although PAPS, non-oxidized form, shows no protective effects.
## Inhibition of lps recognition by tlr4
LPS is a potent activator of macrophages and a causal agent of endotoxin shock. [bib_ref] LPS/TLR4 transduction pathway, Lu [/bib_ref] LPS is well known to induce production of pro-inflammatory mediators such as tumor necrosis factor-alpha (TNF-alpha), IL-1beta, IL-6, ROS and nitric oxide (NO), leading to death from endotoxin shock in animal models. [bib_ref] Induction of heat shock protein 70 by zinc-bis-(DL-hydrogenaspartate) reduces cytokine liberation, apoptosis,..., Klosterhalfen [/bib_ref] [bib_ref] Beneficial pleiotropic actions of melatonin in an experimental model of septic shock..., Carrillo-Vico [/bib_ref] Moreover, it has been shown that LPS enhances cytokine and/or NO pro-duction in macrophages, [bib_ref] 2-Aminopurine inhibits lipopolysaccharide-induced nitric oxide production by preventing IFN-beta production, Sugiyama [/bib_ref] microglial cells [bib_ref] Inhibitory effect of Curcuma comosa on NO production and cytokine expression in..., Jantaratnotai [/bib_ref] and ECs. [bib_ref] Lipopolysaccharide activates endothelial nitric oxide synthase through protein tyrosine kinase, Huang [/bib_ref] The process of these cellular responses involves recognition of LPS by Toll-like receptor (TLR) 4 on the surface of the host cells, followed by activation of nuclear factor (NF)-kappa B.TLR is a receptor family protein related to innate immunity and to date, 12 functional TLRs have been identified in mice and 10 in humans. [bib_ref] Toll-like receptors and their crosstalk with other innate receptors in infection and..., Kawai [/bib_ref] TLR4 has now been established as the receptor for LPS. The activation of innate immunity response by LPS starts from the interaction of LPS with LPS-binding protein (LBP). LBP in the serum removes LPS from the outer membrane of the bacteria and delivers LPS to soluble CD14 (sCD14) and/or membrane-bound CD14 and then LPS bound to CD14 is transferred to TLR4-myeloid differentiation 2 (MD-2) complex. The formation of dimeric LPS/MD-2/TLR4 complex initiates the intracellular signaling. [bib_ref] Molecular simplification of lipid A structure: TLR4-modulating cationic and anionic amphiphiles, Calabrese [/bib_ref] According to a recent study, 52 oxPAPS inhibited LPS-induced elevation of E-selectin mRNA in human umbilical vein endothelial cells (HUVECs), indicating that it acted as an antagonist on induction of TLR4 downstream genes. oxPAPS prevented binding of LPS to LBP in an in vitro competitive assay. Moreover, oxPAPS formed complexes with sCD14 and prevented interaction of LPS with sCD14 in vitro.
Taken together, it was suggested that oxPAPS binds to LBP and sCD14, thus preventing recognition of LPS via MD-2/TLR4 complex, leading to inhibition of LPSinduced inflammatory reaction.
## Inhibition of respiratory burst
Neutrophils are recruited to inflamed sites upon infections and exert microbicidal activities. The assembly and activity of NADPH oxidase (NOX) are essential for neutrophil microbicidal activities. Neutrophils engulf invading microbes and kill them by ROS such as hydroxyl radicals and hypochlorous acids derived from superoxide anions generated by NOX together with microbicidal peptides and proteases in phagolysosomes. [bib_ref] Priming of the neutrophil NADPH oxidase activation: role of p47phox phosphorylation and..., El-Benna [/bib_ref] [bib_ref] Neutrophils at work, Nauseef [/bib_ref] Neutrophil NOX (NOX2), also referred to as respiratory burst oxidase is a multicomponent enzyme system composed of membrane proteins (p22phox and gp91phox, which form cytochrome b558) and cytosolic proteins (p47phox, p67phox, p40phox and Rac2), which assemble at membrane sites upon cell activation. The importance of this enzyme in host defenses is highlighted by the fact that loss of function mutations of NOX subunits cause chronic granulomatous disease in which the phagocyte enzyme is dysfunctional, leading to life-threatening bacterial and fungal infections. In contrast, excessive ROS generation can damage surrounding tissues. Thus, NOX activation and ROS production have to be tightly regulated.
It has been shown that ROS produced by activated neutrophils promotes oxPS production. [bib_ref] NADPH oxidase-dependent oxidation and externalization of phosphatidylserine during apoptosis in DM-SO-differentiated HL-60..., Arroyo [/bib_ref] reported that oxPAPS inhibited ROS production in phorbol myristate acetate (PMA)-or formyl-methionylleucyl-phenylalanine (FMLP)-stimulated neutrophils in a dose dependent manner. [bib_ref] The oxidation state of phospholipids controls the oxidative burst in neutrophil granulocytes, Blüml [/bib_ref] In contrast, unoxidized PAPS dose-dependently increased the production of ROS in neutrophils, suggesting that there is an oxidation state-dependent regulatory role of oxPS on neutrophil respiratory burst formation. oxPAPS did not inhibit upregulation of CD11b which is involved in adhesion of activated neutrophil to the endothelium, or phagocytosis. It has been known that MAPK (ERK1/2 and p38) are activated during respiratory burst and there is involvement of these kinases in the production of ROS. [bib_ref] Phosphorylation of the NADPH oxidase component p67(PHOX) by ERK2 and p38MAPK: selectivity..., Dang [/bib_ref] However, oxPAPS did not inhibit activation of ERK1/2 and p38 in neutrophils stimulated with either PMA or FMLP.
Taken together, it was shown that oxPAPS inhibited respiratory burst in neutrophils induced by various stimuli without inhibiting MAPK activation.
## Atherogenic activities
## Induction of vegf via upr
Endoplasmic reticulum (ER) is the site for proper folding of newly synthesized proteins and formation of three-dimensional conformation of proteins, and only proteins assured of quality in ER are released into the secretary pathway. ER stress induced by hypoxia, nutrient deprivation, acidosis and certain chemicals disturbs the protein quality control leading to accumulation of the incorrect folded proteins in ER, and this triggers the activation of the following three ER transmembrane proteins, which generate an adaptive response called the unfolded protein response (UPR): inositol-requiring enzyme 1 (IRE1), double-stranded RNA-activated protein kinase (PKR)-like ER kinase (PERK) and activating transcription factor-6 (ATF6). [bib_ref] The unfolded protein response as a target for cancer therapy, Nagelkerke [/bib_ref] The activation of these transducer proteins triggers signaling cascades, which induce downstream an adaptive UPR via protein kinases and transcription factors. The PERK/eIF2alpha/ATF4 and IRE1/ASK1/JNK cascades can trigger the induction of vascular endothelial growth factor (VEGF). [bib_ref] Endoplasmic reticulum stress in age-related macular degeneration: trigger for neovascularization, Salminen [/bib_ref] It has been shown that oxPAPS stimulates gene expression of well-characterized angiogenesis inducers (VEGF, IL-8 and COX-2) in HUVECs. [bib_ref] Oxidized phospholipids stimulate angiogenesis via autocrine mechanisms, implicating a novel role for..., Bochkov [/bib_ref] oxPAPS induced VEGF via activation of ATF4 branch of UPR in HUVECs. [bib_ref] ATF4-dependent transcription is a key mechanism in VEGF up-regulation by oxidized phospholipids:..., Oskolkova [/bib_ref] According to a recent report, VEGF induction by oxPAPS in ECs depended on microR-NA-663 (miR-663). [bib_ref] Permissive role of miR-663 in induction of VEGF and activation of the..., Afonyushkin [/bib_ref] Given that miR-663 is required for oxPAPC-induced elevation of ATF4 and its target gene TRIB (tribbles pseudokinase), [bib_ref] Permissive role of miR-663 in induction of VEGF and activation of the..., Afonyushkin [/bib_ref] it is suggested that ox-PAPS may also induce ATF4 and its downstream gene VEGF in a miR-663-dependent manner, playing a role in angiogenesis especially in atherosclerotic plaques.
## Procoagulation
Protein C is a vitamin-K dependent anti-coagulant serine protease serving as a negative feedback regulator against coagulation. [bib_ref] Activated protein C: a regulator of human skin epidermal keratinocyte function, Mckelvey [/bib_ref] When thrombin generated by blood coagulation binds to thrombomodulin (TM) on the EC membrane, protein C is activated by thrombin-TM complex to be an activated protein C (APC). This conversion is augmented by a specific receptor for protein C, endothelial cell protein C receptor. APC converts activated coagulation factors V (Va) and VIII (VIIIa) to the inactive forms by proteolysis with an APC cofactor protein S, resulting in inhibition of blood coagulation. In addition to its role in coagulation, APC decreases inflammatory responses by inhibition of nuclear translocation of NF-kappa B. [bib_ref] Activated protein C: a regulator of human skin epidermal keratinocyte function, Mckelvey [/bib_ref] Protein C inhibitor (PCI) is a serine protease inhibitor belonging to the class of serpins. [bib_ref] The serpins are an expanding superfamily of structurally similar but functionally diverse..., Silverman [/bib_ref] PCI, originally an inhibitor of APC, inactivates lots of other serine pro-teases including blood coagulation factors, fibrinolytic enzymes, tissue kallikrein, and the sperm protease acrosin. [bib_ref] Regulation of protein C inhibitor (PCI) activity by specific oxidized and negatively..., Malleier [/bib_ref] It has been shown that oxPAPS, PAPS and oxPAPE directly bind PCI and stimulate its inhibitory activity toward APC. [bib_ref] Regulation of protein C inhibitor (PCI) activity by specific oxidized and negatively..., Malleier [/bib_ref] Positively charged amino acids in the Hhelix of PCI were involved not only in heparin binding, but also in the binding of oxPAPS, PAPS and oxPAPE.
Given that PCI and annexin V, a protein binding to the head group of PS, were found to be endogenously colocalized in atherosclerotic plaques, it was suggested that oxPAPS in vivo could promote blood coagulation and thrombus formation as well as inflammation at sites of tissue injury by stimulating the inhibition of APC by PCI.
immunosuppressive activities oxPAPS inhibited the proliferation of peripheral blood T cells induced by anti-CD3/CD28 monoclonal antibody, demonstrating a novel immunosuppressive molecule for adaptive immunity. [bib_ref] Oxidized phospholipids induce anergy in human peripheral blood T cells, Seyerl [/bib_ref] In that study, oxPAPC prevented cyt c oxCL the development of Th1-type responses, T cell proliferation and the induction as well as the effector phase of CD8+ effector cytotoxic T lymphocytes induced through stimulation via TCR/CD3 complex, although there were no data for oxPAPS. Furthermore, T cells activated in the presence of oxPAPC failed to proliferate in response to restimulation, a phenomenon called anergy. If oxPS functions similarly to oxPC, oxPS may also induce T cell anergy, resulting in avoiding overwhelming Th1driven immune responses at the inflammation sites. Further examinations for oxPS will be required to resolve the above issues.
## Concluding remarks
The oxidative stress associated with apoptosis results in the selective oxidation of PS in the cytoplasmic layer of the plasma membranes, followed by egress of oxPS to the surface of the apoptotic cells and effective recognition of oxPS by phagocytes. This is no longer than a mere non-specific oxidation of phospholipids, but seems to be a finely tuned system via non-random oxidation of PS for clearance of apoptotic cells by phagocytes to prevent the inflammation [fig_ref] Figure 2: Selective oxidation of PS and externalization of oxPS in apoptotic cells as... [/fig_ref]. oxPS has some biological activities such as anti-inflammatory activities, atherogenic activities, and immunosuppressive activities other than the role in apoptotic cell clearance as mentioned above.
In contrast, oxPC has more novel functions and is recognized as a lipid mediator. In the future, we have to examine using highly pure oxPS whether oxPS has the same functions as oxPC and furthermore detect undiscovered biological activities of oxPS. In addition, it is also important to detect the specific receptor for oxPS in the phagocytes.
The author declares no conflict of interest.
[fig] Figure 1: Effect of caspase inhibitors on cyt c release into the cytosol (A) as well as oxidation of PS exposed to the cell surface (B) in Jurkat cells after treatment with anti-Fas antibody. 28 a P < 0.01 versus control, b P < 0.01 versus anti-Fas antibody alone. [/fig]
[fig] Figure 2: Selective oxidation of PS and externalization of oxPS in apoptotic cells as well as recognition of oxPS on the cell surface by macrophages. Its proposed mechanisms are indicated by bold arrows. [/fig]
|
Analysis of the effect of adjuvant radiotherapy on outcomes and complications after radical hysterectomy in FIGO stage IB1 cervical cancer patients with intermediate risk factors (GOTIC Study)
Background: There are no definitive criteria for identifying which patients with The International Federation of Gynecology and Obstetrics (FIGO) stage IB cervical cancer will benefit from adjuvant therapy after radical hysterectomy. The aims of this study were to clarify the efficacy of adjuvant therapy and assess complications after radical hysterectomy in patients with FIGO stage IB1 cervical cancer with intermediate risk factors. Methods: Between January 2005 and December 2009, the medical records of 75 stage IB1 patients' intermediate risk factors (i.e., tumor size 2-4 cm, lymphovascular involvement, and/or deep stromal invasion >1/2) who underwent radical hysterectomy at six institutions were collected, and these patients were enrolled in this nonrandomized retrospective study. We simplified the criteria of intermediate risk factors as much as possible, as the criteria adopted in some clinical studies are complicated in practice. Results: The patients were grouped according to the receipt of adjuvant therapy as follows: 46 patients, no further treatment; 19 patients, external beam radiation treatment, including 9 patients who received brachytherapy; 5 patients, concurrent chemoradiotherapy (CCRT); and 5 patients, chemotherapy (CT). The clinical outcomes and complications in each group were analyzed. After an average follow-up of 82.6 months (range, 24-135 months), only one patient with all three risk factors who received radiotherapy (RT) experienced recurrence. Excluding this patient, the remaining patients who received RT, CCRT, or CT had two or three risk factors. Lymphedema was significantly more common among patients who received RT or CCRT, whereas the incidence of ileus and ureteral obstruction was not different among the treatment groups. However, an unsutured peritoneum increased the risk of ileus. Conclusions: The findings of this study suggest that RT and CCRT after radical hysterectomy are not beneficial in patients with intermediate risk factors. In particular, RT and CCRT appeared to increase the incidence of lymphedema. A prospective randomized study is needed to verify the findings of this study.
# Background
Cervical cancer is the fourth most common cancer among women worldwide and the second most diagnosed cancer in developing countries [bib_ref] Cancer of the cervix uteri, Bermudez [/bib_ref]. Radical hysterectomy with pelvic lymphadenectomy has been a primary treatment in women with stage IB cervical cancer, and the procedure is associated with a 5-year survival rate of 87-92 % [bib_ref] Primary management of early stage cervical cancer (IA1-IB) and appropriate selection of..., Gray [/bib_ref]. Radiotherapy (RT) is a feasible technique that provides similar outcomes as radical hysterectomy [bib_ref] Primary management of early stage cervical cancer (IA1-IB) and appropriate selection of..., Gray [/bib_ref] [bib_ref] Randomised study of radical surgery versus radiotherapy for stage Ib-IIa cervical cancer, Landoni [/bib_ref] [bib_ref] Concurrent chemotherapy and pelvic radiation therapy compared with pelvic radiation therapy alone..., Peters 3rd [/bib_ref]. Surgery enables pathological examination by surgeons, permitting identification of risk factors for cancer recurrence. In general, patients with parametrial invasion, a positive vaginal margin, or positive pelvic lymph nodes who are diagnosed as being at high risk are assigned to receive adjuvant therapies. By contrast, large tumor size, deep stromal invasion, and lymphovascular invasion are classified as intermediate risk factors, and adjuvant therapy for patients with these risk factors remains controversial. Although the criteria for intermediate risk factors defined by the Gynecologic Oncology Group (GOG) study [bib_ref] A randomized trial of pelvic radiation therapy versus no further therapy in..., Sedlis [/bib_ref] have been widely accepted, each institution offers adjuvant therapy to patients based on its own protocol. Thus, there is always a problem of interpreting discrepant results between studies because of the different study models. On the contrary, it has been recognized that postoperative RT results in a significant increase in the incidence of adverse events affecting quality of life, such as lymphedema, ileus, and ureteral obstruction [bib_ref] Treatment results of adjuvant chemotherapy after radical hysterectomy for intermediateand high-risk stage..., Takeshima [/bib_ref] [bib_ref] Treatment of cervical cancer with adjuvant chemotherapy versus adjuvant radiotherapy after radical..., Hosaka [/bib_ref] [bib_ref] Comparison of adjuvant chemotherapy and radiation in patients with intermediate risk factors..., Lee [/bib_ref].
The GOG study concluded that pelvic RT after radical hysterectomy significantly improves progressionfree survival and benefits patients with histological types of adenocarcinoma and adenosquamous carcinoma [bib_ref] A phase III randomized trial of postoperative pelvic irradiation in stage IB..., Rotman [/bib_ref]. The aims of the current retrospective study were to evaluate the effect of RT and treatment-related morbidity after radical hysterectomy for patients with intermediate risk as defined by simplified criteria.
# Methods
Between 2005 and 2009, the medical records of 89 stage IB1 patients with intermediate risk factors (i.e., tumor size 2-4 cm, lymphovascular involvement, and/or deep stromal invasion >1/2) who underwent type III radical hysterectomy as defined by Piver et al. [bib_ref] Five classes of extended hysterectomy for women with cervical cancer, Piver [/bib_ref] and bilateral pelvic lymphadenectomy were obtained for this retrospective study from six institutions belonging to the Gynecologic Oncology Trial and Investigation Consortium of North Kanto (GOTIC): Gunma University, Tsukuba University, Jichi Medical University, National Defense Medical College, Saitama Medical University International Medical Center, and Tochigi Prefectural Cancer Center. None of the patients had received preoperative treatment such as neoadjuvant chemotherapy (CT) or RT. Informed consent was not obtained from each participant because this was a retrospective study. Instead of that, all participants were given the right to withdraw the use of the data. The protocol of this study was approved based on the necessity of the individual institutions' ethical committees.
Based on each institution's criteria, patients received adjuvant therapy, including RT, CT, or concurrent chemoradiotherapy (CCRT). RT consisted of conventional external beam (EBRT) to the pelvis (28-42 Gy) in fractions of 1.8-2.0 Gy for 28-42 days. Nine patients in the RT group also received vaginal brachytherapy (BRA) in fractions of 4-7 Gy for a total dose of 7-21 Gy. In the CCRT group, cisplatin (40 mg/m 2 ) was infused intravenously every week. In the CT group, patients received paclitaxel (175 mg/m 2 ) plus carboplatin (AUC 6) every 3 weeks for 6 cycles.
Data regarding tumor size (MS), histopathological findings, depth of stromal invasion [bib_ref] Radiotherapy and chemoradiation after surgery for early cervical cancer, Rogers [/bib_ref] , and lymphovascular space invasion (LVSI) are summarized in . All patients received regular follow-up. During follow-up, complications such as ileus and ureteral obstruction were classified according to the Common Toxicity Criteria. The severity of lymphedema was rated according to the staging system of the International Society of Lymphology as follows: stage 0, a latent or sub-clinical condition in which swelling in not evident; stage 1, temporary visible swelling that can be reduced by elevation of the limb; stage 2, clear pitting and limb elevation cannot reduce tissue swelling; and stage 3, also known as lymphostatic elephantiasis, tissue becomes extremely swollen, leading to skin changes such as acanthosis, fat deposits, and warty overgrowths. Statistically significant differences in the severity of adverse events, namely lymphedema, ileus, and ureteral obstruction, as a result of no further treatment (NFT, n = 46), RT and CCRT (n = 24), or CT (n = 5) were analyzed using the Kruskal-Wallis or chi-squared test. All tests were two-tailed, and a P value <0.05 was considered statistically significant.
The presence or absence of complications was determined, and cross-tabulation was performed when the retroperitoneum was opened or closed during surgery and when RT was or was not administered. In addition, odds ratios and 95 % confidence intervals for the presence or absence of complications were calculated using logistic regression analyses based on whether the retroperitoneum was opened or whether RT was performed. JMP ver. 9 (SAS Institution Japan Inc., Tokyo, Japan) was used for all analyses.
# Results
Eighty-nine patients were enrolled in this retrospective study from six institutions; meanwhile, 14 patients were excluded because of incompatibility with the inclusion criteria and incomplete follow-up data. Overall, the study included 51, 20, and 4 patients with squamous cell carcinoma, adenocarcinoma, and adenosquamous histology, respectively . One patient with all three intermediate risk factors who received EBRT and BRA died of a recurrent tumor in the lungs 24 months later. All other patients are alive without recurrence, and they have been followed up for an average of 84.3 months (range, 47-135 months). Lymphedema was observed more frequently in patients who received postoperative RT than in those who did not receive RT or CT (P < 0.001). The risks of ileus and ureteral obstruction were not significantly different between patients who received NFT/CT or RT/CCRT. Other radiationrelated adverse effects such as cystitis and proctitis were extremely rare [fig_ref] Table 3: The frequency of complications by treatment regimens [/fig_ref]. We further assessed the severity of lymphedema associated with RT, finding that more patients who receive RT developed grade 2 or 3 lymphedema (P < 0.001) [fig_ref] Table 4: Incidences of lymphedema grades by treatment regimensLymphedema level is rated according to... [/fig_ref].
Retroperitoneal suturing after lymphadenectomy was also assessed to evaluate the prophylactic effect of an unsutured pelvic peritoneum on lymphedema risk. The overall lymphedema rate was comparable between patients with an unsutured pelvic peritoneum and those with a sutured pelvic peritoneum [fig_ref] Table 5: Adverse events in unsutured and sutured peritoneum status [/fig_ref]. However, an unsutured pelvic peritoneum was significantly associated with ileus among patients who did not receive adjuvant RT .
# Discussion
The outcome of surgery with or without adjuvant RT in patients with early-stage cervical cancer has been debated by many researchers. In general, RT has been demonstrated to significantly decrease the risk of local recurrence. However, overall survival is not conclusively improved, whereas certain subset analyses illustrated a positive effect of RT on the reduction of recurrence, for example, the GOG 92 study suggested that RT is effective for patients with a combination of deep stromal invasion and large tumor size (≥4 cm) [bib_ref] A phase III randomized trial of postoperative pelvic irradiation in stage IB..., Rotman [/bib_ref]. Tumor size ≥4 cm has been strongly correlated with recurrence [bib_ref] A limited role for adjuvant radiotherapy after the Wertheim/Okabayashi radical hysterectomy for..., Van Der Velden [/bib_ref] [bib_ref] Are adenocarcinomas and adenosquamous carcinomas different from squamous carcinomas in stage IB..., Lai [/bib_ref] [bib_ref] The prognostic factors for patients with early cervical cancer treated by radical..., Tsai [/bib_ref] [bib_ref] Prospective surgical-pathological study of disease-free interval in patients with stage IB squamous..., Delgado [/bib_ref] [bib_ref] Intermediate-risk grouping of cervical cancer patients treated with radical hysterectomy: a Korean..., Ryu [/bib_ref]. As we assumed that tumor size ≥4 cm was a high risk factor even in the intermediate risk group, we excluded patients with tumor diameters ≥4 cm in this study. It is also accepted that stromal invasion by tumors is an important prognostic factor, and a majority of studies evaluated invasion depth via measurements in absolute millimeters or fractions of muscle layer [bib_ref] A phase III randomized trial of postoperative pelvic irradiation in stage IB..., Rotman [/bib_ref] [bib_ref] Intermediate-risk grouping of cervical cancer patients treated with radical hysterectomy: a Korean..., Ryu [/bib_ref]. Stromal invasion in fractional seconds was used in this study because variable thickness in the cervical wall might not reflect the extent of stromal invasion. In addition, measurement in fractional thirds appeared difficult, especially in patients with a thin cervical wall, and it might increase discrepancies because of the multi-institutional nature of the study. Previously, a combination of intermediate risk factors, such as large tumor size, LVSI, and stromal invasion, was associated with incremental recurrence rates of up to . In this study, only one patient (1/24) with a single risk factor received adjuvant RT, whereas 28/51 patients with multiple risk factors received adjuvant therapy [fig_ref] Table 2: shows the characteristics of intermediate risk factors in this group of patients [/fig_ref]. However, this study has some limitations, including its retrospective nature and the possible diversity of treatment modalities among institutions, resulting in heterogeneity among the treatment groups. For example, the introduction of CCRT for treating high risk cervical cancer probably influenced some institutions to utilize CCRT, as 4/5 patients who received CCRT had all three risk factors , which may result in bias in interpreting the results. In this study, only one patient with all three intermediate risk factors who received RT and BRA died of recurrent tumor in the lungs 24 months later. This finding suggests that adjuvant RT does not benefit patients with intermediate risk factors as defined in this study. These issues can be addressed via a randomized design study in the future. Another purpose of this study was to assess adverse events associated with RT, as adjuvant RT has long been known to increase complications [bib_ref] Complications of combined radical hysterectomy-postoperative radiation therapy in women with early stage..., Barter [/bib_ref] [bib_ref] Morbidity and survival patterns in patients after radical hysterectomy and postoperative adjuvant..., Fiorica [/bib_ref]. Adjuvant CT alone for post-radical hysterectomy patients has been revealed to provide a better postoperative quality of life by eliminating RT-related morbidities such as small-bowel obstruction or leg edema [bib_ref] Treatment results of adjuvant chemotherapy after radical hysterectomy for intermediateand high-risk stage..., Takeshima [/bib_ref] [bib_ref] Comparison of adjuvant chemotherapy and radiation in patients with intermediate risk factors..., Lee [/bib_ref] , as also supported by our results [fig_ref] Table 3: The frequency of complications by treatment regimens [/fig_ref]. Based on the small number of cases of ileus and ureteral obstruction, there was no difference between NFT/CT and RT/CCRT, which was consistent with a meta-analysis [bib_ref] Radiotherapy and chemoradiation after surgery for early cervical cancer, Rogers [/bib_ref] of two combined trials by Bilek et al. [bib_ref] Radical pelvic surgery versus radical surgery plus radiotherapy for stage Ib carcinoma..., Bilek [/bib_ref] and GOG 92 [bib_ref] A phase III randomized trial of postoperative pelvic irradiation in stage IB..., Rotman [/bib_ref]. Conversely, the incidence of lymphedema was significantly higher and the adverse event was of greater severity for patients who received RT or CCRT (P < 0.001). Several studies indicated that an unsutured peritoneum after pelvic lymphadenectomy reduced the risk of lymphocyst formation [bib_ref] To drain or not to drain: a retrospective study of closed-suction drainage..., Jensen [/bib_ref] [bib_ref] Drainage following radical hysterectomy and pelvic lymphadenectomy: dogma or need?, Lopes [/bib_ref]. In this study, we assessed whether the incidence of lymphedema improved when the retroperitoneum was left open. In contrast to a previous report suggesting that an unsutured peritoneum significantly reduced the risk of lymphedema [bib_ref] Radiotherapy negates the effect of retroperitoneal nonclosure for prevention of lymphedema of..., Tanaka [/bib_ref] , this strategy did not provide a significant advantage in avoiding lymphedema in this study [fig_ref] Table 5: Adverse events in unsutured and sutured peritoneum status [/fig_ref]. Lymphedema is most commonly diagnosed within the first year, but a certain number of patients manifest symptoms in later years Adverse events for patients with an unsutured pelvic peritoneum who did not receive radiotherapy. Horizontal lines represent odds ratio (diamonds) with 95 % confidence intervals (CIs)
[table] Table 2: shows the characteristics of intermediate risk factors in this group of patients. Among the 75 patients, 24 received either RT or CCRT and 5 were treated with CT alone. Only a single patient who had one intermediate risk factor received EBRT and BRA. By contrast, 19/24 (79.2 %) patients who had all three intermediate risk factors as defined in this study received adjuvant therapy. [/table]
[table] Table 3: The frequency of complications by treatment regimens [/table]
[table] Table 5: Adverse events in unsutured and sutured peritoneum status [/table]
[table] Table 4: Incidences of lymphedema grades by treatment regimensLymphedema level is rated according to the staging system by the International Society of Lymphology Abbreviations: NFT no further treatment, CT chemotherapy, RT radiation therapy Kruskal-Wallis test 25]. Thus, the possible reason for discrepant results between studies may be attributable to the follow-up duration; specifically, the average observation period was 82.6 months in our study, whereas other studies evaluated patients for 3 years after surgery. Intriguingly, our study demonstrated that an unsutured peritoneum without RT significantly increased the incidence of ileus, whereas there was no difference between an unsutured and sutured peritoneum regarding the risks of lymphedema and ureteral obstruction(Fig. 1).In the present study, postoperative adjuvant RT significantly increased adverse events for intermediate risk patients as defined in this study. To date, no trial has established a solid consensus regarding RT after surgery for early-stage cervical cancer. Although this retrospective study included a limited number of patients, the results provide useful information for further consideration in the management ofpatients with intermediate risk cervical cancer. BRA, brachytherapy; EBRT, external beam radiation therapy; CCRT, concurrent chemoradiotherapy; CT, chemotherapy; LVSI, lymphovascular space invasion; MS, tumor size; NFT, no further treatment; RT, radiation treatment; SI, stromal invasion Department of Obstetrics and Gynecology, Gunma University Hospital, Maebashi, Gunma 371-8511, Japan. 2 Faculty of Medicine, University of Tsukuba, Tsukuba, Ibaraki 305-8577, Japan. 3 Department of Obstetrics and Gynecology, Jichi Medical University Hospital, Shimotsuke, Tochigi 329-0498, Japan. 4 Department of Obstetrics and Gynecology, National Defense Medical College, Tokorozawa, Saitama 359-8513, Japan. 5 Department of Gynecologic Oncology, Saitama Medical University International Medical Center, Hidaka, Saitama 350-1298, Japan. 6 Department of Gynecology, Tochigi Cancer Center, Utusnomiya, Tochigi 320-0834, Japan. 7 Current address: Department of Gynecology, Gunma Cancer Center, Ota, Gunma 373-8550, Japan. 8 Current address: Department of Gynecology, Hyogo Cancer Center, Akashi, Hyogo 673-8558, Japan. [/table]
|
What Might Your Practice Look Like Post–Peak COVID-19?
# Introduction
The ACR Commission on General, Small, Emergency and/or Rural Practice organized a panel to discuss the response of several different types of radiology practices to the coronavirus disease 2019 (COVID-19) pandemic and planning for the resumption of services across multiple practice settings. The practice leaders represented the perspectives of general, small, and rural practices; practices serving in critical-access hospitals; in addition to a community division of a large academic institution, a national radiology practice, and a teleradiology practice. The presenters share their unique practice responses and future outlooks on the basis of the most currently available knowledge at the time while planning the initial stages of recovery during the rapidly evolving COVID-19 pandemic.
## The practice leaders
Anthony Gabriel, MD, MBA, cofounder of Radiology Partners, presented the perspective of a national radiology physician practice management company including 1,600 radiologists in 25 states.
Benjamin W. Strong, MD, chief medical officer of vRAD, discussed the teleradiology company's perspective and its response to a greater than 50% reduction of imaging volume early in the pandemic. The practice includes more than 500 radiologists serving more than 2,100 sites across all 50 states.
Cathrine E. Keller, MD, managing physician of Lake Medical Imaging, reported on the perspective of a 23-member outpatient-only radiology practice servicing primarily elderly patients in a highly competitive local market.
Daniel A. Rodgers, MD, president of Kanawha Valley Radiologists, a sixperson radiology practice in rural West Virginia, assessed the benefits and challenges the pandemic presents for small rural practices and critical-access hospital practices.
Howard B. Fleishon, MD, MMM, FACR, vice chair of the ACR Board of Chancellors, presented the perspective of the community division of Emory University as chief of radiology services. The division provides comprehensive coverage for hospitals, outpatient imaging centers, and other ventures outside of the main downtown Emory campuses in the greater Atlanta area.
Lyndon K. Jordan III, MD, FACR, president and managing partner of Wake Radiology, described the challenges facing his 50-physician radiology group in Raleigh, North Carolina, and the innovative solutions in partnering with local businesses.
## Initial response, recruitment, and future imaging volumes
Multiple panelists commented that current recruitment efforts and service expansion plans would be paused or significantly reconsidered as practices tried to model future volumes and different scenarios for staged recovery. However, all indicated that currently signed contracts would be honored, and most indicated that new hires would start on time. The smaller and outpatient-based practices both indicated that groups were short staffed at the initial pandemic and that these groups dealt with the initial imaging reduction by having partners voluntarily take additional time off the schedule or distributing extra vacation days equitably across all individuals. During the session, the national teleradiology organization highlighted infrastructure that allowed workload balancing by assigning case volumes to radiologists to balance the available cases with the desire of radiologists available to read while the national radiology practice indicated that on the basis of existing practice structure, most units formulated local solutions that varied by the individual practice size and preferences.
A common theme was that the significant volume losses across the country were because of voluntary deferral and widespread (but heterogeneously implemented) stay-at-home orders. At the time of presentation, there was significant uncertainty as to the timing, rate, and overall resumption of outpatient and elective imaging as well as the impact on patient preferences toward outpatient or hospital-based imaging options. Leaders are closely monitoring trends, exploring ways to increase efficiencies and adjust compensation models, and considering options to mitigate the uncertainties lying ahead. A specific note was made that the increased use of initial employment video interviews will likely persist after the pandemic ends to screen potential candidates safely and efficiently. Some predict an opportunity to provide urgent imaging services at outpatient facilities for patients who have been appropriately triaged from urgent care centers and telemedicine visits rather than emergency departments or hospitalbased facilities, especially during surge or peak COVID-19 volumes. Outpatient centers may leverage this opportunity to create new strategic solutions for imaging coverage through and potentially beyond the COVID-19 pandemic. This might include expanded hours of access that would allow greater temporal and physical appointment spacing for more optimal patient safety in addition to the ability to increase same-day add-on case scheduling.
## Practice consolidation and finances
Opinions were mixed on the overall impact of the pandemic on recent trends in radiology consolidation and corporatization. Some felt that highly leveraged entities may take time to reorganize or face increased risk for failure in current lending environments. Others noted that some practices, especially smaller or resource-limited groups, may be reconsidering a "go it alone" approach, especially if the pandemic is prolonged or there is a significant economic downturn. It is possible that the severe stress placed on many practices and health systems could lead to greater marketplace consolidation or bankruptcies [bib_ref] Private equity-backed hospital investments and the impact of the coronavirus disease 2019..., Lexa [/bib_ref].
All practices have been affected by reduced volumes, with some reporting greater than 80% losses [bib_ref] Impact of the coronavirus disease 2019 (COVID-19) pandemic on imaging case volumes, Naidich [/bib_ref]. Leaders are modeling various scenarios for both recovery and potential subsequent waves of infections or stay-at-home orders for the coming years. Most agreed that it is too early to tell what effect these changes may have on practice stability, and all leaders are examining how to remain viable under multiple different scenarios for short-term solvency. Panelists noted that local infection rates, system capacity, and other regional factors will have the biggest impact on imaging recovery. In this regard, singlesite or closely organized regional practices will likely have more uniform recovery conditions facilitating a cohesive plan and implementation compared with geographically dispersed practices.
## Lasting changes
All panelists agreed that remote reading and telehealth services will see a significant increase after the pandemic, even though teleradiology has been a well-established and available service for many years. This corresponds to several anecdotal reports and surveys of practices performed early in the pandemic [bib_ref] Off-site radiology workflow changes due to the coronavirus disease 2019 (COVID-19) pandemic, Quraishi [/bib_ref]. Those practices that had existing remote solutions in place were able to react quickly with off-site reading in the early days of the pandemic, while expanded deployment was significantly accelerated for those practices that had not fully embraced teleradiology previously. Several panelists noted that they were already seeing increases in the numbers of current radiologists as well as job applicants requesting to work from home for personal and family safety concerns. They noted that practices may have to adjust current job offerings to attract the best candidates if volumes return quickly and the job market tightens. During the recovery phase, hospitals may be more willing to negotiate the total number of on-site providers needed to facilitate an increase in teleradiology services.
Panelists highlighted the need to resume imaging operations safely and efficiently in the recovery and postpandemic imaging phases. Several noted that improved cleaning protocols and the use of personal protective equipment are likely to persist long after the pandemic subsides, while other measures such as eliminating waiting rooms may be relaxed gradually on the basis of local infection statistics and patient acceptance. These plans will continue to evolve with experience and knowledge as different parts of the country and world move through the pandemic at different paces [bib_ref] ACR statement on safe resumption of routine radiology care during the coronavirus..., Davenport [/bib_ref].
## Available resources
Panelists noted that it was particularly challenging to understand and apply for the various types of federal aid available during the pandemic, especially with the rapidly evolving relevant legislation and regulations. Success required close coordination with accountants and attorneys, and the ability to use preexisting relationships was highly beneficial. The ACR created a comprehensive online portal early in the pandemic to provide practices with regularly updated resources and education material as it became available and evolved, including both publicfacing and member-only content. ACR leaders who participated in the panel noted that the College is uniquely positioned to convene radiologists from multiple varied practice settings and help them learn from one another. The pandemic may accelerate investment in programs to develop radiology leaders and their ability to participate in strategic decision making for their organizations.
# Conclusions
All radiology practice types have been significantly affected by the COVID-19 pandemic. Leaders representing a diverse group of practices in the general, small, and rural community, critical-access hospitals, a community division of a large academic institution, a national radiology practice, and a teleradiology practice provide different perspectives on the immediate postrecovery phase for radiology. These individuals highlight the need to safely and effectively resume imaging services on the basis of local conditions, while developing contingency plans for potential future forced imaging reductions or service disruptions. |
Advancing Photoelectrochemical Energy Conversion through Atomic Design of Catalysts
Powered by inexhaustible solar energy, photoelectrochemical (PEC) hydrogen/ammonia production and reduction of carbon dioxide to high added-value chemicals in eco-friendly and mild conditions provide a highly attractive solution to carbon neutrality. Recently, substantial advances have been achieved in PEC systems by improving light absorption and charge separation/transfer in PEC devices. However, less attention is given to the atomic design of photoelectrocatalysts to facilitate the final catalytic reactions occurring at photoelectrode surface, which largely limits the overall photo-to-energy conversion of PEC system. Fundamental catalytic mechanisms and recent progress in atomic design of PEC materials are comprehensively reviewed by engineering of defect, dopant, facet, strain, and single atom to enhance the activity and selectivity. Finally, the emerging challenges and research directions in design of PEC systems for future photo-to-energy conversions are proposed. 2104363 (1 of 16)www.advancedsciencenews.com www.advancedscience.comReaction Mechanisms of PEC SystemPEC-driven energy conversion processes are highly complex because photoinduced charge undergoes generation,[57]separation,[58]and transfer process[59]before reaching the surface active sites for final catalytic reactions. In order to
# Introduction
Since industrialization, the global surface temperature has been rising, seriously impacting on the global ecosystem and social and economic environment.Extreme weather such as tornadoes, rainstorms, heatwaves, and droughts occur frequently.This threatens the stability of human society and hinders the sustainable development of human beings.According to the Paris Agreement,by the end of the 21st century, the rise in global average temperature must be limited to less than 2°C above DOI: 10.1002/advs.202104363 preindustrial levels. At present, more than 50 countries have announced that they will achieve carbon neutrality by the middle of this century, and nearly 100 countries are working on setting their own goals. Carbon neutrality has become a global scale movement.
Until now, fossil energy source, such as coal and oil, is still the main consumption force in the energy structure of all countries. In 2019, fossil energy accounted for 85.1% of the primary energy consumption in China,.3% in the United States,.1% in the European Union, and 87.4% in Japan.The extensive use of fossil energy has led to excessive carbon emissions. Moreover, energy-intensive industries are also the main source of carbon emissions. For example, the Haber-Bosch process for ammonia (NH 3 ) preparation (operating at 400-500°C, 100-200 atm) accounts for ≈1.5% of global greenhouse gas emissions.To achieve carbon neutrality, it is imperative to develop clean and renewable energy and upgrade the energyintensive industry to carbon-free processes. In addition, the development of carbon capture and storage technology is also an important route to reduce carbon emissions.Solar energy is an exhaustless natural resource, with 89, 300 TW of solar energy available on the earth's surface, exceeding the world's annual energy consumption.Harvesting energy directly from sunlight provides a desirable approach toward fulfilling the requirement for clean and renewable energy. Since the pioneering work of Fujishima and Honda in 1972,which reported a photoelectrochemical (PEC) system with TiO 2 as an electrode to split water, photoelectrochemistry has been considered as a leading environmentally friendly approach for solar to energy (fuel and chemical) conversion. Until now, significant advances have been achieved in PEC hydrogen (H 2 ) evolution.Recently, promising results have also been shown in PEC-driven NH 3 generation and carbon dioxide (CO 2 ) reduction to valuable chemicals.Undoubtedly, PEC-driven production of H 2 , NH 3 and reduction of CO 2 under eco-friendly and mild conditions provide a highly attractive solution to carbon neutrality [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref].However, the innovation of the PEC devices with high activity, durability, and selectivity for water-to-H 2 , nitrogen-to-NH 3 , and CO 2 -to-chemicals is highly challenging. The common photoelectrodes are composed of light-absorbing materials decorated with catalytically active cocatalysts.The efficiency of solar to fuel/chemical conversion depends on the efficiencies of photocarrier generation,separation,transport,and the final catalytic reaction.The loss of photogenerated carriers in any of these processes mentioned above will affect the final solar-to-fuel/chemical conversion efficiency and restrict the further performance enhancement of the PEC system. Technoeconomic modeling has suggested that a practical water splitting device should be able to produce a solar-to-hydrogen (STH) efficiency of >10% to compete with the conventional industrial processes (methane steam reforming).So far, this requirement has been met through the construction of two series polymer electrolyte membrane electrolyzer driven by In-GaP/GaAs/GaInNAsSb triple junction solar cells, and the overall STH efficiency can reach more than 30%.By contrast, PECdriven nitrogen reduction reaction (PEC-NRR) and CO 2 reduction reaction (PEC-CRR) are still in their infancy, whose performance is far from requirements for wide application.
Recently, thanks to the exciting progress in novel synthetic method,advanced-theoretical computations,and cutting-edge characterization techniques,tremendous breakthroughs have been achieved in nanostructured electrocatalysts toward hydrogen evolution reaction (HER),NRR,and CRR.The rationally designed electrocatalysts with tunable atomicand electronic structurehave been obtained. The performance of many cost-effective and earthabundant electrocatalysts has surpassed those of noble-metalbased electrocatalysts.Until now, special emphasis in PEC system is on the improvement of light-absorbing, charge separation, and transfer processes,but much less attention is given to the surface catalytic process on the photoelectrocatalysts. Since the catalytic reaction is mainly determined by the electronic structure of the surface atoms,the atomic design of the photoelectrocatalysts is urgently needed.Besides, appropriate atomic design can bring additional beneficial effects,such as extending the light absorption range by introducing defects,accelerating charge separation by surface faceting, etc. However, till now, there is no comprehensive review pinpointing the significance of PEC-HER, PEC-NRR, and PEC-CRR with advancements through the atomic design of PEC photoelectrocatalysts.
In this review, we focus on recent progress in the atomic design of photoelectrocatalysts to enhance PEC performance. We focus on the three key reactions for renewable energy and chemical production: HER, NRR, and CRR. For each reaction, the mechanism is discussed briefly. Our intention is to show how atomic design, such as engineering of defect, heteroatom, facet, strain, and single atoms, successfully modulates the electronic structure of photoelectrocatalysts, thus leading to excellent PEC performance. Finally, we provide some insights into the development of photoelectrocatalysts and summarize the remaining challenges and future directions in this emerging field. We hope that this review will offer helpful guidance for oriented design and optimization of photoelectrocatalysts for the research and development of highly efficient PEC systems. understand the limitations of PEC performance and develop design principles for better PEC materials, it is essential to investigate the principles and mechanisms involved in the above four processes [fig_ref] Figure 2: Schematic diagram of four processes of photoinduced charge generation, separation, transfer, and... [/fig_ref]. Given that many excellent works have given comprehensive reviewon the origin of the charge recombination during generation, separation, and transfer processes, and the corresponding strategies to prevent photogenerated charge loss,we will briefly introduce the mechanisms of water splitting, NRR, and CRR, which will help understanding how atomic design of photoelectrocatalysts significantly affects the performance of PEC systems. Note that the mechanisms of these catalytic reactions in the PEC system are very similar to those in electrocatalysis,which have been deeply investigated. In this review, the widely accepted catalytic mechanism for the specific reactions are introduced.
## Mechanisms for water splitting
PEC splitting of water into H 2 and O 2 is a highly attractive way to produce H 2 fuel.The kinetic and thermodynamic barriers in the water oxidation and reduction reactions are one of the important factors restricting the STH conversion efficiency.In PEC water splitting [fig_ref] Figure 3: a [/fig_ref] , the overall reaction is: 2H 2 O → 2H 2 + O 2 , which can be conducted in acidic, neutral, and alkaline environments [fig_ref] Figure 3: a [/fig_ref]. We take alkaline water splitting as an example to describe the generally accepted mechanisms.
For water reduction reaction at the photocathode, there are two possible mechanisms [fig_ref] Figure 3: a [/fig_ref] , i.e., Volmer-Tafel and Volmer-Heyrovsky pathways.For Volmer-Tafel pathway
[formula] M + H 2 O (l) + e − → M − H * + OH − (1) 2M − H * → 2M + H 2(2) [/formula]
For Volmer-Heyrovsky pathway
[formula] M + H 2 O (l) + e − → M − H * + OH −(3)M − * H + H 2 O (l) + e − → M + OH − + H 2 (4) [/formula]
where M is the active site and *H is intermediate during HER. Experimental and theoretical efforts suggest that both advantageous water dissociation and ideal *H adsorption free energy (ΔG *H = 0) are needed to guarantee a high catalytic HER activity in alkaline condition.For water oxidation reaction (oxygen evolution reaction, OER) at the photoanode, proposed mechanism is [fig_ref] Figure 3: a [/fig_ref] www.advancedsciencenews.com www.advancedscience.com ≈1.5% of global greenhouse gas emission.By contrast, PEC-NRR is a very attractive route to produce NH 3 at ambient temperature and atmospheric pressure.Generally, a typical NRR process involves the N 2 adsorption/activation and cleavage of the N≡N bond, the hydrogenation of intermediates and the final desorption of NH 3 .There are two widely accepted mechanisms for NRR,i.e., dissociation pathway [fig_ref] Figure 4: The possible reaction mechanisms of PEC-NRR including a [/fig_ref] and association pathway [fig_ref] Figure 4: The possible reaction mechanisms of PEC-NRR including a [/fig_ref]. In the dissociation pathway, N≡N triple bond in adsorbed N 2 cleaves first, and then the two isolated N atoms adsorbed on the catalyst surface are separately hydrogenated to form two NH 3 molecules. Due to the ultrastable N≡N triple bond, the dissociative mechanism requires extremely high energy. According to the different N 2 adsorption modes and hydrogenation sequences,the association mechanism can be further divided into three pathways, namely alternating, distal, and enzymatic pathways [fig_ref] Figure 4: The possible reaction mechanisms of PEC-NRR including a [/fig_ref].
In the alternating pathway, N 2 molecules are adsorbed on the catalyst surface in the end-on mode, and two N atoms are hydrogenated alternately to release two NH 3 molecules. This process may produce by-products (such as N 2 H 2 and N 2 H 4 ), thus reducing the Faradaic efficiency and yield of NH 3 .In the distal pathway, hydrogenation preferentially occurs on the distal N atom of the N 2 molecule, first releasing an NH 3 molecule and leaving another N atom on the surface to generate another NH 3 .For the enzymatic pathway, N 2 molecules are adsorbed on the catalyst surface in the side-on mode, and the hydrogenation process is similar to that in alternating mode.
Notably, the well-known Haber-Bosch method proceeds in the dissociation pathway, which requires considerable energy to break the ultrastable N≡N triple bond (940.95 kJ mol −1 ).This is the origin of the high energy consumption of the Haber-Bosch method. By contrast, most of catalysts for sustainable NH 3 production such as noble metal, non-noble metal, or nonmetal, are reported to catalyze NRR in the association mechanism.
## Mechanism for crr
PEC reduction of CO 2 is a promising route to convert CO 2 into valuable chemical products such as CO, CH 4 , CH 3 OH, C 2 H 4 , C 2 H 5 OH, etc.CRR process is very complex, and the products are diverse, which are highly dependent on the reaction routes. The widely acknowledged pathways include C 1 and C 2 .For the C 1 pathway [fig_ref] Figure 5: The possible reaction pathways of different CO 2 reduction products [/fig_ref] , the first electron reduces CO 2 into *CO 2 − , and the second proton-electron pair transferring to the C or O atom in *CO 2 − yields HCOOH or CO, respectively. The further proton-electron pairs can reduce the adsorbed CO (CO*) to CH 4 or CH 3 OH.Compared with the straightforward C 1 reaction routes, the reaction routes for C 2 products are more complex. To date, there are still some disputes about the possible pathways for forming C 2 products,such as *CO dimerization, *CO-COH coupling, "carbene," CO insertion mechanism, etc. In the path from *CO to C 2 H 4 or CH 3 CH 2 OH, the formation of *CO-COH is considered as the rate-determining step [fig_ref] Figure 5: The possible reaction pathways of different CO 2 reduction products [/fig_ref].Then, *CO-COH is reduced to *CH 2 CHO via several proton-electron coupled transfer steps. And *CH 2 CHO is subsequently transformed to ethylene or ethanol via hydrogenolysis or hydrogenation processes.Although possible reaction mechanisms of water splitting (OER and HER), NRR, and CRR have been proposed, revealing the detailed mechanism still remains a major challenge because of the short lifetime and low coverage of intermediates under working conditions. Combining cutting-edge in situ characterization techniques with advanced theoretical calculations opens up new exciting opportunities to reveal the "real" catalytic reactions.
## Advancing pec reactions through atomic design of photoelectrodes
As mentioned above, the photo-to-energy (fuel or chemical) conversion efficiency of the PEC system is closely related to the mechanisms of OER, HER, NRR, and CRR occurring at the photoelectrocatalysts. Recent experimental and theoretical worksdemonstrate that the catalytic mechanisms (thermodynamic barrier,kinetic barrier,rate-determining step,etc.) of the PEC-driven catalytic reactions are very sensitive to electrode surface structure. And materials with different compositions/structures exhibit quite different reactivity. Therefore, adjusting the atomic structure of photoelectrocatalysts by creating defects, strain [fig_ref] Figure 6: Schematic diagram of enhancing PEC-driven solar-tofuel/chemical conversions by the atomic design of... [/fig_ref] , dopants, or single atoms is an effective way to improve the performance of PEC devices.
## Defect engineering
Defects [fig_ref] Figure 7: a [/fig_ref] can modulate the local electronic structure of the catalysts, facilitate the rate-determining step, reduce the reaction energy barrier, and sometimes become active sites themselves for PEC-driven energy conversion.Recently, Shi and co-workers fabricated O-vacancy-rich BiOI nanosheets for PEC-NRR.They demonstrated that the Ovacancies significantly enhanced the adsorption of N 2 molecules on BiOI nanosheets, evidenced by a temperature programmed desorption experiment. Moreover, they claimed that O-vacancies could effectively inject the photogenerated electrons into the antibonding orbital of the adsorbed N 2 molecule, thus weakening the N≡N bond and promoting N 2 reduction.
Hao and co-workers reported the control of the selectivity of PEC-CRR by adjusting the S-vacancy on the surface of Cu 2 ZnSnS 4 /CdS photocathode by a facile heat treatment in different atmospheres (air and nitrogen).They revealed that after heat treatment in air, the S-vacancies on the Cu 2 ZnSnS 4 /CdS surface were filled by oxygen, leading to better CO 2 and CO adsorption capability with enhanced CRR activity and higher selectivity toward methanol/ethanol [fig_ref] Figure 7: a [/fig_ref]. After heat treatment in N 2 , more generated S-vacancies facilitated the surficial CO desorption process and higher CO selectivity [fig_ref] Figure 7: a [/fig_ref]. www.advancedsciencenews.com www.advancedscience.com Feng and co-workers reported porous cobalt phosphoselenide nanosheets with Se-vacancies for highly efficient PEC water splitting.Their density functional theory (DFT) calculations revealed that the Se-vacancies and P atoms synergistically regulated the electronic structure of the catalysts, thus optimized the ΔG *H and promoting HER kinetics [fig_ref] Figure 7: a [/fig_ref]. As a result, their PEC water-splitting device using the cobalt phosphoselenide nanosheet as both the anode and cathode achieved a hydrogen generation current density of 10 mA cm −2 at a low voltage of 1.64 V, which exceeded the Ir/C-Pt/C reference device.
Recently, engineering dual-/multivacancy defects demonstrated the great potential to enhance the PEC performance of photoelectrodes.Wang et al. developed a multistep heat treatment strategy to synthesize a g-C 3 N 4 /reduced graphene oxide (RGO)/TiO 2 composite with both N-and O-vacancies. They showed that benefiting from the extended visible-light absorption range and the enhanced carrier separation/transfer rate arising from the defects, the composite exhibited a greatly enhanced photocatalytic H 2 evolution rate of 4760 μmol h −1 g −1 , surpassing most reported g-C 3 N 4 /TiO 2 composites.
In addition to anion vacancies, Du and co-workers reported manipulating Cu-vacancies in Cu 2-x S nanoflake arrays to modulate the PEC-HER performance.They showed that the Cu 2-x S nanoflake arrays with surface-rich Cu-vacancies exhibited a considerably enhanced PEC-HER performance [fig_ref] Figure 7: a [/fig_ref]. They claimed that such superior performance originates from the Cuvacancies optimized surface catalytic property, which facilitated charge separation and resulted in surface plasmon resonance effect as well.
Besides point defects, such as vacancies described above, planar defectswere employed to regulate the PEC performance of catalysts. Gong and co-workers reported grainboundary-rich gold-catalyst-decorated amorphous silicon photocathode to produce syngas with adjustable CO/H 2 ratios [fig_ref] Figure 7: a [/fig_ref].They demonstrated that CO selectivity was strongly associated with the grain-boundary density of the Au catalysts. They showed that the grain boundary induced lattice strain at the surface, which effectively stabilized the *COOH intermediate, thus enhancing the CRR activity. As a result, the designed system achieved a half-cell efficiency of 0.42% for CRR under 1 sun illumination. Moreover, Maschmeyer and co-workers synthesized high-index faceted rutile TiO 2 thin films with twin grain boundaries for efficient PEC water splitting.They demonstrated that the twin boundaries improved the interaction at the TiO 2 /electrolyte interface, thus enhancing the catalytic activity of the PEC device.
Despite substantial progress in enhancing PEC performance through introducing defects, there are still many unsolved issues.
For example, precise control of the concentration and distribution of defects is highly required,because the effects of defect concentrations, distribution, and interaction between different types of defects on catalyst activity are different. Besides, in photoelectrocatalysts, the presence of large quantities of defects usually sacrifices the efficiency in separation and transport of photogenerated electrons/holes, so the defects should be located at the outermost surface of photoelectrocatalysts to just facilitate the catalytic reactions. In synthesis, the post-treatmentor heat treatmentof presynthesized catalysts is more conducive to the distribution of defects on the outermost surface of the catalysts. More importantly, compared with point defects, planar defects can provide more active sites in catalysts.However, planar defects are not stable in small-sized catalysts, so it is still a great challenge to construct well-controlled planar defects in catalysts.
## Heteroatom doping
Heteroatom doping can significantly change the physical and chemical properties of materials,such as 1 ppm boron (B) doping can alter the resistivity of silicon by five orders of magnitude. In electrocatalysis, heteroatom doped carbon-based metal-free materials are widely used for oxygen reduction reaction (ORR),OER,HER,and NRR.Recently, heteroatom doping [fig_ref] Figure 8: a [/fig_ref] has shown great potential in PEC-driven catalytic reactions.For example, Xiong and co-workers incorporated B in Bi catalysts for enhanced PEC-NRR.Note that the first hydrogenation step of N 2 to form *NNH is the rate-determining step for NRR. Their theoretical calculation revealed that, in the case of Bdoped Bi catalyst, the energy barrier of *NNH formation was reduced to 2.0 eV, much lower than the pristine catalyst [fig_ref] Figure 8: a [/fig_ref]. Meanwhile, the B dopants could activate nearby Bi atoms as active sites for NRR and donate electrons to Bi, bringing more electrons into the antibonding orbital of *NNH [fig_ref] Figure 8: a [/fig_ref]. As a result, their PEC system achieved an ammonia yield rate of 29.2 mg NH 3 g cat. Liu and co-workers reported a P-doped MoS 2 as a HER cocatalyst for p-Si electrode.Their extended X-ray absorption fine structure spectra and DFT calculations indicated that the doped P atoms created more exposed edges and S-vacancies. In particular, S-vacancies activated the catalytic activity of the inert basal plane. Moreover, Bemana and Nadimi successfully doped S atoms into the hematite matrix,and found that the incorporated S atoms were more probable in the form of cationic substitution (S 4+ ) rather than anionic substitution (S 2− ). This resulted in more Ovacancies and Fe sites with adjustable valence state, which improved the mobility of the photoinduced charge carrier, thus favoring the PEC performance.
Generally, heterodopants are incorporated into the host PEC materials during the growth of the host materials in traditional chemical approachesor post-treatment of the presynthesized host materials.However, heteroatoms are often repelled out of the host material due to the "self-purification" effect,which results in poor dopant stoichiometric control. Similar to the defect engineering discussed above, precise control of the concentration and distribution of heterogeneous dopants on the surface of electrocatalysts still remains challenging.
## Facet engineering
Due to the different adsorption/desorption characteristics of reactants and reaction intermediates on the various exposed facets of catalysts, facet-dependent catalysis has been widely observed in thermocatalysis, photocatalysis, and electrocatalysis.Undoubtedly, controlling the exposure of crystal facets [fig_ref] Figure 9: a [/fig_ref] on the PEC material is one of the most effective ways to improve the PEC activity.Water interaction with the photoelectrode surface is crucial for PEC water splitting. H 2 O adsorption on TiO 2 surfaces has been thoroughly investigated.Theoretical and experimental studies have revealed that unsaturated Ti and O atoms on the surface of TiO 2 adsorb water molecules by forming Ti-O andCopyright 2021, Wiley-VCH. d) DFT-calculated HER free-energy diagrams.Copyright 2021, Wiley-VCH. e) Linear sweep voltammetry curves of different Cu 1.94 S nanoflake arrays under the chopped full-spectrum illumination. Reproduced with permission.Copyright 2017, Wiley-VCH. f) CO to H 2 ratios at different potentials on various amorphous-Si/TiO 2 /Au electrodes. Reproduced with permission.Copyright 2019, The Royal Society of Chemistry.Copyright 2020, Elsevier B.V. e) Linear sweep voltammetry curves of PEC-HER photocathodes. Reproduced with permission.Copyright 2020, American Chemical Society. O-H bonds, respectively. Apparently, surface facets with varied densities of unsaturated Ti and O atoms exhibit significantly different water adsorption and dissociation characteristics. Besides TiO 2 , WO 3 is also widely used as photoelectrocatalyst for PEC water splitting.Wang and co-workers fabricated a WO 3 film with preferentially exposed {002} facets.They showed that the PEC performance of this film was much better than that of WO 3 nanoplate arrays exposing {200} and {002} facets. In particular, WO 3 film with the dominant {002} facet exhibited a high photocurrent density of 3.7 mA cm −2 at 1.23 V versus RHE, ≈93% of the theoretical photocurrent of WO 3 . Their theoretical calculations revealed that the overall energy barrier to achieve water oxidation reaction was lower on the {002} facet than that on the {200} facet. Moreover, the excellent PEC performance of the WO 3 {002} facet was observed by the work of Gong and coworkers.In addition to PEC water splitting, Buonsanti and co-workers reported the facet-dependent activity and selectivity of differently shaped Cu nanoparticles in a gas-fed flow cell configuration [fig_ref] Figure 9: a [/fig_ref].Their experimental results showed that Cu cubes with exposed {100} facets increased ethylene selectivity up to 57%, and Cu octahedra with exposed {111} facets achieved a methane selectivity up to 51% [fig_ref] Figure 9: a [/fig_ref]. In addition, Gao et al. prepared Co 3 O 4 hexagonal platelets with mainly exposed {112} crystal facets for CRR.The combination of experiment and calculation disclosed that the exposed {112} facets of Co 3 O 4 hexagonal platelets were essential for capturing and activating CO 2 molecules [fig_ref] Figure 9: a [/fig_ref]. Specifically, the proton transfer step www.advancedsciencenews.com www.advancedscience.com [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref]. a) Schematic illustration of lattice strain. b) Contour plots of the strain components on a CoO catalyst. c) HER volcano plot. Reproduced with permission.Copyright 2017, Springer Nature. d) Schematic of the lattice change of LiCoO 2 . Reproduced with permission.Copyright 2016, American Association for the Advancement of Science. e) NRR free-energy diagrams of graphene. Reproduced with permission.Copyright 2021, Elsevier B.V. f) Schematic diagram of Co 2+ electronic configurations in high-and low-spin states. Reproduced with permission.Copyright 2020, Wiley-VCH.
was greatly facilitated on the {112} facets of Co 3 O 4 , which is generally considered as the rate-determining step for the overall CRR.
The above work shows that controlling the exposed facets of photoelectrocatalysts is an effective way to improve the performance. At present, the exposed facets of noble metal nanocatalysts can be realized by controlling the crystal morphology, such as icosahedron,dodecahedra,decahedron,tetrahedron,cube,etc. However, the accurate regulation of the exposed facets on most non-noble metal catalysts is still a big challenge. Moreover, the more active facets are also prone to change in the PEC catalytic reactions. Therefore, special attention should be paid to the stability of the exposed crystal facets in working conditions and the influence of crystal facets on catalytic activity.
## Strain engineering
Strain engineering [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref] , by expanding or compressing atoms in a crystal lattice, has recently proven to be a powerful strategy altering the electronic structure and reactivity of materials.For example, the exceptionally active Pt 3 Ni ORR catalysts were obtained by adopting Ni atoms in the underlying atomic layers to compress the Pt surface layer, resulting in a significant d-band center downward shift of the Pt surface, thus leading to optimum adsorption of oxygenated intermediate species and ultrahigh ORR activities.We have reported boosting the HER performance of CoO nanorods by engineering surface strain.Notably, transition metal oxides are generally known to be inactive toward HER due to their inappropriate ΔG *H (too strong H* adsorption on O, but too weak on metal ion). Our experimental and theoretical work showed that the tensile strain on the surface of CoO nanorods significantly reduced the O-vacancy formation energy, thus promoting the formation of massive O-vacancies as the active sites for water dissociation [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref]. Moreover, we found that the tensile strain can tune the electronic structure of CoO nanorods to achieve the optimum ΔG *H . Moreover, we emphasized that the intrinsic activity of the tensile strain engineered CoO nanorods was located at the top of the HER volcano plot [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref].
Cui and co-workers used an electrode material in lithium-ion battery, LiCoO 2 , as the medium to control the lattice strain of the loaded Pt catalyst.They showed that the volume and lattice spacing of layered LiCoO 2 could be changed by the chemical extraction/intercalation of Li ions in the electrochemical charge/discharge process [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref]. This resulted in the compressive or tensile strain in Pt catalyst, which could effectively regulate the adsorption binding energy of intermediates, and thus the catalytic ORR activity of Pt catalysts. They observed a 90% enhancement in Pt ORR activity under compressive strain.
Mao and co-workers investigated the effect of strain on the catalytic NRR performance of graphene by DFT calculations.They found that the increase in strain on graphene led to a decrease in the energy barrier of NRR [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref] , and the graphene with a strain magnitude of 8% exhibited an ultralow energy barrier of 0.35 eV and an overpotential of 0.19 V.Copyright 2017, Springer Nature. c) HER free energy diagram on the Pt(111) surface and single-atom Pt anchored on Ni surface, referred to as Pt 1 /N-Ni. Reproduced with permission.Copyright 2021, The Royal Society of Chemistry. d) PEC-NRR performance of cobalt single-atom catalyst on CoPi/Ti-Fe 2 O 3 photoanode. Reproduced with permission.Copyright 2021, Elsevier B.V.
In addition, the enlargement or contraction of the crystal lattice can affect the spin state of magnetic catalysts, which plays a key role in determining their catalytic activity.Our computational results demonstrated that the contraction of Co-O octahedral in Co 2 VO 4 strengthened the crystal field and increased the energy gap between t 2g and e g , thus causing the spin transition of Co ions from the high-spin state (t 2g 5 e g 2 ) to the low-spin state (t 2g 6 e g 1 ). This is highly advantageous for ORR and OER [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref].Although the works mentioned above are mainly in the field of electrocatalysis, we believe that strain engineering may provide a new approach for the development of high-performance PEC electrodes. We note that lattice strain, including compression or tension, is usually generated through lattice mismatch between materials, which requires the construction of core-shell structures or the introduction of substrate materials.This undoubtedly poses a challenge to the design of PEC electrodes, because the PEC electrode also undertakes the functions of light absorption, charge separation, and transfer at the same time. One possible solution to introduce lattice strain only on the surface of PEC electrode is through facet engineering or incorporating defects or dopants,which will not affect the light absorption, charge separation, and transfer in the bulk electrode.
## Single-atom engineering
Single-atom catalysts [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref] have become one of the frontiers and hot topics in the fields of materials, chemistry, and physics.Recently, single-atom catalysts have shown promising applications in PEC devices.Cui et al. developed a Fe 2 O 3 -NiO x -Ir single-atom electrode for PEC-OER,in which NiO x acted as the "movable bridge" connecting Fe 2 O 3 with Ir single atom, capturing the photogenerated holes from Fe 2 O 3 and transferring them to Ir single atom; while Ir single atom served as the super active sites for OER [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref]. As a result, they observed an unusual OER activity with a turnover frequency of 2.4-12.7 s −1 per Ir site at 1.23 V versus RHE and a high stability of the hybrid catalyst (>80 h) in alkaline electrolyte.
Very recently, Zhang and co-workers dispersed Pt single atoms on the nitrogen modified Ni substrate (Pt 1 /N-Ni) for PEC-HER.They demonstrated that the lone pair electrons in nitrogen atoms were prone to coordinate with the unoccupied d orbital of Pt atoms, stabilizing Pt single atoms. This was supported by the clustering energy calculation [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref]. They further integrated the Pt 1 /N-Ni catalyst into the Cu 2 O nanowire www.advancedsciencenews.com www.advancedscience.com photocathodes for PEC-HER. The hybrid electrode achieved an extremely high photocurrent density of 11.9 mA cm −2 at 0 V versus RHE and a high photo-to-H 2 conversion efficiency of 1.75%.
Zhao and co-workers reported an efficient PEC-NRR system by integrating the Fe 2 O 3 -based photoanode for water oxidation and the Co single-atom-based photocathode for NRR.This system afforded an NH 3 generation rate of 1021.5 μg mg Co −1 h −1 and a Faradaic efficiency of 11.9% at 1.2 V versus RHE [fig_ref] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2... [/fig_ref].
In the above works, the deposited single atoms on the photoelectrodes act as the active sites for PEC reactions. Apparently, the low loading mass of single atoms on the photoelectrodes greatly limits the photo-to-energy (fuel/chemical) conversion of the PEC system. Therefore, it is highly required to develop a synthetic approach for high loading of single atoms. Very recently, a graphene-quantumdot-mediated method has promoted the single-atom loading up to 40 wt%,which provides a new direction in this field. On the other hand, single-atom-modified photoelectrodes are also expected to find wide applications for reactions beyond HER and OER, such as NRR and CRR.
## Conclusions and future perspectives
This review summarizes the recent advances in the atomic design of photoelectrocatalysts toward HER, NRR, and CRR from the perspective of surface catalytic reactions. We introduce basic catalytic mechanisms and show the commonly used strategies for the design of efficient photoelectrocatalysts at the atom scale, including engineering of defect, dopant, facet, strain, and single atom. Although some achievements have been made, surface catalytic reactions have not been paid adequate attention to PEC systems, and most research works are focusing on improving light absorption and charge separation/transfer. Further research efforts should be devoted to filling this gap in order to improve the performance of PEC. The ultimate goal is to understand the PEC reactions at the molecular level and design PEC materials atom-by-atom to boost the catalytic activity and selectivity. Future research direction is suggested but is not limited to: 1) At present, the understanding of catalytic mechanisms of HER, NRR, and CRR reactions in the PEC system is mainly inherited from the corresponding electrocatalytic reactions. But compared with electrocatalytic reactions, PEC catalytic reactions are distinct in some aspects. For example, a large amount of hot electrons are generated under light illumination, which can significantly change the catalytic reaction mechanisms. More experimental and theoretical efforts should be devoted to revealing the "real" PEC catalytic reaction mechanisms. 2) Recently, rapid progress in insitu surface-enhanced spectroscopies, such as infrared absorption,Raman,and advanced theoretical calculationsopens up new exciting opportunities for monitoring the reactive intermediates with short lifetime and low coverage under operando conditions, which have achieved great success in thermocatalysisand electrocatalysis.Extending these advanced technologies and successful research experiences to PEC systems will undoubtedly bring new breakthroughs in PEC-driven photo-to-energy conversions. The main challenge now is to design optical paths to eliminate the interference of the incident light on the Raman or infrared signals in these cutting-edge in situ spectroscopies. 3) PEC-NRR and PEC-CRR are still in their infancy, whose activity and selectivity are far from satisfactory for industrial application. In PEC-NRR and PEC-CRR, PEC-HER is a competing reaction with favorable thermodynamics and kinetics, which largely limits the production rate and selectivity of the desired products in PEC-NRR and PEC-CRR. Recently, taking advantage of the unique localized surface plasmon resonance,photothermal effect in the PEC systemto activate the reactants and intermediates has been proved a highly promising way to decrease the reaction barrier, facilitate the rate-determining step, and thereby promote the overall catalytic activity and selectivity of PEC-NRR and PEC-CRR. Therefore, more attentions should be paid to these unique light-induced effects. 4) Exploring a new PEC catalyst paradigm with an efficient solar energy utilization in the untapped near-infrared range can promote the development of full-spectrum-driven PEC systems. In this field, metal-assisted or self-doped plasmonic semiconductors have shown great potential due to their strong absorption in the near-infrared range.
## Conflict of interest
The authors declare no conflict of interest.
[fig] Figure 1: Schematic illustration of PEC energy conversion for the production of H 2 , NH 3 and the reduction of CO 2 to valuable chemicals. [/fig]
[fig] Figure 2: Schematic diagram of four processes of photoinduced charge generation, separation, transfer, and final catalytic reactions in PEC devices. [/fig]
[fig] Figure 3: a) Schematic diagram of PEC cell for water splitting under alkaline conditions. b) HER mechanism. c) OER mechanism. [/fig]
[fig] Figure 4: The possible reaction mechanisms of PEC-NRR including a) dissociative and b) associative pathways. [/fig]
[fig] Figure 5: The possible reaction pathways of different CO 2 reduction products: a) C 1 pathway; b) C 2 pathway. [/fig]
[fig] Figure 6: Schematic diagram of enhancing PEC-driven solar-tofuel/chemical conversions by the atomic design of photoelectrocatalysts. [/fig]
[fig] − 1 h: −1 and Faradaic efficiency of 8.3% at a bias of 0.48 V versus reversible hydrogen electrode (RHE) (Figure 8d). Shen and co-workers reported the activation of the MoS 2 basal plane for PEC-HER by in situ O-doping. [143] Based on combined experimental results and theoretical calculations, they demonstrated that the O-Mo-S sites on the MoS 2 basal planes enhanced the intrinsic conductivity of MoS 2 by strengthening the hybridization between Mo-d orbital and S-p orbital. Moreover, the O-doping effectively modulated ΔG *H on the MoS 2 basal plane. This significantly promoted the PEC-HER activity of the MoS 2 basal plane (Figure 8e). [/fig]
[fig] Figure 7: a) Schematic diagram of surface vacancy defects. b) DFT-optimized structures of CO 2 and CO adsorbed on CdS and O-doped CdS with Svacancy. c) Yield rates of CO 2 reduction products under illumination. Reproduced with permission. [/fig]
[fig] Figure 8: a) Schematic diagram of heteroatom doping. b) DFT-calculated NRR free-energy diagrams on pristine and B-doped Bi catalysts. c) Projected density of states (pDOS) and projected crystal orbital Hamilton populations (pCOHP) for the N atoms of *NNH directly bonded to Bi atom on optimum B-doped Bi catalysts. d) NH 3 yield rate and Faradaic efficiency of B-doped Bi nanorolls. Reproduced with permission. [/fig]
[fig] Figure 9: a) Schematic diagram of faceted catalyst surface. b) Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images of various faceted Cu nanocrystals. c) Faradaic efficiencies of Cu catalysts with spherical, cubic, and octahedral morphologies in a gas-fed flow cell in 1 m KOH. Reproduced with permission.[163] Copyright 2020, American Chemical Society. d) DFT-optimized structures and free energy diagram of CO 2 and reduction intermediates on Co 3 O 4 {111} and Co 3 O 4 {112} surfaces. Reproduced with permission.[164] Copyright 2016, WILEY-VCH. [/fig]
|
An Atypical Presentation of Chronic Atrophic Gastritis: Hemolytic Anemia and Mesenteric Panniculitis
Microangiopathic hemolytic anemia (MAHA) requires an aggressive approach since primary thrombotic microangiopathy syndromes such as thrombotic thrombocytopenic purpura (TTP) can progress rapidly to a fatal outcome. Differential diagnosis can be challenging even for an experienced hematologist. We present a case of a 52-year-old male who presented with symptoms of mesenteric panniculitis and showed signs of MAHA. His condition was attributed to severe vitamin B12 deficiency secondary to chronic atrophic gastritis and initiation of appropriate therapy was met with complete resolution of symptoms and normalization of hematologic parameters.
# Introduction
Thrombotic microangiopathies are a diverse group of disorders that may be congenital or acquired, present in childhood or in adults, and manifest an acute or a more gradual progressive course [bib_ref] Syndromes of thrombotic microangiopathy, George [/bib_ref]. Clinically they are characterized by thrombocytopenia, microangiopathic hemolytic anemia (MAHA), and extremely elevated serum lactate dehydrogenase (LDH) levels with or without signs of end-organ damage [bib_ref] Syndromes of thrombotic microangiopathy, George [/bib_ref] [bib_ref] Thrombotic microangiopathies, Moake [/bib_ref]. Differential diagnosis can be challenging as primary thrombotic microangiopathy syndromes such as thrombotic thrombocytopenic purpura (TTP) can progress rapidly to a fatal outcome without prompt initiation of effective therapy [bib_ref] Thrombotic thrombocytopenic purpura, George [/bib_ref]. In this article, we discuss a case of severe vitamin B12 deficiency presenting as concurrent microangiopathic hemolytic anemia and mesenteric panniculitis.
## Case report
A 52-year-old Hispanic male with past medical history significant for essential hypertension, gastroesophageal reflux disease (GERD), and recently diagnosed anemia attributed to iron deficiency presented to the Emergency Department with three weeks' history of left flank pain. Pain was described as severe, sharp, and constant in nature, made worse with prolonged fasting or following a meal. No recognizable alleviating factors were reported. Associated symptoms included nausea, vomiting of nonbloody, nonbilious secretions, and retrosternal burning. There was no hematemesis, melena, perrectal bleeding, or change in bowel habits. Patient also reported progressive weakness and fatigue for two weeks and weight loss of approximately ten pounds over the last two months. Patient was prescribed omeprazole and iron supplements for newly diagnosed anemia and abdominal pain by his primary care physician. He denied taking other prescription or over-the-counter medications, herbal supplements, or illicit drugs. Patient denied alcohol or tobacco use. There was no personal or family history of solid or hematologic malignancies or bleeding or thrombotic events.
Patient was afebrile on admission with normal hemodynamic parameters. Physical examination revealed conjunctival pallor and scleral icterus. Cardiovascular and pulmonary examination were within normal limits. Abdominal examination revealed mild to moderate left flank and epigastric tenderness with no rebound or involuntary guarding. There was no palpable hepatosplenomegaly or peripheral lymphadenopathy. The rest of the examination was largely unremarkable.
Complete blood count (CBC) on admission revealed hemoglobin 6.3 g/dL, hematocrit 17.9%, white blood cell 2 Case Reports in Hematology Computed tomography (CT) scan of the abdomen and pelvis showed acute mesenteric inflammation involving central small bowel mesentery, compatible with the diagnosis of mesenteric panniculitis (see [fig_ref] Figure 1: CT scan of the abdomen and pelvis showing inflammatory fat stranding in... [/fig_ref].
Peripheral blood smear revealed an abundance of schistocytes and occasional hypersegmented neutrophils (see [fig_ref] Figure 3: Peripheral blood smear showing poikilocytosis, schistocytosis, thrombocytopenia, and a hypersegmented neutrophil [/fig_ref]. This, along with the initial laboratory findings, was concerning for an atypical presentation of thrombotic thrombocytopenic purpura (TTP). At this point, the plan was to start urgent plasmapheresis. However, within the next hour, serum vitamin B12 level was reported at <109 pg/dL (normal range: 213-816 pg/dL) with serum homocysteine level of 86.8 mol/L (normal range: 5.1-15.4 mol/L). Plasmapheresis plans were stopped. Bone marrow smear and biopsy were performed confirming erythroid hyperplasia and megaloblastic anemia (see . Cyanocobalamin repletion was promptly initiated.
Over the next 2 days, hemoglobin remained stable, reticulocyte count increased to 1.17%, and major improvement in the degree of hemolysis was noted. Furthermore, abdominal pain resolved and patient was tolerating regular diet. Gastroenterology team recommended upper endoscopy and considered mesenteric panniculitis reactive to underlying hematologic disorder. Late during the course of this admission, the sent-out laboratory tests showed methylmalonic acid level of 4080 nmol/L (normal range: 87-318 nmol/L), while the serologic tests for Intrinsic Factor Blocking Antibody and Anti-Gastric Parietal Cell Antibody were strongly positive. Together, these findings suggested autoimmune chronic atrophic gastritis as the underlying etiology of current presentation.
Esophagogastroduodenoscopy (EGD) was performed as outpatient showing two large, sessile polyps in the gastric body (see [fig_ref] Figure 7: EGD showing two firm sessile polyps [/fig_ref] that were completely removed. The remainder of the EGD was within normal limits. Gastric polyp histologic examination was positive for well-differentiated neuroendocrine tumor and background changes of chronic atrophic gastritis.
Patient continues to receive cyanocobalamin replacement therapy as outpatient. Vitamin B12 replacement has been met with complete normalization of hematologic parameters and resolution of symptoms of mesenteric panniculitis. This is the first case report of vitamin B12 deficiency secondary to chronic autoimmune gastritis, presenting with concurrent microangiopathic hemolytic anemia and mesenteric panniculitis.
# Discussion
TTP has been classically diagnosed based on a pentad of MAHA, thrombocytopenia, neurologic abnormalities, renal dysfunction, and fever [bib_ref] Thrombotic thrombocytopenic purpura, George [/bib_ref] [bib_ref] Thrombotic thrombocytopenic purpura: Report of 16 cases and review of the literature, Amorosi [/bib_ref]. However, with the introduction of plasmapheresis as the effective treatment modality, Case Reports in Hematology presumptive diagnosis is commonly made on the basis of MAHA and thrombocytopenia without an apparent cause [bib_ref] Thrombotic thrombocytopenic purpura, George [/bib_ref] [bib_ref] Comparison of plasma exchange with plasma infusion in the treatment of thrombotic..., Ga [/bib_ref]. As a result, up to 10% of patients diagnosed with TTP and receiving plasmapheresis may be later found to have an alternate diagnosis [bib_ref] The Oklahoma Thrombotic Thrombocytopenic Purpura-Hemolytic Uremic Syndrome (TTP-HUS) Registry: A Community Perspective..., George [/bib_ref] , underlining the importance of continuous reevaluation.
The most common symptoms of TTP at presentation are generalized weakness, abdominal pain, nausea, and vomiting [bib_ref] Thrombotic thrombocytopenic purpura, George [/bib_ref]. With suggestive clinical symptoms and characteristic laboratory findings, it is not surprising that our patient was presumptively diagnosed with TTP and planned for initiation of plasmapheresis. However, the possibility of vitamin B12 deficiency was also entertained in view of relatively mild thrombocytopenia, extremely elevated LDH level, and presence of hypersegmented neutrophils on blood smear.
Vitamin B12 deficiency causes dyssynchrony between nuclear and cytoplasmic maturation, resulting in ineffective erythropoiesis, intramedullary hemolysis, and characteristic features of megaloblastic anemia on blood smear: macrocytosis and hypersegmented neutrophils [bib_ref] Vitamin B 12 deficiency, Stabler [/bib_ref]. Other less common hematologic findings associated with vitamin B12 deficiency include thrombocytopenia, pancytopenia, and hemolytic anemia [bib_ref] Current hematological findings in cobalamin deficiency. A study of 201 consecutive patients..., Andrès [/bib_ref]. Furthermore, vitamin B12 deficiency has been implicated as a cause of thrombotic microangiopathy mimicking TTP [bib_ref] Pseudothrombotic thrombocytopenic purpura: a rare presentation of pernicious anemia, Tadakamalla [/bib_ref] [bib_ref] Therapeutic dilemma in the management of a patient with the clinical picture..., Walter [/bib_ref] [bib_ref] To treat or not to treat: a rare case of pseudo-thrombotic thrombocytopenic..., Malla [/bib_ref] [bib_ref] When the picture is fragmented: Vitamin B12 deficiency masquerading as thrombotic thrombocytopenic..., Panchabhai [/bib_ref] [bib_ref] Hemolysis and schistocytosis in the emergency department: consider pseudothrombotic microangiopathy related to..., Noël [/bib_ref]. The mechanism of thrombotic microangiopathy and resultant MAHA is not fully understood, but the presence of hyperhomocysteinemia has been suggested as the plausible link between vitamin B12 deficiency and endothelial dysfunction and microvascular thrombi formation [bib_ref] Hemolysis and hyperhomocysteinemia caused by cobalamin deficiency: Three case reports and review..., Acharya [/bib_ref] [bib_ref] High frequency of vitamin B12 deficiency in asymptomatic individuals homozygous to MTHFR..., Zittan [/bib_ref].
