{"paper_id":"6dbaba4c-82f1-45de-bedd-763f4540e497","body_text":"Asian Pacific Journal of Cancer Prevention, Vol 22\n553\nDOI:10.31557/APJCP .2021.22.2.553\nARID1A Expression in Endometriosis Model with Endometrial Autoimplantation and DMBA Induction\nAsian Pac J Cancer Prev, 22 (2), 553-558\nIntroduction\nOvarian malignancy is one of the most deadly types \nof malignancy in gynecology with a 5-year survival rate \nof less than 40% (Bray et al., 2018; Mangili et al., 2012; \nReid et al., 2017). GLOBOCAN data in 2018 revealed \nthat the incidence of ovarian malignancy ranks seventh in \nwomen, with 780,000 deaths worldwide. According to the \nIAPI Cancer Registry, 1,351 cases of ovarian malignancy \nwere recorded in 2013. It ranked 7th in the primary organ \ntumors and is the 3rd most common malignancy in women \n(Badan Registrasi Kanker Perhimpunan Dokter Spesialis \nPatologi Indonesia, 2017).\nOvarian malignancy comes from three cell types, \nnamely epithelial, stromal, and germinal cells. Epithelial \nmalignancy is recorded in more than 90% in all ovarian \nmalignancies. Clear cell ovarian carcinoma (CCOC) \nAbstract\nBackground: Ovarian carcinoma is one of the most deadly malignancies in the gynecologic field. The cause is not \nyet known, and the clinical symptoms are not specific. Endometrioid carcinoma and ovarian clear cell carcinoma can \noriginate from endometriosis and are known as endometriosis-related ovarian carcinoma (EAOC). Development of \nEAOC experimental animal models is needed for basic research and clinical preparation of human tissue tests. This \nstudy aimed to determine the role of the ARID1A gene mutation in the carcinogenetic process of EAOC in experimental \nanimal models induced with DMBA. Methods: In this study, the EAOC experimental model was developed using \nthe autoimplantation technique and DMBA induction. This study involved placebo surgery mice (sham), endometrial \nautoimplantation, and a combination of endometrial autoimplantation and DMBA induction, which were sacrificed \nat weeks 5, 10, and 20, respectively. Histopathological assessment and immunohistochemical ARID1A staining with \nan assessment of positive percentages were carried out on 200 cells. Results: This study produced 1 (20%) atypical \nendometriosis and 1 (20%) clear cell carcinoma at implantation and after 10 weeks of DMBA induction, and 100% \nendometrioid carcinoma in the DMBA-induced group. ARID1A staining did not show any significant difference \n(p = 0.313) in all groups. Conclusion: The combination of endometrial autoimplantation techniques and DMBA induction \nin the ovary produced atypical endometriosis, clear cell carcinoma, and endometrioid carcinoma, where time is an \nimportant factor. There was no significant difference in ARID1A expression between the treatment and control groups.\nKeywords: Endometriosis- EAOC- experimental animal model- DMBA- ARID1A\nRESEARCH ARTICLE\nHistopathology and ARID1A Expression in Endometriosis-\nAssociated Ovarian Carcinoma (EAOC) Carcinogenesis Model \nwith Endometrial Autoimplantation and DMBA Induction\nand endometrioid ovarian carcinoma (EOC) are often \nassociated with benign endometriotic lesions. (Ellenson \nLH et al., 2014; Gilks et al., 2014; Kurman and Shih, 2010)\nEndometriosis is a gynecological disorder that is often \nfound in women of reproductive age. Approximately \n1% of endometriosis has the potential to be malignant \n(Esmaili et al., 2016; Xiao et al., 2012). Ovarian \nmalignancy associated with endometriosis is known as \nendometriosis-associated ovarian carcinoma (EAOC). \nPredisposing factors that cause endometriosis to \nturn into ovarian malignancy are known to involve \nmany factors including genetic factors and epigenetic \nfactors and/or effects of the tumor microenvironment \n(Akbarzadeh-Jahromi et al., 2015)\nOvarian carcinoma and adjacent endometriosis lesions \nhave similar genetic changes (Ness, 2003). Wiegand et al., \n(2010) found mutations of the ARID1A tumor suppressor \nEditorial Process: Submission:08/12/2019   Acceptance:02/08/2021\n1Department of Anatomical Pathology, Faculty of Medicine Universitas Indonesia. 