{"paper_id":"313387dd-b466-4bf5-b91e-697129983707","body_text":"ENDOMETRIOSIS: ORIGINAL ARTICLE\nAssociations Between Endometriosis and Gut Microbiota\nAgnes Svensson1 & Louise Brunkwall2 & Bodil Roth1 & Marju Orho-Melander2 & Bodil Ohlsson1\nReceived: 4 November 2020 / Accepted: 21 February 2021\n# The Author(s) 2021\nAbstract\nThe gut microbiota has been associated with many diseases, including endometriosis. However, very few studies have been\nconducted on this topic in human. This study aimed to investigate the association between endometriosis and gut microbiota.\nWomen with endometriosis (N=66) were identified at the Department of Gynaecology and each patient was matched with three\ncontrols (N=198) from the general population. All participants answered questionnaires about socioeconomic data, medical\nhistory, and gastrointestinal symptoms and passed stool samples. Gut bacteria were analyzed using 16S ribosomal RNA se-\nquencing, and in total, 58 bacteria were observed at genus level in both patients with endometriosis and controls. Comparisons of\nthe microbiota between patients and controls and within the endometriosis cohort were performed. Both alpha and beta diversities\nwere higher in controls than in patients. With the false discovery rate q<0.05, abundance of 12 bacteria belonging to the classes\nBacilli, Bacteroidia, Clostridia, Coriobacteriia, and Gammaproteobacter differed significantly between patients and controls.\nDifferences observed between patients with or without isolated ovarian endometriosis, involvement of the gastrointestinal tract,\ngastrointestinal symptoms, or hormonal treatment disappeared after calculation with false discovery rate. These findings indicate\nthat the gut microbiota may be altered in endometriosis patients.\nKeywords Endometriosis . Gastrointestinal symptoms . Gut microbiota . Pathophysiology\nIntroduction\nEndometriosis is an inflammatory, estrogen-dependent dis-\nease defined by presence of endometrial tissue outside the\nuterine cavity, affecting approximately 6 –10% of reproduc-\ntive women [ 1, 2]. Besides gynecological symptoms [ 3, 4],\ngastrointestinal symptoms affect up to 90% of patients with\nendometriosis [5]. The most common gastrointestinal symp-\ntom is bloating, followed by nausea, constipation, diarrhea,\nand vomiting [5, 6].\nThe gastrointestinal tract is a complex ecosystem with a\nsymbiosis of food molecules, gut mucosal cells, immune sys-\ntem cells, and microorganisms. It is a dynamic environment,\nand the commensal bacteria, or microbiota, are contently\nchanging [7]. The gut microbiota is thought to play a major\nrole in the maintenance of health and development of disease\nand has through inflammatory and metabolic changes been\nproven to affect conditions both inside and outside of the\ngastrointestinal tract [8–10].\nSystemic levels of estrogen in post-menopausal women\nhave been associated with fecal microbiome richness and\nlevels of fecal Clostridia taxa [ 11]. Therefore, the gut micro-\nbiota has been suggested to have an impact on estrogen levels\nin men and post-menopausal women and to be involved in\nestrogen-dependent diseases [11, 12]. Higher estrogen levels\nstimulate epithelial proliferation in the female reproductive\ntract and have been shown to drive diseases such as endome-\ntriosis and endometrial cancer [ 13]. Recent studies have\nshown that the gut microbiota is a major regulator of inflam-\nmatory processes outside the gastrointestinal tract [14], factors\nwhich may be involved in the pathogenesis of endometriosis\n[15]. The cytoplasmic protein AXIN1 is involved in the reg-\nulation of apoptosis and has been reported to be a potential\nnew biomarker for endometriosis [ 16], with correlations to\nclinical data such as gastrointestinal symptoms and hormone\ntreatment [17].\nDue to the impact of immunological and hormonal changes\nin patients with endometriosis, and the impact of gut microbi-\nota on immune and estrogen responses, it has been\n* Bodil Ohlsson\nBodil.ohlsson@med.lu.se\n1 Department of Internal Medicine, Skåne University Hospital, Lund\nUniversity, Jan Waldenströms street 15, floor 5,\n20502 Malmö, Sweden\n2 Department of Clinical Sciences in Malmö, Lund University,\nMalmö, Sweden\nhttps://doi.org/10.1007/s43032-021-00506-5\n/ Published online: 3 March 2021\nReproductive Sciences (2021) 28:2367–2377\n\nhypothesized that the gut microbiota is involved in the patho-\ngenesis of endometriosis [18]. The primary aim of the present\nstudy was to investigate the gut microbiota in endometriosis in\ncomparison with healthy controls. The secondary aim was to\nexamine any differences regarding microbiota abundance\nwithin the endometriosis cohort, dependent on disease locali-\nzation, symptoms, or treatment.