Mesenteric panniculitis is a relatively uncommon benign condition characterized by varying degrees of inflammation, fat necrosis, and fibrosis of the small bowel mesentery [bib_ref] Mesenteric panniculitis: various presentations and treatment regimens, Issa [/bib_ref]. It may occur spontaneously or in association with other disorders, including abdominal trauma, hematologic and solid organ malignancies, and autoimmune conditions [bib_ref] Mesenteric panniculitis: various presentations and treatment regimens, Issa [/bib_ref] [bib_ref] CT evaluation of mesenteric panniculitis: prevalence and associated diseases, Daskalogiannaki [/bib_ref]. The majority of cases are diagnosed incidentally during the work-up of an unrelated condition [bib_ref] CT evaluation of mesenteric panniculitis: prevalence and associated diseases, Daskalogiannaki [/bib_ref]. The most common symptoms in clinically apparent cases are abdominal pain, nausea and vomiting, change in bowel habits, and weight loss [bib_ref] Sclerosing Mesenteritis: Diverse clinical presentations and dissimilar treatment options. A case series..., Vlachos [/bib_ref]. Confirmation of diagnosis generally requires histologic proof of mesenteric panniculitis; however, diagnosis is often suggested by typical radiologic findings [bib_ref] Mesenteric panniculitis: various presentations and treatment regimens, Issa [/bib_ref] [bib_ref] CT evaluation of mesenteric panniculitis: prevalence and associated diseases, Daskalogiannaki [/bib_ref]. Treatment options are varied, nonstandardized, and typically reserved for symptomatic patients [bib_ref] Mesenteric panniculitis: various presentations and treatment regimens, Issa [/bib_ref] [bib_ref] Sclerosing Mesenteritis: Diverse clinical presentations and dissimilar treatment options. A case series..., Vlachos [/bib_ref]. A myriad of antiinflammatory and hormonal therapies, including systemic corticosteroids and tamoxifen, alone or in combination, have been used with variable success. Most cases tend to resolve spontaneously with no intervention [bib_ref] Mesenteric panniculitis: various presentations and treatment regimens, Issa [/bib_ref] [bib_ref] CT evaluation of mesenteric panniculitis: prevalence and associated diseases, Daskalogiannaki [/bib_ref] [bib_ref] Sclerosing Mesenteritis: Diverse clinical presentations and dissimilar treatment options. A case series..., Vlachos [/bib_ref].
Mesenteric panniculitis has been reported in association with vitamin B12 deficiency and chronic atrophic gastritis [bib_ref] A clinical case of mesenteric panniculitis associated with chronic gastritis and biliary..., Ribeiro [/bib_ref]. The patient was treated with prolonged course of systemic corticosteroids with complete symptomatic and radiologic remission. We did not consider treatment of mesenteric panniculitis in our patient as the disease process was assumed to be secondary to concurrent hematologic abnormalities. To our knowledge, there have been no case reports of concurrent mesenteric panniculitis and thrombotic microangiopathy secondary to vitamin B12 deficiency.
Gastric carcinoid tumors, also called gastroenteropancreatic neuroendocrine tumors, comprise approximately 10% to 30% of all carcinoid tumors [bib_ref] Gastric carcinoid tumors, Wardlaw [/bib_ref]. There are four different types of gastric carcinoid tumors with type 1 comprising the majority of cases [bib_ref] Gastric carcinoid tumors, Wardlaw [/bib_ref] [bib_ref] Gastric carcinoids: Biologic behavior and prognosis after differentiated treatment in relation to..., Borch [/bib_ref]. Type 1 carcinoid tumors are found in association with chronic atrophic gastritis and achlorhydria and are usually multiple, nodular, or polypoid in nature and less than 1 cm in size [bib_ref] Gastric carcinoid tumors, Wardlaw [/bib_ref]. Endoscopic resection is the treatment of choice for type 1 gastric carcinoid tumors and they generally follow a benign course with >95% 5-year survival rate [bib_ref] Gastric carcinoid tumors, Wardlaw [/bib_ref] [bib_ref] Gastric carcinoids: Biologic behavior and prognosis after differentiated treatment in relation to..., Borch [/bib_ref].
# Conclusion
Thrombotic microangiopathies are a diverse group of disorders that share microangiopathic hemolytic anemia and thrombocytopenia in common. Vitamin B12 deficiency may present in a similar fashion and pose a diagnostic and therapeutic challenge to the clinicians. We propose exploring further work-up for exclusion of vitamin B12 deficiency as the cause of thrombotic microangiopathy, especially for cases with extremely elevated LDH levels (>2500 IU/L), low reticulocyte counts, mild to moderate thrombocytopenia (>50,000/ L), and/or atypical findings on blood smear such as hypersegmented neutrophils.
[fig] Figure 1: CT scan of the abdomen and pelvis showing inflammatory fat stranding in the small bowel mesentery (axial view). [/fig]
[fig] Figure 2: Redemonstration of inflammatory fat stranding in the small bowel mesentery (coronal view). (WBC) count 4,100/ L, platelet count 88,000/ L, and reticulocyte count 0.86%. Mean corpuscular volume (MCV) was 96.2 fL and red blood cell distribution width (RDW) was 23.4%. Basic metabolic panel (BMP) showed normal serum electrolyte concentrations and normal renal function with creatinine 0.8 mg/dL and blood urea nitrogen (BUN) 14 mg/dL. Liver function tests (LFTs) on admission revealed elevated total bilirubin concentration of 4.5 mg/dL with direct fraction of 0.5 mg/dL and mildly elevated alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels of 67 IU/L and 165 IU/L, respectively. Coagulation profile showed international normalized ratio (INR) of 1.2 and partial thromboplastin time (PTT) of 25.4 seconds. Direct antiglobulin test and antibody screen were negative. Lactate dehydrogenase (LDH) was 6,235 IU/L while haptoglobin was <8.0 mg/dL. [/fig]
[fig] Figure 3: Peripheral blood smear showing poikilocytosis, schistocytosis, thrombocytopenia, and a hypersegmented neutrophil (arrow). [/fig]
[fig] 3, Figure 4: Magnified view of hypersegmented neutrophils. [/fig]
[fig] Figure 5: Bone marrow smear showing multinucleated red cell (white arrow), typical megaloblast (arrowhead), and mitotic phase cell (yellow arrow). [/fig]
[fig] Figure 6: Bone marrow smear showing erythroid hyperplasia with dysplastic features and megaloblastic changes. [/fig]
[fig] Figure 7: EGD showing two firm sessile polyps. [/fig]
|
Comparison of the Values of Solar Cell Contact Resistivity Measured with the Transmission Line Method (TLM) and the Potential Difference (PD)
## Tlm and pd methods-presentation of statistical data based on electronic databases
Information on the use of TLM transmission lines in electronics is collected in the IEEE Explore database by IEEE (Institute of Electrical and Electronics Engineers) and IET (Institution of Engineering and Technology). The method was originally proposed by Shockley in 1964and later modified by Berger. Based on the data contained in the database, it can be concluded that in the period from 1990 to 2020, 17 related articles were published in journals from the Philadelphia list, including the Journal of Photovoltaics, Electron Device Letters, Transactions on Electron Devices, and the Journal of Display Technology.
Information on the use of the transmission line method was also analysed based on data contained in the Scopus database using the SciaVal tool. Based on studies from 100 countries from 2015 to 2020, a ranking of the scientific achievements of 35 countries was prepared in which Poland placed 14th. In the case of Poland, only the author of this publication published three works in the period of 2015-2020, so this is an interesting issue and a subject that requires further investigation.
Articles covering the application of the transmission line method in various fields of science from 2015 to 2020 were also analysed from the Scopus database, following the ASJC (All Science Journal Classification) classification of thematic areas. Based on this analysis, it was found that the most significant applicability of the TLM method was observed in the field of science-materials engineering (32%). In contrast, the value did not exceed 5% in energy, chemistry, or other fields. In the field of photovoltaics, from 1989 to 2020 the most According to the Web of Science database, the topic of TLM covers the literature from 1973 to 2021, for a total of 12,996 publications. After limiting the search to the electronics category, we obtained 7722 scientific papers.
From 1983 to 2020, the IEEE Explore database in the field of electronics and electrics collected 34 publications; we searched these using the term "contact resistance and PD method". However, from 1978 to 2020, 30 publications were recorded, of which four were in the field of photovoltaics. These were found using the keyword "contact resistance scanning,". According to the search "contact resistance scanning" in the Scopus database, there were eight publications from 1979 to 2021, while in the Web of Science database, there were seven publications from 2005 to 2020.
Certain discrepancies in the scientific circulation of published research results in the form of scientific journals within the aforementioned bibliographic databases may result from the emergence of an increasing number of scientific journals which themselves fulfil the role of a sieve, separating the valuable scientific publications while omitting those of poor quality or that are non-scientific .
This work aims to test the value of contact resistivity of the front metallisation of solar cells. These cells were made only for measurement purposes. This work uses two methods, TLM and PD, to determine the same parameters in different ways and with different measuring stands, depending on the needs and expectations of the person performing the measurement.
The correct performance of this technological process requires the precise selection of parameters. The production of metallisation requires the correct execution of the front electrode of the correct size and shape and proper cavity in the semiconductor material. In the area where the front electrode connects to the semiconductor, junctions are formed, introducing additional resistances to the electrical circuit that limit the photocurrent flow. The use of electrode pastes minimises these losses and the level of emitter doping can be determined. There are also leakages of separated charge carriers in the junction area, causing a photovoltage drop in the solar cell. The power loss of a photovoltaic cell is influenced, among other things, by resistivity, which is a feature of cells that has a fundamental impact on their quality.
The determined parameters of the technological process strongly depend on the composition of the conductive paste or the metallic powder from which the paste is made. This is because the conductivity of the tested paste depends on the granulation, particle shape, and powder content of the paste. Ceramic glaze (e.g., SiO 2 ) binds the particles of the base material with the silicon substrate; the rest is an organic carrier mixture that makes the paste sticky. The thickness of the applied layer on the front electrode and the morphology of the substrate affect the resistivity value obtained as a result of the measurements.
## Methods applied to measuring selected parameters of the electrical properties of photovoltaic cells
Currently, research and development in electronics and photovoltaics are focused on developing and producing electrical contacts using various techniques and determining the dependence of the resistivity of these contacts between multicomponent metallic components and conductive layers. The goal is to optimise the technology of contact metallisation. The metallic element layer should meet various requirements to ensure the low resistivity in the vicinity of the metal-semiconductor junction. Of particular importance in a correctly performed technological process are the proper selection of the material (conductive coating and substrate); its thickness; its coating geometry (shape and size); the conditions of its production; the adhesion of the metallic element to the substrate; and the substrate morphology (e.g., structure and roughness).
The external operating parameters that characterise a silicon solar cell include the cell open-circuit voltage (V oc ), short circuit current (I sc ), fill factor (FF), maximum obtainable power (Pm), and photovoltaic conversion efficiency (E ff ). These values depend, for example, on technological factors (the reflectance factor; cover factor; incomplete absorption factor due to limited material thickness; and the Q i collection factor, taking into account that not all generated charge carriers reach the pn junction, where they can be separated) and are determined by the influence of the manufacturing process on the parameters, the material and physical properties of the individual layers, and the structural elements of a solar cell. The most important of these physical parameters include: the electromagnetic radiation reflection coefficient (R ref ); the thickness of the base material of the cell (D c ); the concentration of charge carriers (ni); the mobility (µ) and lifetime of charge carriers (τ); the length of the carrier diffusion path charge (L); the speed of surface recombination (SRV); the resistance (R) and resistivity (ρ) of areas and structural elements of the cell; and the types and concentration of defects and the resulting density of recombination centres (Nit).
Basic measurements of photovoltaic cells include the current-voltage characteristics (I-V), which determine the physical parameters of the manufactured solar cell. A oneor two-diode model can be matched numerically to the measured I-V characteristics. In the measuring system of the stand used for measuring the I-V characteristics of light and dark cells, four basic elements that determine the quality of the measurement can be distinguished: a light source, measuring system, table, and contact probes.
The I-V characteristics must be measured under the strictly defined conditions of a specific radiation spectrum and temperature. The standard used is the AM1.5 spectrum, which has an intensity of 1000 W/m 2 at a cell temperature of 25 - C. The measured cell is placed on a brass table that acts as a current electrode for the back contact, while gold-plated probes provide the contact to the front electrode with telescopic pressure. The lighting elements are four independently positioned halogen lamps powered by a highly stabilised power supply with a "ramp" voltage increase (the so-called soft start). A reverse-biased reference photodetector controls the stability of the light intensity during the measurement. The measurement of the I-V characteristics of the cell consists of the simultaneous measurement of the voltage biasing the reverse link in the range of ±0.75 V and the measurement of the current, the value of which is calculated for the selected load resistance and the measured voltage drop.
One of the operations used for producing photovoltaic cells is the application of electrical contacts. As numerous studies show, the electrode layer should meet various requirements to ensure the low resistance of the electrode connection zone with the substrate. Of particular importance is the appropriate selection of the material, including the electrode and substrate; the conditions of its production; the shape and size of the electrode and its adhesion to the substrate; and the substrate morphology. The resistivity is understood to be the quantity that characterises the metal-semiconductor junction, considering the area above and below the junction. The value of contact resistance, which depends on the type of paste used, the substrate resistance, and the temperature of the metallisation process, can be determined experimentally-for example, by using the TLM (transmission line model) method or the PD (potential differences) method . In the case of the TLM method, the measurement consists of forcing an electric current signal between the selected pair of adjacent front conductive lines on the tested sample through the supply soda and the spontaneous generation of a potential difference in them through the measuring probes. In the PD method, local lighting produces a current and the voltage is measured with a metal probe placed on the front metallisation of the test sample. During the measurement, the sample is short-circuited with an external receiver.shows the procedure for determining the selected electrical parameters (including resistivity) of semiconductor structure contacts using the aforementioned measurement methods. In order to standardise the results obtained from the electrical properties tests, this article adopts one determination of contact resistance (R) and resistivity (ρ) for two methods.shows the method for measuring using two measurement techniques.shows the procedure for determining the selected electrical parameters (including resistivity) of semiconductor structure contacts using the aforementioned measurement methods. In order to standardise the results obtained from the electrical properties tests, this article adopts one determination of contact resistance (R) and resistivity (ρ) for two methods.shows the procedure for determining the selected electrical parameters (including resistivity) of semiconductor structure contacts using the aforementioned measurement methods. In order to standardise the results obtained from the electrical properties tests, this article adopts one determination of contact resistance (R) and resistivity (ρ) for two methods.
[formula] A1. R p = U I · K, Ω/ [17,20-25] B1 R cl = C · V I = C · V d·J , Ω·cm I = I L = J · d, A/cm [26-34] A2. R T = U I , Ω A3. R T = 2R c + d·R p k , Ω R c = C · V I = C · V·w J·d , Ω·cm 2 I = I (L·w) = J·d w , A/cm 2 A4. L T = ρ c /R p 0.5 , cm R c = √ ρ c ·R p k coth (L · L T ), Ω A5. If, L T ≥ 2 L, to ρ c = R c · k · L T Ω·cm 2 If, L T < 2 L, to ρ c = R c · k · L Ω·cm 2 [/formula]
Where C-correction factor (·1.8) for the current leakage out of the spot and shading by the probe.
Measurements of the resistivity of the front contacts of the semiconductor structures were carried out in two research centres and designated as the research centre A_TLM method and the research centre B_PD method.presents the differences between the methods used and the research equipment used for this purpose. In the case of the TLM method, the measurement is performed manually but is repeatable. The mere collection and processing of the measurements were time-consuming. The user has to develop a form for the graphical presentation of the received data.
In the PD method, the measurement is performed automatically, and the user decides how long it is to last. This stand enables the immediate recording of the measurement results in 2D and 3D formats with the calculated data and their direct printout. However, it is impossible to perform the measurement again because the tested sample is very damaged. The patterns of the applied front metallisation, sample size, cost and size of the test stand, number of patterns used to determine the parameters sought, and the symbols of these parametersare the main differences between these methods. The symbols of the same parameters may differ because they often depends on the manufacturers bringing something new to the market. In the case of Mechatronics, this was a stand equipped with a Corescan device.
# Methodology
The test apparatus with the applied method was used to test the selected electrical parameters (i.e., resistance and resistivity): (1) TLMand (2) it is impossible to perform the measurement again because the tested sample is very damaged. The patterns of the applied front metallisation, sample size, cost and size of the test stand, number of patterns used to determine the parameters sought, and the symbols of these parametersare the main differences between these methods. The symbols of the same parameters may differ because they often depends on the manufacturers bringing something new to the market. In the case of Mechatronics, this was a stand equipped with a Corescan device.
# Methodology
The test apparatus with the applied method was used to test the selected electrical parameters (i.e., resistance and resistivity): (1) TLMand (2) The measuring stand used for the TLM method is a measuring system developed as part of a research project. The proposed solution used for measuring the selected electrical parameters with the TLM method was granted a patent by the Patent Office of the Republic of Poland in 2014-P.398223. On the other hand, the measuring stand used for the PD method was introduced to the market in 2000. It was designed by SunLab (an ECN spin-off) and manufactured and delivered to the market by Mechatronics (under license from SunLab) .
As part of the experiment, the selected electrical parameters (i.e., resistance and resistivity of photovoltaic cells) were tested, and then, using the formulae contained in, their values were determined. In the TLM method, a row of five front electrodes (strip size-0.1 mm × 8 mm) with a variable distance between them (2.5; 5; 10; 30 mm) was used. The electrode height was 15 μm and the surface resistance of the emitter diffusion layer was 50 Ω/□. A current value of 30 mA was used for the TLM method, while for the PD method a current density value of 30 mA/cm 2 was used. The front electrode was made of DuPont standard PV19b silver paste, which is commonly used in the industry. The front electrode was applied using a template. The tests were carried out on two series of samples, which differed in terms of the firing temperature used in the front metallisation in the belt furnace, selected to obtain the objectively most advantageous product featurei.e., the solar cell and the measurements performed.
A total of 32 samples were used for this investigation. This paper presents the selected results of samples produced with the use of the methods described. The research was carried out in a narrow scope. For this article, samples were selected with layer parameter distributions similar to a Gaussian distribution. The limit values from the interval were given for each series, which meant the minimum value of the resistivity of the front electrode connection with the substrate. The measuring stand used for the TLM method is a measuring system developed as part of a research project. The proposed solution used for measuring the selected electrical parameters with the TLM method was granted a patent by the Patent Office of the Republic of Poland in 2014-P.398223. On the other hand, the measuring stand used for the PD method was introduced to the market in 2000. It was designed by SunLab (an ECN spin-off) and manufactured and delivered to the market by Mechatronics (under license from SunLab) .
As part of the experiment, the selected electrical parameters (i.e., resistance and resistivity of photovoltaic cells) were tested, and then, using the formulae contained in, their values were determined. In the TLM method, a row of five front electrodes (strip size-0.1 mm × 8 mm) with a variable distance between them (2.5; 5; 10; 30 mm) was used. The electrode height was 15 µm and the surface resistance of the emitter diffusion layer was 50 Ω/ . A current value of 30 mA was used for the TLM method, while for the PD method a current density value of 30 mA/cm 2 was used. The front electrode was made of DuPont standard PV19b silver paste, which is commonly used in the industry. The front electrode was applied using a template. The tests were carried out on two series of samples, which differed in terms of the firing temperature used in the front metallisation in the belt furnace, selected to obtain the objectively most advantageous product feature-i.e., the solar cell and the measurements performed.
A total of 32 samples were used for this investigation. This paper presents the selected results of samples produced with the use of the methods described. The research was carried out in a narrow scope. For this article, samples were selected with layer parameter distributions similar to a Gaussian distribution. The limit values from the interval were given for each series, which meant the minimum value of the resistivity of the front electrode connection with the substrate.shows the results of measurements of the electrical properties of the first series of photovoltaic cells. Based on the results obtained by testing the electrical properties of this series of photovoltaic cells, it was found that the lowest values for resistance and resistivity were obtained from samples with a firing temperature of 940 - C for both methods (TLM and PD). These values were R = 0.4 Ω, ρ = 18 mΩ cm 2 for the TLM method and R = 2.5 Ω cm, ρ = 20 mΩ cm 2 for the PD method. In the case of cells where the electrodes were fired at a low temperature-e.g., 840 - C-it can be concluded that the discrepancies in the measurements of the resistance and resistivity values from both methods resulted from the poor connection of the electrode with the substrate.shows the results of measurements of the electrical properties of the first series of photovoltaic cells. Based on the results obtained by testing the electrical properties of this series of photovoltaic cells, it was found that the lowest values for resistance and resistivity were obtained from samples with a firing temperature of 940 °C for both methods (TLM and PD). These values were R = 0.4 Ω, ρ = 18 mΩcm 2 for the TLM method and R = 2.5 Ωcm, ρ = 20 mΩcm 2 for the PD method. In the case of cells where the electrodes were fired at a low temperature-e.g., 840 °C-it can be concluded that the discrepancies in the measurements of the resistance and resistivity values from both methods resulted from the poor connection of the electrode with the substrate.shows the results of measurements of the electrical properties of the first series of photovoltaic cells. Based on the results obtained by testing the electrical properties of this series of photovoltaic cells, it was found that the lowest values for resistance and resistivity were obtained from samples with a firing temperature of 940 °C for both methods (TLM and PD). These values were R = 0.4 Ω, ρ = 18 mΩcm 2 for the TLM method and R = 2.5 Ωcm, ρ = 20 mΩcm 2 for the PD method. In the case of cells where the electrodes were fired at a low temperature-e.g., 840 °C-it can be concluded that the discrepancies in the measurements of the resistance and resistivity values from both methods resulted from the poor connection of the electrode with the substrate.shows the results of measurements of the electrical properties of the second series of photovoltaic cells. Based on the results of this series of tests, it was found that the lowest values of resistance and resistivity were obtained for samples with a firing temperature of 930 - C for both methods (TLM and PD). These values were R = 0.5 Ω, ρ = 30 mΩ cm 2 for the TLM method and R = 4 Ωcm, ρ = 35 mΩ cm 2 for the PD method.presents an original measurement printout from the Corescan device of the resistance distribution of the cells.
## List of test results performed with the use of the tlm and pd methods
## List of test results performed with the use of the tlm and pd methods
## List of test results performed with the use of the tlm and pd methods
Materials 2021, 14, x FOR PEER REVIEW 8 of 12shows the results of measurements of the electrical properties of the second series of photovoltaic cells. Based on the results of this series of tests, it was found that the lowest values of resistance and resistivity were obtained for samples with a firing temperature of 930 °C for both methods (TLM and PD). These values were R = 0.5 Ω, ρ = 30 mΩcm 2 for the TLM method and R = 4 Ωcm, ρ = 35 mΩcm 2 for the PD method.presents an original measurement printout from the Corescan device of the resistance distribution of the cells. To summarise, based on the comparative analysis of the first series of photovoltaic cells using the TLM and PD methods, it was found that the measurements of resistivity in a temperature range of 900 to 960 °C are of the same or similar order. It can also be stated that in a temperature range of 840 to 900 °C, the cells showed the highest and most non-uniform contact resistance, which resulted in the dispersion of the obtained resistivity values. At 940 °C, the cell with the lowest contact resistance without damage was obtained.
In the second series, the results of the resistance values at the metal-semiconductor contact were scattered. Discrepancies in the obtained data may have resulted from, among other things, the manual application of the front metallisation-e.g., using a template-or another stage in the technological process of producing a finished photovoltaic cell. The results of the resistivity measurements in the accepted temperature range are of a similar order. One can also notice some defects in all samples. This is confirmed by the results of the risk at the metalsemiconductor interface. The lowest resistance value was obtained at 930 °C.
After the analysis, it can be unequivocally stated that each of the described methods can be used to collect data and information that can be used in the cell manufacturing process. The dispersion of their values may result from, among other things, the technology used in manufacturing the finished product; production automation (which reduces costs and significantly increases productivity); the robotisation of control stations; the methods and indicators used in the product quality assessment at every stage of production; and the mathematical formulae available in the literature for their calculation and analysis. To summarise, based on the comparative analysis of the first series of photovoltaic cells using the TLM and PD methods, it was found that the measurements of resistivity in a temperature range of 900 to 960 - C are of the same or similar order. It can also be stated that in a temperature range of 840 to 900 - C, the cells showed the highest and most non-uniform contact resistance, which resulted in the dispersion of the obtained resistivity values. At 940 - C, the cell with the lowest contact resistance without damage was obtained.
In the second series, the results of the resistance values at the metal-semiconductor contact were scattered. Discrepancies in the obtained data may have resulted from, among other things, the manual application of the front metallisation-e.g., using a template-or another stage in the technological process of producing a finished photovoltaic cell. The results of the resistivity measurements in the accepted temperature range are of a similar order. One can also notice some defects in all samples. This is confirmed by the results of the risk at the metal-semiconductor interface. The lowest resistance value was obtained at 930 - C.
## Summary
1. Both series of samples were made using the same technology. The measurement methods used in two independent research centres made it possible to compare their advantages and disadvantages. The potential difference method allows the modelling of the contact resistance of the front electrode's frontal contact mesh and the optimisation of the burnout process. In addition, it is possible to graphically present the contact resistance measurements of the front electrode of the photovoltaic cell in 2D/3D, which is very useful for detecting possible defects at this stage of the technological process. This is not possible in the transmission line method. The PD method is destructive, while in the TLM method, depending on the measuring probes used, the measurement can be repeated. Automatic measurement and adjustment of measurement parameters is another advantage of the PD method, but the dimensions and the cost of purchasing the entire stand can be classified as disadvantages. Performing the measurements manually is associated with an extended time of implementation, so this can be considered a disadvantage of the TLM method, while the cost of purchasing the station itself is low, which is clearly an advantage. 2. The researcher/research team, depending on their requirements or the needs of the industrial market, decide on the choice of method and a suitable measuring station. 3. Based on the research analysis, it can be concluded that the measured values of resistivity may differ slightly from each other. This is mainly due to the formulae available in the literature, the number of measurements performed, and After the analysis, it can be unequivocally stated that each of the described methods can be used to collect data and information that can be used in the cell manufacturing process. The dispersion of their values may result from, among other things, the technology used in manufacturing the finished product; production automation (which reduces costs and significantly increases productivity); the robotisation of control stations; the methods and indicators used in the product quality assessment at every stage of production; and the mathematical formulae available in the literature for their calculation and analysis.
## Summary
## 1.
Both series of samples were made using the same technology. The measurement methods used in two independent research centres made it possible to compare their advantages and disadvantages. The potential difference method allows the modelling of the contact resistance of the front electrode's frontal contact mesh and the optimisation of the burnout process. In addition, it is possible to graphically present the contact resistance measurements of the front electrode of the photovoltaic cell in 2D/3D, which is very useful for detecting possible defects at this stage of the technological process. This is not possible in the transmission line method. The PD method is destructive, while in the TLM method, depending on the measuring probes used, the measurement can be repeated. Automatic measurement and adjustment of measurement parameters is another advantage of the PD method, but the dimensions and the cost of purchasing the entire stand can be classified as disadvantages.
Performing the measurements manually is associated with an extended time of im-plementation, so this can be considered a disadvantage of the TLM method, while the cost of purchasing the station itself is low, which is clearly an advantage.
## 2.
The researcher/research team, depending on their requirements or the needs of the industrial market, decide on the choice of method and a suitable measuring station. 3.
Based on the research analysis, it can be concluded that the measured values of resistivity may differ slightly from each other. This is mainly due to the formulae available in the literature, the number of measurements performed, and the technological processes used, often the lack of automatic line technology in the cell manufacturing process. The analysis mentioned above proves the credibility of the obtained results of electrical measurements using the TLM method in research centre A and the PD method in research centre B, as well as the legitimacy of their use in testing photovoltaic cells. 4.
The TLM method has a few problems. Firstly, the fit of the straight line to the measurement points, expressed by the value of its slope, is subject to uncertainty. A more significant problem, however, is the correct determination of the value of the layered resistance used to calculate the resistance between the two contact lines. This is because, in the process of forming the contact between the paste and the substrate, there is a change in the layer resistance directly below the contact line. Secondly, the method does not take into account the resistance of the metal on the contact line. The advantage of the TLM method is a simple sample preparation procedure (no need to use a ready-made sample of solar cell) and quick and direct measurement of the resistance values. It is also a non-destructive method. The required components of the stand are calibrated according to the guidelines of the manufacturers of these components. The measuring station enables:
- Setting the current value on the C401B calibrator in a range from 0 to 110 mA with a resolution of 0.01 mA, with an accuracy of ± (0.1% of the set value + 6 digits); - Voltage measurement with the BM859CF voltmeter in a range from 0 to 500 mV with a resolution of 0.01 mV, and in a range from 0.5 to 50 V with a resolution of 0.1 mV and an accuracy of ± (0.02% of the set value +2 digits).
In the case of the PD method, the problem is the precise determination of the current density value necessary in the measurement procedure. This can be performed using a high-class monochromator, measuring the spectral efficiency of a cell-e.g., silicon-in a range of 300-1100 nm; multiplying it by the photon flux; and then integrating it over the entire range and calculating the J sc of the cell. The second method is to measure the bright I-V characteristic and calculate J sc , but this is also subject to uncertainty. The second disadvantage of the method is its destructive procedure that does not allow repeated measurements of the sample in the same area. This necessitates the use of samples in the form of a ready cell. The manufacturer is responsible for the calibration of the device, and the measurement accuracy refers to a constant value in the formulae (i.e., "correction factor" (* 1.8) for current leakage out of the spot and shading by the probe), which is beyond the control of the user. On the other hand, the advantage of the PD method is the ability to determine the quality of the sample (e.g., its defects) based on the graphical observation of the result and, more importantly, the determination of contact homogeneity (i.e., the low or high uniformity of the contact resistance value). |
The association between MTHFR 677C>T genotype and folate status and genomic and gene-specific DNA methylation in the colon of individuals without colorectal neoplasia1234
Background: Decreased genomic and increased gene-specific DNA methylation predispose to colorectal cancer. Dietary folate intake and the methylenetetrahydrofolate reductase polymorphism (MTHFR 677C.T) may influence risk by modifying DNA methylation. Objective: We investigated the associations between MTHFR 677C.T genotype, folate status, and DNA methylation in the colon. Design: We conducted a cross-sectional study of 336 men and women (age 19-92 y) in the United Kingdom without colorectal neoplasia. We obtained blood samples for measurement of serum and red blood cell folate, plasma homocysteine, and MTHFR 677C.T genotype and colonic tissue biopsies for measurement of colonic tissue folate and DNA methylation (genomic-and genespecific, estrogen receptor 1, ESR1; myoblast determination protein 1, MYOD1; insulin-like growth factor II, IGF2; tumor suppressor candidate 33, N33; adenomatous polyposis coli, APC; mut-L homolog 1, MLH1; and O 6 -methylguanine-DNA methyltransferase, MGMT) by liquid chromatography/electrospray ionization mass spectrometry and pyrosequencing, respectively. Results: Of the 336 subjects recruited, 185 (55%) carried the CC, 119 (35%) the CT, and 32 (10%) the TT alleles. No significant differences in systemic markers of folate status and colonic tissue folate between genotypes were found. The MTHFR TT genotype was not associated with genomic or gene-specific DNA methylation. Biomarkers of folate status were not associated with genomic DNA methylation. Relations between biomarkers of folate status and gene-specific methylation were inconsistent. However, low serum folate was associated with high MGMT methylation (P = 0.001). Conclusion: MTHFR 677C.T genotype and folate status were generally not associated with DNA methylation in the colon of a folate-replete population without neoplasia. This trial was registered at clinicaltrials.gov as ISRCTN43577261.Am J Clin Nutr 2013;98:1564-74. . 5 Abbreviations used: APC, adenomatous polyposis coli; ESR1, estrogen receptor 1; IGF2, insulin-like growth factor II; LINE-1, long interspersed nuclear element-1; MGMT, O 6 -methylguanine-DNA methyltransferase; MLH1, mut-L homolog 1; MTHFR, methylenetetrahydrofolate reductase; MYOD1, myoblast determination protein 1; N33, tumor suppressor candidate 33; PCR, polymerase chain reaction; QC, quality control; SFRP1, secreted frizzled-related protein 1.
# Introduction
There is conflicting evidence regarding the role of folate in colorectal cancer. Meta-analyses of cohort studies generally point to a negative association between folate intake and risk [bib_ref] Folate intake and colorectal cancer risk: a meta-analytical approach, Sanjoaquin [/bib_ref] [bib_ref] Pooled analyses of 13 prospective cohort studies on folate intake and colon..., Kim [/bib_ref]. More recently, 2 US cohorts reporting postfortification dietary folate intakes showed significant negative associations [bib_ref] High levels of folate from supplements and fortification are not associated with..., Stevens [/bib_ref] , whereas another US cohort showed no effect [bib_ref] B vitamin intakes and incidence of colorectal cancer: results from the Women's..., Zschäbitz [/bib_ref]. Results from randomized controlled trials investigating whether folic acid supplementation decreases the risk of colorectal adenoma (precursor of colorectal cancer) are largely negative [bib_ref] A randomized trial on folic acid supplementation and risk of recurrent colorectal..., Wu [/bib_ref] [bib_ref] Folate supplementation and adenomatous colonic polyps, Paspatis [/bib_ref] [bib_ref] Folic acid for the prevention of colorectal adenomas: a randomized clinical trial, Cole [/bib_ref] , although the follow up of these trials has been probably too short to result in changes in risk of colorectal cancer, which develops over decades.
Folate provides one-carbon units for DNA methylation and DNA synthesis, and disruption in both of these processes is the hallmark of neoplasia. Modifications to DNA methylation occur on a genomic-and gene-specific level in colorectal neoplasia: a decrease in genomic DNA methylation (hypomethylation) [bib_ref] Hypomethylation of DNA from benign and malignant human colon neoplasms, Goelz [/bib_ref] is accompanied by an increase in methylation of CpG islands in the promoter regions of many tumor-suppressor genes (hypermethylation) [bib_ref] Epigenetics in cancer, Esteller [/bib_ref] [bib_ref] CpG island methylator phenotype in cancer, Issa [/bib_ref].
It is uncertain whether folate status influences DNA methylation in the colon. Some reports have suggested associations between serum and red blood cell folate and genomic DNA methylation in people with and without neoplasia [bib_ref] Folate status, genomic DNA hypomethylation, and risk of colorectal adenoma and cancer:..., Pufulete [/bib_ref] [bib_ref] Influence of folate status on genomic DNA methylation in colonic mucosa of..., Pufulete [/bib_ref] , whereas others found none [bib_ref] Global DNA hypomethylation (LINE-1) in the normal colon and lifestyle characteristics and..., Figueiredo [/bib_ref]. One study found a positive association between red blood cell folate and hypermethylation of estrogen receptor 1 (ESR1) [bib_ref] High levels of folate from supplements and fortification are not associated with..., Stevens [/bib_ref] and secreted frizzled-related protein 1 (SFRP1) genes in the colon of subjects with adenoma [bib_ref] Association between folate levels and CpG Island hypermethylation in normal colorectal mucosa, Wallace [/bib_ref] , whereas another found no associations between markers of folate status and methylation in ESR1 and mut-L homolog 1 (MLH1) [bib_ref] Methylation of estrogen receptor alpha and mutL homolog 1 in normal colonic..., Al-Ghnaniem [/bib_ref]. Folic acid intervention trials in individuals with adenoma have reported equally mixed results, with some suggesting that supplementation can reverse genomic DNA hypomethylation in the colon [bib_ref] Effect of folate supplementation on DNA methylation of rectal mucosa in patients..., Cravo [/bib_ref] [bib_ref] Effect of folic acid supplementation on genomic DNA methylation in patients with..., Pufulete [/bib_ref] and others showing no effect on gene-specific methylation [bib_ref] Association between folate levels and CpG Island hypermethylation in normal colorectal mucosa, Wallace [/bib_ref] [bib_ref] Folic acid and vitamin B-12 supplementation does not favorably influence uracil incorporation..., Van Den Donk [/bib_ref].
The common 677C.T polymorphism in the gene encoding the key enzyme methylenetetrahydrofolate reductase (MTHFR) provides an additional factor that could modify associations between folate status and DNA methylation in the colon. Although associated with a lower risk of colorectal cancer [bib_ref] Meta-and pooled analyses of the methylenetetrahydrofolate reductase (MTHFR) C677T polymorphism and colorectal..., Taioli [/bib_ref] , the exact interaction of this polymorphism with DNA methylation in the colon is not known, although interactions with folate status and genomic DNA methylation in leukocytes have been reported [bib_ref] The MTHFR 677TT genotype and folate intake interact to lower global leukocyte..., Axume [/bib_ref] [bib_ref] A common mutation in the 5,10-methylenetetrahydrofolate reductase gene affects genomic DNA methylation..., Friso [/bib_ref].
The current study was designed to assess the association between the MTHFR 677C.T polymorphism, markers of folate status, and genomic-and gene-specific DNA methylation in the colon of subjects without colorectal neoplasia. We investigated gene-specific DNA methylation in 7 tumor suppressor genes: ESR1, myoblast determination protein 1 (MYOD1), insulin-like growth factor II (IGF2), tumor suppressor candidate 33 (N33), adenomatous polyposis coli (APC), MLH1, and O 6 -methylguanine-DNA methyltransferase (MGMT). These genes were selected because they all show increased methylation and silencing in colorectal tumors [bib_ref] CpG island methylator phenotype in colorectal cancer, Toyota [/bib_ref] , and some also show age-related increases in methylation in the normal colonic mucosa of individuals with and without colorectal neoplasia [bib_ref] Methylation of estrogen receptor alpha and mutL homolog 1 in normal colonic..., Al-Ghnaniem [/bib_ref] [bib_ref] Methylation of the ESR1 CpG island in the colorectal mucosa is an..., Belshaw [/bib_ref] [bib_ref] Methylation of hMLH1 promoter correlates with the gene silencing with a region-specific..., Deng [/bib_ref] [bib_ref] DNA hypermethylation in the normal colonic mucosa of patients with colorectal cancer, Kawakami [/bib_ref] [bib_ref] Age-related hypermethylation of the 5# region of MLH1 in normal colonic mucosa..., Nakagawa [/bib_ref] [bib_ref] MGMT promoter methylation and field defect in sporadic colorectal cancer, Shen [/bib_ref].
# Subjects and methods
## Study population
Subjects were patients (.18 y of age with no upper age limit) who were referred for a clinically indicated colonoscopy at the Endoscopy Departments at King's College Hospital and Guy's and St Thomas' Hospital NHS Foundation Trusts between December 2006 and February 2009. Subjects were eligible for inclusion if they had a normal colonoscopy (to the ileo-caecal valve) with no macroscopic evidence of colorectal tumors (polyps or cancer) or inflammation. Subjects were excluded if they had any of the following: tumors at current colonoscopy; a strong family history of colorectal cancer as defined in the British Society of Gastroenterology guidelines (one first-degree relative with colorectal cancer younger than 45 y; 2 first-degree relatives with colorectal cancer; a family history of endometrial and colorectal cancer, where one colorectal cancer occurs in a relative younger than 50 y); a history of familial polyp syndromes; history of colorectal tumors; inflammatory bowel disease; a current or history of gluten-sensitive enteropathy; documented renal or liver disease; pregnancy; epilepsy; alcoholism; pernicious anemia; and use of antifolate medication. Ethical approval was obtained from King's College Hospital NHS Trust (REC number: 06Q0703/29).
Patients were approached at the preassessment stage before admission to the Endoscopy Department. Initially, medical notes were screened to exclude those who clearly did not meet the inclusion criteria; those who were deemed suitable were approached, informed about the study, and asked for consent. On the day of colonoscopy, those who provided written consent were asked to complete a health screening questionnaire that collected basic demographic information and information on smoking habits, alcohol intake, current medication, and supplement use. Subjects were defined as supplement users if they reported taking any nutritional supplements [vitamins (including folic acid), minerals, amino acids, and fatty acids] at any time in the previous 3 mo. Subjects were defined as folic acid supplement users if they reported taking folic acid supplements or multivitamin preparations containing folic acid at any time in the previous 3 mo. Subjects were also weighed and their heights were measured and BMI calculated as weight/height 2 . All patients had prepared for colonoscopy by consuming an orally administered colonic lavage solution (KleanPrep; Norgine).
A fasting blood sample was collected before colonoscopy by using the evacuated tube technique (red top containing no anticoagulant for serum folate and vitamin B-12 and liver function tests and purple top containing EDTA for full blood count, red blood cell folate, plasma homocysteine, and DNA extraction). The EDTA-treated samples for plasma homocysteine were chilled on ice. Blood samples were spun within 2 h of collection; serum and plasma were stored at 2808C until assayed. At the end of colonoscopy, 6 mucosal biopsies were removed from the rectum (w12 cm from the anal verge) and immediately snapfrozen in liquid nitrogen. Of these, only 4 biopsies were available for use in the current study. After colonoscopy, subjects were given a food-frequency questionnaire (31) designed to assess dietary folate intake to take home and complete; they were asked to return the completed food-frequency questionnaire in the stamped addressed envelope provided.
## Sample size and primary endpoints
The primary endpoints were differences in genomic-and genespecific DNA methylation between the MTHFR 677C.T genotypes (CC, CT, and TT). Because there are no data on differences in gene-specific methylation between genotypes, sample size was calculated by using data from a previous observational study in which leukocyte DNA methylation was measured in healthy volunteers [bib_ref] A common mutation in the 5,10-methylenetetrahydrofolate reductase gene affects genomic DNA methylation..., Friso [/bib_ref]. On the basis of differences in genomic DNA methylation between CC and TT individuals in that study, a comparison between 32 TT individuals and 128 CC individuals would be adequate to detect a 50% difference (0.5 SD, 30 ng mCyt/mg DNA) in genomic DNA methylation between CC and TT individuals (5% significance level, 2 tailed, and 80% power; specifying the ratio of CC to TT as 4).
## Biochemical analyses
Markers of folate status (serum and red blood cell folate and plasma homocysteine), serum vitamin B-12, and liver function variables were all measured in the Departments of Clinical Biochemistry and Haematology at King's College Hospital NHS Trust, London. Liver function was determined by using the Advia 2400 Chemistry System (Siemens Health Care Diagnostics). The hematocrit (required to calculate red blood cell folate) was measured in fresh whole blood within 24 h of collection by using an ADVIA 2120 Hematology System (Siemens Health Care Diagnostics). Serum folate, vitamin B-12, red blood cell folate, and plasma homocysteine were analyzed by competitive protein binding enzyme immunoassay with direct chemiluminescent technology on an Advia Centaur XP Immunoassay System (Siemens Health Care Diagnostics). For both folate and homocysteine assays, the within-run CV was ,2.2% and the betweenrun CV was ,5.2%.
Genotyping for the MTHFR 677C.T mutation DNA was extracted from whole blood by using the GenElute Blood Genomic DNA Kit (Sigma-Aldrich). The DNA was quantified and assessed for purity by using the Nano Drop ND-1000 ultraviolet spectrophotometer (Thermo Scientific) and run on agarose gels to determine size (. 20 kb for all samples). The presence of the MTHFR 677C.T mutation was determined by DNA amplification with appropriate forward and reverse primers followed by Hinf I digestion of polymerase chain reaction (PCR) product as previously described [bib_ref] A candidate genetic risk factor for vascular disease: a common mutation in..., Frosst [/bib_ref].
# Tissue folate analysis
Folate concentrations in colorectal biopsies were measured by standard microbiological assay with the use of Lactobacillus rhamnosus (ATCC 7469) [bib_ref] Colonic mucosal concentrations of folate are accurately predicted by blood measurements of..., Kim [/bib_ref]. Briefly, tissue biopsies (5-64 mg) were weighed and homogenized (by crushing and shearing the sample in a standard Eppendorf tube with a polytetrafluoroethylene pestle until the sample was barely visible) in 20 volumes of folate extraction buffer [5 mmol b-mercaptoethanol/L and 0.1 mol sodium ascorbate/L in 0.1 mol bis(2-hydroxyethyl) imino-tris(hydroxymethyl) methane/L, pH 7.85] and immersed in boiling water for 20 min. The homogenate was cooled in an ice bath and centrifuged at 48C, 2000 rpm, for 10 min. The supernatant fluid (1 mL) was transferred to sterile Eppendorf tubes and deconjugated with 1 mL chicken pancreas solution (5 mg/mL) and 3 mL 0.1 mol sodium phosphate buffer/L for 16 to 18 h at 378C to convert folylpolyglutamates to corresponding diglutamate derivatives. Deconjugated folate samples were incubated with L. rhamnosus on 96-well plates for 42 h at 378C. Plates were read on a plate reader at 620 nm. Folate concentrations in the biopsies were calculated from a standard curve of serial dilutions of a folic acid standard. Folate extracted from pooled colorectal tissue samples served as the quality control (QC). For this assay, the within-run CV (30 QC samples) was 3.2% and the between-run CV (QC sample assayed over 4 d) was 6.8%.
## Dna extraction from tissue biopsies
Genomic DNA was extracted (simultaneously with RNA) from the tissue biopsies by using the AllPrep DNA/RNA Mini Kit (Qiagen). Before extraction, frozen samples were disrupted with a micropestle on dry ice followed by homogenization with the QIAshredder (Qiagen). DNA was quantitated and assessed for purity and size as described in the section on genotyping.
## Genomic dna methylation
Genomic DNA methylation was determined by liquid chromatography/electrospray ionization mass spectrometry (LC/ESI-MS); 1-5.6 mg DNA was denatured by heating to 1008C for 5 min and then immediately chilled on ice followed by addition of 0.45 mg [ 13 C, 15 N 2 ]2#-deoxycytidine (TRC Toronto Research Chemicals Inc) (the internal standard) and 1/10 volume of 0.1 mol ammonium acetate/L. The mixture was then digested with 4 U S1 nuclease (Sigma-Aldrich) for 2 h at 458C, 0.004 U snake venom phosphodiesterase (GE Health Care), 1/10 volume of 1 mol ammonium bicarbonate/L for 2 h at 378C, and 1 U shrimp alkaline phosphatase (GE Health Care) for 1 h at 378C.
Digested DNA (40 mL) was chromatographed on a Phenomenex Synergi Polar RP column (250 3 2 mm, 4-mm particle size, 80-A pore size) by using an isocratic mobile phase consisting of 10 mmol/L ammonium formate solution + 0.1% formic acid:methanol (95:5) pumped at a flow rate of 0.2 mL/min. The column temperature was 308C, and the sample rack temperature was 48C. The eluent was diverted from the mass spectrometer to waste from 0 to 3 min. The total run time was 10 min.
The chromatography column was coupled to a Micromass Quattro Ultima tandem mass spectrometer (Micromass Quattro Ultima Altrincham) operating in ESI-positive mode by using the following settings: source block temperature of 1208C, desolvation temperature of 4508C, cone gas (N 2 ) flow of w100 L/h, desolvation gas (N 2 ) flow of w650 L/h, collision gas (Ar) pressure of w2.7 3 10 23 mbar, capillary voltage of 2.5 KV, cone voltage of 15 V. The instrument was operated at unit mass resolution (LM & HM, 1& 2 resolution set to 13), collision energy voltage of 12 V, and multiplier voltage of 650V.
2#-Deoxycytidine and 5-methyl-2#-deoxycytidine (Sigma-Aldrich) were used as external standards. Quantification was carried out in multiple reaction monitoring mode by monitoring the precursor/product ion pair of m/z 228/112 for 2#-deoxycytidine and m/z 242/126 for 5-methyl-2#-deoxycytidine; the dwell time was 500s for each pair. The chromatographic peaks for 2#-deoxycytidine and 5-methyl-2#-deoxycytidine eluted at 5.0 and 6.5 min, respectively. Chromatograms were analyzed by using the Quanlynx v 4.1 software package.
The cytosine and 5-methylcytosine contents in DNA were calculated from calibration curves, where the area ratios of 2#-deoxycytidine or 5-methyl-2#-deoxycytidine to internal standard (labeled deoxycytidine) were plotted against known concentrations of 2#-deoxycytidine or 5-methyl-2#-deoxycytidine. The genomic DNA methylation status was expressed as the relative amount of 5-methyl-2#-deoxycytidine compared with total deoxycytidine residues (ie, % methylcytosine). The within-and between-day CVs (determined from QC samples assayed on the same day and on 12 different days) were 3.9% and 10.0%, respectively.
## Gene-specific dna methylation
Bisulfite modification of DNA (500 ng per sample) was carried out by using the EZ-96 DNA Methylation-Gold Kit (Zymo Research Corp) according to the manufacturer's protocol. Bisulfite-modified DNA was stored at 2208C until used.
Promoter sequences were obtained from the ENSEMBL human genome database (http://www.ensembl.org/Homo_sapiens/Info/ Index). Primers were designed by using the PSQ Assay Design Software The PCR mix (50 mL volume) included the following: 50 ng bisulfate-converted DNA, 10 pmol forward and reverse primer, 103 PCR buffer (containing 1.5 mmol MgCl 2 /L), and an additional 0-1.5 mmol MgCl 2 /L, 200 mmol/L of each dNTP, 1.25 U HotStarTaq DNA polymerase (Qiagen), and nucleasefree water to make up to final volume. The final concentration of MgCl 2 ranged from 1.5 to 3.0 mmol/L. The cycling conditions were as follows: 958C for 15 min; 50 cycles of 958C for 20 s, 54-608C for 20 s, and 728C for 20 s; 728C for 5 min; and 48C hold. Gel electrophoresis was carried out on all PCR products by using 10 mL PCR product, which was loaded onto a 1.5% (wt:vol) agarose gel. The gels where then visualized by using ultraviolet transilluminator (Alpha Imager; Alpha Innotech Corporation).
Pyrosequencing was carried out by using the PSQHS 96 system (Qiagen). Sample preparation was carried out by using the Pyro-Mark Vacuum Prep Workstation (Qiagen) according to the manufacturer's protocol: 10 mL PCR product (biotinylated strand only) was immobilized to 2 mL Streptavidin Sepharose HP beads (GE Health Care) by using the Vacuum Prep Workstation; the beads were released onto a PSQ HS 96 plate containing 10 pmol sequencing primer (ESR1, GAGGGGTGTTTAGAGTTT; MYOD1, GAAAGTTAGTTTAGAGGTGA; IGF2, GAGGTAGATGA-AGAGGA; N33, AGGAGGAGATTGATAGAGTA; MLH1, GTAGTATTGTGTTTAGTTT; MGMT, GATTTGGTGAGTGTT-TGGGT; APC, GGAGAGAAGAGAGTTATAG), and the plate was heated to 808C for 2 min to anneal the primers. Sequencing was carried out on the PSQHS 96 system according to standard procedures. The Pyro Q-CpG (Qiagen) software was used to quantify CpG methylation. The degree of methylation at each CpG site was calculated as a percentage by using the allele quantification functionality of the PSQHS 96 software and is expressed as a percentage [methylation (%) = peak height methylated/(peak height methylated -peak height nonmethylated) 3 100]. For each gene, we calculated the average methylation (%) across all CpG dinucleotides in the sequence studied.
Universal methylated and unmethylated control DNA (Zymo Research Corp) were used as positive and negative controls. A pooled colorectal tissue DNA sample was bisulfite modified and served as the QC for the entire set of genes. Within-run and between-run CVs for all assays were 3% and 7%, respectively.
All laboratory personnel who carried out the blood and tissue folate and plasma homocysteine analysis and the genomic-and gene-specific methylation analysis were blinded to the MTHFR 677C.T genotype status of participants.
# Statistical analysis
Data were analyzed by using STATA 11 for Windows (Sta-taCorpLP). Differences in categorical variables (sex, ethnic group, smoking, and supplement use) between genotypes were compared by using chi-square tests. Differences in single con-tinuous variables between genotypes were compared by using ANOVA (age, BMI, dietary folate intake, and genomic DNA methylation) or Kruskal-Wallis tests, which were used for variables that did not have a normal distribution or for which there was a violation of the equality of variance assumption (alcohol intake, serum, red blood cell and colonic tissue folate, serum vitamin B-12, plasma homocysteine, and gene-specific methylation). The relation between demographic variables (age, sex, ethnic group, BMI, smoking, supplement use, and alcohol intake), biomarkers of folate status (serum, red blood cell and colonic tissue folate, and plasma homocysteine), MTHFR 677C.T genotype, and genomic-and gene-specific DNA methylation was investigated by using Spearman's rank correlation coefficients. The association between plasma homocysteine and demographic variables, biomarkers of folate status, serum vitamin B-12, and MTHFR 677C.T genotype was assessed by using multiple linear regression. Plasma homocysteine was skewed, and no transformation improved the model fit; therefore, we performed the regression analysis after removing 4 outliers. Removal of these outliers did not change the conclusions.
Genomic-and gene-specific DNA methylation values are expressed as a proportion and are therefore bounded between 0 and 1. We assessed the association between genomic-and genespecific methylation and biomarkers of folate status and MTHFR 677C.T genotype by using generalized linear models with the binomial family distribution and logit link function, including the robust option to obtain robust SEs. The coefficient from the model gives the amount of change in the log odds of having high methylation for each unit increase in biomarkers of folate status. All models were adjusted for age (continuous), sex (male or female), ethnic group (white or nonwhite), supplement use (user or nonuser), and serum vitamin B-12 (continuous). For genomic DNA methylation, both unadjusted and adjusted models are presented. For gene-specific methylation, we presented only the adjusted models because the differences between adjusted and unadjusted models were marginal. For all models, we checked model fit visually using plots of residuals versus fitted values and versus each of the predictors in turn and normal QQ plots of standardized deviance residuals. We used Stata's linktest to check whether logit was the correct link function to use and assessed model fit by using the Hosmer and Lemeshow goodness-of-fit test and Pearson's correlation test. For all models, results of these diagnostic tests suggested good model fit (P . 0.05). We corrected for multiple comparisons (8 repeated analyses) by using the Bonferroni correction to reduce the chance of obtaining false-positive results (type I error). We therefore considered a P value of 0.006 to be significant (0.05/8). All statistical tests were 2 sided. We repeated the analysis without adding supplement use to the models to prevent possible overadjustment (because supplement use is clearly linked with the provision of B vitamins). The results were not materially altered (data not shown). For gene-specific methylation, we presented coefficients and 95% CIs in graphic form for ease of interpretation.
# Results
Of the 3112 patients screened, 295 did not consent to participate and 2466 did not meet the inclusion criteria, leaving a total of 351 eligible subjects. Of these, 4 had no blood biochemistry results and 11 had abnormal liver function tests; this left a total of 336 participants for analysis (96% of those eligible). Of these, 185 (55%) were homozygous for the wild-type allele (CC) of the MTHFR 677C.T genotype, 119 (35%) carried one mutant allele (CT) and 32 (10%) were homozygous for the mutant allele (TT). Subject characteristics are shown in [fig_ref] TABLE 1: Subject characteristics by MTHFR 677C [/fig_ref]. There were no significant differences in age, BMI, folate or alcohol intakes, or the distribution of sex, smoking, and supplement use between genotypes. The distribution of white and black subjects differed between genotypes (white: 50% CC, 39% CT, and 11% TT; black: 72% CC, 23% CT, and 4% TT). Only 60 subjects (18%) returned the food-frequency questionnaire, hence dietary folate intakes were only available for these individuals.
The biomarkers of folate and vitamin B-12 status by MTHFR 677C.T genotype are shown in [fig_ref] TABLE 2: Biomarkers of folate and vitamin B-12 status by MTHFR 677C [/fig_ref]. There were no significant differences in serum, red cell and tissue folate, serum vitamin B-12 or plasma homocysteine between genotypes. Serum and red cell folate were positively correlated (r = 0.525, P , 0.001); both were negatively correlated with plasma homocysteine (r = 20.309 and r = 20.261, P , 0.001, respectively). Tissue folate was positively correlated with serum and red cell folate (r = 0.464 and r = 0.292, P , 0.001) and negatively correlated with plasma homocysteine (r = 20.140, P = 0.01). There were no significant correlations between dietary folate intake and any of the biomarkers of folate status in the subset of subjects (n = 60) who completed the food-frequency questionnaire (P . 0.05). However, the use of supplements containing folic acid were significantly correlated with serum folate (r = 0.200, P , 0.001), tissue folate (r = 0.121, P = 0.03) and plasma homocysteine (r = 20.205, P , 0.001), and weakly with red blood cell folate (r = 0.100, P = 0.08).
Multiple linear regression analysis with plasma homocysteine as the dependent variable showed associations with age (P , 0.001), sex (with higher homocysteine concentration in men, P , 0.001), serum vitamin B-12 (P = 0.001) and serum folate (P = 0.002) but no association with MTHFR 677C.T genotype (P . 0.05).
## Genomic dna methylation
The average methylcytosine content of colonic DNA was 4.2% with a range of 1.5-6.8%. There was no influence of MTHFR 677C.T genotype on genomic DNA methylation (P = 0.60, [fig_ref] FIGURE 1: Genomic DNA methylation by MTHFR 677C [/fig_ref]. There was no correlation between genomic DNA methylation and serum, red blood cell or tissue folate, serum vitamin B-12 or plasma homocysteine (all P values .0.05, data not shown). The coefficients and 95% CIs (adjusted and unadjusted) from generalized linear models for the association between genomic DNA methylation and biomarkers of folate status and MTHFR 677C.T genotype are shown in [fig_ref] TABLE 3: Association between genomic DNA methylation in colonic mucosa, MTHFR 677C [/fig_ref]. In the adjusted model, there were no significant associations between any of the folate biomarkers or MTHFR 677C.T genotype and genomic DNA methylation. In the unadjusted model, a significant negative association was found between genomic DNA methylation and red blood cell folate (P = 0.04), but this disappeared after adjustment (P = 0.07).
## Gene-specific methylation
The median (range) methylation across all CpG sites analyzed for ESR1, MYOD1, IGF2, and N33 were as follows: ESR1, 13% (1-40%); MYOD1, 10% (2-28%); IGF2, 8% (1-33%); and N33, 14% (3-67%). For APC, MLH1, and MGMT, the medians (ranges) were as follows: APC, 2% (0-7%); MLH1, 2% (0-31%); and MGMT, 2% (0-5%). MLH1 had a small proportion of subjects (15%) with methylation values .10% (11-31%), but most (85%) had values between 0% and 10%.
ESR1, N33, and MYOD1 methylation were positively associated with age (r = 0.373, r = 0.29, and r = 0.46, respectively; P , 0.001) and with each other (ESR1 and MYOD1, r = 0.408; ER and N33, r = 0.247; N33 and MYOD1, r = 0.233, P , 0.001). IGF2, APC, MLH1, and MGMT methylation were not associated with age (P . 0.05). APC methylation was positively associated with IGF2 methylation (r = 0.117, P = 0.03) and MGMT methylation (r = 0.135, P = 0.01). MLH1 methylation was positively associated with ESR1 methylation (r = 0.110, P = 0.04). No association was found between genomic-and genespecific DNA methylation (P . 0.05).
Of the other demographic variables assessed, sex was associated with IGF2 methylation (16% higher in men than in women; P = 0.02), ethnicity (white compared with nonwhite) was associated with N33 methylation (5% higher in nonwhites; P = 0.04), and supplement use was associated with APC methylation (17% lower in users; P = 0.03). No other significant associations were noted. All subsequent statistical analyses were adjusted for age, sex, ethnic group, and supplement use.
Gene-specific methylation did not differ between CC, CT, and TT genotypes [fig_ref] FIGURE 2: Gene-specific DNA methylation by MTHFR 677C [/fig_ref]. The coefficients and 95% CIs for the association between gene-specific methylation and biomarkers of folate status and MTHFR 677C.T genotype, assessed by using generalized linear models (see [fig_ref] TABLE 1: Subject characteristics by MTHFR 677C [/fig_ref] under "Supplemental data" in the online issue), are shown in. The TT genotype did not influence methylation in any of the genes investigated. However, the CT genotype was associated with lower N33 methylation (P = 0.02) and higher MYOD1 methylation (P = 0.02), although these relations were not significant after the Bonferroni correction for multiple comparisons (which raised the threshold for significance to P = 0.006).
A positive association was found between red blood cell folate and N33 methylation (P = 0.04), a negative association between serumand tissuefolate and MGMT methylation (P = 0.001 and P = 0.01, respectively), and a positive association between plasma homocysteine and MLH1 methylation (P = 0.02). Of these associations, only that between MGMT methylation and serum folate (P = 0.001) remained significant after Bonferroni correction. No other significant associations were noted.
# Discussion
In the current study, we investigated the association between MTHFR 677C.T genotype and biomarkers of folate status and genomic and gene-specific methylation in the colon in 336 subjects without colorectal neoplasia. This population was replete with regards to folate and vitamin B-12 status. The distribution of the MTHFR T allele (35% CT and 10% TT) in our population was in Hardy-Weinberg equilibrium ( x 2 = 3.80).
The MTHFR TT genotype was not associated with either genomic-or gene-specific methylation in the colon. This result confirms our previous finding of no association between the MTHFR TT genotype and either genomic- [bib_ref] Influence of folate status on genomic DNA methylation in colonic mucosa of..., Pufulete [/bib_ref] or gene-specific [bib_ref] Methylation of estrogen receptor alpha and mutL homolog 1 in normal colonic..., Al-Ghnaniem [/bib_ref] DNA methylation in the colon in a smaller sample of subjects without neoplasia (68 and 73 subjects, respectively). Although there was a suggestion of a relation between the CT genotype and N33 and MYOD1 methylation, the relevance of this is not clear, particularly because this relation was not consistent and not significant after Bonferroni correction for multiple comparisons. Only 3 studies have investigated the association between MTHFR 677C.T genotype and DNA methylation, but in individuals with adenoma and cancer. Figueiredo et al [bib_ref] Global DNA hypomethylation (LINE-1) in the normal colon and lifestyle characteristics and..., Figueiredo [/bib_ref] and Wallace et al [bib_ref] Association between folate levels and CpG Island hypermethylation in normal colorectal mucosa, Wallace [/bib_ref] reported no association between MTHFR 677C.T genotype and long interspersed nuclear element-1 (LINE-1) methylation (a measure of genomic DNA methylation) and ESR1 and SFRP1 methylation, respectively, in the colonic mucosa of 388 subjects with previously resected colorectal adenoma. Iacopetta et al (34) measured LINE-1 methylation in the normal mucosa of 178 individuals with colorectal cancer and showed that individuals with the mutant alleles for both MTHFR 677C.T and A1298C (ie, the lowactivity genotype 677TT/1298CC) had lower LINE-1 methylation than did those with a high-activity genotype (677CC/1298AA) (P = 0.03). It is possible that having 2 mutations exerts a stronger effect on DNA methylation than does either mutation in isolation or that the relation between MTHFR 677C.T genotype and DNA methylation differs in subjects with and without cancer.
Our finding of no association between MTHFR TT genotype and DNA methylation is confirmed by other studies in healthy subjects that have measured DNA methylation in lymphocyte DNA [bib_ref] Associations between two common variants C677T and A1298C in the methylenetetrahydrofolate reductase..., Narayanan [/bib_ref] [bib_ref] Association of dietary and genetic factors related to one-carbon metabolism with global..., Ono [/bib_ref] [bib_ref] Global DNA methylation measured by liquid chromatography-tandem mass spectrometry: analytical technique, reference..., Kok [/bib_ref] [bib_ref] The association between first trimester micronutrient intake, MTHFR genotypes, and global DNA..., Merrill [/bib_ref] [bib_ref] Metabolic, hormonal and immunological associations with global DNA methylation among postmenopausal women, Ulrich [/bib_ref]. As with our study population, subjects in these studies were folate replete. In subjects with poor folate status, the TT genotype does influence genomic DNA methylation in lymphocytes [bib_ref] A common mutation in the 5,10-methylenetetrahydrofolate reductase gene affects genomic DNA methylation..., Friso [/bib_ref] [bib_ref] Methylenetetrahydrofolate reductase 677C.T variant modulates folate status response to controlled folate intakes..., Guinotte [/bib_ref] [bib_ref] Methylenetetrahydrofolate reductase 677C.T polymorphism affects DNA methylation in response to controlled folate..., Shelnutt [/bib_ref] [bib_ref] Genomic DNA hypomethylation, a characteristic of most cancers, is present in peripheral..., Stern [/bib_ref] , and TT individuals have lower levels of genomic DNA methylation [bib_ref] A common mutation in the 5,10-methylenetetrahydrofolate reductase gene affects genomic DNA methylation..., Friso [/bib_ref] [bib_ref] Genomic DNA hypomethylation, a characteristic of most cancers, is present in peripheral..., Stern [/bib_ref] or show greater increases in genomic DNA methylation in response to folate repletion [bib_ref] Methylenetetrahydrofolate reductase 677C.T variant modulates folate status response to controlled folate intakes..., Guinotte [/bib_ref] [bib_ref] Methylenetetrahydrofolate reductase 677C.T polymorphism affects DNA methylation in response to controlled folate..., Shelnutt [/bib_ref]. It is possible that an association between MTHFR 677C.T genotype and DNA methylation in the colon is only evident in individuals with low folate status. Friso et al [bib_ref] A common mutation in the 5,10-methylenetetrahydrofolate reductase gene affects genomic DNA methylation..., Friso [/bib_ref] investigated the interaction between folate status and MTHFR 677C.T genotype in determining genomic DNA methylation in 187CC and 105TT subjects. In their study, the difference in genomic DNA methylation in lymphocytes between CC and TT individuals was observed only when serum folate was below the median (,12 nmol/L). In our study the median serum folate concentration was 25 nmol/L, which is more than double that reported by Friso et al [bib_ref] A common mutation in the 5,10-methylenetetrahydrofolate reductase gene affects genomic DNA methylation..., Friso [/bib_ref]. When we stratified our population using the cutoff value used by Friso et al, we observed a 26% decrease in genomic DNA methylation between CC and TT subjects in those with serum folate values ,12 nmol/L, although the number of subjects was too small (12 CC and 2 TT) to allow a statistical comparison.
Another possible reason for the lack of association between the MTHFR TT genotype and DNA methylation in our study was the lack of differences in serum and red blood cell folate between genotypes. Although plasma homocysteine was 13% higher in TT individuals than in CC individuals, this difference was not statistically significant. Generally, although population studies have shown that plasma homocysteine is higher in TT individuals [bib_ref] Relation between folate status, a common mutation in methylenetetrahydrofolate reductase, and plasma..., Jacques [/bib_ref] , this effect is modified by folate statusa high folate status is associated with a reduced effect of MTHFR 677C.T on homocysteine concentrations [bib_ref] Alcohol intake and methylenetetrahydrofolate reductase polymorphism modify the relation of folate intake..., Chiuve [/bib_ref] [bib_ref] Effect of the methylenetetrahydrofolate reductase 677C.T mutation on the relations among folate..., De Bree [/bib_ref] [bib_ref] The methylenetetrahydrofolate reductase 677C.T polymorphism as a modulator of a B vitamin..., Hustad [/bib_ref]. In our multiple linear regression analysis, age, sex, and serum folate had stronger effects on plasma homocysteine concentrations (P # 0.01) than did MTHFR 677C.T genotype (P . 0.05).
The current study showed no association between folate biomarkers and genomic DNA methylation in colonic mucosa [fig_ref] TABLE 3: Association between genomic DNA methylation in colonic mucosa, MTHFR 677C [/fig_ref] after adjustment for various factors. In the unadjusted model, red blood cell folate was negatively associated with genomic DNA methylation (ie, DNA becomes hypomethylated as red blood cell folate increases). This is contrary to our previous report in which we showed a weak positive association between serum and red blood cell folate and genomic DNA methylation (P = 0.07 and P = 0.08, respectively) in subjects without neoplasia [bib_ref] Influence of folate status on genomic DNA methylation in colonic mucosa of..., Pufulete [/bib_ref]. In that study, we used the methyl acceptance assay to measure DNA methylation, and, although studies using the methyl acceptance assay have shown associations between folate status
[formula] g ¼ b 0 þ b1X 1 þ b2X 2 . bkX k ) [/formula]
. For the continuous predictor variables (biomarkers of folate status), the coefficient gives the amount of change in the log odds of having high methylation for each unit increase in the biomarker (ie, a positive coefficient suggests an increase in the odds, whereas a negative coefficient suggests a decrease in the odds of having high methylation with increasing biomarker concentration). For the categorical predictor variable (MTHFR 677C.T), the coefficient gives the log odds of having high methylation in the CT or TT genotype compared with the CC genotype (reference group), ie, a negative coefficient suggests a decrease in the odds of having high methylation in individuals carrying the CT or TT genotype compared with individuals carrying the CC genotype.
2 Threshold for significance using Bonferroni correction for 8 repeated tests is P = 0.006. [bib_ref] Pooled analyses of 13 prospective cohort studies on folate intake and colon..., Kim [/bib_ref] Adjusted for age, sex (male or female), ethnicity (white or nonwhite), supplement use (user or nonuser), serum vitamin B-12, and MTHFR 677C.T genotype. [bib_ref] Pre-and postfortification intake of folate and risk of colorectal cancer in a..., Gibson [/bib_ref] Adjusted for age, sex (male or female), ethnicity (white or nonwhite), supplement use (user or nonuser), and serum vitamin B-12.
and DNA methylation [bib_ref] Effect of folate supplementation on DNA methylation of rectal mucosa in patients..., Cravo [/bib_ref] [bib_ref] Effect of folic acid supplementation on genomic DNA methylation in patients with..., Pufulete [/bib_ref] [bib_ref] DNA methylation as an intermediate biomarker in colorectal cancer: modulation by folic..., Cravo [/bib_ref] , this has not been replicated in studies using other methylation measurements [bib_ref] Global DNA hypomethylation (LINE-1) in the normal colon and lifestyle characteristics and..., Figueiredo [/bib_ref] [bib_ref] The MTHFR 677TT genotype and folate intake interact to lower global leukocyte..., Axume [/bib_ref] [bib_ref] Low colonocyte folate is associated with uracil misincorporation and global DNA hypomethylation..., Mcglynn [/bib_ref]. We measured gene-specific methylation in 7 genes (ESR1, MYOD1, IGF2, N33, APC, MLH1, and MGMT). The first 4 genes (ESR1, MYOD1, IGF2, and N33) had a wide range of methylation values (1-67%), and all except IGF2 were strongly associated with age (P , 0.001). The latter 3 genes (APC, MLH1, and MGMT) were not associated with age and had a narrow range of methylation values (0-7%), except for MLH1, for which a proportion of subjects (19%) had methylation values .10%. These findings are similar to what has been previously reported in the literature for these genes [bib_ref] DNA hypermethylation in the normal colonic mucosa of patients with colorectal cancer, Kawakami [/bib_ref] [bib_ref] Accelerated age-related CpG island methylation in ulcerative colitis, Issa [/bib_ref] [bib_ref] Methylation of the oestrogen receptor CpG island links ageing and neoplasia in..., Issa [/bib_ref].
We found no significant associations between markers of folate status and methylation of ESR1, MYOD1, IGF2, and APC in the colon. For N33, MGMT, and MLH1, some associations were noted, although these were in opposite directions (increased methylation with higher red blood cell folate for N33 and decreased methylation with higher serum and colonic folate for MGMT). The importance of these findings is not certain, given that these relations were not consistent, either between genes or for biomarkers within the same gene and all but one of these associations (MGMT methylation and serum folate) were not significant after correction for multiple comparisons. In our previous study, we found no relation between markers of folate status and ESR1 and MLH1 methylation in 73 subjects without colorectal neoplasia [bib_ref] Influence of folate status on genomic DNA methylation in colonic mucosa of..., Pufulete [/bib_ref]. Wallace et al [bib_ref] Association between folate levels and CpG Island hypermethylation in normal colorectal mucosa, Wallace [/bib_ref] reported that higher red blood cell folate concentrations were associated with higher levels of ESR1 (P = 0.03) and SFRP1 (P = 0.01) methylation in normal colonic mucosa of subjects with adenoma, which suggests a negative effect of high folate status on gene-specific methylation. However, this study included subjects with previously resected adenomas who participated in a randomized controlled trial of folic acid supplementation (1 mg/d for 3 y) for the prevention of further adenomas. Individuals with adenoma may have different gene-specific methylation patterns in their normal mucosa than individuals who are neoplasia free [as has been shown previously [bib_ref] Epigenetics in cancer, Esteller [/bib_ref] [bib_ref] Influence of folate status on genomic DNA methylation in colonic mucosa of..., Pufulete [/bib_ref] [bib_ref] Methylation of the ESR1 CpG island in the colorectal mucosa is an..., Belshaw [/bib_ref] ] and so may respond differently to increasing folate status, particularly when folic acid is administered in pharmacologic doses.
Our study had several strengths. We used a population without neoplasia, which avoided the potential confounding effects that neoplasia may have on DNA methylation. We also used quantitative assays with high sensitivity to assess genomic and gene-specific DNA methylation. We measured both systemic and tissue folate to allow a comprehensive assessment of the relation between folate status and DNA methylation. Colonic tissue folate was strongly associated with serum and red blood cell folate, so it appears that systemic markers of folate status are good surrogates of folate status in the colon. The limitations of our study include the fact that our population may not be representative of general population without neoplasia, because they presented with symptoms (eg, bleeding, change in bowel habit, pain, and bloating) that prompted referral for colonoscopy. These individuals may have different dietary habits and other factors that may have influenced DNA methylation in the colon. Another limitation is that none of the methylation assays were performed in duplicate because of the limited availability of colonic tissue biopsies. This may have influenced our findings in view of the variability in the methylation assays (7-10%). Also, we used rectal biopsies, but there is evidence that methylation levels differ between the left and right side of the colon [bib_ref] Association between folate levels and CpG Island hypermethylation in normal colorectal mucosa, Wallace [/bib_ref].
In conclusion, the findings of the current study suggest that MTHFR 677C.T genotype and biomarkers of folate status do not influence DNA methylation in this population. The suggestion that a higher folate status is associated with lower MGMT methylation needs to be interpreted with caution given the inconsistencies across the folate measures and between different genes. Future studies need to examine these relations in populations with low folate status and examine the influence of other mutations in folate pathway genes on DNA methylation in the colon.
We thank LGC, UK, for providing their facilities and expertise for the genomic DNA methylation measurements in colorectal tissue. Supported by the Biotechnology and Biological Sciences Research Council (BB/D523178/1).
The authors' responsibilities were as follows-MP: designed the study; J Hanks, IA, NK, AG, and CLL: conducted the research; J Harris, CR, and MP: analyzed the data; J Hanks, MP, and PE: wrote the manuscript; and MP . Association between gene-specific methylation in colonic mucosa, biomarkers of folate status, and MTHFR 677C.T genotype. Dots and horizontal lines represent the coefficients and 95% CIs, respectively, obtained from generalized linear models adjusted for age, sex (male or female), ethnicity (white or nonwhite), supplement use (user or nonuser), serum vitamin B-12, and MTHFR 677C.T genotype. The coefficients give the amount of change in the log odds of having high methylation in each gene investigated for each unit increase in the independent variables (with all other variables held constant) serum folate (nmol/L) (A), red blood cell folate (nmol/L) (B), tissue folate (nmol/g tissue) (C), and plasma homocysteine (mmol/L) (D) or the log odds of having high methylation in individuals carrying the MTHFR CT genotype (E) or TT genotype (F) compared with individuals carrying the CC genotype. The vertical dotted line (coefficient of 0) shows the line of no effect. Coefficients .0 indicate an increased odds of having high methylation (positive association), whereas coefficients ,0 indicate a decreased odds of having high methylation (negative association), for each unit increase in biomarker concentrations, or for the CT or TT genotype compared with the CC genotype (reference). The threshold for significance with Bonferroni correction for 8 repeated tests is P = 0.006. 1 P = 0.001 (significant after Bonferroni correction); 2 P = 0.04; 3 P = 0.01; 4 P = 0.02; 5 P = 0.02; 6 P = 0.02. and PE: had primary responsibility for the final content. All authors read and approved the final manuscript. None of the authors had a potential conflict of interest.