2Animal Research Facilities, Indonesian Medical \nEducation and Research Institute, Faculty of Medicine Universitas Indonesia. 3Specialty Programme in Anatomical Pathology, \nDepartment of Anatomical Pathology, Faculty of Medicine Universitas Indonesia. 4Department of Obstetrics and Gynecology, \nFaculty of Medicine Universitas Indonesia. 5Human Reproduction, Infertility, and Family Planning, Indonesian Medical Education \nand Research Institute, Faculty of Medicine Universitas Indonesia. *For Correspondence: ekawuyung@yahoo.com\nPuspita Eka Wuyung1,2*, Familia Bella Rahadiati3, Hartono Tjahjadi1, Salinah \nSalinah1, Kusmardi Kusmardi1, Ria Kodariah1, Budi Wiweko4,5\n\nPuspita Eka Wuyung et al\nAsian Pacific Journal of Cancer Prevention, Vol 22\n554\ngene in endometrioid carcinoma and clear ovarian cell \ncarcinoma. The same mutation was found in atypical \nendometriosis associated with ovarian carcinoma, so it was \nestimated that ARID1A mutation occurs at the beginning of \nthe neoplastic transformation of endometriosis. Research \non the transformation of malignancies in endometriosis is \nof current interest. This study was conducted to determine \nthe role of the ARID1A gene mutation in the carcinogenetic \nprocess of  EAOC in experimental animal models induced \nwith DMBA.\nMaterials and Methods\nThis study was carried out with an experimental design. \nEthical approval was given by the ethics committee of the \nFaculty of Medicine, Universitas Indonesia, under file \nnumber 1028/UN2.F1/ETIK/2017. Twenty-five rats were \nused from the Center for Research and Development of \nBiomedical and Basic Health Technology, Indonesian \nMinistry of Health, which were divided into 5 treatment \ngroups: endometrial autoimplantation group without \nDMBA induction, a combination of autoimplantation and \nDMBA induction techniques. DMBA (Sigma Chemical \nCo., St. Louis, MO) was heated to 124°C to reach the \nfusion point. A 3.0 silk thread with a length of 0.5 cm was \nimmersed in the melted DMBA. Silk thread would then \ncontain 1 mg of DMBA.\nThe surgery was carried out at IMERI FKUI, 3 days \nafter acclimatization. Anesthesia was carried out with \nketamine hydrochloride (73 mg/BW) and xylazine (8.8 \nmg/BW) intraperitoneally. Before surgery, the aseptic \nprocedure was performed on the abdominal area using \npovidone iodine, and a vertical incision was made in \nthe middle of the abdomen. Left uterine tissue was cut 1 \ncm long and reversed so that the endometrium appeared \non the outer surface, and tissue was then placed in a \ncold 0.9% NaCl solution. After that, the endometrial \ntissue and DMBA-coated threads were implanted in the \nright ovary using nylon threads sized 4.0. Ceftriaxon \nwas then administered at 20 mg/KgBB for 3 days \nintraperitoneally. At 5, 10, and 20 weeks, euthanasia \nusing ketamine was performed together with the sham \ngroup, and then the implanted tissue in the right ovary \nwas taken and fixed with 10% formalin buffer. It was \nthen made into Hematoxylin-Eosin (HE) preparations \nand immunohistochemical staining was performed using \nARID1A-PSG3 Santa Cruz biotechnology primary \nantibodies and seen under a light microscope.\nAssessment of ARID1A expression was carried out \nin normal endometrial epithelial cells, epithelial and \nglandular cells of the autoimplanted endometriosis, and \nin induced tumor cells. The loss of ARID1A expression \nwas assessed by loss of expression in the cell nucleus. \nSemiquantitative assessment using Image G refers \nto the study by (Fadare et al., 2012), which relies the \nassessment on the percentage of stained cells regardless \nof the intensity of the staining. The assessment was carried \nout on 200 cells. The data obtained were then analyzed \nstatistically with the SPSS 22.