{"paper_id":"074290f9-56e7-4b38-a389-10535da1d6fe","body_text":"Tubal origin of ovarian endometriosis\nZeng Yuan1,2,3,*, Lijie Wang 1,2,3,*, Yiying Wang 3, Tingguo Zhang 4,L iL i 4,\nJaniel M Cragun 5,6, Setsuko K Chambers 5,6, Beihua Kong 1 and Wenxin Zheng 1,3,5,6\n1Department of Obstetrics and Gynecology, Qilu Hospital, Shandong University, Shandong, China; 2Key\nLaboratory of Gynecologic Oncology, Department of Obstetrics and Gynecology, Qilu Hospital, Shandong\nUniversity, Shandong, China;\n3Department of Pathology, University of Arizona College of Medicine, Tucson,\nAZ, USA; 4Department of Pathology, Shandong University School of Medicine, Shandong, China;\n5Department of Obstetrics and Gynecology, University of Arizona, Tucson, AZ, USA and 6Arizona Cancer\nCenter, University of Arizona, Tucson, AZ, USA\nEndometriosis is a puzzling and debilitating disease that affects millions of women around the world. Ovary is\nthe most common organ site involved by endometriosis. Despite various hypotheses about its cell of origin,\nuncertainty remains. On the basis of our clinicopathologic observations, we hypothesize that fallopian tube\nmay contribute the histogenesis of ovarian endometriosis. To examine if the hypothesis, tubal origin of ovarian\nendometriosis, has scientific supporting evidence, we identified a set of novel genes, which are either highly\nexpressed in the normal fallopian tube or in the endometrium through a gene differential array study. Among\nmany differentially expressed genes, FMO3 and DMBT1 were selected as the initial biomarkers to test the\nhypothesis. These biomarkers were then validated in ovarian sections with foci of endometriosis by comparing\ntheir expression levels in the fallopian tube and the endometrium within the same patients with real-time PCR,\nwestern blot and immunohistochemistry analysis. FMO3 was highly expressed in the tubal epithelia while low in\nthe paired endometrium. In contrast, DMBT1 was high in the endometrium but low in the fallopian tube. In 32\novarian endometriosis cases analyzed by real-time PCR, 18 (56%) showed a high level of FMO3 and a low level\nof DMBT1 expression. However, 14 (44%) endometriosis cases showed a reversed expression pattern with\nthese two markers. Results were similarly seen in the methods of western blot and immunohistochemistry. The\nfindings suggest that approximately 60% of the ovarian endometriosis we studied may be derived from the\nfallopian tube, whereas about 40% of the cases may be of endometrial origin. The fallopian tube epithelia may\nrepresent one of the tissue sources contributing to ovarian endometriosis. Such novel findings, which require\nconfirmation, may have a significant clinical impact in searching for alternative ways of prevention and\ntreatment of endometriosis.\nModern Pathology (2014) 27, 1154–1162; doi:10.1038/modpathol.2013.245; published online 3 January 2014\nKeywords:\nbiomarkers; endometrium; fallopian tube; ovarian endometriosis\nIntroduction\nEndometriosis is defined by the presence of endo-\nmetrial tissue outside the uterus. It is one of the\nmost common benign gynecologic disorders asso-\nciated with pelvic pain and infertility. Endometrio-\nsis, most commonly involving the ovary, affects\napproximately 10% of women in their reproductive\nage and up to 50% of women suffering infertility\nand abdominal pain.\n1–5 The pathogenesis of endo-\nmetriosis remains unclear and elusive since it was\nfirst described by Von Rokitansky over 100 years\nago.6 Sampson’s retrograde menstruation theory 7,8\nhas been widely accepted since 1920s, but remains\ncontroversial as retrograde menstruation occurs in\nup to 90% of women in reproductive age\n9 but only\n6–10% of those women have endometriosis. Retro-\ngrade menstruation may explain occurrence of\nendometriosis within the pelvis or abdominal cavity\nbut fails to explain the presence of endometriosis in\nremote sites outside the peritoneal cavity. These\ncontroversial points led Iwanoff and Meyer to\npropose the coelomic metaplasia theory,\n8 which\nexplained that endometriosis may derive from\nmesothelial cells through metaplasia. The presence\nof endometriosis in remote areas and the rare\nendometriosis in males support the metaplasia\nCorrespondence: Dr B Kong, MD, FCAP , PhD, Department of\nObstetrics and Gynecology, Qilu Hospital, Shandong University,\n107 W.Wenhua Road, Ji’nan, Shandong 250012, China or\nDr W Zheng, MD, FCAP , Department of Pathology, University of\nArizona, 1501 N. Campbell A venue, #5224A, Tucson, AZ 85724, USA.\nE-mail: kongbeihua@sdu.edu.cn or zhengw@email.arizona.edu\n*These authors contributed equally to this study.\nReceived 11 September 2013; revised 5 December 2013; accepted\n6 December 2013; published online 3 January 2014\nModern Pathology (2014) 27, 1154–1162\n1154 & 2014 USCAP, Inc All rights reserved 0893-3952/14 $32.00\nwww.modernpathology.org\n\ntheory.10 However, metaplastic theory also lacks a\ncellular or molecular basis. 11,12 Lymphatic and\nvascular transportation of the endometrium has\nalso been proposed as a complimentary theory to\nexplain rare cases of endometriosis occurring in\nunusual locations far from the pelvis,\n13 but this is\nunlikely to be the primary mechanism of disease\nspread.\n14–16 Another compelling proposal suggests\nthat bone marrow-derived stem cells may differen-\ntiate into endometriotic tissue within and outside of\nendometrial cavity.\n17 But the stem cell theory is\nprimitive and not completely understood, although\nit is interesting. 