[fig] FIGURE 1: Genomic DNA methylation by MTHFR 677C.T genotype. Bars represent means and 95% CIs for unadjusted values (CC, n = 185; CT, n = 119; TT, n = 32; P = 0.60). P = 0.60 for comparison between genotypes (ANOVA). [/fig]
[fig] FIGURE 2: Gene-specific DNA methylation by MTHFR 677C.T genotype. Boxes represent medians and IQRs; whiskers are the expected range and outliers. A: ESR1, N33, MYOD1, and IGF2 methylation. B: APC, MGMT, and MLH1 methylation (CC, n = 185; CT, n = 119; TT, n = 32). Kruskal-Wallis tests were used for comparison between genotypes (ESR1, P = 0.92; N33, P = 0.12; MYOD1, P = 0.24; IGF2, P = 0.77; APC, P = 0.45; MGMT, P = 0.68; MLH1, P = 0.89). [/fig]
[fig] FIGURE 3: Association between gene-specific methylation in colonic mucosa, biomarkers of folate status, and MTHFR 677C.T genotype. Dots and horizontal lines represent the coefficients and 95% CIs, respectively, obtained from generalized linear models adjusted for age, sex (male or female), ethnicity (white or nonwhite), supplement use (user or nonuser), serum vitamin B-12, and MTHFR 677C.T genotype. The coefficients give the amount of change in the log odds of having high methylation in each gene investigated for each unit increase in the independent variables (with all other variables held constant) serum folate (nmol/L) (A), red blood cell folate (nmol/L) (B), tissue folate (nmol/g tissue) (C), and plasma homocysteine (mmol/L) (D) or the log odds of having high methylation in individuals carrying the MTHFR CT genotype (E) or TT genotype (F) compared with individuals carrying the CC genotype. The vertical dotted line (coefficient of 0) shows the line of no effect. Coefficients .0 indicate an increased odds of having high methylation (positive association), whereas coefficients ,0 indicate a decreased odds of having high methylation (negative association), for each unit increase in biomarker concentrations, or for the CT or TT genotype compared with the CC genotype (reference). The threshold for significance with Bonferroni correction for 8 repeated tests is P = 0.006. 1 P = 0.001 (significant after Bonferroni correction); 2 P = 0.04; 3 P = 0.01; 4 P = 0.02; 5 P = 0.02; 6 P = 0.02. [/fig]
[table] TABLE 1: Subject characteristics by MTHFR 677C.T genotype [/table]
[table] TABLE 2: Biomarkers of folate and vitamin B-12 status by MTHFR 677C.T genotype 1 [/table]
[table] TABLE 3: Association between genomic DNA methylation in colonic mucosa, MTHFR 677C.T genotype, and biomarkers of folate status 1 [/table]
|
An improved group teaching optimization algorithm for global function optimization
This paper proposes an improved group teaching optimization algorithm (IGTOA) to improve the convergence speed and accuracy of the group teaching optimization algorithm. It assigns teachers independently for each individual, replacing the original way of sharing the same teacher, increasing the evolutionary direction and expanding the diversity of the population; it dynamically divides the students of the good group and the students of the average group to meet the different needs of convergence speed and population diversity in different evolutionary stages; in the student learning stage, the weak self-learning part is canceled, the mutual learning part is increased, and the population diversity is supplemented; for the average group students, a new sub-space search mode is proposed, and the teacher's teaching method is improved to reduce the diversity in the population evolution process. and propose a population reconstruction mechanism to expand the search range of the current population and ensure population diversity. Finally, the experimental results on the CEC2013 test suite show that IGTOA has clear advantages in convergence speed and accuracy over the other five excellent algorithms. OPEN www.nature.com/scientificreports/ simulating its predator behavior, the algorithm assigns predation tasks such as encircling, chasing, and attacking gray wolves to different levels of gray wolves according to the gray wolf 's social level to complete the predation behavior, thereby achieving global optimization. The ABC algorithm 12 is inspired by the honey harvest behavior, where bees conduct different activities according to their respective division of labor, and realize the sharing and communication of swarm information, so as to find the optimal solution to the problem. The teaching and Learning Optimization (TLBO) algorithm 13 simulates the class-based learning method, and individual students reserve knowledge through two stages: "Teaching" and "Learning", so as to improve the ability of each individual.In order to meet the requirements of the solution accuracy and speed in the engineering application field, the scholars have made a more profound exploration of the group intelligence evolution algorithm, mainly focusing on the following two aspects:On the one hand, the existing group intelligence evolution algorithm is improved to further improve its optimization performance. The more representative algorithms are as follows: In 2017, Wei Sun et al. proposed an all-dimensional particle swarm algorithm with randomly selected neighbors 14 (ADN-RSN-PSO). This algorithm early adopted the randomly selected neighborhood (RSN) learning strategy to improve group diversity, and later adopted the all-dimensional neighborhood (ADN) learning strategy to improve the convergence rate. In 2018, Deng Xianli et al. proposed a multiple group based self-adaptive migration PSO algorithm (MSMPSO) 15 , the algorithm integrates two commonly used neighbor topology to give individuals more information sources, based on the parallel evolution of multiple sub-populations, gives different search characteristics to each subpopulation, realize the collaboration and reasonable allocation of sub-population resources, and finally improve the comprehensive performance of the algorithm. In 2019, Gurcan Yavuz and Dogan Aydın proposed an algorithm for artificial swarm based on self-adaptive search equations (SSEABC )16 , the algorithm integrates selfadaptive strategies, local search strategies, incremental population size strategies, and conversion conditions for local and internal search, showing some advantages through comparative experiments. In 2019, Juan Li et al.proposed an improved adaptive knowledge-learning cuckoo search algorithm (I-PKL-CS) 17 , the proposed algorithm introduces a learning model with individual history knowledge and group knowledge in the original CS algorithm, and adopts threshold statistical learning strategies to develop the potential of individual knowledge learning and group knowledge learning, providing a good trade-off for exploration and development. In 2020, an efficient dual-adaptive random standby enhanced whale optimization algorithm (RDWOA) 18 was proposed by Huiling Chen et al. the algorithm introduces two strategies in the original algorithm: one is random standby or random replacement strategy to improve the convergence speed of the algorithm; the other is to introduce double adaptive weight strategy to improve the overall search ability of the algorithm. In 2021, An improved artificial tree algorithm with two populations (IATTP) 19 was proposed by Yaping Xiao et al., The algorithm proposes the competition mechanism between populations, and the migration can expand the population, reduces the inefficient population size, and realizes the reasonable interaction between the population and the branches In 2022, Zhongkai Feng et al. proposed an enhanced sine cosine algorithm (ESCA) 20 , the algorithm uses reverse learning strategy to expand the search scope, adaptive evolution strategy to improve global exploration, community search strategy to increase population diversity, and greedy selection strategy guarantees solution quality, thus improving the convergence accuracy of the algorithm.On the other hand, many new types of group intelligent evolution algorithms with excellent performance have emerged, such as: In 2020, Selim Yilmaz and Sevil Sen proposed the electric fish optimization (EFO) algorithm 21 according to the way that the electric fish determine the prey orientation and transfer information to each other, the active and passive electric positioning ability of such fish are thought to be able to well balance local and global search. In 2020, a Side-Blotched Lizard Algorithm(SBLA) 22 was proposed for the polymorphic populations of lizards simulated by Oscar Maciel et al., the algorithm utilizes three operators to embody the lizard state and uses a sub-population management strategy to simulate the variation of each state lizard population over time. In 2021, Zhang Kaifeng et al. inspired by the modern corporate teamwork behavior and proposed a Cooperation Search Algorithm (CSA) 23 , it uses team communication, reflection learning, and internal competition to complete the global optimization during repeated iterations. In 2020, Yiying Zhang and Zhigang Jin were inspired by the group teaching mechanism to propose the Group Teaching Optimization Algorithm (GTOA) 24 which is different from the TLBO algorithm, which only includes two stages of "Teaching" and "Learning", GTOA includes four phases: Teacher assignment phase, Ability group phase, Teacher phase, and Student phase, it adopts different teaching methods for students with different amounts of knowledge. Experimental results show that the convergence speed and convergence accuracy of GTOA are significantly better than those of PSO 25 , DE 26 , WOA 27 , NNA 28 , SCA 29 and TLBO 13 .Extensive experiments show that compared with the classical swarm intelligence evolutionary algorithm, the recently proposed evolutionary algorithm usually has higher convergence accuracy and faster convergence speed. However, for more complex optimization problems, they also inevitably have many defects, such as insufficient population diversity, easy to fall into local optimal. In order to meet the requirements of practical engineering applications, it is necessary to further improve the optimization performance of the proposed evolutionary algorithm, which is bound to become a research hot-spot in the field of evolution and engineering applications in recent years. Based on the above background, this paper only studies GTOA, proposes an improved group teaching optimization algorithm, and further improves its ability to solve complex optimization problems. First, assign teachers independently to each body, replace the original way of sharing the same teacher, increase the evolutionary direction, and expand population diversity. Second, dynamically divide the good group of students and average group students to replace the original fixed distribution model to meet the different needs of different evolutionary stages for convergence speed and population diversity. Third, the student learning phase cancels the self-learning part with weak effect, increases the mutual learning part, and supplements the population diversity. Fourth, for the average group of students, a new sub-space search model is proposed, and the teaching method of teachers is improved to reduce the loss of diversity in the process of population evolution. Fifth, a new
In many engineering areas, in order to obtain the maximum economic or social benefits, the best solution needs to be sought under certain conditions, such as: In the issue of cargo transport, the planned transport scheme meets both the quantity and loading conditions and minimizes total transport cost; In the process of product production, it is required to reduce the use of manpower and equipment to maximize the profit while meeting the product specifications. Mathematically, the essence of this process is the solution of optimization problems 1 . Due to the increasingly complex optimization problems proposed in the fields of science, engineering, and economics, even if complex mathematical models are built, it is difficult to obtain more desirable results. However, scholars have found that creatures in nature can accomplish complex tasks such as predation, risk avoidance, and evolution by assembling in groups, collaborating with each other, interacting, and other simple actions 2 . It presents a kind of group intelligence from which people take inspiration and design multiple evolutionary algorithms that can better solve optimization problems. Therefore, the swarm intelligence evolutionary algorithm has become the most effective and widely used method at present [bib_ref] Research on Group Intelligence Optimization algorithm and Its Application (The University of..., Tan [/bib_ref].
At present, in engineering application fields, such as surface roughness modeling and optimization of tungsten-copper alloys in micro-milling processes [bib_ref] Surface roughness modeling and optimization of tungsten-copper alloys in micro-milling processes, Quiza [/bib_ref] ; a decision-making framework for dynamic scheduling of cyber-physical production systems based on digital twins 5 ; evolving fuzzy models of shape memory alloy wire actuators [bib_ref] Evolving fuzzy models of shape memory alloy wire actuators, Precup [/bib_ref] , the representative group intelligent evolution algorithm is mainly as follows: Genetic algorithm (GA), Differential Evolution (DE) algorithm , Particle Swarm Optimization (PSO) algorithm, Whale Optimization algorithm (WOA), Grey Wolf Optimizer (GWO), Artificial Bee Colony Algorithm (ABC) algorithm. Among them, the GA 7 is a biological evolution process that simulates the natural selection and genetic mechanism of Darwin's theory of biological evolution, chromosomes are the main carrier of genetic material, that is, a collection of multiple genes, constitutes the genetic operation of genetic algorithm through selection, crossover and mutation. The DE 8 is proposed based on the evolutionary idea of genetic algorithm, algorithm variant vectors are generated by the parent difference vector and crossed with the parent individual vector to generate new individual vectors, selected directly with its parent individual. The PSO algorithm 9 is inspired by the social behavior of flocks or fish, each individual in a particle population is called a particle, representing a position or possible candidate solution in a multidimensional search space, updating positions by cooperating with each other until the optimal solution is found. The WOA 10 is based on the behavior of whale prey, the location of each whale represents a viable solution, and during each generation of swimming, the whales randomly choose driving or rounding up to gradually approach their prey. The GWO 11 is a group intelligent optimization algorithm for www.nature.com/scientificreports/ population reconstruction mechanism is proposed to increase the possibility of population jumping out of local optimum. Tested on the CEC2013 test set, the results show that IGTOA has certain advantages in convergence speed, convergence accuracy and stability compared with the other five optimization algorithms.The remaining structure of this article is as follows: The "GTOA" Section introduces how the original algorithm GTOA works. In the "IGTOA" Section, the improved algorithm IGATOA is proposed, and its overall structure and improvement methods are elucidated. The "Experiment and analysis" Section will improve the algorithm and the original algorithm and other excellent algorithms, based on the CEC2013 test function simulation experiments, and obtain results. Finally, "Conclusion" Section provides a concluding overview.
## Gtoa
In 2020, Zhang Yiying et al. proposed the GTOA through the simulation group teaching mechanism to solve the continuous function optimization problem. In this algorithm, the decision variable is equal to the discipline, and the individual is equal to the student, that is, the individual student is composed of multiple disciplines, and the fitness value is equal to the knowledge level of the student. For solving the minimization problem, the smaller the fitness value of the individual, the higher the knowledge level of the student. Group the students according to their knowledge level, combine the characteristics of the group, adopt different learning methods, and constantly increase the amount of knowledge of the students in the process of evolution. The pseudo-code of the GTOA is shown in Algorithm 1, and its key operations are described below.
Population initialization. Assume that the population size is N, and the dimension of the problem to be optimized is D, the i-th individual in the initial population X 0 = [X 0 1 , X 0 2 , ... , X 0 N ] T can be described as www.nature.com/scientificreports/ X o i = [x 0 i,1 , x 0 i,2 , ..., x 0 i,D ](i = 1,2,…,N; j = 1,2,…,D), its j-th dimension x i,j is randomly generated according to Eq. (1):
Among them, Um and Lm are the upper and lower limits of the search range of the optimization problem, respectively, and rand(0,1) is a random number between 0 and 1.
Teacher assignment. For the current population X t , a teacher is determined according to Eq. [bib_ref] Review of typical algorithms of group intelligence, Yu [/bib_ref] to teach students in each iteration.
Among them, X t 1 , X t 2 and X t 3 are respectively the three students with the best fitness value, and t is the current number of iterations.
Update good student group. The individual update of the good students group needs to go through the teacher teaching phase and the student learning phase in turn, as follows.
Teacher teaching phase. Generally, students in good group have strong ability to accept knowledge, and teachers pay attention to improving the average knowledge of the whole group when teaching. In view of this, the GTOA designed the teacher teaching method as shown in Eq. (3).
Among them,X t+1 GT i represents the individual update of student X t Gi after the teacher's teaching; F is the teaching factor, with a random value of 1 or 2; a, b, c are random numbers in [0,1], and b + c = 1; M t is the average subject knowledge of students in the good group in the t iteration process, as shown in Eq. [bib_ref] Surface roughness modeling and optimization of tungsten-copper alloys in micro-milling processes, Quiza [/bib_ref].
If the knowledge level of students is improved after teaching, the subject knowledge of students should be updated; otherwise, the original students should remain unchanged. The details are shown in Eq. [bib_ref] A decision-making framework for dynamic scheduling of cyber-physical production systems based on..., Villalonga [/bib_ref].
Student learning phase. According to Formula [bib_ref] Evolving fuzzy models of shape memory alloy wire actuators, Precup [/bib_ref] , students learn from each other within the group to acquire new knowledge of each subject, and decide whether to update the current individual students according to the method described in Eq. [bib_ref] Feature subset selection using a genetic algorithm, Yang [/bib_ref]. Among them, e and g are two random numbers within [0,1]; X t+1 GS i is the individual updated by X t+1 GT i after learning through the student phase during the t + 1 iteration; X t+1 GT j is another individual student randomly selected in this group, and j ≠ i.
Update average student group. Similar to the good group, the individual renewal of the students in the average group also needs to go through the teaching phase and the learning phase successively. Among them, the learning phase of the students in the average group is exactly the same as that of the good group. And according to the differences in the knowledge level of the two groups, the teaching phase of the teachers is different, which is as follows.
In view of the relatively poor knowledge level of students in the average group, teachers are more inclined to improve the knowledge level of individual students in the learning process. The GTOA has developed a teaching plan for the students in the average group, such as Eq. [bib_ref] A fuzzy adaptive differential evolution algorithm, Liu [/bib_ref]. Similar to the teaching of good group teachers, after the teaching, they should also judge whether to update their existing subject knowledge according to Eq. (5).
## Igtoa
A large number of experiments show that for more complex function problems, similar to other swarm intelligence evolutionary algorithms, GTOA also has shortcomings such as slow convergence speed and easy to fall into local optimum. This paper deeply analyzes the reasons for the above problems, and proposes an improved group teaching optimization algorithm (IGTOA), the flow chart of which is shown in Teacher assignment. As described in "Teacher assignment" Section in the teacher assignment stage of the GTOA, from the optimal individual and the average of the three better individuals, the winner is selected as the teacher of all students, and all students in the subsequent teaching stage of the good group and the teaching phase of the average group are selected. Students learn only from that teacher. Obviously, such a single-teacher learning mode will make individuals approach to it quickly, resulting in a single direction of individual explora- www.nature.com/scientificreports/ tion, rapid decline of population diversity, and easy to fall into local optimal. In order to solve the above problems, in IGTOA, all students independently select teachers for learning in the following way, as shown in Eq. (9).
Among them, T t represents the teacher assigned to the students during the teacher teaching stage; P _teacher represents teacher assignment probability. Generally, P _teacher = 0.5 can achieve better results.
By the Eq. (9) can be seen, each individual is independent of choosing the best individual and one of the three is the center of the optimal individual as a teacher, and learn from it, evolutionary direction is no longer a single individual, can better maintain the population diversity, and the best individual and all three is the center of the optimal individual carries the better the evolution of the information, will not be too much lower convergence speed. In short, this new allocation method of independent teachers can balance the convergence rate and population diversity.
Adaptive grouping. In GTOA, according to the fitness of individuals, half of the individuals with better fitness value in the whole population are divided into good group, and the rest are divided into average group. Through the overall analysis of GTOA, it can be found that the evolution of average students is relatively slow, and the main function is to provide evolutionary information for good students to explore and develop new positions, so that they can quickly approach the global optimal position. In short, the average group focused on maintaining population diversity, while the good group was mainly responsible for exploration and search.
Generally, different evolutionary stages have different requirements for algorithm performance: In the early stage of iteration, the fitness value gap between individuals is large, and the population diversity is good. Usually, it is hoped that the algorithm will quickly converge to the region where the optimal solution is located. As the evolution progresses, the fitness value gap between individuals is decreasing. , the individuals become more and more similar, and the population diversity gradually deteriorates. It is expected that the algorithm can increase the population diversity in order to have the ability to jump out of the local optimum. In order to better meet the needs of different evolutionary stages of the algorithm, in the early stage of evolution, the scale of the average group should be appropriately reduced and the scale of the good group should be expanded; and vice versa. Based on the above ideas, this paper proposes a dynamic allocation method of the number of good group and average group students as shown in Eq. [bib_ref] An improved whale algorithm and its application in truss optimization, Jiang [/bib_ref] , and the allocation process is shown in [fig_ref] Figure 2: Individual number of students in the two groups changed [/fig_ref] Among them, N A t and N G t represent the number of students in the average group and the good group in the t-th iteration respectively; P _group is the grouping ratio, generally, P _group = 0.1 can achieve better results; ⌊- ⌋ represents rounding; t is the current iteration number of times; max_t maximum number of iterations.
As can be seen from [fig_ref] Figure 2: Individual number of students in the two groups changed [/fig_ref] , in the process of 1500 iterations, the number of students in the good group was significantly higher than that in the average group in the early stage of evolution. With the increase of iterations, the number of students in the good group gradually decreased, in line with the different needs of good group size and average group size in different evolutionary stages. Improvement of update method for good group in student learning phase. GTOA divides the population into good and average groups according to the merits of the fitness value, and each group goes www.nature.com/scientificreports/ through the teacher teaching phase and the student learning phase accordingly, and the two groups share the same student learning strategy. Unlike the teacher phase, the student learning phase does not learn from good teachers, but from other individuals in the group and themselves. Therefore, the student learning phase is mainly responsible for maintaining the population diversity within the group. An in-depth analysis of the way in which individuals are updated at the student learning stage as shown in Eq. [bib_ref] Evolving fuzzy models of shape memory alloy wire actuators, Precup [/bib_ref] shows that the new individual is actually constituted by "Original individual + Mutual learning + Self-learning". Among them, the mutual learning part randomly selects other individuals in the group to learn, which can provide population diversity to a certain extent; the self-learning part is that the current individual learns from the individual who has not undergone the teacher stage, if the individual is not retained after going through the teacher stage, the current individual is exactly the same as the individual who experienced the teacher stage before. Obviously, self-learning is completely meaningless, and if the individual is retained after going through the teacher stage, the current individual is better than the individual before the teacher stage, and it is difficult to learn from it to produce new individuals who are better than themselves, then the self-learning stage will still not play a role. In short, the self-learning component of GTOA is extremely weak. In order to further enhance its population diversity, the self-learning part is abolished, learning from other individuals is added, and the new student learning stage individual renewal method is proposed as shown in Eq..
[formula] (9) T t ≡ X t 1 , if rand < P_teacher X t 1 + X t 2 + X t 3 3 , otherwise(10) N t A = P_group × N + � N − 2 × P_group × N � × t max _t N t G = N − N t A [/formula]
where X t+1 GS j , X t+1 GT k are two different individuals randomly selected in the good group, i ≠ j ≠ k, respectively, and the definitions of parameters a and b are shown in Eq. [bib_ref] A powerful and efficient algorithm for numerical function optimization: Artificial bee colony..., Karaboga [/bib_ref].
where, sgn(- ) is a symbolic function. Improvement of update method for average student group. As described in "Adaptive grouping" Section, the good group in GTOA is mainly responsible for exploration and search, while the average group focuses on maintaining population diversity. Similar to the updating method of students in the good group, students in the average group also conduct a complete search in the D-dimensional search space. Such a wide range of communication in the search space is very likely to make the population quickly close to several superior individuals and gather in a certain area or several regions, resulting in a serious loss of population diversity. If the average group of students does not conduct a complete large-scale search in the D-dimensional search space, but only conducts a small-scale search in some dimensions, it is easy to overcome the above shortcomings.
Based on the above analysis, average group of students in teachers' teaching and students' learning phase are small range subspace search, among them, the average group of students to study with good group of students learning phase in the same way, and the new design of ordinary teachers teaching phase and sub-space model way choice of dimension to search in the specific as follows.
Dimension selection in subspace patterns. In order to maintain the diversity of the population as much as possible, in the subspace search mode we designed for the average group of students to update, the number of dimensions to be updated by each individual and the specific dimensions are randomly generated, as follows: First, for each individual, a random integer k is randomly generated in , which is the total number of dimensions that the individual needs to update; Then, k random integers are randomly generated in the dimension space , and the subsequent subspace search will be performed in its corresponding dimension.
To further understand the dimension selection in the above subspace pattern, a concrete example is shown in [fig_ref] Figure 3: Dimension selection in subspace search mode [/fig_ref]. Given that the dimension of the problem to be optimized is 30, for example, for individual X2, the number of randomly generated dimensions to be updated is 5, and 5 random integers are randomly generated within [bib_ref] Oppositionbased differential evolution, Rahnamayan [/bib_ref] , which are {4,12,5,29,17} respectively, indicating that individual X2 will only search on the dimension {4,12,5,29,17} when conducting subspace search, the other dimensions don't change.
Improvement of teacher teaching phase. It can be seen from Algorithm 1 that the students in the good group and the average group in GTOA only rely on the students in this group to update their subject knowledge. When the next iteration is regrouped, the good group and the average group can exchange information with each other. However, since the students in the good group are better than the students in the average group, after the knowledge update, the students in the good group are basically better than the students in the average group. Therefore, even after regrouping, only a very small number of students from the average group entered the good group. Obviously, students in the good group only absorbed a very small amount of the evolutionary information of the average group, and did not really get the diversity supplement. Basically, they still relied on their original evolutionary information to evolve, which was easy to quickly converge to a local optimum. In order to make the good group get the full diversity supplement, it is necessary to further improve the subject knowledge of the students in the average group and increase the opportunity for them to enter the good group. In view of this, the teaching methods of average group teachers are improved as follows, as shown in Eq. [bib_ref] Teaching-Learning-Based Optimization: An optimization method for continuous nonlinear large scale problems, Rao [/bib_ref]. represents the teachers assigned according to Eq. (10), and one teacher is independently selected from each dimension; X t+1 GT m,j represents an individual selected randomly from X t+1 GT , and likewise, an individual selected independently from each dimension; F 1 is the random number between [bib_ref] Review of typical algorithms of group intelligence, Yu [/bib_ref] , and a is the random number between [0,1].
[formula] (11) X t+1 GS i = X t+1 GT i + a × (X t+1 GT j − X t+1 GT i ) + b × (X t+1 GT k − X t+1 GT i ) (12) a = rand × sgn(f (X t+1 GT i ) − f (X t+1 GT j )) b = rand × sgn(f (X t+1 GT i ) − f (X t+1 GT k )) (13) X t+1 AT i,j = X t A i,j + F 1 × ((a × T t+1 j + (1 − a) × X t+1 GT m,j ) − X t A i,j ) [/formula]
A comparison of Eqs. [bib_ref] A fuzzy adaptive differential evolution algorithm, Liu [/bib_ref] and [bib_ref] Teaching-Learning-Based Optimization: An optimization method for continuous nonlinear large scale problems, Rao [/bib_ref] can be found that: First, in the GTOA, the learning objects in the average group are all teachers. In this section, all the students of the teachers in the good group are also listed as the learning objects. Because the good group students are the result of learning from teachers, the gap between the average group and the good group is further shortened, and the possibility of the average group gene flowing into the good group is increased. Second, compared with only learning from teachers themselves, the new teaching method of teachers has more combinations of learning objects, which greatly improves the diversity of students in the average group. Although the evolutionary information of some outstanding group students will be mixed into the average group students, the evolutionary genes flowing in different dimensions come from different outstanding students, and the genes of the average group students are completely preserved in the dimension without subspace search, so that there is a big difference between the evolutionary information of the average group students and the evolutionary information of the good group students. Therefore, when the students of the average group flow into the good group, they can be supplemented with a certain population diversity. To sum up, the new teaching method of average group teachers proposed in this section has certain advantages.
In order to further understand the subspace search mode of teacher teaching phase and student learning phase in the average group, a specific example is given in [fig_ref] Figure 4: Subspatial learning mode of teacher teaching stage of average group individual X... [/fig_ref]. Given that the dimension of optimization problem is 10 and the number of individuals is 5. In the teacher teaching phase, the number of dimensions to be updated in the randomly generated subspace of individual XA2 is 3, and 3 dimensions are randomly selected for subspace search, including {2, 5, 9}, and other dimensions remain unchanged. For dimension 2, random teacher T1 and good group student XG2 learn from their 2-nd dimension according to formula [bib_ref] Teaching-Learning-Based Optimization: An optimization method for continuous nonlinear large scale problems, Rao [/bib_ref] ; for dimension 5, teacher T2 and the 5th dimension of good group students XG3 were randomly selected for learning; and for dimension 9, the teacher T1 and the 9-th dimension of good group XG4 were randomly selected for learning. Assuming that the newly generated individual is superior to the original, the original individual will update its knowledge of each subject, otherwise unchanged.
It should be noted here that the students in the average group adopt the same learning method as the good group students in the learning stage, as shown in Eq.. The difference is that they do not use the full space search mode, but use the subspace search mode.
Population reconstruction mechanism. A large number of experimental studies have shown that, similar to other swarm intelligence evolutionary algorithms, for very complex optimization problems, with the progress of evolution, GTOA also has the defects of slow evolution and difficulty in obtaining the global optimal solution. In order to solve this problem, this section proposes a new population reconstruction mechanism, which mainly includes the starting conditions of the population reconstruction mechanism and the population reconstruction strategy. The details are as follows. ... www.nature.com/scientificreports/ Initiating conditions of population reconstruction mechanism. As we all know, swarm intelligence evolutionary algorithm will show the following obvious characteristics when it evolves slowly: in the process of successive iterations, the optimal value obtained by the population does not change. Therefore, this excerpt uses the change of global optimal value as the starting condition of population reconstruction mechanism. Specific methods are as follows: Initialization parameters change_flag = 1 and flag, where change_flag is used to record the number of times that the global optimal value does not change continuously, flag is the threshold that determines the local optimal value; Then, the Euclidean distance of the globally optimal individual in two consecutive iterations is calculated according to Eq. [bib_ref] All-dimension neighborhood based particle swarm optimization with randomly selected neighbors, Sun [/bib_ref] , if move = 0, change_ flag = change_ flag + 1, otherwise, change_ flag = 1. When change_ flag = flag, the population reconstruction policy is started and change_ flag is set to 1.
[formula] 30 t X 2 t X 1 t X 3 t i X t N X individual [/formula]
Among them, bestX t and bestX t−1 represent the optimal individuals in the t-th and (t-1)-th iterations, respectively.
Population reconstruction strategy. Experimental results show that for complex optimization problems, the improved algorithm proposed in this paper has a good ability to maintain population diversity. Even if the phenomenon of slow evolution occurs, the individual differences in the current population are relatively obvious, and the clustering and high similarity of individuals are not presented. Through in-depth analysis of the overall optimization process of GTOA, it is not difficult to find the essential reasons for the above phenomenon as follows: At the beginning of its evolution, GTOA started from a very limited number of individuals. Under the guidance of excellent teachers, it drove each individual to move from the area to the better area, and gradually narrowed the search range, so that outstanding individuals could use the help of a smaller range. Other individuals conduct further in-depth and refined exploration, thereby stimulating better evolutionary information to improve teachers. However, for very complex optimization problems, there are often multiple local optimal solutions, and the global optimal solution is hidden in a narrow region. In the huge search space, each individual will be drawn to several local optimal regions with a high probability, because the local optimal values are relatively similar, it is difficult for each individual to jump out of the local optimal region. Although good population diversity can be maintained, the search area cannot be reduced, thus it is difficult to provide motivation for teachers to further fine exploration. To sum up, in order to force teachers to have the motivation to search carefully, the search scope must be appropriately narrowed and other evolutionary information must be introduced.
Generally, GTOA in slow evolution of several iterations, the better individual has acquired when the area before the local optimum, obviously, compared with extensive search in large search space, surrounded by where l is the scaling coefficient of the lens, typically, l = 10.
[formula] (14) move = D i=1 (bestX t (i) − bestX t−1 (i)) 2 [/formula]
where k is a random integer of . The above population reconstruction methods have the following advantages: First, in several iterations, although the optimal value has not changed, sufficient communication has been carried out between individuals, and effective evolutionary information in poor individuals has been contained in excellent individuals. Therefore, retaining some excellent individuals and removing half of the poor individuals can basically represent the current evolutionary information and will not affect the exploration ability of the population. Second, the reverse learning is carried out on the individuals in the sub-population pop2, which not only further effectively narrows the search range, but also introduces other evolutionary information, which further provides the impetus for the evolution of outstanding individuals. Third, the individual parts and even all dimensions of the population pop3' are randomly selected from the best part of the individuals, and obviously, they all belong to the partially sampled individuals in the region formed by the best part of the individuals. It not only effectively reduces the search area, but also because they are the recombination of the various dimensions of the better individuals, although they contain part of the evolutionary information of the better individuals, they are quite different from them, which supplements the population diversity to a certain extent, so that the algorithm has the power to further fine-tune the search. [fig_ref] Figure 6: Comparison of individual distribution before and after population reconstruction [/fig_ref] shows the comparison before and after population reconstruction. Assuming that the number of individuals is 50, the problem to be optimized is a sphere function with a dimension of 2. It can be seen that the search range is significantly reduced after population reconstruction, and it is closer to the global optimum.
The complexity analysis of the algorithm. The population size of IGTOA algorithm is N; the number of students in excellent and ordinary groups is N good and N average respectively; and the problem dimension is D. As shown in , the IGTOA algorithm includes the following four main steps: good group teacher stage (T _gt ), good group student stage (T _gs ), general group teacher stage (T _at ), general group student stage (T _as ). The time overhead of the IGTOA algorithm also mainly comes from these four operations. At each generation of the IGTOA algorithm runs, the worst-time complexity of the above four operations is analyzed as follows:
The good group teacher stage (T _gt ) needs to calculate N good × D times at most formula (3), so its corresponding worst time complexity is O(N good × D); the good group student stage (T _gs ) needs to calculate N good × D times at most formula (11), so its corresponding worst time complexity is O(N good × D); the average group teacher stage
[formula] (15) pop2 ′ i,j = Um − Lm 2l − pop2 i,j 2l , i ∈ [1 , N2] (16) pop3 ′ i,j = pop1 k,j , i ∈ [ 1 , N3]× D) + O(N good × D) + O(N average × D) + O(N average × D)) ≈O(2 N × D). [/formula]
# Experiment and analysis
In this part, we test the performance of the proposed IGTOA algorithm through four experiments: the first is the parameter sensitivity analysis; the second is the effectiveness of each improvement strategy; and the third is the performance comparison with other excellent intelligent optimization algorithms. The fourth is to compare the results of each algorithm in engineering applications.
"Effect of the parameters on the algorithm", "Proof of the effectiveness of the improvement measures", "Compared with other excellent algorithms" Sections is tested on a CEC2013 test set containing 28 functions. According to the characteristics of these functions, they can be divided into three groups: the first group is the unimodal function F1-F5, the second group is the multimodal function F6-F20, and the third group is the combination function F21-F28. Detailed information about this test set can be found in the literature [bib_ref] Problem Definitions and Evaluation Criteria for the CEC 2013 Special Session on..., Liang [/bib_ref]. Meanwhile, the algorithms are implemented in Matlab 2021a. All the tests are conducted on a personal computer equipped with a core i7-11800H CPU (2.30 GHz) and a 16.0-GB memory.
Effect of the parameters on the algorithm. The proposed algorithm IGTOA adds to the original GTOA algorithm with the following parameters: including change_flag, Pg, l, N1 and N2. When examining the influence of one of the above parameters on the performance of the IGTOA algorithm, the parameter is set to 5 sets of values, and the other parameters remain unchanged. In all experiments, the population size N = 50, the problem dimension D = 30, the maximum function evaluation number MaxFEs = 5000*D, and the other parameters were set as follows:
(1) When examining the effect of change_flag on the performance of the IGTOA algorithm, change_flag is set to 10, 30, 50, 70 and 100, respectively. Other parameters are set as follows: All of the above experiments were run independently 30 times on the CEC2013 test set, and the average value of the optimal value obtained from each independent run when the same number of preset function evaluation times was reached was counted. The specific data are shown in Tables 1 and 2. In Tables 1 and 2, the parameters that perform best on each function are shaded, and the number of functions is calculated on the last line. In order to further compare the performance of the algorithm, the parameters of the data in [fig_ref] Table 1: Effect of parameters change_flag and Pg on IGTOA [/fig_ref] shown as in [fig_ref] Figure 7: Results of the Friedman test for each parameter [/fig_ref] , in which the height of the bar graph represents the size of the average rank of each algorithm, the higher the bar graph, the higher the average rank and the higher the rank, the overall performance of the algorithm of the parameter at the value.
[formula] Pg = 0.1, l = 10, N1 = N2 = 0.2*N. [/formula]
According to the data in [fig_ref] Table 1: Effect of parameters change_flag and Pg on IGTOA [/fig_ref] , when change_flag = 30, the relatively best results are achieved on the 22 functions; when change_flag is 10 or 70, it works relatively best on 16 functions; and when the change_flag = 50 www.nature.com/scientificreports/ or 100, the relatively best results are achieved on 14 and 13 functions, respectively. As can also be seen from the results in [fig_ref] Figure 7: Results of the Friedman test for each parameter [/fig_ref] , IGTOA algorithm performance is optimal when change_flag = 30, when change_flag = 10,50 and 70, IGTOA algorithm performance is little different, but it is inferior to IGTOA algorithm performance when change_flag = 30, while when change_flag = 100, IGTOA algorithm performance is not satisfactory. In "Conclusion" Section, the IGTOA algorithm is more sensitive to the parameter change_flag, and it performs the best when change_flag = 30. Analyzing other data in [fig_ref] Table 1: Effect of parameters change_flag and Pg on IGTOA [/fig_ref] and [fig_ref] Figure 7: Results of the Friedman test for each parameter [/fig_ref] in the same way, we can find that the IGTOA algorithm is sensitive to both parameters l and N1 and N2, with slightly decreased sensitivity to parameter Pg, and the IGTOA algorithm performs best when Pg = 0.1, l = 10 and N1 = N2 = 0.2*N. To sum up. when the parameters change_flag, Pg, l, N1, N2 are set to 30, 0.1, 10, and 0.2 N,0.2 N, respectively, IGTOA has a good optimization effect. If you want to obtain a better effect for a certain actual optimization problem, the above parameters can also be adjusted several times.
## Proof of the effectiveness of the improvement measures.
To verify the effectiveness of each improvement measure in Parts 3.1-3.3, a corresponding improvement strategy was removed in IGTOA, five new improvement algorithms were formed, including the improved algorithm for removing the teacher allocation strategy of "Teacher assignment" Section in IGTOA, the improved algorithm for removing the adaptive grouping strategy of "Adaptive grouping" Section in IGTOA, the improved algorithm for removing the good group improvement strategy of "Improvement of update method for good group in student learning phase" Section in IGTOA, the improved algorithm for removing the common group improvement strategy of "Improvement of update method for average student group" Section in IGTOA and improved algorithms for the population reconstruction strategy of removing "Improvement of teacher teaching phase" Section in IGTOA. For simplicity, the above five new algorithms are called IGTOA1, IGTOA2, IGTOA3, IGTOA4 and IGTOA5, respectively. The above five improved algorithms and IGTOA were compared on the CEC2013 test set. To ensure the fairness of the comparison, the parameters of each algorithm are set exactly the same, as follows: population size N = 50, problem dimension D = 30, maximum evaluation number MaxFEs = 5000*D, change_flag = 30, Pg = 0.1, l = 10, N1 = N2 = 0.2*N. The mean and variance of the optimal values obtained by running the algorithms www.nature.com/scientificreports/ independently 30 times on each function are counted, and the specific results are shown in [fig_ref] Table 3: Comparison results of each improvement strategy with IGTOA on the 30-dimensional CEC2013... [/fig_ref]. Among them, the data from outside and in parentheses represent the mean and standard deviation of the optimal values obtained in 30 independent experiments, respectively. The bold value indicates that the performance of the other improved algorithms are significantly worse than IGTOA on the corresponding functions, counting the number of functions for which the improved algorithm is significantly worse than IGTOA in the penultimate row, and giving the average rank and the ranking results of IGTOA and the other five improved algorithms for Friedman test in the last two lines. As can be seen from [fig_ref] Table 3: Comparison results of each improvement strategy with IGTOA on the 30-dimensional CEC2013... [/fig_ref] , the functions of the five improved algorithms performing perform IGTOA performance are 14, 10, 15, 14 and 11 respectively. This shows that the five improvements proposed in "Teacher assignment", "Adaptive grouping", "Improvement of update method for good group in student learning phase", "Improvement of update method for average student group" , "Improvement of teacher teaching phase" Sections have some effectiveness. Furthermore, the rank of each algorithm obtained by Friedman detection is visible, with the smallest rank of the corresponding improvement algorithm after removing the good group improvement strategy of "Improvement of update method for good group in student learning phase" Section compared to IGTOA. This shows that among the five improvements, the improvement measure in "Improvement of update method for good group in student learning phase" Section has the greatest impact on the performance of the IGTOA algorithm, while several other improvement algorithms have little difference on the performance of IGTOA algorithm.
Compared with other excellent algorithms. In order to fully investigate the performance of IGTOA algorithm, the algorithm, basic GTOA and four recently proposed representative optimization algorithms are analyzed in four aspects of convergence accuracy, convergence speed, stability and running time. Algorithms for comparison include IATTP(2021) [bib_ref] Improved self-adaptive search equation-based artificial bee colony algorithm with competitive local search..., Yavuz [/bib_ref] ; MSMPSO(2018) [bib_ref] A powerful and efficient algorithm for numerical function optimization: Artificial bee colony..., Karaboga [/bib_ref] ; ADN-RSN-PSO(2017); ESCA(2022) [bib_ref] An improved cuckoo search algorithm with self-adaptive knowledge learning, Li [/bib_ref]. To ensure the fairness of the algorithm, the number of individuals in the population is N = 50, the test problem dimen- www.nature.com/scientificreports/ sions D = 30, 100, the maximum function evaluation times MaxFEs = 5000*D, and the remaining parameters are shown in .
Comparative experiment on algorithm convergence. [fig_ref] Table 5: Results of each algorithm on the 30-dimensional CEC2013 test set [/fig_ref] respectively show the experimental results of each algorithm for solving the 30-and 100-dimensional function problems on the CEC2013 test set, with in-and out-parenthesis values being the standard deviation and mean of the optimal values obtained in 30 independent experiments. Among them, the bold data indicates that the corresponding algorithm has obtained the best solution effect on the test function, and the last line counts the number of functions for obtaining the best performance of each algorithm. To further compare the performance differences of the above algorithms, a Friedman test with a significant level of 0.05 was performed on the above data, and the specific results are shown in [fig_ref] Table 7: Results of the Wilcoxon rank sum test by IGTOA with other algorithms [/fig_ref]. Among them, the size of the p-value reflects the difference between the two algorithms in the current function. When the p-value is less than 0.05, it indicates that the IGTOA and the corresponding comparison algorithm have obvious differences in the current function, while otherwise, there is no significant difference between the two algorithms. In addition, the "+/=/−" in the last row indicates the number of functions that the IGTOA algorithm is significantly better than, not significantly different from, and significantly inferior to the corresponding comparison algorithm, respectively. As can be seen from the data in [fig_ref] Table 5: Results of each algorithm on the 30-dimensional CEC2013 test set [/fig_ref] www.nature.com/scientificreports/ show that IGTOA has obvious advantages in convergence accuracy over the remaining five algorithms. In addition, as the dimension of the optimization problem increases, the advantages of the IGTOA algorithm are also greater. To further compare the comprehensive performance of each algorithm on all functions, [fig_ref] Table 8: The Friedman detection results for each algorithm [/fig_ref] presents the results of Friedman detection. As can be seen from the Friedman test results in [fig_ref] Table 8: The Friedman detection results for each algorithm [/fig_ref] , the rank of IGTOA is significantly lower than the other five methods, indicating that the IGTOA algorithm performs best in terms of convergence accuracy.For 30-dimensional function optimization problems, the comprehensive performance of each algorithm is IGTOA > GTOA > IATTP > ADN-RSN-PSO > MSMPSO > ESCA; for 100-dimensional function optimization problems, the comprehensive performance of each algorithm is IGTOA > IATTP > GTOA > ADN-RSN-PSO > MSMPSO > ESCA.
Comparative test of the convergence rate of the algorithm. In order to compare the convergence rate of the algorithm more intuitively, [fig_ref] Figure 8: Convergence curves of each algorithm on the test suit [/fig_ref] gives the iterative process curve where each algorithm is run randomly once when the test function dimension is 30.The horizontal and vertical coordinates represent the logarithm of the function evaluation times and the fitness values, respectively.Parameter settings for each algorithm are performed as in .
As can be seen in [fig_ref] Figure 8: Convergence curves of each algorithm on the test suit [/fig_ref]. For the functions F1, F3, F5, F7, F9, F11, F12, F13, and F20, the IGTOA all converge to the theoretical optimal results; the ESCA converges to the theoretical optimum on F3, F9, F11, and F13; the IATTP algorithm converges to the theoretical optimum on F3, F7, F9, F11, F12, and F13; the GTOA converges to the theoretical optimum on the F3, F7, F9, F11, and F13; the ADN-RSN-PSO converges to the theoretical optimum only on the F11; while MSMPSO does not obtain the theoretical optimal results on any function. Compared with IGTOA, GTOA and IATTP showed faster convergence on F9, F11 and F13, while ESCA converged only faster on F11, while other algorithms converged slower on the remaining functions, including F1, F3, F5, F7, F12 and F20. For the remaining 19 functions, each algorithm converged to the local optima, including F2, F4, F6, F8, F10, F14-F19 and F21-F28. For F2, F14, F19 and F21, IGTOA only converged slightly slower than GTOA in the early evolution, but all faster than the other four algorithms, especially in the late evolution, IGTOA showed faster convergence than the other five algorithms. For F4, IGTOA converges slower than IATTP, but faster than the other 4 contrast algorithms. For F15, F23 and F26, the IGTOA showed the fastest convergence compared with the other five algorithms, whereas the IGTOA decreased.Later in evolution, IGTOA converged only faster on F15 than IATTP, slower on F23 than GTOA and ADN-RSN-PSO, and only slightly slower than MSMPSO on F26. For F16, IGTOA converges only converged slightly slower than MSMPSO in early evolution, but by later evolution, IGTOA showed the fastest convergence. For F24, IGTOA showed the fastest convergence rate in the early evolution, slowing down as evolution progressed and being comparable to that of IATTP and GTOA. For F27, IGTOA converged slightly slower than MSMPSO and IATTP in the early evolution, the convergence of each algorithm decreased, but ADN-RSN-PSO decreased more slowly, and by the later evolution ADN-RSN-PSO, MSMPSO and IATTP all converged faster than IGTOA. But for other functions, including F6, F10, F17, F18, F22, F25, and F28, IGTOA showed the fastest convergence rate compared to the other five evolutionary algorithms. In Conclusion Section, IGTOA has some advantages in convergence speed over the other five algorithms.
Comparative test of the algorithm stability. To intuitively compare the stability of each algorithm, we draw the box plot of the optimal results obtained from 30 independent runs of each algorithm.Limited to space, this section selects only nine different types of functions for comparison, including: F1, F2 and F5 in uni-modal functions; F6, F14 and F16 in multi-modal functions; and F22, F25 and F28 in combined functions. As shown in [fig_ref] Figure 9: Box plots of the 6 algorithms on the test function [/fig_ref]. www.nature.com/scientificreports/ [fig_ref] Figure 9: Box plots of the 6 algorithms on the test function [/fig_ref] follows from the fact that for the uni-modal functions F2 and F5 and the multi-modal functions F22 and F14, IGTOA is more stable than the other five algorithms.For the uni-modal function F1, the stability of IGTOA and IATTP was flat, and both significantly outperformed the other four algorithms, including GTOA, MSMPSO, ADN-RSN-PSO, and ESCA.For multi-modal function F16, IGTOA is only slightly less stable than GTOA; for combined function F25, IGTOA is slightly less stable than ESCA, but its solution accuracy is [fig_ref] Table 9: Comparison of time complexity between IGTOA and other algorithms [/fig_ref]. The dimension of the test function is 30, the maximum number of evaluation is Max_FEs = 5000*D, and the other parameter settings are shown in .
As seen from [fig_ref] Table 9: Comparison of time complexity between IGTOA and other algorithms [/fig_ref] , for the unimodal function F1-F20, I GTOA runs for slightly longer times compared to GTOA, MSMPSO, ADN-RSN-PSO, and ESCA. For the combined functions F21-F28, the running time of each algorithm is not very different. However, the running time of the algorithms is not much different, which means that the time complexity of I GTOA is slightly higher compared with G T O A and other contrast algorithms. This is due to the multiple improvement strategies employed by IGTOA, requiring more manipulation when looking for better individuals. Combined with the convergence rate, with the same convergence accuracy, IGTOA does not increase compared with the other algorithms. www.nature.com/scientificreports/ Comparison of the engineering application effect. In order to further compare the effects of IGTOA algorithm and other comparison algorithms in practical application, this section uses each algorithm to handle the cooperative beam forming optimization problem. The cooperative beam forming optimization problem is a typical problem in the antenna array. By optimizing the amplitude and phase of the emission signal weight of each cooperative node, the peak side valve level PSL minimization as shown in formula (17) is realized. www.nature.com/scientificreports/ where, AF(θ, w) represents the array factor, as shown in formula [bib_ref] An efficient double adaptive random spare reinforced whale optimization algorithm, Chen [/bib_ref]. φ is the main beam direction. θ SL is the direction corresponding to the peak point in the range θ ∈ [−π, φ) ∪ (φ, π ] beam chart except for the main lobe peak point, is called the lateral lobe direction. The denominator AF(φ,w) is the main beam power and the molecule max |AF(θ SL , w)| is the maximum beam power in the side flap.
where, w k is the complex number weight coefficient of the signal emitted by the k-th cooperative node, as shown in formula [bib_ref] An improved artificial tree algorithm with two populations (IATTP), Xiao [/bib_ref]. www.nature.com/scientificreports/ www.nature.com/scientificreports/ where, ξ k and α k are the amplitude and initial phase of the emission signal weights of the k-th cooperative node, respectively, and ξ ∈ [0, 1] , α ∈ [−π , π ]. The beam forming scenario in this section is shown in . Among them, the wavelength of the sending signal is , and the six cooperative nodes are distributed in the circle domain with a radius of 4 , and one cooperative node is located in the center of the circle domain. Each algorithm is optimized as the objective function shown in formula [bib_ref] An improved cuckoo search algorithm with self-adaptive knowledge learning, Li [/bib_ref]. For comparative fairness, in this experiment, the problem scale N is 50, the maximum function evaluation times Max_FEs = 5000*D, the number of nodes k = 6, the polar radius r k = 4, and other parameters are shown in . In order to avoid the adverse effects of contingency on the algorithm evaluation, each algorithm runs independently for 10 times, and selects the best collaborative beam optimization scheme corresponding to the PSL median of each algorithm is compared. [fig_ref] Figure 11: Beam map of IGTOA and each contrast algorithm in a rectangular coordinate... [/fig_ref] intuitively gives the beam diagram of IGTOA and each comparison algorithm in the right Angle coordinate system, and then the PSL corresponding to each algorithm is annotated in the graph.
As can be seen from [fig_ref] Figure 11: Beam map of IGTOA and each contrast algorithm in a rectangular coordinate... [/fig_ref] , for the above collaborative beam forming scenario, the best PSL obtained from GTOA, IATTP, MSMPSO, ADN-RSN-PSO, ESCA, and IGTOA are: − 3.7667 dB, − 4.0599 dB, − 3.1973 dB, − 4.0369 dB, − 3.7997 dB and − 4.3917 dB, respectively. Each algorithm achieved better cooperative beam optimization than unoptimized (− 1.8909 dB), and IGTOA achieved the best synergistic beam optimization than the other five algorithms. In Conclusion Section, the proposed IGTOA also has excellent performance in engineering applications.
# Conclusion
This paper proposes an improved algorithm-IGTOA, which assigns teachers by probability and introduces different excellent genes in the group to ensure the population diversity; at the same time, the adaptive grouping method, combined with the different learning abilities of students in the two groups, put forward suitable search methods and learning methods, balancing the diversity loss rate in the evolution process and the algorithm convergence rate; in addition, this paper proposes a population reconstruction mechanism that starts with whether the population optimal individual has continuous changes and provides new genes for the population while maintaining excellent genes, which ensures the convergence rate of the algorithm and better maintains the population diversity; finally, simulation results from multiple experiments of this algorithm in the CEC2013 test suite show that IGTOA has good comprehensive performance, and IGTOA has obvious advantages in convergence speed and solution accuracy compared with many other comparative algorithms.
## Data availability
The datasets used or analysed during the current study available from the corresponding author on reasonable request.
[fig] Figure 2: Individual number of students in the two groups changed. Scientific Reports | (2022) 12:11267 | https://doi.org/10.1038/s41598-022-15170-1 [/fig]
[fig] Figure 3: Dimension selection in subspace search mode. Scientific Reports | (2022) 12:11267 | https://doi.org/10.1038/s41598-022-15170-1 [/fig]
[fig] Figure 4: Subspatial learning mode of teacher teaching stage of average group individual X A2 . Scientific Reports | (2022) 12:11267 | https://doi.org/10.1038/s41598-022-15170-1www.nature.com/scientificreports/ individual only in multiple local optimal point exploration of smaller scope, it is easier to search the global optimal peak of the narrow scope. In addition, references30 and 31 have known that the reverse learning strategy of replacing the original candidate solution by the relative points of the candidate solution is a better estimation of the original candidate solution, and generally achieves better optimization results compared with the method of replacing the original candidate solution by random points. Based on the above reasons, a new population reconstruction method is proposed in this section, as shown in Fig. 5, as follows: Firstly, the population was divided into three parts according to fitness, including pop1, pop2 and pop3, and the numbers of sub-populations were N1 = 0.1 N, N2 = 0.4 N and N3 = 0.5 N, respectively. Then, the individuals in the sub-population pop1 are retained directly as pop1' , individuals in the sub-population of pop2 randomly choose the dimensions to learn backwards to form the pop2' according to the Eq. (15), and the individuals in the population pop3 are combined into new individuals by randomly selecting different dimensions of the individuals in pop1 according to formula (16), thereby forming pop3' . Finally, sub-populations pop1' , pop2' and pop3' were merged to form a new population, newpop, to participate in the next iteration. It should be noted that the reconstructed population does not preferentially retain the original population, but directly participate in the evolution of the next generation. [/fig]
[fig] Figure 5: Population reconstruction mechanism. [/fig]
[fig] 2: In the new improvement strategy, the number of people in the average group should be greater than 2, so the minimum Pg should be 0.06. When examining the effect of Pg on the performance of the IGTOA algorithm, Pg was set to 0.06, 0.1, 0.4, 0.7, and 0.9, respectively. Other parameters are set as follows: change_flag = 30, l = 10, N1 = N2 = 0.2*N. (3) When examining the effect of l on the performance of the IGTOA algorithm, l was set to 5, 10, 20, 30, and 50, respectively. Other parameters are set as follows: change_flag = 30, Pg = 0.1, N1 = N2 = 0.2*N. (4) When examining the effect of N1 and N2 on the performance of IGTOA algorithm, N1 and N2 are set to the following five values: N. Other parameters are set as follows: change_flag = 30, Pg = 0.1, l = 10. [/fig]
[fig] Figure 6: Comparison of individual distribution before and after population reconstruction. Scientific Reports | (2022) 12:11267 | https://doi.org/10.1038/s41598-022-15170-1 [/fig]
[fig] Figure 7: Results of the Friedman test for each parameter. Scientific Reports | (2022) 12:11267 | https://doi.org/10.1038/s41598-022-15170-1 [/fig]
[fig] Table 4: Initial parameters setting of each algorithm.AlgorithmInitial parametersIGTOA change_flag = 30, Pg = 0.1, l = 10, N1 = N2 = 0.2 N GTOA b = rand(0,1), c = 1-b IATTP h1 = h2 = h3 =0.5; h4 = 0.8; m = 50, q = 0.8 MSMPSO pop1: c1 = 2.0, c2 = 1.0, c3 = 0.2; pop2: c1 = 0.1, c2 = 1.0, c3 = 2.0; pop3: c1 = c2 = c3 = 1.0; cycle = 10 ADN-RSN-PSO α 1 = α 2 = 2.05, n s = 2, r g = 4, s r = 0.6, ρ = 0.4, P AND = 0.2, L 0 = 0.35*range, L min = 10 -8 *range, x = 0.7298ESCA Pc = 0.6, r2 = 2*pi*rand(0,1), r3 = 2*rand(0,1), r4 = r5 = r6 = r7 = r8 = rand( [/fig]
[fig] Figure 8: Convergence curves of each algorithm on the test suit. Scientific Reports | (2022) 12:11267 | https://doi.org/10.1038/s41598-022-15170-1 [/fig]
[fig] 17: PSL = 20 log 10 max |AF(θ SL , w)| AF(φ, w) (18) AF(θ, w) = k k=1 w k e j(2π/ )r k [cos (θ −ψ k )] [/fig]
[fig] Figure 9: Box plots of the 6 algorithms on the test function. Scientific Reports | (2022) 12:11267 | https://doi.org/10.1038/s41598-022-15170-1 [/fig]
[fig] Figure 11: Beam map of IGTOA and each contrast algorithm in a rectangular coordinate frame. Scientific Reports | (2022) 12:11267 | https://doi.org/10.1038/s41598-022-15170-1 [/fig]
[table] Table 1: Effect of parameters change_flag and Pg on IGTOA. Note: The parameters that perform best on each function are bolded. [/table]
[table] Table 3: Comparison results of each improvement strategy with IGTOA on the 30-dimensional CEC2013 test suite. Note: The parameters that perform best on each function are bolded. [/table]
[table] Table 5: Results of each algorithm on the 30-dimensional CEC2013 test set. Note: The parameters that perform best on each function are bolded. www.nature.com/scientificreports/ 30 times under the same number of function evaluation times. The specific results are shown in [/table]
[table] Table 7: Results of the Wilcoxon rank sum test by IGTOA with other algorithms. [/table]
[table] Table 8: The Friedman detection results for each algorithm. Scientific Reports | (2022) 12:11267 | https://doi.org/10.1038/s41598-022-15170-1 [/table]
[table] Table 9: Comparison of time complexity between IGTOA and other algorithms.Figure 10. Distribution of the cooperative nodes. https://doi.org/10.1038/s41598-022-15170-1www.nature.com/scientificreports/(19) w k = ξ k e jα k [/table]
|
Long‐lasting response to third‐line crizotinib treatment in a patient with non‐small cell lung cancer with brain metastases and poor performance status
Identification of the anaplastic lymphoma kinase (ALK) gene has refined the classification of non-small cell lung cancer (NSCLC) and promoted research on molecularly targeted drugs such as crizotinib, an ALK inhibitor with good efficacy, in ALK-rearranged NSCLC. At present, few studies have reported the efficacy of crizotinib in patients with ALK-rearranged NSCLC with brain metastases. In a patient with NSCLC harboring ALK-rearrangement who had brain metastases and poor performance status (PS), we obtained a durable response with crizotinib administered following multi-line chemotherapy regimens.
# Introduction
Anaplastic lymphoma kinase (ALK) rearrangement is identified in approximately 5% of patients with nonsmall cell lung cancer (NSCLC), and these patients have been found to respond well to crizotinib, an oral ATPcompetitive inhibitor of ALK. [bib_ref] Anaplastic lymphoma kinase inhibition in non-small-cell lung cancer, Kwak [/bib_ref] [bib_ref] Activity and safety of crizotinib in patients with ALK-positive non-small-cell lung cancer:..., Camidge [/bib_ref] In the recent PROFILE clinical trials, a promising intracranial disease control rate (I-DCR) and median intracranial time to progression (MI-TTP) were reported with crizotinib in patients with ALK-rearranged NSCLC with brain metastases. [bib_ref] Clinical experience with crizotinib in patients with advanced ALK-rearranged nonsmall-cell lung cancer..., Costa [/bib_ref] [bib_ref] Intracranial efficacy of crizotinib versus chemotherapy in patients with advanced ALKpositive non-small-cell..., Solomon [/bib_ref] In this report, we document the response to third-line crizotinib therapy in a patient with ALK-rearranged NSCLC with brain metastases and poor performance status (PS) who had previously received multi-line chemotherapy.
## Case report
A 44-year-old female patient with no history of smoking presented with right chest pain and a cough in June 2012. A chest computed tomography (CT) scan revealed a mass with a maximum diameter of 4.5 cm in the right upper lung and right pleural effusion, but an upper abdominal CT scan and bone emission CT revealed no abnormalities. Magnetic resonance imaging (MRI) revealed multiple nodules in the brain with a maximum diameter of 1 cm, but no peripheral edema. Biopsy of the pleura confirmed a poorly differentiated lung adenocarcinoma. An epidermal growth factor receptor (EGFR) gene mutation test was negative. The patient was diagnosed with NSCLC with brain metastases (cT4N0M1, stage IV). Her Eastern Cooperative Oncology Group PS was 0.
Initially, the patient received first-line chemotherapy with six cycles of pemetrexed and cisplatin (AP). A partial response (PR) was achieved in the lung (shrinkage of the pulmonary tumor and pleural effusion) and a complete response (CR) was achieved in the brain, with elimination of the metastases. The patient subsequently received maintenance treatment with pemetrexed and palliative radiotherapy for the right upper lung lesion. However, after six cycles of pemetrexed maintenance treatment, the disease progressed. A recurrent nodule was detected by brain MRI in the left frontal lobe (maximum diameter, 3 mm), and bilateral lung and liver metastases (multiple small bilateral lung lesions with maximum diameters of 5 mm and single liver nodules with maximum diameters of 3 cm) were detected by an 18 F-labeled fluorodeoxyglucose-positron emission tomography-CT scan. As her PS was adequate (PS = 1), the patient received second-line chemotherapy with docetaxel and cisplatin (DP), which was administered for four cycles. In addition, she was treated with γ-knife radiotherapy for her recurrent brain lesion. Following two cycles of DP therapy, tumor assessments revealed stable disease (SD). However, CT and MRI scans after four cycles of DP showed enlargement of the bilateral lung tumors and brain lesions (maximum diameters, 4 and 6 mm, respectively; [fig_ref] Figure 1: Prior to the administration of crizotinib, [/fig_ref].
Around this time, the patient developed severe anemia, fatigue, and neurotoxicity as a result of the platinum-based chemotherapy, and her PS worsened (PS = 2). Because of the stressful nature of further biopsies of the small lesions, the original pleural sample was used for detection of ALK rearrangement by fluorescence in-situ hybridization testing. As the tumor was positive for ALK rearrangement, the patient was started on crizotinib 250 mg orally twice daily in June 2014. Two months after the initiation of crizotinib therapy, CT and MRI examinations revealed a decrease in the size of the bilateral lung metastases, and the brain and liver metastases had disappeared . After a year of crizotinib therapy, a brain MRI showed two new nodules in the left temporal lobe (maximum diameter, 3 mm) and the right frontal lobe (maximum diameter, 2.5 cm), but the pulmonary lesions remained stable. X-knife radiotherapy was then administered to control the two new brain metastatic lesions.
By June 2016, the patient had received crizotinib treatment for 24 months without any evidence of disease progression. Although she experienced a slight blurring of vision, fatigue, and nausea during the first month of crizotinib therapy, these symptoms gradually disappeared after a month.
# Discussion
Approximately 10-20% of patients with NSCLC are found to have brain metastases at the time of initial diagnosis, and several retrospective studies have reported that 20-30% of patients with brain metastases have ALKrearranged NSCLC. [bib_ref] Brain metastasis development and poor survival associated with carcinoembryonic antigen (CEA) level..., Arrieta [/bib_ref] [bib_ref] Three-arm, randomized, phase 2 study of carboplatin and paclitaxel in combination with..., Hanna [/bib_ref] [bib_ref] Brain metastases in patients with EGFR-mutated or ALKrearranged non-small-cell lung cancers, Rangachari [/bib_ref] [bib_ref] Oncogene status predicts patterns of metastatic spread in treatment-naive nonsmall cell lung..., Doebele [/bib_ref] Because of the blood-brain barrier (BBB), antineoplastic drugs commonly exhibit poor penetration into the brain; therefore, the central nervous system (CNS) has been considered a sanctuary for many types of cancer. However, studies have shown that the BBB can be disrupted by brain metastases, and targeted therapies, such as EGFR and ALK tyrosine kinase inhibitors (TKIs), have shown great potential in treating brain metastases. [bib_ref] Clinical experience with crizotinib in patients with advanced ALK-rearranged nonsmall-cell lung cancer..., Costa [/bib_ref] [bib_ref] Intracranial efficacy of crizotinib versus chemotherapy in patients with advanced ALKpositive non-small-cell..., Solomon [/bib_ref] [bib_ref] The seed and soil hypothesis: Vascularisation and brain metastases, Fidler [/bib_ref] [bib_ref] Brain metastases from lung cancer responding to erlotinib: The importance of EGFR..., Porta [/bib_ref] [bib_ref] Erlotinib as second-line treatment in patients with advanced non-small-cell lung cancer and..., Wu [/bib_ref] Crizotinib is a first-generation ALK inhibitor approved by the United States Food and Drug Administration because of its effectiveness in the treatment of ALKrearranged NSCLC. [bib_ref] Anaplastic lymphoma kinase inhibition in non-small-cell lung cancer, Kwak [/bib_ref] Recently, some studies have focused on the CNS activity of crizotinib in NSCLC patients with brain metastases. [bib_ref] Clinical experience with crizotinib in patients with advanced ALK-rearranged nonsmall-cell lung cancer..., Costa [/bib_ref] [bib_ref] Intracranial efficacy of crizotinib versus chemotherapy in patients with advanced ALKpositive non-small-cell..., Solomon [/bib_ref] A pooled retrospective study analyzed the clinical benefits of crizotinib in ALK-rearranged NSCLC with brain metastases after failure of at least one prior chemotherapy regimen in patients who had participated in the PROFILE 1005 and PROFILE 1007 clinical trials. [bib_ref] Clinical experience with crizotinib in patients with advanced ALK-rearranged nonsmall-cell lung cancer..., Costa [/bib_ref] The 12-week I-DCR with crizotinib in patients with previously untreated asymptomatic brain metastases was 56%, compared with 62% in patients with previously treated brain metastases, and the MI-TTP were 7 and 13.2 months, respectively. [bib_ref] Clinical experience with crizotinib in patients with advanced ALK-rearranged nonsmall-cell lung cancer..., Costa [/bib_ref] In a prospective study (PROFILE 1014) that compared first-line crizotinib with pemetrexed plus cisplatin or carboplatin chemotherapy in ALK-rearranged NSCLC, a significantly higher I-DCR at 12 weeks was achieved with crizotinib in patients with stable treated brain metastases (85% vs. 45%, respectively; P < 0.001), and the median progression-free survival (PFS) was also significantly longer with crizotinib (9 vs. 4 months, respectively; P < 0.001). [bib_ref] Intracranial efficacy of crizotinib versus chemotherapy in patients with advanced ALKpositive non-small-cell..., Solomon [/bib_ref] In our patient, thirdline therapy with crizotinib achieved intracranial progression-free survival (I-PFS) of 12 months, and an additional 12 months of brain lesion control was achieved when X-knife radiotherapy was performed, together with continued crizotinib therapy for the new lesions. As crizotinib has demonstrated good efficacy in the subset of ALK-rearranged NSCLC patients with brain metastases and is relatively well tolerated, patients with a poor PS score should be given the opportunity of treatment with crizotinib. Because the patient we treated showed sensitivity to the first-line therapy administered, it may be that the efficacy of crizotinib is better in chemotherapysensitive patients. This has been reported in EGFR-TKI therapy, and could be one of the reasons for the long PFS. [bib_ref] Erlotinib as maintenance treatment in advanced non-small-cell lung cancer: A multicentre, randomised,..., Cappuzzo [/bib_ref] Although isolated CNS progression occurred in our patient after 12 months of crizotinib treatment, no progression was found in her extracranial disease. Two new nodules were detected in the brain but the previously identified brain metastases were stable. The likely reason for this may be poor drug penetration in the nonmetastatic brain area. 14,15 Isolated CNS relapse is a common disease progression pattern in patients with ALKrearranged NSCLC who are treated with crizotinib. [bib_ref] Clinical experience with crizotinib in patients with advanced ALK-rearranged nonsmall-cell lung cancer..., Costa [/bib_ref] In a retrospective study, Takeda et al. analyzed the clinical impact of continuing crizotinib treatment after radiotherapy for isolated CNS progression in patients with ALKrearranged NSCLC. [bib_ref] Clinical impact of continued crizotinib administration after isolated central nervous system progression..., Takeda [/bib_ref] In addition to controlling the extracranial lesions, the brain metastases were controlled for another 5.5 months. [bib_ref] Clinical impact of continued crizotinib administration after isolated central nervous system progression..., Takeda [/bib_ref] In our patient, isolated CNS progression occurred after one year of crizotinib treatment. However, continued crizotinib therapy combined with irradiation of the brain lesion achieved long-lasting survival. Two months after beginning crizotinib therapy, (a,b) computed tomography and (c,d) magnetic resonance imaging scans showed a decrease in the size of the bilateral lung metastases, and the brain metastases had disappeared.
Continuous EGFR-TKI treatment combined with local treatment has been proven an effective strategy for patients with locally advanced NSCLC. [bib_ref] Local ablative therapy of oligoprogressive disease prolongs disease control by tyrosine kinase..., Weickhardt [/bib_ref] [bib_ref] Continuous EGFR-TKI administration following radiotherapy for non-small cell lung cancer patients with..., Shukuya [/bib_ref] Our findings indicate that similar benefits can be obtained with continuous ALKinhibitor treatment for locally advanced disease. [bib_ref] Clinical impact of continued crizotinib administration after isolated central nervous system progression..., Takeda [/bib_ref]
[fig] Figure 1: Prior to the administration of crizotinib, (a,b) computed tomography and (c,d) magnetic resonance imaging scans revealed enlargement of the bilateral lung tumors and brain lesions (maximum diameters, 4 and 6 mm, respectively). [/fig]
|
Investigating the utility of clinical outcome-guided mutual information network in network-based Cox regression
Background: Network-based approaches have recently gained considerable popularity in high-dimensional regression settings. For example, the Cox regression model is widely used in expression analysis to predict the survival of patients. However, as the number of genes becomes substantially larger than the number of samples, the traditional Cox or L 2 -regularized Cox models are still prone to noise and produce unreliable estimations of regression coefficients. A recent approach called the network-based Cox (Net-Cox) model attempts to resolve this issue by incorporating prior gene network information into the Cox regression. The Net-Cox model has shown to outperform the models that do not use this network information.Results: In this study, we demonstrate an alternative network construction method for the outcome-guided gene interaction network, and we investigate its utility in survival analysis using Net-Cox regression as compared with conventional networks, such as co-expression or static networks obtained from the existing knowledgebase. Our network edges consist of gene pairs that are significantly associated with the clinical outcome. We measure the strength of this association using mutual information between the gene pair and the clinical outcome. We applied this approach to ovarian cancer patients' data in The Cancer Genome Atlas (TCGA) and compared the predictive performance of the proposed approach with those that use other types of networks. Conclusions: We found that the alternative outcome-guided mutual information network further improved the prediction power of the network-based Cox regression. We expect that a modification of the network regularization term in the Net-Cox model could further improve its prediction power because the properties of our network edges are not optimally reflected in its current form.
# Background
The Cox regression model [bib_ref] Regression Models and Life-Tables, Cox [/bib_ref] has been commonly used in survival analysis to detect important features and to predict patient survival. Due to the advance of sequencing technology, the number of genes or features in these analyses is becoming substantially larger than the number of samples. Although several variant models, such as L 1 −regularization [bib_ref] The lasso method for variable selection in the Cox model, Tibshirani [/bib_ref] or L 2 −regularization in Hilbert space [bib_ref] Kernel Cox regression models for linking gene expression profiles to censored survival..., Li [/bib_ref] , have been proposed to resolve this issue, those variant models are still prone to noise and over-fitting [bib_ref] Penalized Cox regression analysis in the high-dimensional and low-sample size settings, with..., Gui [/bib_ref].
Recently, network-based approaches have gained considerable popularity for expression quantitative trait loci (eQTL) studies and clinical outcome predictions in highdimensional regression settings. These approaches incorporate prior network information of either the features [bib_ref] Network-constrained regularization and variable selection for analysis of genomic data, Li [/bib_ref] or outcomes [bib_ref] A multivariate regression approach to association analysis of a quantitative trait network, Kim [/bib_ref] , or both [bib_ref] Graph-regularized dual Lasso for robust eQTL mapping, Cheng [/bib_ref] [bib_ref] Leveraging input and output structures for joint mapping of epistatic and marginal..., Lee [/bib_ref]. For survival analysis, Zhang et al. [bib_ref] Network-based survival analysis reveals subnetwork signatures for predicting outcomes of ovarian cancer..., Zhang [/bib_ref] recently proposed a variant of the L 2 −regularized Cox model called the Net-Cox model, which is a networkregularized Cox regression model. The typical types of networks used in such approaches are either coexpression networks, which are constructed by computing the correlation between every pair of variables, or static networks, such as the protein-protein interaction (PPI) network, which can be obtained from the existing knowledgebase. In the Net-Cox model, co-expression and functional linkage networks were incorporated in the survival analysis, and the results showed enhanced performance when compared with the conventional methods, which do not use the network information.
A potential limitation of these conventional networks is that the edges only reflect the information of withinfeatures or within-outcomes relations, and do not consider the association between features and outcomes, which may be useful in improving the predictive power. In this study, we show that the outcome-guided mutual information network improves the performance of survival analysis in the Net-Cox regression [bib_ref] Network-based survival analysis reveals subnetwork signatures for predicting outcomes of ovarian cancer..., Zhang [/bib_ref]. We demonstrate the utility of this outcome-guided gene network with the analysis of a TCGA ovarian cancer dataset and compare its performance with those of survival analyses that use other types of networks.
# Methods
## Background on survival analysis using cox regression
First, we describe the basic formulation of Cox regression [bib_ref] Regression Models and Life-Tables, Cox [/bib_ref] for survival analysis using expression and survival data. Given a gene expression profile X, which consists of n patients and p genes, the risk of an event at time t for the ith patient with gene expression
[formula] X i = (X i1 ,...,X ip )′ is defined as h (t|X i ) = h 0 (t)e X i β , where b = (b 1 ,...,b p ) [/formula]
′ is a regression coefficient vector, and h 0 (t) denotes the baseline hazard function at t. The coefficient b and the function h 0 are generally unknown and need to be estimated. In a traditional Cox regression model, the estimation of b is based on the maximization of the partial log-likelihood, pl(b):
[formula] pl(β) = n i=1 δ i X i β − log j∈R(t i ) e X j β . [/formula]
Here, δ i is the observed status (δ i = 1 implies observed; δ i = 0 implies censored), t i represents the event time of the ith patient, and R(t i ) is the subset of patients who survived to time t i . Once the coefficient vectorβ is obtained, a Breslow estimator can estimate the baseline hazard function h 0 as follows:
[formula] h 0 (t i ) = 1 j∈R(t i ) e X jβ . [/formula]
## Regularized cox regression in high-dimensional setting
When p ≫ 1, the Cox regression model is prone to noise and tends to produce unreliable estimations of regression coefficients. Several solutions that shrink the coefficients have been proposed. A common solution is the L 2 −Cox model, which uses the penalized total log-likelihood:
[formula] l pen (β, h 0 ) = n i=1 −e X jB H 0 (t i ) + δ i log h 0 (t i ) + X β − 1 2 λ p j=1 β 2 j , [/formula]
where λ p j=1 β 2 j is the regularization term and l controls the amount of shrinkage. Net-Cox regression is an extension of the L 2 −Cox model and uses the following penalized log-likelihood:
[formula] l pen (β, h 0 ) = n i=1 −e X jB H 0 (t i ) + δ i log h 0 (t i ) + X β − 1 2 λβ β, [/formula]
in which lb′Γb is the penalty term and Γ = (1 -α) L + aI conveys prior network information between genes. Here, L = I − S, where I is the identity matrix and S is a normalized Laplacian matrix, which comes from the gene network.
The parameter α ∈ (0,1] controls the contribution of network information to the model. When α is small, more network information is incorporated into the penalty term than when α is large. We note that when α = 1, the network-based cox regression model reduces to the L 2 −Cox model that does not use any network information.
The typical network used for this type of networkbased approach is the coexpression network, in which the edge weights correspond to the correlation between expression vectors of two genes [bib_ref] A multivariate regression approach to association analysis of a quantitative trait network, Kim [/bib_ref]. In [bib_ref] Network-based survival analysis reveals subnetwork signatures for predicting outcomes of ovarian cancer..., Zhang [/bib_ref] , a functional linkage network was constructed from the existing knowledgebase. We note that both types of networks do not reflect information about the relation between the features and the outcomes inferred from a given dataset. Therefore, we consider an alternative approach of constructing and utilizing outcome-guided mutual information network in survival analysis. The detailed procedures are explained in the following sections.
## Gene interaction networks associated with clinical outcome
Mutual information is an information-theoretic measure that reveals the dependency or association between two random variables [bib_ref] The mutual information: detecting and evaluating dependencies between variables, Steuer [/bib_ref] and is defined as follows:
[formula] I(X; Y) = H(X) + H(Y) − H(X, Y), [/formula]
where H(X) and H(Y) denote the entropies of the respective variables, X and Y, and H(X, Y) denotes the joint entropy of X and Y. It can be used to detect both linear and nonlinear relations between two random variables [bib_ref] An approximation to the distribution of finite sample size mutual information estimates, Goebel [/bib_ref]. While many previous studies generally used mutual information to detect dependency between two features of the same type, more recent approaches in [bib_ref] Fast detection of high-order epistatic interactions in genome-wide association studies using information..., Leem [/bib_ref] [bib_ref] Characterizing genetic interactions in human disease association studies using statistical epistasis networks, Hu [/bib_ref] have also extended the mutual information to assess the association between expressions of two genomic features, X 1 and X 2 , and the clinical outcome Y as follows:
[formula] I(X 1 , X 2 ; Y) = H(X 1 , X 2 ) + H(Y) − H(X 1 , X 2 , Y). [/formula]
Note that genomic feature values, such as gene expressions or clinical outcome values, are often numeric. To compute the mutual information, we discretize the genomic feature values using a histogrambased technique as in [bib_ref] Mutual information relevance networks: functional genomic clustering using pairwise entropy measurements, Butte [/bib_ref]. In the case of survival data, a simple thresholding scheme converts the outcome variable of the survival month to a binary variable. In the case of the TCGA dataset used in this study, we split the patients into short-term living (≤ 3 years) and longterm living (> 3 years). Patients who are reported as living and have an overall survival time less than 3 years are filtered out of this study.
Computation of mutual information for every pair of genes and the clinical outcome variable produce a complete network between genes. We can further filter out less significant edges by using the permutation testing scheme proposed in [bib_ref] Mutual information relevance networks: functional genomic clustering using pairwise entropy measurements, Butte [/bib_ref]. Through the repeated permutation of clinical outcome labels and the re-computation of mutual information values, a threshold θ is defined as the maximum of the average mutual information values for each gene pair. This θ can be used as a base threshold to remove insignificant edges. However, this threshold still leaves a large number of edges that may not be fully beneficial to the downstream analysis. In this study, another parameter s, which amplifies the significance level by a factor of (1 + s), was proposed to create a stricter cut-off. In particular, the filtered mutual information network with significance level s is defined by
[formula] G σ = g i , g j g i , g j ∈ P and I g i , g j ; Y ≥ θ (1 + σ ) , [/formula]
where P is the set of all genes, Y denotes the binary survival status of patients, and I(g i , g j ; Y) is the mutual information of the gene pair (g i , g j ) and Y. We adopt this thresholding scheme and test the performance of different parameter values. In order to apply the network to the Net-Cox model, we also normalize G s and create a normalized Laplacian matrix
[formula] S = R − 1 2 G σ C − 1 2 , where R ii = j G σ ij and C ii = j G σ ij both of which are diagonal matrices. [/formula]
## Dataset
We use genomic profiles and clinical outcome data from patients with ovarian serous cystadenocarcinoma in TCGA. Genomic profiles of copy number alterations (CNAs), messenger RNA (mRNA), and methylation (METH) are used in the experiments. We remove genes or patients with missing values and extract genes that are common to all three profiles. As a result, our data matrix consists of expressions and alterations of 10,022 genes and 340 patients for each of the three profiles.
## Evaluation measure
We compare the performance of the Net-Cox model using the proposed mutual information networks with those of models using a gene co-expression and gene functional linkage network as seen in [bib_ref] Network-based survival analysis reveals subnetwork signatures for predicting outcomes of ovarian cancer..., Zhang [/bib_ref]. In order to create a baseline, an analysis of the L 2 − Cox model, which does not use any network information, is performed. This is the same as setting α = 1 in the Net-Cox model.
With each run of the Net-Cox analysis, the prognostic indices PI = X β are calculated as prediction markers, where X is the test data not used in training, and the regression coefficientβ is estimated from given training data, network information, and the parameters. We evaluate the prediction performance using a time-dependent area under the curve (AUC) [bib_ref] Time-dependent ROC curves for censored survival data and a diagnostic marker, Heagerty [/bib_ref].
Specifically, time-dependent sensitivity and specificity functions are defined as follows:
[formula] sensitivity[c, t|f (X)] = Pr{f (X) > c|δ(t) = 1}, specificity[c, t|f (X)] = Pr{f (X) > c|δ(t) = 0}, [/formula]
in which c is the cut-off point, t is the survival time, f(X) are the prognostic indices, f (X) = X β and δ(t) is the event indicator at time t [bib_ref] Partial Cox regression analysis for high-dimensional microarray gene expression data, Li [/bib_ref]. Upon examination of sensitivity[c, t|f(X)] and 1specificity[c, t|f(X)], we can define ROC[t|f(X)] as receiving operating characteristic (ROC) curves at any time t and AUC[t|f (X) ] as the area under the ROC curves at any time t. The larger AUC[t|f(X)] is, the better our prediction model performs at time t.
In addition, we evaluate the performance, using a validation set and the log-rank test [bib_ref] Evaluation of survival data and two new rank order statistics arising in..., Mantel [/bib_ref] , to analyze whether the patients are properly classified: high-risk or low-risk. We rank patients in descending order by their prognostic indices (PI) and divide them into a high-risk group containing the top 40% of patients and a low-risk group containing the bottom 40%. We then test the validity of this group assignment by using the log-rank test and the true survival information.
## Cross validation
We select the optimal parameters, λ and α, that respectively control the degree of sparsity and the amount of network constraint in the Net-Cox model using a 5-fold cross validation process. We reserve 20% of all the samples in this study for validation (the validation set) and use the remaining 80% (the training set) for cross validation in order to optimize the parameters and check the cross-validation error. In the case of mutual information networks, we also choose the optimal s using cross-validation. The average time-dependent AUC is used as an evaluation measure for each genomic profile and the different kinds of networks. We trained each model five times using a 5-fold cross validation process on the training set. At each run, we estimate the regression coefficientβ and the mutual information network (or co-expression network) on four folds of the training data. The estimated coefficientβ and the remaining fold, consisting of 20% of the training data, are then used to calculate the prognostic indices PI = X′b, which are, in turn, used to rank the patients according to their expected survival time, or month.
To examine how the mutual information network filtered by significance level s contributes to the performance of the Net-Cox model, we experiment with the complete mutual information network and those networks filtered by s, where s = [0, 0.05, 0.1, 0.15, 0.2, 0.25, 0.3] for each profile. This corresponds to eight mutual information networks on each profile, including the complete network. We vary the parameters λ = [10 −4 ,10 −3 , 10 −2 , 10 −1 ] and α = [0.1, 0.3, 0.5, 0.7, 0.9, 1]. As a result, we choose the optimal (λ, α) pair with the best performance, in terms of the time-dependent ROC curves, for each network and each profile.
# Enrichment analysis
In order to assess the biological significance of the results, we performed a gene list enrichment analysis for the 100 genes with the largest regression coefficients for each profile using ToppGene http://toppgene.cchmc.org [bib_ref] ToppGene Suite for gene list enrichment analysis and candidate gene prioritization, Chen [/bib_ref]. The enrichment analysis is based on Gene Ontology [bib_ref] Gene ontology: tool for the unification of biology. The Gene Ontology Consortium, Ashburner [/bib_ref] and Pathway and Disease. The terms that have adjusted p-values under 0.05 when using the Benjamini-Hochberg correction [bib_ref] Discovering the false discovery rate, Benjamini [/bib_ref] are considered biologically significant. We also ran the enrichment test on the genes in the network consisting of the 100 edges with the largest mutual information values for each profile.
# Results and discussion
Optimal parameter selection using cross validation
We first examined the performance behavior of the mutual information network based Net-Cox model as a function of the parameter s . Note that as the value of s increases, fewer edges remain in the network. In , the leftmost x-tick label "comp" represents the result obtained using the complete network, and the following x-ticks correspond to different s values. Overall, the best performance resulted not from the complete network but from the network filtered by s. Therefore, reducing less significant information and using only significant edges seem to improve the performance of the networkregularized regression model. When the CNA and methylation profiles were examined, the best result was obtained using the largest test value of s = 0.3, for which 9,896 and 6,011 edges, respectively, remained, which was about 0.01% of the total number of pairs. In contrast, the mRNA profile preferred a larger network consisting of 568,486 edges (about 1%) with the optimum at s = 0.1.
We also optimized the parameters l and a using a 5-fold cross validation process [fig_ref] Table 1: Optimal parameters and averaged time-dependent AUCs of each network using 5-fold cross... [/fig_ref]. We found that the optimal a value for the mutual information network was smaller than those of the co-expression and functional linkage networks. This implies that the mutual information network contributed more to the regression Prediction accuracy of the mutual information network-based Net-Cox model as a function of the network significance parameter s. Average time-dependent AUC using 5-fold cross validation on training data is shown across σ. http://www.biomedcentral.com/qc/1752-0509/9/S1/S8 model than the other approaches. Moreover, the average time-dependent AUC was highest when the mutual information based network was used in all profiles. However, we also note that the median time-dependent AUC was not always highest for the mutual information network, which may imply complementary properties of different types of networks. In [fig_ref] Figure 2: Performance comparison of different network types in terms of 5-fold cross-validation accuracy [/fig_ref] , the bar plot for the average time-dependent AUC and the boxplot for the distribution of the time-dependent AUC across all the time points and the 5-fold experiments are shown.
## Performance comparison on validation set
After selection of the best parameters for each network, we re-trained each model using the entire training set and then applied the obtained regression model to the holdout validation set. [fig_ref] Figure 3: Performance comparison of different network types on validation set [/fig_ref] shows a comparison between the resulting time-dependent AUCs. The mutual information based network had the best performance for the mRNA profile and comparable results for the other profiles. [fig_ref] Figure 4: Kaplan-Meier survival curves [/fig_ref] shows the Kaplan-Meier survival curves and p-values from the log-rank tests with respect to the patient group assignment for each approach. Examining the CNA and mRNA profiles, every method revealed significant results (p-value < 0.05) with the log-rank test, except for the functional linkage network on the CNA profile. However, all the methods showed insignificant results for the methylation profile. It appears that the interaction effect in the methylation profile is substantially less than in other profiles.