\nResults\nIn the endometrial autoimplantation group without \nDMBA induction, the graft tissue proliferated to resemble \nendometriosis in humans, in which no atypia was found. \nIn the group with combination of autoimplantation \ntechnique and 5-week DMBA induction, graft tissue \ngrew to resemble non-atypical endometriosis in humans, \n1 of which had squamous metaplasia. In the group with \ncombination of autoimplantation technique and 10-week \nDMBA induction, 1 malignant tumor was found, as well as \n1 endometriosis cyst with hard atypia and 3 endometriosis \nwithout atypia, 2 of which were squamous metaplasia. All \nsubjects in the sham group showed normal endometrial \nhistology (Figure 1).\nThe sham (control) group shows normal endometrial \nFigure 1. A. Endometrial cyst (HE40x). B. Endometrial cyst with squamous metaplasia (HE400x). C. Cyst with \nfibrotic wall, with glands lined with atypical cells in between (HE 100x). B. Stratified epithelium, tufting, and nuclear \natypia (HE 400x)\n\nAsian Pacific Journal of Cancer Prevention, Vol 22\n555\nDOI:10.31557/APJCP .2021.22.2.553\nARID1A Expression in Endometriosis Model with Endometrial Autoimplantation and DMBA Induction\nFigure 2. A-B. Clear cell carcinoma. A tubulocystic structure with a hobnail core and a solid structure resembling \ncobblestone can be found. Tumor cells appeared polygonal in shape with a pleomorphic nucleus and a pronounced \nnucleolus. The cytoplasm appears clear, mostly eosinophilic (HE400x). C. Endometrioid carcinoma. The tumor mass \n(♦) destructed the ovarium (●). The endometrial cyst wall was found (↑) (HE 40x). D. Endometrioid carcinoma \nforming a glandular structure. Tumor cells had round, oval, and columnar nuclei, with rough chromatin, which \nappeared vesicular with the nucleolus. Signs of mitosis were found. Cytoplasm,  eosinophilic (HE400x).\nFigure 3. ARID1A Expression. A. Negative control. B. Positive control from intestinal tissue. C. ARID1A expres-\nsion in normal endometrium in the sham group. D. ARID1A expression in non-atypical endometriosis. E. ARID1A \nexpression in atypical endometriosis. F. Endometrioid carcinoma showed patchy negative ARID1A expression (A-E, \nHE 400x).\n\n\nPuspita Eka Wuyung et al\nAsian Pacific Journal of Cancer Prevention, Vol 22\n556\nhistology. The endometrial autoimplantation group \nwithout DMBA induction showed similar histology to \nhuman tissue, with no identified atypia. Cysts covered \nby flat epithelium, cuboid to columnar, could be seen. \nInflammatory cell infiltration was also found.\nThe subjects in the endometrial autoimplantation \ngroup with 5-week DMBA induction all showed \nnon-atypical endometriosis, one of which happened to be \nsquamous metaplasia (Figure 1.A, B). The endometrial \nautoimplantation group with 10-week DMBA induction \ngave rise to non-atypical endometriosis in 3 rats, 2 \nof which were squamous metaplasia, 1 was atypical \nendometriosis (Figure 1.C, D), and 1 was clear cell \ncarcinoma (Figure 2.A, B.).\nThe group of rats treated with a combination of \nendometrial autoimplantation and 20-week DMBA \ninduction all had endometrioid carcinomas. Histologic \nfeatures showed a tumor mass that formed a back-to-back \nglandular structure to a solid structure. The tumor cells \nhad a moderate pleomorphic nucleus, rough chromatin, \nwhich appeared vesicular with the nucleolus. Cytoplasm \nappeared eosinophilic. Signs of mitosis were found \n(Figure 2.C, D).\nARID1A expression\nARID1A staining (Figure 3) was assessed with Image \nJ. ARID1A expression in the sham group was 98.5-100% \nwith a median of 99.7%. ARID1A expression in the \nendometrial autoimplantation group yielded a range of \nvalue of 86-100% with a median of 95.10%.  ARID1A \nexpression in the endometrial autoimplantation group \nwith 5-week DMBA induction was 94-100% with a \nmedian of 92.2%. ARID1A expression in the endometrial \nautoimplantation group with 10-week DMBA induction \nwas 91.5-100% with a median of 92.2%.  