18 Overall, no single theory perfectly\naccounts for the pathogenesis of all cases of\nendometriosis.\nThe fallopian tube was previously recognized\nonly as a carrier for the menstrual endometrium to\npass into the peritoneal cavity or onto the ovarian\nsurface. The fallopian tube has never been examined\nas a possible tissue or cellular source of endome-\ntriosis. However, we thought the fallopian tube\nlikely contributes to ovarian endometriosis forma-\ntion based on our clinicopathological observations.\nTubal mucosa is known to be able to form endo-\nmetrial-like tissue. For instance, endometrialization\nis commonly seen within the tubal lumen after tubal\nligation.\n8,19 It is also known that tubal epithelia\nshed viable cells onto the ovarian surface forming\nendosalpingiosis or ovarian epithelial inclusions,\na common finding seen within the ovary in\napproximately 30% of the cases.\n20,21 In a recent\nstudy on the cell origin of ovarian low-grade\nserous carcinoma, we demonstrated that the\nmajority of the ovarian epithelial inclusions are\nderived from tubal epithelia\n20 and the ovarian\nepithelial inclusions could be transformed into\novarian endometriosis through a probable meta-\nplastic process.\n22 Therefore, we hypothesize that\ntubal epithelium contributes to the formation of\novarian endometriosis. In this study, we identified\na set of novel genes, which are either highly\nexpressed in the fallopian tube or in the endo-\nmetrium through a gene differential array study.\nWe validated these unique genes and their corres-\nponding protein expression in ovarian endometriosis\nby comparing their expression levels to paired\nspecimens of fallopian tube and endometrium\nwithin individual patients.\nMaterials and methods\nTissue Specimens\nTissue samples including human tubal fimbria,\npaired endometrium and ovarian endometriosis\nwere obtained from surgical pathology specimens\nwithin 30 min of resection at the Department of\nGynecology, Qilu Hospital of Shandong University,\nShandong, China. The human subject research\nprotocols were approved by the institutional review\nboard. A total of 147 specimens derived from 56\npatients were studied. Among them, each of 35\npatients with ovarian endometriosis generated a set\nof samples including fallopian tube, endometrium\nand ovarian endometriosis. The remaining non-\nendometriosis patients generated 21 paired fallo-\npian tube and endometrial samples. All these\npatients underwent total hysterectomy and bilateral\nsalpingo-oophorectomy for either ovarian endo-\nmetriosis or benign gynecologic disease without\nendometriosis. The patients’ age ranged from 35 to\n51 years with a mean age of 42. No patient studied\nreceived hormonal treatment within the 6 months\nbefore surgical resection. Representative portions of\nthe same tissue specimen were either snap frozen\nand stored in liquid nitrogen until use or fixed in\n10% neutral formaldehyde overnight and embedded\nin paraffin for routine histological examination.\nTissue identification and histologic diagnosis of\novarian endometriosis were confirmed by gynecolo-\ngic pathologists. Tissue sections containing areas of\nendometriosis, tubal mucosa and endometrium with\nboth glandular epithelia and stroma were confirmed\nunder microscope and hand microdissected for\neither real-time PCR or western analysis.\nMicroarray and Data Analysis\nIn order to identify tissue-specific biomarkers, we\ncompared the gene expression between the fallopian\ntube and the endometrium from patients without\nevidence of endometriosis by gene array analysis.\nThree pairs of fresh human endometrium (two\nproliferative phase and one secretory phase) and\ncorresponding tubal fimbria specimens were selected\nfrom the pool of the 21 paired samples mentioned\nabove, labeled, and sent to Kang Chen Bio-Tech\n(Shanghai, China) to perform whole-genome expres-\nsion microarray analysis using the Agilent array\nplatform. All the endometrial samples were histolo-\ngically examined under routine microscope and no\ntubal metaplasia was found in any of the samples we\nanalyzed. Total RNA from three pairs of hand\nmicrodissected epithelial samples were prepared\nusing TRIzol (Invitrogen, Gaithersburg, MD, USA),\nfurther quantified by the NanoDrop 1000 and RNA\nintegrity was confirmed by standard denaturing\nagarose gel electrophoresis. The Human Gene\nExpression Array was manufactured by Agilent with\n41 000 þ genes and transcripts represented, all with\npublic domain annotations.\nSample labeling and array hybridization were\nperformed according to the Agilent One-Color\nMicroarray-Based Gene Expression Analysis proto-\ncol (Agilent Technologies, Palo Alto, CA, USA) and\ndescribed elsewhere.\n23 Median normalization and\nsubsequent data processing were performed by\nusing the GeneSpring GX v11.0 software package.\nAfter median normalization of the raw data, genes\nthat at least six out of six samples have flags in\nOviduct contributing to ovarian endometriosis\nZ Yuan et al 1155\nModern Pathology (2014) 27, 1154–1162\n\npresent (‘all targets value’) were chosen for further\ndata analysis. Differentially expressed genes\nwere identified through fold change filtering.\nHierarchical clustering was performed using the\nAgilent GeneSpring GX software (version 11). Gene\nontology analysis and pathway studies were per-\nformed in the standard enrichment computation\nmethod.\nThe criteria for selection of differentially expressed\ngenes were as follows: (1) the cutoff value differen-\ntially expressed level between the tubal fimbria and\nthe endometrium is more than twofold; and (2)\nPo0.05 in fold change filtering. Genes that fit\nthese criteria were considered significant for\ndiscrimination.