## Signature genes for each profile
In order to examine the genes that have the strongest marginal association with survival, we displayed the five largest regression coefficients in each profile as a heatmap [fig_ref] Figure 5: Heat-map for the regression coefficients of 15 selected genes [/fig_ref]. The top five genes from each profile were all distinct with no overlap, but some genes had large coefficients in multiple profiles. The top genes from the CNA profile had smaller overall regression coefficients than the top genes in other profiles. Moreover, the top genes in the CNA profile had larger regression coefficients than those in other profiles, which suggests that the roles of these genes are more prominent in other genomic levels. We find that the identified genes are associated with ovarian cancer in many previous studies. The gene with the largest coefficient in the CNA profile was URI1, prefoldin-like chaperone (URI1), which is reported to be an oncogene that amplified in ovarian cancer cells [bib_ref] URI is an oncogene amplified in ovarian cancer cells and is required..., Theurillat [/bib_ref]. Dual-Specificity Tyrosine-(Y)-Phosphorylation Regulated Kinase 1B (DYRK1B), which is over-expressed in a wide spectrum of ovarian cancer cell lines and human specimens [bib_ref] The involvement of FoxO in cell survival and chemosensitivity mediated by Mirk/Dyrk1B..., Gao [/bib_ref] , was the fourth highest gene. The largest coefficient gene from the mRNA profile, Oviductal Glycoprotein 1 (OVGP1), is reported to be a more accreted detection marker than other markers of ovarian epithelial cancers [24]. Over-expression of the Elafin/Peptidase Inhibitor 3 (PI3) gene from the mRNA profile is associated with poor overall survival. We found no reported associations of the top five genes in the methylation profile with ovarian cancer from previous studies. An enrichment analysis was performed for the 100 genes having the largest coefficients in each profile [fig_ref] Table 2: Significantly enriched terms [/fig_ref]. Many significantly enriched terms demonstrated an association with ovarian cancer. For example, the genes URI1, Processing Of Precursor 4 (POP4), Pleckstrin Homology Domain Containing, Family F Member 1 (PLEKHF1), and DYRK1B from the CNA profile were enriched in the Ovarian Neoplasms term (ctd:D010051) of Disease. For Gene Ontology, the top 100 genes in the CNA profile were primarily enriched in the Biological Process (BP) terms. The Hox family of homeobox genes was enriched in the terms related with the embryonic skeletal system (GO:0048706; GO:0048704), and these genes are critical for cell migration and DNA repair [bib_ref] Multilevel omic data integration in cancer cell lines: advanced annotation and emergent..., Liu [/bib_ref]. The genes in the mRNA profile that were enriched in the Chitinases-related terms were already known for their cancer indication roles [bib_ref] Human chitinases and chitinaselike proteins as indicators for inflammation and cancer, Kzhyshkowska [/bib_ref]. Therefore, we can conclude that the genes with large regression coefficients are expected to be related to ovarian cancer.
# Network analysis
To illustrate the general topology of the mutual information network and its effect on prediction performances more closely, we constructed gene interaction sub-networks by using the 100 edges with the largest mutual information values for each profile [fig_ref] Figure 6: Gene-gene interaction sub-network constructed by using the 100 edges with the largest... [/fig_ref]. The color of each node represents the strength of its marginal effect. We also see many genes with weak marginal effects appear in the network. Topologies of the networks were analyzed by Cytoscape [bib_ref] Cytoscape: A Software Environment for Integrated Models of Biomolecular Interaction Networks, Shannon [/bib_ref] and summarized in [fig_ref] Table 3: Network properties of mutual information subnetwork for each profile [/fig_ref]. The three networks reveal different network structures and topologies. The CNA network consists of a smaller number of connected components than the others (7 versus 54 and 50). It also shows a denser connection between the genes. The network centralization value is about five times higher, and its average number of neighbors is about twice as much. In addition, the overall coefficients of the genes in the CNA network were smaller than those in the other profiles. Considering that those genes have high mutual information values and, hence, strong interactive effects, this may imply that the interaction effect on survival in the CNA profile is more dominant than the marginal effect with each gene. The R 2 value for the power-law distribution of the mRNA and methylation networks were 0.922 and 0.909, respectively, which shows strong scale freeness [bib_ref] A general framework for weighted gene coexpression network analysis, Zhang [/bib_ref] , as is the case with many other biological networks [bib_ref] The large-scale organization of metabolic networks, Jeong [/bib_ref] [bib_ref] The use of network analyses for elucidating mechanisms in cardiovascular disease, Diez [/bib_ref] [bib_ref] Gene co-expression network topology provides a framework for molecular characterization of cellular..., Carter [/bib_ref] [bib_ref] Comprehensive curation and analysis of global interaction networks in Saccharomyces cerevisiae, Reguly [/bib_ref]. [fig_ref] Table 4: Significantly enriched terms [/fig_ref] summarizes the enrichment test results for the genes in each sub-network. The network from the methylation profile was mainly enriched in GO Molecular Function (MF) terms. Networks for the CNA and mRNA profiles were not enriched in GO terms but in terms of diseases related to ovarian cancer, such as Neoplasms (ctd:D009369), Carcinoma (ctd:D002277), and Neoplasm Recurrence, Local (ctd:D009364). Further, the CNA profile network was enriched in other related disease terms, such as Obesity (ctd: D009756) and BMI (Body Mass Index) [bib_ref] Height, Body Mass Index, and Ovarian Cancer: A Follow-Up of 1.1 Million..., Engeland [/bib_ref] , and Insulin Resistance (ctd:D007333) [bib_ref] Insulin resistance and cancer risk: an overview of the pathogenetic mechanisms, Arcidiacono [/bib_ref] and Hyperlipidaemia (ctd:D006949) [bib_ref] Ovarian cancer risk and history of selected medical conditions linked with female..., Parazzini [/bib_ref].
# Conclusions
In this study, we investigated the utility of an alternative network construction approach based on mutual information in network-based Cox regression. Our results show that the mutual information based network can further improve prediction performance in survival analyses. Moreover, the permutation testing scheme used to discard insignificant pairs improved the prediction performance.
Overall, the performance gain of this alternative approach over existing methods was rather marginal. It seems due to a mismatch between the high mutual information value and the small value of the penalty term (ψ(β)) = 1 2 p i,j S i,j (β i − β j ) 2 ) in the Net-Cox model-it does not necessarily mean that the gene pairs containing high mutual information with respect to survival should have similar marginal effects because the mutual information measure is more concerned with the interaction effect. Even with this discrepancy, results based on the mutual information network are still promising. In future studies, the regularization term could be modified to better reflect the information contained in the mutual information network and, hence, further improve the performance. Another direction would be to apply this network scheme to network-based approaches in other domains.
[fig] Figure 2: Performance comparison of different network types in terms of 5-fold cross-validation accuracy. Jeong et al. BMC Systems Biology 2015, 9(Suppl 1):S8 [/fig]
[fig] Figure 3: Performance comparison of different network types on validation set. [/fig]
[fig] Figure 4: Kaplan-Meier survival curves (high-risk group: red line; low risk group: blue line) and log-rank test results on validation set for each profile and network (A-Mutual information based network; B-Correlation network; C-Functional linkage network; D-L 2 -Cox). [/fig]
[fig] Figure 5: Heat-map for the regression coefficients of 15 selected genes. The five genes having the largest regression coefficients are selected for each profile and merged together. [/fig]
[fig] Figure 6: Gene-gene interaction sub-network constructed by using the 100 edges with the largest mutual information values on each profile (A: CNA, B: mRNA, C: METH). [/fig]
[table] Table 1: Optimal parameters and averaged time-dependent AUCs of each network using 5-fold cross validation on training data [/table]
[table] Table 2: Significantly enriched terms (using ToppGene) in the 100 genes that have the largest regression coefficients for each profile [/table]
[table] Table 3: Network properties of mutual information subnetwork for each profile [/table]
[table] Table 4: Significantly enriched terms (using ToppGene) in the sub-network consisting of the gene pairs that have the 100 largest mutual information values [/table]
|
Kienböck's disease: a case report
Kienböck disease is a condition characterized by avascular necrosis of the lunate bone. Advanced imaging can aid in the diagnosis and staging of Kienböck disease. Magnetic resonance imaging (MRI) is an important adjunct to diagnosis. In particular, MRI is helpful early in the disease when plain radiographs may not reveal abnormalities. A 17 -year-old man with Kienböck disease who underwent radiography and MR is described in this article.
# Introduction
Kienböck disease is a condition marked by avascular necrosis of the lunate bone. MRI is useful in diagnosis and staging and should be considered, after conventional radiography, for patients with suspected Kienböck disease [bib_ref] Imaging of Kienböck Disease, Arnaiz [/bib_ref]. There are few reports in children and therapeutic recommendations in the literature about this condition [bib_ref] Kienböck's Disease in a 14-Year-Old Gymnast: a Case Report, Herzberg [/bib_ref]. We report here a case of a 17-year-old male teenager.
## Patient and observation
A 17-year-old, left-handed, suffered progressive dorsal left wrist pain, six months before his consultation. No specific wrist injury was reported. The patient complained of a slight loss of the arc of wrist flexion/extension along with substantial loss of grip strength. . Clinical examination showed marked tenderness over the dorsal aspect of the wrist. Flexion/extension of the left wrist was 85°/85° compared with 90°/90° on the right side. Pronation and supination were normal. Grip strength was 10 kg on the dominant left side compared with 20 kg on the right.On standard PA and lateral radiographs wrist showed a densification of the lunate with a flattened and irregular appearance associated with a fixed rotation of the scaphoid, without signs of osteoarthritis .The diagnosis was made of stage III Kienböck's disease according to Lichtman's classification . An MRI was performed confirming the diagnosis by objectifying a fragmented and irregular appearance of the lunate with a revamped appearance on T2 and STIR associated with a thickening of the subcutaneous soft tissue and minimal effusion at the styloid recess [fig_ref] Figure 2: MRI of the left wrist [/fig_ref]. Surgical treatment option was refused by the patient and his parents. It was decided to immobilize the wrist in a splint and to stop sport activities the patient will be convened within 3 months for clinical and radiological control.
# Discussion
Kienböck's disease is rare in children, only a few cases have been published [bib_ref] Kienböck's Disease in a 14-Year-Old Gymnast: a Case Report, Herzberg [/bib_ref] [bib_ref] Kienböck's disease in a 12-year-old girl, Luc De [/bib_ref]. It's most commonly affects men between the ages of 20 and 40 years. The disease commonly affects the dominant wrist [bib_ref] Imaging of Kienböck Disease, Arnaiz [/bib_ref]. Many patients describe a history of trauma, but this is not always present as is the case of our patient.
Kienböck disease was described by the Austrian radiologist Robert Kienböck in 1910 as a condition characterized by avascular necrosis of the lunate bone. It is also known as osteonecrosis, lunatomalacia, and aseptic or ischemic necrosis of the lunate [bib_ref] Imaging of Kienböck Disease, Arnaiz [/bib_ref].
The pathophysiologic mechanism of this entity is multifactorial [bib_ref] Imaging of Kienböck Disease, Arnaiz [/bib_ref].There is no single definitive cause of Kienböck disease, a complex interplay of vascular and anatomic variations, combined with varying degrees of microtrauma and insults, contribute to its development.
The symptoms of Kienböck disease can vary depending on their stage at initial presentation, patients typically present with pain localized to the radiolunate facet, decreased motion, swelling, and weakness in the affected hand. Pain is classically insidious in onset, often related to activity, and can be present for extended periods before presentation.
Radiography is the initial imaging technique for assessing Kienböck disease and also can be used to rule out other pathologic conditions, such as arthrosis and fractures [bib_ref] Imaging of Kienböck Disease, Arnaiz [/bib_ref].Plain radiography allows the disease to be classified into 4 stages according to Lichtman and associates [bib_ref] E13-4. PubMed | Google Scholar Figure 1: Lateral radiographs wrist showed a..., Hurley [/bib_ref]. This classification is highly reliable and reproducible and has the most clinical relevance because it helps in determining the most appropriate treatment [bib_ref] Imaging of Kienböck Disease, Arnaiz [/bib_ref].
MRI is likely to be the next best imaging examination after routine radiography [bib_ref] Imaging of Kienböck Disease, Arnaiz [/bib_ref]. In the early stages of the disease, the use of MRI can aid in making the diagnosis and is more sensitive and specific than bone scanning [bib_ref] E13-4. PubMed | Google Scholar Figure 1: Lateral radiographs wrist showed a..., Hurley [/bib_ref]. Furthermore, MRI is useful for longitudinal assessment of the postoperative response to direct and indirect revascularization procedures. Contrast-enhanced MRI is important for determining the degree of necrotic tissue and the most appropriate treatment of stage II and IIIA disease. Contrastenhanced, MRI is not necessary in stages I, IIIB, IIIC, or IV because the degree of necrosis does not change treatment in these stages [bib_ref] Imaging of Kienböck Disease, Arnaiz [/bib_ref].
Kienböck disease remains a challenging clinical problem. Kienböck disease is often a progressive disorder resulting in joint destruction within 3-5 years if untreated [bib_ref] Imaging of Kienböck Disease, Arnaiz [/bib_ref]. There remains no definitive treatment of this entity. There are several treatment options, largely based on the stage at presentation. Although options vary, they typically fall into several broad categories: unload the lunate, revascularize the lunate, or treat carpal instability and collapse with salvage procedures.
# Conclusion
Kienböck disease is a condition marked by avascular necrosis of the lunate bone. MRI can help in visualizing of the bone anatomy, the staging of Kienböck disease, and ruling out alternative diagnoses that mimic Kienböck disease (pseudo-Kienböck lesions). MRI therefore should be considered after conventional radiography in the care of patients with suspected Kienböck disease. Lateral radiographs wrist showed a densification of the lunate with a flattened and irregular appearance of the lunate, without signs of osteoarthritis
# Figures
[fig] Figure 2: MRI of the left wrist: (A) sagital protons density showing collapse of the lunate with mixesd signal intensity; (B) sagital T2weighted fat-suppressed MR image showing fragmented and irregular appearance of the lunate with hyper signal intensity; (C) coronal STIR showing fragmented and irregular appearance of the lunate with hyper signal intensity [/fig]
|
Deep-inspirational breath-hold (DIBH) technique in left-sided breast cancer: various aspects of clinical utility
Background: Studying the clinical utility of deep-inspirational breath-hold (DIBH) in left breast cancer radiotherapy (RT) was aimed at focusing on dosimetry and feasibility aspects.Methods:In this prospective trial all enrolled patients went through planning CT in supine position under both DIBH and free breathing (FB); in whole breast irradiation (WBI) cases prone CT was also taken. In 3-dimensional conformal radiotherapy (3DCRT) plans heart, left anterior descending coronary artery (LAD), ipsilateral lung and contralateral breast doses were analyzed. The acceptance of DIBH technique as reported by the patients and the staff was analyzed; post-RT side-effects including radiation lung changes (visual scores and lung density measurements) were collected.Results: Among 130 enrolled patients 26 were not suitable for the technique while in 16, heart or LAD dose constraints were not met in the DIBH plans. Among 54 and 34 patients receiving WBI and postmastectomy/nodal RT, respectively with DIBH, mean heart dose (MHD) was reduced to < 50%, the heart V 25 Gy to < 20%, the LAD mean dose to < 40% and the LAD maximum dose to about 50% as compared to that under FB; the magnitude of benefit was related to the relative increase of the ipsilateral lung volume at DIBH. Nevertheless, heart and LAD dose differences (DIBH vs. FB) individually varied. Among the WBI cases at least one heart/LAD dose parameter was more favorable in the prone or in the supine FB plan in 15 and 4 cases, respectively; differences were numerically small. All DIBH patients completed the RT, inter-fraction repositioning accuracy and radiation side-effects were similar to that of other breast RT techniques. Both the patients and radiographers were satisfied with the technique.Conclusions: DIBH is an excellent heart sparing technique in breast RT, but about one-third of the patients do not benefit from that otherwise laborious procedure or benefit less than from an alternative method.Trial registration: retrospectively registered under ISRCTN14360721 (February 12, 2021)
After breast-conserving surgery most of the patients need postoperative whole breast irradiation (WBI) with or without a tumor bed boost, while in low-risk cases partial breast irradiation (PBI) is sufficient. The need of chest wall (CW) irradiation is infrequent unless if combined with nodal irradiation, postmastectomy irradiation (PMI). The practice of nodal irradiation shows a broad spectrum from the sole irradiation of different axillary levels to that of all regional nodes depending on the risk status and surgery performed. Clinical studies indicate that while after sentinel lymph node biopsy (SNB) axillary block dissection may be replaced with axillary level I-II nodal radiotherapy in limited nodal involvement cases, full nodal radiotherapy although at a price of radiation sequelae, may contribute to improved survival of axillary lymph node positive cases.
Most breast cancer patients become long-survivors, hence therapeutic interventions should not endanger the patients' general health and well-being. The main radiation-related hazards are radiogenic heart and lung damage resulting in significant morbidity many years or decades after the radiotherapy. The excess relative risk of secondary lung cancer or cardiac mortality has been estimated as 0.11 and 0.04 per one Gray increase of the dose to the whole lung and heart, respectively. Radiation-induced heart disease (RIHD) most frequently manifests in the damage of the coronary and capillary vessels of the heart which induces a progressive fibrotic process leading to circulatory changes with potentially fatal ischemic heart disease (IHD). The dose-dependent harmful effect of radiation exposure of the heart has been demonstrated in retrospective analyses and simulations of radiotherapies of breast cancer (BC) patients with IHD. The dose to the heart and hence RIHD incidence is higher in left-sided cases and the risk is more significant in young patients. It is estimated that every 1 Gy mean heart dose increases the incidence of IHD event by 7.4% that may be potentiated by preexisting cardiovascular risk factorsand smoking. In a systematic review of contemporary publications, a radiation dose to the heart of 5.2 and 3.7 Gy in left and right sided cases, respectively, is still demonstrated while the dose to the ipsilateral lung is 9 Gy. The risk of radiogenic heart damage is linearly dose-dependent, but no lower threshold with the absence of risk has been identified. Hence all efforts should be made to avoid or lessen heart exposure as much as possible.
There are many approaches to protect the heart from radiation exposure. Prone positioning and the breathholding techniques operate by separating the heart and the radiation fields; the advanced IMRT and proton irradiation techniques are not widely applied, while the reduction of the volume to be irradiated during partial breast irradiation (PBI) or the omittance of radiotherapy are options in low-risk cases. These methods result in variable effects on lung and heart exposures: while prone radiotherapy dramatically reduces lung doses, heart doses individually differ.
The breath-holding technique first described in breast cancer in 2001only recently became widespread. Its greatest impact is reduced dose to the heart and LAD, and to a lesser extent to the lung. Nevertheless, the magnitude of benefit individually varies according to the patient's anatomical features and lung capacity; occasionally cases with the absence of advantageor even elevated heart doseshave been described. Since earlier we found prone positioning individually helpful for heart sparing in the RT of Three-dimensional conformal whole breast irradiation plans of 2 cases illustrating the benefit (upper series) or the lack of benefit (lower series) of the DIBH manoeuvre; transversal slices are shown during free-breathing (FB), DIBH and in the prone position (PTV contoured in red, heart contoured in orange, LAD contoured in yellow) left breast cancer cases, we wished to study the practical aspects of DIBH to find out its optimal place in routine practice. We tested the feasibility of the method among all patients, analyzed the relative dosimetry benefits according to the indications, and in the WBI only cohort, analyzed nodal coverage; also, radiation lung changes were prospectively followed.
# Methods
This prospective cohort study had been approved by the Institutional Ethics Review Board of the University of Szeged (#272/2017), and all the enrolled patients gave their written informed consent to participation. Inclusion criteria were left-sided breast cancer needing postoperative WBI/PMI and informed consent, while the exclusion criteria were the presence of chronic obstructive pulmonary disease, bronchial asthma or other severe comorbidity that would hinder cooperation during DIBH (extreme obesity, mental disorder, hypacusis).
For DIBH, the protocol of the voluntary breath-holding technique as described by http:// www. jove. com/ video/ 51578/]. All enrolled patients were assessed for baseline breath-holding capacity and trained for DIBH by a physiotherapist (JS), who reinvited the patient for further support if needed. Those patients who could not practise breath-holding for a minimum of 20 s were withdrawn from DIBH. These, together with those who despite sufficient breath-hold did not have improved heart and/or LAD doses (and hence DIBH was irrelevant) were excluded from further analyses.
Planning CT series were acquired in supine position with the arms elevated under both normal breathing and DIBH; in WBI cases CT was performed also in prone position. In-room lasers, skin marks, and verbal instructions through an audiovisual system with high performance video cameras (RMC-190 controller, 2 PTC-120 robotic HD cameras, Datavideo Technologies Co., Taiwan) ensured consistent breath-hold; beam-gating was applied manually.
Target volumes and organs at risk (OARs) were outlined in all CT series, and treatment plans were generated in all setups. In the first series of 37 cases irradiated with DIBH technique, portal imaging 3-times a week with 2 orthogonal setup beams of kilovoltage photon energy was performed, necessary couch translations were recorded and inter-fractional systematic and random setup errors were calculated; also, daily treatment times were recorded, and questionnaires regarding the comfort of the RT procedure (patients, RT fractions 6 and 25, cumulative score 0-9) and the efforts needed from the staff during planning CT or RT (radiographers, planning CT and RT fractions 6 and 25, cumulative score 0-12) similarly to were completed.
High scores indicated satisfactory acceptance (Additional file 1).
## Radiotherapy and dosimetry data
Radiotherapy techniques and facilities were described previously. Briefly, patients were positioned on the supine thorax and prone breast modules (WBI cases) of the AIO (All In One) Solution (ORFIT, Wijnegem, Belgium) system without mask fixation. Planning CT images were acquired throughout the entire planning volume. Target volumes (whole breast or chest wall with or without nodal regions including axillary levels I-IV and internal mammary lymph nodes) and OARs i.e. the heart, LAD, ipsilateral lung, contralateral breast were contoured according to the European Society for Radiotherapy and Oncology (ESTRO) guidelines. The aim was to achieve equivalent target and nodal volume contouring among all setups. The main objectives were a mean dose to the planning target volume (PTV) of 50 Gy (25 fractions), and V 47,5 Gy ≥ 90%, V 53,5 Gy ≤ 1%, while the dose constraints were the following: heart mean dose < 2.5 Gy, heartV 25Gy < 3%, LAD mean dose: < 12 Gy, lung mean dose < 10 Gy (WBI) and < 16 Gy (nodal RT), lung V 20 Gy < 15% (WBI) and 30% (nodal RT), contralateral breast V 10Gy < 5%. All plans were generated in the Varian Eclipse v13.6 (Varian Oncology Systems, Palo Alto, CA, USA) treatment planning system with the Analytical Anisotropic Algorithm (AAA) v8.0. 3DCRT plans applyed opposing tangential 6MV photon fields set up isocentrically and a median of 2 (1-3) 6/10 MV segmental fields (WBI, CW) . For nodal irradiation, a single matched field was used. Sequential tumor bed boosts given according to the protocol were not included in the dosimetric comparisons. Radiotherapy was delivered with a Varian TrueBeamSTx (Varian Oncology Systems, Palo Alto, CA, USA) linear accelerator.
For plan evaluation, conformity and homogeneity indiceswere calculated as follows: CN = TV RI TV × TV RI V RI
(TV: Target volume, i.e. PTV; TV RI : Target volume covered by the reference isodose; V RI : Volume of the reference isodose, ideal is 1), Homogeneity Index (HI)(D 2% , D 50% , D 98% = dose received by 2%, 50% and 98% of PTV, respectively, ideal is 0):
The following dose-volume parameters of the PTV and OARs were collected: heart mean dose, heart V 25 Gy , LAD mean dose, LAD maximum dose, ipsilateral mean lung dose, ipsilateral lung V 20 Gy , contralateral breast V 10 Gy . In further analyses we compared heart and LAD dose reductions (the differences between the heart mean dose,
[formula] HI = D 2% − D 98% D 50% [/formula]
heart V 25 Gy and LAD mean dose in RT plans under FB vs. DIBH, respectively) with the relative increase of the ipsilateral lung volume.
In the WBI only cohort, nodal regions were retrospectively contoured both in the DIBH and free breathing (FB) planning CTs for nodal dosimetry (V 25 Gy , V 45 Gy , V 47.5 Gy ) comparisons.
## Follow-up of patients including lung density change measurements
Data on clinical symptoms, skin changes and breast fibrosis were collected in the first cohort of patients at the completion of RT, and 3 months and 1 year thereafter using the CTCAE v. 4 system. Also, chest CTs under FB were performed at these time points for assessing postirradiation lung changes as described. Briefly, CT scans were evaluated both visually (category 0: no visible changes, category 1: increased density, hazy opacity, category 2: strand-like thickening) and with measuring the density of the lungs in a CT slice at the level of the left heart ventricle in the area outlined by the chest wall and a line between the edge of the sternum and the midheight of the chest in comparison with the planning CT. Density measurements were performed by excluding visible changes, and were corrected by subtracting the density of the unexposed contralateral lung. Visual assessment scores and mean lung density changes (MLDC) were analyzed and compared to the same parameters in a previous cohort of patients irradiated between 2010 and 2011 with 3DCRT under FB in supine position.
# Statistical analysis
Patient-related and dosimetry data were summarized using descriptive statistics; quantitative variables were expressed as mean ± SD. For the comparison of data, repeated ANOVA test (3 variables) and paired t-test (2 variables) were used. Lung density changes were analyzed with one-way ANOVA, while visual scores in the DIBH vs. FB groups with the chi-square test. The effects of patient-related variables or RT categories on DIBHrelated dosimetry benefit were evaluated with linear regression.
# Results
According to the inclusion and exclusion criteria, 130 patients were enrolled between January 2018 and November 2019, nevertheless, among them 42 patients did not partake in DIBH and hence were not included in the dosimetry analysis of DIBH cases. Twenty-six patients were not suitable for DIBH (18 could not withhold breathing for 20 s, 3 patients were stressed, 2 withdrew consent, 1 could not follow the instructions, 1 was diagnosed with pulmonary embolism following enrollment and 1 with disease progression), while in 16 cases OAR doses were apparently not acceptable despite the DIBH manoeuvre. Out of the 16 patients with suboptimal 3DCRT plans under DIBH, 10 received RT with an alternative technique in supine position, and 6 received 3DCRT in prone position. Altogether 88 patients received RT under DIBH, their age was 57.3 ± 11.7 years (mean ± SD), their weight, length and BMI were 72.5 ± 13.6 kg, 163.6 ± 7.1 cm and 27.1 ± 5.1 kg/m 2 (mean ± SD), respectively. Among them 35 received chemotherapy before the RT, and 56 received endocrine therapy during (aromatase inhibitors) or after (tamoxifen) the RT.
## Dosimetry data
In the group of 88 patients who received 3DCRT under DIBH, MHD was reduced by > 50%, the heart V 25 Gy by > 80%, the LAD mean dose by > 60% and the LAD maximum dose by about 50% as compared to that under FB. The extent of dose reduction did not differ according to the indication of WBI, PMI or nodal RT. Lung doses were lower under DIBH than under FB and were dramatically reduced in the prone RT plans. The dose to the contralateral breast did not significantly change according to DIBH vs. FB. Nevertheless, heart and LAD dose differences (DIBH vs. FB) individually varied. Among the WBI cases at least one dose parameter was more favorable in the prone plan or in the supine WBI + FB plan in 15 or 4 cases, respectively; differences were numerically small (Table 2/A). None of the heart or LAD doses were superior under FB in the group of patients receiving WBI/CW + nodal RT. As mentioned earlier, in altogether 16 cases of all enrolled patients dose constraints were not met in the supine 3DCRT + DIBH plans; these patients received irradiation using an alternative technique with improved dose parameters as listed in
[formula] /B. [/formula]
The relative increase in the volume of the ipsilateral lung was 1.75 ± 0.06 (range 1.25-2.41).
Weak correlations were found in the entire DIBH population between the relative increase of the ipsilateral lung volume and the MHD (R = 0.400, p < 0.000), heart V 25Gy (R = 0.386, p < 0.001) and LAD mean dose (R = 0.242, p = 0.037). None of the studied patient-related or RT parameters were associated with the benefit of DIBH.
In a dosimetry analysis, we studied the nodal doses in a cohort of 30 WBI cases under FB vs. DIBH. WBI with DIBH delivered significantly less dose to the level 1 axillary lymph nodes but larger doses to the interpectoral and internal mammary nodes than WBI under FB. Although the average V 45 Gy was < 30% in all subregions, individual dose coverage differed.
## Feasibility
All the patients having started DIBH completed the treatment. Inter-fractional random and systematic setup errors were 3.9 and 2.7 mm, respectively. The average and median time for daily radiotherapy with DIBH took mean: 10.7minutes.
In 43 cases including 27 patients with WBI only, 4 with WBI + nodal RT and 12 with CW + nodal RT, post-RT follow-up data were analyzed. Radiodermatitis of grade 1 was usual breast fibrosis occurred in 40% of the patients 3 months after the RT but was much less thereafter. Asymptomatic inflammatory/ fibrotic radiogenic lung changes were present in most cases, while in 2 patients also chest pain and cough developed, but medical intervention was not needed (pneumonitis of grade 1) ( Patients were asked about their comfort during the DIBH procedure, the stability of the position and the easiness of the intervention, while radiographers were asked about their efforts needed and the compliance of the patient during intervention. High total scores indicate satisfaction with the method from both sides. Heart and LAD doses in patients for whom DIBH did not provide advantage WB, whole breast, CW, chest wall, WBI, whole breast irradiation, IMRT, intensity-modulated radiotherapy, 3DCRT, 3D conformal radiotherapy A, Patients treated with 3DCRT + DIBH (n = 88): the difference in heart and LAD doses (DIBH vs. FB/prone) differed according to the case and the dosimetry parameter B, List of individual dose differences among the 16 patients excluded from DIBH due to unacceptable heart and LAD doses; in these, alternative Discussion DIBH could not be implemented for all patients. We found that even among patients seemingly appropriate for the technique, about 20% could not practise DIBH, and 15% of the rest due to dosimetry concerns had to receive RT using an alternative technique. Furthermore, although in the remaining cases the DIBH maneuver resulted in reduced heart and LAD and lung doses in most, still, in some cases FB or prone positioning yielded superior or equal dosimetry results.
In a large database of 272 patients, similar experience was found: more than 40% of the patients were not suitable for or did not benefit from 3DCRT + DIBH. Tanguturi et al. found that among 146 patients deemed potential candidates for breast radiotherapy with DIBH, the DIBH technique provided a neutral change in 25 (17%) or even increased MHD in 14 (10%) cases. A beneficial effect was favored by younger age, greater body mass index (BMI) and larger inspirational lung volume changes. Dell'Oro et al. demonstrated that the benefit provided by DIBH in OAR doses individually differs. Notably, 3 out of 20 patients had higher heart, 4 had increased LAD and 6 had increased lung doses under DIBH as compared to FB. It was found that larger total lung volume increase predisposed to more significant DIBH-related heart and LAD dose reductions. Lin et al. found that in comparison to FB, DIBH resulted larger benefit in heart and LAD doses during postmastectomy RT than during WBI. No similar relationship could be seen in our study; in this patient cohort heart and LAD dose reductions were weakly correlated with DIBH-related lung volume increases, and none of the studied parameters seemed to have predictive power.
Due to the mentioned obstacles of benefiting the most from DIBH and for the optimal use of resources, every RT center should decide its strategy how to include DIBH into routine practice. One approach is the use of anatomical attributes for individual patient selection. Lin et al. tested artificial intelligence-based predictive models to choose the preferred RT technique DIBH vs. prone in a set of 16 patients needing WBI. Taking into consideration heart doses only, breast volume and the distance between the center of the breast and the heart, while respecting composite OAR exposure including the ipsilateral lung, the volumes of the heart and breast, and breast-lung distance had predictive power. If just heart doses were considered, in more cases was preferred the DIBH technique, while if weighted OAR toxicity (dose to heart, ipsilateral lung, and contralateral breast) was considered, the opposite was the finding.
Another possibility is to apply DIBH only in a selected group of patients for example if heart exposure seems unfavorable under FB. Such an optimization-selection approach was the development of a trained model (using RapidPlan ™ , Varian) to develop RT preplans for the estimation of heart doses under FB: if the parameters under FB are satisfying, there is no need for a second series of CT under DIBH. Another strategy to perform screening of all left-sided cases for DIBH using a quick, pragmatic, and systematic assessment protocol by radiographers to consider whether the technique is beneficial, and the patient is suitable. In both cases, primary CT scanning under FB was needed. Tanna et al. compared 4 selection methods, and found optimal the London Cancer Alliance's upfront selection process (without the need of CT scanning under FB) that recommends the use of DIBH if the tumor bed is situated in the lower quadrants of the breast or is extensive, and in all CW RT cases. This method should be preferred in centers with resource constraints.
In most cases similar heart and LAD dose reductions to published results were found at DIBH. We consider DIBH one of the heart sparing techniques which, has the mainstay of good reproducibility and the position-stabilization of the PTV. In WBI cases an alternative solution could be prone positioning; in this, although repositioning accuracy is inferior, ipsilateral lung dose is significantly reduced. In nodal RT cases, if DIBH cannot be utilized, the IMRT technique may provide solution. Actually, Zhao et al. found that unacceptable heart and LAD doses in 3DCRT plans with DIBH could be solved using the IMRT technique, but no further dosimetric improvement was seen in low OARdose cases. In fact, the consideration of the dose constraints of the heart and its substructures seems the most important for adequate heart sparing in modern RT as suggested by the DEGRO breast cancer expert panel; the appropriate technique should be selected accordingly.
Thanks to the successful use of new technologies and raised attention to solving heart sparing, the control of ipsilateral lung doses is becoming a challenge. It has been estimated that DIBH by reducing the dose to the ipsilateral lung volume and lung mass by 15-24%, results in an about 20% improvement in lung normal tissue complication probability (NTCP) as compared to that after FB. In fact, lung doses approximate zero if prone WBI is performed. Yan et al. based on literature data, compared the risks of cardiac death, significant cardiac event, or lung cancer mortality in 34 breast cancer patients having RT plans both under DIBH and in the prone position. Both techniques provided excellent cardiac sparing, nevertheless, with DIBH, an excess lung cancer mortality of 0.5% evolved, which was absent with the prone WBI technique. We studied the occurrence of acute and late radiation changes of the ipsilateral lung by means of visual evaluation and lung density measurements. That special interest was due to the known stimulatory effect of oxygen on radiosensitivity. The incidence of Range 5-9 5-9 4-12 5-12 5-12 asymptomatic abnormalities was higher, but not significantly different from that in an earlier patient cohort. We believe that the protocol-defined followup period of 1 year post-radiotherapy was sufficient to detect all adverse pulmonary reactions. A significant proportion of the enrolled patients was not able to effectively practise DIBH. Those whose voluntary breath-holding capacity or chest wall movements were not sufficient for DIBH were identified before or at simulation. In a retrospective analysis, Nissen et al. found similar proportion and similar reasons for omitting DIBH in a left-sided BC patient population: among 20 patients non-compliant to DIBH, 7 could not use the equipment, 7 was not able to maintain breath-hold for at least 20 s, 7 had psychological problems, and 4 had other reasons. Training for DIBH is essential. Kim et al. detected lower heart doses in patients who received 5-day preparatory coaching before starting DIBH versus those who did not. There is controversy whether thoracic or abdominal DIBH provides better heart sparing. Zhao et al. found that most patients practice thoracic DIBH, but may be easily trained for abdominal DIBH which in fact, better lowers heart, LAD and lung doses. Proper DIBH technique is essential since its beneficial effect depends primarily on the lung volume increase achieved. We consider very important the role of the physiotherapist expert who should provide standardized coaching and support to the patients. The DIBH technique was easily adapted, and inter-fractional repositioning accuracy seemed similar to that of other techniques (supine, prone, PBI).
As demonstrated by Pazos et al. and Borm et al., during DIBH, the relative position of the nodal regions to that of the PTV changes and as a consequence accidental doses to axillary levels I and II become less or more relevant as compared to that without DIBH. We separately analyzed the 6 nodal regions, and found that due to DIBH while the dose to axillary level I is lower, the doses to the interpectoral lymph node region and the ipsilateral internal mammary (IM) region may be higher than during FB, however, differences show individual variations. We think that this finding should raise attention to always considering the doses to axillary levels I-II if their irradiation is needed due to limited axillary surgery and low-volume SNB positivity. In these cases, supine RT with DIBH should be favored instead of prone WBI since during the latter the axillary lymph node regions are excluded from irradiation.
Since the dose to the OARs is rarely a problem during PBI, limited experience is published with PBI and DIBH. The DIBH method utilized independently of the laterality of the disease was found useful for the PBI of 37 patients. In our practice, heart and LAD exposure clearly depended on the size and location of the tumor bed. Recently, the use of DIBH occasionally, in cases with relatively high heart or LAD doses in PBI plans the DIBH method provided prompt solution.
# Conclusions
DIBH is one of the powerful heart sparing techniques in breast cancer RT. Alertness is needed to identify those patients who do not benefit from that laborious procedure or benefit less than from an alternative method: if WBI is needed, prone positioning, if nodal RT is necessary the IMRT technique could be alternate options.
## Abbreviations
3DCRT : 3-Dimensional conformal radiotherapy; CW: Chest wall; DIBH: Deepinspirational breath-hold; ESTRO: European Society for Radiotherapy and Oncology; FB: Free breathing; LAD: Left anterior descending coronary artery; MHD: Mean heart dose; OAR: Organ at risk; PBI: Partial breast irradiation; PMI: Postmastectomy irradiation; PTV: Planning target volume; RIHD: Radiationinduced heart disease; RT: Radiotherapy; SNB: Sentinel node biopsy; V xGy : A subvolume of a volume that is exposed to ≥ x Gy; WBI: Whole breast irradiation. |
Pneumomediastinum, pneumorrhachis and subcutaneous emphysema associated with viral infections: Report of three cases
Spontaneous pneumomediastinum is usually secondary to alveolar rupture in the pulmonary interstitium, associated with subcutaneous emphysema and occasionally with pneumothorax, but is rarely associated with pneumorrhachis. The leaked air into the pulmonary perivascular interstitium follows the path of least resistance from the mediastinum to the fascial planes of the neck. Air freely communicates via the neural foramina and collects in the epidural space. Pneumorrhachis is defined as the presence of air in the spinal canal, either in the intradural and/or extradural spaces. It is a very rare clinical entity and mostly asymptomatic, hence most probably underdiagnosed. Many pathological and physiological events can lead to alveolar rupture, and these clinical findings can be related to various, mainly traumatic and iatrogenic etiologies. Herein we report three cases of pneumomediastinum, subcutaneous emphysema, interstitial emphysema and pneumorrhachis in two cases, which were related to rhinovirus, human bocavirus and respiratory syncytial virus infection.
Pneumomediastinum is defined as the presence of interstitial air in the mediastinum. Pneumomediastinum is a benign condition that is occasionally associated with subcutaneous emphysema and occasionally with pneumothorax, but is rarely associated with pneumorrhachis (PR). [bib_ref] Spontaneous pneumomediastinum, Fugo [/bib_ref] Pneumorrhachis, "air within the spinal epidural space", occurs in a variety of settings, including skull and spinal fractures, epidural abscess, epidural anesthesia, lumbar puncture, and traumatic pneumothorax and pneumomediastinum. Although it is rare, there have been documented cases in the literature in which bronchial asthma also contributed to the development of PR. [bib_ref] Pathogenesis, diagnosis and management of pneumorrhachis, Oertel [/bib_ref] In PR, increase in the intra-alveolar pressure secondary to vigorous cough leads to rupture of alveoli, resulting in air leaks in the peribronchovascular space, where it follows the path of least resistance to the mediastinal, pleural spaces and fascial planes of the neck. There are no fascial barriers to prevent communication of the posterior mediastinum or the retro-pharyngeal space with the epidural space. Air thus freely communicates via the neural foramina and collects in the epidural space. [bib_ref] Pneumorrhachis, pneumomediastinum, pneumopericardium and subcutaneous emphysema as complication of bronchial asthma, Manden [/bib_ref] There are no standard guidelines for the management of symptomatic PR, and its treatment is often individualized.
Herein we report three cases of pneumomediastinum, subcutaneous emphysema, interstitial emphysema and PR in two cases, which were related to rhinovirus, human bocavirus and respiratory syncytial virus (RSV) infection.
## Case reports
## Case 1
Patient 1 was an 8-year-old girl who presented with chest pain and dyspnea. She had a history of upper respiratory tract infection with fever, neck swelling and sore throat before these symptoms started. She had no history of asthma or foreign body aspiration, but did have a history of suspected abdominal trauma. Physical examination showed subcutaneous emphysema of the extremities and above the clavicles. Chest radiograph showed subcutaneous emphysema and pneumomediastinum. Also, thorax computed tomography (CT) indicated pneumomediastinum, interstitial emphysema and epidural emphysema. Pulmonary function tests and oxygen saturation were normal. Polymerase chain reaction (PCR) of the nasal secretions obtained on the fourth day was positive for human rhinovirus. Mycoplasma pneumoniae and Chlamydophila pneumoniae IgG and IgM were serologically negative. Antinuclear antibody (ANA) and anti-double stranded DNA were investigated to rule out connective tissue diseases and were found to be negative. α-1-Anti-trypsin was assessed for interstitial emphysema and was normal. The patient was given 100% oxygen therapy and also given antibiotics empirically. She recovered uneventfully and was discharged after 6 days.
## Case 2
Patient 2 was a 5-year-old boy who presented to the emergency department with a 7 day history of cough, sputum, sore throat and dyspnea, and a 1 day history of fever. He also had a history of recurrent bronchiolitis. On physical examination he had bilateral wheezing. He had tachypnea and required supplemental oxygen to maintain oxygen saturation >92%. Chest radiograph indicated suspected bilateral paracardiac and hilar infiltration. Thorax CT indicated pneumomediastinum, interstitial emphysema and infiltration of the lingular segment of the left upper lobe [fig_ref] Figure 2: Axial computed tomography showing interstitial emphysema within the right middle lobe, lingula... [/fig_ref].
He was vigorously treated with broad-spectrum antibiotics, inhalation bronchodilators, systemic corticosteroids, and highflow oxygen. PCR of nasal secretions obtained on the fifth day was positive for human bocavirus. M. pneumoniae and C. pneumoniae IgG and IgM were serologically negative. Skin prick test was negative and the patient was discharged with inhaled corticosteroid treatment because of history of recurrent bronchiolitis.
## Case 3
Patient 3 was a 12-year-old boy who presented to hospital with crackling under the skin and a 4 day history of cough, nasal congestion and rhinorrhea. He had no history of asthma or trauma. Physical examination indicated mild subcutaneous crepitus over the arms, and breath sounds were normal. Chest radiograph demonstrated pneumomediastinum with moderate amounts of subcutaneous air, and thorax CT indicated pneumomediastinum, subcutaneous emphysema and epidural emphysema [fig_ref] Figure 3: Axial computed tomography of the chest showing pneumomediastinum [/fig_ref]. Pulmonary function tests and oxygen saturation were normal. PCR of nasal secretion on the first day of treatment was negative, also M. pneumoniae and C. pneumoniae IgG and IgM were serologically negative. After 1 week, however, the second nasal secretion sample was positive for RSV. Oxygen was administered to facilitate the absorption of the mediastinal air and a few days later the symptoms completely resolved.
All nasal samples were tested using Multiplex-Polymerase Chain Reaction System (Epicentre Biotechnologies, Madison, WI, USA) for the presence of RSV, human rhinoviruses, human metapneumovirus, adenoviruses, influenza viruses A and B, human coronaviruses, parainfluenza viruses 1-4, enteroviruses, and human bocavirus.
# Discussion
Spontaneous pneumomediastinum is related to Valsalva maneuvers (e.g. coughing, vomiting, labor, sneezing, use of inhaled medication), severe bronchopulmonary infection (e.g. measles, M. pneumoniae), foreign body ingestion, esophageal rupture and dental surgery, although there may be no identifiable cause. Pneumomediastinum can also occur secondary to chest injury, mechanical ventilation and thoracic surgery. 4,5 Pneumomediastinum usually results from mechanisms that increase alveolar pressure, which leads to alveolar rupture. By direct extension, air enters the interstitial tissues and spreads easily into the mediastinal borders. Air in the mediastinal tissues may originate from the respiratory tract, such as after blunt or penetrating trauma to the facial bones, pharynx, trachea and mainstem bronchi. In this report patient 1 had a history of suspected abdominal mild trauma not associated with the pneumomediastinum. [bib_ref] Spontaneous pneumomediastinum in children: A literature review, Bullaro [/bib_ref] In spontaneous pneumomediastinum, PR develops when air travels along fascial planes from the posterior mediastinum or the retropharyngeal space through the neural foramens into the epidural space, which is not protected by a true fascial envelope. PR is becoming somewhat more often diagnosed as imaging techniques improve. Here, we described PR in two patients. Most recent surveys suggest that PR associated with spontaneous pneumothorax and pneumomediastinum with asymptomatic neurological status is self-limiting in 98% of cases. [bib_ref] Pathogenesis, diagnosis and management of pneumorrhachis, Oertel [/bib_ref] Two of the present patients were neurologically asymptomatic.
The association of spontaneous PR with pneumomediastinum is rare and restricted to very few cases in the literature. [bib_ref] Pathogenesis, diagnosis and management of pneumorrhachis, Oertel [/bib_ref] The association of spontaneous pneumomediastinum and human bocavirus, a recently described respiratory pathogen, has been newly reported in the literature. [bib_ref] Spontaneous pneumomediastinum as a complication in human bocavirus infection, Pekcan [/bib_ref] In the present patient 2, interstitial pneumonitis due to bocavirus infection may have been related to pneumomediastinum, but few cases of pneumomediastinum and PR associated with influenza and rhinovirus have been reported. [bib_ref] Spontaneous pneumomediastinum complicating pneumonia in children infected with the 2009 pandemic influenza..., Hasegawa [/bib_ref] [bib_ref] Pneumothorax, pneumomediastinum, subcutaneous emphysema and pneumorrhachis as complications of common flu, Patel [/bib_ref] In the present study, violent coughing in all three patients, which increases intrathoracic pressure, as well as probable rupture of a peripheral pulmonary alveolus by increased intra-alveolar pressure, preceded the development of pneumomediastinum and PR.
Respiratory syncytial virus is a leading cause of hospitalization in children <1 year old and premature infants, although it has been frequently detected in adults and children as a significant pathogen. The association of pneumomediastinum and subcutaneous emphysema with RSV infection has been reported by in an infant. [bib_ref] Pneumomediastinum and subcutaneous emphysema: An uncommon presentation of respiratory-syncytial virus infection in..., Dehmel [/bib_ref] In the present study we could not detect RSV because of low viral burden initially, but repeated nasal sample confirmed RSV.
Spontaneous pneumomediastinum is a self-limiting and benign condition that is frequently over0investigated and over0treated. [bib_ref] Pathogenesis, diagnosis and management of pneumorrhachis, Oertel [/bib_ref] [bib_ref] Spontaneous pneumomediastinum: A comparative study and review of the literature, Caceres [/bib_ref] Clear guidelines regarding the diagnostic and therapeutic interventions, however, are currently unavailable. In most cases, PR is asymptomatic and constitutes a radiological curiosity. The intra-spinal air is reabsorbed spontaneously and completely into the circulation. Asymptomatic patients and those with minimal neurological symptoms due to PR associated with pneumomediastinum or pneumothorax are efficiently managed conservatively with high-flow oxygen inhalation, which enhances reabsorption of the air by nitrogen washout, and further helps in symptomatic improvement in high-pressure PR, in which air under pressure has entered the epidural space. [bib_ref] Spontaneous pneumomediastinum in children: A literature review, Bullaro [/bib_ref] Thus PR should be treated based on etiology and severity of symptoms; surgery, antibiotics, foreign body removal from the trachea, and bronchodilators in chronic bronchitis will be used on a case-by-case basis. [bib_ref] Pneumorrhachis, pneumomediastinum, pneumopericardium and subcutaneous emphysema as complication of bronchial asthma, Manden [/bib_ref] In conclusion, spontaneous pneumomediastinum, pneumothorax and PR secondary to viral infection is a self-limiting and benign condition that should not be over-investigated or over-treated.
[fig] Figure 1: (a) Chest radiograph showing subcutaneous emphysema (long arrows) and pneumomediastinum (short arrows). (b) Axial computed tomography showing subcutaneous emphysema (arrowhead) and pneumomediastinum (white arrow). Also note the extracranial epidural emphysema (black arrow). [/fig]
[fig] Figure 2: Axial computed tomography showing interstitial emphysema within the right middle lobe, lingula of the left lung and lower lobe (white arrows), and pneumomediastinum. [/fig]
[fig] Figure 3: Axial computed tomography of the chest showing pneumomediastinum (white arrow), subcutaneous emphysema (arrowheads) and extracranial epidural emphysema (black arrow). [/fig]
|
Aggregation modeling of the influence of pH on the aggregation of variably charged nanoparticles
The aggregation of variably charged nanoparticles is usually induced by the changes in internal and external conditions, such as solution temperature, pH, particle size, van der Waals force, and electrostatic repulsion among particles. In order to explore the effect of pH on the aggregation of variable charge nanoparticles, this paper proposed an extended model based on the 3D on-lattice Cluster-Cluster Aggregation (CCA) model. The extended model successfully established the relationship between pH and sticking probability, and used Smoluchowski theory to calculate the aggregation rate of nanoparticles. The simulation results showed that: (1) the change of the aggregation rate of the variable charge nanoparticles with pH conforms to the Gaussian distribution, (2) the initial particle concentration has a significant effect on the aggregation rate of the nanoparticles, and (3) pH can affect the competition between van der Waals force and electrostatic repulsion between particles, thereby affecting the degree of openness of clusters. The research demonstrated the extended CCA model is valuable in studying the aggregation of the variably charged nanoparticles via transforming the corresponding influence factors into the influence on the sticking probability.The research on the aggregation of particles in the nano and micrometer size range is relevant in numerous processes, such as in the production of chemical materials and medicine, in environmental protection, and so forth 1-4 . The main factors affecting the aggregation of particles: pH 4-7 , solution temperature 7 , ionic strength 8,9 , long-range van der Waals forces 10 , gravity 11,12 , electrolyte concentration 12 , and electrolyte type 13 , particle size, the surface charge density of particles, etc. The researches showed that the surface charges of particles is mainly divided into variable and permanent surface charge 5 . The charge of the variably charged surface depends not only on the chemical surface, but also on the composition of the solution. The variably charged surface is a hydrated oxide surface (i.e. M-OH), and the H + in the solution with lower pH can combine with M-OH on the surface of particles, which leads to the increase in positive charges on the surface of particles. On the contrary, the OH − in higher pH solution can induce the dissociation of H + on the surface of particles, resulting in the increase in negative charges on the surface of particles. Therefore, there exists a pH value which makes the surface of the particles uncharged, the pH is called the point of zero charge (PZC) of the variably charged particle 5 . Li et al. proved that only the two DLVO forces, electrostatic and van der Waals forces should be considered in the study of the aggregation of variably charged particles 5,14 . As mentioned above, the electrostatic forces of variably charged particles depend on pH, so the aggregation of variably charged particles would be pH dependent.It is a great challenge for the researchers to precisely control the dynamic aggregation process in laboratory experiments. The aggregation model bridges the study of particle aggregation by computer simulation and laboratory experiments. There are some common aggregation models, such as the Diffusion-Limited Aggregation (DLA) model, the Reaction-Limited Aggregation (RLA) model, and the Cluster-Cluster Aggregation (CCA) model. Among those models, the Cluster-Cluster Aggregation (CCA) model is considered to be suitable for describing particle aggregation 15,16 . The CCA model takes the Monte Carlo simulation method as the core idea and was divided into Diffusion-Limited Cluster Aggregation (DLCA) model and Reaction-Limited Cluster Aggregation (RLCA) model by the probability of aggregation of particles after a collision, and the probability of aggregation can be expressed as the sticking probability[16][17][18]. The DLCA model assumes two particles (clusters) can stick together once they collide, which means the sticking probability is 1, which reflects the rapid OPEN
www.nature.com/scientificreports/ aggregation. And the RLCA model assumes the two particles (clusters) may not stick together after the collision, but stick together with a sticking probability less than 1, which reflects the slow aggregation. James and Sasikumar 6 studied the effects of solvent evaporation and pH on the gelation rate and gel structure based on the classic CCA model. However, in their experiments, many simplifications were performed on the evaporation and pH change of the solution, and the pH of the solution was limited to less than or equal to 7. And their research did not directly calculate the gelation rate, instead, the rate of gelation is reflected by the number of clusters in the system at the same time.
In this paper, an extended CCA model was proposed to study the influence of pH on the aggregation of variably charged nanoparticles. In this study, pH is the only variable, and the influence of other factors is not considered. In the extended CCA model, the sticking probability in the classic CCA model was modified as a function related to the point of zero charge (PZC)of variably charged nanoparticles and pH based on the research of J. James and R. Sasikumar. In order to reflect the change of the aggregation rate with pH more completely and accurately, the pH is set as a time parameter to ensure that the pH changes continuously throughout the simulation process. And the aggregation rate was calculated by Smoleuchowski theory. The specific simulation process and results are described as follows.
# Simulation methods
The diffusion probability. The diffusion process is consistent with the classic CCA model. According to the Stokes-Einstein formula, the initial diffusion coefficient of the single particle can be obtained by where K is the Boltzmann constant, T is the absolute temperature, µ is the viscosity of the solution, and R f is the hydrodynamic radius (larger than the geometric radius) of the single particle [bib_ref] Effect of the hydrodynamic radius of colloid microspheres on the estimation of..., Du [/bib_ref]. The diffusion coefficient of the clusters consisting of i single particles is [bib_ref] 3D simulation of the Cluster-Cluster Aggregation model, Li [/bib_ref] [bib_ref] Coupling effects of diffusive model and sticking model on aggregation kinetics of..., Xiong [/bib_ref] [bib_ref] Structure and kinetics of reaction-limited aggregation, Meakin [/bib_ref] where S is the mass of the selected cluster, which is expressed by the number of the single particles contained in the selected cluster, and γ is the diffusion exponent. A cluster is randomly selected to diffusion and its diffusion probability P move is calculated by where D max is the maximum diffusion coefficient for any clusters in the current system. A random number X uniformly distributed in [0,1] is generated and the selected cluster is moved only if X ≤ P move .
The sticking probability. If a collision occurs between two clusters (one consists of i single particles, and another consists of j single particles), they can stick together to form a new larger cluster with the sticking probability P ij where P 1 is the sticking probability of the single particle, σ is the sticking probability exponent. P ij is set to 1 if P ij ≥ 1. When a collision takes place, a random number Y uniformly distributed in [0,1] is generated and compared with the given P ij . The collision is considered effective only when Y ≤ P ij is verified. The sticking probability of particles is set as a function related to pH and PZC in the research of James and Sasikumar 6 :
The PZC was set as 7, and the influence of pH change caused by the evaporation of liquid medium on the particle aggregation rate and gel structure during the gelation process was discussed in the study of James and Sasikumar 6 . Although the difference of the evaporation rate of solute and solvent leads to the fluctuation of pH, the pH will never exceed 7. Therefore, the study did not include the situation when pH is greater than 7. In this paper, pH is the only variable, which means that the surface charge of variably charged particles is only related to pH 5 . So when the pH is equal to PZC, the net surface charge of the particles is zero, which makes the electrostatic repulsion force between the particles zero. At this time, once the particles collide, they will stick together, which corresponds to the rapid aggregation. When the pH is smaller than PZC, the number of positive charges on the surface of the particles will increase. Conversely, when the pH is greater than PZC, the number of negative charges on the surface of the particles will increase. In both cases, the electrostatic repulsion force among particles will increase, which will reduce the sticking probability, and lead to slow aggregation. In the classic CCA model, when P ij = 1, it corresponds to the rapid aggregation, and when P ij < 1, it corresponds to the slow aggregation. According to the above analysis and Eq. (2), the sticking probability of two particles (clusters) after the collision is modified to the relationship related to pH and PZC:
The modified Eq. (6) can not only reflect the change of the sticking probability when the pH is greater than PZC, but also the change of the sticking probability when pH is less than or equal to the PZC. When the initial www.nature.com/scientificreports/ sticking probability P 1 is set to 1, Eq. (6) includes both the rapid aggregation (pH = PZC) and the slow aggregation (pH = PZC). Similarly, a random number Z is generated between [0,1]. If Z ≤ P ij , the collision is considered an "effective collision", which means the two colliding clusters will stick together.
[formula] (1) D 0 = KT 6πµR f (2) D i = D 0 S γ (3) p move = D i D max (4) P ij = P 1 (ij) σ (5) P ij = P 1 10 (pH−PZC) .( [/formula]
## The aggregation rate of particles. smoluchowski modeled the aggregation rate of particles by calculat-
ing the rate of diffusion of particles toward a central reference particle [bib_ref] Detailed model of the aggregation event between two fractal clusters, Lattuada [/bib_ref] [bib_ref] Applications of fractals in soil and tillage research: A review, Perfect [/bib_ref]. Du 22 deduced the theoretical formula for calculating the aggregation rate constant in the rapid aggregation and the slow aggregation of charged particles based on the Smoluchowski theory. In the CCA model, the aggregation process can be regarded as a process in which a moving particle or cluster collides with another static particle or cluster,and aggregates together with a certain probability [bib_ref] Simulation of deformation and breakup of large aggregates in flows of viscous..., Higashitani [/bib_ref]. According to Smoluchowski theory, the circulation flux of collisions between particles is considered firstly. When the particle j is fixed, the flux expression for the particle i type of particle that can collide with particle j is 22 :
where n i is the concentration of i particles or clusters; R ij is the distance between two particles (clusters) when they collide. For a mono-disperse system, R ij is the diameter of the single particles. D i is the diffusion coefficient of the particle (or cluster) i. As mentioned above, particle j is fixed, so Eq. (7) is the number of collisions of particles per unit time and unit volume. For the rapid aggregation, each collision will lead to the aggregation. Therefore, the rapid aggregation rate constant of particles (or clusters) i can be deduced as follows:
According to Eqs. (1) and (2), Eq. (8) can be modified as where the value of γ (diffusion exponent) can be set to 0 in the CCA model [bib_ref] Coupling effects of diffusive model and sticking model on aggregation kinetics of..., Xiong [/bib_ref]. It is obvious from Eq. (2) when the value of γ is 0, the diffusion probability of the selected cluster is independent of the mass of clusters (the number of single particles contained in clusters). Therefore, Eq. (9) can be modified as:
Equation (10) represents the formula for the rapid aggregation rate obtained by the Smoluchowski theory combined with the characteristics of the CCA model. For the slow aggregation rate, the particles need to overcome the repulsive barrier after the collision, so the slow aggregation rate can be expressed as where W is the stability ratio. According to Eqs. [bib_ref] Computer simulation of colloidal aggregation induced by directionalism of long range van..., Xiong [/bib_ref] and (11), the slow aggregation rate can be expressed as:
It is obvious from Eqs. [bib_ref] Computer simulation of colloidal aggregation induced by directionalism of long range van..., Xiong [/bib_ref] and (12), the aggregation rate of particles or clusters is independent of the mass of clusters when the diffusion index is 0. The stability ratio W in Eqs. [bib_ref] A Brownian dynamics simulation to study the influence of gravity on the..., Li [/bib_ref] and (12) can be defined as the ratio of "the total number of collisions between particles" to "the effective number of collisions between particles" in the CCA model [bib_ref] Relating colloidal particle interactions to gel structure using Brownian Dynamic simulations and..., Mellema [/bib_ref] [bib_ref] Topological properties of diffusion limited aggregation and cluster-cluster aggregation, Meakin [/bib_ref]. For the rapid agglomeration, once the particles collide, they can agglomerate together, so the number of collisions between particles is equal to the effective number of collisions, which means W = 1 . For the slow agglomeration, because the repulsion force is considered, most collisions are reflective which enables the particle or cluster to skip from one side to another, and penetrate deeper within the aggregate to make a denser structure [bib_ref] Topological properties of diffusion limited aggregation and cluster-cluster aggregation, Meakin [/bib_ref] , so the number of collisions between particles is always greater than the "effective number of collisions", that is, W > 1 . Replacing 1/W with α which can be called attachment efficiency, the calculation of k slow can be expressed as:
In the Eqs. [bib_ref] Computer simulation of colloidal aggregation induced by directionalism of long range van..., Xiong [/bib_ref] and [bib_ref] Effect of gravity and electrolyte concentration on the fractal structure of colloidal..., Yuan [/bib_ref] , the R (the geometric radius of the single particle) should be smaller than the R f (the hydrodynamic radius of the single particle) [bib_ref] Effect of the hydrodynamic radius of colloid microspheres on the estimation of..., Du [/bib_ref] [bib_ref] Specific ion effects on particle aggregation induced by monovalent salts within the..., Oncsik [/bib_ref]. Du et al. [bib_ref] Effect of the hydrodynamic radius of colloid microspheres on the estimation of..., Du [/bib_ref] proved that the slow aggregation rate calculated by considering the hydrodynamic radius is about 8% lower than the theoretical value, and it has almost no effect on the fast aggregation rate. Therefore, in this paper, it is considered that R is equal to R f , without considering the specific values of R and R f for specific particle types. www.nature.com/scientificreports/
[formula] (7) J i = 4πR ij D i n i (8) k i = 4πR ij D i . (9) k i = 4πR ij ( KT 6πR f µ )S γ (10) k i = 4π(2R) KT 6πR f µ = 4KTR 3µR f . (11) k slow = k i W (12) k slow = k i W = 4KTR 3µWR f .(13) [/formula]
## Experimental parameters
As the above discussions, for variably charged particles, the aggregation process depends on pH when the electrolyte concentration, electrolyte type, and temperature of electrolyte solution remain unchanged. To accurately reflect the relationship between aggregation rate and pH, pH was set as a time-dependent parameter in this paper.
To simplify, the simulation only considered the influence of pH on particle aggregation and assumed that the electrolyte concentration remained constant throughout the simulation, namely, the ion concentration of the system remained constant. The change of pH was defined as:
where v is the change rate of pH and pH ( t 0 ) is the initial pH (when t = 0 ). The whole simulation was a timedependent process. After each cluster was randomly selected to move, the simulation time was increased by t 15 :
where N(t) is the number of clusters (including single particles) in the system at time t. However the influence of pH on the diffusion process is not clear, it is assumed that pH has the same effect on the diffusion of clusters of different sizes, so the diffusion index γ is set to 0. Referring to the research of James and Sasikumar 6 , the initial sticking probability P 1 is set to 1. The spherical single particle diameter d 1 = 100 nm, Brownian motion step length l B = d 1 , cubic side length L = 100d 1 and the initial particle concentration c is set to 0.01, 0.005, 0.0025 (the number of the single particles of the initial system equals to c * L 3 ). The simulation terminates when only one cluster remains in the system or the aggregation rate k is 0 (the aggregation no longer occurs in the system).
# Results and discussion
The influence of pH on the aggregation rate. The changes of aggregation rate with pH at different initial particle concentration and pH change rate are shown in [fig_ref] Figure 1: The change of aggregation rate, k, with pH for different initial particle... [/fig_ref]. According to the simulation results [fig_ref] Figure 1: The change of aggregation rate, k, with pH for different initial particle... [/fig_ref] , the changes of particle aggregation rate with pH conform to the Gaussian distribution. The simulation results are consistent with the conclusions obtained by Zhu 5 through experiments, so the simulation results are reliable. When the pH is infinitely close to PZC, the electrostatic repulsion force between the particles is very small, and may even be zero. Therefore, the aggregation between the particles can be regarded as "rapid aggregation", so the particles aggregation rate reaches the maximum. When pH increases or decreases along PZC, the surface charge of the variably charged particles increases, resulting in the increase of the electrostatic repulsion force between the particles, which leads to the decrease of the number of the effective collisions. The "slow aggregation" occurs between the particles, so the aggregation rate decreases. As shown in [fig_ref] Figure 1: The change of aggregation rate, k, with pH for different initial particle... [/fig_ref] , the aggregation rate reaches the maximum when pH is always slightly greater than PZC. The main reasons: in the small range where the pH is infinitely close to PZC from the left and right ends, the sticking probability P ij between colliding particles is infinitely close to or equal to 1, and the aggregation can almost be regarded as "rapid aggregation". In this range, α keeps increasing, and it can be seen from Eq. (13) that the aggregation rate k increases in this range. Therefore, the maximum aggregation rate k appears when the pH value is slightly higher than PZC. Noticeably, the main purpose of setting different pH change rates is to reflect the change of aggregation rate with pH in more detail and to avoid the case that pH is greater than PZC after the first t.
Compared with the theoretical prediction chart proposed by Zhu , the change of particles aggregation rate with pH is mainly divided into four stages. The main reason why the "1" stage is not obvious in the simulation results [fig_ref] Figure 1: The change of aggregation rate, k, with pH for different initial particle... [/fig_ref] is that the simulation time from the beginning ( t = 0 ) to the first t is relatively large, and pH changes rapidly from 0, resulting in the aggregation rate almost growth linearly, so the change process of "1" stage is not obvious. Referring to Eq. (6), when the pH value is far greater or smaller than PZC, the sticking probability particles will be very small. At this time, there is almost no aggregation between particles, so the "1" and "4" stages in would appear. According to the simulation results [fig_ref] Figure 1: The change of aggregation rate, k, with pH for different initial particle... [/fig_ref] , the aggregation rate changes sharply with pH in the "1" and "2" stages, while the "3" and "4" stages are relatively stable. The possible reason is that in the initial stage of aggregation, the concentration of particles is higher, and there are more collision opportunities between particles, which can increase the number of collisions and effective collisions to a certain extent. However, the sticking probability at this stage is still relatively small. Thus, the conflict among the sticking probability, the number of collisions, and the effective number of collisions leads to the violent fluctuation in the "1" and "2" stages.
Through this model, the experiment carried by Xu et al. [bib_ref] Impact of environmental conditions on aggregation kinetics of hematite and goethite nanoparticles, Xu [/bib_ref] in which they studied the effect of pH on the stability of hematite nanoparticles and goethite nanoparticles (PZC of hematite is 7.5, PZC of goethite is 7.8) was reproduced. The relationship of attachment efficiency [in Eqs. [bib_ref] Effect of gravity and electrolyte concentration on the fractal structure of colloidal..., Yuan [/bib_ref] and [bib_ref] Impact of environmental conditions on aggregation kinetics of hematite and goethite nanoparticles, Xu [/bib_ref] ] with pH values is shown in the . It can be found that when the pH of the suspension is less than 6 or greater than 10, the nanoparticle suspension is very stable because the electrostatic repulsion force on the surface of the nanoparticle is much greater than the van der Waals force. When the pH is between 6 and 10, the situation reverses, and aggregation occurs between the nanoparticles. It is obvious from that when pH is 7.5, the aggregation rate of hematite reaches the maximum, and when pH is 7.8, the aggregation rate of goethite reaches the maximum. The experimental results obtained in this study are basically consistent with those obtained by Xu et al., and are in line with the theory described above. Therefore, the model proposed in this paper can predict the aggregation rate of variably charged nanoparticles and predict the stability pH interval of variably charged nanoparticles.
The changes of the weighted average cluster size with time under different initial particle concentrations and pH change rates are illustrated in [fig_ref] Figure 4: A log-log plot of the weighted average cluster size, S [/fig_ref]. The change of weighted average cluster size can reflect the growth rate www.nature.com/scientificreports/ of clusters. As shown in [fig_ref] Figure 4: A log-log plot of the weighted average cluster size, S [/fig_ref] , the higher the initial particle concentration is, the earlier the particles start to aggregate and reach a stable state. At the same time and the same pH change rate, the weighted average cluster size increases quickly when the initial particle concentration increases. The results indicated that the effect of initial particle concentration on the aggregation rate cannot be ignored.
The influence of pH on cluster morphology. The final cluster structures under different pH conditions with the initial particle concentration of 0.01 are illustrated in [fig_ref] Figure 5: The final cluster structure under different solution pH in the extended 3D... [/fig_ref] [bib_ref] Topological properties of diffusion limited aggregation and cluster-cluster aggregation, Meakin [/bib_ref] [bib_ref] Diffusion-limited cluster aggregation: Impact of rotational diffusion, Jungblut [/bib_ref]. When pH is 5 and 9, the net charge on the surface of the particles is not zero, and there is electrostatic repulsion force between the particles. Corresponding to the slow aggregation, it easier for the single particles (or small clusters) to enter the interior of large clusters and form dense clusters [bib_ref] Three-dimensional off-lattice Monte Carlo simulations on a direct relation between experimental process..., Kim [/bib_ref]. According to Eq. (6), when pH is 5 and 9, the sticking probability is 0.01. However, due to the randomness of the CCA model, the fractal dimensions of clusters generated under the two conditions are slightly different. When pH is 6, the electrostatic repulsion force between the particles is between the values when pH is 5 (or pH = 9) and 7, so the sticking probability and the fractal dimension of the clusters formed are also between the values when pH is 5 (or pH = 9) and 7. . When pH is equal to 7, the weighted average cluster size increases fastest and reaches the stable state first. The changes of the weighted average cluster size at pH = 5 and pH = 9 are almost the same, and the change speed is the slowest. Therefore, it can be concluded that when pH value is equal to PZC, the rapid aggregation occurs between the particles, with the highest aggregation rate and the most open cluster structure; when the difference between pH and PZC is larger, the sticking probability between particles is smaller, the aggregation rate is also smaller, and the cluster structure is more compact.
# Conclusions
By employing the theory that the aggregation of the variably charged particles depends on pH under given temperature, electrolyte concentration, and electrolyte type, this paper proposed an extended model based on the classic 3D on-lattice CCA model. In the extended CCA model, the sticking probability was set as a function of pH and PZC, and the change of pH was set as a time-dependent parameter. Through this extended CCA model, this research successfully verified that the aggregation rate of variably charged particles with pH changes conforms to Gaussian distribution, which is consistent with the results obtained by traditional experimental methods. And this model can be used to calculate the aggregation rate and the fractal dimension at different pH, and the stability pH interval of variably charged nanoparticles, as well as to predict the trend of the aggregation rate of different types of variably charged nanoparticles with pH. Besides, these simulation results indicated that the closer the pH is to PZC, the tighter the cluster structure is. And the greater the initial concentration of particles, the greater the rate of aggregation. This paper is a useful attempt and supplement to the study of the aggregation process of nanoparticles and provides a new method and idea for studying the effect of pH on the aggregation of nanoparticles. The extended model could provide new technical support for related scholars to study the aggregation of nanoparticles. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
[fig] Figure 1: The change of aggregation rate, k, with pH for different initial particle concentration and pH change rate. [/fig]
[fig] Figure 2, Figure 3: Theoretical prediction diagram of the change of colloidal aggregation rate constant, k, with pH 5 . The attachment efficiency (1/W) of hematite and goethite nanoparticles in suspensions with different pH obtained by this model. [/fig]
[fig] Figure 4: A log-log plot of the weighted average cluster size, S(t), as a function of time t, for different initial particle concentrations and pH. [/fig]
[fig] Figure 5: The final cluster structure under different solution pH in the extended 3D on-lattice CCA model. The changes of the weighted average cluster size with time under different pH conditions are shown in [/fig]
[table] 6: P ij = P 1 10 −|pH−PZC| . [/table]
|
Tip-Enhanced Stokes–Anti-Stokes Scattering from Carbyne
Carbyne, a linear chain of carbon atoms, is the truly one-dimensional allotrope of carbon and the strongest known Raman scatterer. Here, we use tip-enhanced Raman scattering (TERS) to further enhance the Raman response of a single carbyne chain confined inside a double-walled carbon nanotube. We observe an increase of the anti-Stokes scattering by a factor of 3290 and a 22-fold enhancement of the anti-Stokes/Stokes ratio relative to far-field measurements. The power dependence of the anti-Stokes/Stokes ratio under TERS conditions is indicative of coherent Stokes−anti-Stokes scattering mediated by an excited phonon. The role of resonance effects and laser-induced heating are discussed and potential opportunities are outlined.
# ■ introduction
The Raman effect describes the inelastic scattering of light by quantized vibrational excitations (phonons) in matter. [bib_ref] A New Type of Secondary Radiation, Raman [/bib_ref] In the Stokes process, an incoming photon of energy ℏω L creates a phonon of energy ℏω ph and a red-shifted photon of energy ℏω S = ℏω L − ℏω ph , where ℏ is the reduced Planck constant and ω denotes frequency. On the other hand, in the anti-Stokes process a phonon is annihilated and the scattered photon is blueshifted to an energy ℏω aS = ℏω L + ℏω ph . Requiring the previous existence of a phonon in the system, the intensity of anti-Stokes scattering is proportional to the phonon occupation number n described by Bose−Einstein statistics, n = {exp[ℏω ph /(k B T)] − 1} −1 , where k B is the Boltzmann constant and T is the temperature. In contrast, Stokes scattering can also occur in the absence of a phonon, rendering the Stokes intensity proportional to n + 1.Taking into account the ω 4 dependence of Raman scattering, a result of classical field theory, 4 the anti-Stokes/Stokes intensity ratio can thus be written as
Because of the temperature dependence of eq 1, the anti-Stokes/Stokes ratio can be used for calibration-free thermometry. [bib_ref] Quantitative tradeoffs between spatial, temporal, and thermometric resolution of nonresonant Raman thermometry..., Mcgrane [/bib_ref] [bib_ref] Anti-Stokes Raman Scattering of Single Carbyne Chains, Tschannen [/bib_ref] However, there have been many reports of anti-Stokes/Stokes ratios deviating from this simple relation, for instance, in carbon nanotubes or transition metal dichalcogenides. [bib_ref] Raman Scattering and Anomalous Stokes-Anti-Stokes Ratio in MoTe2 Atomic Layers, Goldstein [/bib_ref] Such anomalous anti-Stokes/Stokes ratios are often related to material resonances, that is, an electronic transition matching the energy of Raman-scattered pho-tons.Alternatively, the same effect can also be induced by the presence of an external optical resonance, such as a cavity [bib_ref] Controlling Nonequilibrium Phonon Populations in Single-Walled Carbon Nanotubes, Steiner [/bib_ref] or a plasmon. [bib_ref] A study of local heating of molecules under surface enhanced Raman scattering..., Maher [/bib_ref] Another scenario leading to anomalous anti-Stokes/Stokes ratios occurs when the phonon population n is no longer dominated by thermal phonons but by phonons that are induced optically by a Stokes process. Such a regime can be reached by reducing the number of thermal phonons using a large ratio of phonon energy to thermal energy, ℏω ph ≫ k B T, or by increasing the number of optically induced phonons by using large laser powers or providing local field enhancement. Whenever the majority of anti-Stokes scattered photons originates from a Raman process involving the annihilation of a Stokes-induced phonon, meaning that the two scattering events share the same quantum of vibration, then the Stokes and anti-Stokes photons are correlated. [bib_ref] Correlation between the Stokes and anti-Stokes components in inelastic scattering of light, Klyshko [/bib_ref] This phenomenon has therefore been termed correlated Stokes−anti-Stokes (SaS) scattering and has been demonstrated for diamond [bib_ref] Stokes−anti-Stokes correlations in diamond, Kasperczyk [/bib_ref] and water 18 by means of photon arrival statistics violating classical descriptions.
Another manifestation of the same effect can be seen in the excitation power dependence of the anti-Stokes scattering intensity. For spontaneous Raman scattering, both Stokes and anti-Stokes signals scale linearly with excitation power (in the absence of laser heating) and the anti-Stokes/Stokes ratio remains constant. On the other hand, in the SaS regime the anti-Stokes scattering intensity assumes a quadratic dependence and, consequently, the anti-Stokes/Stokes ratio increases linearly with laser power. Such nonlinear anti-Stokes scattering intensities have been demonstrated for diamond using highpower ultrafast laser pulses, [bib_ref] Stokes−anti-Stokes correlations in diamond, Kasperczyk [/bib_ref] in twisted bilayer graphene tuned into resonance with the anti-Stokes emission, [bib_ref] Optical-Phonon Resonances with Saddle-Point Excitons in Twisted-Bilayer Graphene, Jorio [/bib_ref] and for molecules under surface-enhanced Raman scattering (SERS) conditions. The motivation for investigating SaS scattering stems from its potential as a source of entangled photons as well as possible applications in quantum information. [bib_ref] Correlation between the Stokes and anti-Stokes components in inelastic scattering of light, Klyshko [/bib_ref] [bib_ref] Stokes−anti-Stokes correlations in diamond, Kasperczyk [/bib_ref] [bib_ref] Temporal Quantum Correlations in Inelastic Light Scattering from Water, Kasperczyk [/bib_ref] [bib_ref] Optical-Phonon Resonances with Saddle-Point Excitons in Twisted-Bilayer Graphene, Jorio [/bib_ref] [bib_ref] Macroscopic non-classical states and terahertz quantum processing in room-temperature diamond, Lee [/bib_ref] [bib_ref] Stokes−Anti-Stokes Correlation in the Inelastic Scattering of Light by Matter and Generalization..., Parra-Murillo [/bib_ref] Candidate materials should host phonons with a large energy that are unlikely to be populated at room temperature. Moreover, a large Raman scattering cross section is beneficial to reduce the threshold in terms of local excitation power for the onset of the SaS phenomenon. A material which fulfils these requirements is carbyne, a long linear chain of sphybridized carbon atoms, which represents the truly onedimensional allotrope of carbon. [bib_ref] The Era of Carbon Allotropes, Hirsch [/bib_ref] The only Raman-active mode of carbyne, termed C-mode, is characterized by a phonon energy of around 1800 cm −1 (ℏω ph ≈ 9k B T at 293 K) and a Raman cross section per atom that exceeds that of any other known material by 2 orders of magnitude or more. [bib_ref] Raman Scattering Cross Section of Confined Carbyne, Tschannen [/bib_ref] Here, we combine the large phonon energy and Raman cross section of a carbyne chain confined inside a double-walled carbon nantoube with the strong local field enhancement afforded by tip-enhanced Raman scattering (TERS). We observe an anti-Stokes/Stokes ratio that is more than 2 orders of magnitude larger than expected based on thermal phonon statistics. Furthermore, the dependence of the anti-Stokes/ Stokes ratio on excitation power suggests that the Stokes and anti-Stokes photons are correlated. Our work thus establishes tip-enhanced Raman scattering of carbyne as a potential source of entangled photon pairs and extends the observation of the SaS effect to one-dimensional systems.
## ■ results and discussion
Carbyne chains are grown inside double-walled carbon nanotubes following the procedure reported by Shi et al. [bib_ref] Confined Linear Carbon Chains as a Route to Bulk Carbyne, Shi [/bib_ref] and dispersed on a thin glass coverslip using the recipe described in ref [bib_ref] Raman Resonance Profile of an Individual Confined Long Linear Carbon Chain, Heeg [/bib_ref]. To identify an individual confined carbyne chain (see [fig_ref] Figure 1: Tip-enhanced Raman scattering of confined carbyne [/fig_ref] , we capitalize on the nanoscale imaging capabilities of TERS, as described in Methods. [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref] shows the Raman spectrum of a single confined carbyne chain. The spectrum in [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref] represents a conventional far-field Raman measurement, while the spectrum in [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref] was recorded by TERS. The peaks at ±1834 cm −1 correspond to the C-mode of the carbyne chain, while the feature just below 1600 cm −1 is the G-mode signature of the encasing carbon nanotube structure, likely a bundle of doublewalled tubes judging from the topography map in [fig_ref] Figure 1: Tip-enhanced Raman scattering of confined carbyne [/fig_ref]. A small D-peak is visible above 1300 cm −1 , indicating the presence of some defects in the nanotubes.Looking at [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref] , it is striking that the Stokes and anti-Stokes signal strengths are within the same order of magnitude. This observation violates the usual intuition from thermal phonon statistics as expressed by eq 1. To understand the extent to which this large anti-Stokes/Stokes ratio can be attributed to the presence of the TERS tip, let us first discuss the far-field reference measurement depicted in [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref]. We fit both Stokes and anti-Stokes peaks with a Lorentzian and obtain the corresponding Raman signals I S and I aS by integration. Taking into account the frequency-dependent response of our setup (see Supporting Information), we find that the experimental far-field anti-Stokes/Stokes ratio amounts to I aS /I S = 3.4 × 10 −3 . On the other hand, eq 1 at room temperature (293 K) predicts a value of I aS /I S = 2.4 × 10 −4 , where we have considered that one order of ω cancels because we are counting photons rather than measuring intensities.Hence, there is a factor of 14 discrepancy between calculated and experimental anti-Stokes/Stokes ratio. This discrepancy can be decomposed into two separate Nano Letters pubs.acs.org/NanoLett Letter contributions, a material resonance and heating by the laser.
To disentangle those two effects, we record power-dependent far-field measurements to extract the rate of laser-induced heating (see Supporting Information). We find that laser irradiation using 0.65 mW excitation power increases the local temperature to 310 K, thus increasing the anti-Stokes/Stokes ratio by a factor of 1.6 compared to room temperature. The remaining factor of 9 that the experimental anti-Stokes/Stokes ratio deviates from eq 1 we ascribe to a resonance effect. This assignment can be understood by considering the band gap energy of the investigated carbyne chain. It is well-established that for confined carbyne, the band gap energy is linearly related to the C-mode frequency and dictated by the diameter of the encasing inner nanotube. [bib_ref] Electronic Band Gaps of Confined Linear Carbon Chains Ranging from Polyyne to..., Shi [/bib_ref] [bib_ref] Carbon Nanotube Chirality Determines Properties of Encapsulated Linear Carbon Chain, Heeg [/bib_ref] On the basis of the Raman shift of 1834 cm −1 observed in [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref] , we can calculate a band gap energy of 2.1 eV, which is larger than the excitation energy of 1.96 eV. As a consequence, a larger enhancement of the anti-Stokes scattering is expected due to the resonance Raman effect compared to Stokes scattering.Having evaluated the separate contributions of thermal and resonance effects on the far-field anti-Stokes/Stokes ratio, we now focus on the spectrum depicted in [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref] , which was acquired in the presence of the TERS tip. Relative to the farfield measurement in [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref] , a marked tip-induced enhancement of the Raman response is apparent. The Cmode Stokes peak is enhanced by a factor of 150, mediated by the large field enhancement provided by the used TERS probe geometry (see Methods). The enhancement of the anti-Stokes signal is even more pronounced, amounting to 3290, which implies a 22-fold increase of the anti-Stokes/Stokes ratio by the TERS tip.