ARID1A \nexpression in the endometrial autoimplantation group with \n20-week DMBA induction was in the range of 71.5-100% \nwith a median of 90.40 %. \nIn the endometriosis group and groups with \ncombination of autoimplantation and DMBA induction, \nthe median percentage of ARID1A expression was lower \ncompared to the sham group. However, the Kruskal-Walis \ntest revealed that this difference was not significant \n(p=0.313). \nDiscussion\nSpontaneous endometriosis cannot occur in rats. Rats \ndo not experience the menstrual cycle but experience an \nestrus cycle. In the estrus cycle, there is no endometrial \nshedding, but resorption of the damaged endometrium. \nImplantation of the endometrium into the ovary is \nanalogous to retrograde menstruation which can occur \nspontaneously in humans, causing endometrium cells to \nmigrate to the ovary. In rats, this spontaneous process does \nnot occur so that the process of transfer of endometrial \ntissue can be manipulated by autoimplantation techniques \nto the ovaries, where ectopic endometrial tissue can be \nobtained (Berkley et al., 2004; Grümmer, 2006; King \net al., 2016). The combination of autoimplantation \ntechniques and DMBA induction for 5 weeks resulted in \nnon-atypical endometriosis, one of which was squamous \nmetaplasia. Similar to the human condition, metaplastic \nlesions are often found in endometriosis.\nOvaries are likened to fertile soil for the growth of \nendometriosis cells. When ovulation occurs, the process \nof rupture and repetitive repairs cause local inflammation \nthat changes the condition of the microenvironment. \nCytokines and growth factors are produced at the site of \novarian rupture, producing chemotactic factors that attract \ncells outside the ovary, both benign and malignant cells. \nPost-ovulatory inflammation and pro-repair condition \nalso function as tumorigenesis factors that support the \ntransformation of malignancy and allow malignant cells \nto survive (Yang-Hartwich et al., 2014).\nEndometriosis cysts in humans contain many iron \nelements which are thought to play a role in the process \nof transformation of endometriosis malignancies. The \nfree-form or catalytic accumulation of iron mediates \nthe formation of reactive oxygen species (ROS) \nthrough Fenton reactions and induces oxidative stress. \nOxygen-free radicals produced from free iron induce \nmutations and DNA damage. This high iron content plays \nan important role in carcinogenesis originating from \nendometriosis (Yamaguchi et al., 2008).\nInflammation is considered as the hallmark \nof endometriosis. The high abnormal activity of \nintraperitoneal macrophages plays a role in the optimization \nof endometriosis growth by releasing angiogenic factors, \ntherefore increasing the microvascularization of the \nparietal peritoneum. Endometrial implants in the ovary \ncause estrogen persistence and reduction of 2 progesterone \nreceptor isoforms resulting in physiological milieu \nchanges around the surface of the ovary. Progesterone can \nsuppress cell proliferation and induce apoptosis. Changes \nin hormone receptors give rise to non-physiological \nhormonal conditions and can cause further progression \ntowards malignancy (Yamaguchi et al., 2008). This is \nsupported by the results of this study, which observed \nthat the group with endometrial autoimplantation and \n10-week DMBA induction obtained 1 (20%) atypical \nendometriosis. Atypical endometrosis is known to be an \nEAOC precursor lesion. In this group, there was also 1 \n(20%) malignancy with clear cell carcinoma. In the group \nobserved until the 20th week, all malignancies were \nfound with endometrioid carcinoma. Therefore, it can be \ndrawn that time influences the emergence of malignant \ntransformation in endometriosis.