\nValidation of Microarray Data by Real-Time PCR\nTo verify the gene expression data obtained from the\nmicroarray, real-time PCR was performed on two\nselected genes, using total RNA from 21 paired tubal\nfimbria and corresponding endometrial samples\nincluding those three paired specimens for the gene\narray analysis. Among the 21 paired specimens,\n17 were in proliferative phase and the remaining\n4 were in secretory phase within the menstrual\ncycle. The target genes for this study were selected\nbased on the level of expression after comparing\nwith tubal expression: FMO3 (55.72-fold) and\nDMBT1 ( /C0 44.57-fold). FMO3, flavin-containing\nmonooxygenase 3 belongs to the FMO family,\nencoding microsomal flavin-containing monooxy-\ngenase, which was involved in the oxidative\nmetabolism of a variety of xenobiotics. In human\nbeings, FMO3 is predominantly expressed in the\nadult liver, but it appears not sex dependent. The\ngene DMBT1 codes for a protein belonging to the\nsuperfamily of scavenger receptor cysteine-rich\nproteins. DMBT1 is involved in the immune defense\nand may partially have role in epithelial differentia-\ntion. GAPDH expression was used as the internal\ncontrol. Thirty-five cases with ovarian endometrio-\nsis and paired tubal and endometrial samples were\nanalyzed by real-time PCR. Primers were designed\nusing Primer 3 software and the sequences were as\nfollows: FMO3:F 5\n0-AATTCGGGCTGTGATATTGC-30\nand R5 0-TTGAGGAAGGTTCCAAATCG-30; DMBT1\nF50-TGCTCTGTCTGCCAAATCAC-30 and R5 0-GTCA\nTTGTCTGCCTGCTTGA-30.\nThe protocol of real-time PCR is described else-\nwhere.24 Data analysis was performed with\nStepOnePlust Real-Time PCR System software,\nversion 2.2 (Applied Biosystem, Hercules, CA, USA).\nFor relative quantification of gene expression, the\ncomparative Ct method ( DDCt) was used. To\ncalculate the quantification, amplification products\nwere normalized to GAPDH (DCt). Statistical analysis\nwas performed using paired two-tailed t-test to\ncompare relative mRNA expression levels in the\nfallopian tube and the corresponding endometrium,\nor ovarian endometriosis lesions. Statistical signifi-\ncance was defined as a P-valueo0.05.\nWestern Blot Analysis\nMonoclonal antibodies for FMO3 and DMBT1 were\nobtained from Abcam, USA, FMO3 for 1:8000\ndilution, DMBT1 for 1:1000 dilution. All samples\nmentioned above were subsequently evaluated for\nprotein expression by western blot and the detailed\nmethod is described elsewhere.\n25 GAPDH antibody\nwas used as the loading control.\nImmunohistochemistry\nImmunohistochemistry with antibodies to FMO3\nand DMBT1 was performed as described pre-\nviously.\n20 The fallopian tubal mucosa and prolifera-\ntive endometrial sections served as positive controls\nfor FMO3 and DMBT1, respectively. Negative\ncontrols were carried out by replacing primary\nantibodies with class-matched mouse and rabbit\nIgGs on parallel sections. The subcellular staining\nlocalization for FMO3 was cytoplasmic, whereas for\nDMBT1 it was both cytoplasmic and membranous.\nMore than 25% positively stained epithelial cells\nwere arbitrarily defined as positive, whereas 25% or\nless stained were negative.\nThe potential stromal cell contamination from\nthose microdissected endometrial or tubal epithelia\nwas examined by using immunohistochemical\nstaining with CD10 (positive for endometrial stromal\ncells) and vimentin (positive for tubal stromal cells).\nThe purity of the microdissected epithelia reached\n499% as there were barely any CD10 or vimentin-\npositive cells observed in representative samples.\nStatistical Analysis\nA t-test was used for analyzing the differences of\nmRNA and protein expression between the tubal\nfimbria and the paired endometrium. ANOVA was\nused to analyze the different expression levels\namong the tubal fimbria, the paired endometrium\nand the foci of ovarian endometriosis. Multiple\ncomparisons by LSD t-test were carried out among\ngroups of ovarian endometriosis, fallopian tube and\nendometrium. Analyses were performed using the\nSPSS statistical software program version 13.0\n(SPSS, Chicago, IL, USA). Po0.05 was considered\nto be statistically significant.\nResults\nIn order to examine if the fallopian tube truly has a\nrole in the pathogenesis of ovarian endometriosis,\nwe have to examine the possibility of the endome-\ntrial origin in the same setting. On the basis of\nthis understanding, we have used total mRNA\nModern Pathology (2014) 27, 1154–1162\nOviduct contributing to ovarian endometriosis\n1156 Z Yuan et al\n\nmicroarray analysis to identify unique genes and\ncorresponding proteins from the tubal and endome-\ntrial samples of patients without endometriosis.\nIdentified unique products were then studied in\nthe lesions of ovarian endometriosis as well as in\npaired tubal and endometrial tissues.\nMultiple Unique Genes Identified in the Fallopian\nTube Over the Endometrium\nA total of 4114 and 3451 genes were identified in the\nfallopian tube and endometrium, respectively. The\ngene expression profile of the fallopian tube and\npaired endometrium was compared by using a\nVolcano Plot. The threshold for the gene expression\nlevel comparison between the fallopian tube and the\nendometrium was Z2.0 fold change. There were\n1796 genes identified with more than twofold\ndifferential expression between human fallopian\ntube and endometrial tissues. All these differentially\nexpressed genes were further scrutinized and the\nhighly differentially expressed genes were summar-\nized. Compared with the endometrium, the fallo-\npian tube showed a total 911 upregulated genes.