We investigate the origin of this tip-induced enhancement of the anti-Stokes/Stokes ratio by performing power-dependent TERS measurements (see Methods). In [fig_ref] Figure 3: Power-dependent TERS measurements of confined carbyne [/fig_ref] , we plot the measured tip-enhanced anti-Stokes/Stokes ratio I aS /I S as a function of laser power P L . To exclude a corruption of the results by the TERS tip drifting away from the laser focus and the carbyne chain, the same measurement series is performed twice in a row. The data points represent the mean values of these measurements. The indicated standard deviations are small (below 5%), suggesting that the validity of our data is not affected by tip drift. Along with the data we plot fits of two different models. The solid black line is a fit of eq 1 with a power-dependent temperature, T = 293 K + C T P L , and an additional resonance factor R
[formula] I I R k CP exp (293 K ) . aS S aS 3 S 3 ph B T L ω ω ω = × × −ℏ + Ä Ç Å Å Å Å Å Å Å Å Å Å É Ö Ñ Ñ Ñ Ñ Ñ Ñ Ñ Ñ Ñ Ñ(2) [/formula]
Note that compared to eq 1, we have now explicitly replaced the (ω aS 4 /ω S 4 ) term by (ω aS 3 /ω S 3 ) in order for I aS /I S to reflect the ratio of anti-Stokes and Stokes photons rather than intensities.The resonance factor R is set to its experimentally determined value of 9, extracted from the farfield data in [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref]. In doing so, we are implying that R is not affected by the TERS tip. This assumption is justified because the plasmonic resonance of the tip is engineered such that the field enhancement is equivalent at the Stokes and anti-Stokes frequencies, as elaborated in Methods. Thus, the only free fitting parameter is the rate of laser-induced heating C T in the presence of the tip. For the fit represented by the red dashed line, an additional term linear in laser power is included in the model
[formula] I I R k CP C P exp (293 K ) . aS S aS 3 S 3 ph B T L SaS L ω ω ω = × × −ℏ + + l m o o n o o Ä Ç Å Å Å Å Å Å Å Å Å Å É Ö Ñ Ñ Ñ Ñ Ñ Ñ Ñ Ñ Ñ Ñ | } o õ o o(3) [/formula]
The coefficient C SaS describes the anti-Stokes signal arising from Stokes-induced anti-Stokes scattering, that is, it accounts for the correlated SaS process. Equation 3 is an approximation, which is generally valid for high-frequency phonons. [bib_ref] Stokes−Anti-Stokes Correlation in the Inelastic Scattering of Light by Matter and Generalization..., Parra-Murillo [/bib_ref] [bib_ref] Raman Studies of Carbon Nanostructures, Jorio [/bib_ref] While the coefficient C SaS can in principle be expressed in terms of fundamental constants, such as coupling and decay rates, [bib_ref] Stokes−Anti-Stokes Correlation in the Inelastic Scattering of Light by Matter and Generalization..., Parra-Murillo [/bib_ref] it is used here as a phenomenological fitting parameter. The assumption that the local temperature increases linearly with laser power implied in eqs 2 and 3 is discussed in Supporting Information.
Looking at the fit of eq 2 in [fig_ref] Figure 3: Power-dependent TERS measurements of confined carbyne [/fig_ref] , it is apparent that the exponential Boltzmann factor with a power-dependent temperature alone fails to reproduce the evolution of the anti-Stokes/ Stokes ratio with laser power. Moreover, the heating rate C T returned by this fit amounts to 316 K/mW, which appears unphysical given that the heating rate experimentally determined from the far-field measurements is an order of magnitude lower (see Supporting Information). Further evidence that speaks against such pronounced heating is given by the nanotube G-mode signature, as also detailed in the Supporting Information.
Let us now consider the fit of eq 3 (red dashed line in [fig_ref] Figure 3: Power-dependent TERS measurements of confined carbyne [/fig_ref] , which allows for the possibility of correlated SaS scattering. We observe an excellent agreement of this fit with the experimental data. The fitted coefficient C SaS , a measure of the significance of the SaS process, amounts to 6.3 × 10 −3 mW −1 , while the returned heating rate C T is close to zero. We emphasize that this does not imply that there is no heating at all, because assuming a heating rate of several tens of K/mW does not notably compromise the quality of the fit. However, it is clear that the fit of eq 3 is dominated by the SaS term and that this fit captures the evolution of the anti-Stokes/Stokes ratio with laser power much better than the purely thermal model. Importantly, it is only under TERS conditions that we observe this effect. In power-dependent far-field measurements, no significant contribution of SaS scattering is apparent Nano Letters pubs.acs.org/NanoLett Letter (see Supporting Information). Our data thus provides strong evidence for tip-mediated SaS scattering. To understand the observed effect, we recall that Raman scattering in general scales roughly with the fourth power of the local field enhancement. [bib_ref] Rigorous justification of the |E| 4 enhancement factor in surface enhanced Raman..., Ru [/bib_ref] Therefore, the pronounced local field enhancement provided by the used TERS probe geometry (see Methods) strongly increases the Stokes scattering and, as a result, the number of Stokes-induced phonons. The annihilation of these photogenerated phonons by anti-Stokes scattering results in the observed characteristic linear scaling of the anti-Stokes/Stokes ratio with excitation power [bib_ref] Stokes−anti-Stokes correlations in diamond, Kasperczyk [/bib_ref] [bib_ref] Optical-Phonon Resonances with Saddle-Point Excitons in Twisted-Bilayer Graphene, Jorio [/bib_ref] [bib_ref] Population Pumping of Excited Vibrational States by Spontaneous Surface-Enhanced Raman Scattering, Kneipp [/bib_ref] [bib_ref] Vibrational Pumping in Surface Enhanced Raman Scattering (SERS), Maher [/bib_ref] and imposes a correlation between Stokes and anti-Stokes photons. [bib_ref] Correlation between the Stokes and anti-Stokes components in inelastic scattering of light, Klyshko [/bib_ref] [bib_ref] Temporal Quantum Correlations in Inelastic Light Scattering from Water, Kasperczyk [/bib_ref] [bib_ref] Stokes−Anti-Stokes Correlation in the Inelastic Scattering of Light by Matter and Generalization..., Parra-Murillo [/bib_ref] We emphasize, however, that the local field enhancement present in our experiment is not sufficient to trigger the SaS phenomenon in any material. This is exemplified by the absence of a notable anti-Stokes G-peak in [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref] , which indicates that for the carbon nanotubes no significant SaS scattering is occurring. Similarly, for graphene, we observe that the TERS tip increases the G-mode anti-Stokes/Stokes ratio only by a factor of 3.1 (see Supporting Information), in stark contrast to the 22-fold enhancement measured for carbyne. The fact that SaS scattering, under identical TERS conditions, only occurs for carbyne and not for other carbon nanomaterials clearly highlights that the SaS phenomenon is related to the intrinsic properties of carbyne. Specifically, it is the high phonon energy and the large Raman scattering cross section of carbyne that provide a much lower threshold in terms of local excitation power for the onset of SaS scattering compared to other materials. Finally, let us briefly contrast TERS, which is used in our experiments to transition the Raman scattering of carbyne into the SaS regime, with SERS, which in principle is able to provide similar conditions with even stronger local field enhancements. [bib_ref] Surface Enhanced Raman Scattering Enhancement Factors: A Comprehensive Study, Ru [/bib_ref] A shortcoming of commonly used SERS substrates such as roughened metal surfaces or colloidal aggregates is that they give rise to spectra of limited reproducibility regarding intensity and appearance. [bib_ref] Electronic and vibrational surface-enhanced Raman scattering: from atomically defined Au (111) and..., Inagaki [/bib_ref] These limitations originate from the heterogeneity of the plasmonic resonances present on the substrate as well as the stochastic placement of individual sample molecules with respect to local electromagnetic hotspots. On the other hand, TERS probes like the one used in this study can be fabricated with a high degree of reproducibility and a precisely tuned plasmonic resonance behavior (see Methods). Moreover, the positioning of the TERS tip and the associated field intensity hotspot relative to the sample can be precisely controlled. Lastly, direct contact between sample and plasmonic nanostructure is avoided in TERS but standard in typical SERS experiments. This direct contact with the metal can affect the electronic structure of the sample and thus alter the resonance conditions in an unpredictable way. [bib_ref] Present and Future of Surface-Enhanced Raman Scattering, Langer [/bib_ref] To summarize, TERS provides a level of control that SERS cannot offer, which is of particular importance for reproducible single-molecule studies as performed in the present work.
## ■ conclusion and outlook
In summary, we have reported an anomalously large anti-Stokes/Stokes ratio in the tip-enhanced Raman spectrum of confined carbyne. This observation is in part due to the generation of correlated SaS photon pairs, as evident from the linear excitation power dependence of the anti-Stokes/Stokes ratio under TERS conditions. Another sizable contribution to the measured anti-Stokes/Stokes ratio arises from the resonance Raman effect enhancing the anti-Stokes emission more efficiently than the Stokes emission, which we have demonstrated using far-field data as a reference.
From the theory of resonance Raman scattering in solids, the anti-Stokes emission is expected to be largest when in resonance with an electronic transition, that is, for excitation energies that are lower than the band gap energy by exactly the energy of the phonon that is annihilated in the anti-Stokes process.In the present study, the investigated carbyne chain has a phonon energy of roughly 230 meV and a band gap of 2.1 eV. Hence, the excitation energy of 1.96 eV used in our experiments presumably does not fully exploit the potential of the Raman resonance effect in enhancing the anti-Stokes/ Stokes ratio. Optimizing the resonance conditions can increase the efficiency of the SaS process, as demonstrated for twisted bilayer graphene. [bib_ref] Optical-Phonon Resonances with Saddle-Point Excitons in Twisted-Bilayer Graphene, Jorio [/bib_ref] Moreover, a recent theoretical study suggests that the correlations between Stokes and anti-Stokes photons can be enhanced by modifying the statistical properties of the incident light. [bib_ref] Enhancement of Stokes-anti-Stokes correlations by the classical incoherent incident light, Shishkov [/bib_ref] In terms of potential applications, SaS scattering has been investigated as a write-read protocol for solid state quantum memory, where a Stokes-induced phonon stores information that can be read out by an anti-Stokes process during the lifetime of the phonon. [bib_ref] Macroscopic non-classical states and terahertz quantum processing in room-temperature diamond, Lee [/bib_ref] In our experiments, the transition of the Raman scattering from carbyne into the SaS regime is mediated by the plasmonic field enhancement afforded by a TERS probe. While TERS is arguably a nontrivial experimental technique compared to SERS, it allows for a precise tuning of the scattering enhancement by varying the tip−sample distance. Thereby, the correlated character of the SaS photons can potentially be tuned continuously from classical to quantum. [bib_ref] Correlation between the Stokes and anti-Stokes components in inelastic scattering of light, Klyshko [/bib_ref] [bib_ref] Temporal Quantum Correlations in Inelastic Light Scattering from Water, Kasperczyk [/bib_ref] [bib_ref] Stokes−Anti-Stokes Correlation in the Inelastic Scattering of Light by Matter and Generalization..., Parra-Murillo [/bib_ref] ■ METHODS TERS Measurements. Our home-built TERS setup is based on an inverted optical microscope with a scanning probe microscope on top. A radially polarized laser beam (excitation energy 1.96 eV) is strongly focused by a high numerical aperture oil-immersion objective. Positioning a TERS probe into the laser focus generates a nanoscale excitation source for Raman scattering. The TERS probe is attached to a piezoelectric tuning fork and its distance from the sample surface is controlled by a shear-force feedback system. The backscattered light is collected by the same objective lens and analyzed by a CCD-equipped spectrometer after passing a notch filter that removes the Rayleigh component. A motorized neutral density filter wheel placed in the beam path allows for varying the excitation power while the TERS probe is engaged. All power values reported in this work were measured before the back aperture of the objective. For imaging, the spectral region of carbyne's C-mode Stokes peak is isolated by a narrow bandpass filter and detected by an avalanche photodiode while the sample is raster-scanned with a piezo stage. Optical and topographic information is acquired simultaneously (see [fig_ref] Figure 1: Tip-enhanced Raman scattering of confined carbyne [/fig_ref]. The integration times are 25 ms/pixel for imaging and 5 s for full TERS spectra (up to 900 s for far-field spectra). An illustration of our TERS setup is provided in the Supporting Information. More details about the principles and experimental implementation of TERS can be found in ref 39.
TERS Probe. In this work we employ a plasmon-tunable tip pyramid (PTTP) made from gold as TERS probe. This probe geometry consists of a micropyramidal frustum with a Nano Letters pubs.acs.org/NanoLett Letter nanopyramidal end. The size of the nanopyramid is set by the fabrication process to 470 nm, such that the tip supports a localized surface plasmon resonance monopolar mode at the excitation wavelength of our laser. As a consequence, the PTTP probe provides a stronger field enhancement at its apex compared to regular gold micropyramids, [bib_ref] Highly reproducible near-field optical imaging with sub-20-nm resolution based on template-stripped gold..., Johnson [/bib_ref] resulting in the large signal enhancement observed in [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref]. Importantly, the plasmonic resonance of the PTTP probe is not only centered by design at the excitation energy of our laser but is also symmetric to within 0.3 eV from the resonance peak maximum. [bib_ref] Optical Properties of Plasmon-Tunable Tip Pyramids for Tip-Enhanced Raman Spectroscopy, Miranda [/bib_ref] Equal field enhancement can therefore be expected at the Stokes and anti-Stokes frequencies of carbyne. Consequently, the resonance factor R in eqs 2 and 3 does not differ from its far-field value obtained from [fig_ref] Figure 2: Raman spectroscopy of a single confined carbyne chain [/fig_ref]. The plasmonic spectral shaping effect, [bib_ref] Electronic and vibrational surface-enhanced Raman scattering: from atomically defined Au (111) and..., Inagaki [/bib_ref] [bib_ref] Plasmonic photoluminescence for recovering native chemical information from surface-enhanced Raman scattering, Lin [/bib_ref] which typically needs to be considered when evaluating anti-Stokes/Stokes ratios in plasmon-enhanced spectroscopy, does therefore not play a role in the TERS experiments presented in this work.
■ ASSOCIATED CONTENT
[fig] Figure 1: Tip-enhanced Raman scattering of confined carbyne. (a) Optical image of the C-mode scattered Stokes photons. The full width at half-maximum of the intensity extracted along the white line indicates a spatial resolution of 20 nm. (b) Topographic image acquired simultaneously with the optical information shown in (a). The length of the scale bars represents 250 nm. [/fig]
[fig] Figure 2: Raman spectroscopy of a single confined carbyne chain. (a) Far-field Raman spectrum. (b) Tip-enhanced Raman spectrum of the same carbyne chain. The power used for both measurements is 0.65 mW. The anti-Stokes spectra in (a) and (b) are scaled by a factor of 70 and 7, respectively. [/fig]
[fig] Figure 3: Power-dependent TERS measurements of confined carbyne. The data points show the anti-Stokes/Stokes ratio as a function of laser power P L . A correction for the frequency-dependent sensitivity of our setup has been applied. The solid black line is a fit of eq 2, the red dashed line is a fit of eq 3. The parameters returned by the fits are given in the main text. [/fig]
|
Solitary Metastasis of Hepatocellular Carcinoma to the Rectus Abdominis 13 Years After the Initial Treatment
Solitary muscle metastasis of hepatocellular carcinoma (HCC) is extremely rare, and late metastasis is also rare. We present a 59year-old man who had received initial treatment for HCC 13 years previously. Ultrasonography revealed a tumor between the abdominal wall and the liver surface. Tumor resection was performed with suspected intrahepatic metastasis or abdominal wall metastasis of HCC, and the tumor was found to be within the rectus abdominis without an association with the liver. Histologically, the resected material was confirmed to be a muscle metastasis of HCC. We discuss the management of muscle metastasis of HCC.Pathologic examination showed the proliferation of atypical cells with eosinophilic vesicles presenting a pseudo-tubular pattern and immunohistochemically stained with glypican-3, which is compatible with poorly to moderately differentiated HCC. The pathology ACG Case Rep J 2021;8:e00607.
# Introduction
Hepatocellular carcinoma (HCC) is a common primary hepatic cancer. The most frequent site of extrahepatic metastases is the lungs, followed by the lymph nodes, bones, and adrenal glands. [bib_ref] Clinical features of hepatocellular carcinoma with extrahepatic metastases, Natsuizaka [/bib_ref] [bib_ref] Clinical features and prognosis of patients with extrahepatic metastases from hepatocellular carcinoma, Uka [/bib_ref] [bib_ref] Hepatocellular carcinoma with extrahepatic metastasis: Clinical features and prognostic factors, Uchino [/bib_ref] However, skeletal muscle metastasis is rare, with a reported incidence rate of 0%-1.5%, and solitary muscle metastasis is extremely rare. [bib_ref] Clinical features of hepatocellular carcinoma with extrahepatic metastases, Natsuizaka [/bib_ref] [bib_ref] Clinical features and prognosis of patients with extrahepatic metastases from hepatocellular carcinoma, Uka [/bib_ref] [bib_ref] Patterns and clinicopathologic features of extrahepatic recurrence of hepatocellular carcinoma after curative..., Yang [/bib_ref] [bib_ref] The psoas muscle as an unusual site for metastasis of hepatocellular carcinoma:..., Wu [/bib_ref] [bib_ref] Muscle metastasis from hepatocellular carcinoma, Michalaki [/bib_ref] [bib_ref] Solitary metastasis to the intercostal muscle from hepatocellular carcinoma: A case report, Furumoto [/bib_ref] [bib_ref] Solitary muscle metastasis of hepatocellular carcinoma to the biceps femoris muscle with..., Orita [/bib_ref] Most recurrences occur relatively early after initial treatment; the reported median duration between initial treatment and extrahepatic metastases is 23.2 months. [bib_ref] Extrahepatic spread from hepatocellular carcinoma: Who are candidates for aggressive anti-cancer treatment?, Ishii [/bib_ref] Late metastasis more than 10 years after initial treatment is rare. [bib_ref] Late solitary pelvic metastasis of hepatocellular carcinoma mimicking alpha-fetoprotein-producing gynaecologic tumour, Kim [/bib_ref] [bib_ref] Late retroperitoneal recurrence of hepatocellular carcinoma 12 years after initial diagnosis, Wong [/bib_ref]
## Case report
The patient was a 59-year-old man who had neither hepatitis B virus nor hepatitis C virus infection, with no chronic liver disease or risk factor of HCC. At age 46, he was diagnosed with moderately differentiated HCC and underwent radiofrequency ablation (RFA) twice and transcatheter arterial embolization. After that, he had undergone regular follow-up once in 6 months with abdominal ultrasonography. At age 59, a 24.8 3 14.0-mm tumor was found between the abdominal wall and the surface of liver segment 5. Subsequent contrast-enhanced computed tomography (CT) imaging revealed that the lesion was contrastenhanced in the early phase, and the contrast was washed out in the late phase [fig_ref] Figure 2: Contrast-enhanced computed tomography revealed a 24 [/fig_ref]. No other lesions indicating metastases were observed.
Laboratory investigations revealed a normal blood cell count, normal liver enzyme levels, and coagulation. Although the serum a-fetoprotein (AFP) level was normal at 4.5 ng/mL, the protein induced by the vitamin K absence/antagonist-II (PIVKA-II) level was high at 93 mAU/mL. Subsequently, intrahepatic metastasis or abdominal wall metastasis of HCC was established as a differential diagnosis, and tumor resection was planned. During surgery, the tumor was found to be located within the rectus abdominis, not continuous with the liver. The tumor was resected with a margin of approximately 1 cm, surrounding some parts of the rectus abdominis . There was a fistula cranial to the tumor, which seemed to be the needle tract of the former RFA, but it was evidently located away from the resected tumor.
is the same as the initial HCC, and muscle metastasis of HCC to the rectus abdominis was confirmed. Three months after the surgery, the serum PIVKA-II level decreased to 20 mAU/mL. Three years after the surgery, there was no evidence of recurrence.
# Discussion
Skeletal muscle metastasis of HCC is rare, with the reported incidence rate of 0%-1.5%. [bib_ref] Clinical features of hepatocellular carcinoma with extrahepatic metastases, Natsuizaka [/bib_ref] [bib_ref] Clinical features and prognosis of patients with extrahepatic metastases from hepatocellular carcinoma, Uka [/bib_ref] [bib_ref] Patterns and clinicopathologic features of extrahepatic recurrence of hepatocellular carcinoma after curative..., Yang [/bib_ref] Although the reason remains unclear, this rarity may be due to several factors, including muscle motion resulting in mechanical tumor destruction, inhospitable muscle pH, and muscles' ability to remove tumorproduced lactic acid. [bib_ref] Muscle metastasis of carcinoma, Molina-Garrido [/bib_ref] In our case, the tumor was located away from the scar of the former RFA; hence, we consider it to be hematogenous or lymphogenous metastasis to the rectus abdominis rather than needle track seeding after RFA, whose reported incidence rate is 0%-1.1%. [bib_ref] Radiofrequency ablation for subcapsular hepatocellular carcinoma, Poon [/bib_ref] [bib_ref] Seeding after radiofrequency ablation of hepatocellular carcinoma in patients with cirrhosis: A..., Latteri [/bib_ref] [bib_ref] Long-term effectiveness of radiofrequency ablation for solitary small hepatocellular carcinoma: A retrospective..., Francica [/bib_ref] Solitary muscle metastasis of HCC is extremely rare among all muscle metastases, with only 4 cases being reported. [bib_ref] The psoas muscle as an unusual site for metastasis of hepatocellular carcinoma:..., Wu [/bib_ref] [bib_ref] Muscle metastasis from hepatocellular carcinoma, Michalaki [/bib_ref] [bib_ref] Solitary metastasis to the intercostal muscle from hepatocellular carcinoma: A case report, Furumoto [/bib_ref] [bib_ref] Solitary muscle metastasis of hepatocellular carcinoma to the biceps femoris muscle with..., Orita [/bib_ref] In all 4 cases, surgical resection was performed. In 2 of those cases, no recurrence was detected in the follow-up period. In 1 case, a pelvic lesion was detected 2 months after the surgery. In another case, intrahepatic recurrence was detected 10 months after the surgery but could be controlled with transcatheter arterial chemoembolization, and recurrence was no longer observed after that. In our case, tumor resection was performed, and no recurrence was detected 3 years after the surgery. Although the management of solitary extrahepatic metastasis remains controversial, surgical resection may be effective for patients with 1 or 2 isolated extrahepatic metastases who exhibit good hepatic functional reserve and general performance status and successful treatment of intrahepatic HCC. [bib_ref] Efficacy of surgical resection in management of isolated extrahepatic metastases of hepatocellular..., Chan [/bib_ref] Late recurrence of HCC after initial treatment is rare, and there are only 2 previous reports of extrahepatic recurrence occurring after 10 years. [bib_ref] Late solitary pelvic metastasis of hepatocellular carcinoma mimicking alpha-fetoprotein-producing gynaecologic tumour, Kim [/bib_ref] [bib_ref] Late retroperitoneal recurrence of hepatocellular carcinoma 12 years after initial diagnosis, Wong [/bib_ref] In the first case, the tumor of the right kidney was revealed by CT 12 years after the initial surgery, and surgical resection was performed for suspected renal cell carcinoma. In the second case, the tumor of the right pelvis was revealed by CT and magnetic resonance imaging 28 years after the initial surgery. Hence, surgical resection was performed after the suspicion of degenerative uterine myoma or ovarian fibroma. Late recurrent lesions tend to be diagnosed as a wrong primary tumor because of the rarity of late metastasis of HCC.
In our case, the PIVKA-II level was elevated, whereas the AFP level was normal. Among the 4 cases of solitary muscle metastases previously reported, the serum AFP level did not increase in 2 cases. [bib_ref] The psoas muscle as an unusual site for metastasis of hepatocellular carcinoma:..., Wu [/bib_ref] [bib_ref] Muscle metastasis from hepatocellular carcinoma, Michalaki [/bib_ref] [bib_ref] Solitary metastasis to the intercostal muscle from hepatocellular carcinoma: A case report, Furumoto [/bib_ref] [bib_ref] Solitary muscle metastasis of hepatocellular carcinoma to the biceps femoris muscle with..., Orita [/bib_ref] However, in 1 case, PIVKA-II was elevated. [bib_ref] Solitary muscle metastasis of hepatocellular carcinoma to the biceps femoris muscle with..., Orita [/bib_ref] It has been reported that the PIVKA-II level may be a predictive marker for extrahepatic HCC metastases, regardless of AFP level. [bib_ref] Protein induced by vitamin K absence or antagonist-II production is a strong..., Bae [/bib_ref] It is important to measure both AFP and PIVKA-II levels to diagnose recurrence and metastasis of HCC. In contrast to CT images, the most common pattern of skeletal muscle metastasis is a focal intramuscular mass with homogeneous contrast enhancement. [bib_ref] CT imaging features of skeletal muscle metastasis, Ong [/bib_ref] Among the previously reported cases of muscle metastases of HCC, the common pattern was heterogeneous enhancement in both early and late phases. [bib_ref] The psoas muscle as an unusual site for metastasis of hepatocellular carcinoma:..., Wu [/bib_ref] [bib_ref] Solitary metastasis to the intercostal muscle from hepatocellular carcinoma: A case report, Furumoto [/bib_ref] [bib_ref] Hepatocellular carcinoma with disseminated skeletal muscle metastasis, Rahim [/bib_ref] In our case, the lesion was contrast-enhanced in the early phase, and the contrast was washed out in the late phase, which was compatible with HCC. This finding may be occasionally beneficial in diagnosing HCC metastasis to the skeletal muscle.
We present the case of late solitary muscle metastasis of HCC. It is important to recognize that HCC can metastasize to skeletal muscle even long time after initial treatment. Although it is difficult to distinguish between solitary metastasis of HCC and another kind of tumor, laboratory examination and CT imaging may be useful for diagnosis.
# Disclosures
Author contributions: R. Kato wrote the article. A. Sakamoto, T. Noguchi, and S. Matsuda revised the article for intellectual content. H. Terajima approved the final article. A. Sakamoto is the article guarantor.
Financial disclosure: None to report.
Previous presentation: This case was presented at the Kinki Surgical Association; September 2, 2017; Kyoto, Osaka.Informed consent was obtained for this case report.
Received July 22, 2020; Accepted February 16, 2021
[fig] Figure 1: (A) Ultrasonography revealed a 24.8 3 14.0-mm tumor between the abdominal wall and the surface of liver segment 5. (B) Color Doppler ultrasonography showed blood flow in the tumor. [/fig]
[fig] Figure 2: Contrast-enhanced computed tomography revealed a 24.6 3 16.9-mm tumor on the surface of liver segment 5 that was contrastenhanced in (A) the early phase and the contrast was washed out in (B) the late phase. [/fig]
|
Longitudinal structural brain development and externalizing behavior in adolescence
Background: Cross-sectional studies report relations between externalizing behavior and structural abnormalities in cortical thickness of prefrontal regions and volume reductions in subcortical regions. To understand how these associations emerge and develop, longitudinal designs are pivotal. Method: In the current longitudinal study, a community sample of children, adolescents and young adults (N = 271) underwent magnetic resonance imaging (MRI) in three biennial waves (680 scans). At each wave, aspects of externalizing behavior were assessed with parentreported aggression and rule-breaking scores (Child Behavior Checklist), and self-reported aggression scores (Buss-Perry Aggression Questionnaire). Regions of interest (ROIs) were selected based on prior research: dorsolateral prefrontal (dlPFC), orbitofrontal (OFC), anterior cingulate cortex (ACC), insula, and parahippocampal cortex, as well as subcortical regions. Linear mixed models were used to assess the longitudinal relation between externalizing behavior and structural brain development. Structural covariance analyses were employed to identify whether longitudinal relations between ROIs (maturational coupling) were associated with externalizing behavior. Results: Linear mixed model analyses showed a negative relation between parent-reported aggression and right hippocampal volume. Moreover, this longitudinal relation was driven by change in hippocampal volume and not initial volume of hippocampus at time point 1. Exploratory analyses showed that stronger maturational coupling between prefrontal regions, the limbic system, and striatum was associated with both low and high externalizing behavior. Conclusions: Together, these findings reinforce the hypothesis that altered structural brain development coincides with development of more externalizing behavior. These findings may guide future research on normative and deviant development of externalizing behavior.
# Introduction
Adolescence, the transition period between childhood and adulthood, is marked by substantial cognitive, affective, and social development [bib_ref] Heightened stress responsiveness and emotional reactivity during pubertal maturation: Implications for psychopathology, Dahl [/bib_ref]. An interesting adolescent-specific pattern concerns the increase in risk-taking, sensation seeking, and novelty seeking, which is often interpreted as a normative pattern in the path toward autonomy and identity development [bib_ref] Annual Research Review: Neural contributions to risk-taking in adolescence-developmental changes and individual..., Crone [/bib_ref] [bib_ref] Self-development: Integrating cognitive, socioemotional, and neuroimaging perspectives, Pfeifer [/bib_ref]. At the same time, deviant behavior such as substance abuse, aggression, and delinquency emerge in adolescence in a subset of individuals [bib_ref] Research review: Evaluating and reformulating the developmental taxonomic theory of antisocial behaviour, Fairchild [/bib_ref] [bib_ref] Male antisocial behaviour in adolescence and beyond, Moffitt [/bib_ref] , but why and how this occurs is not yet well understood. Although research has shown consistent patterns in developmental changes in brain structure across adolescence [bib_ref] Development of subcortical volumes across adolescence in males and females: A multisample..., Herting [/bib_ref] [bib_ref] Structural brain development between childhood and adulthood: Convergence across four longitudinal samples, Klein [/bib_ref] [bib_ref] Development of the cerebral cortex across adolescence: A multisample study of interrelated..., Tamnes [/bib_ref] , few studies examined how the developmental pathway to externalizing behavior is associated with changes in brain development.
Longitudinal studies have demonstrated that adolescent brain development is associated with continuing changes in cortical and subcortical gray matter [bib_ref] Typical development of basal ganglia, hippocampus, Wierenga [/bib_ref]. Multisample studies have confirmed that the overall patterns are comparable across datasets and cultures, providing consistent evidence for cortical gray matter reductions across adolescence, continuing into young adulthoodwith the speed of change being region-dependent . A crucial, but poorly understood question concerns how these developmental patterns are related to individual differences in behavioral development, which can only be directly examined using longitudinal designs. Although longitudinal studies do not allow for the test of causality, they do provide a better understanding of how patterns coincide within individuals, and how early markers predict later behavioral outcomes [bib_ref] Longitudinal modeling in developmental neuroimaging research: Common challenges, and solutions from developmental..., King [/bib_ref].
Our current understanding of the neurobiological correlates of externalizing behavior is, however, mainly derived from cross-sectional studies comparing clinical samples to healthy controls. These studies showed that externalizing behavior in adolescents (including aggression, psychopathy, and conduct problems) is related to volumetric reductions in prefrontal regions, such as the orbitofrontal (OFC), dorsolateral prefrontal (dlPFC) and anterior cingulate (ACC) cortices (for a meta-analysis, see [bib_ref] Prefrontal structural and functional brain imaging findings in antisocial, violent, and psychopathic..., Yang [/bib_ref] , and also to volumetric reductions in subcortical regions, like the striatum, amygdala, and hippocampus [bib_ref] A systematic review and meta-analysis of neuroimaging in oppositional defiant disorder (ODD)..., Noordermeer [/bib_ref] [bib_ref] Cortical and subcortical abnormalities in youths with conduct disorder and elevated callous-unemotional..., Wallace [/bib_ref]. Externalizing problems, like many other psychiatric symptoms, exist on a continuum in the general population [bib_ref] The national institute of mental health research domain criteria and clinical research..., Garvey [/bib_ref]. Instead of using a categorical approach, which is defined by cut-offs between adaptive and maladaptive behavior (e.g., diagnostic criteria), a dimensional approach of psychopathology may better capture the variation between mental health and mental illness and may therefore provide novel insights in brain-behavior association across development [bib_ref] A neurodevelopmental perspective on the research domain criteria (RDoC) framework, Casey [/bib_ref].
There are a few longitudinal studies that examined the relation between structural brain development and externalizing behavior in community samples. In a study including children and adolescents (6-18 years), [bib_ref] Cortical thickness, cortico-amygdalar networks, and externalizing behaviors in healthy children, Ameis [/bib_ref] showed that externalizing behavior was negatively associated with cortical thickness in OFC and cingulate cortex, but they did not find a relation with hippocampus or amygdala volume. Findings from a three-wave longitudinal study showed that groups of adolescents with distinct developmental patterns of externalizing behavior showed different developmental structural brain changes, such that a group with desisting conduct problems showed an attenuation of the typical pattern of cortical thinning in dlPFC and ACC compared to groups with stable low or intermediate levels of externalizing behavior [bib_ref] Trajectories of adolescent conduct problems in relation to cortical thickness development: A..., Oostermeijer [/bib_ref]. Furthermore, this 'desisting' group showed an exaggeration of growth in hippocampal volume [bib_ref] Trajectories of adolescent conduct problems in relation to cortical thickness development: A..., Oostermeijer [/bib_ref]. Yet, a recent large-scale longitudinal study focusing on late childhood reported that higher levels of externalizing behavior was predictive of attenuated maturation of total subcortical volume [bib_ref] Tracking brain development and dimensional psychiatric symptoms in children: A longitudinal population-based..., Muetzel [/bib_ref]. Taken together, existing longitudinal findings to date are inconsistent both in regions that are associated with externalizing behavior as well as the direction of this association.
Structural neuroimaging studies also examined associations between brain regions to test relations between structural networks and cognitive and affective processes, which potentially reveal more about distributed networks and their behavioral associations [bib_ref] Networks of anatomical covariance, Evans [/bib_ref]. Structural covariance refers to correlations between properties of brain regions across individuals (Alexander-Bloch, [bib_ref] The convergence of maturational change and structural covariance in human cortical networks, Alexander-Bloch [/bib_ref]. Longitudinal designs allow for examination of maturational coupling (i.e., patterns of correlated change across subjects), which may reflect coordinated development between regions (Alexander-Bloch et al., 2013). In line with this framework, [bib_ref] Cortical thickness, cortico-amygdalar networks, and externalizing behaviors in healthy children, Ameis [/bib_ref] demonstrated that orbitofrontal-amygdala structural network properties were associated with externalizing behavior. Specifically, low levels of externalizing behavior were associated with stronger maturational coupling within the OFC-amygdala network.
We aimed to test the association between brain developmental patterns and externalizing behavior, with a threefold focus: (a) confirm the longitudinal relations between externalizing behavior and cortical brain structure, specifically in the dlPFC, OFC, and ACC [bib_ref] Trajectories of adolescent conduct problems in relation to cortical thickness development: A..., Oostermeijer [/bib_ref] , (b) test the longitudinal developmental relations between externalizing behavior and subcortical brain structures in more detail [bib_ref] Cortical thickness, cortico-amygdalar networks, and externalizing behaviors in healthy children, Ameis [/bib_ref] [bib_ref] Tracking brain development and dimensional psychiatric symptoms in children: A longitudinal population-based..., Muetzel [/bib_ref] [bib_ref] Trajectories of adolescent conduct problems in relation to cortical thickness development: A..., Oostermeijer [/bib_ref] , and (c) examine how maturational coupling between volume of these brain regions relate to externalizing behavior [bib_ref] Cortical thickness, cortico-amygdalar networks, and externalizing behaviors in healthy children, Ameis [/bib_ref]. For this purpose, we tested participants from a community sample between ages 8 and 29 years, who were scanned at three occasions, each separated by a 2-year-interval. Aspects of externalizing behavior were measured with parent-report measures of aggression and rule-breaking (i.e., Child Behavior Checklist (CBCL) and with a self-reported measure of aggression (Buss-Perry Aggression Questionnaire (BPAQ)). Both questionnaires have previously been shown to be valid and reliable [bib_ref] The aggression questionnaire, Buss [/bib_ref].
We hypothesized that the development of prefrontal cortical regions negatively coincides with externalizing behavior. Hence, we expected that higher levels of externalizing behavior would be associated with attenuated cortical thinning in dlPFC, OFC, and ACC [bib_ref] Cortical thickness, cortico-amygdalar networks, and externalizing behaviors in healthy children, Ameis [/bib_ref] [bib_ref] Trajectories of adolescent conduct problems in relation to cortical thickness development: A..., Oostermeijer [/bib_ref]. Moreover, we expected a negative relation between externalizing behavior and subcortical development, specifically in striatum, hippocampus, and amygdala volume [bib_ref] Tracking brain development and dimensional psychiatric symptoms in children: A longitudinal population-based..., Muetzel [/bib_ref] [bib_ref] Cortical and subcortical abnormalities in youths with conduct disorder and elevated callous-unemotional..., Wallace [/bib_ref]. Finally, we examined the relation between maturational coupling and externalizing behavior. Although exploratory in nature, we expected that low externalizing behavior was associated with stronger maturational coupling between prefrontal regions and subcortical regions [bib_ref] Cortical thickness, cortico-amygdalar networks, and externalizing behaviors in healthy children, Ameis [/bib_ref].
# Methods and materials
## Participants and procedure
This study was part of the longitudinal research project BrainTime [bib_ref] Goal-directed correlates and neurobiological underpinnings of adolescent identity: A multimethod multisample longitudinal..., Becht [/bib_ref] [bib_ref] Emerging depression in adolescence coincides with accelerated frontal cortical thinning, Bos [/bib_ref] [bib_ref] Increased striatal activity in adolescence benefits learning, Peters [/bib_ref] [bib_ref] Contributions of reward sensitivity to ventral striatum activity across adolescence and early..., Schreuders [/bib_ref] [bib_ref] Unraveling age, puberty and testosterone effects on subcortical brain development across adolescence, Wierenga [/bib_ref] , conducted in Leiden, The Netherlands. For this study, a community sample of children, adolescents, and young adults were recruited from local schools and advertisement. At the first time point (TP1), 299 participants were included (ages: 8-25 years; 146 males). All participants were fluent in Dutch, right-handed, and had normal or corrected-to-normal vision. Inclusion criteria were the absence of neurological or mental health problems or the use of psychotropic mediation at TP1. One participant reported to have an attention deficit disorder diagnosis at TP1 and was therefore excluded.
Participants were invited to participate in three consecutive assessment waves approximately every 2 years. Intelligence was assessed at TP1 and TP2 using two subtests of ageappropriate Wechsler Intelligence Scales (TP1: Similarities and Block Design; TP2: Vocabulary and Picture Completion). All participants had an estimated overall IQ >80. The institutional Review Board at Leiden University Medical Center approved the study. Informed consent was obtained from participants or from parents in case of minors at each time point. Participants received a financial reimbursement for their participation in the study.
# Materials
Parent-reported aggression and rule-breaking. The CBCLis a parentreport instrument that was used to assess levels of aggressive behavior and rule-breaking in children and adolescents under age 18. The CBCL is a well-validated questionnaire and widely used to assess a broad range of behavior problems in children (ages 6-18) as observed in the previous 2 months. In this study, we used the raw scores of the aggressive behavior and rule-breaking scales. The aggressive behavior scale consists of 18 items, and the rule-breaking scale of 17 items. Parents were asked to rate each statement on a 3-point scale ranging from 'not true' to 'very true/often true'. Cronbach's alpha for aggressive behavior was good for all three TPs (a = .79 -.83). Cronbach's alpha for rule-breaking was poor at TP1 (a = .50) and acceptable at TP2 (a = .66) and TP3 (a = .72).
Self-reported aggression. The BPAQ [bib_ref] The aggression questionnaire, Buss [/bib_ref] is a self-report measure of aggressive tendencies and comprises 29 items. The BPAQ distinguishes four subscales: physical aggression (e.g., 'If someone hits me, I hit back'), verbal aggression (e.g., 'I tell my friends openly when I disagree with them'), anger (e.g., 'I have trouble controlling my temper'), and hostility (e.g., 'When people are especially nice, I wonder what they want'). Participants indicated on a 7-point scale the degree to which each item characterized them (ranging from 'extremely uncharacteristic of me' to 'extremely characteristic of me'). The cumulative score on total aggression was used for analyses. Cronbach's alpha was good across all TPs (a = .81-.86). The BPAQ data from the project have been reported previously in relation to sex steroid hormones and diffusion weighted imaging measures of white matter [bib_ref] Short fused? associations between white matter connections, sex steroids, and aggression across..., Peper [/bib_ref].
## Image acquisition and analysis
Structural magnetic resonance images (MRI) were acquired on the same 3 Tesla Philips Achieva whole body scanner, with a standard 32-channel whole-head coil. T1-weighted anatomical scans were obtained at each time point (TR = 9.8 ms, TE = 4.6 ms, flip angle = 8°, 140 slices, 0.875 mm 9 0.875 mm 9 1.2 mm, and FOV=224 9 177 9 168 mm). Scan time for this sequence was 4 min 56 s. There were no major scanner hardware or software upgrades during the data collection period. A radiologist reviewed all T1-weighted scans; no anomalous findings were reported.
## Image processing
Whole-brain volumetric segmentation and cortical surface reconstruction was performed using the longitudinal pipeline of FreeSurfer 5.3 (http://surfer.nmr.mgh.harvard.edu/). The technical details and specific processing steps are described elsewhere [bib_ref] Whole brain segmentation: Automated labeling of neuroanatomical structures in the human brain, Fischl [/bib_ref] [bib_ref] Within-subject template estimation for unbiased longitudinal image analysis, Reuter [/bib_ref]. Detailed post-processing quality control was performed on all scans (see Appendix S1 for details). For each scan, volumetric estimates for subcortical regions were extracted per hemisphere. Parcellation of the cerebral cortex into gyral regions was performed using the Desikan-Killiany-Tourville atlas [bib_ref] Structural brain development between childhood and adulthood: Convergence across four longitudinal samples, Klein [/bib_ref]. Per hemisphere, 31 cortical regions were labeled.
## Regions of interest
Based on previous research on externalizing behavior, we selected the following regions of interest (ROIs): the OFC, the dlPFC, the ACC, the insula, and the parahippocampal cortex. For each hemisphere, the mean cortical thickness was created by combining the following parcellation units: OFC, medial and lateral OFC; dlPFC, superior frontal, rostral middle frontal and caudal middle frontal cortex; ACC: rostral and caudal ACC. For the cortical thickness index, we took the size of each included ROI into account. For example, we calculated cortical thickness of the OFC as follows:
[formula] ðCT medialOFC ÃSA medialOFC ÞþðCT lateralOFC ÃSA lateralOFC Þ ðSA medialOFC þSA lateralOFC Þ [/formula]
Additionally, we investigated the following subcortical volumes: amygdala, hippocampus, caudate, putamen, pallidum, nucleus accumbens, and thalamus.
## Statistical analyses
Statistical analyses were performed using SPSS 23.0 (IBM SPSS Statistics, IBM Corporation) and R 3.2.0 (R Core Team, 2014). To examine reliability over time, we calculated intraclass correlations using SPSS (see [fig_ref] Table 1: Sample characteristics [/fig_ref]. To examine the developmental trajectory of parent-reported aggression scores, parent-reported rule-breaking scores, and self-reported aggression scores we used mixed model analyses, using the nlme package in R(see Appendix S1 for model selection procedure). Next, we examined the longitudinal relations between parent-reported aggression, parent-reported rule-breaking, self-reported aggression and thickness of the specified cortical ROIs, and volume of the subcortical ROIs. Each ROI was added to the best age model, separately. Note, that sex did not improve growth model fit of parent-reported aggression, parent-reported rule-breaking, nor self-reported aggression. To control for age and sex effects on structural brain measures, we used regression residuals of MRI measures in our mixed models to predict externalizing behavior. That is, we first assessed the best age model for each MRI variable and tested whether sex improved model fit. Next, we extracted the regression residuals for the best fitting growth model for each MRI variable. The model below outlines our general mixed model:
[formula] Y ij ¼ age ij þ residual MRIij þ 1jSubject [/formula]
The i subscript denotes subject, the j denotes TP. Additionally, we examined whether an interaction between age and MRI explained additional variance to the models. For significant findings, we additionally tested whether initial level or brain development contributed to development of externalizing behavior. To do so, we added TP1 and change related to TP1 as predictors in the mixed model analyses (i.e., indices of TP1, TP2, and TP3 minus index of TP1). Also, for these indices, we used regression residuals based on the best fitting growth model. The model below outlines our mixed model using initial level of the MRI variable (residual_baselineMRI) and change level of the MRI variable (residual_changeMRI):
[formula] Y ij ¼ age ij þ residual baseline MRI i þresidual change MRI ij þ 1jSubject [/formula]
The i subscript denotes subject, the j denotes TP. Additional analyses on cortical volume and cortical surface area are reported in the supplement .
To control for multiple comparisons, we used a Bonferroni correction procedure adjusted for correlated variables (http:// www.quantitativeskills.com/sisa/calculations/bonfer.htm) [bib_ref] What's wrong with Bonferroni adjustments, Perneger [/bib_ref] [bib_ref] Some comments on frequently used multiple endpoint adjustment methods in clinical trials, Sankoh [/bib_ref]. The average correlation between cortical thickness variables (five ROIs for both hemispheres) was r = 0.53, yielding a significance level for a (2-sided adjusted) = .018 for cortical thickness analyses. The average correlation between subcortical volumes (seven ROIs for both hemispheres) was r = 0.40, which resulted in an a (2-sided adjusted) = .010 for analyses on subcortical volume.
Finally, we explored whether maturational coupling in MRI volume measures were related to parent-reported aggression, parent-reported rule-breaking, and self-reported aggression scores by using anatomical correlations [bib_ref] A key characteristic of sex differences in the developing brain: Greater variability..., Wierenga [/bib_ref]. This method assesses the inter-regional anatomical associations by defining the statistical similarity between pairs of ROIs. To investigate maturational coupling, we used individual slopes for each ROI assessed over TPs. Hence, for each individual, we extracted the random slope from the best fitting age model using linear mixed models. Note that for the maturational coupling analyses, we only used participants with MRI data of all three TPs. The Pearson correlation coefficient between rates of change in any two regions (i.e., slope) i and j was assessed. Next, differences in correlations were compared between groups of high and low dimensions of externalizing behavior using permutation testing (see Appendix S1). Groups were defined using median split across all three TPs (median parentreported aggression: 4.17; median parent-reported rule-breaking: 1.93; median self-reported aggression 86.00). [fig_ref] Table 1: Sample characteristics [/fig_ref] summarizes the sample characteristics. After quality control, we included 271 participants who had at least one MRI scan of good quality and available questionnaire data (total number of scans = 680). For the analyses concerning parent-reported aggression and rule-breaking we used data of participants that had CBCL data at least at one TP and were younger than 18 years old at TP3 (n = 147). For the analyses on self-reported aggression, we used participants who had Buss-Perry aggression scores on at least at one TP (n = 269). For the maturational coupling analyses, 69 participants were included for parent-reported aggression and parent-reported rule-breaking, and 168 participants for self-reported aggression.
# Results
Correlational analyses showed that self-reported aggression scores were positively correlated with parent-reported aggression scores (TP1: r = 0.49, p < .001; TP2: r = 0.42, p < .001; TP3: r = 0.45, p < .001) and parent-reported rule-breaking scores (TP1: r = 0. 37, p < .001; TP2: r = 0.33, p < .001; TP3: r = 0.37, p < .001). Parent-reported aggression scores and rule-breaking scores correlated positively at each TP (r = 0.58-0.63).
## Developmental trajectory of externalizing behavior
In [fig_ref] Figure 1: For self-reported aggression, the mixed model analyses revealed negative relations between self-reported... [/fig_ref] -c, the developmental trajectories of parent-reported aggression, rule-breaking, and selfreported aggression are depicted as a function of age. For parent-reported aggression and rule-breaking, we found an increase across adolescence [Aggression (Age 1 ): b = 7.96, t (247) =2.59, p = .01; Rule-breaking (Age 1 ): b = 14.00, t (247) =7.00, p < .001]. Adding a random slope effect improved model fit for rulebreaking only, showing that for rule-breaking individuals differ both in intercept and developmental pattern. For self-reported aggression scores, the random intercept model fitted best. Adding a main effect of sex and/or an interaction effect with sex did not improve model fit for parent-reported aggression or, rule-breaking or, self-reported aggression (see for model fit indices).
## Longitudinal relation between parent-reported aggression and rule-breaking scores with cortical thickness and subcortical volumes
Mixed model analyses showed a longitudinal relation between parent-reported aggression scores with right hippocampus volume (residuals) (b = À7.34, t (178) = À2.87, p = .005) and left pallidum volume (residuals) (b = 6.24, t (178) = 2.41, p = .017) . The latter effect was, however, not significant when correcting for multiple comparisons. As depicted in [fig_ref] Figure 2: Mixed model analyses for the longitudinal relation between parent-reported aggression and hippocampal... [/fig_ref] , larger right hippocampus volume was associated with lower parent-reported aggression scores. In contrast, larger left pallidum volume was associated with higher parent-reported aggression scores [fig_ref] Figure 1: For self-reported aggression, the mixed model analyses revealed negative relations between self-reported... [/fig_ref]. Note that when we replaced outliers (Z ≥ 3.0) to 3 standard deviations from the mean, similar results were obtained. Adding age by ROI interactions did not improve model fit.
Next, we assessed whether the above-described significant relations between parent-reported aggression scores and regional brain indices were related to change relative to TP1 in subcortical volume or to initial level of these regional brain indices at TP1. Follow-up mixed model analyses showed that these relations were driven by change in right hippocampal volume (b = À7.11, t (166) = À2.48, p = .014) and change in left pallidum volume (b = 6.31, t (166) = 2.20, p = .029) . Decreases in right hippocampal volume relative to TP1 were associated with increases in parent-reported aggression, independent of hippocampus volume at TP1. Increases in left pallidum volume relative to TP1 were associated with increases in parent-reported aggression, independent of level of pallidum volume at TP1.
For parent-reported rule-breaking scores, mixed models revealed a longitudinal relation between right OFC (b = À2.88, t (178) = À1.97, p = .050) and left dlPFC thickness (b = À2.93, t (178) = À2.01, p = .046), but neither of these effects survived correction for multiple comparisons [fig_ref] Figure 2: Mixed model analyses for the longitudinal relation between parent-reported aggression and hippocampal... [/fig_ref] and . Follow-up mixed model analyses revealed that these longitudinal associations were driven by change in right OFC (b = À8.40, t (165) = À5.02, p < .001) and left dlPFC thickness (b = À8.89, t (165) = À5.38, p < .001) . These results indicate that thinning in these regions and decreases in rule-breaking behavior are associated, independent of initial level of cortical thickness at TP1. Note that when we replaced outliers (Z ≥ 3.0) to 3 standard deviations from the mean, similar results were obtained.
Longitudinal relation between self-reported aggression scores with cortical thickness and subcortical volumes p = .031) [fig_ref] Figure 3: Difference matrix for rates of anatomical change [/fig_ref] ; . None of these relations survived correction for multiple comparisons. Also for self-reported aggression, adding an interaction term between age and ROIs did not improve model fit. Note that when we replaced outliers (Z ≥ 3.0) to 3 standard deviations from the mean, similar results were obtained. Follow-up analyses to examine whether development of self-reported aggression coincided with change in subcortical ROIs relative to TP1 or to level of these ROIs at TP1 revealed that the relation between self-reported aggressive behavior in right caudate volume was driven by change in right caudate volume (b = À36.86, t (353) = À2.55, p = .011). That is, decreases in right caudate volume were associated with decreases in self-reported aggression regardless of initial level of right caudate volume at TP1. For right putamen volume and left thalamus volume, neither change nor initial level of these brain indices was significantly associated with developmental trajectories of aggressive behavior .
## Group differences in maturational coupling
We further explored whether maturational coupling between slopes of MRI volume measures were related to parent-reported aggression, parent-reported rulebreaking, and self-reported aggression scores. [fig_ref] Figure 3: Difference matrix for rates of anatomical change [/fig_ref] depicts the difference matrix that was derived from the subtraction of the correlated rates of anatomical change for high parent-reported Aggression group from the matrix of correlated rates of anatomical change for the low parent-reported Aggression group. Differences between groups (High vs. Low parent-reported Aggression Group) are depicted in the lower half of the matrix. Permutation testing showed that three correlations of anatomical change differed significantly between groups. Two correlations between rates of anatomical change were stronger in the high parent-reported Aggression group; left pallidum volume with left putamen volume and right thalamus volume with left putamen volume. The correlation between right hippocampal volume and right ACC volume was stronger in the low parent-reported Aggression group (see [fig_ref] Figure 4: Difference matrix for rates of anatomical change [/fig_ref].
Also, for parent-reported rule-breaking groups, permutation testing showed three significant differences between correlations of anatomical change between groups [fig_ref] Figure 4: Difference matrix for rates of anatomical change [/fig_ref]. Again, the correlations between left pallidum volume and left putamen volume, as well as left pallidum volume and left caudate volume, were stronger in the high parentreported rule-breaking group. For the low parentreported rule-breaking group, the correlation between right hippocampal volume and left pallidum volume was stronger [fig_ref] Figure 5: Difference matrix for rates of anatomical change [/fig_ref].
Median split analyses on self-reported aggression showed five significant differences between maturational coupling in the high versus low group [fig_ref] Figure 5: Difference matrix for rates of anatomical change [/fig_ref]. Four correlations were stronger in the high self-reported aggression group: left amygdala volume with left ACC volume, right OFC volume with right insula volume, left pallidum volume with right thalamus volume, and left pallidum volume with right caudate volume. The correlation between left dlPFC volume and left ACC volume was stronger in the low self-reported aggression group .
# Discussion
This study tested the relation between externalizing behavior and cortical and subcortical brain development in a typically developing sample. The behavioral analyses demonstrated an increase in parent-reported aggression and rule-breaking across adolescence, consistent with prior studies showing that adolescence is not only an important transition period for novelty seeking, impulsiveness, and sensation seeking , but also a period in which disruptive-behavior disorders emerge (Moffitt, 2018). Self-reported aggression scores, however, showed a stable pattern across age, but with large individual differences. An important question we addressed was whether individual trajectories of externalizing behavior were associated with individual differences in brain development; a question for which longitudinal designs are pivotal. This study yielded two main findings: (a) right hippocampal volume was negatively associated with parent-reported aggression, and (b) structural maturational coupling between regions within striatum, limbic system, and prefrontal regions were associated with externalizing behavior. Smaller right hippocampal volume was associated with higher parent-reported aggression scores. Interestingly, follow-up analyses showed that this relation was driven by change in hippocampal volume regardless of initial volume of hippocampus at TP1; reinforcing the idea that longitudinal measures capture trajectories that cannot be observed using cross-sectional analyses. Stronger decline in hippocampal volume was related to increases in parentreported aggression. Our findings demonstrated that hippocampal development and externalizing behavior coincide, yet we cannot infer the causal direction of this relation. Traditionally it is suggested that brain development shapes behavior, but it is likely that this shaping process goes both ways. Indeed, a recent study reported that externalizing behavior merely affects subcortical brain structure, and this relation was not found the other way around [bib_ref] Tracking brain development and dimensional psychiatric symptoms in children: A longitudinal population-based..., Muetzel [/bib_ref].
The hippocampus is part of the limbic system of the brain, which is generally involved in emotional processes. It has been suggested that the hippocampus plays an important role in regulating aggressive behavior. For example, regional stimulation of hippocampus can facilitate or inhibit aggression [bib_ref] Brain structures and neurotansmitters regulating aggression in cats: Implications for human aggression, Gregg [/bib_ref]. Our finding that developmental changes in hippocampal volume were associated with developmental changes in externalizing behavior is also in line with findings of a previous longitudinal study with adolescents. [bib_ref] Trajectories of adolescent conduct problems in relation to cortical thickness development: A..., Oostermeijer [/bib_ref] showed that adolescents who displayed a desisting pathway of conduct problems showed aberrant development of hippocampal volume. Yet, it remains elusive whether our observed coinciding developmental pattern is related to deviant behavior. Our participants showed only mild problems of externalizing behavior, all within the healthy range of psychological functioning. Nevertheless, prior studies in clinical samples showed volumetric reductions of hippocampal volume in individuals with borderline personality disorder [bib_ref] Hippocampal volume reduction and history of aggressive behaviour in patients with borderline..., Zetzsche [/bib_ref] and conduct disorder [bib_ref] Morphometric brain abnormalities in boys with conduct disorder, Huebner [/bib_ref] , which supports the idea that using a dimensional approach in a community sample may enhance our understanding of clinical externalizing problems. It should be noted that hippocampal volume development was related to parent-reported aggression, but not to self-reported aggression. Parentreported aggression and self-reported aggression correlated moderately in the current study indicating both substantial overlap as well as distinction between these two behavioral constructs.
In contrast to our expectation, we did not find relations between cortical thickness of dlPFC, OFC, and ACC and parent-reported nor self-reported aggressive behavior. Our results showed that parent-reported rule-breaking showed a negative relation with cortical thinning in right OFC and left dlPFC; although these results did not survive correction for multiple comparisons. Previous crosssectional and longitudinal studies using both clinical and community samples emphasized the relation between cortical thickness of prefrontal regions and externalizing behavior [bib_ref] Cortical thickness, cortico-amygdalar networks, and externalizing behaviors in healthy children, Ameis [/bib_ref] [bib_ref] Trajectories of adolescent conduct problems in relation to cortical thickness development: A..., Oostermeijer [/bib_ref] [bib_ref] Prefrontal structural and functional brain imaging findings in antisocial, violent, and psychopathic..., Yang [/bib_ref]. Speculatively, differences in severities of externalizing behavior, age-range, and methodological differences (e.g., analytical approach of sMRI) may explain inconsistencies between studies.
Exploratory analyses on maturational coupling showed that adolescents with low levels of parentreported aggressive behavior showed stronger synchronous development of right hippocampal volume and right ACC volume. This finding fits well with Difference matrix for rates of anatomical change (High parent-reported rule-breaking group vs. low parent-reported rulebreaking group). Significant differences are depicted in the lower half of the matrix prior longitudinal research that revealed similar outcomes for stronger amygdala-prefrontal cortex structural connectivity in relation to lower levels of psychopathology [bib_ref] Cortical thickness, cortico-amygdalar networks, and externalizing behaviors in healthy children, Ameis [/bib_ref] [bib_ref] Cortico-amygdalar maturational coupling is associated with depressive symptom trajectories during adolescence, Vijayakumar [/bib_ref]. Given that previous research focused on specific ROIs, future research needs to examine whether externalizing behavior is related to maturational coupling within certain regions of the limbic system (e.g., amygdala and/or hippocampus) and prefrontal cortex, or whether a more general limbic system-prefrontal cortex network is involved. Furthermore, we found that stronger maturational coupling was not merely related to lower levels of externalizing behavior. Adolescents with higher levels of parent-reported aggression and parentreported rule-breaking showed stronger maturational coupling between the striatum and limbic system. Moreover, higher levels of self-reported aggression were related to stronger subcortical-subcortical coupling and cortical-subcortical coupling. Future research is warranted to replicate this pattern of mixed positive and negative associations between externalizing behavior and maturational coupling.
This study also had some limitations that should be addressed in future research. First, the sample is large in comparison to prior studies, but relatively small to test for age interactions. Second, the behavioral assessments included both the parentreport and self-report and were therefore multiinformant, but these informants did not report on exactly the same behavioral construct. In future studies, it will be important to better align measurements to have a general index of externalizing/ aggressive behavior and to test whether results of different informants tap into different constructs and subsequently distinct neural underpinnings. Third, we used a normative sample and none of the participants reported clinical levels of externalizing behavior. Future studies could benefit from examining the full range of externalizing behavior. Last, we only examined the relation between externalizing behavior and brain development without considering other factors that might contribute to this relationship. That is, during adolescence, substantial changes occur in several domains that might mediate this relationship, for example, social factors including not only school transitions and friendships, but also biological factors such as sex steroid hormones [bib_ref] A testosteronerelated structural brain phenotype predicts aggressive behavior from childhood to adulthood, Nguyen [/bib_ref]. To conclude, this study revealed that aspects of externalizing behavior were associated with lower right hippocampal volume. Furthermore, we found that a stronger maturational network between ACC and limbic system was associated with low externalizing behavior, whereas stronger subcortical-subcortical maturational coupling was associated with relatively higher levels of externalizing behavior. Our findings highlight the need to investigate dynamic changes in brain structure and their relation to behavioral outcomes using within-person comparisons as well as examining developmental trajectories between regions. Unraveling the neural underpinnings of externalizing behavior across a continuum may provide important new insights about when normative development becomes deviant.
## Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article: Appendix S1. Supplementary methods.
[fig] © 2018: The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for Child and Adolescent Mental Health. [/fig]
[fig] Figure 1: For self-reported aggression, the mixed model analyses revealed negative relations between self-reported aggression scores and left thalamus volume (b = À29.15, t (364) = À2.24, p = .026), right putamen volume (b = À26.64, t (364) = À2.04, p = .042), and right caudate volume (b = À27.87, t (364) = À2.Developmental trajectories of (A) parent-reported aggression, (B), parent-reported rule-breaking, and (c) self-reported aggression. The line represents the optimal fitting model, shade represents 95% confidence interval. Lines represent individual participants, and participants who were measured once are represented by dots. Note, circle is TP1, triangle is TP2 and square is TP3© 2018 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for Child and Adolescent Mental Health. [/fig]
[fig] Figure 2: Mixed model analyses for the longitudinal relation between parent-reported aggression and hippocampal volume. Left panel represents individual data for three TPs for parentreported aggression and hippocampal volume. Right panel depicts predicted model for parent-reported aggression and hippocampal volume (residuals). Note, circle is TP1, triangle is TP2 and square is TP3 © 2018 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for Child and Adolescent Mental Health. [/fig]
[fig] Figure 3: Difference matrix for rates of anatomical change (High parent-reported aggression group vs. low parent-reported aggression group). Significant differences are depicted in the lower half of the matrix © 2018 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for Child and Adolescent Mental Health. [/fig]
[fig] Figure 4: Difference matrix for rates of anatomical change (High parent-reported rule-breaking group vs. low parent-reported rulebreaking group). Significant differences are depicted in the lower half of the matrix [/fig]
[fig] Figure 5: Difference matrix for rates of anatomical change (High self-reported aggression group vs low self-reported aggression group). Significant differences are depicted in the lower half of the matrix © 2018 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for Child and Adolescent Mental Health. [/fig]
[table] Table 1: Sample characteristics [/table]
|
Brachial artery thrombosis following bee sting, case report
# Introduction
Severe reactions may occur following bee stings and other insects. These reactions can be local or systemic. Local reactions are often characterized by pain, swelling, erythema, itching, and blister, and in general, type 1 mast cells surround the sting site and this is an anaphylactic reaction [bib_ref] Fatal infection after a bee sting, Truskinovsky [/bib_ref]. It differs in poisoning clinics due to honey bee or wasp. Most people are usually locally allergic, and after insertion, reactions including pain, redness, swelling occur in the area and usually disappear within a few hours. It can cause a wide variety of reactions such as bee sting, neurological findings, seizure, ischemic attack. However, in less frequent severe cases, myocardial infarction, pulmonary edema, bleeding, kidney failure, and life-threatening anaphylactic shock may also occur [bib_ref] Quadriparesis following wasp sting: an unusual reaction, Agarwal [/bib_ref]. This case is presented in order to emphasize this very rare and non-literature situation that physicians should be aware of such a complication in order to make an emergency diagnosis and to start early treatment.
## Presentation of case
A 47-year-old man without comorbid disease applied to the emergency room in the morning with a complaint of paleness, coldness, numbness and bruising in the left palm developing at 1 h after bee sting from his left leg. It was determined that the patient had a history of medication, a family history including genetic information or smoking status, and did not engage in beekeeping. The patient did not have any known chronic conditions. On examination, there were signs of coldness, partial parasy and ischemia on the left palm and fingers. Brachial radial and ulnar pulses were nonpalpable. Cardiac rhythms were in sinus rhythm. Doppler ultrasonography performed on the patient revealed stenosis in the brachial artery. Digital angiography was performed by entering the upper extremity femoral artery immediately. Vascular occlusion was observed after brachial bifurcation. It was seen that filling was made from collateral. After the brachial artery was reached with the catheter tip, the metalyse bolus was performed. Then 25,000 of heparin was given. Then, metalyse brachial trunktan was given again. Balloon angioplasty was performed on the radial artery. Then, it was observed that the filling of the brachial trunk and ulnar, radial artery increased. Increased flow in collaterals was observed (Figs. 1-4). During peroperative angiography, it was observed that the vessel was opened and the blood supply increased. All these procedures were performed by the cardiovasculer surgeon. The patient was taken to cardiovascular intensive care after the operation. Heparin infusion of 1000 units / h iv was given for 24 h. Oral coumadin and aspirin (100 mg) were started on the patient who was taken into the service on the 2nd day. Cyanosis of the patient's arm disappeared and radial and ulnar artery pulses were palpable. There were no complications related to the intervention and the patient tolerated this procedure easily. The patient stated that his complaints disappeared after the treatment, he received after the operation. On the 10th day, the patient was discharged with necessary recommendations.
# Discussion
After an allergic reaction to bee sting; skin reaction symptoms (erythema, itching, urticaria and angioedema), respiratory system effects (larynx edema and bronchospasm) and cardiovascular (myocardial depression, hypotension and shock) and gastroin- testinal tract events (nausea, vomiting, fecal incontinence) may occur [bib_ref] Acute pulmonary hemorrhage following a honeybee sting: a case report, Lam [/bib_ref] [bib_ref] The natural history and epidemiology of insect venom allergy: clinical implications, Bilo [/bib_ref] [bib_ref] Unusual reactions to insect stings, Reisman [/bib_ref] [bib_ref] Effects on behaviour and EEG of single chain phospholipases A2 from snake..., Gandolfo [/bib_ref]. Consequently, severe anaphylactic shock may also lead to cerebral or myocardial ischemia accompanied by permanent sequelae [bib_ref] Histamine-induced coronary artery spasm: the concept of allergic angina, Kounis [/bib_ref]. The clinical picture of bee sting may rarely include diffuse alveolar bleeding, rhabdomyolysis, thrombocytopenic purpura, and vasculitis [bib_ref] The natural history and epidemiology of insect venom allergy: clinical implications, Bilo [/bib_ref] [bib_ref] Unusual reactions to insect stings, Reisman [/bib_ref] [bib_ref] Effects on behaviour and EEG of single chain phospholipases A2 from snake..., Gandolfo [/bib_ref] [bib_ref] Histamine-induced coronary artery spasm: the concept of allergic angina, Kounis [/bib_ref]. Although our case did not have any history, it was found appropriate to present because vasospasm and thrombosis developed in the brachial artery following bee sting. This is a very rare clinical presentation of bee stings according to the literature. It contains various chemical components that contribute to the clinical picture such as bee venom, enzymes (phospholipases, hyaluronidase) and biological amines (histamine, serotonin, dopamine, norepinephrine and acetylcholine) (9). Our patient was admitted to the emergency department with a bruise extending from his left hand to his arm after bee stings and was generally thought to be related to the dose of this poison.
It is associated with a variety of reasons that can activate the pathways of hypersensitivity and induce the degranulation of mast cells. In the light of this information, the results of our case were interpreted as toxic findings. Providing life support should be a priority in cases of anaphylaxis. Oxygen delivery, adrenaline (0.5 mg intramuscularly in 1/1000 adult), intervals of 10−15 min, diphenhydramine, prednisolone, ranitidine, fluids or, if necessary, positive inotropes in the presence of bronchospasm. It is recommended to give 2 agonists and aminophylline. In the emergency department, anti-allergic and prophylactic anti-anaphylactic treatments were applied in line with the treatment protocols recommended to our patient.
Phospholipase A2 is a key enzyme for the activation of the metabolism of arachidonic acid. During the metabolism of arachidonic acid, a number of cytokines and chemokines are released. These include leukotrienes via lipoxygenase and prostaglandins such as thromboxane via cyclooxygenase. These mediators have been accused in many clinical and laboratory studies to induce coronary artery spasm and / or acute myocardial infarction. This shows that the event in our case occurred thrombosis in the vascular bed and secondary vasospasm. Acute allergic attack after bee sting, such as Type II variant of Kounis syndrome, most likely caused plaque erosion or rupture, and vasospasm developed distal to the brachial artery after thrombosis formation [bib_ref] Allergic angina and allergic myocardial infarction: a new twist on an old..., Nikolaidis [/bib_ref]. Our study has been reported according to SCARE 2018 criteria [bib_ref] The SCARE 2018 statement: updating consensus Surgical CAse REport (SCARE) guidelines, Agha [/bib_ref]. We did not have any limitations in our approach to this case in terms of our treatment strategy, potential risks and complications.
# Conclusion
The case presented here, albeit rarely, emphasizes the fact that thrombosis may occur due to bee stings and that emergency services should recognize these cases.
## Note
Conflicts of Interest do not exist between authors. No funding source was included in this study. All procedures were done with prior permission from the hospital and the patient.
## Conflicts of interest
There is no financial disclosures of the authors. Author Mehmet Hüseyin Akgül declares that he has no conflict of interest.
Author Eldar Bagırov declares that he has no conflict of interest.
# Sources of funding
There is no funding sources of the study.
# Ethical approval
Ethics committee no approval was obtained. This studies is case report.
## Consenṫ
Informed consent is take to patient.
# Author contribution
All studies was wrote from to writer.
## Registration of research studies
Not applicable.
## Guarantor
The writer was guarantor. The name is Mehmet Hüseyin Akgül and Eldar Bagırov.
[fig] Figure 1: During the percutaneous intravascular angiographic intervention, the image taken before treatment showed that the flow in the brachial artery distal decreased due to occlusion. [/fig]
[fig] Figure 2: During percutaneous intravascular angiographic intervention, the image taken after treatment showed a significant increase in flow in the distal of the brachial artery. [/fig]
[fig] Figure 3: Cyanotic hand fingers are observed in the patient's emergency image taken before angiography. [/fig]
[fig] Figure 4: In the image taken after the angiography of the patient, it is observed that the cyanosis in the toes disappeared. [/fig]
|
Isolation and characterization of centromeric repetitive DNA sequences in Saccharum spontaneum
Sugarcane (Saccharum hybrids spp.)is the most important sugar crop that accounts for ~75% of the world's sugar production. Recently, a whole-genome sequencing project was launched on the wild species S. spontaneum. To obtain information on the DNA composition of the repeat-enriched region of the centromere, we conducted a genome-wide analysis of the DNA sequences associated with CenH3 (a mutant of histone H3 located in eukaryote centromeres) using chromatin immunoprecipitation followed by sequencing (ChIP-seq) method. We demonstrate that the centromeres contain mainly SCEN-like single satellite repeat (Ss1) and several Ty3/gypsy retrotransposon-related repeats (Ss166, Ss51, and Ss68). Ss1 dominates in the centromeric regions and spans up to 500 kb. In contrast, the Ty3/gypsy retrotransposon-related repeats are either clustered spanning over a short range, or dispersed in the centromere regions. Interestingly, Ss1 exhibits a chromosome-specific enrichment in the wild species S. spontaneum and S. robustum, but not in the domesticated species S. officinarum and modern sugarcane cultivars. This finding suggests an autopolyploid genome identity of S. spontaneum with a high level of homology among its eight sub-genomes. We also conducted a genome-wide survey of the repetitive DNAs in S. spontaneum following a similarity-based sequence clustering strategy. These results provide insight into the composition of sugarcane genome as well as the genome assembly of S. spontaneum.The centromere is a chromosomal domain that directs the assembly of kinetochore, which mediates chromosome segregation by interacting with spindle microtubules. A typical feature for centromeric chromatin is the presence of CenH3 (CENP-A in mammals), which is a mutant of histone H3. Studies have revealed that CenH3 present in all eukaryote centromeres studied 1-4 . Thus, centromeric chromatin is defined by the presence of CenH3. Centromeric DNA is composed of satellite DNAs and highly repeated centromeric retrotransposons (CRs). Centromeric satellite DNAs are usually mega base-sized arrays with a monomer size ranging from 100-200 to thousands of base pairs 5-7 . In humans, satellite DNAs may dominate the entire functional region of the centromere3,4,[8][9][10]. CRs are also enriched in eukaryote centromeres, especially in plants4,9,11. In the centromeres of maize and rice, CRs are clustered, span over a long range, and are always intermingled with centromeric satellites 12-14 . Most identified CRs in plants belong to the Ty3/gypsy group of long terminal repeat (LTR) retrotransposons 11,15 . The most striking structural feature of a plant CR is a C-terminal chromodomain of the integrase gene, which may be responsible for centromere-specific insertion 15 . Moreover, sequence analyses have demonstrated that some centromeric satellite DNAs originated from retrotransposons7,9,[16][17][18][19][20]. Thus, the CRs may play a core role in the occurrence of classical repeat-enriched centromeres.
Although many draft genomes of variant species have been reported in recent decades, whole centromere DNA sequencing, even the fine-scale genetic and physical mapping of centromeres, remains a challenge because of highly repetitive nature of centromeric DNA. The chromatin immunoprecipitation followed by sequencing (ChIP-seq) method allowed us to isolate the DNA sequences associated with CenH3 (CENP-A in mammals), which is a mutant of histone H3 and presents in all eukaryote centromeres studied [bib_ref] The centromere paradox: stable inheritance with rapidly evolving DNA, Henikoff [/bib_ref] [bib_ref] A molecular view of plant centromeres, Jiang [/bib_ref] [bib_ref] The molecular basis for centromere identity and function, Mckinley [/bib_ref]. In plants, extensive studies have been conducted to analyze the DNA composition of CenH3-associated centromeric chromatin using ChIP-seq [bib_ref] Maize centromeres expand and adopt a uniform size in the genetic background..., Wang [/bib_ref] [bib_ref] Maize centromere structure and evolution: sequence analysis of centromeres 2 and 5..., Wolfgruber [/bib_ref] [bib_ref] Boom-Bust Turnovers of Megabase-Sized Centromeric DNA in Solanum Species: Rapid Evolution of..., Zhang [/bib_ref] [bib_ref] De novo centromere formation on a chromosome fragment in maize, Fu [/bib_ref] [bib_ref] Sequential de novo centromere formation and inactivation on a chromosomal fragment in..., Liu [/bib_ref] [bib_ref] Stable Patterns of CENH3 Occupancy Through Maize Lineages Containing Genetically Similar Centromeres, Gent [/bib_ref] [bib_ref] Formation of a functional maize centromere after loss of centromeric sequences and..., Zhang [/bib_ref] [bib_ref] Inbreeding drives maize centromere evolution, Schneider [/bib_ref]. However, poorly assembled reference centromeres often hamper the application of ChIP-seq in the characterization of centromeric DNAs. Recently, an alternative approach based on the evaluation of the enrichments of clustered repeats from the whole genome was used in centromeric DNA studies [bib_ref] Repeatless and repeat-based centromeres in potato: Implications for centromere evolution, Gong [/bib_ref] [bib_ref] Graph-based clustering and characterization of repetitive sequences in next-generation sequencing data, Novak [/bib_ref] [bib_ref] Repetitive DNA in the pea (Pisum sativum L.) genome: comprehensive characterization using..., Macas [/bib_ref] [bib_ref] RepeatExplorer: a Galaxy-based web server for genome-wide characterization of eukaryotic repetitive elements..., Novák [/bib_ref]. In this method, a reference repeat sequence from the whole genome was first generated and then clustered by evaluating mutual similarities between sequences to identify groups of densely connected reads [bib_ref] Graph-based clustering and characterization of repetitive sequences in next-generation sequencing data, Novak [/bib_ref] [bib_ref] Repetitive DNA in the pea (Pisum sativum L.) genome: comprehensive characterization using..., Macas [/bib_ref] [bib_ref] RepeatExplorer: a Galaxy-based web server for genome-wide characterization of eukaryotic repetitive elements..., Novák [/bib_ref]. Then, the potential centromeric repeat clusters were identified by evaluating the relative enrichments of the ChIP-seq reads with respect to the reference control. Several studies conducted on plants using this approach have demonstrated that it is an efficient method to assay the centromeric repeat DNA compositions of various species, especially those without a genome assembly or poorly assembled centromere regions [bib_ref] Repeatless and repeat-based centromeres in potato: Implications for centromere evolution, Gong [/bib_ref] [bib_ref] Repeat Composition of CenH3-chromatin and H3K9me2-marked heterochromatin in Sugar Beet (Beta vulgaris), Kowar [/bib_ref] [bib_ref] Stretching the rules: monocentric chromosomes with multiple centromere domains, Neumann [/bib_ref] [bib_ref] Rapid proliferationand nucleolar organizer targeting centromeric retrotransposonsin cotton, Han [/bib_ref].
Sugarcane (Saccharum hybrids spp.) is the most important sugar crop that accounts for ~75% of the world's sugar production [bib_ref] Microcollinearity between autopolyploid sugarcane and diploid sorghum genomes, Wang [/bib_ref]. The Saccharum genus is composed of two wild species, i.e., S. robustum and S. spontaneum, and four groups of formerly cultivated clones: S. officinarum, S. barberi, S. sinense, and S. edule. Both wild species contain large amounts of natural genetic variations and have a wide range of chromosome numbers and ploidy levels with 2n = 6x − 8x = 60-170 for S. robustum and 2n = 32-128 for S. spontaneum [bib_ref] Saccharum species as horticultural classes, Irvine [/bib_ref] [bib_ref] Determination of basic chromosome numbers in the genus Saccharum by physical mapping..., D'hont [/bib_ref]. All modern sugarcane cultivars are derived from interspecific crosses between S. officinarum and S. spontaneum [bib_ref] Cytological studies in Saccharum chromosome transmission in inter-specific and inter-generic crosses, Roach [/bib_ref] [bib_ref] Molecular cytogenetic investigation of chromosome composition and transmission in sugarcane, Piperidis [/bib_ref]. The wild clones of S. spontaneum were used to introduce disease resistance, vigor, stubbling, and other traits into S. officinarum and thus, a series of backcrosses with S. officinarum were conducted to restore the high sugar content trait, a process called Noblization [bib_ref] Cytological studies in Saccharum chromosome transmission in inter-specific and inter-generic crosses, Roach [/bib_ref]. The resulting modern sugarcane cultivars (Saccharum spp. hybrids) are highly polyploid interspecific hybrids and represent the most genetically complex crop ever studied [bib_ref] Molecular cytogenetic investigation of chromosome composition and transmission in sugarcane, Piperidis [/bib_ref]. Estimated monoploid sugarcane genome size is approximately 930 Mb, which is similar to the genome size of sorghum, closely related grass [bib_ref] Microcollinearity between autopolyploid sugarcane and diploid sorghum genomes, Wang [/bib_ref]. Thus, the genome of sugarcane cultivars will have a DNA content of about 10 Gb 34 . Recently, a whole-genome sequencing project was launched on a S. spontaneum clone SES208 (2n = 8x = 64). However, the octaploid complexity of its nuclear genome, especially the high-level content of repetitive sequences (approximately 60%) has hindered severely the assembly process of the whole genome (Ray Ming, personal communication). In this study, we conducted a genome-wide assay of centromeric repetitive DNA sequence repeats on SES208 following CenH3-based ChIP-seq approach [bib_ref] Graph-based clustering and characterization of repetitive sequences in next-generation sequencing data, Novak [/bib_ref] [bib_ref] Repetitive DNA in the pea (Pisum sativum L.) genome: comprehensive characterization using..., Macas [/bib_ref] [bib_ref] RepeatExplorer: a Galaxy-based web server for genome-wide characterization of eukaryotic repetitive elements..., Novák [/bib_ref]. The isolated centromeric repeats were confirmed by cytological analyses. We revealed the origin, structure and distribution of each repeat in the centromeres of SES208. These results will contribute to our understanding of sugarcane centromeres and also facilitate sugarcane whole-genome sequencing.
# Results
Composition of the repetitive DNAs in S. spontaneum clone SES208. To provide a reference repeat database for the analysis of the centromere repetitive sequences, the input DNA was sequenced using HiSeq 2500 platform. A total of 75.9 million 100-bp pair-end reads were obtained. Of which, 3.9 million reads were randomly selected to generate repeat clusters using the RepeatExplorer software [bib_ref] RepeatExplorer: a Galaxy-based web server for genome-wide characterization of eukaryotic repetitive elements..., Novák [/bib_ref]. This analysis resulted in a total of 359, 342 repeat clusters and 883, 890 single/non-clustered reads. The 359, 342 clusters represented different repeat families in SES208 genome that accounted for 77.4% of the analyzed 3.9 million reads. Among these clusters, 226 clusters that accounted for 46.7% of the genomic reads were relatively enriched in the SES208 genome (genome proportion> 0.01%) [fig_ref] Figure 1: The composition and annotation of repetitive DNAs in S [/fig_ref]. Thus, these 226 most highly repetitive clusters were annotated to characterize the most repeat families.