\nIn rats with endometrial implantation only, changes \nwere also observed in the microenvironment. However, \nin these experimental animals, no iron accumulation \nwas found. Components of ROS derived from iron \ndegradation are not found in experimental animals like \nthey are in humans. Therefore DMBA was used to induce \nmalignancy. DMBA is an indirect carcinogen that will \nbe metabolized by CYP1B1. DMBA metabolic results \nare DMBA-DE, which become DNA adducts and can \ninduce mutation. CYP isoform is known to produce ROS \nduring the metabolism of carcinogens through uncoupled \nreactions. ROS can enter diffusely into cells and interact \n\nAsian Pacific Journal of Cancer Prevention, Vol 22\n557\nDOI:10.31557/APJCP .2021.22.2.553\nARID1A Expression in Endometriosis Model with Endometrial Autoimplantation and DMBA Induction\nwith macromolecules such as lipids, proteins, and DNA to \ninduce oxidative modification which eventually triggers \nthe oncogenic process (Madden et al., 2014; Muqbil et \nal., 2006; Priyadarsini and Nagini, 2012).\nARID1A mutations are one of the most frequently \nreported molecular genetic changes in CCOC and EOC. \nARID1A mutations cause loss of expression of the \nBAF250a protein, which normally suppresses cellular \nproliferation through the regulation of the transcription \nof p53-dependent suppressor tumors such as CDKN1A \nand SMAD3. ARID1A inactivation occurs early in the \ndevelopment of CCOC, or EOC (Ayhan et al., 2012). \nBorrelli et al., (2016) shows a partial loss of ARID1A \nexpression in rectovaginal deep-infiltrating endometriosis \n(DIE), endometrioma, and in endometrial tissue used \nas controls. ARID1A mutations are also often found in \nendometriod carcinoma of the uterus (Gounaris et al., \n2011).\nEOC and CCOC in humans do not all experience \nARID1A mutations. This is supported by observations of \nARID1A expression that were not statistically significant \nbut showed a tendency to decline in some tumors in the \n20-week induction group. Wiegand et al., (2010) suggested \nthat 46% of CCOC had an ARID1A mutation, whereas, \nin EOC, it was stated that 30% had an ARID1A mutation. \nEOC and CCOC without ARID1A mutations are referred \nto as wild-type ARID1A. However, tumors with ARID1A \nmutations can also show ARID1A expression, which \nis totally lost or partially lost. This missing ARID1A \nexpression partly describes clonal mutations and the \npresence of tumor heterogeneity (Win et al., 2016).\nThe results of this study can complement \ncurrent knowledge of ARID1A  expression in the \nimmunohistochemical profiles of DMBA-induced EAOC. \nEOC and CCOC can show mutations other than ARID1A. \nA research conducted by Salinah et al., (2018) revealed \nthat DMBA-induced endometrioid carcinoma showed a \ndecrease in p16INK4a protein expression.\nIn conclusion, histopathological changes occurred in \nrats given endometriosis autoimplantation and ovarian \nDMBA induction along with the emergence of malignant \ntransformation in endometriosis. The combination of \nendometrial autoimplantation techniques and DMBA \ninduction in the ovary produced atypical endometriosis, \nclear cell carcinoma, and endometrioid carcinoma. There \nwas no significant difference in ARID1A  expression \nbetween the treatment and control groups.\nAcknowledgments\nWe thank the Ministry of Research, Technology, and \nHigher Education of Indonesia for the 2018-2019 Hibah \nPenelitian Dasar Unggulan Perguruan Tinggi (University \nResearch Excellence Grant).\nReferences\nAkbarzadeh-Jahromi M, Shekarkhar G, Sari Aslani F, et al \n(2015). Prevalence of endometriosis in malignant epithelial \novarian tumor. Arch Iran Med, 18, 844–8. (True)\nAyhan A, Mao TL, Seckin T, et al (2012). Loss of ARID1A \nexpression is an early molecular event in tumor progression \nfrom ovarian endometriotic cyst to clear cell and endometrioid \ncarcinoma. Int J Gynecological Cancer, 22, 1310–5. \nBadan Registrasi Kanker Perhimpunan Dokter Spesialis  \nPatologi Indonesia. Kanker di Indonesia tahun 2013. Data \nhistopatologik. Jakarta 2017. \nBerkley KJ, Dmitrieva N, Curtis KS, Papka RE (2004). \nInnervation of ectopic endometrium in a rat model of \nendometriosis. Proc Nat Acad Sci U S A, 101, 11094–8. \nBorrelli GM, Abrão MS, Taube ET, et al (2016). (Partial) loss \nof BAF250a (ARID1A) in rectovaginal deep-infiltrating \nendometriosis, endometriomas and involved pelvic sentinel \nlymph nodes. Mol Hum Rep, 22, 329–37. \nBray F, Ferlay J, Soerjomataram I, et al (2018). Global cancer \nstatistics 2018: GLOBOCAN estimates of incidence and \nmortality worldwide for 36 cancers in 185 countries. CA \nCancer J Clin, 68, 394–424. \nEllenson LH, SG C, Cho KR, et al (2014). Endometrioid \ntumours. In K. RJ, C. ML, H. CS, and Y . RH (Eds.), WHO \nclassification of tumours of female reproductive organs, \nIARC, Lyon, pp 29–32.