\nMaterials and Methods\nStudy Design\nPatients with endometriosis ( N=66) were recruited from the\nDepartment of Gynaecology at Skåne University Hospital and\nwere matched with three controls each from the Malmö\nOffspring Study (MOS). The study participants answered a\nquestionnaire concerning sociodemographic data and medical\nhistory, completed the visual analog scale for irritable bowel\nsyndrome (VAS-IBS), and passed stool samples. The gut mi-\ncrobiota was identified at genus level using 16S rRNA se-\nquencing. Comparisons in gut microbiota were performed be-\ntween patients and controls, and within the patient cohort,\nusing two-tailed Mann-Whitney U test, Fisher ’s exact test,\nSpearman’s correlations test, Shannon diversity index, and\nBray Curtis dissimilarity index.\nPatients\nPatients were recruited from the Department of Gynaecology\nat Skåne University Hospital, Malmö, according to the ICD-\n10 classification of endometriosis, N80. The main inclusion\ncriteria were to have a diagnosis of endometriosis, confirmed\nby laparoscopy or laparotomy. General inclusion criteria were\nan age above 18 years and to comprehend the Swedish or\nEnglish languages. Exclusion criteria were an uncertain diag-\nnosis of endometriosis, current pregnancy, diagnosed inflam-\nmatory bowel disease (IBD), living far from the hospital, and\nmultiple or severe somatic or psychiatric comorbidities.\nBetween September 2016 and March 2017, 266 women who\nfulfilled the inclusion criteria were identified. Of these, 196\nwomen were excluded because they were as follows: unwill-\ning to participate (N=162), had moved too far from the hospi-\ntal (N=23), had a non-surgically confirmed diagnosis ( N=7),\nor an uncertain diagnosis of endometriosis ( N=2). This re-\nduced the number of women to 72. Out of these, 66 women\npassed stool samples and were thereby included in the present\nstudy.\nControls\nThe controls were recruited from 2644 individuals who had\npreviously been extracted from the MOS [19], which consists\nof descendants to participants in the Malmö Diet and Cancer\nCardiovascular Cohort (MDC-CC). The recruitment of partic-\nipants to the MDC-CC and the MOS took place in the 1990s\nand 2010s, respectively [20, 21]. Each case was matched with\nthree controls according to sex (only women), age (± 730\ndays), body mass index (BMI) (± 2 BMI units), and smoking.\nThe participants from MOS who had not answered question-\nnaires and passed stool samples or were diagnosed with celiac\ndisease, Crohn ’s disease, ulcerative colitis, irritable bowel\nsyndrome (IBS), or lactose intolerance were excluded from\nthe matching process.\nStudy Questionnaires\nThe patients with endometriosis answered questions regarding\ntheir endometriosis-associated symptoms such as onset of\nsymptoms, trigger factors, and treatment. They also answered\na questionnaire regarding education, occupation, marital sta-\ntus, smoking habits, alcohol habits, physical activity, medical\nhistory, and pharmacological treatments. The participants in\nthe MOS answered a similar questionnaire. They were also\nasked the following question: “Have you experienced bowel\nsymptoms during the last 2 weeks?” All participants from\nMOS who had answered “yes” were excluded from the pres-\nent study.\nVAS-IBS\nThe VAS-IBS is a psychometrically validated questionnaire\nused to estimate gastrointestinal symptoms in patients with\nfunctional bowel disease [ 22]. With the VAS-IBS question-\nnaire, the patients estimated the severity of symptoms over the\nlast 2 weeks regarding abdominal pain; constipation; diarrhea;\nbloating and flatulence; vomiting and nausea; and intestinal\nsymptom’s influence on daily life. The VAS-IBS question-\nnaire measures each of the symptoms on a continuous scale\nfrom 0 to 100 mm where 0 represents no problems and 100\nrepresents very severe problems. The scales were inverted\nfrom the original scales [ 22]. Reference values from 52\nhealthy women were used as controls [ 23].\nGut Microbiota and Laboratory Analyses\nStool samples were collected by the patients and controls at\nhome in sterile tubes (Sarstedt, Numbrecht, Germany) and put\nin the freezer until they were brought to the lab. The samples\nwere kept at –80 °C until the extraction of microbial DNA.