The LTR retrotransposons were the most abundant repeat families, accounting for 38.45% of the SES208 genome [fig_ref] Figure 1: The composition and annotation of repetitive DNAs in S [/fig_ref]. Among them, the Ty3/gypsy retrotransposons were the most enriched, representing 24.97% of the genome, followed by LTR/copia, accounting for 13.48% of the genome. Satellite repeats and six DNA transposons (DNA/CMC-EnSpm, DNA/MULE MuDR, DNA/PIF-Harbinger, RC/Helitron, DNA/hAT-Ac, and DNA/ hAT-Tag1) were also found in the genome, representing 3.43% and 3.70% of the genome, respectively. Among the six DNA transposons, only DNA/CMC-EnSpm and DNA/MULE MuDR represented relatively high genome proportions (1.53% and 1.20%, respectively), other four DNA transposons showed less than 1% genome proportions. The rest of the repeat families, including ribosome DNA, long-interspersed nuclear elements (LINEs), and uncharacterized repeats, also showed a relatively low genome proportion of < 1%.
Computational isolation of the centromere-specific repeats in S. spontaneum clone SES208.
Our immunostaining assay confirmed that the rice CenH3 antibody could also specifically recognize the CenH3 of SES208 [fig_ref] Figure 2: Immunoassay on rice CenH3 binding-specificity and FISH analyses of CenH3 ChIP DNA... [/fig_ref]. We then conducted CenH3 ChIP on S. spontaneum SES208. To evaluate the enrichment of centromeric DNA in our ChIP DNA, we labeled the ChIP DNA and conducted FISH analysis. We detected highly enriched signals from the centromeres of SES208, but the signals were not evident with the negative control DNA sample under the same conditions (mock, see Methods) [fig_ref] Figure 2: Immunoassay on rice CenH3 binding-specificity and FISH analyses of CenH3 ChIP DNA... [/fig_ref].
A total of 73.2 million 100-bp pair-end sequence reads from ChIP DNA were obtained. These CenH3 ChIP DNA reads and the 75.9 million input DNA reads were mapped to all repeat clusters using BLAST (e-value, 1e-8). The read number ratios of ChIP DNA relative to the input DNA, which indicated the level of enrichment of each repeat family in the centromere, were calculated for all clusters [fig_ref] Figure 3: Relative enrichments of repeat DNA families in the ChIP-seq data and SES208... [/fig_ref]. The sequence proportion of each repeat family was estimated based on the number of sequence reads associated with individual clusters. Seven repeat clusters showed obvious higher ratios (> 2) than the others, suggesting that these seven clusters were most likely the centromeric DNAs. These seven centromeric repeat clusters having a ChIP/input ratio > 2 represented a total of 3.14% of the genome proportion. Only the centromeric repeat cluster 1 showed a relatively high genome proportion (2.5%), all other six repeat clusters had a very low genome proportion (from 0.046% for cluster 218 to 0.260% for cluster 51) [fig_ref] Figure 3: Relative enrichments of repeat DNA families in the ChIP-seq data and SES208... [/fig_ref] , , indicating that the repeat elements in these six clusters had relatively low copy numbers when compared with those of repeat cluster 1 in the SES208 genome.
Cytological confirmation of computationally identified centromeric repeats. One contig with the highest read depth for each of the seven repeat clusters was selected as the representative. In the FISH assay, all seven repeat sequences generated strong FISH signals in the centromeres [fig_ref] Figure 4: FISH mapping of centromeric repeats in S [/fig_ref] and S1), confirming that these repeat sequences were components of centromeric DNAs. Five of these seven repeats, namely, Ss1, Ss166, Ss51, Ss262, and Ss68, generated centromere-specific signals from all 64 chromosomes of SES208, indicating these repeats colonized and spread specifically in the centromeres. However, the signal intensities from different centromeres varied for individual repeats, suggesting that each centromere might have a different copy number. Interestingly, we always observed eight higher intensity FISH signals for repeat Ss1 than other centromeric repeats [fig_ref] Figure 4: FISH mapping of centromeric repeats in S [/fig_ref] , suggesting an enrichment of this repeat in these eight chromosomes. For repeats Ss268 and Ss218, in addition to the centromeric FISH signals, dispersed FISH signals on chromosome arms were also detected [fig_ref] Figure 1: The composition and annotation of repetitive DNAs in S [/fig_ref] , suggesting that these two repeats are not centromere-specific. Therefore, these two repeats were excluded from further centromeric analyses.
To mine the centromeric repeats at a wider scale, we also analyzed four other repeat clusters (Cl242, Cl53, Cl203, and Cl279), which had relatively high ChIP/input ratios (> 1.5) [fig_ref] Figure 3: Relative enrichments of repeat DNA families in the ChIP-seq data and SES208... [/fig_ref]. However, no centromeric FISH signals were detected using these repeats, implying that these four as well as other repeat clusters with lower ChIP/input ratios are likely non-centromeric. Based on all these results, the five repeat sequences (Ss1, Ss166, Ss51, Ss262, and Ss68) that generated strong centromere-specific FISH signals were selected for further analyses.
## Sugarcane centromeres contain both satellite and retrotransposon dnas. the five repeat
sequences that showed centromere-specific FISH signals were blasted in the nucleotide database. Repeat Ss1 contained eight ~140-bp repetitive units, which shared high sequence similarity (74-99%) with the sugarcane centromeric satellite SCENs from an Egyptian breeding variety no. 37185 [bib_ref] A novel repetitive sequence of sugar cane, SCEN family, locating on centromeric..., Nagaki [/bib_ref]. For the other four repeats, varying degrees of similarities to the characterized centromeric retrotransposons were detected. Repeats Ss51 and Ss68 showed a high sequence similarity (> 70%) to a centromeric retrotransposon in maize (CRM) (Gene Bank: AY129008.1), which has a full size of 7572-bp with two 931-bp long terminal repeats (LTR) [fig_ref] Figure 5: Dot-plot similarity comparisons of sugarcane centromeric repeats and other retrotransposons [/fig_ref]. Specifically, Ss51 showed a 969-bp region of similarity to partial LTR regions (722 bp) and a 247-bp internal conjunction region of 5′ LTR [fig_ref] Figure 5: Dot-plot similarity comparisons of sugarcane centromeric repeats and other retrotransposons [/fig_ref]. For repeat Ss68, nearly all of its sequence (97.3% or 3,855 of the full size of 3,962 bp, was highly similar (88%) to the gag-pol region of the maize retrotransposon CRM [fig_ref] Figure 5: Dot-plot similarity comparisons of sugarcane centromeric repeats and other retrotransposons [/fig_ref]. This indicated that repeats Ss51 and Ss68 may derive from different sections of one retrotransposon. The major part (84.4%, 1,225 bp) of repeat Ss166 shared 73% sequence similarity with the gag-pol region of another maize centromeric Ty3/gypsy retrotransposon (Gene Bank: AF078917.1) [fig_ref] Figure 5: Dot-plot similarity comparisons of sugarcane centromeric repeats and other retrotransposons [/fig_ref]. For repeat Ss262, no significant similarity was found, except for a short region (~136 bp) that shared 72% sequence similarity with the gag-pol region of a rice centromeric retrotransposon CRR3 (Gene Bank: DQ458292.1) [fig_ref] Figure 5: Dot-plot similarity comparisons of sugarcane centromeric repeats and other retrotransposons [/fig_ref]. Additional BLAST search in the protein database using repeats Ss51, Ss68 and Ss166 as quires revealed a large number of similarities (30-88%) with the Ty3/gypsy retrotransposon protein, thereby confirming their retrotransposon origin.
The structure of centromeric repetitive DNA sequence repeats revealed by fiber-FISH. As expected, long contiguous fiber-FISH signals were observed with the repeat Ss1 probe [fig_ref] Figure 6: Fiber-FISH mapping of the centromeric repeats in S [/fig_ref] , confirming its tandem repeat structure. We selected the ten longest fiber-FISH signals for length measurements. The length varied from 132.0 to 181.4 μ m, with an average of 153.5 μ m , suggesting that the centromeric repeat Ss1 could span up to ~400 to 500 kb in S. spontaneum clone SES208 (1 μ m = 3.24 kb) [bib_ref] Systematic application of DNA fiber-FISH technique in cotton, Wang [/bib_ref].
Continuous signal spots were also detected when using repeats Ss51 and Ss68. Interestingly, fiber-FISH signals from Ss51 and Ss68 were consistently associated with the same DNA fibers [fig_ref] Figure 6: Fiber-FISH mapping of the centromeric repeats in S [/fig_ref] in our dual-color fiber-FISH experiment, confirming that repeats Ss51 and Ss68 represented different parts of the same repeat unit (one retrotransposon). However, the majority of the signals from this retrotransposon were short, spanning an average of 101 kb , n = 19) in the genome. Moreover, the densities of the fiber-FISH signal spots of these two probes were consistently lower than those of Ss1 [fig_ref] Figure 6: Fiber-FISH mapping of the centromeric repeats in S [/fig_ref] , suggesting that there were other sequence(s) inserted within this retrotransposon array. To find out the relationship between the centromeric satellite repeat Ss1 and the CR represented by repeat Ss51 and Ss68, dual-color fiber-FISH experiments were also conducted using labeled probes of Ss1, Ss51 and Ss68 [fig_ref] Figure 6: Fiber-FISH mapping of the centromeric repeats in S [/fig_ref]. Among the sixteen fibers obtained, eight showed overlapped or partial overlapped signals [fig_ref] Figure 6: Fiber-FISH mapping of the centromeric repeats in S [/fig_ref]. For other eight fibers, six showed only the signal of repeat Ss1 probe, while two showed signals of the CR probe, either Ss51, or Ss68, or both probes that were labeled with the same color [fig_ref] Figure 6: Fiber-FISH mapping of the centromeric repeats in S [/fig_ref]. The results indicated that this highly repeated retrotransposon was intermingled with satellite repeat Ss1 in some centromeres of the SES208 genome.
No continuous fiber-FISH signal spots were found for Ss166 and Ss262 probes. In dual-color fiber-FISH using these two probes with repeat Ss1, we detected dispersed signal spots that overlapped with the signals of repeat Ss1 [fig_ref] Figure 6: Fiber-FISH mapping of the centromeric repeats in S [/fig_ref] , suggesting that both repeats Ss166 and Ss262 were dispersed in centromeres.
Enrichment of the SCEN-like satellite repeat was chromosome-specific in sugarcane. An interesting finding of the FISH assay was that we consistently observed eight centromeres showing much stronger FISH signals with the repeat Ss1 probe than other probes [fig_ref] Figure 4: FISH mapping of centromeric repeats in S [/fig_ref] , suggesting that these eight centromeres contain significantly more copies of repeat Ss1. Previous studies have demonstrated a high similarity within the eight su-bgenomes (chromosome set of a monoploid) for S. spontaneum SES208 and suggested its auto-octaploid identity of 2n = 8x = 64 [bib_ref] Microcollinearity between autopolyploid sugarcane and diploid sorghum genomes, Wang [/bib_ref] [bib_ref] Determination of basic chromosome numbers in the genus Saccharum by physical mapping..., D'hont [/bib_ref]. Thus, it is most likely that there are eight homologous chromosomes for one specific chromosome (designated as the Chromosome Ss1) distribution among the eight sub-genomes. However, a recent study revealed that rapidly evolving centromeric DNAs could colonize and proliferate in the centromeres of chromosomes belonging to one sub-genome after polyploidy formation [bib_ref] Rapid proliferationand nucleolar organizer targeting centromeric retrotransposonsin cotton, Han [/bib_ref]. Thus, we cannot rule out the possibility that these eight chromosomes also belong to one sub-genome. To clarify this, we conducted a FISH assay using both repeat Ss1 and 5 S rDNA probes [fig_ref] Figure 7: FISH assay of the satellite repeat Ss1 in other sugarcane species [/fig_ref]. The results showed that there were eight chromosomes bearing the 5S rDNA signals [fig_ref] Figure 7: FISH assay of the satellite repeat Ss1 in other sugarcane species [/fig_ref] , and those chromosomes are theoretically the eight homologous chromosomes derived from one specific chromosome (designated as the Chromosome 5S) in this octaploid sugarcane. If these eight chromosomes bearing the brighter Ss1 signals were a complete chromosome set of one sub-genome, we would have seen one chromosome bearing FISH signals from both Ss1 and 5S rDNA. However, this was not the case in this study [fig_ref] Figure 7: FISH assay of the satellite repeat Ss1 in other sugarcane species [/fig_ref] , indicating that the eight chromosomes with brighter Ss1 signals were not members of one sub-genome. Furthermore, the morphologies of these eight chromosomes were not significantly different [fig_ref] Figure 2: Immunoassay on rice CenH3 binding-specificity and FISH analyses of CenH3 ChIP DNA... [/fig_ref]. All these results suggested that the eight chromosomes bearing brighter Ss1 signals were homologous to one another and distributed among the eight sub-genomes.
Additional FISH analyses were also conducted on another wild species S. robustum (clone Molokai6081), S. officinarum (clone LA Purple), and two modern sugarcane cultivars (ROC22 and Funong38). Significantly brighter FISH signals of repeat Ss1 were also detected in one set of eight chromosomes in S. robustum [fig_ref] Figure 7: FISH assay of the satellite repeat Ss1 in other sugarcane species [/fig_ref] , which did not bear the 5S rDNA signals either [fig_ref] Figure 7: FISH assay of the satellite repeat Ss1 in other sugarcane species [/fig_ref]. However, the signal intensities of the Ss1 satellite repeat showed a graduated difference, and obviously brighter Ss1 signals were hardly seen in S. officinarum [fig_ref] Figure 7: FISH assay of the satellite repeat Ss1 in other sugarcane species [/fig_ref] and cultivars ROC22 and Funong38 [fig_ref] Figure 7: FISH assay of the satellite repeat Ss1 in other sugarcane species [/fig_ref]. As S. officinarum is domesticated from wild species S. robustum, and cultivars are the interspecific hybrids of S. officinarum and wild species S. spontaneum 37 , it is concluded that the SCEN-like satellites which has a equal chromosome-Ss1 enrichments in S. robustum and S. spontaneum may have undergone uninformed deletion or proliferation in different centromeres during domestication. Interestingly, SCEN-like satellites have equal chromosome-Ss1 enrichment patterns in S. robustum and S. spontaneum.
# Discussion
In eukaryotes, a significant fraction of the genome is comprised of repetitive DNAs, the component of which is often greater than the coding sequence component and is also referred to as the "dark matter" of the genome. Researchers showed that the repetitive DNAs play a role in numerous cell processes [bib_ref] Why repetitive DNA is essential to genome function, Shapiro [/bib_ref] [bib_ref] Repetitive sequences in plant nuclear DNA: types, distribution, evolution and function, Mehrotra [/bib_ref]. Therefore, understanding the contents and origins of repetitive DNAs represents an important step towards completely deciphering the organization and function of the genome sequence [bib_ref] Repetitive elements may comprise over two-thirds of the human genome, De Koning [/bib_ref]. However, the contents of repetitive DNAs derived from whole-genome sequencing are likely under-estimated because the highly repeated DNAs regularly fail to be assembled due to the technology barrier. For example, approximately half of the human genome was previously identified as TEs and other repeats in a previous human genome sequencing dataset [bib_ref] Repetitive DNA and next-generation sequencing: computational challenges and solutions, Treangen [/bib_ref]. However, an additional 20% or more repetitive or repeat-derived DNAs were identified using an alternative de novo strategy recently [bib_ref] Repetitive elements may comprise over two-thirds of the human genome, De Koning [/bib_ref]. This phenomenon was also found in animals and plants [bib_ref] Repeat Composition of CenH3-chromatin and H3K9me2-marked heterochromatin in Sugar Beet (Beta vulgaris), Kowar [/bib_ref] [bib_ref] Rapid proliferationand nucleolar organizer targeting centromeric retrotransposonsin cotton, Han [/bib_ref] [bib_ref] De novo identification of repeat families in large genomes, Price [/bib_ref]. In this study, we applied a method to assess repetitive DNA composition using similarity-based sequence clustering and annotation rather than the genome assembly. This approach has been applied effectively to evaluate the repetitive DNA composition for whole genome or centromeric regions in other plants [bib_ref] Repeatless and repeat-based centromeres in potato: Implications for centromere evolution, Gong [/bib_ref] [bib_ref] Repeat Composition of CenH3-chromatin and H3K9me2-marked heterochromatin in Sugar Beet (Beta vulgaris), Kowar [/bib_ref] [bib_ref] Stretching the rules: monocentric chromosomes with multiple centromere domains, Neumann [/bib_ref] [bib_ref] Rapid proliferationand nucleolar organizer targeting centromeric retrotransposonsin cotton, Han [/bib_ref]. It is notable that approximately 77% of the SES208 genome is comprised of mobile elements or other repeat structures [fig_ref] Figure 1: The composition and annotation of repetitive DNAs in S [/fig_ref] , demonstrating a high proportion of repetitive elements in the SES208 genome. We believe that this survey of the compositions and genome proportions of the repetitive families will contribute to SES208 genome sequencing and further sugarcane genome studies.
In rice, maize and sorghum, the centromeric satellite repeats can span from hundreds of Kb to several Mb [bib_ref] Functional rice centromeres are marked by a satellite repeat and a centromere-specific..., Cheng [/bib_ref] [bib_ref] Maize centromeres: organization and functional adaptation in the genetic background of oat, Jin [/bib_ref] [bib_ref] Cloning and characterization of a centromere-specific repetitive DNA element from Sorghum bicolor, Miller [/bib_ref] , which has hampered the DNA sequence assembly of the centromeric region. Our Fiber-FISH assay revealed that the repeat Ss1 could span up to 500 kb in SES208. We cannot determine if the data represent the entire size of the Ss1 arrays in one centromere because the satellite repeats can be interrupted by CRs [fig_ref] Figure 6: Fiber-FISH mapping of the centromeric repeats in S [/fig_ref]. However, the data indicate that the DNA sequence assembly of the centromeric regions remains a challenge for SES208 because reconstruction of such long tandem arrays is not feasible, even with the latest PacBio sequencing platform (read length up to 40 kb with an average of 10-15 kb). Alternatively, cytological analyses based on chromosome-specific markers [bib_ref] Functional rice centromeres are marked by a satellite repeat and a centromere-specific..., Cheng [/bib_ref] or individual chromosome addition lines 6,14 might be a feasible approach to elucidate the composition and structure of centromere in sugarcane.
A common feature of typical centromeres is enriched satellites and TE repeats. Moreover, the centromeric satellites are often homogenized and thus, a single type of satellite can dominate all centromeres in most higher eukaryotes. For example, the centers of centromeres are always occupied by one type of satellite repeat in major crops, such as rice (CentO) [bib_ref] Functional rice centromeres are marked by a satellite repeat and a centromere-specific..., Cheng [/bib_ref] , maize (CentC) [bib_ref] Maize centromeres: organization and functional adaptation in the genetic background of oat, Jin [/bib_ref] , sorghum (pSau3A9) [bib_ref] A conserved repetitive DNA element located in the centromeres of cereal chromosomes, Jiang [/bib_ref] , Brachypodium (Bd_CENT) [bib_ref] Genome sequencing and analysis of the model grass Brachypodium distachyon, International Brachypodium [/bib_ref] , and Arabidopsis (pAL1) [bib_ref] Chromatin immunoprecipitation reveals that the 180-bp satellite repeat is the key functional..., Nagaki [/bib_ref] [bib_ref] Localization of tandemly repeated DMA sequences in Arabidopsis thaliana, Maluszynska [/bib_ref]. In humans, the alpha satellite dominates the entire functional region of the centromere 8 . Here, our results show that the DNA composition of the SES208 centromere is also characterized as a typical centromere. First, satellite and retrotransposon-like DNAs are the major repeat components in the centromeres . Second, only one centromeric satellite repeat is found that accounts for 2.5% of the genome. This value is much higher than the value of TE-related or other repeats (0.57% genome proportion for the four centromere-specific repeats) [fig_ref] Figure 1: The composition and annotation of repetitive DNAs in S [/fig_ref]. In addition, long and compact signal spots for satellite Ss1 are detected by fiber-FISH [fig_ref] Figure 6: Fiber-FISH mapping of the centromeric repeats in S [/fig_ref] , in contrast to the short and loose signal spots for the TE-related or other repeats [fig_ref] Figure 6: Fiber-FISH mapping of the centromeric repeats in S [/fig_ref]. All these data indicated that the Ss1 satellite repeat is dominantly abundant in SES208 centromeres.
Centromeric satellites evolve rapidly and can differ greatly, even among closely related species in eukaryotes 27 . However, once a satellite repeat evolves into a structure that would be favorable for the function and structure of the centromere, it will be fixed and spread to other centromeres. For example, the alpha-satellite repeat has occupied the centromeres of primate species for nearly 40 million years [bib_ref] Boom-Bust Turnovers of Megabase-Sized Centromeric DNA in Solanum Species: Rapid Evolution of..., Zhang [/bib_ref] [bib_ref] Primate comparative genomics: lemur biology and evolution, Horvath [/bib_ref]. The results from this and several other studies have revealed that the SCEN-like satellite Ss1 colonized the centromeres of all studied clones of S. spontaneum, S. robustum, S. officinarum, and modern sugarcane cultivars [bib_ref] A novel repetitive sequence of sugar cane, SCEN family, locating on centromeric..., Nagaki [/bib_ref] [bib_ref] Characterization of CENH3 and centromere-associated DNA sequences in sugarcane, Nagaki [/bib_ref]. This may suggest that SCEN-like satellite is relatively conserved or resides in the centromeres across all the species and modern sugarcane varieties of the Saccharum genus. Further efforts have to be devoted to verify this because the extent of diversity with regard to basic chromosome number and ploidy level is unclear in this genus.
It is known that all modern sugarcane cultivars are derived from interspecific crosses between S. officinarum and S. spontaneum conducted a century ago 38,39 and S. officinarum was assumed to originate from S. robustum [bib_ref] Saccharum species as horticultural classes, Irvine [/bib_ref]. However, our results showed that the SCEN-like satellite repeat Ss1 exhibited chromosome-specific enrichment in S. spontaneum and S. robustum, but not in S. officinarum and the cultivars [fig_ref] Figure 7: FISH assay of the satellite repeat Ss1 in other sugarcane species [/fig_ref]. A recent study in maize showed that the domestication selection for centromere-like genes could cause amplification or deletion of the satellite centromere repeats in maize [bib_ref] Inbreeding drives maize centromere evolution, Schneider [/bib_ref]. In cotton, the D sub-genome-derived centromeric repeats could invade the centromeres of other sub-genomes and rapidly amplify after polyploidization, which might result in indistinguishable enrichment among different centromeres for each centromeric repeat [bib_ref] Rapid proliferationand nucleolar organizer targeting centromeric retrotransposonsin cotton, Han [/bib_ref]. Thus, a potential explanation for our results is that the SCEN-like satellite repeat Ss1 found in S. officinarum and sugarcane cultivars may have undergone rapid turnover events among different centromeres during the interspecific hybridization and domestication selection process, whereas asexual vegetative propagations may have restrained the chromosome-specific enrichment pattern of the SCEN-like satellite repeat in the two wild species.
More importantly, the highly conserved pattern of the SCEN-like satellite repeat Ss1 between eight homologous chromosomes may have occurred before the polyploidization event. Another possibility is that Ss1 repeat may have undergone either uniform amplification for the eight individual chromosomes or deletion in other chromosomes. However, the latter hypothesis is unlikely because the SCEN-like satellite repeat Ss1 colonized all centromeres, and there is no evidence showing that the centromeric satellite repeat can turnover synchronously in the multiple copies of specific centromere(s) in polyploid individuals. By contrast, our results from S. officinarum and other cultivars suggest that either deletion or amplification of the satellite repeat Ss1 was not limited to specific centromeres [fig_ref] Figure 7: FISH assay of the satellite repeat Ss1 in other sugarcane species [/fig_ref]. Given the rapid evolution of the centromeric satellite repeat, we can anticipate that the formation of the octaploid occurs through a rapid doubling process without dramatic heterogenization between sub-genomes. This means that the S. spontaneum clone SES208 has an auto-octaploid genome with eight homologous sets of sub-genomes. ChIP and ChIP-seq. To isolate the centromeric DNA sequences, we performed CenH3 ChIP on S. spontaneum SES208. A polyclonal antibody against rice CenH3 was used in this study [bib_ref] Sequencing of a rice centromere uncovers active genes, Nagaki [/bib_ref]. ChIP experiments were conducted according to published protocol [bib_ref] Sequencing of a rice centromere uncovers active genes, Nagaki [/bib_ref]. Young leaf tissues were collected for ChIP experiment. A fraction of micrococcal nuclease-digested genome DNAs was preserved as genomic control (designated as input DNA).
Normal rabbit serum was used in the mock control. A ChIP-seq library was constructed by using NEBNext ® Ultra ™ DNA Library Prep Kit (New England BioLabs Inc., Ipswich, MA, USA) according to product instructions. The library was sequenced on HiSeq 2500 (Illumina, San Diego, CA, USA) following the 100 bp paired-end genomic DNA sequencing protocol.
## Data treatment and centromeric repeat identification. the sequence reads from chip and input
DNAs (ChIP-seq and input-seq reads, respectively) were first treated using FastUniq [bib_ref] FastUniq: a fast de novo duplicates removal tool for paired short reads, Xu [/bib_ref] and Trimmomatic 57 to remove PCR duplication and low quality reads. A two-step procedure 7 was adopted with some modifications to determine the centromeric repetitive sequences. First, the input DNA was sequenced using HiSeq 2500 platform. A portion of randomly selected input-seq reads were used to perform graph-based clustering using the RepeatExplorer software (http://repeatexplorer.umbr.cas.cz/) with default parameters [bib_ref] Graph-based clustering and characterization of repetitive sequences in next-generation sequencing data, Novak [/bib_ref] [bib_ref] RepeatExplorer: a Galaxy-based web server for genome-wide characterization of eukaryotic repetitive elements..., Novak [/bib_ref]. Repeats (contigs) were then identified and classified as individual repeat clusters based on their sequence similarity. Second, the ChIP-seq and input-seq reads were mapped to repeat clusters using BLASTn with E-value threshold 1e-8 [bib_ref] Gapped BLAST and PSI-BLAST: a new generation of protein database search programs, Altschul [/bib_ref]. Reads were assigned to one cluster based on their highest similarity. The numbers of aligned reads from ChIP-seq and input-seq were counted, and the read proportion of each repeat family was calculated based on aligned read number/total read number. The ChIP/input ratios were used to evaluate the relative enrichment of repeat families in the centromere.
PCR primers were designed from bioinformatically putative repeat contigs (see . The PCR products with the corresponding sizes were recovered using a gel extraction kit (Promega, USA). Amplicons were then cloned into bacterial DH5α cells and sequenced to confirm the presence of the desired repeats. The cloned amplicons were isolated and labeled for further FISH analysis.
Chromosomal Immunoassay. The immunoassay was performed as previously described 7 with some modifications. Briefly, fresh root tips of SES208 were harvested and fixed in 2% (w/v) paraformaldehyde for 15 min at room temperature (RT). Root tips were washed in 1 × PBS for three times for 5 min each. A single root tip was then squashed on a glass slide with a cover slip. The cover slip was removed after freezing in liquid nitrogen, followed by dehydration in 70% ethanol for 5 min. The rice CenH3 antibody (20 μ g/mL) was applied to the chromosome slides and incubated at 37 °C for 3 hr. The slides were washed in 1 × PBS at RT for three times and incubated with Alexa Fluor 594 Chicken anti-Rabbit IgG (Invitrogen, USA) at 37 °C for another hour. After three times of 5-min washing at RT in 1 × PBS, chromosomes were counterstained with 4, 6-diamidino-2-phenylindole and were examined under an Olympus BX63 fluorescence microscope (Olympus, Japan).
Fluorescence in situ hybridization (FISH). FISH and fiber-FISH were carried out according to published protocols [bib_ref] Systematic application of DNA fiber-FISH technique in cotton, Wang [/bib_ref] [bib_ref] Complete assignment of the chromosomes of Gossypium hirsutum L. by translocation and..., Wang [/bib_ref] [bib_ref] Application of fiber-FISH in physical mapping of Arabidopsis thaliana, Jackson [/bib_ref]. The DNAs labeled with digoxigenin-dUTP (Roche Diagnostics, USA) and Biotin-dUTP (Roche Diagnostics, USA) were detected using rhodamine-conjugated anti-digoxigenin (Roche Diagnostics, USA) and fluorescein-conjugated avidin (Life Technologies, USA), respectively. DNAs were labeled with digoxigenin-dUTP and Biotin-dUTP for FISH analysis. Slides were examined under Olympus BX63 fluorescence microscope (Olympus, Japan). Chromosome and signal images were captured and merged using CellSens Dimension software (Olympus, Japan). Fiber-FISH was conducted to reveal the organization of the centromeric repeat sequences in SES208 genome. The fiber-FISH signals were measured and converted into kb using a 3.21-kb/μ m conversion rate [bib_ref] Systematic application of DNA fiber-FISH technique in cotton, Wang [/bib_ref].
[fig] Figure 1: The composition and annotation of repetitive DNAs in S. spontaneum clone SES208. (A) Summary of the contents of the repeat family and single copy reads. A total of 359, 342 repeat clusters were generated, among which 226 repeat clusters (accounting for 46.7% of the genome) were analyzed further. (B). Annotation and the genome proportions of the 226 repeat clusters. [/fig]
[fig] Figure 2: Immunoassay on rice CenH3 binding-specificity and FISH analyses of CenH3 ChIP DNA enrichment and localizations. (A) Metaphase chromosome spread from a S. spontaneum SES208 root tip cell. (B) Immunofluorescence signals detected at the primary chromosome constriction. (C) Merged image of (A,B). (D) Metaphase chromosome spread from a S. spontaneum SES208 root tip cell. (E) FISH signals of CenH3 ChIP DNA probe (F) Merged image of (D,E). (G) Metaphase chromosome spread from a S. spontaneum SES208 root tip cell. (H) FISH signals of mock DNA probe.(I) Merged image of (G,H). Bar = 10 μ m. Scientific RepoRts | 7:41659 | DOI: 10.1038/srep41659 [/fig]
[fig] Figure 3: Relative enrichments of repeat DNA families in the ChIP-seq data and SES208 genome. Repeat clusters are represented by dots. The x-axis is the genome proportion for each cluster. The y-axis is the ratio of the ChIP-seq reads to input-seq reads, representing the enrichment of each corresponding cluster from the ChIP-seq data. [/fig]
[fig] Figure 4: FISH mapping of centromeric repeats in S. spontaneum clone SES208. Five repeats, namely, Ss166 (A), Ss51 (B), Ss262 (C), Ss68 (D), and Ss1 (E), were mapped to the metaphase chromosomes. Centromerespecific FISH signals were detected from these five repeat probes. Bar = 10 μ m. [/fig]
[fig] Figure 5: Dot-plot similarity comparisons of sugarcane centromeric repeats and other retrotransposons. Dot-plots are drawn using the Unipro UGENE software. The sequence similarities exceeding 80% over a 15-bp sliding window are displayed as black or red dots or diagonal lines. (A) Comparisons between the repeats Ss51 and Ss68 with a maize centromeric retrotransposon (Gene Bank: AY129008.1). (B) Comparisons between the repeats Ss166 with a maize centromeric retrotransposon (Gene Bank: AF078917.1). (C) Comparisons between the repeats Ss262 with a rice centromere retrotransposon (CRR3) (Gene Bank: DQ458292.1). Scientific RepoRts | 7:41659 | DOI: 10.1038/srep41659 [/fig]
[fig] Figure 6: Fiber-FISH mapping of the centromeric repeats in S. spontaneum clone SES 208. (A) Two complete fiber-FISH signals derived from the satellite repeat Ss1 probe. (B) Two partial fiber-FISH signals derived from Ss68 (green) and Ss51 (red) probes. (C) One complete fiber-FISH signal derived from Ss68 + Ss51 (green) and Ss1 (red) probes. (D) One complete fiber-FISH signal derived from Ss166 (green) and Ss1 (red) probes. The signal spots of Ss166 probe are indicated by arrows. (E) One complete fiber-FISH signal derived from Ss 262 (green) and Ss1 (red) probes. The signal spots of Ss262 probe are indicated by arrows. Bar = 10 μ m. Methods Plant Materials. Plant materials used in this study included two sugarcane cultivars, ROC22 and Funong38, and three clones of wild species, SES208 (S. spontaneum, 2n = 8x = 64), Molokai6081 (S. robustum), and LA Purple (S. officinarum, 2n = 80). All plants were grown at 30 °C in the greenhouse under natural sunlight. [/fig]
[fig] Figure 7: FISH assay of the satellite repeat Ss1 in other sugarcane species. The probes of repeat Ss1 and 5S rDNA were hybridized simultaneously to the metaphase chromosomes of S. spontaneum clone SES208 (A-D), S. robustum clone Molokai6081 (E-H), S. officinarum clone LA Purple (I-L) and two sugarcane cultivars, ROC22 (M-P) and Funong38 (Q-T). Bar = 10 μ m. Scientific RepoRts | 7:41659 | DOI: 10.1038/srep41659 [/fig]
|
Phytochemicals, Nutrition, Metabolism, Bioavailability, and Health Benefits in Lettuce—A Comprehensive Review
Citation: Shi, M.; Gu, J.; Wu, H.; Rauf, A.; Emran, T.B.; Khan, Z.; Mitra, S.; Aljohani, A.S.M.; Alhumaydhi, F.A.; Al-Awthan, Y.S.; et al.
# Introduction
Lettuce (Lactuca sativa L.) belongs to the Asteraceae family and originates in the Mediterranean. It is a successful and diverse plant distributed worldwide [bib_ref] Everywhere but Antarctica: Using a supertree to understand the diversity and distribution..., Funk [/bib_ref]. The first cultivated lettuce appears in several primitive writings in early 2680 BCE as a medicinal
## Lettuce types and production
In Australia, Crisphead (iceberg), romaine (cos), butterhead, and looseleaf are the most common groups, while there are significant differences in growth habit, leaf color, shape, and texture between them [fig_ref] Figure 2: The frequent growth types of lettuce [/fig_ref].
## Lettuce types and production
In Australia, Crisphead (iceberg), romaine (cos), butterhead, and looseleaf are the most common groups, while there are significant differences in growth habit, leaf color, shape, and texture between them [fig_ref] Figure 2: The frequent growth types of lettuce [/fig_ref].
## Lettuce types and production
In Australia, Crisphead (iceberg), romaine (cos), butterhead, and looseleaf are the most common groups, while there are significant differences in growth habit, leaf color, shape, and texture between them [fig_ref] Figure 2: The frequent growth types of lettuce [/fig_ref]. Countries such as America, Asia, Australia, and Europe have become more vigorous markets for lettuce. The gross yield of lettuce in China is the maximum globally, occupying 56.4% of the harvesting market, followed by the United States (14.3%) and India (4.1%) [bib_ref] Analysis of the world lettuce market, Shatilov [/bib_ref]. In Australia, the production and consumption of lettuce have increased. Lettuce production increased by about 118% between 1980 and 2005, while other vegetables such as maize, potato, paddy rice, and a tomato increased by 149%, 135%, 146%, and 128%, respectively. Compared to yield increases, the increased production was mainly attributed to an increasing area cultivated and improving the resistance to insects and disease. About 6000 hectares are cultivated annually in Australia, manufacturing 135,000 tonnes for different markets, including fresh and processing. It is estimated that the lettuce industry in Australia is worth around $100 million and $8 million belongs to export.
## Taxonomical classification
Kingdom-Plantae Subkingdom-Viridiplantae Infrakingdom-Streptophyta Superdivision-Embryophyta Division-Tracheophyta Subdivision-Spermatophytina Class-Magnoliopsida Superorder-Asteranae Order-Asterales Family-Asteraceae Genus-Lactuca L. Species-Lactuca sativa L.
## Crisphead
It is also known as the iceberg, one species of Lactuca sativa var. capitata family, which forms a tight and dense head of spherical leaves that fold each other. It is known for its thick and crisp leaves and prominent veins. The wide range of leaf colors from light green to deep green and most varieties are mid-green during the mature period, while some specific genotypes contain anthocyanins [bib_ref] Expansion of conservation tillage initiatives in California, Mitchell [/bib_ref].
## Butterhead
Butterhead also belongs to Lactuca sativa var. capitata and shows a small head of spherical leaves, such as Crisphead, which the butterhead group can also form, but the leaves are not folded tightly. The inner leaves are cream green due to the lack of light, while the outer leaves are darker colored or brownish and easier to be torn and bruised than the Crisphead group. Their pliability and oily leaf texture are attractive to the consumer.
## Loose-leaf
It is also known as a leaf (Lactuca sativa). These groups form no heads but have many leaves in the center of lettuce that include smooth margins, deeply lobed, or even curled and fringed leaves. The leaf lettuce is characterized by shorter and softer leaves than romaine that are more perishable during the shipment. Looseleaf lettuce can be marketed as a whole product and salad mix ingredient, usually classified into three types: green, red, and oak. The various leaves' colors range from green to yellowish-green and some contain a red shade relying on the content of anthocyanins in lettuce and the light intensity used when growth.
## Romaine
Romaine, also called Cos or Lactuca sativa L., is the most upright growing one of the four major types of lettuce. This specie forms an oblong-shaped head with long and rigid leaves with a prominent midrib, which is upright, and can grow to a maximum of 30 cm [bib_ref] Northern field production of leaf and romaine lettuce using a high tunnel, Rader [/bib_ref]. The outer leaves are medium green, while the leaves folded tightly are greenish-white in the center. Their sweet flavor is more apparent than in any other lettuce cultivar [fig_ref] Figure 1: Stages involved in selecting published data for inclusions in the current study... [/fig_ref]. Many countries worldwide cultivate lettuce that can be classified into several consumer uses such as whole head, bulk harvest (for salad manufacturing or food service), and baby leaf [bib_ref] Lettuce and spinach, Simko [/bib_ref].
## Phytochemicals and nutritional composition of lettuce
## Phenolic compounds
Phenolic compounds are vital to plant secondary metabolites, which play an important role in the plant protection mechanism [bib_ref] Plant phenolics: Extraction, analysis and their antioxidant and anticancer properties, Dai [/bib_ref]. It is believed that phenolics can prolong the lettuce shelf life and improve plant stress resistance, contributing to decreasing the loss of postharvest [bib_ref] Postharvest ascorbate metabolism in two cultivars of spinach differing in their senescence..., Hodges [/bib_ref]. This is characteristic because the reactive oxygen species can be scavenged, which is believed to be involved in leaf senescence and also in the plant's antioxidant defense mechanism [bib_ref] Chemical composition and antioxidant capacity of lettuce: Comparative study of regular-sized (Romaine)..., López [/bib_ref]. Firstly, the most common phenolic acids in lettuce are caffeic acid, chlorogenic acid, as well as their derivatives [bib_ref] Influences of organic fertilization, high tunnel environment, and postharvest storage on phenolic..., Zhao [/bib_ref]. Phenolic acids account for 70% and 94% of the total phenolic content in some green varieties, whereas only 35% and 45% for red lettuce [bib_ref] Characterisation of polyphenols and antioxidant properties of five lettuce varieties and escarole, Llorach [/bib_ref]. Secondly, flavonoid composition and flavonols in lettuce are quercetin and kaempferol derivatives, anthocyanins, and flavone luteolin [bib_ref] Characterisation of polyphenols and antioxidant properties of five lettuce varieties and escarole, Llorach [/bib_ref]. Furthermore, quercetin (flavonol) was followed by isorhamnetin, an essential (O-methylated flavonol) flavonoid in some lettuce types. There are free and bound phenolic compounds found in lettuce [bib_ref] Chemical composition and antioxidant capacity of lettuce: Comparative study of regular-sized (Romaine)..., López [/bib_ref]. The free phenolic compounds that can be identified in the methanolic extract were protocatechuic, chlorogenic, caffeic and p-coumaric acids, kaempferol, luteolin, apigenin, and phlorizin. Chlorogenic acid is the composition mainly found among the free phenolics. Generally, containing the highest chlorogenic and caffeic acids were romaine lettuces, but with great variability among cultivars [bib_ref] Chemical composition and antioxidant capacity of lettuce: Comparative study of regular-sized (Romaine)..., López [/bib_ref].
Another main difference between red and green lettuce is anthocyanins since this type of flavonoids is responsible for the red/blue/purple color pigments in some vegetables and fruits, but not all [bib_ref] Phytochemicals and overall quality of leafy lettuce (Lactuca sativa L.) varieties grown..., Mampholo [/bib_ref]. There was an association between the color of leaves and lettuce type. Generally, lettuce's qualitative and quantitative phenolic content relies on their variety with different genetic information. In recent years, red lettuce has been increasingly popular in the diet as a salad mix ingredient because of its anthocyanin content and higher phenolic contents that contribute to greater health benefits than the green one [bib_ref] Anthocyanin levels in nine lettuce (Lactuca sativa) cultivars: Influence of planting date..., Gazula [/bib_ref]. However, various agronomic or environmental factors such as light, and climatic and postharvest conditions and their tissue type can influence the phenolic content found in lettuce [fig_ref] Table 1: Chemical structure of main phytochemicals identified in lettuce [/fig_ref] [bib_ref] Polyphenol content and antioxidative activity in some species of freshly consumed salads, Heimler [/bib_ref]. Therefore, further studies can focus on this area. Recent research has shown that a number of dietary polyphenolic compounds produced from plants are more potent antioxidants in vitro than vitamins E or C and may therefore contribute to the protective benefits in vivo. It is now feasible to establish the antioxidant activities of plant-derived flavonoids in the aqueous and lipophilic phases and to determine the degree to which the overall antioxidant potentials of wine and tea may be explained by the activities of individual polyphenols [bib_ref] Antioxidant properties of phenolic compounds, Rice-Evans [/bib_ref].
## Carotenoids
Carotenoids are a kind of lipid-soluble pigment widely found in fruits and vegetables that are yellow-orange and leafy vegetables with dark color [bib_ref] Actual knowledge on food sources, intakes, stability and bioavailability and their protective..., Maiani [/bib_ref]. β-carotene content was highest presented in all the lettuce cultivars, accounting for half of the total carotenoids, which was followed by lutein (20%), lactucaxanthin (13%), violaxanthin (11%), and neoxanthin (6%) [bib_ref] Chemical composition and antioxidant capacity of lettuce: Comparative study of regular-sized (Romaine)..., López [/bib_ref]. However, the content of carotenoids is also varying in different lettuce types. There was a moderate association between β-carotene content and leaf color value L*, suggesting that the increasing leaf brightness or luminosity is highly related to β-carotene content [bib_ref] Phytochemicals and overall quality of leafy lettuce (Lactuca sativa L.) varieties grown..., Mampholo [/bib_ref]. Carnat et al. [bib_ref] Characterisation and variation of antioxidant micronutrients in lettuce (Lactuca sativa folium), Carnat [/bib_ref] reported that the amount of β-carotene and lutein were generally higher in red oak leaf than in green lettuce types such as butterhead and Batavia, which confirmed that the carotenoid concentration and the green color chlorophyll content are positively correlated [bib_ref] Distribution and metabolism of dietary carotenoids in humans as a criterion for..., Khachik [/bib_ref]. In contrast, these findings remain controversial in other academic studies. For example, Mou [bib_ref] Genetic variation of beta-carotene and lutein contents in lettuce, Mou [/bib_ref] indicated higher levels of carotenoids in green leaf lettuce than the red lettuce, and a similar content of β-carotene is found in red leaf lettuce and the green one. Therefore, the carotenoid contents are not completely related to leaf pigmentation.
In addition, previous reports have demonstrated that cultivar and other conditions could affect the carotenoid content of lettuce. As for Crisphead with a closed head shape, its leaves obtain less light than the leaves with a relatively open head, this leads to less carotenoid synthesis [bib_ref] Genetic variation of beta-carotene and lutein contents in lettuce, Mou [/bib_ref]. The outer leaves' carotenoids are higher than the inner leaves because of the positive influences of light intensity on carotenoid biosynthesis in the outer leaves [bib_ref] Nutritional quality of outer and inner leaves of green and red pigmented..., Baslam [/bib_ref].
## Chlorophyll
As the main pigment of leafy green vegetables, chlorophyll plays a significant role in determining the health of vegetables. The chlorophyll content quantification can be considered a visible factor in monitoring the nutritional statement of green leafy vegetables such as lettuce since most of the nitrogen is integrated into leaf chlorophyll. In addition, with the senescence after harvest, lettuce chlorophyll will gradually be degraded, which can be considered an indicator of quality [bib_ref] Polyphenol content and antioxidative activity in some species of freshly consumed salads, Heimler [/bib_ref]. Moreover, chlorophyll is thought to be a good source of vitamins E, A, C, K, and β-carotene, and important minerals such as magnesium, potassium, iron, calcium, and essential fatty acids. Chlorophyll has been claimed to have protective effects such as anti-carcinogenic and antimutagenic activities because of its antioxidant properties [bib_ref] Effects of cooking methods on chlorophylls, pheophytins and colour of selected green..., Turkmen [/bib_ref].
López et al. [bib_ref] Chemical composition and antioxidant capacity of lettuce: Comparative study of regular-sized (Romaine)..., López [/bib_ref] demonstrated the differences found in some types of nutritional and bioactive compounds due to the various structures and sizes of the lettuce head. For example, open lettuce heads, such as Romaine with a higher photosynthetic area, had increased compounds such as chlorophylls, sugars, and other relevant metabolites [bib_ref] Effect of leaf position on the distribution of phytochemicals and antioxidant capacity..., Ozgen [/bib_ref]. In contrast, the cultivars having closed heads can block the sunlight penetration, resulting in a lower content of metabolites. On the basis of Mou [bib_ref] Genetic variation of beta-carotene and lutein contents in lettuce, Mou [/bib_ref] , similar effects of head structure can be observed on carotenoid content, which has been mentioned before. The relationship between β-carotene and chlorophyll plays a part in photosynthesis, while the former acts as an accessory pigment. β-carotene can absorb light energy from the chlorophyll at various wavelengths and transfer it to the chlorophyll [bib_ref] Nutrient content of lettuce and its improvement, Mou [/bib_ref]. Chlorophyll is the main photosynthetic pigment that produces the green color in plants and many algae. Thus, a higher total chlorophyll content can be found in green lettuce varieties than in red ones.
## Vitamins
Vitamins are one of the important micronutrients needed for metabolism [bib_ref] Biotechnological production of vitamins, Survase [/bib_ref]. In lettuce, folate, and vitamins C and E are the most common types. Firstly, Wang et al. [bib_ref] Fate of folates during vegetable juice processing-Deglutamylation and interconversion, Wang [/bib_ref] reported total folate is higher in romaine lettuce than in spinach or another popular leafy vegetable. Secondly, two forms of vitamin C, ascorbic acid (AA) and dehydroascorbic (DHAA), both show biological functions in lettuce [bib_ref] Evaluation of analgesic, anti-inflammatory, thrombolytic and hepatoprotective activities of roots of Premna..., Mahmud [/bib_ref] [bib_ref] The oxidation of dehydroascorbic acid and 2,3-diketogulonate by distinct reactive oxygen species, Dewhirst [/bib_ref]. The simultaneous analysis of both formations is necessary. For instance, ascorbic acid also contributes to 24.5% of the overall antioxidant activity in lettuce, which shows that vitamin C is an essential antioxidant [bib_ref] Characterisation and variation of antioxidant micronutrients in lettuce (Lactuca sativa folium), Carnat [/bib_ref]. In contrast, Szeto et al. [bib_ref] Total antioxidant and ascorbic acid content of fresh fruits and vegetables: Implications..., Szeto [/bib_ref] found that the level of vitamin C in lettuce accounts for only 1% of the total antioxidant power. Such distinct findings can be clarified by having dehydroascorbic acid in lettuce. Even though dehydroascorbic acid seems like vitamin C, it has no antioxidant activity [bib_ref] Characterisation and variation of antioxidant micronutrients in lettuce (Lactuca sativa folium), Carnat [/bib_ref]. Unlike folate, lettuce is not a rich source of vitamin C like other vegetables such as broccoli, cauliflower, or pepper [bib_ref] Total phenol, flavonoid, proanthocyanidin and vitamin C levels and antioxidant activities of..., Bahorun [/bib_ref]. Thirdly, vitamin E is found in different forms of tocopherols among four lettuce [bib_ref] Tocopherol and tocotrienol contents of raw and processed fruits and vegetables in..., Chun [/bib_ref]. Loose-leaf has the highest content of 1.06 mg/100 g of edible weight. Indeed, α-tocopherols and γtocopherols are the principal forms in lettuce and the former is the most biologically active. In addition, Szymańska and Kruk [bib_ref] Tocopherol content and isomers' composition in selected plant species, Szymańska [/bib_ref] indicated that α-tocopherol dominates in lettuce leaves, while γ-isomer is abundant in part of the lettuce head. Based on the United States Department of Agriculture (New York), the total tocopherol in lettuce is about 0.5 µg/g −1 with 0.26 between αand γ-conformation [bib_ref] Changes in compositions of galactolipids, triacylglycerols, and tocopherols of lettuce varieties (Lactuca..., Byrdwell [/bib_ref]. In addition, Saini et al. [bib_ref] Stability of carotenoids and tocopherols in ready-to-eat baby-leaf lettuce and salad rocket..., Saini [/bib_ref] observed that the high content of ascorbic acid and tocopherol in baby-leaf lettuce contributes to the shelf life, preserving phenolic compounds.
## Minerals
In terms of minerals, it is the best way to eat many leafy vegetables in the diet to obtain abundant minerals in the body [bib_ref] Dietary iron supplements and Moringa oleifera leaves influence the liver hepcidin messenger..., Saini [/bib_ref]. It is believed that sodium (Na) and potassium (K) play a crucial role in the balance of water and electrolytes and other metabolic functions [bib_ref] The importance of mineral elements for humans, domestic animals and plants-A review, Soetan [/bib_ref]. Using the recommended consumption of Na each day as the standard, the content of Na from lettuce is low. In contrast, the content of K in lettuce is higher than that of Na in lettuce. In addition, calcium (Ca) is another important mineral that benefits bone health and minimizes the risk of osteoporosis. Other minerals such as phosphorus (P), magnesium (Mg), iron (Fe), and zinc (Zn) can also be found in lettuce and vary from the head type and are affected by soil conditions. Similarly, Santos et al. [bib_ref] Multi-elemental analysis of ready-to-eat "baby leaf" vegetables using microwave digestion and high-resolution..., Santos [/bib_ref] also found that the primary mineral in lettuce is iron (Fe). Meanwhile, the content of potassium (K) and calcium (Ca) are also the highest, but low in Na.
## Anti-nutrients compounds
Anti-nutrients are also detected in lettuce species. Such compounds have direct and indirect effects ranging from a mild reaction to a death, which need to be considered. The main anti-nutrients in lettuce include nitrates, phytates, tannins, oxalates, and cyanogenic glycosides. For instance, nitrogen is important for the nutrition and function of lettuce so that plants can influence the metabolic control of lettuce by nitrate and other forms of nitrogen. It is principally detected in cell vacuoles and can be transported in the xylem [bib_ref] Review on: Antinutritional factors in vegetable crops, Sinha [/bib_ref]. The water and nutrients can be carried by xylem from roots to leaves, while the photosynthetic product can be carried by phloem between the leaves and the growth points of the plant, affecting the distribution of nitrate, which means the leaf crops such as lettuce and spinach have relatively high nitrate concentrations. Another influence of this transport mechanism is that old leaves have higher nitrate concentrations than younger leaves [bib_ref] Migration and employment change: Empirical evidence on the spatial and temporal dimensions..., Greenwood [/bib_ref]. The accumulation of nitrate can cause harmful actions. Another anti-nutrient in lettuce is alkaloids, which are the natural bitter compositions of plants with some pharmacological properties. Alkaloids result in gastrointestinal tract and nervous system disorders that disrupt or improperly augment electrochemical transmission [bib_ref] Allelochemicals in plant foods and feedingstuffs: 1. Nutritional, biochemical and physiopathological aspects..., Aletor [/bib_ref]. It is evident that alkaloids have a physiological impact on humans [bib_ref] Goitrogenic food and prevalence of goitre in Sri Lanka, Fernando [/bib_ref]. For example, high tropane alkaloids intake can result in heartbeat rapidly, paralysis, and even death in fatal cases. In addition, a high dose of tryptamines alkaloids can cause staggering and death.
Many plants, including some crops, contain plant toxins that are natural compounds. In many cases, these toxins can be considered defense mechanisms that plants produce to prevent insects from overeating. If one of the plants were eaten enough by an insect, the toxic chemicals could accumulate in their body and reach a high level causing the disrupted function of cells and tissue, then the insect will die. In this way, plant toxins are vital to the natural balance system [bib_ref] An overview of nutritional and antinutritional factors in green leafy vegetables. Hortic, Natesh [/bib_ref]. In addition, green leafy vegetables containing anti-nutrients such as nitrates, oxalates, phytates, cyanogenic glycosides, and tannins influence micronutrient absorption. Some thermal processing can decrease the content of anti-nutrients in lettuce by boiling, cooking, and blanching. Further research is also necessary on various varieties and adopting agronomic practices that can decrease the content and impact of anti-nutrients in green leafy vegetables such as lettuce and improve their nutritional value.
## Bioavailability of bioactive compounds from lettuce
It is critical to mention the ultimate delivery of these compounds when considering food rich in bioactive compounds and their related health, and thus bioavailability is introduced as a parameter. The definition of bioavailability is the extent to which ingested bioactive compounds reach the circulatory system and act on specific sites [bib_ref] Bioaccessibility and bioavailability of phenolic compounds, Shahidi [/bib_ref]. Furthermore, bioaccessibility means the process by which the fraction of bioactive compounds are released from the food matrix through the upper digestive system [bib_ref] Intake and bioaccessibility of total polyphenols in a whole diet, Saura-Calixto [/bib_ref]. Chlorogenic acid, which is the ester of caffeic acid and quinic acid, is one of the most prevalent polyphenols in the human diet, being found mostly in coffee, fruits, and vegetables [bib_ref] Intake and bioaccessibility of total polyphenols in a whole diet, Saura-Calixto [/bib_ref]. For eight days, four groups of rats were given either a food enriched with chlorogenic, caffeic, or quinic acids (250 µmol/d) or a control diet. HPLC-electrospray ionization-tandem mass spectrometry was used to determine the parent chemicals and their metabolites in urine and plasma. There were significant disparities in their levels across the groups.
The recovery of chlorogenic acid in urine was minimal and the total urinary excretion of caffeic acid freed by chlorogenic acid hydrolysis and its tissular methylated metabolites (ferulic and isoferulic acids) was less than 0.5% (mol/mol) of the amount consumed. On the other hand, m-coumaric acid and derivatives of phenylpropionic, benzoic, and hippuric acids were the most abundant metabolites in both urine and plasma. Hippuric acid is derived primarily from the quinic acid moiety, whereas all other metabolites are derived from the caffeic acid moiety. These microbial metabolites account for 57.4% of chlorogenic acid consumption (mol/mol). Such a significant quantity of microbial metabolites indicates that chlorogenic acid's bioavailability is highly dependent on its metabolism by the gut microflora. They are highlighted for their possible role in understanding the biological effects of dietary polyphenols [bib_ref] Chlorogenic acid bioavailability largely depends on its metabolism by the gut microflora..., Gonthier [/bib_ref]. Caffeic acid has a 14.7% absolute bioavailability in rats and a 12.4% intestinal absorption in the caco-2 cell model [bib_ref] Bioavailability of caffeic acid in rats and its absorption properties in the..., Wang [/bib_ref]. However, other research discovered that caffeic acid, p-coumaric acid, and ferulic acid are much more bioavailable than chlorogenic acid, despite being excreted faster. The bioavailability of ferulic acid in humans was investigated by tracking the pharmacokinetics of excretion in proportion to intake. The findings indicate that the peak period for maximum urinary excretion is around 7 h and that the recovery of ferulic acid in the urine is approximately 11-25% of that consumed, based on the total free ferulic acid and feruloyl glucuronide excreted [bib_ref] Bioavailability of ferulic acid, Bourne [/bib_ref]. According to a literature review on kaempferol digestion and absorption, the most absorbable forms include kaempferol rutinoside and glucoside in tea, endive glucuronide and glucoside, and broccoli sophoroside. Kaempferol metabolites were detected in the plasma as aglycone and glucuronide, and in the urine as sulfate; however, further human investigations are required to measure kaempferol absorption from meals [bib_ref] Dietary quercetin and kaempferol: Bioavailability and potential cardiovascular-related bioactivity in humans, Dabeek [/bib_ref]. Apigenin obtained from food enters the human circulation in vivo and may therefore display physiologic effects [bib_ref] Bioavailability of apigenin from apiin-rich parsley in humans, Meyer [/bib_ref].
## Digestion, absorption, and metabolism
Bioactive compounds absorption process from food starts with the chewing and digestion of food in the mouth, which is the oral phase. The roles of chewing and digestive enzymes, especially α-amylase, play in reducing food particle size and the beginning of phenolic compounds release [bib_ref] Overview of metabolism and bioavailability enhancement of polyphenols, Lewandowska [/bib_ref]. After that, the digest will transfer to the stomach, where the gastric phase begins. During this phase, the pepsin, as the chief digestive enzyme in the stomach, acidic pH, and mixing can enhance the breakdown of the food matrix, causing further release [bib_ref] Dietary factors affecting polyphenol bioavailability, Bohn [/bib_ref]. Following the gastric phase, the gastric chyme is gradually released into the upper small intestine, also called the duodenum, where the small intestine phase begins under an alkaline environment. At this phase, the pancreas and liver release enzymes and bile salts enhance the most food degradation. As the food matrix is completely digested into absorbable macronutrient units, further compounds will subsequently be released. The undigested phenolic compounds continue to move to the lower intestinal compartments, which are the colon parts, where they can further metabolize by native microbiota. The produced metabolites can circulate in the bloodstream or be delivered to target tissues by plasma proteins and lipoproteins. The metabolism of phenolic compounds in humans, as an example, was summarized in [fig_ref] Figure 3: Schematic layout of the digestive pathway of phenolic compounds[81] [/fig_ref].
Bioactive compounds absorption process from food starts with the chewing and digestion of food in the mouth, which is the oral phase. The roles of chewing and digestive enzymes, especially α-amylase, play in reducing food particle size and the beginning of phenolic compounds release [bib_ref] Overview of metabolism and bioavailability enhancement of polyphenols, Lewandowska [/bib_ref]. After that, the digest will transfer to the stomach, where the gastric phase begins. During this phase, the pepsin, as the chief digestive enzyme in the stomach, acidic pH, and mixing can enhance the breakdown of the food matrix, causing further release [bib_ref] Dietary factors affecting polyphenol bioavailability, Bohn [/bib_ref]. Following the gastric phase, the gastric chyme is gradually released into the upper small intestine, also called the duodenum, where the small intestine phase begins under an alkaline environment. At this phase, the pancreas and liver release enzymes and bile salts enhance the most food degradation. As the food matrix is completely digested into absorbable macronutrient units, further compounds will subsequently be released. The undigested phenolic compounds continue to move to the lower intestinal compartments, which are the colon parts, where they can further metabolize by native microbiota. The produced metabolites can circulate in the bloodstream or be delivered to target tissues by plasma proteins and lipoproteins. The metabolism of phenolic compounds in humans, as an example, was summarized in [fig_ref] Figure 3: Schematic layout of the digestive pathway of phenolic compounds[81] [/fig_ref].
## Phenolic compounds
In nature, phenolics are mainly present in the form of esters, glycosides, and polymers, which need to be hydrolyzed and metabolized in the environment or by the natural microbiota of the digestive system before being absorbed [bib_ref] Changes in bioactive compounds and antioxidant activity of plant-based foods by gastrointestinal..., Ketnawa [/bib_ref]. Their bioavailability regulates the release and activation of phenolic compounds, and the metabolic form is transferred to the target organ through the blood and then passes through the gastrointestinal tract. According to previous literature, phenolic compounds are digested around 48% in the stomach and intestinal stages, with the remaining 42% bioaccessible in the colon. They also reported that 10% of phenolic compounds are undigested and remain intact within the food matrix. As for most phenolic compounds, only the aglycone form with highly hydrophobic characteristics can pass across physiological membranes through diffusion. In addition, the glycosidic form of phenolic compounds has undoubtedly influenced absorption in the intestine phase [bib_ref] Digestion and bioavailability of bioactive phytochemicals, Karaś [/bib_ref].
## Carotenoids
Carotenoids will be released from the food matrix and then incorporated into the mixed micelles that are formed by carotenoids, acylglycerols, cholesterol, and phospholipids before they can be absorbed. The mixed micelles can transport liposoluble compounds to the intestinal epithelium [bib_ref] Carotenoids bioavailability from foods: From plant pigments to efficient biological activities, Fernandez-Garcia [/bib_ref]. Hence, the formation of micelles can be considered one of the important parameters that influence the bioavailability of carotenoids. Furthermore, Hof et al. [bib_ref] Dietary factors that affect the bioavailability of carotenoids, Hof [/bib_ref] demonstrated that the bioaccessibility of carotenoids in animals and humans is quite low due to the structure of carotenoids providing interaction with other compounds in the food matrix.
## Vitamin c
There are reduced and oxidized forms in vitamin C. Both can be digested and absorbed throughout the small intestinal epithelium by various mechanisms. A sodium-dependent active transporter (SVCT1) was used for ascorbic acid absorption [bib_ref] Glucose Modulates Vitamin C Transport in Adult Human Small Intestinal Brush Border..., Malo [/bib_ref]. Additionally, with the help of a glucose transporter, dehydroascorbic acid can be absorbed by promoting diffusion in the duodenum and jejunum [bib_ref] Distinct mechanisms of transport of ascorbic acid and dehydroascorbic acid in intestinal..., Fujita [/bib_ref].
## Factors affecting the bioavailability
The internal and external parameters can potentially affect the ultimate biological activity of bioactive compounds in the diet [fig_ref] Figure 3: Schematic layout of the digestive pathway of phenolic compounds[81] [/fig_ref]. For example, enzyme action is another important factor, which causes molecular transformation in compounds depending on their solubility as well as destroying chemical bonds in phenolics and proteins, carbohydrates, and lipids with cell wall tissues, thereby enhancing the phenolic compounds released [bib_ref] Effects of cooking and in vitro digestion of rice on phenolic profiles..., Ti [/bib_ref]. A different pH in each phase environment may impact a fraction of the bioactive compounds by degradation, modification, or transformation of compounds into various structural forms with distinct chemical characteristics, bioaccessibility, bioavailability, and biological activities. Moreover, when food comes into the mouth and ingestion starts, different compounds release from the food matrix, with interactions that influence phytochemical bioavailability [fig_ref] Figure 4: Parameters affecting the bioavailability of bioactive compounds[89] [/fig_ref].
## Health benefits and risks of lettuce
Patients with severe or end-stage renal illness exhibit uremic pruritus often and prominently. The lack of effective therapeutics for pruritus associated with a renal illness often causes several complications for these individuals and makes it difficult to choose a suitable treatment. Using a molecular docking approach, researchers explored the interactions between nineteen natural bioactive components of lettuce (Lactuca sativa L.) and human kappa opioid receptors. According to the in silico docking experiments, most ligands conferred greater antipruritic effectiveness than gabapentin. γ-tocopherol, δ-tocopherol, and campesterol had the greatest protein-binding affinities [bib_ref] Pharmacological evidence for the use of Cissus assamica as a medicinal plant..., Dutta [/bib_ref].
Various kinds of lettuces on the market provide texture, taste, and flavor and provide a variety of phytochemicals with related health benefits [bib_ref] Nutritional attributes of salad vegetables, Hedges [/bib_ref]. The frequency of consumption means that lettuce nutrients account for a considerable proportion of dietary intake. Altunkaya et al.observed a synergistic antioxidant effect in lettuce, which via inhibiting polyphenol oxidase (PPO) activity associated with quercetin and α-tocopherol.
## Health benefits and risks of lettuce
Patients with severe or end-stage renal illness exhibit uremic pruritus often and prominently. The lack of effective therapeutics for pruritus associated with a renal illness often causes several complications for these individuals and makes it difficult to choose a suitable treatment. Using a molecular docking approach, researchers explored the interactions between nineteen natural bioactive components of lettuce (Lactuca sativa L.) and human kappa opioid receptors. According to the in silico docking experiments, most ligands conferred greater antipruritic effectiveness than gabapentin. γ-tocopherol, δ-tocopherol, and campesterol had the greatest protein-binding affinities [bib_ref] Pharmacological evidence for the use of Cissus assamica as a medicinal plant..., Dutta [/bib_ref].
Various kinds of lettuces on the market provide texture, taste, and flavor and provide a variety of phytochemicals with related health benefits [bib_ref] Nutritional attributes of salad vegetables, Hedges [/bib_ref]. The frequency of consumption means that lettuce nutrients account for a considerable proportion of dietary intake. Altunkaya et al.observed a synergistic antioxidant effect in lettuce, which via inhibiting polyphenol oxidase (PPO) activity associated with quercetin and α-tocopherol. Meanwhile, the existence of anthocyanins in iceberg lettuce triggers the antioxidant capacity, through both enzyme inhibition and the sequestration of trace elements involved in the production of free radicals [bib_ref] Antioxidant properties of selected varieties of lettuce (Lactuca sativa L.) commercially available..., Gan [/bib_ref] [bib_ref] Action mechanism and cardiovascular effect of anthocyanins: A systematic review of animal..., Reis [/bib_ref]. Consequently, the antioxidative combination of several bioactive compounds in lettuce contributes to numerous pharmacological properties, including cardio-protective effect, anti-cancer, and anti-diabetic [fig_ref] Figure 5: Health-promoting benefits of lettuce [/fig_ref].
## Healthy benefits
## Cardio-protective effect
Cardiovascular diseases (CVDs) are types of chronic non-infectious diseases that are one of the major threats to people's health, and are related to a large number of complex risk factors such as hypertension, hyperlipidemia, diabetes, and metabolic syndrome [bib_ref] Cardiovascular disease risk factors: A childhood perspective, Praveen [/bib_ref]. Some previous epidemiological studies prove that consuming vegetables is positively related to treating CVDs, especially for some green leafy vegetables such as lettuce. For ex-
## Healthy benefits
## Cardio-protective effect
Cardiovascular diseases (CVDs) are types of chronic non-infectious diseases that are one of the major threats to people's health, and are related to a large number of complex risk factors such as hypertension, hyperlipidemia, diabetes, and metabolic syndrome [bib_ref] Cardiovascular disease risk factors: A childhood perspective, Praveen [/bib_ref]. Some previous epidemiological studies prove that consuming vegetables is positively related to treating CVDs, especially for some green leafy vegetables such as lettuce. For example, Nicolle et al. [bib_ref] Health effect of vegetable-based diet: Lettuce consumption improves cholesterol metabolism and antioxidant..., Nicolle [/bib_ref] investigated the diet of rats containing 20% lettuce can have a cardioprotective effect by improving cholesterol metabolism and plasma antioxidant ability.
Lettuce tends to have a number of interesting impacts on CVD factors in various mechanisms owing to its fiber, the availability of antioxidants, and probably other phytochemicals. It is believed that fiber and antioxidant compounds have positive effects on cholesterol metabolism and the prevention of CVDs. Firstly, the lower cholesterol impact of lettuce can be attributed to the fiber part. Pectin, soluble fibers affecting lipid metabolism both in animal and human bodies, has been verified to reduce digestive cholesterol absorption [bib_ref] Plasma cholesterol-lowering effect on rats of dietary fiber extracted from immature plants, Nishimura [/bib_ref]. The cholesterol absorption inhibition mechanisms primarily involve disrupting micelle formation and delaying the cholesterol transfer through the unstirred layer [bib_ref] Interaction of bile acids and cholesterol with non-systemic agents having hypocholesterolemic properties, Stedronsky [/bib_ref]. Hypocholesterolemia impacts can also be found in dietary fiber by promoting fecal excretion of overall steroids as bile acids. That reflects the retention of bile acids in the viscous medium and the acceleration of biliary flow. In addition, there are also some indirect impacts on the cholesterol mechanism exerted by fiber. Short-chain fatty acids such as propionate, which is considered the most effective cholesterol-lowering agent, are produced by fiber fermentation in the large intestine. All these mechanisms combined can reduce the accumulation of plasma cholesterol [bib_ref] Health effect of vegetable-based diet: Lettuce consumption improves cholesterol metabolism and antioxidant..., Nicolle [/bib_ref]. Secondly, lettuce can also provide many antioxidant compounds, such as vitamin E, C, carotenoids, and polyphenols, which have health benefits [bib_ref] The antioxidant activity of regularly consumed fruit and vegetables reflects their phenolic..., Proteggente [/bib_ref]. The heart's defense against lipid peroxidation can have protective effects against cardiovascular diseases, while the heart is one of the vital target tissues for reactive oxygen species (ROS). Nicolle et al. [bib_ref] Health effect of vegetable-based diet: Lettuce consumption improves cholesterol metabolism and antioxidant..., Nicolle [/bib_ref] measured that lower peroxidation in the heart has a relationship with the increase in the vitamin E/TG ratio in the plasma. After consuming lettuce, the strong antioxidant ability in plasma was related to a significant increase in vitamin C and vitamin E, and is less effective than vitamin C in the post-prandial period in improving the antioxidant ability. However, various phenolic compounds in lettuce may also improve the antioxidant capacity. LDL oxidation, which can be inhibited effectively by polyphenols in lettuce, is one of the main causes of atherosclerosis [bib_ref] Pomegranate juice consumption reduces oxidative stress, atherogenic modifications to LDL, and platelet..., Aviram [/bib_ref]. García-Lafuente et al. [bib_ref] Flavonoids as anti-inflammatory agents: Implications in cancer and cardiovascular disease, García-Lafuente [/bib_ref] reported that other mechanisms of phenols to protect against CVDs are antiplatelet, anti-inflammatory effects, increasing HDL, and improving endothelial function.
## Anti-cancer effect
Lettuce can positively affect cancer prevention because it contains some phytochemicals, and some clinic researchers have stated that these have health benefits. Iodinebiofortified lettuce causes cancer cells to produce reactive oxygen species (ROS), which leads to anticancer effects via the activation of programmed cell death in cancer cells [bib_ref] Anti-and pro-oxidant potential of lettuce (Lactuca sativa L.) biofortified with iodine by..., Sularz [/bib_ref].
Firstly, β-carotene and ascorbic acid are potential nutrients in lettuce that can decline the risk of colorectal cancer. Case-control research has been conducted in Italy, having 1584 colorectal patients with a family history of disease and 2879 control objectives. Assessing the daily food consumption suggested an adverse relationship between lettuce consumption and colorectal cancer. Whereas Ca, vitamin E, and folate in lettuce have no significant association with this cancer [bib_ref] Risk factors for colorectal cancer in subjects with family history of the..., Fernandez [/bib_ref]. Moreover, an investigation of non-smoking patients reported that a daily diet with lettuce could reduce lung cancer incidence owing to carotenoids, β-carotene, and retinol [bib_ref] A multicenter case-control study of diet and lung cancer among non-smokers, Brennan [/bib_ref]. Secondly, the chlorophyll pigment in green varieties with dark colors can have a protective effect against specific cancer, such as colon cancer and liver cancers. The mechanism of chlorophyll is binding together hydrocarbons, aflatoxins, and other hydrophobic molecules, which are related to cancer, rather than eliminating them [bib_ref] Nutrition and cancer: A review of the evidence for an anti-cancer diet, Donaldson [/bib_ref]. In contrast, other studies have claimed that low consumption of green leafy vegetables in the diet could increase cancer risk. Thirdly, lettuce contains a high content of phenolic acids, dicaffeoyl tartaric acid, and chlorogenic acid, while having a low content of flavonols, namely, quercetin 3-O-glucuronide. One of the most vital utilizations of plant-derived therapeutics formed by phenols is for preventing cancer and chemotherapy. Polyphenols can decline the growth rate of tumors, thus having an anti-cancer effect [bib_ref] Bioactive phenolic compounds: Production and extraction by solid-state fermentation. A review, Martins [/bib_ref] [bib_ref] Comprehensive review on naringenin and naringin polyphenols as a potent anticancer agent, Rauf [/bib_ref] [bib_ref] Berberine as a potential anticancer agent: A comprehensive review, Rauf [/bib_ref]. The anti-cancer mechanism has been reported by Terry, whereby chemoprevention impacts polyphenols such as antioxidants, reducing free radicals and ROS, thus decreasing their damaging effects on DNA. The MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) test and flow cytometry were used to determine the antiproliferative and apoptotic effects while simultaneously studying the genes governing apoptosis and cell cycle in this cell culture model. Additionally, we examined the inhibitory and kinetic properties of the aqueous L. taraxacifolia (African lettuce) extract using recombinant human CYP450 isozyme model systems (CYP1A2, CYP2C9, and CYP2C19). L. taraxacifolia significantly inhibited the development of WHC01 cancer cells. The majority of genes involved in the cell cycle were downregulated. A cell cycle study revealed that L. taraxacifolia extract induced a G0-G1 cell cycle arrest in WHC01 cells, followed by morphological alterations. The treatment with L. taraxacifolia extract resulted in a 50-70% reduction in the expression levels of CYP1A2 and CYP2C19. In a reversible and time-dependent manner, L. taraxacifolia extract inhibited recombinant CYP1A2, CYP2C9, and CYP2C19. This research sheds fresh light on the potential anticancer properties of L. taraxacifolia, an extensively utilized medicinal plant in regions of Africa and around the globe, particularly by cancer patients [bib_ref] African lettuce (Launaea taraxacifolia) displays possible anticancer effects and herb-drug interaction potential..., Thomford [/bib_ref]. The anti-proliferative effects of phenolic extracts from lettuce grown under low nitrogen conditions were superior to those of phenolic extracts from lettuce grown under adequate nitrogen conditions, in part because they interfered with the cell cycle and induced apoptosis, when compared to those of phenolic extracts from lettuce grown under adequate nitrogen conditions [bib_ref] Alteration of phenolic composition in lettuce (Lactuca sativa L.) by reducing nitrogen..., Zhou [/bib_ref]. Apigenin is a flavonoid produced from edible plants that has been shown to be effective as an anticancer drug in a number of experimental and biological investigations. In many kinds of tumors, including breast, lung, liver, skin, and blood cancers, as well as prostate cancer and pancreatic cancer and cervical cancer and stomach cancer, it causes cell development to be arrested and apoptosis to occur via altering multiple signaling pathways. Apigenin promotes apoptosis by the activation of the extrinsic caspase-dependent pathway, which is characterized by the upregulation of the mRNA expressions of caspase-3, caspase-8, and TNF-α. It stimulates the intrinsic apoptotic pathway, as indicated by the elevation of cytochrome c, Bax, and caspase-3 in human prostate cancer PC-3 cells, but the levels of caspase-8, TNF-α, and B-cell lymphoma 2 remained unaltered. Apigenin therapy results in a considerable downregulation of matrix metallopeptidases-2, -9, Snail, and Slug, which inhibits the invasion of cancer cells. After treatment with apigenin, the expression of the transcription factors NF-κB p105/p50, PI3K, and Akt and the phosphorylation of p-Akt all drop. On the other hand, Apigenin-mediated therapy drastically lowers the expression of the pluripotency marker Oct3/4 protein, which may be related to the downregulation of PI3K/Akt/NF-κB signaling in the cells [bib_ref] Apigenin as an anticancer agent, Imran [/bib_ref].
## Anti-diabetic effect
One of the treatment plans can target diabetes, such as α-amylase and α-glucosidase [bib_ref] Phytochemical screening, Antioxidant, Thrombolytic, alpha-amylase inhibition and cytotoxic activities of ethanol extract..., Kabir [/bib_ref] [bib_ref] The multifunctional role of herbal products in the management of diabetes and..., Rahman [/bib_ref] , in which starch can be hydrolyzed into glucose entering the blood circulation. Thereby, some interesting studies investigated the natural bioactive compounds having an antidiabetic effect. Lettuce leaves can be considered a phytonutrient storehouse with high carotenoid content such as lutein, lactucaxanthin, and β-carotene [bib_ref] Characterisation of polyphenols and antioxidant properties of five lettuce varieties and escarole, Llorach [/bib_ref] , while lactucaxanthin is not commonly found in other plants. researched that the purity of lactucaxanthin separated from lettuce is 96% verified by HPLC and LCMS. This molecule can inhibit α-amylase and α-glucosidase activities, providing a way to reduce post-prandial hyperglycemia by controlling the breakdown of starch during digestion. On the other hand, plants rich in polyphenols also have the effect of anti-diabetic. The inhibitory effect of polyphenols on intestinal glycosidases and glucose transporter has been researched by Dembinska-Kiec, et al. [bib_ref] Antioxidant phytochemicals against type 2 diabetes, Dembinska-Kiec [/bib_ref]. Cheng et al.experimented that some obese hyperglycemic mice fed with the diet with the addition of red lettuce varieties at the dose of 100 or 300 mg/kg to alleviate diabetes diseases due to the high content of polyphenols.
Anthocyanins, as the polyphenolic pigments, were attributed to the red color of lettuce with antioxidant, anti-inflammatory, and anti-diabetes effects. Dietary addition anthocyanins extracted from bilberry and lettuce can promote hyperglycemia and insulin sensitivity in mice with type 2 diabetes. Furthermore, dietary supplementation of anthocyanin cyanidin 3-glucoside feeding for 8 weeks by diabetic mice can decrease the damage of hepatic oxidants and avoid hepatic steatosis [bib_ref] Nutraceutical-Based Pharmacological Intervention in the Management of Liver Diseases, Shah [/bib_ref]. A similar result was found in another study, the addition of anthocyanin 3-glucoside to the high-fat diet feeding for 5 weeks of mice can decrease the level of blood glucose, increase insulin sensitivity, attenuate liver steatosis, and reduces inflammatory cytokines in adipose tissue [bib_ref] Cyanidin 3-glucoside attenuates obesity-associated insulin resistance and hepatic steatosis in high-fat diet-fed..., Guo [/bib_ref].
## Immunomodulatory effect
In one experiment, the polysaccharides from stem lettuce (SLP) were further purified by anion exchange chromatography followed by size exclusion chromatography to provide two pure polysaccharides, SLP-1 and SLP-2. Galacturonic acid, galactose, and arabinose were the most abundant sugars in SLP-1, which had a molecular weight of 90 KDa and a molar ratio of 17.6:41.7:33.9. In the molar ratio of 11.5:69.5:9.3:8.2, mannose, galacturonic acid, galactose, and arabinose were the primary constituents of SLP-2, which had a molecular weight of 44 KDa and was mostly consisted of mannose, galacturonic acid, galactose, and arabinose. Furthermore, both pure polysaccharides include sulfate radicals, have triple-helical topologies, and have the ability to enhance macrophage proliferation without causing cytotoxicity in the presence of other factors. SLP-2 was shown to be more effective than SLP-1 in stimulating phagocytic and nitric oxide production. As a consequence of the findings, it is possible that polysaccharides from stem lettuce may be investigated as immunomodulatory agents in the pharmaceutical and functional food industries [bib_ref] Purification, characterization and immunomodulatory activity of polysaccharides from stem lettuce, Nie [/bib_ref]. A study on the murine macrophage cell line RAW264.7 was conducted to determine the immunomodulatory properties of Green lettuce extracts (GLE) and their probable mechanisms of action. GLE considerably boosted NO levels in RAW264.7 cells and the expression of immunomodulators such as iNOS, COX-2, IL-1, IL-6, IL-12, TNFalpha, and MCP-1 in the cells. GLE was also shown to increase NO levels in RAW264.7 cells dramatically. Despite the fact that GLE stimulated ERK1/2, p38, JNK, and NF-κB, GLE-mediated production of immunomodulators was reliant on p38, JNK, and NF-κB for its effectiveness. In addition, TLR4 suppression prevented the GLE-mediated production of immunomodulators as well as activation of the p38, JNK, and NF-κB transcriptional factors. Together, our findings revealed that the TLR4-MAPK/NF-κB-signaling pathways were involved in GLE-induced macrophage activation and that GLE might be developed as a potential immunomodulatory functional food ingredient [bib_ref] Immune-Enhancing Effects of Green Lettuce (Lactuca sativa L.) Extracts through the TLR4-MAPK/NF-κB..., Seo [/bib_ref] [bib_ref] Exploring the Immune-Boosting Functions of Vitamins and Minerals as Nutritional Food Bioactive..., Mitra [/bib_ref].