\nThese included 50-fold or more ( n ¼ 8), 20-fold or\nmore ( n ¼ 28) and 2-fold or more ( n ¼ 875). Com-\npared with the fallopian tube, the endometrium\nshowed a total of 885 upregulated genes including\n20-fold or more ( n ¼ 7) and 2-fold or more ( n ¼ 878).\nThere were no genes with 450-fold upregulation\nfound in the endometrial tissue. The representative\nup- or downregulated genes are listed in Table 1 as\nSupplementary Data.\nFMO3 was Highly Expressed in the Fallopian Tube,\nwhereas DMBT1 was in the Endometrium\nThere were many upregulated genes either in the\nfallopian tube or in the endometrium identified\nthrough the gene array analysis. We sought target\ngenes, which were uniquely stable, conservative,\nnot easily regulated by menstrual cycle hormones or\ncell proliferative status as the endometrial and tubal\ntissue are more or less regulated by menstrual\nhormones and such hormone-regulated genes may\nnot be representative for current study. Meanwhile,\nin a practical level, the corresponding protein\nproducts should have antibodies that are known to\nwork both in immunohistochemistry and in western\nblot assays. After screening, we identified two\ngenes, one was highly expressed in the fallopian\ntube ( FMO3) and the other in the endometrium\n(DMBT1), which matched our specified conditions.\nWe validated these two genes with real-time PCR,\nwestern blot and immunohistochemistry.\nBoth the FMO3 and DMBT1 genes follow the trend\ndifferences of the microarray results. In the 21 pairs\nof the tubal and corresponding endometrial samples\nin real-time PCR analysis, FMO3 was highly\nexpressed in the fallopian tube compared with the\npaired endometrium, with fold increment ranging\nfrom 4 to 324 (average fold change ¼ 44.38,\nPo0.001). In contrast, DMBT1 was highly expressed\nin endometrium compared with the fallopian tube\nwith fold change ranging from 5 to 456 (average fold\nchange ¼ 22.11, Po0.001). The data pool of the 21\npaired cases is shown in Figures 1a and b. Among\nthe 21 paired tubal and endometrial samples, there\nwere 17 pairs in proliferative phase and 4 pairs in\nsecretory phase of the menstrual cycle. In either\nfallopian tube or endometrial tissue, the gene\nexpression level of FMO3 and DMBT1 were in-\ndifferent between the proliferative or secretory\nphases ( P40.10). In addition, none of the 21 pairs\nof samples showed discordance with the trend\ntoward co-expression of FMO3 and DMBT1 in\nfallopian tube vs endometrial tissue.\nAmong the 21 pairs of tubal and endometrial\nsamples, 14 pairs were adequate for western blot\nanalysis. FMO3 protein expression was significantly\nhigher in the fallopian tube samples than that in the\nendometrium, with an average fold of increment\n11.05 ( P ¼ 0.006). In contrast, the DMBT1 protein\nlevel was significantly lower in the fallopian tube\n(average decreasing fold ¼ 32.08) compared with the\nexpression in the endometrium ( P ¼ 0.001). These\nresults were compatible with the findings from real-\ntime PCR validation, indicating FMO3 and DMBT1\ndo not change significantly at the transcriptional\nand post-transcriptional levels. Representative pro-\ntein expression level is shown in Figures 1c and d.\nAll 21 pairs of the tissue samples were studied for\nthe cellular location of both FMO3 and DMBT1\nby immunohistochemistry. Both biomarkers were\nmainly cytoplasmic. No nuclear stainings were\nidentified for these two genes. Of the 21 paired\nsamples, 19 (91%) showed moderately to strongly\npositive staining of FMO3 in the majority epithelial\ncells of the fallopian tube, whereas the remaining\ntwo tubal samples were weakly stained. In contrast,\nFMO3 was only weakly and focally expressed in the\nthree (14%) endometrial samples, mainly within the\nendometrial glands. There were some stromal and\nendothelial cell stainings identified with no specific\npattern. DMBT1 cellular localization was totally\ndifferent from FMO3. DMBT1 was strongly and\ndiffusely positive in the majority of glandular cells\nin all 21 endometrial samples, but not in the tubal\nsections we studied. These results will be further\npresented in the next section when ovarian endo-\nmetriosis was compared.\nFMO3 and DMBT1 Expression in Ovarian\nEndometriosis\nAfter validation of differentially expressed FMO3\nand DMBT1 in paired tubal and endometrial samples,\nwe further examined these genes and corresponding\nprotein expression in patients with ovarian endo-\nmetriosis as well as its corresponding fallopian tubal\nModern Pathology (2014) 27, 1154–1162\nOviduct contributing to ovarian endometriosis\nZ Yuan et al 1157\n\nand endometrial samples. There were a total of 35\npatients with the requisite three samples for the\nstudy. These gene expression levels were compared\nin patients with ovarian endometriosis and those\nwithout ovarian endometriosis from the data pre-\nsented above. Both FMO3 and DMBT1 expression\nlevels in the fallopian tubal and endometrial\nsamples showed no statistical difference between\nthe patients with ovarian endometriosis and those\nwithout ovarian endometriosis. In other words,\nFMO3 remained high in the fallopian tube, whereas\nDMBT1 was high in the endometrium (data not\nshown).