\nEsmaili H, Vahedi A, Mohajeri S, et al (2016). Evaluation of \nthe pathogenesis of tumor development from endometriosis \nby estrogen receptor, P53 and Bcl-2 immunohistochemical \nstaining. Asian Pac J Cancer Prev, 17, 5247. \nFadare O, Renshaw IL, Liang SX (2012). Does the loss of \nARID1A (BAF-250a) expression in endometrial clear cell \ncarcinomas have any clinicopathologic significance? A pilot \nassessment. J Cancer, 3, 129–36. \nGilks C, Bell D, Huntsman D, et al (2014). Clear cell tumours. In \nK. RJ, C. ML, H. CS, and Y . RH (Eds.), WHO classification \nof tumours of female reproductive organs. IARC, Lyon, pp \n33–34.\nGounaris I, Charnock-Jones DS, Brenton JD (2011). Ovarian \nclear cell carcinoma-bad endometriosis or bad endometrium?. \nJ Pathol, 225, 157–60. \nGrümmer R (2006). Animal models in endometriosis research. \nHum Reprod Update, 12, 641–9.\nKing CM, Barbara C, Prentice A, et al (2016). Models of \nendometriosis and their utility in studying progression to \novarian clear cell carcinoma. J Pathol, 238, 185–96.\nKurman RJ, Shih IM (2010). The origin and pathogenesis of \nepithelial ovarian cancer: A proposed unifying theory. Am \nJ Surg Pathol, 34, 433–43. \nMadden JA, Hoyer PB, Devine PJ, Keating AF (2014). Acute \n7,12-dimethylbenz[a]anthracene exposure causes differential \nconcentration-dependent follicle depletion and gene \nexpression in neonatal rat ovaries. Toxicol Appl Pharmacol, \n276, 179–87.\nMangili G, Bergamini A, Taccagni G, et al (2012). Unraveling the \ntwo entities of endometrioid ovarian cancer: A single center \nclinical experience. Gynecol Oncol, 126, 403–7. \nMuqbil I, Azmi AS, Banu N (2006). Prior exposure to restraint \nstress enhances 7,12-dimethylbenz(a)anthracene (DMBA) \ninduced DNA damage in rats. FEBS Lett, 580, 3995–9. \nNess RB (2003). Endometriosis and ovarian cancer: Thoughts on \nshared pathophysiology. Am J Obstet Gynecol, 189, 280–94. \nPriyadarsini RV , Nagini S (2012). Quercetin suppresses \ncytochrome P450 mediated ROS generation and NFκB \nactivation to inhibit the development of 7,12-dimethylbenz[a]\nanthracene (DMBA) induced hamster buccal pouch \ncarcinomas. Free Radical Res, 46, 41–9.\nReid BM, Permuth JB, Sellers TA (2017). Epidemiology of \novarian cancer: a review. Cancer Biol Med, 14, 9-32. \nSalinah, Ekawuyung P, Wiweko B, Kodariah R, Kusmardi \n(2018). Expression of p16INK4a in Endometrioid Cancer \nRat Model. Adv Sci Lett, 24, 6584–6. \n\nPuspita Eka Wuyung et al\nAsian Pacific Journal of Cancer Prevention, Vol 22\n558\nWiegand KC, Shah SP, Al-Agha OM, et al (2010). ARID1A \nmutations in endometriosis-associated ovarian carcinomas. \nN Engl J Med, 363, 1532–43.\nWin TT, Nik M, NMZ Ma, SO, Ismail M (2016). Bilateral \novarian clear cell carcinoma Arising in 17 year longstanding \nhistory of bilateral ovarian endometriosis. Iran J Pathol , \n11, 478–82. \nXiao W, Awadallah A, Xin W (2012). Loss of ARID1A/\nBAF250a expression in ovarian endometriosis and clear cell \ncarcinoma. Int J Clin Exp Pathol, 5, 642–50. \nYamaguchi K, Mandai M, Toyokuni S, et al (2008). Contents \nof endometriotic cysts, especially the high concentration of \nfree iron, are a possible cause of carcinogenesis in the cysts \nthrough the iron-induced persistent oxidative stress.  Clin \nCancer Res, 14, 32–40. \nYang-Hartwich Y , Gurrea-Soteras M, Sumi N,et al (2014). \nOvulation and extra-ovarian origin of ovarian cancer. Sci \nRep, 4, 6116. \nThis work is licensed under a Creative Commons Attribution-\nNon Commercial 4.0 International License.","source_license":"CC0","license_restricted":false}