## Hepatoprotective effect
The preventive effect of L. serriola against paracetamol-induced hepatotoxicity has been shown. Silymarin was prescribed as a routine medication. The hepatoprotective impact of the extract was assessed using biochemical markers in the liver, antioxidant enzymes, and a lipid profile in the blood. It was discovered that the methanolic extract was the most powerful in in vitro antioxidant experiments. Treatment with paracetamol raised the levels of hepatic biomarkers and the lipid profile of the blood while decreasing the levels of antioxidant enzymes. Restoration of liver biomarkers, blood lipid profile, and antioxidant enzymes was seen after pre-treatment with Lactuca serriola. L. serriola has hydroxyl (-OH), carboxylic acid (C=O), and alkene (C=C) groups, according to the results of the FTIR analysis. The presence of polyphenolic components in the methanolic extract of L. serriola was discovered using HPLC analysis. It is concluded that the methanolic extract of Lactuca serriola has hepatoprotective potential, which may be attributed to the presence of polyphenolic components and the antioxidant activities of these compounds [bib_ref] Anti-oxidant and hepatoprotective effects of Lactuca serriola and its phytochemical screening by..., Awan [/bib_ref]. When compared to the group that received only CCl 4 , the ethanol leaf extract of Launaea taraxacifolia reduced the acute increase in serum aspartate aminotransferase (AST), alanine aminotransferase (ALT), and alkaline phosphatase (ALP) levels caused by CCl 4 . As a result, it has been hypothesized that the extract has hepatoprotective properties against CCl 4 -induced liver damage through attenuating oxidative stress [bib_ref] Hepatoprotective Potential of Ethanol Leaf Extract of Launaea taraxacifolia Willd.(Asteraceae) on Carbon..., James [/bib_ref]. The efficacy of an ethanolic extract of lettuce (Lactuca sativa L. var. longifolia) leaves in protecting the reproductive system of rats against the toxicity produced by carbon tetrachloride (CCl 4 ) was investigated. The induction of oxidative stress by CCl 4 in the rat results in a reduction in the rise in body weight and relative testis weight. It also causes a significant rise in the levels of thiobarbituric acid reactive substances and nitrites in the testicles, as well as a corresponding drop in reduced glutathione and different antioxidant enzymes. CCl 4 therapy resulted in a drop in the serum levels of testosterone, luteinizing hormone, and follicle-stimulating hormone, as well as a rise in the levels of estradiol and prolactin. The histopathology of CCl 4 -treated rats revealed a partial degeneration of germ and Leydig cells, as well as abnormalities in spermatogenesis, as a result of the treatment. With the use of lettuce extract once a week for ten weeks, the levels of reactive chemicals such as thiobarbituric acid and nitrite are reduced, while the levels of antioxidant enzymes such as catalase, peroxidase, superoxide dismutase, glutathione peroxidase, and glutathione are increased. When CCl4-intoxicated rats were given lettuce extract, the serum levels of testosterone, luteinizing hormone, follicle-stimulating hormone, estradiol, prolactin, histology, body weight, and relative testis weight were all returned to near-normal levels, as did their histology. The findings clearly suggest that therapy with lettuce extract enhances the antioxidant defense system against CCl 4 -induced toxicity. They give evidence that it may have a therapeutic function in treating free radical-mediated diseases in humans [bib_ref] Protective effects of Lactuca sativa ethanolic extract on carbon tetrachloride induced oxidative..., Hefnawy [/bib_ref].
## Neuroprotective activity
The ethyl acetate portion of lettuce prevented the neurotoxic effects of glucose/serum deprivation (GSD) (Lactuca sativa). In PC12 cells, lettuce reduced the increased Bax and caspase-3 proteins and decreased Bcl-2 caused by GSD [bib_ref] Mechanism of protective effect of lettuce against glucose/serum deprivation-induced neurotoxicity, Ghorbani [/bib_ref]. Lactate dehydrogenase release and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide reduction assays were used to assess the neuroprotective effects of the phenolic extract of romaine lettuce and its pure caffeic acid derivatives (isochlorogenic, chlorogenic, chicoric, and caffeic acids) in PC-12 cells. In 100 g of fresh romaine lettuce, total phenolics and total antioxidant capacity averaged 22.7 mg of gallic acid equivalents and 31.0 mg of vitamin C equivalents, respectively. In a dose-dependent manner, the phenolic extract of romaine lettuce rescued PC-12 cells from oxidative damage produced by H 2 O 2 . One of the phenolics in romaine lettuce, isochlorogenic acid, demonstrated more neuroprotection than the other three caffeic acid derivatives discovered in the lettuce. Despite having lower levels of phenolics and antioxidant capacity than other common vegetables, romaine lettuce's contribution to total antioxidant capacity and anti-neurodegenerative impact in human diets would be greater due to a higher daily per capita intake than other common vegetables [bib_ref] Antineurodegenerative effect of phenolic extracts and caffeic acid derivatives in romaine lettuce..., Im [/bib_ref]. Lettuce glycoside B has a neuroprotective effect on cerebral ischemia-reperfusion damage, which may be due to its capacity to upregulate nerve growth factor and neurotrophin-3 expression in the brain cortex during the late stages of ischemia [bib_ref] Lettuce glycoside B ameliorates cerebral ischemia reperfusion injury by increasing nerve growth..., Zhan [/bib_ref] [fig_ref] Figure 5: Health-promoting benefits of lettuce [/fig_ref].
## Antioxidant properties
Polyphenol oxidase was extracted from fresh lettuce for the purpose of a research investigation. Both the temperature, which should be kept at 40 - C, and the pH level should be maintained at 7.0. It has been shown that catechin, catechol, chlorogenic acid, caffeic acid, and gallic acid may all be used as substrates by lettuce polyphenol oxidase. The results showed that chlorogenic acid had the highest substrate specificity compared to the other substrates that were tested. Some compounds easily inhibited the activity of lettuce polyphenol oxidase. The enzyme lettuce polyphenol oxidase was subjected to tests against possible inhibitors, including ascorbic acid, cysteine, oxalic acid, and citric acid. Cysteine was the most efficient inhibitor of the three. It also evaluated how much total phenol and total antioxidant activity were present in the sample when these inhibitors were present at room temperature and refrigerator temperature. The total antioxidant activity of lettuce was raised by the addition of ascorbic acid and cysteine, while the addition of citric and oxalic acids had no influence on the level of total antioxidant activity. Ascorbic acid and cysteine worked together to prevent oxidation of the phenolic compounds in lettuce [bib_ref] Effect of various inhibitors on enzymatic browning, antioxidant activity and total phenol..., Altunkaya [/bib_ref]. In a study, the antioxidant activity, both hydrophilic and lipophilic, as well as the total phenolic content of the leaves of three distinct cultivars of Lactuca sativa L. (Iceberg, Romaine, and Baby head) were compared and contrasted (stem, inner, medium, and outermost leaves). Romaine had the greatest levels of hydrophilic and lipophilic antioxidant activity, and its phenolic content was also greater than that of Iceberg and Baby head lettuces. Romaine also had the highest phenolic content. While the hydrophilic antioxidant activity exhibits a non-specific distribution, the lipophilic antioxidant activity rose dramatically from the stem to the outermost leaves. This suggests that the lipophilic antioxidant may have a protective function in mature or light-exposed leaves [bib_ref] pH dependent antioxidant activity of lettuce (L. sativa) and synergism with added..., Altunkaya [/bib_ref]. The influence of pH on the antioxidant activities of combinations of lettuce extract with quercetin, green tea extract, or grape seed extract was investigated. This was undertaken to reduce Fremy's salt in an aqueous solution using direct electron spin resonance spectroscopy and in l-α-phosphatidylcholine liposome peroxidation assay measured following the formation of conjugated dienes. When employing both of these approaches, the radical-scavenging impact of each and every phenolic antioxidant that was investigated exhibited an increasing trend with rising pH values. When used in combination with lettuce extract, quercetin, green tea extract, and grape seed extract operated in a synergistic manner to prevent the oxidation of peroxidating liposomes, with quercetin demonstrating the greatest benefit. It is owing to an increase in their electron-donating capacity upon deprotonation as well as to their stability in alkaline solutions leading to the polymerization process that the antioxidant activity of the phenols raises with increasing pH. This rise is pH dependent. These polymerization processes of polyphenolic antioxidants might result in enhanced radical scavenging activity as a consequence of the formation of additional oxidizable hydroxyl moieties in the polymeric products of these reactions [bib_ref] pH dependent antioxidant activity of lettuce (L. sativa) and synergism with added..., Altunkaya [/bib_ref]. Interestingly, both the yield and the antioxidant activity of lettuce are greatly improved by using LED lighting with the spectrum B435/R663 at a ratio of 1.25 ± 0.1 [bib_ref] A study of the prevalence of thalassemia and its correlation with liver..., Palit [/bib_ref]. A higher concentration of certain vitamins, antioxidant enzymes, and polyphenolic compounds is produced in the enhanced plants because of the application of iodine compounds, which improves the antioxidant capacity of lettuce [bib_ref] Anti-and pro-oxidant potential of lettuce (Lactuca sativa L.) biofortified with iodine by..., Sularz [/bib_ref].
## Other health benefits
The mechanism of aging is an accumulation of numerous adverse modifications in the cells and tissues when the ages increase because of free radical/oxidative changes leading to the risks of disease increases. The oxidative damage will happen even in normal conditions. While the damage rates will be high when the aging is due to the poor capacity of antioxidation and repairing [bib_ref] Alterations in antioxidant enzymes during aging in humans, Rizvi [/bib_ref]. It has been shown that the consumption of antioxidantrich food, such as a polyphenolic diet, is efficient in decreasing the harmful influence of aging and behavior. Firstly, polyphenols in lettuce are good for alleviating the deleterious effects of aging on the brain and even the nervous system. More specifically, the most important role of food polyphenols in protecting the brain during aging is that these composites can cross the blood-brain barrier (BBB), which strictly controls the flow of metabolites, nutrients, and drugs into the brain. Secondly, Vitamin A belongs to retinoids, which are beneficial for anti-aging skin treatments. Moreover, as another compound, vitamin C is good for your skin that can maintain skin at a young status and accelerate wound healing. Sufficient consumption of vitamin A and vitamin C, which lettuce can be considered a good source, is necessary for developing and maintaining collagen, which provides the structure for the skin.
Regular intake of lettuce can help reduce bone loss. The leafy green vegetable can be considered one of the good natural sources of vitamin K such as romaine lettuce. Some research has found that vitamin K 2 can help increase bone density and even better reduce the risk of osteoporosis than calcium can. Apart from bone-building and maintaining a skeletal structure in healthy status, vitamin K can also play an important role in blood clotting, treating bruises, and helping bone calcification [bib_ref] Vitamin K-The ignorant nutrient, Chaudhary [/bib_ref]. Furthermore, we need to metabolize the food we intake properly because it can provide the energy needed for our daily routine. Lettuce consumption can promote the process of metabolism in our body due to the presence of minerals such as iron, magnesium, and potassium. Moreover, the vitamin B complex contained in lettuce is also able to help our metabolism.
## Health risks
Anti-nutrients are the primary factor to cause consumption risks in lettuce. Though the content of these compounds is controlled under a dangerous level in lettuce cultivation, it cannot be ignored; regarding the above-mentioned nitrate, which is perceived as a purely harmful dietary component, its excessive intake can lead to infantile methemoglobinemia, carcinogenesis, and possibly even teratogenesis [bib_ref] Nitrate in vegetables: Toxicity, content, intake and EC regulation, Santamaria [/bib_ref]. Ysart et al. [bib_ref] Monitoring for nitrate in UK-grown lettuce and spinach, Ysart [/bib_ref] detected an average concentration of 1085 mg/kg (50 to 3209 mg/kg) in several lettuce samples, which was a little over, in the comparison with the European Normative. In the meantime, some metabolites, such as nitrite and nitrogen oxide production, may cause methemoglobinemia and gastric cancer [bib_ref] Monitoring of nitrites and nitrates levels in leafy vegetables (Spinach And Lettuce):..., Iammarino [/bib_ref]. In addition, the physiological effects alkaloids have on humans are evident [bib_ref] Goitrogenic food and prevalence of goitre in Sri Lanka, Fernando [/bib_ref]. High tropane alkaloid consumption causes fast pulse, paralysis, and death in the worst-case scenario [bib_ref] Review on: Antinutritional factors in vegetable crops, Sinha [/bib_ref]. Negri et al. [bib_ref] The case of tryptamine and serotonin in plants: A mysterious precursor for..., Negri [/bib_ref] compared a large number of vegetables, tryptamine content in lettuce is high at 24.5 ± 2.4 µg/g −1 of dry weight. Ingestion of large amounts of tryptamine alkaloids causes final death. Moreover, Besharat et al. [bib_ref] Wild lettuce (Lactuca virosa) toxicity, Besharat [/bib_ref] listed some side effects and toxicity resulting from an overdose of lettuce, including mydriasis and photophobia, dizziness, diaphoresis, auditory hallucination, and cardiovascular and respiratory difficulties caused by dysrhythmia; however, with the herb-use of lettuce, it acts as a spasmolytic and a sedative.
## Future perspectives
The extraction of bioactive compounds is one of the new trends observed in plants. Some previous studies have shown that lettuce is one of the fascinating and rich sources of antioxidant polyphenols that support human health [bib_ref] Lettuce and chicory byproducts as a source of antioxidant phenolic extracts, Llorach [/bib_ref]. Some studies have focused on the extraction of lettuce by various methods. Fresh-cut processing can facilitate the extraction of polyphenols in lettuce tissue, which is a response to cell injury such as leaf-cutting or shredding. As for the procedure used previously, lettuce needs to be homogenized by extractive solution, maintained at a temperature of 4-50 - C for 48 h. Regardless of the time consumed, these procedures cannot provide food-grade polyphenol extracts. Therefore, Chemat et al. [bib_ref] Review of green food processing techniques. Preservation, transformation, and extraction, Chemat [/bib_ref] reported some new methods using "green" extraction technologies by non-toxic and GRAS solvents with ultrasound assistance that can effectively extract food-grade polyphenol from lettuce waste. The polyphenols and gluconate extraction from lettuce waste can be suitable for functional foods and nutraceuticals.
Another new trend is to make functional flour. Apart from extracting polyphenols from lettuce, it can be used in some other ways. For example, lettuce with high phenol and fiber content is air-dried and ground to add to flour, producing functional baked products. Traditionally, it is common for a bakery to use whole meal flour because of the high antioxidants and fiber contents in bran and aleurone [bib_ref] Current trends in the enhancement of antioxidant activity of wheat bread by..., Dziki [/bib_ref]. The same objectives can be reached by flours made from vegetables or their waste, which contains high contents of these compounds. Lettuce flour can be considered an ideal ingredient for bread to increase functionality. Plazzotta et al. [bib_ref] Exploitation of lettuce waste flour to increase bread functionality: Effect on physical,..., Plazzotta [/bib_ref] verified that lettuce flour greatly improved the phenolic content of bread and increased the antioxidant activity by 200%. Lettuce flour added to bread can gradually reduce the leavening ability of dough and increase the moisture and hardness of bread. However, the utilization of lettuce flour will influence bread processing and sensory aspects of the bread, while bread containing 170 to 575 g/kg of lettuce flour is more acceptable than commercial whole meal bread.
# Conclusions
In conclusion, lettuce has high nutritional value due to its contribution to dietary fiber, several important dietary minerals, various vitamins, and bioactive compounds such as carotenoids, phenolic compounds, chlorophyll, and even sesquiterpene lactones. The bioavailability and metabolism of phytonutrients can better understand related health benefits and the ultimate delivery of these compounds. Especially due to its antioxidant compounds, lettuce can provide some potential health benefits in cardio-protective, anticancer, anti-diabetic, and anti-aging. Due to these potentials, both in vitro and in vivo evidence verify the preclinical and clinical application of lettuce extracts, which can be further investigated. Meanwhile, the purified phytochemicals extracted from lettuce may reveal the possibility of producing efficient functional foods. Moreover, further studies can focus on the growing conditions and development strategies to increase the nutritional value of lettuce and widely apply them in other areas.
[fig] Figure 1: Stages involved in selecting published data for inclusions in the current study are depicted in a flow chart; n: number of literature reports. [/fig]
[fig] Figure 2: The frequent growth types of lettuce: Crisphead, Butterhead, Looseleaf, and Romaine. [/fig]
[fig] Figure 3: Schematic layout of the digestive pathway of phenolic compounds[81]. [/fig]
[fig] Antioxidants 2022 ,: 11, x FOR PEER REVIEW 11 of 24 [/fig]
[fig] Figure 4: Parameters affecting the bioavailability of bioactive compounds[89]. [/fig]
[fig] Figure 5: Health-promoting benefits of lettuce. [/fig]
[fig] Author: Contributions: Conceptualization, M.S., J.G., H.W. and H.A.R.S.; supervision and funding acquisition, H.A.R.S.; methodology and writing-original draft preparation, M.S., J.G. and H.W.; data collection, drawing figures, and critically revised the article, A.R., T.B.E., Z.K., S.M., A.S.M.A., F.A.A., Y.S.A.-A., O.B., M.T. and H.A.R.S. All authors have read and agreed to the published version of the manuscript. Funding: Hafiz Suleria is the recipient of an "Australian Research Council-Discovery Early Career Award" (ARC-DECRA-DE220100055) funded by the Australian Government. This research was funded by the University of Melbourne under the "McKenzie Fellowship Scheme" (Grant No. UoM-18/21), the "Faculty Research Initiative Funds (Grant No. UoM-19/20)" and "Collaborative Research Development Grant (Grant No. UoM-21/23)" funded by the Faculty of Veterinary and Agricultural Sciences, the University of Melbourne, Australia. [/fig]
[table] Table 1: Chemical structure of main phytochemicals identified in lettuce. [/table]
|
Onset of efficacy and tolerability following the initiation dosing of long-acting paliperidone palmitate: post-hoc analyses of a randomized, double-blind clinical trial
Background: Paliperidone palmitate is a long-acting injectable atypical antipsychotic for the acute and maintenance treatment of adults with schizophrenia. The recommended initiation dosing regimen is 234 mg on Day 1 and 156 mg on Day 8 via intramuscular (deltoid) injection; followed by 39 to 234 mg once-monthly thereafter (deltoid or gluteal). These post-hoc analyses addressed two commonly encountered clinical issues regarding the initiation dosing: the time to onset of efficacy and the associated tolerability. Methods: In a 13-week double-blind trial, 652 subjects with schizophrenia were randomized to paliperidone palmitate 39, 156, or 234 mg (corresponding to 25, 100, or 150 mg equivalents of paliperidone, respectively) or placebo (NCT#00590577). Subjects randomized to paliperidone palmitate received 234 mg on Day 1, followed by their randomized fixed dose on Day 8, and monthly thereafter, with no oral antipsychotic supplementation. The onset of efficacy was defined as the first timepoint where the paliperidone palmitate group showed significant improvement in the Positive and Negative Syndrome Scale (PANSS) score compared to placebo (Analysis of Covariance [ANCOVA] models and Last Observation Carried Forward [LOCF] methodology without adjusting for multiplicity) using data from the Days 4, 8, 22, and 36 assessments. Adverse event (AE) rates and relative risks (RR) with 95% confidence intervals (CI) versus placebo were determined.Results: Paliperidone palmitate 234 mg on Day 1 was associated with greater improvement than placebo on Least Squares (LS) mean PANSS total score at Day 8 (p = 0.037). After the Day 8 injection of 156 mg, there was continued PANSS improvement at Day 22 (p ≤ 0.007 vs. placebo) and Day 36 (p < 0.001). Taken together with results in the 39 mg and 234 mg Day 8 arms, these findings suggest a trend towards a dose-dependent response. During Days 1 to 7, AEs reported in ≥2% of paliperidone palmitate subjects (234 mg) and a greater proportion of paliperidone palmitate than placebo subjects were: agitation (3.2% vs. 1.3%; RR 2.52 [95% CI 0.583, 10.904]), headache (4.0% vs. 3.8%; RR 1.06 [95% CI 0.433, 2.619]), and injection site pain (6.7% vs. 3.8%; RR 1.79 [95% CI 0.764, 4.208]). Days 8 to 36 AEs meeting the same criteria in the 156 mg Day 8 arm were: anxiety (3.1% vs. 2.5%; RR 1.24 [95% CI 0.340, 4.542]), psychotic disorder (2.5% vs. 1.3%; RR 1.99 [95% CI 0.369, 10.699]), dizziness (2.5% vs. 1.3%; RR 1.99 [95% CI 0.369, 10.699]), and injection site pain (2.5% vs. 1.3%; RR 1.99 [95% CI 0.369, 10.699]). Corresponding Days 8 to 36 AEs in the 39 mg Day 8 group were: agitation (4.5% vs. 4.4%; RR 1.03 [95% CI 0.371, 2.874]), anxiety (3.9% vs. 2.5%; RR 1.55 [95% CI 0.446, 5.381]), and psychotic disorder (2.6% vs. 1.3%; RR 2.07 [95% CI 0.384, 11.110]) while in the 234 mg Day 8 group it was anxiety (3.1% vs. 2.5%, RR 1.25 [95% CI 0.342, 4.570]).
Conclusions: Significantly greater symptom improvement was observed by Day 8 with paliperidone palmitate (234 mg on Day 1) compared to placebo; this effect was maintained after the 156 mg Day 8 injection, with a trend towards a dose-dependent response. No unexpected tolerability findings were noted in the first week or month after the initiation dosing. Trial registration: ClinicalTrials.gov: NCT#00590577
# Background
For individuals with schizophrenia-whether a first episode or a relapse-the rapid and robust control of symptoms at well tolerated medication dosages are primary goals to reduce emotional distress, minimize disruption to the patient's life, and reduce the risk of dangerous behaviors . Evidence also suggests that the prompt improvement in symptoms may improve long-term outcomes . To realize these benefits, the effectiveness of a therapeutic agent measured as an improvement in symptoms, acceptable tolerability, and an early onset of effect are important considerations in the choice of an antipsychotic agent.
Rapid symptom control is a strong predictor of treatment success and may be a valuable indicator of longterm symptom control as well as a low rate of relapse and rehospitalization, which may contribute to reducing healthcare costs [bib_ref] Relapse and rehospitalisation rates in patients with schizophrenia: effects of second generation..., Csernansky [/bib_ref] [bib_ref] The cost of relapse and the predictors of relapse in the treatment..., Ascher-Svanum [/bib_ref]. While some data suggest that most symptom amelioration occurs within the first 2 weeks after introduction of antipsychotic treatment, some patients require a longer time to respond. In a recently published investigation of patterns of response (defined as ≥30% reduction in Positive and Negative Symptom Score [PANSS] from baseline) with an atypical antipsychotic, approximately 36% of patients responded within 2 weeks, while an additional 20% responded by week 6 [bib_ref] Onset and persistence of antipsychotic response in patients with schizophrenia, Glick [/bib_ref]. The American Psychiatric Association guidelines recommend a 2-to 4-week therapeutic trial prior to changing a treatment regimen .
Paliperidone palmitate is a long-acting injectable formulation of paliperidone, which is also formulated for daily oral administration as paliperidone extended-release (ER). Paliperidone palmitate is the palmitate ester of paliperidone. The dosing of paliperidone palmitate may be expressed in terms of milligrams (mg) of paliperidone palmitate or in terms of milligram equivalents (mg eq) of the pharmacologically active fraction, paliperidone. Paliperidone palmitate expressed as 39, 156, and 234 mg is equivalent to 25, 100, and 150 mg eq, respectively, of the active fraction paliperidone. The pharmacokinetic properties of paliperidone palmitate allow for once-monthly injections following two initiation doses given 1 week apart [bib_ref] Paliperidone Palmitate) Package Insert, Invega ® Sustenna ® [/bib_ref] [bib_ref] Population pharmacokinetics of intramuscular paliperidone palmitate in patients with schizophrenia, Samtani [/bib_ref] [bib_ref] Practical guidance for dosing and switching paliperidone palmitate treatment in patients with..., Gopal [/bib_ref]. Pharmacokinetic data indicate higher median peak concentrations following paliperidone palmitate administration into the deltoid rather than the gluteal muscle, with similar area-under-the-curve (AUC) values [bib_ref] Population pharmacokinetics of intramuscular paliperidone palmitate in patients with schizophrenia, Samtani [/bib_ref]. Given this, it is recommended that administration of initiation doses of paliperidone palmitate be in the deltoid muscle with maintenance dose administration being interchangeable between deltoid and gluteal administration [bib_ref] Population pharmacokinetics of intramuscular paliperidone palmitate in patients with schizophrenia, Samtani [/bib_ref].
Paliperidone palmitate has been studied in several randomized, double-blind controlled trials using various dosing regimens [bib_ref] Efficacy and safety of paliperidone palmitate in adult patients with acutely symptomatic..., Gopal [/bib_ref] [bib_ref] Paliperidone palmitate maintenance treatment in delaying the time-torelapse in patients with schizophrenia:..., Hough [/bib_ref] [bib_ref] Paliperidone palmitate, a potential long-acting treatment for patients with schizophrenia. Results of..., Kramer [/bib_ref] [bib_ref] Safety and tolerability of deltoid and gluteal injections of paliperidone palmitate in..., Hough [/bib_ref] [bib_ref] A randomized, placebo-controlled study to assess the efficacy and safety of three..., Pandina [/bib_ref]. A recently completed phase 3 trial was the first placebo-controlled study to assess paliperidone palmitate administered at the recommended Day 1 dose of 234 mg by deltoid injection. Subjects then received 39, 156, or 234 mg on Day 8 and monthly thereafter (deltoid or gluteal). In this study, paliperidone palmitate, without oral antipsychotic supplementation, was associated with significant improvements in symptomatology with no unexpected tolerability findings in adults with symptomatic schizophrenia, at all doses tested [bib_ref] A randomized, placebo-controlled study to assess the efficacy and safety of three..., Pandina [/bib_ref].
An early, well-tolerated response to antipsychotic treatment has important down-stream implications for long-term symptom control, treatment adherence, healthcare costs, and, consequently, clinical decisionmaking. These post-hoc analyses of data from the published trial [bib_ref] A randomized, placebo-controlled study to assess the efficacy and safety of three..., Pandina [/bib_ref] was designed to address two commonly encountered clinical questions associated with the initiation regimen of paliperidone palmitate: 1) when is the onset of efficacy and; 2) how well is this initiation dose tolerated. This report focuses on the subjects who received 234 mg on Day 1 (deltoid) followed by 156 mg on Day 8 (deltoid or gluteal). Data are also presented for those who received 234 mg on Day 1 followed by 39 or 234 mg on Day 8.
# Methods
Design A 13-week double-blind, randomized, placebo-controlled phase 3 trial (NCT#00590577) was conducted from March 2007 to March 2008 at 72 centers in 8 countries in North America, Europe, and Asia. Subjects with schizophrenia and a PANSS total score of 70 to 120 (inclusive) at screening and 60 to 120 (inclusive) at doubleblind baseline were eligible for study enrollment. Key exclusion criteria included primary DSM-IV Axis I diagnosis other than schizophrenia, DSM-IV diagnosis of active substance dependence within 3 months before screening, history of treatment resistance (failure to respond to 2 adequate courses of different antipsychotic medications with a minimum of 4 weeks duration at the patient's maximum tolerated dose), history of neuroleptic malignant syndrome, a relevant history of any significant or unstable systemic disease, morbid obesity (body mass index ≥40 kg/m 2 ), and circumstances that could increase the risk of the occurrence of Torsade de Pointes or sudden death. Further details of the study design are reported by Pandina et al. [bib_ref] A randomized, placebo-controlled study to assess the efficacy and safety of three..., Pandina [/bib_ref].
## Study medications
The study consisted of a screening period of up to 7 days to washout disallowed psychotropic medications followed by a 13-week double-blind treatment period. On Day 1, eligible patients were randomly assigned (1:1:1:1) to fixed doses of paliperidone palmitate 39, 156, or 234 mg (equivalent to 25, 100, or 150 mg eq of the active fraction paliperidone), or placebo, based on a computer-generated randomization schedule balanced by using permuted blocks of treatments and stratified by center. On Day 1, all patients received a deltoid injection of paliperidone palmitate 234 mg or matching placebo. On Day 8, and then on Days 36 and 64, patients received their assigned treatment per the randomization schedule, injected in the deltoid or the gluteal muscle at the discretion of the investigator. Patients were hospitalized from Day 1 (first injection) until at least after the second injection of study drug on Day 8. Antipsychotics except study drug were prohibited during the doubleblind treatment period. Prior antiparkinsonian medications were to be washed out prior to baseline, but were allowed during the study at the discretion of the investigator if extrapyramidal symptoms [EPS] emerged or worsened. Oral benzodiazepines were allowed for agitation, anxiety, or sleep difficulties at the permitted maximum daily doses.
## Assessments
Efficacy was assessed using PANSS total scores at Days 4, 8, 22, 36, 64, and 92 (or study endpoint). Tolerability assessments included treatment-emergent adverse events reports and adverse-event related study discontinuations.
## Analysis sets and statistical evaluations
Onset of efficacy and tolerability analyses were performed on the intent-to-treat (ITT) analysis set, which included all randomized patients who received at least one dose of double-blind study medication and had both the baseline and at least one post baseline efficacy assessment. Changes from baseline in PANSS total scores were estimated by Least Squares (LS) means and compared between groups using an Analysis of Covariance (ANCOVA) model and Last Observation Carried Forward (LOCF) methodology, without adjustment for multiplicity. Results on Days 4 and 8 were pooled for the paliperidone palmitate dose arms (all received 234 mg of paliperidone palmitate on Day 1). At Days 22 and 36, data were analyzed for each paliperidone palmitate dose group (corresponding to 39, 156, or 234 mg that was administered at Day 8). The effect size (Cohen's d) for paliperidone palmitate relative to placebo in PANSS change from baseline was calculated using Cohen's d with LS means and mean (standard error [SE]) from ANCOVA model for treatment comparison. Onset of efficacy was defined as the first timepoint at which the change from baseline in the PANSS total score in the paliperidone palmitate group was significant compared to placebo (at the 2-sided nominal 5% level of significance). Responder rates were defined as the proportion of subjects with a ≥30% reduction from baseline in PANSS total score. Pairwise comparisons were performed using Cochran-Mantel-Haenszel tests controlling for country.
Adverse events that occurred in ≥2% of paliperidone palmitate subjects were summarized (by dose arm) and compared to placebo, with determination of the relative risk (RR) and 95% confidence interval (95% CI) associated with paliperidone palmitate. RRs were considered statistically significant when 95% CIs did not include 1. No adjustments were made for multiplicity. Tolerability associated with the initiation dosing regimen was also assessed through analyses of treatment discontinuation and adverse event reports (including extrapyramidal-, metabolic-and potentially prolactin-related events) during post-injection time periods of Days 1 to 7 and 8 to 36. Extrapyramidal events included akathisia, tremor, dyskinesia, extrapyramidal disorder, movement disorder, or parkinsonism. Metabolic events included metabolism or nutritional disorders such as increased/decreased appetite, increased/decreased weight, dyslipidemia(s), or malnutrition. Potentially prolactin-related events included reproductive system events, breast disorders, and ejaculation disorders.
# Results
## Patient disposition and characteristics
Of 855 subjects screened, 652 (76%) were randomized to either paliperidone palmitate (n = 488) or placebo (n = 164); 476 and 160, respectively, were in the ITT analysis set [fig_ref] Figure 1: Subject Randomization [/fig_ref]. All ITT subjects randomized to paliperidone palmitate received a Day 1 dose of 234 mg; 161 were randomized to the 156 mg Day 8 treatment arm. One-hundred sixty (160) and 155 subjects were randomized to the 234 and 39 mg Day 8 treatment arms, respectively. Administration of the Day 1 doses were primarily (99%) in the deltoid muscle (3 paliperidone palmitate and 1 placebo subject received the injection in the gluteus). The Day 8 dose was administered in the gluteus in 48% to 53% of those in the paliperidone palmitate treatment arms and in 58% of those in the placebo treatment arm.
Baseline demographics and disease characteristics in the ITT analysis set were similar across treatment arms with a mean age of 39 years, 67% male, and 54% Caucasian [bib_ref] A randomized, placebo-controlled study to assess the efficacy and safety of three..., Pandina [/bib_ref]. Mean (Standard Deviation [SD]) PANSS total score scores were 86.8 (10.31) in the placebo group and 86.9 (11.99), 86.2 (10.77), and 88.4 (11.70) in the paliperidone palmitate 39, 156, and 234 mg Day 8 arms, respectively. Atypical antipsychotics were commonly used (70% of subjects) prior to enrollment, with oral risperidone use reported by 34% to 41% of subjects across the arms. Prior to baseline, approximately 30% of subjects in each treatment arm were using an anti-EPS medication (24% placebo and 35%, 30% and 33% in the paliperidone palmitate 39, 156, and 234 mg Day 8 arms, respectively) and approximately 60% were using a benzodiazepine (65%, 67%, 58%, and 59%, respectively).
## Effects on panss total scores at day 8 timepoint
Paliperidone palmitate 234 mg administered on Day 1 was associated with a significantly greater improvement than placebo on mean PANSS total score at the [fig_ref] Table 1: Effect size for PANSS total change score [/fig_ref]. These results suggest a dose-related effect.
The responder rates (≥30% reduction from baseline in PANSS total score) were significantly higher with paliperidone palmitate (all dose groups) than with placebo by the Day 36 timepoint [fig_ref] Figure 3: Responders [/fig_ref]. In the group receiving the 156 mg Day 8 dose, the responder rate was 36.6% compared to 20.6% with placebo (p = 0.002) [fig_ref] Figure 3: Responders [/fig_ref].
## Discontinuations and benzodiazepine use days 1 to 7
During Days 1 to 7, the percentage of patients who discontinued study participation was 2.9% in those who received paliperidone palmitate (234 mg Day 1) and 4.4% in the placebo group [fig_ref] Table 2: Study Discontinuations and Benzodiazepine Use, by Treatment Group and Time Period [/fig_ref]. During the week following the first injection, the most common reason for discontinuation in both groups was withdrawal of consent (1.9% in placebo and 1.1% in paliperidone palmitate).
Withdrawal due to adverse events was low in both groups (0.8% [n = 4] in paliperidone palmitate and 1.3% [n = 2] in placebo). Events that resulted in discontinuation in the paliperidone palmitate group were: gastroesophageal reflux, pain in extremity, suicidal ideation, and toothache (1 subject); injection site pain (1 subject); insomnia, schizophrenia, and tremor (1 subject); and psychotic disorder (1 subject). Adverse events leading to discontinuation in the two placebo-treated subjects were schizophrenia (1 subject) and schizophrenia, increased aspartate aminotransferase and increased blood lactate dehydrogenase (1 subject).
Approximately half the patients in both groups reported benzodiazepine use [fig_ref] Table 2: Study Discontinuations and Benzodiazepine Use, by Treatment Group and Time Period [/fig_ref].
## Days 8 to 36
In the month following the Day 8 injection (Days 8 to 36), discontinuation rates were 32.5% in the placebo group and 23.6% in the paliperidone palmitate 156 mg Day 8 group [fig_ref] Table 2: Study Discontinuations and Benzodiazepine Use, by Treatment Group and Time Period [/fig_ref]. Discontinuation due to adverse events was 3.1% (n = 5 in each group) in the placebo group as well as the paliperidone palmitate 156 mg Day 8 treatment arm. Adverse events leading to discontinuation in the 5 placebo-treated subjects were: increased alanine aminotransferase and increased aspartate aminotransferase (1 subject); nausea and vomiting (1 subject); delusional disorder-persecutory type, musculoskeletal stiffness, and tremor (1 subject); anxiety and schizophrenia (1 subject); and insomnia and schizophrenia (1 subject). Events leading to discontinuation in the paliperidone palmitate 156 mg Day 8 group were: schizophrenia (2 subjects); schizophrenia-paranoid type (1 subject); insomnia, otitis media-chronic, psychotic disorder, and toothache (1 subject); and injection site swelling (1 subject).
No subject discontinued due to adverse events in the paliperidone palmitate 39 mg Day 8 arm; 5 discontinued in the 234 mg Day 8 treatment arm. Adverse events in the latter group were psychiatric disorder and toothache (1 subject); anxiety (1 subject); agitation, aspartate aminotransferase increase, toothache, and white blood cell count decrease (1 subject); agitation, insomnia, and schizophrenia (1 subject); and blood amylase increased and cerebrovascular accident (1 subject).
Benzodiazepine use during this period was reported by 43.8% of the placebo arm and 33.5% of the paliperidone palmitate 156 mg Day 8 arm. Rates were 42.6% in the 39 mg Day 8 arm and 40.0% in the 234 mg Day 8 arm [fig_ref] Table 2: Study Discontinuations and Benzodiazepine Use, by Treatment Group and Time Period [/fig_ref].
## Adverse events days 1 to 7
The overall rate of adverse events during the week following the paliperidone palmitate 234 mg Day 1 [fig_ref] Figure 4: Adverse Events in ≥2% of Paliperidone Palmitate and in a Higher Percentage... [/fig_ref]. The RRs were not statistically significant as determined by 95% CIs.
The incidence of any EPS-related event reports during Days 1 to 7 was 3.6% (17/476) in the paliperidone palmitate 234 mg group and 3.1% (5/160) in the placebo group. The use of anti-EPS medications was 5.5% (26/476) and 7.5% (12/160), respectively.
Metabolic and potentially prolactin-related events were reported in < 2% of those administered paliperidone palmitate 234 mg or placebo on Day 1.
## Days 8 to 36
The adverse event rate during the month following the A total of 39 subjects reported adverse events that were rated as serious during Days 8 to 36: 29 paliperidone palmitate subjects (6.1%) and 10 placebo subjects (6.3%). The serious adverse events reported in the placebo arm were: acute psychosis (1 subject); persecutory type delusional disorder (1 subject); psychotic disorder (2 subjects); schizophrenia (5 subjects); electrocardiogram change (1 subject); and non-cardiac chest pain (1 subject). Serious adverse events reported in those receiving the recommended Day 8 dose of 156 mg were: anxiety (1 subject); psychotic disorder (4 subjects); schizophrenia, paranoid type (1 subject); and schizophrenia (4 subjects). Serious adverse events reported in the 39 mg Day 8 arm during this period were: agitation (1 subject); depression (1 subject); auditory hallucination (1 subject); insomnia (1 subject); psychotic disorder (2 subjects); schizophrenia (5 subjects); suicidal ideation (3 subjects); diverticulitis (1 subject); and syncope (1 subject). Those reported in the 234 mg Day 8 arm were: anxiety (1 subject); depression (1 subject); psychotic disorder (1 subject); schizophrenia (4 subjects); and cerebrovascular accident (1 subject). Note that a given patient may have reported more than one serious adverse event.
In the paliperidone palmitate group receiving the recommended initiation dosing (234 mg Day 1/156 mg Day 8), the adverse events reported in ≥2% of this group and in a greater percentage of paliperidone palmitate than placebo subjects were anxiety [fig_ref] Figure 4: Adverse Events in ≥2% of Paliperidone Palmitate and in a Higher Percentage... [/fig_ref]. These RRs were not statistically significant, as determined by 95% CIs.
In the paliperidone palmitate 39 mg Day 8 arm [fig_ref] Figure 4: Adverse Events in ≥2% of Paliperidone Palmitate and in a Higher Percentage... [/fig_ref] , the adverse events reported in ≥2% of this group and in a greater percentage of paliperidone palmitate than placebo subjects were agitation (4.5% vs. [bib_ref] The cost of relapse and the predictors of relapse in the treatment..., Ascher-Svanum [/bib_ref] The incidence of any EPS-related events during Days 8 to 36 were 3.7% (6/161) in the paliperidone palmitate 156 mg Day 8 arm and 4.4% (7/160) in the placebo arm (RR 0.8518; 95% CI 0.293, 2.48). Specific extrapyramidal symptoms were reported in < 2% of subjects in any treatment arm, with the exception of akathisia, reported in 2.5% (4/160) of the paliperidone palmitate 234 mg Day 8 arm and 3.1% (5/160) of the placebo arm (RR 0.800, 95% CI 0.219, 2.925).
The use of anti-EPS medications during Days 8 to 36 was 9.0% (14/155), 9.9% (16/161), and 6.3% (10/160) in those receiving 39, 156, and 234 mg on Day 8, respectively. The rate was 6.3% (10/160) in the placebo group.
Metabolic events and potentially prolactin-related events were reported in < 2% of subjects in each paliperidone palmitate arm and the placebo arm.
# Discussion
Two common clinical questions regarding the initiation dosing of paliperidone palmitate, specifically the time to onset of efficacy and the associated tolerability, were addressed in these post-hoc analyses of a large, doubleblind, placebo-controlled trial. The question of when clinicians and patients can anticipate an improvement in symptoms is integral to clinical decision-making, particularly when managing a symptomatic patient with schizophrenia and planning a treatment strategy. While some clinicians may prefer to initiate paliperidone palmitate at a lower than the recommended initiation regimen due to tolerability concerns, previously published data suggest this may result in sub-therapeutic plasma levels and poor longer-term clinical response in some patients [bib_ref] Efficacy and safety of paliperidone palmitate in adult patients with acutely symptomatic..., Gopal [/bib_ref]. Thus, data were presented in this report for the early days and weeks following the initiation regimen to examine the efficacy and tolerability of the recommended initiation doses for paliperidone palmitate.
Findings showed significantly greater symptom improvement by Day 8 with paliperidone palmitate (234 mg on Day 1) compared to placebo, without oral antipsychotic supplementation, with this effect maintained after the 156 mg injections through Day 64, as well as at study endpoint [bib_ref] A randomized, placebo-controlled study to assess the efficacy and safety of three..., Pandina [/bib_ref]. When looking across the treatment arms, a trend towards a dose-dependent response was observed during the first 36 days of this study, again consistent with the data reported through study endpoint [bib_ref] A randomized, placebo-controlled study to assess the efficacy and safety of three..., Pandina [/bib_ref]. Also of note, the effect size vs. placebo for PANSS data illustrate an increasing improvement over time with the 156 mg dose (0.30, 0.43, 0.42, and 0.49 at Days 22, 36, 64, and endpoint, respectively), and the 234 mg dose (0.41, 0.40, 0.48, and 0.55, respectively). The 39 mg arm had lower and relatively constant effect sizes from Day 22 through endpoint (0.27, 0.28, 0.26, and 0.28, respectively). The early reduction in mean PANSS score shown here is supported by that from a non-inferiority trial [bib_ref] A double-blind study of paliperidone palmitate and risperidone long-acting injectable in adults..., Pandina [/bib_ref] , where PANSS improvement was similar at the Day 4 timepoint for subjects receiving an initial injection of paliperidone palmitate at 234 mg compared to oral risperidone given at 1 to 6 mg per day.
The clinical improvement observed with the initiation doses of paliperidone palmitate in this study is supported by the attainment of therapeutic serum concentrations of paliperidone reported in clinical and pharmacokinetic modeling analyses [bib_ref] Population pharmacokinetics of intramuscular paliperidone palmitate in patients with schizophrenia, Samtani [/bib_ref] [bib_ref] Practical guidance for dosing and switching paliperidone palmitate treatment in patients with..., Gopal [/bib_ref] [bib_ref] A randomized, placebo-controlled study to assess the efficacy and safety of three..., Pandina [/bib_ref]. Following a single intramuscular dose, the release of paliperidone into the systemic circulation occurs as early as Day 1, with a gradual rise to reach maximum plasma concentrations at a median of 13 days [bib_ref] Paliperidone Palmitate) Package Insert, Invega ® Sustenna ® [/bib_ref]. The two initial doses of paliperidone palmitate (234 mg Day 1/156 mg Day 8) into the deltoid help attain therapeutic concentrations rapidly, with the AUC profiles being dose proportional over the 39 to 234 mg dose range [bib_ref] Paliperidone Palmitate) Package Insert, Invega ® Sustenna ® [/bib_ref]. In studies that used lower doses of paliperidone palmitate and initiation dose administration into the gluteal muscle, an onset of efficacy by Day 8 was not consistently observed [bib_ref] Efficacy and safety of paliperidone palmitate in adult patients with acutely symptomatic..., Gopal [/bib_ref].
With respect to tolerability concerns with the recommended paliperidone palmitate initiation dosing, this study did not reveal unexpected adverse events or high rates of specific adverse events in the first week or subsequent month after the initiation injections. In addition, overall treatment discontinuations and discontinuations due to adverse events were generally low during this time. However, these are data from a single clinical study. Further, the relative risk analysis requires comment. This analysis was undertaken with the intent of providing a useful way identifying adverse events that may be more likely to occur with active treatment as compared with placebo. Findings were that events such as agitation, anxiety, dizziness, headache, injection site pain, and psychotic disorder had a relative risk ranging from approximately 1.1 to 2.5 during the first month of treatment. Although these relative risks were not statistically significant, as determined by the 95% CIs, they may be clinically relevant providing useful information for clinicians to consider when initiating treatment with paliperidone palmitate. Additionally, it must be noted that the analysis of this relatively small database is not sufficient to identify rare treatment-related events.
Extrapyramidal symptoms such as parkinsonism, akathisia, dyskinesia, and dystonia are also an area of concern with respect to the tolerability of an antipsychotic regimen. Substantial literature supports that the incidence of these events as well as the time of onset differ substantially [bib_ref] Neurological complications of psychiatric drugs: clinical features and management, Haddad [/bib_ref] [bib_ref] Extrapyramidal symptoms with atypical antipsychotics: incidence, prevention and management, Pierre [/bib_ref]. In terms of onset, dystonic reactions and akathisia generally occur within the first few hours to days of treatment while parkinsonism occurs within the first few weeks and tardive dyskinesia or dystonia generally appearing after months or years of treatment [bib_ref] Neurological complications of psychiatric drugs: clinical features and management, Haddad [/bib_ref] [bib_ref] Extrapyramidal symptoms with atypical antipsychotics: incidence, prevention and management, Pierre [/bib_ref]. Broadly speaking the risk for extrapyramidal symptoms is generally considered to be lower with atypical compared with typical antipsychoticshowever, the risk for these events varies among the agents in each class. Within the atypical class of agents the risk for extrapyramidal events is often dose-related [bib_ref] Neurological complications of psychiatric drugs: clinical features and management, Haddad [/bib_ref]. In this analysis, the incidence of extrapyramidal symptoms was less than 2%, with akathisia being the only extrapyramidal symptom having an incidence of > 2% (2.5%) during Days 8 to 36 at the highest dose of paliperidone palmitate (234 mg).
One must also consider that this study was not designed to assess onset of efficacy or the tolerability associated with the initiation regimen. Therefore, these findings are somewhat limited by the timepoints that were assessed (i.e., [bib_ref] Practical guidance for dosing and switching paliperidone palmitate treatment in patients with..., Gopal [/bib_ref] and data collected at these visits. For example, more timepoints would be valuable to assess onset. It should also be noted that while commonly used criteria were applied to define onset as well as response, other criteria could result in different outcomes. Further, these criteria were applied to a population of subjects enrolled in a large doubleblind clinical trial and these findings may not generalize to patient populations with different characteristics. Also, the results presented here are population-based data that do not fully address the heterogeneity that is associated with individual treatment response. That is, mean responses from a population address probabilities of clinical response but do not predict the response for a particular patient. Finally, it should be pointed out that there was a substantial placebo response observed in this trial. This is not uncommon in studies of patients with schizophrenia and, nevertheless, the effect size data for paliperidone palmitate compared to placebo suggests a clinically meaningful dose-and time-dependent treatment effect in this population.
# Conclusions
In this study, the initiation regimen of paliperidone palmitate of 234 mg on Day 1 and 156 mg on Day 8 was associated with a significant improvement in symptoms by Day 8 that continued at the subsequent Day 22 and Day 36 timepoints among subjects with symptomatic schizophrenia. There was a trend towards a dose-dependent response observed across the dosage groups. There were no unusual or unexpected tolerability findings noted during either the first week or month following paliperidone palmitate treatment initiation.
Endnote a. The dosing used in this clinical study aligns with the recommended initiation regimen of paliperidone palmitate (i.e., 234 mg on Day 1, 156 mg on Day 8); however, the dosage regimen recommends that these injections are both given in the deltoid muscle, with gluteal muscle injections being an option after the Day 8 dose [bib_ref] Paliperidone Palmitate) Package Insert, Invega ® Sustenna ® [/bib_ref].
[fig] Figure 1: Subject Randomization. Of 652 subjects enrolled in the double-blind treatment period, 488 were randomized (1:1:1:1) to paliperidone palmitate (fixed dose of 39, 156, or 234 mg) and 164 to placebo. All those randomized to the fixed doses of paliperidone palmitate received 234 mg as the first initiation dose on Day 1, followed by administration of their fixed dose on Days 8, 36, and 64. [/fig]
[fig] Figure 3: Responders: ≥30% Improvement from Baseline in PANSS Total Scores. The responder rates were significantly higher with paliperidone palmitate (all dose groups) than with placebo by the Day 36 timepoint. [/fig]
[fig] Figure 4: Adverse Events in ≥2% of Paliperidone Palmitate and in a Higher Percentage of Paliperidone Palmitate than Placebo Subjects. Adverse events meeting these criteria during Days 1 to 7 are shown in Panel A for subjects received paliperidone palmitate 234 mg Day 1 (rates and relative risks versus placebo with 95% CIs); none were statistically significant as determined by 95% CIs. Adverse events that met these criteria during Days 8 to 36 are shown in Panel B for the paliperidone palmitate 156 mg Day 8 group, Panel C for the 39 mg Day 8 group, and Panel D for the 234 mg Day 8 group. None were statistically significant as determined by 95% CIs. [/fig]
[table] Table 1: Effect size for PANSS total change score: Paliperidone palmitate vs. placebo (95% CI) initiation dose was similar to that seen with placebo (38.0% [181/476] vs. 43.1% [69/160], respectively).With the exception of one report of schizophrenia in a placebo-treated subject, no other adverse events were rated as serious.Adverse events reported in ≥2% of paliperidone palmitate treated subjects and in a greater proportion of paliperidone palmitate than placebo-treated subjects were agitation (3.2% vs. 1.2%; RR 2.52 [95% CI 0. [/table]
[table] Table 2: Study Discontinuations and Benzodiazepine Use, by Treatment Group and Time Period [/table]
|
Nuclear envelope morphology constrains diffusion and promotes asymmetric protein segregation in closed mitosis
# Introduction
Asymmetric cell division is a major mechanism for the generation of cell diversity in eukaryotes, particularly through unequal segregation of cell fate determinants between daughter cells. Asymmetric cell division is a two-step process. First, the cell polarizes its cortex and accumulates fate determinants to one pole of the cell. Second, the mitotic spindle aligns with the cell polarity axis, defining the division axis of the cell. This places the polarized material on one side of the future division plane and, therefore, in only one of the two daughters. Subsequently, the asymmetrically segregated cell fate determinants specify the cell's identity and drive differentiation from its sister.
Several transcription factors, such as Prospero in Drosophila melanogaster, Pie-1 in Caenorhabditis elegans, and Ash1 in budding yeast, also segregate asymmetrically despite their nuclear localization [bib_ref] The prospero transcription factor is asymmetrically localized to the cell cortex during..., Spana [/bib_ref] [bib_ref] Asymmetric accumulation of Ash1p in postanaphase nuclei depends on a myosin and..., Bobola [/bib_ref] [bib_ref] The PIE-1 protein and germline specification in C. elegans embryos, Mello [/bib_ref] [bib_ref] Identification of asymmetrically localized determinant, Ash1p, required for lineage-specific transcription of the..., Sil [/bib_ref]. To allow this asymmetry, Prospero and Ash1 exit the nucleus before division and relocalize to the cell cortex, similarly to the other fate determinants [bib_ref] The prospero transcription factor is asymmetrically localized to the cell cortex during..., Spana [/bib_ref] [bib_ref] Mating type switching in yeast controlled by asymmetric localization of ASH1 mRNA, Long [/bib_ref] [bib_ref] Actin-dependent localization of an RNA encoding a cell-fate determinant in yeast, Takizawa [/bib_ref] [bib_ref] Staufen-dependent localization of prospero mRNA contributes to neuroblast daughter-cell fate, Broadus [/bib_ref] [bib_ref] Miranda mediates asymmetric protein and RNA localization in the developing nervous system, Schuldt [/bib_ref].
The transcription factor Ace2, however, does not appear to use this mechanism. This protein, which contributes to the asymmetry of budding yeast division, accumulates specifically in the nucleus of the bud. Indeed, yeast cells divide by budding of a future daughter cell off the surface of the original mother cell. By inducing transcription of several daughterspecific genes, Ace2 triggers many aspects of daughter-specific transcription and governs the physical separation of the newborn cell from its mother after completion of cytokinesis [bib_ref] Parallel pathways of gene regulation: homologous regulators SWI5 and ACE2 differentially control..., Dohrmann [/bib_ref] [bib_ref] Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by..., Spellman [/bib_ref] [bib_ref] Yeast Cbk1 and Mob2 activate daughter-specific genetic programs to induce asymmetric cell..., Colman-Lerner [/bib_ref].
Unlike Prospero and Ash1, Ace2 localizes fully symmetrically to the cytoplasm throughout mitosis [bib_ref] Parallel pathways of gene regulation: homologous regulators SWI5 and ACE2 differentially control..., Dohrmann [/bib_ref]. Indeed, Ace2 asymmetry is established at the level of D uring vegetative growth, Saccharomyces cerevisiae cells divide asymmetrically: the mother cell buds to produce a smaller daughter cell. This daughter asymmetrically inherits the transcription factor Ace2, which activates daughter-specific transcriptional programs. In this paper, we investigate when and how this asymmetry is established and maintained. We show that Ace2 asymmetry is initiated in the elongated, but undivided, anaphase nucleus. At this stage, the nucleoplasm was highly compartmentalized; little exchange was observed for nucleoplasmic proteins between mother and bud. Using photobleaching and in silico modeling, we show that diffusion barriers compartmentalize the nuclear membranes. In contrast, the behavior of proteins in the nucleoplasm is well explained by the dumbbell shape of the anaphase nucleus. This compartmentalization of the nucleoplasm promoted Ace2 asymmetry in anaphase nuclei. Thus, our data indicate that yeast cells use the process of closed mitosis and the morphological constraints associated with it to asymmetrically segregate nucleoplasmic components. mCherry-Tub1. We repeatedly bleached Ace2-GFP in one quarter of the daughter nucleus and monitored fluorescence levels in the opposite quarter . The Ace2-GFP signal rapidly decayed over the entire nucleus. As a control, we repeated this procedure with cells expressing the histone Htb2-GFP , which is stably incorporated into chromatin [bib_ref] Twenty-five years of the nucleosome, fundamental particle of the eukaryote chromosome, Kornberg [/bib_ref]. In contrast to Ace2-GFP, Htb2-GFP fluorescence decayed only slowly in the nonbleached part of the nucleus compared with the bleached area. Similarly, paraformaldehyde fixation of the cells stopped Ace2-GFP exchange between the bleached and unbleached parts of the nucleus . This suggests that Ace2-GFP freely diffuses in the nucleoplasm. We conclude that the selective retention of Ace2-GFP in one half of the anaphase nucleus is not caused by immobilization of Ace2 in the bud nucleus.
## Exchange between the two halves of the late anaphase nucleus is restricted
We next investigated whether the anaphase nucleoplasm is compartmentalized. In this case, nucleoplasmic proteins should not freely exchange between mother and daughter. To test this possibility, we performed photobleaching experiments on tetracycline repressor (TetR)-GFP [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref]. We performed these FLIP experiments at early and late stages of nuclear division, i.e., anaphase, as determined by nuclear morphology. The early stage of anaphase begins as soon as the nucleus enters the bud and contracts in the bud neck. This stage ends when a thin nuclear bridge forms between the two future rounded up nuclei. In late anaphase, the nucleus adopts a fully elongated, dumbbelllike shape, touching mother and daughter cell cortexes. Repetitive photobleaching of TetR-GFP fluorescence in a small area of the mother part of late anaphase nuclei led to rapid fluorescence loss in the mother, whereas fluorescence loss was much slower in the daughter nucleoplasm, and vice versa (Figs. 2 A and S1 A). Thus, the TetR-GFP exchange between the two halves of the nucleus is restricted. As reported previously, we detected no compartmentalization in early anaphase [bib_ref] A mechanism for asymmetric segregation of age during yeast budding, Shcheprova [/bib_ref].
To extract quantitative information from these FLIP experiments, we determined the time required to lose 30% of the GFP fluorescence in mother and bud parts of the nucleus. The ratio of these two durations (bud over mother) expresses the different fluorescence decay rates in the two compartments upon bleaching the mother part of the nucleus. It is inversely proportional to the exchange rate between the two compartments. We termed this ratio "degree of compartmentalization" (°CP). In early anaphase, the °CP of TetR-GFP was around two . In late anaphase, it was 30-fold higher [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref].
Control FLIP experiments established that the vast majority of TetR-GFP exchange between mother and daughter occurs through the nuclear tube , B-E). When we photobleached the cytoplasm of TetR-GFP-expressing cells, fluorescence barely decayed in the nucleus. Thus, TetR-GFP shuttling between nucleoplasm and cytoplasm had very little impact on the TetR-GFP exchange between mother and bud parts of the anaphase nucleus. Furthermore, we observed no exchange of TetR-GFP between nuclei in cells that had undergone nuclear import/export, before cytokinesis [bib_ref] Parallel pathways of gene regulation: homologous regulators SWI5 and ACE2 differentially control..., Dohrmann [/bib_ref] [bib_ref] Regulated nuclear localisation of the yeast transcription factor Ace2p controls expression of..., O'conallain [/bib_ref] [bib_ref] Yeast Cbk1 and Mob2 activate daughter-specific genetic programs to induce asymmetric cell..., Colman-Lerner [/bib_ref] [bib_ref] The Saccharomyces cerevisiae Mob2p-Cbk1p kinase complex promotes polarized growth and acts with..., Weiss [/bib_ref] [bib_ref] The NDR/LATS family kinase Cbk1 directly controls transcriptional asymmetry, Mazanka [/bib_ref]. Ace2 enters both mother and daughter nuclei but efficiently exits the nucleus located in the mother cell and not its counterpart in the bud [bib_ref] The NDR/LATS family kinase Cbk1 directly controls transcriptional asymmetry, Mazanka [/bib_ref] [bib_ref] Sequential counteracting kinases restrict an asymmetric gene expression program to early G1, Mazanka [/bib_ref]. Retention in the daughter part of the nucleus in late mitosis is facilitated through Ace2 phosphorylation by the LATS-related kinase Cbk1, the downstream effector of the regulation of Ace2 and morphogenesis (RAM) network [bib_ref] The Saccharomyces cerevisiae Mob2p-Cbk1p kinase complex promotes polarized growth and acts with..., Weiss [/bib_ref] [bib_ref] The NDR/LATS family kinase Cbk1 directly controls transcriptional asymmetry, Mazanka [/bib_ref] [bib_ref] The hippo signaling pathway in development and cancer, Pan [/bib_ref].
Here, we show that the asymmetric distribution of Ace2 in the nucleus precedes karyofission. Using photobleaching techniques and spatial stochastic modeling and simulations, we investigated the mechanisms allowing Ace2 accumulation specifically in one half of the nucleus.
# Results
## Asymmetric ace2 localization in the nucleus precedes nuclear division
To determine when Ace2 asymmetry is established relative to nuclear division, we followed Ace2-GFP using live-cell fluorescence microscopy. We visualized nuclear shape using the reporter protein dsRed-HDEL, which localizes to the lumen of the ER and perinuclear space between outer and inner nuclear membranes (ONM and INM, respectively). We followed the different steps of nuclear elongation and nuclear division and correlated them with Ace2 localization. As previously reported, Ace2-GFP begins to enter the nucleus in midanaphase and localizes first to the entire nucleus [bib_ref] Yeast Cbk1 and Mob2 activate daughter-specific genetic programs to induce asymmetric cell..., Colman-Lerner [/bib_ref] [bib_ref] The Saccharomyces cerevisiae Mob2p-Cbk1p kinase complex promotes polarized growth and acts with..., Weiss [/bib_ref]. However, in cells with undivided dumbbellshaped nuclei, the Ace2-GFP signal accumulated asymmetrically in the nuclear lobe located in the bud . In time-lapse experiments, Ace2 asymmetry preceded fission of the bridge linking the two future nuclei by 8 min.
Fission of the inner membrane of the nucleus might precede disassembly of the bridge, but light microscopy does not have the resolution required to observe such an event. We therefore used nucleoplasmic spindle microtubules as a second marker to determine whether the nucleoplasm remains continuous during the establishment of Ace2 asymmetry. Using timelapse microscopy, we monitored Ace2-GFP localization in cells coexpressing mCherry-Tub1 to visualize the relative timing of Ace2 asymmetry and spindle breakdown. In all cases (n = 51), Ace2 distribution became strongly asymmetric in cells with elongated, intact spindles . Thus, Ace2 asymmetry is established before spindle breakdown and karyofission.
This observation suggests that some mechanism prevents Ace2 in the daughter part of the nucleus from diffusing back into the mother part. One possible mechanism for Ace2 retention in one half of the nucleus could be tethering of the protein to chromatin or other structurally constrained nucleoplasmic components. In this case, Ace2 would become immobilized in the daughter nucleus. To assess Ace2 mobility in the daughter nucleoplasm, we used fluorescence loss in photobleaching (FLIP) on cells with elongated, intact mitotic spindles visualized with Figure 1. Ace2 asymmetry precedes nuclear division and is independent of Ace2 immobilization. (A and B) Ace2-GFP localization in late anaphase cells expressing dsRed-HDEL (A) or mCherry-Tub1 (B); 0 min = Ace2 asymmetry. (C-E) FLIP in daughter nuclei of Ace2-GFP (n = 9; C), HTB2-GFP (n = 9; D), or Ace2-GFP in fixed cells (n = 9; E). In images, representative daughter nuclei are shown; in graphs, fluorescence levels over time in bleached (red) and opposing nonbleached (green) quarters (means ± SD) are given. Blue boxes indicate sections of the image shown on the right. White lines indicate cell outlines. Bars: (A-E, left) 2 µm; (C-E, right) 3 µm. t50%, time to lose 50% of initial fluorescence.
(Fig. S1 D) but fail to undergo cytokinesis [bib_ref] Spatial coordination of cytokinetic events by compartmentalization of the cell cortex, Dobbelaere [/bib_ref].
TetR-GFP does not recognize any specific sequence in the yeast genome. Nevertheless, its ability to bind random DNA karyofission but not cytokinesis. This was most obvious upon FLIP TetR-GFP in wild-type cells with separated nuclei and in cdc12-6 mutant cells incubated at 30°C . At this temperature, cdc12-6 cells still divide their nucleus Mean fluorescence levels ± SD over time in the mother (red) and daughter part (green) are shown; bleaching area is indicated in blue. White lines indicate cell outlines. (E) °CP values for the indicated markers during early (blue) and late (orange) stages (for n see ; means ± SEM). Numbers indicate relative increase of °CP from early to late anaphase. (F) Diffusion constants of the nucleoplasmic proteins measured by FCS (means ± SD; n TetR = 168, n TetRHTH = 204, and n NLS-3GFP = 179). ***, P < 0.0001 (t test). Bars, 3 µm. the two nuclear halves was detectable in late anaphase. Using GFP-Src1, an INM protein homologous to LEM domain proteins in mammalian cells, and the artificial INM proteins LR1 and LR2, we confirmed that the exchange of INM proteins is rapid between mother and bud in early anaphase cells ; [bib_ref] SRC1: an introncontaining yeast gene involved in sister chromatid segregation, Rodríguez-Navarro [/bib_ref] [bib_ref] The inner nuclear membrane protein Src1 associates with subtelomeric genes and alters..., Grund [/bib_ref] [bib_ref] Long unfolded linkers facilitate membrane protein import through the nuclear pore complex, Meinema [/bib_ref]. However, as for Src1, a fiveto sevenfold increase in compartmentalization was observed in late anaphase cells. Because LR1 and LR2 are fully artificial proteins, this compartmentalization is not caused by a specific interaction with, for example, scaffolding proteins. Thus, compartmentalization of the nuclear membrane increases in parallel to that of the nucleoplasm as anaphase progresses. These results show that, throughout nuclear division, compartmentalization increases not only in the nucleoplasm but also in both nuclear membranes. Compartmentalization of the ONM occurs first in early anaphase and increases at least by a factor of six as the nucleus adopts a dumbbell shape. Compartmentalization of the INM appears to occur only at this late stage.
## Compartment boundaries are localized in the nuclear bridge
We next wanted to determine where the boundary between the mother and daughter compartments is located within the internuclear bridge. Several options are possible: the boundary could be localized only to one specific part of the bridge, delocalized over the entire bridge, or each end of the bridge could form boundaries [fig_ref] Figure 4: The bridge connecting the two future nuclei forms the compartments' boundary [/fig_ref]. To distinguish between these possibilities, we performed FLIP experiments, bleaching each time a different small area in the bridge connecting the two future nuclei of late anaphase cells, and determined the °CP between the bud and mother parts of the nucleus, asking on what side of the compartments' boundary we bleached the nucleus [fig_ref] Figure 4: The bridge connecting the two future nuclei forms the compartments' boundary [/fig_ref]. This became apparent when we correlated the °CP and its inverse (°CP 1 ) with the position of the bleaching region in the tube [fig_ref] Figure 4: The bridge connecting the two future nuclei forms the compartments' boundary [/fig_ref]. The length of the entire nucleus was normalized to 100% to allow comparison between cells. The mean position of the bud neck was displaced slightly toward the bud half of the spindle, to position 58.1 ± 2.2% (mean ± SD, n = 20). The ends of the internuclear bridge were around positions 32.0 ± 3.4% in the mother and 70.9 ± 2.8% in the daughter. The ends of the tube were defined as the intersection points of the nuclear length axis with the extrapolated spherical outline of the respective daughter nuclei. We averaged the °CP values by bins of five cells with increasing distance of the photobleached area relative to the end of the nucleus in the mother cell. The resulting °CP value was plotted at the mean photobleaching position in the cells of the bin. The two curves described by the data points intersect at a °CP of one, marking the boundary between compartments. When bleaching on the mother side of the boundary, results in a °CP > 1, whereas bleaching on the daughter side results in a °CP between 1 and 0; the °CP 1 should mirror this behavior. The curves for all three markers tested-GFP-Src1, Nsg1-GFP, and TetR-GFP-intersected at the spindle center rather than at the bud neck, suggesting that the boundary is in the bridge but not at the position of the bud neck. Furthermore, none of the curves showed any strong discontinuity sequences with low affinity might affect its diffusion in the yeast nucleus and thereby contribute to its compartmentalization [bib_ref] Dynamics of repressor-operator recognition: the Tn10-encoded tetracycline resistance control, Kleinschmidt [/bib_ref]. To test whether nonspecific DNA binding influences TetR diffusion, we measured the diffusion rates of TetR-GFP, TetR-helix-turn-helix (HTH)-GFP, which lacks the DNA-binding motif, and a fusion of three GFPs with an NLS (NLS-3GFP) in vivo using fluorescence correlation spectroscopy (FCS; [bib_ref] Structural basis of gene regulation by the tetracycline inducible Tet repressoroperator system, Orth [/bib_ref]. The nuclear diffusion coefficient of TetR-GFP was 1.9 ± 0.5 µm 2 /s (mean ± SD, n = 168; [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref] , whereas TetR-HTH-GFP and NLS-3GFP diffused significantly faster (2.7 ± 0.9 µm 2 /s, n TetRHTH = 204; 4.4 ± 2.5 µm 2 /s, n NLS-3GFP = 179). Hence, nonspecific binding of TetR to DNA significantly affected its diffusion properties inside the nucleus. Therefore, transient binding of TetR to the separating chromatin masses during late anaphase might contribute to its compartmentalization.
For these reasons, we next investigated whether DNA binding is a prerequisite for compartmentalization. Using FLIP, no compartmentalization was observed for TetR-HTH-GFP and NLS-3GFP during early anaphase [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref]. However, in cells with dumbbell-shaped nuclei, the °CP increased 10-fold for TetR-HTH and 15-fold for NLS-3GFP [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref] , B-E; and . Thus, compartmentalization of nucleoplasmic proteins in late anaphase nuclei does not require DNA binding capability per se. The lower °CPs of TetR-HTH-GFP and NLS-3GFP compared with TetR-GFP indicate that DNA binding only increases compartmentalization by a factor of two to three. We noticed that both NLS-3GFP and TetR-HTH-GFP shuttled more rapidly between nucleus and cytoplasm than TetR-GFP (Fig. S1 F and not depicted). Therefore, DNA binding may enhance compartmentalization of TetR-GFP through nucleoplasmic retention of the protein.
Analysis of a soluble marker of the lumen between ONM and INM, GFP-HDEL, indicated that compartmentalization did not always increase as nuclear division progressed. First, no major compartmentalization effect was observed for GFP-HDEL during early anaphase . Second, the °CP of GFP-HDEL increased only threefold in late anaphase . Thus, compartmentalization of the nucleoplasm is specific and does not affect the perinuclear space to the same extent.
## Effect of mitotic progression on the compartmentalization of the onm and inm
Analysis of protein diffusion in the ONM and INM revealed yet another set of compartmentalization patterns. To assess the diffusion of nuclear pore complexes (NPCs) and ONM proteins, we performed FLIP experiments on the NPC protein Nup49-GFP and the ONM proteins Nsg1-GFP, Hmg1-GFP, and Hmg2-GFP [bib_ref] Saccharomyces cerevisiae contains two functional genes encoding 3-hydroxy-3-methylglutarylcoenzyme A reductase, Basson [/bib_ref] [bib_ref] A new family of yeast nuclear pore complex proteins, Wente [/bib_ref] [bib_ref] INSIG: a broadly conserved transmembrane chaperone for sterolsensing domain proteins, Flury [/bib_ref]. As reported previously, NPCs and ONM proteins were compartmentalized in early anaphase [bib_ref] A mechanism for asymmetric segregation of age during yeast budding, Shcheprova [/bib_ref]. During the early phase of nuclear division, Nsg1-GFP had a °CP of 7, whereas that of Nup49-GFP was much higher . The compartmentalization of the ONM markers increased six-to ninefold as the cells progressed through anaphase. Concerning Nup49-GFP, no exchange between [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref]. Mean fluorescence levels ± SD over time in the mother (red) and daughter part (green) are shown; bleaching area is indicated in blue. White lines indicate cell outlines. (D) °CP values for the indicated markers during early (blue) and late (orange) stages of nuclear division (n ≥ 13 also see ; means ± SEM). Numbers indicate relative increase of °CP from early to late anaphase. Bars, 3 µm. at the spindle midzone. The ase1 mutation causes premature spindle breakdown and shortening of the dumbbell-shaped nucleus [bib_ref] APC-mediated proteolysis of Ase1 and the morphogenesis of the mitotic spindle, Juang [/bib_ref] [bib_ref] Ase1p organizes antiparallel microtubule arrays during interphase and mitosis in fission yeast, Loïodice [/bib_ref]. To compare wild-type and ase1 cells, we restricted our FLIP experiments to dumbbell-shaped nuclei with a total length of 6-7 µm [fig_ref] Figure 5: Compartmentalization of the nucleoplasm is specifically decreased in long ase1 nuclei [/fig_ref]. Nucleoplasmic compartmentalization decreased significantly in ase1 cells, as seen using TetR-GFP and NLS-3GFP as reporters (Figs. 5 B and S2 A and . However, the compartmentalization of the ONM and INM markers GFP-Src1 and Nsg1-GFP was unaffected compared with wild-type cells [fig_ref] Figure 5: Compartmentalization of the nucleoplasm is specifically decreased in long ase1 nuclei [/fig_ref]. Interestingly, deleting BUD6 or SHS1, which has been shown to significantly decrease the °CP of Nsg1-GFP in early in the tube or at its extremities, suggesting that, in all three cases, the entire bridge forms a gradual boundary. Therefore, the bridge might restrict exchange either through its geometry or through a diffusion barrier along the length of the bridge.
## Mutations interfering with the geometry of the dumbbell-shaped nucleus alter nucleoplasmic compartmentalization
Because compartment boundaries were located on the spindle center in late anaphase, we next investigated whether the spindle midzone contributes to boundary formation. We deleted ASE1, which encodes a microtubule-binding protein localized ; means ± SEM). *, P < 0.05 (t test). Bars, 3 µm. [bib_ref] A genetic screen for ribosomal DNA silencing defects identifies multiple DNA replication..., Smith [/bib_ref] [bib_ref] Association of the RENT complex with nontranscribed and coding regions of rDNA..., Huang [/bib_ref] [bib_ref] Kinetochore recruitment of two nucleolar proteins is required for homolog segregation in..., Rabitsch [/bib_ref] [bib_ref] Role for perinuclear chromosome tethering in maintenance of genome stability, Mekhail [/bib_ref]. The slk19 mutant cells had nuclei with longer and slightly wider, but correctly placed, bridges. Slk19 is a signaling protein involved in spindle dynamics and exit from mitosis [bib_ref] Slk19p is a centromere protein that functions to stabilize mitotic spindles, Zeng [/bib_ref] [bib_ref] Separase, polo kinase, the kinetochore protein Slk19, and Spo12 function in a..., Stegmeier [/bib_ref] [bib_ref] A midzone-based ruler adjusts chromosome compaction to anaphase spindle length, Neurohr [/bib_ref]. To assess the contribution of the individual geometries to nucleoplasmic compartmentalization, we correlated the °CP of TetR-GFP in the knockout strains with the bridge length normalized to its cross section. Remarkably, across all strains and situations considered, °CP and normalized bridge length remained proportional to each other [fig_ref] Figure 6: Changes in nuclear shape influence nucleoplasmic compartmentalization [/fig_ref]. Mutants that displaced their anaphase nucleus from the bud neck, as well as those placing it correctly, follow the same linear correlation between bridge geometry and compartmentalization of the nucleoplasm.
To investigate the role of the spindle as an obstacle restricting exchange between mother and daughter nucleoplasm, we compared the spindle in dumbbell-shaped nuclei of ase1 and cin8 cells using Nsg1-GFP to outline the nuclear periphery and mCherry-Tub1 to visualize the spindle. Both mutants had low °CPs and similar geometry defects, although in the ase1 mutants cells, 60% of the dumbbell nuclei showed no continuous mCherry-Tub1 signal in the bridge, whereas virtually all cin8 cells did [fig_ref] Figure 6: Changes in nuclear shape influence nucleoplasmic compartmentalization [/fig_ref] , D and E). We conclude that the absence of microtubule bundles in the bridge is not the main cause of the compartmentalization defect observed in ase1.
Overall, the correlation between the geometry of the dividing nucleus and the °CP of TetR-GFP strongly suggests that nuclear morphology is the main determinant for nucleoplasmic compartmentalization. The spindle midzone itself and the bud neck have much less impact on this process, if any.
## Stochastic spatial modeling of compartmentalization
To theoretically dissect the impact of geometry and potential diffusion barriers on nuclear compartmentalization, we next used off-lattice spatial stochastic simulations in idealized cells during early and late stages of nuclear division. In off-lattice simulations, the simulation domain is usually discretized to efficiently localize particles, yet each particle can diffuse and react while preserving explicit coordinates at all time points [bib_ref] Stochastic simulation for spatial modelling of dynamic processes in a living cell, Burrage [/bib_ref] [bib_ref] Anomalous diffusion and multifractional Brownian motion: simulating molecular crowding and physical obstacles..., Marquez-Lago [/bib_ref]. Such particle methods yield the highest available accuracy of microscopic cellular dynamics, which was set to 20 nm in our case. Therefore, they provide the ideal basis to test whether diffusion barriers need to be introduced to reproduce in silico the compartmentalization observed in our FLIP experiments in vivo.
After defining simulation geometries for the nucleus during early and late stages of division, separate simulations were run using Smoldyn 2.22 for each molecular species, in each distinctive domain of the nuclear space, and within different timescales (see Materials and methods; [fig_ref] Figure 3: Nuclear membranes are compartmentalized during late stages of nuclear division [/fig_ref] ; [bib_ref] Detailed simulations of cell biology with Smoldyn 2.1, Andrews [/bib_ref]. First, we determined four necessary parameters to carry out simulations of our FLIP experiments presented in Figs. 2 and 3: the number of diffusing particles, the rate of photobleaching in the irradiated area, the diffusion rates of the different molecules, and the strength of any potential diffusion barrier. The determination of the first three parameters can be found in the Materials anaphase [bib_ref] A mechanism for asymmetric segregation of age during yeast budding, Shcheprova [/bib_ref] , had no effect on Nsg1-GFP compartmentalization in dumbbell nuclei [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref]. Thus, the ONM barrier is regulated differently in early compared with late anaphase. The effect of the ASE1 deletion on TetR-GFP exchange indicates that Ase1 contributes to nucleoplasmic compartmentalization. In contrast, it does not affect the compartmentalization of INM and ONM, indicating that the nucleoplasm and the membranes are compartmentalized through different mechanisms.
Deleting ASE1 shortens the overall length of the anaphase nucleus but also reduces bridge length [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref]. We next asked how important this change in nuclear geometry is for compartmentalization. Therefore, we plotted the °CP of individual cells against their bridge length to test whether these parameters correlate with each other [fig_ref] Figure 6: Changes in nuclear shape influence nucleoplasmic compartmentalization [/fig_ref]. The compartmentalization of TetR-GFP correlated well with bridge length, whereas that of GFP-Src1 and Nsg1-GFP did less. Consequently, in wild-type and ase1 strains, the °CP of TetR-GFP was on average higher in nuclei with longer bridges than in those with shorter bridges. Thus, the effect of the ase1 mutation on bridge length at the end of anaphase probably contributed to the effect of this mutation on compartmentalization. However, ase1 mutant cells showed an additional loss of nucleoplasmic compartmentalization that was independent of bridge length, as demonstrated by the observation that at a given length, the ase1 mutant showed a lower °CP for TetR-GFP than the respective wild type. This could be caused either directly by the midzone plugging the tube of wild-type cells or indirectly by a change of bridge diameter in ase1 cells.
Thus, we next investigated which of plugging the bridge with the spindle midzone, bridge dimensions, and correct positioning of the dumbbell-shaped nucleus relative to the bud neck is most important for nucleoplasmic compartmentalization. Therefore, we first measured the length and width of the bridge in nuclei of several knockout strains (Figs. 6 B and S2, C-E). For this, the bridge width was assessed through the total fluorescence intensity of the Nsg1-GFP marker per length unit of the bridge [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref]. Indeed, the intensity of GFP fluorescence per bridge length is expected to correlate linearly with the diameter of the bridge because the density of Nsg1-GFP fluorescence in the ONM was not significantly different in all strains tested [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref]. Several knockout strains showed reduced bridge length and increased bridge diameter, including ase1, the kinesin cin8, the dynein heavy chain dyn1, and a strain lacking both the spindle orientation checkpoint component BUB2 and the dynactin component ARP1 [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref] [bib_ref] S. cerevisiae genes required for cell cycle arrest in response to loss..., Hoyt [/bib_ref] [bib_ref] Two Saccharomyces cerevisiae kinesin-related gene products required for mitotic spindle assembly, Hoyt [/bib_ref] [bib_ref] Kinesin-related proteins required for assembly of the mitotic spindle, Roof [/bib_ref] [bib_ref] Kinesin-related proteins required for structural integrity of the mitotic spindle, Saunders [/bib_ref] [bib_ref] Cytoplasmic dynein is required for normal nuclear segregation in yeast, Eshel [/bib_ref] [bib_ref] Disruption of mitotic spindle orientation in a yeast dynein mutant, Li [/bib_ref] [bib_ref] A yeast actin-related protein homologous to that in vertebrate dynactin complex is..., Muhua [/bib_ref]. Whereas nuclei measured in ase1 or cin8 cells crossed the bud neck, nuclei measured in dyn1 and bub2arp1 divided in the mother cell, i.e., displaced relative to the bud neck. Nuclei of cnm67 cells dividing away from the bud neck also showed slightly increased bridge diameters but no changes in bridge length. Cnm67 is a spindle pole body component [bib_ref] Saccharomyces cerevisiae cells with defective spindle pole body outer plaques accomplish nuclear..., Brachat [/bib_ref]. Deleting the cohibin complex components LRS4 and CSM1 had the same consequences for nuclear architecture as the CNM67 deletion but did not displace the dividing nuclei from the bud neck ; means ± SEM) over bridge length normalized by a value proportional to the bridge's cross section (means ± SD). Cells that divide the nucleus in the mother cell are indicated in red. (A and C) Lines indicate linear regression of the data point using Prism 5.0b. (D) mCherry-Tub1 spindles in wild type, ase1, and cin8 dumbbell nuclei outlined by Nsg1-GFP. (B and D) White lines indicate cell outlines. (E) Percentage of dumbbell-shaped nuclei with (green) and without (red) mCherry-Tub1 staining (MTs) in the bridge (means ± SD; n = 3). Bars, 3 µm. Nuclear compartmentalization during closed mitosis - Boettcher et al.
Simulations performed on TetR-GFP in nuclei with changed bridge morphologies further support this idea [fig_ref] Figure 4: The bridge connecting the two future nuclei forms the compartments' boundary [/fig_ref] : dilating the bridge 3.4-fold decreased the compartmentalization of TetR-GFP approximately eightfold. Shortening the tube to a third of its original length reduced compartmentalization 2.5fold. The combination of both conditions abolished compartmentalization. These results are consistent with the dimensions of the tube having a strong impact on nucleoplasmic compartmentalization, as suggested by the experimental results [fig_ref] Figure 6: Changes in nuclear shape influence nucleoplasmic compartmentalization [/fig_ref].
A strikingly different situation was observed for Nsg1, Nup49, and Src1: all simulations that did not assume the existence of a lateral diffusion barrier in the plane of the membrane failed to accurately reproduce the experimental data of fluorescence loss [fig_ref] Figure 7: Averaged deviations of stochastic model simulations from the experimental mean [/fig_ref] , B, C, E, and F). In the case of Nsg1 particles, the probability of crossing the boundary had to be reduced to 12% of unrestrained conditions in early anaphase and 5% in late anaphase to replicate the behavior seen in vivo [fig_ref] Figure 7: Averaged deviations of stochastic model simulations from the experimental mean [/fig_ref]. For Nup49 and Src1, these probabilities were 2.5 and 25%, respectively, independently of whether the nuclei were in early or late stages of nuclear division (Figs. 7, C and F; and S4, A and B). Details on how deviations from experimental data arose in mother and bud compartments are presented in [fig_ref] Figure 5: Compartmentalization of the nucleoplasm is specifically decreased in long ase1 nuclei [/fig_ref].
The simulations show that the nucleoplasm is compartmentalized because of its geometry. In contrast, diffusion barriers play a major role in subdividing the nuclear membranes into a mother and a daughter compartment throughout nuclear division. and methods section. Here, we describe the outcome of the simulations concerning the contribution of potential diffusion barriers in the process of matching experimental and simulated data.