\nAmong the 35 cases with ovarian endometriosis,\n32 paired samples were adequate for real-time PCR\nanalysis. FMO3 was highly expressed in 18 (56%) of\nthe 32 samples studied, with the fold increment\nranging from 4.9 to 123 (average fold change ¼ 7.21,\nP ¼ 0.016) compared with the endometrium. How-\never, FMO3 expression in the remaining 14 ovarian\nendometriosis samples was significantly lower than\nthat in FT ( Po0.01), but similar to the level of\nexpression in the corresponding endometrium\n(P ¼ 0.184). In contrast, the 18 patients with high\nlevel of FMO3 expression showed a significantly\nlow expression of DMBT1 in the ovarian endo-\nmetriosis lesions compared with the paired endo-\nmetrial samples (average decreasing fold ¼ 6.94,\nP ¼ 0.022). Meanwhile, DMBT1 expression showed\nno statistical differences between the ovarian endo-\nmetriosis and fallopian tube samples ( P ¼ 0.144).\nThe data are summarized in Figures 2a and b. There\nwere no statistical difference of age between the\n18 cases with high FMO3 expression and the\nremaining 14 samples.\nThe 18 ovarian endometriosis samples with high\nFMO3 and low DMBT1 expression also showed con-\ncordant level of the protein expression by western\nblot. The FMO3 expression increased ranging from\n3.3 to 58.5-fold with an average fold increment of 7.2\n(P ¼ 0.007) compared with the expression in the\nendometrium. Among the remaining 17 ovarian\nendometriosis samples, 8 showed a similar level of\nFMO3 expression to the endometrium but lower\nthan the tube, whereas 9 showed no statistical\ndifference compared with either the endometrium\nor the fallopian tube. Among the 35 paired patients,\n19 cases showed that DMBT1 protein expression in\novarian endometriosis samples was similarly low in\nthe fallopian tube, whereas significantly high in the\nendometrium. DMBT1 in the remaining 16 cases\nshowed high expression in areas of ovarian endo-\nmetriosis and the endometrium ( n ¼ 12) and no\ndifference ( n ¼ 4) between ovarian endometriosis\nand paired endometrium. Representative western\ndata are presented in Figures 2c and d.\nFigure 1 FMO3/DMBT1 expression level in paired tubal and endometrial tissue samples. Comparisons of mRNA expression levels in the\nfallopian tube and the endometrium by quantitative PCR in 21 paired samples (pooled data): FMO3 was significantly higher in the\nfallopian tube than that in the endometrium ( a, Po0.001); whereas DMBT1 showed a complete reverse pattern ( b, Po0.001).\nComparisons of protein level of expression in the fallopian tube and the endometrium by western blots in two representative paired cases\n(c, d). Similarly, FMO3 was higher in the fallopian tube than the endometrium ( c, P ¼ 0.006), whereas DMBT1 was higher in the\nendometrium ( d, P ¼ 0.001). The western bands were in the lower panel, whereas the bar graphs reflected the bands density after\nbalanced with the GAPDH control. Each experiment was conducted three times, and the data were expressed as mean ±s.e.m.\n*Statistically significant by comparing with the paired cases. E, endometrium; FT, fallopian tube.\nModern Pathology (2014) 27, 1154–1162\nOviduct contributing to ovarian endometriosis\n1158 Z Yuan et al\n\nCellular localization of these gene products was\nexamined by immunohistochemistry. Among the 35\npaired samples, 19 (54%) showed that FMO3 was\npositively staining in the cytoplasm of ovarian\nendometriosis epithelium, whereas the remaining\n16 ovarian endometriosis samples showed either\nlow expression in focal areas ( n ¼ 10) or inadequate\nexpression for analysis due to loss of glandular\nepithelia within the sections ( n ¼ 6). In terms of\nDMBT1, 17 (49%) ovarian endometriosis samples\nwere negative, whereas the remaining 18 ovarian\nendometriosis samples showed positive staining\n(n ¼ 10) or were inadequate ( n ¼ 8). Representative\npictures for the FMO3 and DMBT1 immunostain-\nings in the fallopian tube, endometrium and ovarian\nendometriosis samples are illustrated in Figure 3.\nDiscussion\nEndometriosis is a worldwide problem affecting a\nsignificant number of women in their reproductive\nage. The histogenesis of endometriosis has per-\nplexed investigators for decades. Despite various\nhypotheses, no single explanation fits all observed\nclinical manifestations.\nThe tubal fimbriated (distal) end, which ‘floats’ in\nthe pelvis over the ovarian surface and the perito-\nneum of the pouch of Douglas, has a unique and\ndelicate role in capturing the egg released from the\novary and provides a conduit for transport and\nfertilization of the egg. In the past, the attention to\nthe fallopian tube has been primarily in younger\nwomen with infertility or the potentially life-\nthreatening situation of ectopic pregnancy. 26\nRecently, the distal fallopian tube has attracted\nconsiderable attention not only as site of origin for\nserous ovarian cancer in women with BRCA muta-\ntions, but also as a distinct anatomical entity where\nthe majority of pelvic serous carcinomas apparently\ndevelop. Awareness is growing that the fallopian\ntube requires attention by both pathologists and\ngynecological surgeons. Through recent understand-\ning of the tubal origin of ovarian ‘high-grade’ and\nFigure 2 FMO3/DMBT1 expression level in paired tubal, endometrial and ovarian endometriosis samples. Comparisons of mRNA\nexpression levels in the fallopian tube, the endometrium and the foci of endometriosis by quantitative PCR in 18 paired cases (pooled\ndata): FMO3 level in fallopian tube was similar to the level of endometriosis, but both were higher than that in the endometrium ( a,\n*P ¼ 0.184 (fallopian tube vs ovarian endometriosis) and ** P ¼ 0.016 (endometrium vs ovarian endometriosis)); DMBT1 level in fallopian\ntube was similar to the level of endometriosis, but both were lower than that in the endometrium ( b,* P ¼ 0.144 (fallopian tube vs ovarian\nendometriosis) and ** P ¼ 0.022 (endometrium vs ovarian endometriosis)). Comparisons of protein level of expression in the tubal,\nendometrial and ovarian endometriosis by western blot in one representative patient ( c, d). FMO3 showed a strong band in both tubal\nand ovarian endometriosis samples, whereas barely detected in the corresponding endometrium ( c,* P ¼ 0.102 (fallopian tube vs ovarian\nendometriosis) and ** P ¼ 0.007 (endometrium vs ovarian endometriosis)); whereas DMBT1 showed a strong band only in the\nendometrial sample, not in the samples of the fallopian tube and endometriosis ( d,* P ¼ 0.221 (fallopian tube vs ovarian endometriosis)\nand **P ¼ 0.006 (endometrium vs ovarian endometriosis)). The western bands were in the lower panel, whereas the bar graphs reflected\nthe bands density after balanced with the GAPDH control. Each experiment was conducted three times, and the data were expressed as\nmean±s.e.m. E, endometrium; FT, fallopian tube.\nModern Pathology (2014) 27, 1154–1162\nOviduct contributing to ovarian endometriosis\nZ Yuan et al 1159\n\n‘low-grade’ serous carcinoma,20,27,28 we have learned\nmany biologic properties of the fallopian tube,\nwhich were previously unrecognized. Within\nwomen’s pelvis, the fallopian tube has a close\ncontact with the ovary, 20,27,29–31 the tubal epithelial\ncells are easily detached from the tubal mucosa, 27,32\nand the majority of the ovarian epithelial inclusions\nare derived from the fallopian tube. 20 We proposed\nthat ovarian endometriosis, at least partially, may be\nderived from the fallopian tube. This hypothesis of\ntubal origin of ovarian endometriosis is novel,\nwhich has not been proposed elsewhere in the past.\nThe main obstacle for us to verify this novel\nhypothesis is to challenge the most popular retro-\ngrade menstruation theory. That means we need to\nstudy both the fallopian tube and the endometrium\nas potential sources of ovarian endometriosis. With\nthis understanding, we identified differentially\nexpressed genes between the endometrium and the\nfallopian tubes by using whole-genome expression\nmicroarray. Among many differentially expressed\ngenes, FMO3 and DMBT1 turned out to be two good\nbiomarkers to test the hypothesis. FMO3 was highly\nexpressed in the tubal epithelia while low in the\npaired endometrium. In contrast, DMBT1 was high\nin the endometrium but low in tubal mucosa in\nthose 21 patients without endometriosis. These two\nmarkers were therefore highly specific and therefore\nappropriate for examining our hypothesis as FMO3\nappears to be a marker of tubal epithelia and DMBT1\nof the endometrial glandular cells. We tested the\ntubal and endometrial samples with these two\nmarkers on 32 patients with ovarian endometriosis.\nIt was found that FMO3 and DMBT1 were similarly\ndifferentially expressed in the fallopian tube and in\nthe endometrium as seen in the 21 patients without\nendometriosis. The results were quite consistent\namong the three validation methods we used in this\nstudy. However, when we analyzed FMO3 and\nDMBT1 expression in the samples of ovarian\nendometriosis, two different outcomes emerged. In\nall, 18 (56%) of the 32 ovarian endometriosis lesions\nshowed a high level of FMO3 expression but low\nwith DMBT1. In contrast, 14 (44%) endometriosis\nsamples showed a reversed expression pattern with\nthese two markers. Again, the results were similarly\nseen among the methods of real-time PCR, western\nblot and immunohistochemistry. Our findings in-\ndicate that approximately 60% of the ovarian\nendometriosis we studied is likely to be derived\nfrom the fallopian tubal epithelia, whereas about\n40% of the cases may be of endometrial origin.\nTherefore, this study supports our hypothesis that\novarian endometriosis may be at least partially\nderived from the fallopian tube.\nThe finding of possible tubal origin of ovarian\nendometriosis is interesting and may have many\nclinical impacts, although the study result is\npreliminary and the underlying mechanism is\nunclear. However, evidence that tubal epithelia are\npotentially able to form endometriosis exists. En-\ndometrialization of the fallopian tube representing\nendometrium-like tissue within the proximal end of\nthe tubal segment is commonly observed in patients\nwho underwent tubal ligation for undesired ferti-\nlity.33–35 Although we cannot exclude the possibility\nof endometrial colonization for those tubal ligation\ncases, many investigators believe that metaplasia\nfrom tubal epithelia seems more likely as a trans-\nitional area from normal looking tubal epithelia to\nminimally formed endometriosis or endometrial-\nlike tissue are commonly present. 27,36 In order to\na\nd\nFMO3\nDMBT1\nef\nbc\nFigure 3 FMO3 and DMBT1 protein detection by immunohistochemistry in the tubal, endometrial and ovarian endometriosis samples.\nBoth FMO3 and DMBT1 expression were mainly cytoplasmic. Moderately to strongly FMO3 immunoreaction was observed in the tubal\n(a) and ovarian endometriosis ( c) sections, but it was negative in the endometrium ( b). However, DMBT1 was only strongly expressed in\nthe endometrium ( e), not in the fallopian tube ( d) and the ovarian endometriosis ( f). Original magnification: /C2 40 for d, the remaining\nwere /C2 100.\nModern Pathology (2014) 27, 1154–1162\nOviduct contributing to ovarian endometriosis\n1160 Z Yuan et al\n\nform ovarian endometriosis, tubal cells have to find\na reasonable route to get onto the ovary. The easy\ndetachment of tubal epithelia makes it practical.