To reproduce the in vivo FLIP experiments on Nsg1, Nup49, Src1, and TetR, we performed parameter sweep simulations assuming various diffusion coefficient values and barriers of different strengths. We defined barriers as a constraint on the given particle allowing it to pass from one to the opposite nuclear lobe at a set probability (p). These barriers were formed by a plane covering the area defined by the bud neck in early anaphase or the middle of the cylinder connecting the future mother and bud nucleus in late anaphase. The value p = 1 indicates that no barrier is present. Values <1 indicate the presence of a barrier acting beyond morphology, whereas values >1 indicate directed transport from one compartment to the other. Because there was no experimental evidence for the latter, we did not include directed transport in the current model. Moreover, we assumed that the passage probability is symmetric, i.e., not dependent on the side from which the particle originated. Subsequently, we determined the distances between the modeled and experimental decay curves.
To explain the behavior of TetR in the nucleoplasm during early and late anaphase, no diffusion barrier is needed, as experimental data and simulation already matched at p = 1 (Figs. 7, A and D; and S4, A and B). Therefore, changes in geometry account for most, if not all, changes in nucleoplasmic compartmentalization as cells progress through anaphase. [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref]. Mother is shown in red; daughter is shown in green. t(50%), time to recover 50% of initial fluorescence in the mother lobe (n WT = 20 and n ase1 = 20; means ± SEM). (B) FLIP on Ace2-GFP in wild type (WT) and ase1. Fluorescence levels over time in representative cells, with nonbleached nuclear lobes in the daughter Nuclear compartmentalization during closed mitosis - Boettcher et al.
# Discussion
In this study, we show that during late anaphase, the nucleus of budding yeast is extensively compartmentalized. Unlike in early stages, during which exchange between mother and bud is not limited in nucleoplasmic compartments, all nuclear spaces become strongly compartmentalized in late mitosis, when the nucleus adopts a dumbbell-like shape. This compartmentalization has clear functional consequences for the cell. By limiting the exchange of the transcription factor Ace2 between mother and bud parts of the nucleus, compartmentalization of the nucleoplasm supports the establishment of Ace2 asymmetry and promotes its inheritance by the daughter cell. Thus, when Ace2 asymmetry is important, nucleoplasmic compartmentalization is likely to confer a selectable advantage to the cell. In other words, mechanisms that ensure compartmentalization of the nucleoplasm could have emerged at least partially in coevolution with Ace2 asymmetry.
Using a combination of experimental and simulated FLIP data, together with mutants affecting spindle organization and nuclear morphology, we could distinguish between the contribution of geometry and the permeability of potential diffusion barriers to nuclear compartmentalization. These experiments establish the following three points.
First, compartmentalization of the nucleoplasm is well explained by the geometry of the nucleus alone. Accordingly, parameters such as the length and width of the bridge connecting the two future daughter nuclei determine the °CP.
Second, compartmentalization of nucleoplasmic proteins is increased for those that are able to bind DNA even with low affinity, such as TetR. This twofold increase in compartmentalization is likely to be mediated by three convergent processes: the separation of the segregating chromosomes into two disjoint masses, the reduction of the diffusion speed of the protein, and its increased retention in the nucleus, which limits exchange via the cytoplasm.
Third, in contrast to what is observed in the nucleoplasm, the compartmentalization of the nuclear membranes cannot be explained by their geometry alone. Already in early anaphase, the dynamics of markers in both the INM and ONM are best explained by the presence of a barrier, which limits lateral diffusion around the spindle midzone, in addition to geometry. Accordingly, in both membranes, the length of the internuclear bridge only modestly affected compartmentalization. Most remarkably, the width of the bridge, which had a strong impact on the compartmentalization of the nucleoplasm, made no clear contribution to the compartmentalization of the INM and ONM.
In a previous study, we found that specialized membrane structures are present in the cortical ER at the bud neck [bib_ref] Septin-dependent compartmentalization of the endoplasmic reticulum during yeast polarized growth, Luedeke [/bib_ref]. These structures contribute to the formation of
## Nucleoplasmic compartmentalization contributes to ace2-gfp asymmetry
Our observations show that the morphological events associated with closed mitosis in budding yeast passively promote compartmentalization of the nucleoplasm during late stages of nuclear division. However, this compartmentalization might be needed to promote selected processes, such as Ace2 segregation. Therefore, we tested whether mutations altering nucleoplasmic compartmentalization affect the behavior and distribution of Ace2.
Using FRAP and FLIP, we characterized the dynamics of Ace2. First, we used FRAP to study the compartmentalization of the Ace2-F127V variant, which is trapped in mother and daughter nuclei [bib_ref] Cbk1p, a protein similar to the human myotonic dystrophy kinase, is essential..., Racki [/bib_ref] [bib_ref] The NDR/LATS family kinase Cbk1 directly controls transcriptional asymmetry, Mazanka [/bib_ref]. Consistent with our data on nucleoplasmic compartmentalization, the signal in the mother nucleus recovered slowly upon photobleaching before spindle breakdown as visualized by mCherry-Tub1 [fig_ref] Figure 8: Reduction of Ace2 asymmetry in ase1 cells [/fig_ref]. In contrast, recovery was rapid in ase1 mutant cells coinciding with the loss of fluorescence in the daughter half of the nucleus. Similarly, when Ace2-GFP was bleached continuously in the mother half of the nucleus, fluorescence was lost slowly in the daughter half, whereas fluorescence loss was faster in similarly treated ase1 mutant cells [fig_ref] Figure 8: Reduction of Ace2 asymmetry in ase1 cells [/fig_ref]. We conclude that, similar to our observations of TetR-GFP, the dimensions of the internuclear bridge limit the exchange of Ace2 between the two halves of dumbbell-shaped nuclei. Thus, compartmentalization contributes to the retention of Ace2 in the daughter nucleus before karyofission.
To test whether compartmentalization contributes to Ace2 asymmetry, we investigated whether compartmentalization mutants displayed altered Ace2 segregation upon cell division. Fluorescence measurements in cells after karyofission as visualized by dsRed-HDEL indicated that Ace2-GFP was distributed more symmetrically in ase1 and cin8 cells [fig_ref] Figure 8: Reduction of Ace2 asymmetry in ase1 cells [/fig_ref] , C and D). Deleting SLK19, which increases nucleoplasmic compartmentalization, slightly increased Ace2 asymmetry. Furthermore, Ace2-S122A distributed more symmetrically in ase1 cells [fig_ref] Figure 8: Reduction of Ace2 asymmetry in ase1 cells [/fig_ref]. Ace2-S122A is mutated in one of its two Cbk1 sites close to the nuclear export signal and is partially defective in establishing Ace2 asymmetry [bib_ref] The NDR/LATS family kinase Cbk1 directly controls transcriptional asymmetry, Mazanka [/bib_ref]. Nevertheless, the symmetry is not as complete as in cells expressing Ace2-AA-GFP lacking both Cbk1 phosphorylation sites [bib_ref] The NDR/LATS family kinase Cbk1 directly controls transcriptional asymmetry, Mazanka [/bib_ref]. Interestingly, the effect of the S122A mutation on Ace2 segregation was partially suppressed in slk19. Collectively, these data indicate that compartmentalization contributes to the maintenance of Ace2 asymmetry, whereas the RAM pathway establishes Ace2 asymmetry and prevents full Ace2 symmetry in ase1 and cin8 cells. We conclude that nucleoplasmic compartmentalization and the RAM network contribute synergistically to the asymmetric segregation of Ace2.
in wild type (filled circles) and ase1 (open circles). t(50%), time to bleach 50% of initial fluorescence in the daughter nucleus (n = 20 for both; means ± SEM). The white circles label the outline of the cells in transmission light pictures, and the blue circles indicate the bleaching area. (C) Localization of Ace2-GFP in wild type and ase1 after nuclear division. (D) Distribution of Ace2-GFP in wild type (n = 175), ase1 (n = 198), cin8 (n = 180), and slk19 (n = 216). (E) Distribution of Ace2(S122A)-GFP (Ace2-A) in wild type (n = 192), ase1 (n = 192), and slk19 (n = 206) and Ace2(S122A,S137A)-GFP (Ace2-AA) in wild type (n = 199). Lines indicate the medians. ***, P < 0.0001 (t test). Bars: (A and B) 3 µm; (C) 2 µm. However, clearly Ase1 and the spindle midzone are not the only factors involved in shaping the anaphase nucleus. Further studies will be required to investigate the role of the other components of the nucleus, such as chromatin, and to identify the machineries involved in membrane dynamics, curvature, and cleavage during the highly sophisticated process of nuclear division.
# Materials and methods
Strains, plasmids, and growth conditions All yeast strains were constructed according to standard genetic techniques [bib_ref] Guide to yeast genetics and molecular biology, Guthrie [/bib_ref] and are isogenic to S288C. The yeast strains used in this study and their references are listed in . All experiments were performed with three individual clones, of which one is listed in . All cultures or plates were grown at 30°C unless indicated otherwise.
## Flip and frap experiments
For all FLIP experiments in [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref] , 5, 6, S1, and S2, cells were grown at 30°C on YPD (yeast, peptone, and dextrose) plates, resuspended in synthetic complete medium, and immobilized on a 2% agar pad containing synthetic complete medium. The cells were imaged on a confocal microscope (LSM 510; Carl Zeiss) with a Plan Apochromat 63×/1.4 NA oil immersion objective, typically using 2% of the laser intensity of an argon laser (488-nm line) at 45% output. The ZEN 2010 software (Carl Zeiss) was used to control the microscope. GFP emission was detected with a 505-nm long pass filter. Photobleaching was applied on the selected area as indicated in the figures. Bleaching pulses were iterated 80 times at 60% of the laser intensity at 45% output. All photobleaching experiments were performed at 30°C.
Photobleaching experiments on Ace2-GFP and HTB2-GFP (Figs. 1, C-E; and 8, A and B) were performed on cells grown in liquid culture in the exponential phase for 5 h at 30°C. Cells were harvested by centrifugation, resuspended in synthetic complete medium, and immobilized on a 2% agar pad containing synthetic complete medium. FLIP and FRAP experiments on Ace2-GFP were performed using a confocal microscope (LSM 710; Carl Zeiss) using a Plan Apochromat 63×/1.4 NA oil immersion objective. GFP was excited with the 488-nm line of an argon laser, whereas mCherry was excited with a 561-nm solid-state laser. The signals were detected with a band pass (bp) 505-540-nm (GFP) and a bp 620-660-nm filter (mCherry).
Quantification was performed using ImageJ 1.42q (National Institutes of Health). The mean fluorescence signal was quantified in the entire daughter, the entire mother part of the nucleus, and in the nuclei of five neighboring cells. After background subtraction, the fluorescence signals of the mother and daughter part were normalized to the mean of the five control nuclei and set to 100% at beginning of the experiment. All experiments were pooled and transferred to Prism 5.0b (GraphPad Software), in which they were fitted a one-phase decay curve constraining the starting point to 100%. The °CP was defined as the ratio of the times needed to lose 30% of the fluorescence signal in the bud over the mother. All images shown in the figures were processed using ImageJ 1.42q and Photoshop CS4 extended version 11.0.2 (Adobe).
## Wide-field microscopy
All experiments using wide-field microscopy ; 6, D and E; 8, C-E; S1 D; S2, C-E, G, and H; and S3) were performed on a microscope (DeltaVision; Applied Precision) with a 100×/1.40 NA U Plan S Apochromat oil immersion objective (Olympus), a 250-W Xenon lamp, and a charge-coupled device camera (CoolSNAP HQ 2 ; Photometrics) using softWoRx software (Applied Precision). Noise was reduced in all widefield images acquired using 3D iterative constrained deconvolution by the software softWoRx. Image analysis was performed with ImageJ 1.42q.
Time-lapse microscopy on Ace2-GFP and on nuclear division using Nsg1-GFP was performed with a z spacing of 400 nm and binning of 2. The pause between the time points was 2 min for Ace2-GFP and 5 min for Nsg1-GFP. Although GFP fluorescence was exited with wavelengths from 461 to 489 nm and detected between 525 and 550 nm, mCherry or dsRed were excited at 563-588 nm and detected between 632 and 660 nm.
The z stacks to describe nuclear shape using Nsg1-GFP (Figs. 6, D and E; S2, C-E, G, and H; and S3), Ace2-GFP distribution [fig_ref] Figure 8: Reduction of Ace2 asymmetry in ase1 cells [/fig_ref] , lateral diffusion barriers at the mother-bud junction. Further studies will be needed to investigate whether membrane compartmentalization in the nuclear envelope follows the same rules. If this is the case, it will be interesting to investigate whether compartmentalizing structures are primarily dedicated to other functions, compartmentalization being a secondary consequence, or whether they are primarily dedicated to barrier formation. For example, functionally specializing the nuclear membranes in the connecting bridge of late anaphase nuclei might be more relevant for spindle stability, chromosome segregation, or karyofission rather than compartmentalization. However, the observation that the passive compartmentalization of the nucleoplasm plays an important role in promoting the asymmetry of cell division suggests that nuclear compartmentalization in its own right is an important function of the interconnecting bridge. Also, the observation that the bud6 mutation, which does not appear to affect nuclear morphology, enhanced the exchange between the mother and bud part of the ONM in early, but not in late, anaphase suggests that several independent mechanisms contribute to ONM compartmentalization. Thus, compartmentalization of this membrane appears to be a tightly controlled process.
Altogether our data support the notion that compartmentalization of the nucleus is an important aspect of the process of closed mitosis, most probably because it serves as a support for the asymmetric segregation of nuclear components, such as daughter-specific transcription factors and nonchromosomal DNA [bib_ref] A mechanism for asymmetric segregation of age during yeast budding, Shcheprova [/bib_ref]. A corollary of this conclusion is that the dumbbell shape of the nucleus is not circumstantial but the result of a complex evolution leading to the adaptation of the process of nuclear division to biologically relevant constraints, beyond karyofission.
Accordingly, it is interesting to note that the morphological events associated with nuclear division are highly variable in fungi. For example, the dividing nucleus of the fission yeast Schizosaccharomyces japonicus does not develop the dumbbell morphology typical of S. cerevisiae and Schizosaccharomyces pombe and undergoes semi-open mitosis instead [bib_ref] Breakage of the nuclear envelope by an extending mitotic nucleus occurs during..., Aoki [/bib_ref] [bib_ref] Divergent strategies for controlling the nuclear membrane satisfy geometric constraints during nuclear..., Yam [/bib_ref]. Hence, it is very likely that this organism, which apparently divides highly symmetrically, shows no compartmentalization of the nucleoplasm. It will be interesting to determine whether it does establish barriers in the nuclear envelope. Investigating the biological consequences of the different forms of nuclear division will shed light on the origins and roles of compartmentalization mechanisms.
From that perspective, gaining a molecular understanding of the mechanisms controlling the morphology and compartmentalization of the dividing yeast nucleus is likely to provide also important information about the process of mitosis in a general sense. Little is known about how the nuclear envelope is shaped and how this process is controlled as the spindle elongates and the nucleus acquires its dumbbell morphology. Understanding the contribution of Ase1, Cin8, and Slk19 in this process will provide insights about how microtubules of the spindle midzone interact with the nuclear membrane in fungi as well as how the same structure interacts with the plasma membrane during the cytokinesis of animal cells.
Bleaching region and rate A bleaching area, in the shape of a cigar, was set to be centered at (1.5595, 0.731, 0) in early stage mother cells, and at (1.01, 0.9982, 0) in late stage mother cells. For all simulations, the radial dimensions of such bleaching region in each dimension were b x = b y = 0.14 µm and b z = 0.23 µm.
It is worth noting that experimental variability may result from samples with different bleaching intensities, while assuming uniform diffusion coefficients. The latter can, in principle, account for parts of the variation bounds in experimental measurements, the major part stemming from differences in single cell sizes.
To make accurate fluorescence loss calculations, one needs to confirm a diffusion constant or bleaching rate experimentally. Because we had already determined the diffusion rate of TetR-GFP by FCS [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref] and because TetR-GFP is not compartmentalized in early anaphase, the bleaching rate could be determined from simulations of the FLIP experiments performed with TetR-GFP, presented in [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref] A. The bleaching rate was defined as a degradation rate for molecules entering the bleaching region. During these simulations, 5,000 TetR particles diffusing at a rate of 1.9 µm 2 /s were uniformly distributed within the nucleoplasm. Fluorescence decay was estimated as a decay rate for particles inside the bleaching region, allowing an individual particle to potentially escape. From this, we estimated an optimal degradation rate of k deg = 110/s. Because this number refers to the speed at which particles can be bleached to fit the experimental data, it does not depend on the compartment where the particles diffuse, i.e., the nucleoplasm, INM, or ONM. Thus, it was used uniformly in fluorescence loss simulations of TetR, Nsg1, Nup49, and Src1. Remarkably, preliminary simulations with TetR-GFP also revealed the impossibility to match experimental FLIP data by using the diffusion rates classically assumed for a protein of this size, stressing out the role of nonspecific DNA binding in TetR dynamics.
## Initial conditions
To determine how many particles should be placed in each of the parts comprising a single simulation compartment, we calculated the nucleoplasm volume (for TetR) and the surface area of the INM (for Src1 and Nup49) and ONM (for Nsg1) in both early and late stages of nuclear division. The number of molecules in each section of a compartment was thus scaled according to the volume or the surface area, depending on whether the molecules diffuse in 3D or 2D.
The volume of an ellipsoid is given by 4/3 r x r y r z , and we made a rough approximation of the missing volume from each truncated ellipsoid by calculating the volume of a similar spherical cap. The latter can be calculated as h/6 (3 + h 2 ), in which the thickness of the cap is denoted by h, and the radius of the base of the cap is . The surface area of a prolate ellipsoid with a circular equator, as is our case, can be calculated as 2 (r x 2 + r z 2 (/tan)), in which = arcos(r x /r y ), whereas the surface area of a spherical cap is ( 2 + h 2 ). The volumes and surface areas of spheres and cylinders are straightforward.
In the last step, we estimated diffusion rates of membrane-bound particles by minimizing the error of the first moment, i.e., the mean of the stochastic simulations, over their ensembles. We performed these simulations in equal numbers of the samples from the FLIP experiments to which they were compared. This approach yielded the following parameters in the ONM and INM simulations: The 2,000 Nsg1 particles equally distributed within mother and bud ONM compartments diffused at a rate of 0.3 µm 2 /s, whereas the 150 Nup49-containing particles uniformly distributed within the ONM/INM diffused at a rate of 0.2 µm 2 /s. The 2,000 particles of Src1 equally distributed within the INM mother and bud compartments diffused at the same rate as the Nsg1 particles (0.3 µm 2 /s). Noticeably, in the cases of Nup49 and Src1, optimal diffusion rates were not found to match either the early or late stages of nuclear division. For the latter, slower diffusion coefficients were estimated (0.125 and 0.2 µm 2 /s, respectively). The curvature of the fluorescence loss profiles could, in principle, be caused by time-varying diffusion rates. A thorough analysis could be performed to determine whether fractional or Brownian diffusion is to be expected [bib_ref] Anomalous diffusion and multifractional Brownian motion: simulating molecular crowding and physical obstacles..., Marquez-Lago [/bib_ref].
Online supplemental material shows that TetR-GFP does not exchange between the cytoplasm and nucleoplasm. [fig_ref] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized [/fig_ref] describes the impact of mutants described in [fig_ref] Figure 6: Changes in nuclear shape influence nucleoplasmic compartmentalization [/fig_ref] on nuclear geometry and nuclear compartmentalization in more detail. [fig_ref] Figure 3: Nuclear membranes are compartmentalized during late stages of nuclear division [/fig_ref] shows the nuclear geometries constructed for the simulations. [fig_ref] Figure 4: The bridge connecting the two future nuclei forms the compartments' boundary [/fig_ref] compares the simulation results to the actual FLIP data. [fig_ref] Figure 5: Compartmentalization of the nucleoplasm is specifically decreased in long ase1 nuclei [/fig_ref] differentiates the deviations between the simulations and the experimental mean arising in the mother and bud compartment. contains °CPs with and spindles in late stages of nuclear division [fig_ref] Figure 6: Changes in nuclear shape influence nucleoplasmic compartmentalization [/fig_ref] , D and E) were recorded with a spacing of 200 nm between the individual planes and without binning. To describe thickness of the tube using Nsg1-GFP fluorescence, the total fluorescence per micrometer of bridge length was measured on summed intensity projections of the full stacks measuring the integrated fluorescence over the whole area of the tube; the background subtracted was measured next to the tube inside the cell. Ace2-GFP and dsRed-HDEL were imaged with different filter sets than described in the previous paragraph (GFP: excitation of 461-489 nm and emission of 523-536 nm; dsRed: excitation of 529-556 nm and emission of 594-645 nm). The distribution of Ace2 was determined based on the integrated density of GFP signal measured in both nuclei outlined by dsRed-HDEL. After subtracting the background inside the cell, a ratio of mother signal over bud signal was calculated.
## Fcs measurements
FCS allows the analysis of fluorescence intensity fluctuations caused by fluorophore diffusion through a small (approximately femtoliter) detection volume [bib_ref] Fluorescence correlation spectroscopy. I. Conceptual basis and theory, Elson [/bib_ref] [bib_ref] A dynamic view of cellular processes by in vivo fluorescence auto-and cross-correlation..., Bacia [/bib_ref] [bib_ref] State of the art and novel trends in fluorescence correlation spectroscopy, Petrov [/bib_ref]. In this study, FCS measurements were performed on a microscope (LSM 710 ConfoCor3; Carl Zeiss) with a 40× 1.2 NA UV-visible infrared C Apochromat water immersion objective using the ZEN 2010 software. Cells were grown at 25°C and immobilized between a glass slide and a coverslip. The experiments were performed at 25°C. FCS measurements were performed in the yeast nucleus for 30 s, with excitation by a 488-nm laser with <0.5 µW. The emission signal was detected with a bp 505-540-nm filter. Curves were fitted to a diffusion model describing one diffusing species, including photophysics of the fluorescent protein, with an additional offset accounting for the effects of photobleaching of the fluorophores. The confocal volume aspect ratio parameter was set to seven.
## Definition of model geometry
To determine the dimensions of the early stage simulation domain, we performed measurements on dividing nuclei expressing Nsg1-GFP imaged on a microscope (DeltaVision) with a 100× oil immersion objective using the softWoRx software. In total, 21 nuclei were analyzed denoting: (a) total length of the mother-bud nuclear complex, (b) length of mother end to bud neck, (c) diameter of mother, (d) diameter of ingression at bud neck, and (e) diameter of bud [fig_ref] Figure 3: Nuclear membranes are compartmentalized during late stages of nuclear division [/fig_ref] , assuming the intermembrane space to be 60 nm wide. The geometry was then built based on the mean dimensions of all 21 cells measured [fig_ref] Figure 3: Nuclear membranes are compartmentalized during late stages of nuclear division [/fig_ref]. From these measurements, idealized truncated ellipsoids joined at their edges with distinct radii were determined, in which the size of the angle determining the truncation was calculated.
Similarly, for the late stage, we obtained a mean spherical radius and mean bridge length connecting to the mother and bud nuclei from 34 cells. The corresponding simulation geometry was idealized as two truncated spheres joined by a cylinder. The diameter of the bridge on the level of the ONM and the thickness of the perinuclear space (early and late) were extracted from transmission EM pictures recorded by [bib_ref] Three-dimensional ultrastructural analysis of the Saccharomyces cerevisiae mitotic spindle, Winey [/bib_ref].
An ellipsoid can be parametrically described as: x = r x cos(u) sin(v) + x 0 , y = r y sin(u) sin(v) + y 0 , and z = r x cos(v) + z 0 , in which u e [0,2), v e [0,), x 0 , y 0 , and z 0 are the coordinates of its center, and r x , r y , and r z are the axis lengths in the x, y, and z directions. In a truncated ellipsoid, the angle v will naturally have a different lower or upper value, and in a sphere, the axis lengths are r x = r y = r z . Accordingly, the model geometries for the simulation of early and late stages of nuclear division were set as described in [fig_ref] Figure 3: Nuclear membranes are compartmentalized during late stages of nuclear division [/fig_ref] , where it should be noted, the radial distances in the z axis are identical to those in the y axis. Once the compartments were defined parametrically, Delaunay triangulations were performed, yielding the final discretization of the simulation domains.
## Particle numbers for simulations
To determine the number of Nsg1-GFP and GFP-Src1 molecules, we relied on the quantification data from the Yeast GFP Fusion Localization Database [bib_ref] Global analysis of protein localization in budding yeast, Huh [/bib_ref] because we needed information about the relative amounts rather than the exact numbers of molecules per cell. Both proteins were present in 2,000 copies per cell (2,100 for Src1 and 1,900 for Nsg1). Based on the total amount of TetR-GFP fluorescence, we estimated 5,000 copies of TetR-GFP per cell. In the case of Nup49-GFP, simulations were run with 150 particles of 16 molecules each per nucleus, which reflects the number of NPCs with the according number of molecules per complex [bib_ref] Nuclear pore complex number and distribution throughout the Saccharomyces cerevisiae cell cycle..., Winey [/bib_ref] [bib_ref] The molecular architecture of the nuclear pore complex, Alber [/bib_ref].
[fig] Figure 2: The nucleoplasm of a dumbbell-shaped nucleus is compartmentalized. (A-D) FLIP experiments on indicated reporter proteins during early and late stages of nuclear division. [/fig]
[fig] Figure 3: Nuclear membranes are compartmentalized during late stages of nuclear division. (A-C) FLIP experiments on the indicated markers of the nuclear envelope during early and late anaphase. Graphs are as in [/fig]
[fig] Figure 4: The bridge connecting the two future nuclei forms the compartments' boundary. (A) Possible scenarios for compartment boundaries. (B) FLIP experiments on GFP-Src1, Nsg1-GFP, or TetR-GFP during late stages of nuclear division. Graphs are as inFig. 2. Fluorescence levels for characteristic cells over time in the mother (red) and daughter part (green) are shown; bleaching area is indicated in blue. White lines indicate cell outlines. (C) °CP 1 (orange) and °CP ratios (blue) of GFP-Src1, Nsg1-GFP, and TetR-GFP plotted against the position of the bleaching region on the nuclear length axis (n = 5 per position; means ± SD). Dashed lines left to right: start of the bridge in mother, bud neck, and end of the bridge in the bud. Lines show section of the illustration on the top right, showing a cell in late anaphase with the outline of the cell and the nuclear envelope in black, the nuclear part of the spindle in green, and the spindle midzone in red. Bars, 3 µm. [/fig]
[fig] Figure 5: Compartmentalization of the nucleoplasm is specifically decreased in long ase1 nuclei. (A) FLIP experiments on the indicated reporters in 6-7-µm-long ase1 nuclei. Graphs are as inFig. 2. Mean fluorescence levels ± SD over time in the mother (red) and daughter part (green) are shown; bleaching area is indicated in blue. White lines indicate cell outlines. (B) °CP values in wild type (WT) and ase1 (for n see [/fig]
[fig] Figure 6: Changes in nuclear shape influence nucleoplasmic compartmentalization. (A) °CP of TetR-GFP, Nsg1-GFP, or GFP-Src1 in individual cells over bridge length; wild type (black) and ase1 (red). (B) Nsg1-GFP outlining nuclear shape in cells of the indicated genotypes. (C) °CP (for n see [/fig]
[fig] Figure 7: Averaged deviations of stochastic model simulations from the experimental mean. Deviations (in percentages) between experimental and simulated data averaged over mother and bud compartments for each experimental time step. (A-F) Simulations for early (A-C) and late (D-F) stages of nuclear division, with TetR-GFP (A and D), Nsg1-GFP (B and E), and GFP-Src1 (C and F). Color code indicates different diffusion barrier permeability. Bold lines show best overlap with experimental data; dashed lines show simulations with larger deviations. [/fig]
[fig] Figure 8: Reduction of Ace2 asymmetry in ase1 cells. (A) FRAP on Ace2(F127V)-GFP in cells with intact spindles. Graphs are as in [/fig]
|
Factors Associated With Adoption of Health Information Technology: A Conceptual Model Based on a Systematic Review
Background:The Health Information Technology for Economic and Clinical Health Act (HITECH) allocated $19.2 billion to incentivize adoption of the electronic health record (EHR). Since 2009, Meaningful Use Criteria have dominated information technology (IT) strategy. Health care organizations have struggled to meet expectations and avoid penalties to reimbursements from the Center for Medicare and Medicaid Services (CMS). Organizational theories attempt to explain factors that influence organizational change, and many theories address changes in organizational strategy. However, due to the complexities of the health care industry, existing organizational theories fall short of demonstrating association with significant health care IT implementations. There is no organizational theory for health care that identifies, groups, and analyzes both internal and external factors of influence for large health care IT implementations like adoption of the EHR.Objective: The purpose of this systematic review is to identify a full-spectrum of both internal organizational and external environmental factors associated with the adoption of health information technology (HIT), specifically the EHR. The result is a conceptual model that is commensurate with the complexity of with the health care sector.Methods:We performed a systematic literature search in PubMed (restricted to English), EBSCO Host, and Google Scholar for both empirical studies and theory-based writing from 1993-2013 that demonstrated association between influential factors and three modes of HIT: EHR, electronic medical record (EMR), and computerized provider order entry (CPOE). We also looked at published books on organizational theories. We made notes and noted trends on adoption factors. These factors were grouped as adoption factors associated with various versions of EHR adoption.Results: The resulting conceptual model summarizes the diversity of independent variables (IVs) and dependent variables (DVs) used in articles, editorials, books, as well as quantitative and qualitative studies (n=83). As of 2009, only 16.30% (815/4999) of nonfederal, acute-care hospitals had adopted a fully interoperable EHR. From the 83 articles reviewed in this study, 16/83 (19%) identified internal organizational factors and 9/83 (11%) identified external environmental factors associated with adoption of the EHR, EMR, or CPOE. The conceptual model for EHR adoption associates each variable with the work that identified it.Conclusions:Commonalities exist in the literature for internal organizational and external environmental factors associated with the adoption of the EHR and/or CPOE. The conceptual model for EHR adoption associates internal and external factors, specific to the health care industry, associated with adoption of the EHR. It becomes apparent that these factors have some level of association, but the association is not consistently calculated individually or in combination. To better understand effective JMIR Med Inform 2014 | vol. 2 | iss. 1 | e9 | p. 1 http://medinform.jmir.org/2014/1/e9/ (page number not for citation purposes) Kruse et al JMIR MEDICAL INFORMATICS XSL - FO RenderX adoption strategies, empirical studies should be performed from this conceptual model to quantify the positive or negative effect of each factor. (JMIR Med Inform 2014;2(1):e9) doi: 10.2196/medinform.3106KEYWORDS electronic health record (EHR); electronic medical record (EMR); health information technology (HIT); medical information systems; computerized provider order entry (CPOE); adoption
# Introduction background
The US Government passed the Health Information Technology for Economic and Clinical Health (HITECH) actto incentivize adoption of the electronic health record (EHR) and to assuage the short run (SR) effects of cost to the health care organization in the adoption process. The three phases of Meaningful Use consume information technology (IT) strategies in the SR because of the HITECH act's timeline for health care organizations to qualify for monetary incentives [bib_ref] Electronic health record functions differ between best and worst hospitals, Elnahal [/bib_ref] [bib_ref] Exploring the business case for ambulatory electronic health record system adoption, Song [/bib_ref].
Adoption of the EHR is a significant goal. International vernacular for the EHR varies; for example, electronic patient record, computerized patient records, electronic medical records (EMRs), and digital medical record. The defining difference, as defined by the Institute of Medicine, the health arm of the US National Academy of Sciences, focuses on the longitudinal and interoperable nature of the electronic patient record. Without these capabilities, the patient record is greatly limited in scope. The longitudinal and interoperable nuances of the EHR are not the only significant advantages; there are eventual cost savings as well.
Studies estimate that adoption of the EHR could eventually save more than $813 billion annually, prevent 200,000 adverse drug events, and enhance the doctor-patient relationship through increased communication [bib_ref] Can electronic medical record systems transform health care? Potential health benefits, savings,..., Hillestad [/bib_ref]. Unfortunately, these benefits are realized in the long run (LR), while the investment to adopt the EHR is expended in the SR. A large deficit in the SR could inhibit a health care organization's ability to compete or survive in heavily competitive environment.
The environment of health care is unique in a competitive environment. The health care organization develops an organizational strategy based on the local environment. To increase an organization's ability to compete, its strategy might also include cost reduction, and EHR adoption runs counter to this goal in the SR. The health care environment faces many sources of influence, including a reluctance to accept technology.
There has been a tremendous amount of research dedicated to the study of acceptance of technology, specifically the Technology Acceptance Model (TAM) [bib_ref] Perceived usefulness, perceived ease of use, and user acceptance of information technology, Davis [/bib_ref]. More recent work has suggested modifications to the TAM that explain a perception of usefulness and intentions from the aspect of social influence and the cognitive instrumental process [bib_ref] A theoretical extension of the Technology Acceptance Model: four longitudinal field studies, Venkatesh [/bib_ref] [bib_ref] User acceptance of information technology: toward a unified view, Venkatesh [/bib_ref]. Several organizational theories have been developed. These focus on the sources of influence and the reason for their existence.
## Organizational theories
Organizational theories address influence, but none adequately addresses the complexity of the health care organization. Payers, providers, and patients all control resources that exert influence. The nature of the competitive environment will also exert influence on decisions. External influence from those who control resources can be explained through resource dependence theory [bib_ref] Environments of organizations, Aldrich [/bib_ref]. Internal and external influences can be explained by the Diffusion of Innovation Theory through its introduction of compatibility, complexity, trialability, observability, and relative advantage [bib_ref] Diffusion of innovations, Rogers [/bib_ref].
According to resource dependence theory, health care organizations with the greatest level of dependence on other organizations that control the resources will feel the greatest level of environmental influence on its decisions. The Resource Dependence Theory describes an external interdependence of organizations. External Control of Organizations,, which is an adaptation of Resource Dependence Theory, provides good insight for this study. The authors' premise is that the external environment creates a social context and plays an important role in how organizational decisions are made. The lack of absolute independence requires some degree of interorganizational exchange of goods or services. As organizations build and negotiate relationships with each other in the exchange of resources, positions of power are established. No one organization can provide all of its own resources, so each organization becomes dependent on the other organizations that control the resources.
Similar to Resource Dependence, the Diffusion of Innovation Theory describes a social system that influences through communication channels [bib_ref] Diffusion of innovations, Rogers [/bib_ref]. Diffusion of Innovation attempts to explain how "an innovation, is communicated through channels over time among members of a social system" [bib_ref] Diffusion of innovations, Rogers [/bib_ref]. This theory accounts for 49-97% of variance in the rate of adoption of innovation through five factors: compatibility, complexity, trialability, observability, and relative advantage. These factors are sorted into three categories of a predictive model for EHR adoption: innovation determinants, organizational determinants, and environmental determinants [bib_ref] User acceptance of information technology: toward a unified view, Venkatesh [/bib_ref]. The next several paragraphs exercise the five factors to this study.
The concept of compatibility [bib_ref] Diffusion of innovations, Rogers [/bib_ref] goes beyond answering the question, "is a product/service right for a market?" It also asks, "is the market ready for the product/service?" For instance, the Chevy Nova failed in Spanish-speaking markets because in Spanish the word "Nova" means "does not go." Promotion of conservation techniques to farmers in the United States initially failed because farmers associated conservation with lower crop yield. Boiling water to sanitize it makes perfect sense to a market that is familiar with germ theory, but primitive tribes in Peru only heated water for sicker, weaker members; as a result, the concept failed when initially introduced and dysentery continued to flourish. In relation to this study, the concept of compatibility might ask, "is the market ready for the EHR?"
The concept of complexity [bib_ref] Diffusion of innovations, Rogers [/bib_ref] is appropriate to this study because innovation can be a double-edged sword. On one hand, it is new and may offer some improvement to a product or service. However, it might also be perceived as too complex; and perception can be a powerful force. If the Baby Boomer generation perceives computers to be too complex, and this perception causes computer anxiety, its users may reject its adoption and use [bib_ref] Factors predicting the use of technology, Czaja [/bib_ref]. The older physicians in a hospital have greater seniority, and are therefore, more influential in the hospital's decision to adopt the EHR. Would this same generation of providers influence the health care organization considering EHR adoption?
The concept of trialability [bib_ref] Diffusion of innovations, Rogers [/bib_ref] applies more to the early-adopter group than the other groups: innovators, early-majority, late-majority, and laggards. In the early phase of promotion for a new product or service, the vendor might lower the risk of adoption by offering free trials or samples to potential users. Once the user is confident of the new item's efficacy, then he/she is more likely to pay full price for its use. When a new producer of an EHR enters the marketplace, it must incentivize the use of its product because it is not known in the industry. The user accepts a risk by trying the new EHR, but the risk is overcome by the incentive. Once the new EHR gains momentum in the industry, adoption enters the early-majority phase. The new EHR has already gained momentum in the industry, and the producer does not need to incentivize its use.
The concept of observability [bib_ref] Diffusion of innovations, Rogers [/bib_ref] is also highly applicable to this study. Decision makers in a hospital that has not yet adopted an EHR will observe the experiences of other hospitals that have adopted it. Vendors will promote or advertise specifically to the nonadopters and help them observe how the EHR can benefit its organization. External players in the health care organization's competitive environment will provide some level of observability.
Relative advantage is a multifaceted concept for this study. In health care, the most important factor is provision of health, as well as the treatment and prevention of disease. If adoption of the EHR speaks directly to the health care organization's primary purpose, then it might provide relative advantage over competitors that have not adopted it. Another concept is that of social prestige [bib_ref] Diffusion of innovations, Rogers [/bib_ref]. Unless a health care organization can serve as an example to other health care organizations (observability), there may not be a sufficient level of relative advantage to be considered.
## Strategy and decision making
Strategy can be a multifaceted concept, and organizations around the world hire strategy experts to help identify and focus on a market forces. An operational definition of strategy is borrowed from education [bib_ref] Patterns of strategies in Swiss higher education institutions, Fumasoli [/bib_ref] and is adapted to health care: strategy is defined as instruments by which health care organizations manage their organizational processes and deal with their environments in order to select a portfolio of activities and find appropriate position in the health care industry (italics indicate a change in wording from the authors' definition). It follows that adoption of an EHR would alter how a health care organization manages its organizational processes, so this definition of strategy is a good fit for the health care industry. However, two significant considerations in the health care environment are the level of local competiveness, and how health care organizations compete.
Studies have shown that decision making in the health care industry is often based on how the organization competes, whether in a single-market or multimarket environment [bib_ref] The efficiency of hospital-based clusters: evaluating system performance using data envelopment analysis, Sikka [/bib_ref]. In either environment, decision-making varies on competition, and the health care industry competes in clusters [bib_ref] The efficiency of hospital-based clusters: evaluating system performance using data envelopment analysis, Sikka [/bib_ref]. The way health care organizations compete will also affect its organizational structure. A four-cluster solution was identified as a reliable, internally valid, and stable model for health networks and a five-cluster solution for health systems [bib_ref] A taxonomy of health networks and systems: bringing order out of chaos, Bazzoli [/bib_ref]. Differentiation and centralization are particularly important in distinguishing unique clusters of organizations. High differentiation typically occurs with low centralization, which suggests that a broader scope of activity is more difficult to centrally coordinate. Integration is also important, but the authors find that health networks and systems typically engage in both ownership-based and contractual-based integration or they are not integrated at all.
Ash and Bates [bib_ref] Factors and forces affecting EHR system adoption: report of a 2004 ACMI..., Ash [/bib_ref] studied the EHR adoption rates and the factors and forces affecting system adoption through surveys (85/650, 13.1%). Only 106 of the 650 (16.3%) of hospitals surveyed had adopted some form of EHR, 63/106 (59.4%) had implemented a full Computerized Provider Order Entry (CPOE) solution, and the other 43/106 (40.6%) implemented a partial CPOE solution. A full one-third of adopters were either Veterans Affairs or military hospitals. Additionally, 481/650 (73.8%) of those who planned to implement a full solution intended to do so within 5 years. Ash and Bates [bib_ref] Factors and forces affecting EHR system adoption: report of a 2004 ACMI..., Ash [/bib_ref] also found that the size of hospital is positively-associated with component adoption; specifically CPOE adoption. The authors inferred from their results that the primary reasons to adopt the EHR is to gain the quality-of-care advantages of CPOE. This inference reinforced our inclusion of CPOE as a dependent variable.
Factors that influence health information system (HIS) adoption in US hospitals have been studied by others as well (n=1441) [bib_ref] Factors influencing health information system adoption in American hospitals, Wang [/bib_ref]. Results showed that HIS adoption is influenced by the hospital market, organizational, and financial factors. Larger, system-affiliated, and for-profit hospitals with more preferred provider organization contracts are more likely to adopt managerial information systems than other hospitals. Operating revenue is positively associated with HIS adoption. The study also identified hostility as an aspect of environmental uncertainty, and that organizations often turn to technological adoption to regain competitive advantage.
A knowledge-based taxonomy of critical factors for adopting an EHR was developed from a systematic literature review [bib_ref] A knowledge-based taxonomy of critical factors for adopting electronic health record systems..., Castillo [/bib_ref]. The researchers selected 68 of 3400 (2.00%) articles to identify six factors of adoption, listed in order of importance: user attitude toward information systems, workflow impact, interoperability, technical support, communication among users, and expert support.
Alternative measures of EHR adoption among hospitals have been studied [bib_ref] Alternative measures of electronic health record adoption among hospitals, Blavin [/bib_ref]. Authors analyzed a 2009 information technology supplement survey distributed by the American Hospital Association (AHA). The survey focused on 24 EHR functionalities in various areas: electronic clinical documentation, results viewing, CPOE, and clinical decision support. They found that 142 of 3937 (3.60%) acute-care hospitals in the United States of responding hospitals have implemented all 24 functions, 386/3937 (9.80%) of hospitals have implemented at least 20 functions, and 1437/3937 (36.50%) have implemented at least one-half of the functions. The researchers added that EHR adoption is a complex process.
Others have studied the relationship between hospital financial position and the adoption of the EHR [bib_ref] Hospital financial position and the adoption of electronic health records, Ginn [/bib_ref]. Through a cross-sectional study of secondary data from several sources, including the AHA (2442/5752, 42.51% acute-care hospitals in the United States), researchers identified five independent and one dependent variable. Of the five independent variables (IVs), only liquidity was positively-associated with EHR. Asset turnover was negatively-associated with EHR adoption. Bed size, a control variable, was positively-associated with EHR adoption. The authors concluded that hospitals adopt EHRs as a strategic move to better align themselves with their environment.
Because commonly used elements of organizational strategy are difficult to change, several of the variables were categorized as internal organizational factors. Research has assessed variables of hospital influence in five categories: (1) capacity as measured by number of beds in groupings by intervals of 100, (2) management, or ownership, (3) organizational focus, or teaching status, (4) competitive location and alternatives, and (5) state regulatory pressures [bib_ref] Case-mix specialization in the market for hospital services, Farley [/bib_ref].
Although resources have been consumed to study factors associated with adoption of HIT, there is a gap in the literature that provides a conceptual model to guide the design of empirical models. It may seem backward to design a conceptual model after so many studies have already been conducted, but the gap remains. The aim of this study was to develop a conceptual model from a systematic literature review that associates both internal and external factors associated with adoption of the EHR. The intent of the conceptual model is to enable future empirical models.
# Methods
## Literature review process
Search terms were selected based on the experience of the authors in the field of health care administration. The time frame of 1993-2013 was selected as convenience. It was assumed that 2 decades would be sufficient to capture trends. [fig_ref] Figure 1: Literature review process [/fig_ref] illustrates the literature review process that identified 83 sources consisting of empirical studies, articles, editorials, commentaries, opinion papers, organizational theories, and text books. The intent of no limits to the type of papers was to mitigate the risk of missing something significant from a study that was not catalogued properly within a key word catalogue like the Dublin Core.
The 83 records were reviewed for content and evidence. After discarding 58 articles for lack of evidence, three additional references were added because they were key concepts upon which other studies were based. Of the remaining 25 articles, a list of factors was identified as IVs. Some factors were grouped under a similar category for the purposes of simplification of the conceptual model. The dependent variable (DV) started as adoption of the EHR, but the studies from those chosen were not as specific. From personal experience, many studies seem to discuss the EHR, but call it something else: most commonly the EMR. That is why EMR was included in the search. Because so few ERHs exist without some form of CPOE, the latter term was included in the search criteria.
Our study combines the influences highlighted by previous work and examines determinants of EHR adoption. Examining EHR adoption at the health care organization level will demonstrate validity between this study and others that have used the hospital as the unit of analysis.
## Ehr adoption and internal organizational influence
Several influences in the environment exert pressure on the health care organization to adopt EHR. Influences range from incentives from the federal government to the nature of local competitive community. US federal incentives provide a heavy influence for EHR implementation, under specific conditions, and imposes penalties for a lack of EHR implementation.
The internal politics of one organization serve as one source of influence. A hospital is part of a community, which serves as an external influence. Further, if a hospital is also part of a larger multihospital system (MHS), then the politics of the broad MHS will also exert influence on local decisions.
## Ehr adoption and external environmental influence
The patient is external to the organization, and for our study, the patient primarily serves as an external influence. Although some employees of the health care organization might also be patients, and this relationship could create a small internal influence, this study considers those few stake holders in the internal organizational factor of users. The providers are internal to the organization, and for our study, providers serve as an internal organizational influence. The payer is a significant influence, and the Center for Medicare and Medicaid Services (CMS) serves as a good example of this significant influence. The HITECH act provides monetary incentives for EHR adoption. Those who do not implement all aspects specified in the stages of adoption are not eligible for the incentives. In this way, the CMS disincentivizes those organizations that do not adopt the EHR. If payments from the CMS were of little consequence to the health care organization's revenue, then the health care organization might decide differently about EHR adoption. A competing health care organization is an external market force in the environment. Third-party payers might compare health care organizations based on maturity of automation because mature clinical components like CPOE will result in more accurate billing.
Such forces incentivize a health care organization to adopt the EHR.
## Overview of the conceptual model
The premise for an EHR adoption conceptual model is that that environmental influences affect organizational strategy of the health care organizations that adopt the EHR [bib_ref] Diffusion of innovations, Rogers [/bib_ref] [bib_ref] Factors and forces affecting EHR system adoption: report of a 2004 ACMI..., Ash [/bib_ref] [bib_ref] A knowledge-based taxonomy of critical factors for adopting electronic health record systems..., Castillo [/bib_ref] [bib_ref] Hospital financial position and the adoption of electronic health records, Ginn [/bib_ref] [bib_ref] Case-mix specialization in the market for hospital services, Farley [/bib_ref]. Diffusion of Innovation theory provides three categories of a predictive model for EHR adoption: innovation determinants, organizational determinants, and environmental determinants [bib_ref] Diffusion of innovations, Rogers [/bib_ref]. Resource Dependence Theory provides a category of a predictive model for EHR adoption, the competitive environment. In construction of the EHR adoption conceptual model, several constructs emerged.
Elements of organizational strategy are not variables that can be easily changed [bib_ref] A taxonomy of health networks and systems: bringing order out of chaos, Bazzoli [/bib_ref] ; therefore, elements typically ascribed to strategy, such as size, ownership, and fiscal stability, will be absorbed into the IVs of influence. This research proposes a model, whereby environmental factors are associated with an organization's decision to adopt the EHR.
Resource Dependence Theory explains environmental influences and the external interdependence of organizations. The authors' premise is that the external environment creates a social context and plays an important role in how organizational decisions are made. The interdependence of organizations widens the field of stakeholders, and this relationship effect should be defined.
Disparate stakeholders have different interests with reference to different components of the EHR. These interests may be different in the SR interests versus the LR interests. SR interests are those that are immediate, such as current year expenditures. LR interests are further out when all inputs are variable. The SR interests of cost can often compete with the LR potential of cost savings and greater safety. Both the SR and LR interests are affected by the external environment.
In a highly competitive environment, SR cost implications could often win over any long-term savings. The number of patients in a market is fixed in the SR, and a highly competitive market will affect each competitor's share of that market. The SR costs of EHR implementation might be insurmountable by an organization in this market because it could not afford to lose ground without significant capital reserves or the ability to borrow cheaply. However, in a less competitive market, the LR interests of potential cost savings have a better chance of influencing the decision to implement an EHR because the costs incurred in the SR are justified by the long-term benefits.
External stakeholders that control resources important to the health care organization can exert significant influence. For instance, a health care organization that receives a significant amount of revenue from the CMS will be influenced more by incentives provided by the CMS than an organization that receives a significant cash flow from private third parties. The relative influence of various external stakeholders may be captured by an analysis of the structure of the market in which a health care organization operates.
Stakeholders have varying interests with regard to the capabilities and effects of EHR components depending upon their relationship with the health care organization. Private payers have both SR and LR interests in the EHR. In the SR, their focus is on minimizing expenditures. Because the health care organization would pass on the implementation costs through higher contract costs, payers would not be equal in the SR. In addition, the disruption of EHR implementation could potentially affect care processes and therefore increase claims. Payers would be interested in the LR benefits of the EHR: potential cost savings, better disease management, and increased safety. However, the SR interests of the private payers might overshadow the LR benefits of the EHR. Public payers enable care of the indigent and elderly. As part of the United States Department of Health and Human Services (HHS), the CMS is highly interested in disease management, public health, safety, and research, and it may value these LR capabilities of the EHR more than the SR costs. The CMS, as part of HHS, would also favor the EHR because it supports the Presidential directive to promote the establishment of the Nationwide Health Information Network that links electronic patient records through health information exchanges.
Providers and patients value face time with each other. During EHR implementation, providers might spend less time in communication with patients. Providers must adapt their processes and clinic-to-administrative schedules. Any disruption or action that is perceived as deleterious to this relationship could result in a negative reaction to EHR implementation. As a result, physicians might oppose EHR adoption, or they might simply support the EHR solution with the shortest implementation time or least administrative burden. Patients might not like the reduced face time with the provider, but they might be attracted to EHR components such as e-prescribing, e-results, personal health records, and email access to the provider. These desirable features are available to the patient when the health care organization chooses to adopt various portions of the CPOE component to the EHR.
# Results
## Chosen articles
The articles chosen for final inclusion were read once more to make a list of variables. The variables from the studies were listed as internal and external. There were significant commonalities in the variables used, so they were combined in the model.
## Ehr adoption and internal organizational influence
As depicted in [fig_ref] Figure 1: Literature review process [/fig_ref] , 16 references identified internal factors [bib_ref] A theoretical extension of the Technology Acceptance Model: four longitudinal field studies, Venkatesh [/bib_ref] [bib_ref] User acceptance of information technology: toward a unified view, Venkatesh [/bib_ref] [bib_ref] Factors predicting the use of technology, Czaja [/bib_ref] [bib_ref] A taxonomy of health networks and systems: bringing order out of chaos, Bazzoli [/bib_ref] [bib_ref] Factors influencing health information system adoption in American hospitals, Wang [/bib_ref] [bib_ref] A knowledge-based taxonomy of critical factors for adopting electronic health record systems..., Castillo [/bib_ref] [bib_ref] Alternative measures of electronic health record adoption among hospitals, Blavin [/bib_ref] [bib_ref] Hospital financial position and the adoption of electronic health records, Ginn [/bib_ref] [bib_ref] The financial performance of hospitals belonging to health networks and systems, Bazzoli [/bib_ref] [bib_ref] Electronic health records' limited successes suggest more targeted uses, Desroches [/bib_ref] [bib_ref] Outcomes of computerized physician order entry in an electronic health record after..., Harshberger [/bib_ref] [bib_ref] Physicians in nonprimary care and small practices and those age 55 and..., Decker [/bib_ref] [bib_ref] Hospitals ineligible for federal meaningful-use incentives have dismally low rates of adoption..., Wolf [/bib_ref] [bib_ref] Healthcare professionals' organisational barriers to health information technologies-a literature review, Lluch [/bib_ref] [bib_ref] Factors associated with adoption of the electronic health record system among primary..., Cheung [/bib_ref]. Six identified size of the health care organization, and six identified strategic alliances. Five identified ownership and five identified complexity of care. Four identified capital expenditures. Three identified users, and three identified teaching status. Two identified user attitude toward HIS, and two identified communication among users. Workflow impact, interoperability, technical support, expert support, physician arrangements, unity of effort, and user computer anxiety were all identified by one study, independently.
The dependent variable was not consistent: seven references used EHR adoption [bib_ref] Alternative measures of electronic health record adoption among hospitals, Blavin [/bib_ref] [bib_ref] Hospital financial position and the adoption of electronic health records, Ginn [/bib_ref] [bib_ref] Case-mix specialization in the market for hospital services, Farley [/bib_ref] [bib_ref] Electronic health records' limited successes suggest more targeted uses, Desroches [/bib_ref] [bib_ref] Physicians in nonprimary care and small practices and those age 55 and..., Decker [/bib_ref] [bib_ref] Hospitals ineligible for federal meaningful-use incentives have dismally low rates of adoption..., Wolf [/bib_ref] [bib_ref] Factors associated with adoption of the electronic health record system among primary..., Cheung [/bib_ref] , two used "electronic capture of clinical data," [bib_ref] Alternative measures of electronic health record adoption among hospitals, Blavin [/bib_ref] [bib_ref] Case-mix specialization in the market for hospital services, Farley [/bib_ref] one used a generic DV of technology adoption [bib_ref] Factors predicting the use of technology, Czaja [/bib_ref] , and six used CPOE [bib_ref] Factors and forces affecting EHR system adoption: report of a 2004 ACMI..., Ash [/bib_ref] [bib_ref] Alternative measures of electronic health record adoption among hospitals, Blavin [/bib_ref] [bib_ref] Case-mix specialization in the market for hospital services, Farley [/bib_ref] [bib_ref] Electronic health records' limited successes suggest more targeted uses, Desroches [/bib_ref] [bib_ref] Outcomes of computerized physician order entry in an electronic health record after..., Harshberger [/bib_ref] [bib_ref] Hospitals ineligible for federal meaningful-use incentives have dismally low rates of adoption..., Wolf [/bib_ref].
## Ehr adoption and external environmental influence
As depicted in [fig_ref] Figure 1: Literature review process [/fig_ref] , nine studies identified external environmental factors [bib_ref] Factors predicting the use of technology, Czaja [/bib_ref] [bib_ref] A taxonomy of health networks and systems: bringing order out of chaos, Bazzoli [/bib_ref] [bib_ref] A knowledge-based taxonomy of critical factors for adopting electronic health record systems..., Castillo [/bib_ref] [bib_ref] Hospital financial position and the adoption of electronic health records, Ginn [/bib_ref] [bib_ref] Electronic health records' limited successes suggest more targeted uses, Desroches [/bib_ref] [bib_ref] Physicians in nonprimary care and small practices and those age 55 and..., Decker [/bib_ref] [bib_ref] Hospitals ineligible for federal meaningful-use incentives have dismally low rates of adoption..., Wolf [/bib_ref]. Five studies identified buyers, four studies identified patients, three studies identified competitiveness, two studies identified location, and one identified interdependence factors external to the organization that are associated with adoption of the EHR.
## Overview of the conceptual model
As previously stated, there was overlap between the sources/theories. There were four internal forces and seven external forces identified through multiple works by three authors [bib_ref] Diffusion of innovations, Rogers [/bib_ref] [bib_ref] A knowledge-based taxonomy of critical factors for adopting electronic health record systems..., Castillo [/bib_ref]. However, it was unclear in existing literature the degree to which these forces can influence a health care organization's decision to adopt the EHR. A complex conceptual model should provide insight into the strength and direction of the influence on the complex health care organization. The resulting conceptual model, depicted in [fig_ref] Figure 2: Conceptual model of factors associated with adoption of the EHR [/fig_ref] , posits a complex relationship between environmental influences, organizational strategy, and EHR adoption.
This framework captures both internal and external factors that influence the adoption of the EHR. The positive (+) and negative (−) signs in the model describe the relationship identified by the associated authors. For instance, Gin et al [bib_ref] Hospital financial position and the adoption of electronic health records, Ginn [/bib_ref] identified a positive relationship between the external environmental factors of public payer (IV) and competitiveness (IV) and an association with the adoption of an EHR (DV). That is to say, the greater the percentages of an organization's reimbursements that come from a public source like CMS, the stronger the association of the organization's adoption of the EHR. Likewise, the greater the Herfindahl Index of the local competitive environment, the stronger the association of the organization's adoption of the EHR. Age is another interesting factor because of its negative relationship with adoption. The older the patient population (external environmental IV) [bib_ref] Factors predicting the use of technology, Czaja [/bib_ref] and provider population (internal organizational IV) [bib_ref] Physicians in nonprimary care and small practices and those age 55 and..., Decker [/bib_ref] , the lower association with the adoption of the EHR (DV). The + and − signs above the arrows between the IVs and DVs indicates the variety of positive and negative associations with the adoption of an EHR.
# Discussion
## Principal findings
The main findings of this study were that nine studies identified external factors and 16 studies identified internal factors associated with the adoption of EHR. These factors were depicted in a conceptual model to describe relationships to EHR adoption.
The conceptual model for EHR adoption illustrates a framework within which both administrators and policy makers can work to understand the levers that exert significant influence in the adoption of EHR. The extensive literature review conducted by this study builds a unique model from which empirical studies can be designed.
Identifying relationships between the adoption factors and adoption of the EHR becomes significant because it identifies levers that will produce a desired action. For instance, if a hospital has a majority of senior providers, perhaps from the Baby Boom generation, the administrators become aware of the additional effort that needs to go into user acceptance. A hospital that has a majority of new providers will not need to expend the resources on user acceptance, because studies already show a penchant for technology in younger generations. Similar inferences on other factors of adoption could be made, and some would require additional study.
For instance, the literature on workflow impact is split. There seems to be evidence that the presence of the EHR both enhances and complicates the providers' workflow. This observation clearly begs additional questions. Were subjects for the data in different phases of adoption of the EHR? Was the hospital that responded negatively in the middle of an EHR implementation? Logically, a large implementation of any technology will become disruptive to the organization. Several studies could emerge from this relationship alone.
Empirical models could easily be designed to further investigate specific relationships between the IVs and DVs. The set of studies on CPOE was interesting. Although there were some overlaps with adoption of the EHR, there were also studies that only looked at CPOE. There does not seem to be an abundance of evidence in the literature about CPOE, and yet the AHA regularly collects data on six different versions of CPOE: laboratory, radiology, pharmacy, nursing, physician notes, and consults. There was no data to be found about the use or efficacy on CPOE consults. It might be interesting to determine the reason for this paucity of data.
# Limitations
The EHR adoption conceptual model associates internal and external factors with the adoption of the EHR, but it is primarily based on an extensive literature review. So far, it is not empirically tested. However, data are available to test the theory. Because the findings of our study are descriptive in nature, we do not opine on appropriate medical use of the information.
Caution should be identified with the interaction of variables. Some variables will most likely confound or mask the effects of others. For instance, is there a direct relationship between the number of beds of a hospital and the number of full-time equivalents? There are staffing models that would most likely answer that question. If there is a similar relationship, then one of these variables should be eliminated in favor of the stronger one. Otherwise, the effects of the weaker variable will be masked by the other. A false conclusion could easily be identified concerning the masked variable.
A majority of references for this study were from the United States, with one exception from Hong Kong. The internal validity of this study is strongest within the US health care sector. The conceptual model might be limited outside the United States because of the nature of competition between hospitals. The analysis of 83 articles identified studies that used similar methods: survey or secondary data analysis. Many authors analyzed data from the AHA, a well-established data set in the United States. These data are self-reported, which comes with limited bias.
# Conclusions
This study also identified overlap between studies in terms of variables. One interpretation of this overlap could be that the variables and associated studies are highly reliable. The key word/phrase searches described in the Methods section identifies the databases queried and results given. Other researchers should be able to duplicate or update this conceptual model going forward.
[fig] Figure 1: Literature review process. [/fig]
[fig] Figure 2: Conceptual model of factors associated with adoption of the EHR. [/fig]
|
Review and Outcome of Prolonged Cardiopulmonary Resuscitation
The maximal duration of cardiopulmonary resuscitation (CPR) is unknown. We report a case of prolonged CPR. We have then reviewed all published cases with CPR duration equal to or more than 20 minutes. The objective was to determine the survival rate, the neurological outcome, and the characteristics of the survivors. Measurements and Main Results. The CPR data for 82 patients was reviewed. The median duration of CPR was 75 minutes. Patients mean age was 43 ± 21 years with no significant comorbidities. The main causes of the cardiac arrests were myocardial infarction (29%), hypothermia (21%), and pulmonary emboli (12%). 74% of the arrests were witnessed, with a mean latency to CPR of 2 ± 6 minutes and good quality chest compression provided in 96% of the cases. Adjunct therapy included extracorporeal membrane oxygenation (18%), thrombolysis (15.8%), and rewarming for hypothermia (19.5%). 83% were alive at 1 year, with full neurological recovery reported in 63 patients. Conclusion. Patients undergoing prolonged CPR can survive with good outcome. Young age, myocardial infarction, and potentially reversible causes of cardiac arrest such as hypothermia and pulmonary emboli predict a favorable result, especially when the arrest is witnessed and followed by prompt and good resuscitative efforts.
# Introduction
Cardiopulmonary resuscitation (CPR) using closed-chest cardiac massage technique was first used in 1960 by Kouwenhoven in 17 patients with cardiac asystole and 3 patients with ventricular fibrillation, with a successful resuscitation in 14 patients (70%) [bib_ref] History of electrocardiology, Moss [/bib_ref]. There are currently no firm guidelines regarding the duration of such resuscitation [bib_ref] American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care, Morrison [/bib_ref]. In one large study, the overall median duration of resuscitation for inhospital cardiac arrest was 17 minutes with an interquartile range of 10-26 min [bib_ref] Duration of resuscitation efforts and survival after in-hospital cardiac arrest: an observational..., Goldberger [/bib_ref]. For out-of-hospital arrest, the National Association of Emergency Medical Services Physicians suggest that resuscitative efforts could be terminated in patients who do not respond to at least 20 minutes of advanced life support care [bib_ref] Termination of resuscitation in the prehospital setting for adult patients suffering nontraumatic..., Bailey [/bib_ref]. The current overall survival rate to hospital discharge is 17% for in-hospital cardiac arrest and 5% for out-of-hospital cardiac arrest [bib_ref] Cardiopulmonary resuscitation of adults in the hospital: a report of 14720 cardiac..., Peberdy [/bib_ref]. Despite this overall poor prognosis, multiple cases of successful prolonged resuscitation with neurologically intact survival have been reported.
In this paper we describe a case of prolonged resuscitation with successful outcome. Since there is no randomized trials that have evaluated the duration of resuscitation and the bulk of information regarding the duration of resuscitation in various conditions such as hypothermia relies on case series and expert opinion, we reviewed all reported cases of prolonged resuscitation in the English literature. Our goal was to determine the success rate and neurological outcome of these patients and to define the characteristics of patients who might benefit from such heroic measures. year-old African American man was admitted to the orthopedics service with a right midtibial fracture. He has a past medical history of diabetes, anemia, obstructive sleep 2 Critical Care Research and Practice apnea, hypertension, and osteoarthritis. He was taken to the operating room (OR) for open reduction and internal fixation under general anesthesia with endotracheal intubation and sciatic nerve block. Pulse oximetry, end tidal CO 2 (ETCO 2 ), bispectral index (BIS), and arterial line blood pressure were continuously monitored. Patient received intravenous midazolam and fentanyl and was paralyzed with cisatracurium.
## Case report
During the procedure, he was in sinus rhythm with heart rate of 120 per minute, blood pressure was 136/71, and temperature was 36.2 ∘ C, with FiO 2 of 0.6 with O 2 saturation of 98%, ETCO 2 of 37 mmHg, and BIS ranging from 39 to 48 during the entire procedure.
After three and half hours of general anesthesia, the tourniquet was released, and immediately the blood pressure dropped to 50/30 mmHg, the heart rate decreased to 20 beats per minute, O 2 saturation dropped to 80, and the ETCO 2 dropped to 7 mmHg. The patient became pulseless and CPR was immediately started. The initial rhythm was pulseless electrical activity (PEA). Epinephrine and vasopressin were given intravenously. Acute massive pulmonary embolism was suspected. Aspiration from left internal jugular central line was done for possible air embolism. Emergent transesophageal ultrasound was performed and showed a dilated right ventricle, with poor systolic function, and hyperdynamic left ventricle. Heparin bolus of 10,000 units was administered. Systemic thrombolytic therapy was contraindicated with the open leg wound. Despite resuscitation for 20 minutes, the patient remained in PEA. The catheterization lab was activated to attempt percutaneous embolectomy. CPR was continued en route with continuous ETCO 2 monitoring which was maintained above 20 mmHg. Multiple providers switched every 2-3 minutes to maintain good quality chest compression, as measured by the end tidal CO 2 . Upon arriving in the catheterization laboratory the patient had return of spontaneous circulation (ROSC) to normal sinus rhythm. Total CPR duration was 86 min during which the rhythm remained PEA. No cardioversion was performed. Medications given during CPR included 5 mg of epinephrine followed by a drip at 50 micrograms/min, 40 units of vasopressin, 2 mg of atropine, 100 meq (100 mmol) of bicarbonate, and 2 gm (13.6 mmol) of calcium chloride for low ionized calcium level (1.09 mmol/L, normal in our lab is greater than 1.12 mmol/L). Bicarbonate was given due to the concern for severe acidosis in the context of prolonged PEA (pH down to 6.9 during the code). BIS was continuously monitored and remained above 80 during the entire resuscitation.
Pulmonary angiogram revealed multiple bilateral distal pulmonary emboli. No saddle or proximal pulmonary artery emboli were seen. A retrievable inferior vena cava filter was placed and patient was transferred to the intensive care unit. Following return from the cath lab, therapeutic hypothermia was instituted given persistent Glasgow coma scale of 8. A temperature of 33 ∘ C was targeted. This was maintained for 24 h, followed by a gradual rewarming (0.25 ∘ C/hour). In addition, he was anticoagulated with a heparin drip after local hemostasis was achieved at the bedside by the orthopedic surgeons. After 24 hours, the patient was rewarmed. He was successfully extubated 2 days later. He returned to neurologic baseline with normal cognition over the next several days. He was transitioned to Coumadin and was discharged home.
The decision to continue CPR beyond 20 minutes was based on the observed nature of the arrest, the availability of a large number of staff to sustain good quality CPR demonstrated by ETCO 2 monitoring, and the persistence of a high BIS score suggesting persistent brain activity.
## Review of published case reports with prolonged resuscitation
# Materials and methods.
We performed a PubMed search of "Prolonged Resuscitation" of all published articles from 1947 to 2013. 3,826 publications were found. Subsequent filters applied included the following: "Human Species", "Case reports", "Abstract", and "English language". 491 articles met the above criteria and were obtained by online access and through interlibrary loan. All 491 articles were reviewed by 2 pulmonary and critical care physicians. Articles were included in this study if they included a detailed description of the resuscitative effort during cardiac arrest (such as immediate cause of arrest, latency to chest compression, duration of chest compression, amount of defibrillation, and number of returns of spontaneous circulation), if such effort lasted at least 20 minutes and if there was a description of the outcome including death or neurological status.
# Results
. We identified 71 case reports published describing prolonged resuscitation after cardiac arrest. Some of the reports had more than 1 patient, and therefore a total of 82 patients are included in our review. All cases were published between 1980 and 2013 . The baseline characteristics of the patients are listed in [fig_ref] Table 1: Baseline characteristics of the 82 patients [/fig_ref]. The mean age was 43 ± 21 years. The main causes for the cardiac arrests were acute myocardial infarction (AMI) (29%), followed by hypothermia (21%) and pulmonary emboli (PE) (12%). Other etiologies include drug overdose (6%) and arrhythmia (7%). There were no cases of sepsis.
The latency to CPR had an average duration of 2 ± 6 minutes [fig_ref] Table 2: Characteristics of the cardiac arrests [/fig_ref]. The initial cardiac arrest rhythm was ventricular tachycardia or fibrillation in 41 cases (50%), and pulseless electrical activity or asystole in 37 cases (45%). The median duration of chest compression was 75 minutes with a range of 20 to 330 minutes (Table 2, [fig_ref] Figure 1: Duration of the cardiopulmonary resuscitation in the 82 patients [/fig_ref]. Seven patients only had 20 to 30 minutes of resuscitation. Quality of chest compression was deemed good in 96% of the patients in these reports, even though the way this was determined is not clarified.
Several types of adjunct therapies were used [fig_ref] Table 3: Use of adjunct therapy [/fig_ref]. Extracorporeal membrane oxygenation (ECMO) was used in 15 cases without the presence of a pulse, for a mean duration of 2.7 ± 3.9 days (range 3 minutes to 11 days). Thirteen patients received thrombolysis, including 9 out of the 10 patients with PE, as well as 3 patients with acute myocardial infarction. Rewarming was performed in 16 of the 17 hypothermic patients (94%). In 1 case of hypothermia, cardiac arrest occurred after starting the rewarming process, and the patient was kept hypothermic thereafter. Hypothermia after return ofwere mainly respiratory (27%), secondary to pulmonary edema and bronchopneumonia. There was 1 case of hemothorax, 2 cases of pneumothorax, and 2 reported flail chest after chest compression. Renal failure occurred in 15 patients (18%), while liver hematoma was reported in 2 cases. There was 1 case of deformed aortic valve that was recently implanted.
Interestingly, signs of life during CPR were reported in 10 cases (12%). These signs were present without evidence of spontaneous circulation and included making respiratory efforts [bib_ref] Resuscitation, prolonged cardiac arrest, and an automated chest compression device, Risom [/bib_ref] [bib_ref] Prolonged retention of awareness during cardiopulmonary resuscitation for asystolic cardiac arrest, Bihari [/bib_ref] [bib_ref] A case of prolonged life support, Bercker [/bib_ref] , sucking on the ET tube [bib_ref] Prolonged cardiac arrest under circulatory support with extracorporeal membrane oxygenation, Ortega [/bib_ref] , hand movements [bib_ref] Prolonged retention of awareness during cardiopulmonary resuscitation for asystolic cardiac arrest, Bihari [/bib_ref] [bib_ref] Ultra long cardiopulmonary resuscitation with intact cerebral performance for an asystolic patient..., Yu [/bib_ref] [bib_ref] Successful extracorporeal life support in cardiac arrest with recurrent ventricular fibrillation unresponsive..., Shin [/bib_ref] [bib_ref] Complete recovery after prolonged resuscitation and cardiopulmonary bypass for hyperkalemic cardiac arrest, Lee [/bib_ref] , opening eyes to call [bib_ref] Prolonged retention of awareness during cardiopulmonary resuscitation for asystolic cardiac arrest, Bihari [/bib_ref] , tracking people around [bib_ref] Prolonged retention of awareness during cardiopulmonary resuscitation for asystolic cardiac arrest, Bihari [/bib_ref] , occasional gaging and blinking [bib_ref] Prolonged resuscitation in accidental hypothermia: use of mechanical cardio-pulmonary resuscitation and partial..., Jones [/bib_ref] , remaining conscious [bib_ref] Prolonged retention of awareness during cardiopulmonary resuscitation for asystolic cardiac arrest, Bihari [/bib_ref] [bib_ref] Ultra long cardiopulmonary resuscitation with intact cerebral performance for an asystolic patient..., Yu [/bib_ref] , responding to name [bib_ref] Prolonged cardiopulmonary resuscitation with preservation of cerebral function in an elderly patient..., Gabrielli [/bib_ref] , and following simple commands [bib_ref] Prolonged retention of awareness during cardiopulmonary resuscitation for asystolic cardiac arrest, Bihari [/bib_ref] [bib_ref] Prolonged cardiopulmonary resuscitation with preservation of cerebral function in an elderly patient..., Gabrielli [/bib_ref].
# Discussion
Studies of the effect of the duration of resuscitation on clinical outcome are few. A recent retrospective analysis of out-ofhospital cardiac arrests showed a decrease in the probability of survival to hospital discharge with a good functional outcome with each minute of cardiopulmonary resuscitation. The probability of good functional outcome after 15 minutes of CPR was down to 2%, compared to 75% for patients resuscitated for 10-15 minutes [bib_ref] Duration of resuscitation efforts and functional outcome after out-of-hospital cardiac arrest: when..., Reynolds [/bib_ref]. In another study from Taiwan, the rate of ROSC was more than 90% in patients resuscitated for less than 10 minutes, compared to 50% when the CPR duration was more than 30 minutes [bib_ref] A web-based Utstein style registry system of in-hospital cardiopulmonary resuscitation in Taiwan, Shih [/bib_ref]. Based on these studies, it appears that the need for longer resuscitation is associated with worse outcome. On the other hand, in a review by Goldberger, longer duration of resuscitation was associated with a higher survival rate, especially in patients with an initial rhythm of PEA or asystole [bib_ref] Duration of resuscitation efforts and survival after in-hospital cardiac arrest: an observational..., Goldberger [/bib_ref]. Finally, an observational study by Cha et al. evaluated the duration of CPR and showed that resuscitation rate was 25.5% and survival rate was 5.6% when duration of CPR was more than 30 minutes [bib_ref] The duration of cardiopulmonary resuscitation in emergency departments after out-of-hospital cardiac arrest..., Cha [/bib_ref]. The difference with our rate is most likely secondary to the publication bias inherent to the retrospective nature of our review.
Our reported and reviewed cases describe a unique group of patients that received cardiopulmonary resuscitation for a duration that far exceeded the average. In fact, the median CPR duration was 75 minutes, compared to an average of 17 minutes for in-hospital cardiac arrest [bib_ref] Duration of resuscitation efforts and survival after in-hospital cardiac arrest: an observational..., Goldberger [/bib_ref] and 17.5 minutes for out-of-hospital cardiac arrest [bib_ref] Duration of resuscitation efforts and functional outcome after out-of-hospital cardiac arrest: when..., Reynolds [/bib_ref]. In addition, in 18% of the reviewed cases, the use of ECMO without a pulse was performed for a median duration of 55 minutes. This adds to the total duration of resuscitation. The 1-year survival was 83%, significantly higher than the overall survival rate to hospital discharge of 17% for in-hospital cardiac arrest [bib_ref] Cardiopulmonary resuscitation of adults in the hospital: a report of 14720 cardiac..., Peberdy [/bib_ref] and of 5% for out-of-hospital cardiac arrest . The neurological outcome was impressive as 78% of the patients achieving a full neurological recovery or CPC1. A major drawback in our study is the presence of publication bias. Positive results' bias occurs when authors are more likely to submit, or editors accept, positive compared to negative or inconclusive results cases [bib_ref] Bias in analytic research, Sackett [/bib_ref]. Cases of prolonged resuscitation with successful outcome are more likely to be written and accepted for publication compared to similar cases with poor outcome. On the other hand, conducting a randomized controlled trial may not be ethical. Despite that we believe that our study has identified a group of patients that can benefit from prolonged resuscitation. These patients were generally young, with no significant comorbidities. The cardiac arrest etiologies were generally reversible and correctable. Most arrests were witnessed, followed by immediate CPR, with advanced life support care, including ECMO when available. These represent the chain of survival factors advocated by the 2010 American Heart Association (AHA) guidelines, which include immediate recognition and activation, early CPR, rapid defibrillation, effective advanced life support, and integrated post-cardiac arrest care [bib_ref] American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care, Travers [/bib_ref]. Physicians are frequently responsible for determining the maximal duration of resuscitation for each patient. We do not believe that an absolute duration of CPR is adequate for Critical Care Research and Practice 5 all patients. Rather, physicians should base the decision to continue CPR on several factors that affect the chances of survival after cardiac arrest. These factors include the patient baseline status, coexisting comorbidities, latency to CPR, latency to defibrillation, and adequacy of chest compression which should be monitored with ETCO 2 and/or diastolic blood pressure [bib_ref] American Heart Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care, Neumar [/bib_ref].
Cerebral perfusion is the goal for CPR. BIS has been used to monitor the hypnotic state in the operating room and to titrate sedation in the ICU [bib_ref] Bispectral analysis measures sedation and memory effects of propofol, midazolam, isoflurane, and..., Glass [/bib_ref]. It has also been shown to reflect changes of cerebral perfusion during resuscitation of an animal model of shock [bib_ref] Effects of cerebral hypoperfusion on bispectral index: a randomised, controlled animal experiment..., Cavus [/bib_ref]. BIS monitoring during CPR was used by some to assess cerebral perfusion as well as adequacy of chest compression [bib_ref] Value of bispectral index monitoring during cardiopulmonary resuscitation, Szekely [/bib_ref] ; some authors even suggested that a level greater than 30 to 40 during CPR is a reasonable target for better neurological outcome [bib_ref] Can bispectral index monitoring predict recovery of consciousness in patients with severe..., Fàbregas [/bib_ref] [bib_ref] Bispectral index is an indicator of adequate cerebral perfusion during cardiopulmonary resuscitation, Chakravarthy [/bib_ref].
Even though BIS can artificially be increased by movement artifact such as chest compression [bib_ref] An observational study of bispectral index monitoring for out of hospital cardiac..., Fatovich [/bib_ref] , providers should consider the need for sedatives in patients undergoing CPR who may exhibit signs of awareness. In our case, the patient had no such signs.
While there is no proven correlation between BIS monitoring and neurological outcome after CPR, our case report shows its potential benefit; we believe that a high level should be encouraging to the team to continue their effort as it reflects adequate perfusion; however a low level is more difficult to interpret.
The underlying disease causing the cardiac arrest can significantly affect the outcome of the resuscitation. Of all etiologies, four specific diagnoses constituted 67% of all reviewed cases. These are acute myocardial infarction, hypothermia, pulmonary emboli, and drug overdose. These etiologies should deserve special consideration by the treating physician for prolongation of CPR duration as well as other adjunctive therapies such as ECMO. In fact out-of-hospital cardiac arrests of cardiac origin have been associated with a better outcome compared to those of noncardiac origin [bib_ref] Out-of-hospital cardiac arrests of non-cardiac origin: epidemiology and outcome, Kuisma [/bib_ref]. In cases of cardiac arrest secondary to hypothermia, multiple case reports indicate survival after prolonged CPR and prolonged downtime. Based on these case reports the AHA recommends continuing resuscitative effort until rewarming has been provided [bib_ref] American Heart Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care, Vanden Hoek [/bib_ref]. In 10 of our 59 reviewed patients (17%), hypothermia was the cause of the arrest. Rewarming was provided in all these patients. Associated conditions included drowning in 1 case, diabetic ketoacidosis in 1 case, and drug overdose in 1 case. Only 1 death was reported in this population after 48 hours of admission. One study found that when cardiac arrest occurs secondary to PE, the diagnosis can be missed in 30% of the cases and thrombolytic therapy can significantly increase the rate of ROSC (81% versus 43%, = 0.03) [bib_ref] Pulmonary embolism as cause of cardiac arrest: presentation and outcome, Kürkciyan [/bib_ref]. In our review, 7 patients (12%) were diagnosed with PE. All of them received thrombolytic therapy and survived to discharge with good neurological outcome. The initial rhythm was pulseless electrical activity (PEA) in 5 of the 7 patients (72%) and asystole in the other 2, consistent with the previously reported high rate of PEA (63%) and asystole (32%) in patient with PE [bib_ref] Pulmonary embolism as cause of cardiac arrest: presentation and outcome, Kürkciyan [/bib_ref].
ECMO may improve outcome after cardiac arrest when compared to standard CPR. It may be particularly useful in cases of witnessed arrest, brief no-flow duration or when the underlying circulatory disease is amenable to immediate cardioversion. ECMO may also be beneficial as a bridge to correcting reversible conditions such as hypothermia and drug intoxications. Fifteen patients (25.4%) in our review were treated with ECMO. The primary cause of arrest was acute myocardial infarction (5/15), hypothermia (4/15), hyperkalemia (2/15), myocarditis (2/15), hypertrophic cardiomyopathy (1/15), and arrhythmia (1/15). There was only one reported death in this group. Fourteen patients were discharged home with good neurological function.
An interesting finding was the presence of signs of life in the absence of spontaneous pulse. These signs ranged from spontaneous respiratory effort to following commands while receiving chest compression. These signs disappeared when CPR was withheld to evaluate for ROSC. This could be explained by good CPR quality leading to a sufficient brain perfusion. In such cases, it is important to continue good resuscitative effort and minimize CPR interruption. Spontaneous pulse should only be checked at the recommended frequency to minimize brain injury from hypoperfusion.
# Conclusion
Our review suggests that the decision to continue or stop CPR should not be based solely on the duration of resuscitation. Factors that affect the outcome include the patient's baseline condition, the reversibility of the cause of the arrest, the latency to starting CPR, the quality of CPR, and the availability and expertise in ECMO. It is possible to encounter signs of life during CPR which should be interpreted as evidence of good perfusion to an intact brain. It is important not to interpret these signs as evidence of ROSC and to minimize any interruption in chest compression.
# Authors' contribution
Drs. Houssein Youness and Ahmed Awab were involved in the conception and design of the study. Drs. Tarek Al Halabi and Houssein Youness were involved with data acquisition. Drs. Jean Keddissi and Houssein Youness were involved in the statistical analysis and interpretation of the data. Dr. Houssein Youness was involved in the drafting of the paper. Drs. Jean Keddissi, Ahmed Awab, Hussein Hussein, and Kellie Jones were involved in the revision of the paper.
[fig] Figure 1: Duration of the cardiopulmonary resuscitation in the 82 patients. [/fig]
[table] Table 1: Baseline characteristics of the 82 patients. [/table]
[table] Table 2: Characteristics of the cardiac arrests. Quality of chest compression reported in 51 patients only. [/table]
[table] Table 3: Use of adjunct therapy. [/table]
|
Subsets and Splits