\nThis is supported by the findings that the tubal\nepithelial cells are easily obtained for culture by\nflushing the fallopian tube.\n27,32 The process is\nfurther facilitated by close spatial anatomic rela-\ntionship between the tubal fimbria and the ovarian\nsurface and their synergistic role of ovulation and\negg capture and subsequent transportation.\n26,29\nFurthermore, ovulation induced wound formation\non the ovarian surface,37–39 pelvic inflammation,40–42\ninflammation induced ovarian surface adhesion 43,44\nand ovarian stromal proliferation under the\ninfluence of gonadotropins 39,45–47 provide all\nnecessary conditions for those tubal epithelia to\nenter the ovarian cortex. Actually, this common\nprocess from morphologic perspective has long been\ndescribed as ‘endosalpingiosis’.\n48–50 Such epithelial\ninclusions found in the ovary are also called as\novarian cortical or epithelial inclusions.\n51,52\nThe question remains how endosalpingiosis or\novarian epithelial inclusions or the tubal type\nepithelia are transformed into endometriosis (en-\ndometrial type cells in morphology). One of our\nrecent studies regarding the cell origin of ovarian\nserous cancers have demonstrated that ovarian\nepithelial inclusions are mainly derived from\nfallopian tube,\n20 supporting the terminology of\nendosalpingiosis, a simple but direct morpholo-\ngically described entity. In addition, the transforma-\ntion from endosalpingiosis or ovarian epithelial\ninclusions can be explained by metaplasia, a\nprocess commonly seen in mu ¨ llerian system.\n53\nThis interpretation is supported by our previous\nobservation of ovarian initial endometriosis. 22\nInitial endometriosis within the ovary defines the\nmorphologic transition of ovarian epithelial inclu-\nsions with half of the glandular epithelia showing\nthe earliest morphologic changes of endometriosis.\n22\nSuch a morphologic transition is certainly not\ninterpretable by retrograde menstruation theory,\nwhereas a metaplasia from tubal epithelia instead\nof from mesothelia via unidentified factors is more\nlikely applicable in this situation. It is interesting to\nnote that tubal stromal cells were positive for FMO3\nstaining (Figure 3a) but negative in the stroma of\novarian endometriosis (Figure 3c). The reasonable\ninterpretation for this observation is that the\nendometriotic epithelial cells were likely derived\nfrom tubal epithelia, whereas the associated stromal\ncells may be derived from ovarian stroma in a de\nnovo process, albeit the phenomenon remains\nclarified in near future.\nThe novel findings of this study may provide\nanother angle to think the histogenesis of ovarian\nendometriosis and such different view may ulti-\nmately shed light of our understanding the patho-\ngenesis of endometriosis and lead us to develop a\npractical way of endometriosis prevention. How-\never, again the findings remain primitive and\ndescriptive. Additional studies from different\nperspectives are necessary to strengthen our under-\nstanding of fallopian tubal origin of ovarian\nendometriosis.\nAcknowledgments\nDr Zeng Yuan is a PhD candidate co-trained at\nUniversity of Arizona, USA and Shandong Univer-\nsity, China. The project was supported in part by\nBetter Than Ever Fund, Arizona Cancer Center\nSupporting Grant, P30 CA23074 from Arizona\nCancer Center and Department of Pathology,\nUniversity of Arizona Startup fund to WZ, and by\nNational Natural Science Foundation of China (no.\n30872738; 81272857), Taishan Scholars Foundation\nof Shandong Province to BK (no. ts20070743).\nDisclosure/conflict of interest\nThe authors declare no conflict of interest.\nReferences\n1 Rogers PA, D’Hooghe TM, Fazleabas A, et al. Defining\nfuture directions for endometriosis research: workshop\nreport from the 2011 World Congress of Endometriosis\nin Montpellier, France. Reprod Sci 2013;20:483–499.\n2 Giudice LC, Kao LC. Endometriosis. Lancet 2004;364:\n1789–1799.\n3 Woodward PJ, Sohaey R, Mezzetti TP Jr. Endometrio-\nsis: radiologic-pathologic correlation. Radiographics\n2001;21:193–216; questionnaire 288-194.\n4 Sainz de la Cuesta R, Eichhorn JH, Rice L W, et al.\nHistologic transformation of benign endometriosis to\nearly epithelial ovarian cancer. Gynecol Oncol 1996;\n60:238–244.\n5 Jenkins S, Olive DL, Haney AF . Endometriosis:\npathogenetic implications of the anatomic distribu-\ntion. Obstet Gynecol 1986;67:335–338.\n6 van der Linden PJ. Theories on the pathogenesis of\nendometriosis. Hum Reprod 1996;11(Suppl 3):53–65.\n7 Nap AW, Groothuis PG, Demir AY, et al. Pathogenesis\nof endometriosis. Best Pract Res Clin Obstet Gynaecol\n2004;18:233–244.\n8 Ridley JH. The histogenesis of endometriosis. Obstet\nGynecol Surv 1968;23:1–35.\n9 Halme J, Hammond MG, Hulka JF , et al. Retrograde\nmenstruation in healthy women and in patients with\nendometriosis. Obstet Gynecol 1984;64:151–154.\n10 Schrodt GR, Alcorn MO, Ibanez J. Endometriosis of the\nmale urinary system: a case report. J Urol 1980;\n124:722–723.\n11 Witz CA. Current concepts in the pathogenesis of\nendometriosis. Clin Obstet Gynecol 1999;42:566–585.\n12 Gazvani R, Templeton A. New considerations for the\npathogenesis of endometriosis. Int J Gynaecol Obstet\n2002;76:117–126.\n13 Javert CT. The spread of benign and malignant\nendometrium in the lymphatic system with a note on\ncoexisting vascular involvement. Am J Obstet Gynecol\n1952;64:780–806.\nModern Pathology (2014) 27, 1154–1162\nOviduct contributing to ovarian endometriosis\nZ Yuan et al 1161\n\n14 Van Schil PE, Vercauteren SR, Vermeire PA, et al.\nCatamenial pneumothorax caused by thoracic endo-\nmetriosis. Ann Thorac Surg 1996;62:585–586.\n15 Cassina PC, Hauser M, Kacl G, et al. Catamenial\nhemoptysis. Diagnosis with MRI. Chest 1997;111:\n1447–1450.\n16 Reid GD, Kowalski D, Cooper MJ, et al. Hepatic\nendometriosis: a case report and review of the\nliterature. Aust N Z J Obstet Gynaecol 2003;43:87–89.\n17 Sasson IE, Taylor HS. Stem cells and the pathogenesis\nof endometriosis. Ann N Y Acad Sci 2008;1127:\n106–115.\n18 Figueira PG, Abrao MS, Krikun G, et al. Stem cells in\nendometrium and their role in the pathogenesis of\nendometriosis. Ann N Y Acad Sci 2011;1221:10–17.\n19 Gardner GH, Greene RR, Ranney B. The histogenesis of\nendometriosis recent contributions. Obstet Gynec\n1953;1:615–637.\n20 Li J, Abushahin N, Pang S, et al. Tubal origin of\n‘ovarian’ low-grade serous carcinoma. Mod Pathol\n2011;24:1488–1499.\n21 Li J, Fadare O, Xiang L, et al. Ovarian serous\ncarcinoma: recent concepts on its origin and carcino-\ngenesis. J Hematol Oncol 2012;5:8.\n22 Zheng W, Li N, Wang J, et al. Initial endometriosis\nshowing direct morphologic evidence of metaplasia in\nthe pathogenesis of ovarian endometriosis. Int J\nGynecol Pathol 2005;24:164–172.\n23 Wang JL, Lin YW, Chen HM, et al. Calcium prevents\ntumorigenesis in a mouse model of colorectal cancer.\nPLoS One 2011;6:e22566.\n24 Wei W, Kong B, Qu X. Alteration of HGF and TSP-1\nexpression in ovarian carcinoma associated with\nclinical features. J Obstet Gynaecol Res 2012;38:57–64.\n25 Hong S, Li X, Zhao Y, et al. 53BP1 suppresses tumor\ngrowth and promotes susceptibility to apoptosis of\novarian cancer cells through modulation of the Akt\npathway. Oncol Rep 2012;27:1251–1257.\n26 Dietl J, Wischhusen J, Hausler SF . The post-reproduc-\ntive fallopian tube: better removed? Hum Reprod\n2011;26:2918–2924.\n27 Kurman RJ, Shih IeM. The origin and pathogenesis of\nepithelial ovarian cancer: a proposed unifying theory.\nAm J Surg Pathol 2010;34:433–443.\n28 Roh MH, Kindelberger D, Crum CP . Serous tubal\nintraepithelial carcinoma and the dominant ovarian\nmass: clues to serous tumor origin? Am J Surg Pathol\n2009;33:376–383.\n29 Eddy CA, Pauerstein CJ. Anatomy and physiology\nof the fallopian tube. Clin Obstet Gynecol 1980;23:\n1177–1193.\n30 Gordts S, Campo R, Rombauts L, et al. Endoscopic\nvisualization of the process of fimbrial ovum retrieval\nin the human. Hum Reprod 1998;13:1425–1428.\n31 Piek JM, Kenemans P , Verheijen RH. Intraperitoneal\nserous adenocarcinoma: a critical appraisal of three\nhypotheses on its cause. Am J Obstet Gynecol 2004;\n191:718–732.\n32 Piek JM, van Diest PJ, Zweemer RP , et al. Tubal ligation\nand risk of ovarian cancer. Lancet 2001;358:844.\n33 Ambekar SV , Mudbhatkal NS, Kothare SN.\nPost-salpingectomy endometriosis (endosalpingiosis);\na report of two cases. J Postgrad Med 1965;11:\n141–144.\n34 deHoop TA, Mira J, Thomas MA. Endosalpingiosis and\nchronic pelvic pain. J Reprod Med 1997;42:613–616.\n35 Chakrabarti I, Ghosh N. Post-salpingectomy endome-\ntriosis: an under-recognized entity. J Mid-life Health\n2010;1:91–92.\n36 Nishida M, Watanabe K, Sato N, et al. Malignant\ntransformation of ovarian endometriosis. Gynecol\nObstet Invest 2000;50(Suppl 1):18–25.\n37 Fathalla MF . Incessant ovulation–a factor in ovarian\nneoplasia? Lancet 1971;2:163.\n38 Casagrande JT, Louie EW, Pike MC, et al. ‘Incessant\novulation’ and ovarian cancer. Lancet 1979;2:170–173.\n39 Chene G, Penault-Llorca F , Le Bouedec G, et al.\nOvarian epithelial dysplasia after ovulation induction:\ntime and dose effects. Hum Reprod 2009;24:132–138.\n40 King SM, Hilliard TS, Wu L Y, et al. The impact of\novulation on fallopian tube epithelial cells: evaluating\nthree hypotheses connecting ovulation and serous\novarian cancer. Endocr Relat Cancer 2011;18:627–642.\n41 Ness RB, Cottreau C. Possible role of ovarian epithelial\ninflammation in ovarian cancer. J Natl Cancer Inst\n1999;91:1459–1467.\n42 Ness RB, Modugno F . Endometriosis as a model for\ninflammation-hormone interactions in ovarian and\nbreast cancers. Eur J Cancer 2006;42:691–703.\n43 Risch HA, Howe GR. Pelvic inflammatory disease and\nthe risk of epithelial ovarian cancer. Cancer Epidemiol\nBiomarkers Prev 1995;4:447–451.\n44 Balkwill F , Mantovani A. Inflammation and cancer:\nback to Virchow? Lancet 2001;357:539–545.\n45 Chakravarti S, Collins WP , Forecast JD, et al. Hormonal\nprofiles after the menopause. Br Med J 1976;2:784–787.\n46 Vanderhyden BC. Loss of ovarian function and the risk\nof ovarian cancer. Cell Tissue Res 2005;322:117–124.\n47 Wong AS, Auersperg N. Ovarian surface epithelium:\nfamily history and early events in ovarian cancer.\nReprod Biol Endocrinol 2003;1:70.\n48 Burmeister RE, Fechner RE, Franklin RR. Endosalpin-\ngiosis of the peritoneum. Obstet Gynecol 1969;34:\n310–318.\n49 Tutschka BG, Lauchlan SC. Endosalpingiosis. Obstet\nGynecol 1980;55:57S–60S.\n50 Stock RJ. Postsalpingectomy endometriosis: a reassess-\nment. Obstet Gynecol 1982;60:560–570.\n51 Scully RE. Pathology of ovarian cancer precursors.\nJ Cell Biochem Suppl 1995;23:208–218.\n52 Resta L, Russo S, Colucci GA, et al. Morphologic\nprecursors of ovarian epithelial tumors. Obstet Gyne-\ncol 1993;82:181–186.\n53 Fukunaga M, Ushigome S. Epithelial metaplastic\nchanges in ovarian endometriosis. Mod Pathol 1998;\n11:784–788.\nSupplementary Information accompanies the paper on Modern Pathology website (http://www.nature.com/\nmodpathol)\nModern Pathology (2014) 27, 1154–1162\nOviduct contributing to ovarian endometriosis\n1162 Z Yuan et al","source_license":"Public-Domain","license_restricted":false}