\nMicrobial DNA was extracted from the stool samples using\na QIAamp column Stool Kit. The V1 –V3 regions of the 16S\nribosomal RNA gene were pairwise (300*2 base pairs) ampli-\nfied and sequenced using a HiSeq Illumina at GATC Biotech\n(Constance, Germany). The sequences were stored as fastq\nfiles which were aligned by FLASH and binned together to\n2368 Reprod. Sci. (2021) 28:2367–2377\n\noperational taxonomic units (OTUs) using QIIME [ 24, 25].\nThe sequences were then matched with the reference database\nGreengenes and classified at genus level. In total, 937,892,146\nreads, with an average of 434,008 reads per sample, were\nincluded in the analysis. Finally, the data was normalized\nusing the cumulative sum scaling with the R package\nmetagenomSeq.\nIn total, 64 bacteria at genus level occurred in the control\ngroup and 66 occurred in the group of patients with endome-\ntriosis. Bacteria that only occurred in <10 of the samples were\nexcluded, leaving 58 bacteria at genus level in the control\ngroup and 62 in the patient group, and thus, 58 bacteria were\nincluded in the statistical analyses between patients and con-\ntrols, and 62 bacteria in calculations within the endometriosis\ncohort.\nLevels of plasma AXIN1 and fecal calprotectin were ana-\nlyzed by ELISA as previously described by Dihm et al. [ 17].\nData Categorization\nSmoking habits were divided into current smoking and no\ncurrent smoking, regardless of previous smoking habits.\nAlcohol intake was divided into <1 or ≥1 standard glass per\nweek. Physical activity was divided into <1 or ≥1hp e rw e e k\nof activity which lead to breathlessness. Hormone treatment\nwas divided into current treatment or no current treatment,\nregardless of previous treatment. Hormonal treatment includ-\ned estrogen, combined oral contraceptives, progestin, and\ngonadotropin-releasing hormone (GnRH) analogs.\nLocalization of endometriosis lesions were divided into iso-\nlated ovarian lesions or spread to any other location and bowel\ninvolvement or no bowel involvement. Gastrointestinal symp-\ntoms were divided into having symptoms or not having any\nsymptoms specified on the VAS-IBS scales, i.e., abdominal\npain; diarrhea; constipation; bloating and flatulence; vomiting\nand nausea; and influence of intestinal symptom’s on daily life\nthe last 2 weeks, considered together as one value. If none of\nthe values exceeded a predetermined value determined in a\nprevious study of healthy female volunteers, the patient was\ncategorized as having no symptoms [23].\nStatistical Analyses\nStatistical analyses were performed using the software SPSS©\nstatistical computer package version 26 for Windows. Since\nthe distribution of the quantitative data was skewed, descrip-\ntive statistics were calculated by the Mann-WhitneyU test and\nSpearman’s correlation test. Fisher ’s exact test was used for\ndichotomous variables. A sensitivity analysis was performed\nwhere all patients and controls who had received antibiotic\ntreatment in the last 6 months prior to inclusion in the study\nwere removed.\nBeta diversity was calculated using the Bray-Curtis dissim-\nilarity index to detect differences in microbiota composition\namong the groups. Beta diversity was calculated using\nvegdist; further statistical difference for dissimilarity index\nwas tested with Adonis, all within the R package vegan.\nAlpha diversity was tested using the Shannon diversity index\nto analyze diversity of genus among the samples. Alpha di-\nversity was calculated usingdiversity. Furthermore, a variance\ntest (ANOVA) was performed [26].\nValues are presented as median and interquartile range or\nnumber and percentage. Q-values are thep-values adjusted for\nFDR set at 5% according to the Benjamini-Hochberg method,\nto adjust for multiple comparisons, and considered our main\nresults [27].\nResults\nBasal Characteristic\nA total of 66 women with endometriosis and 198 controls\nwere included in the study, who showed similar basal\ncharacteristics (Table 1). The median age of the patients\nwas 37.8 (32.8– 43.3) years and the median BMI was 25.0\n(22.0–28.0) kg/m 2. The vast majority of the women had\nan education from secondary school or university (95.4%)\nand were either studying or working (83.4%). Most of the\npatients were non-smokers (84.8%) and drank less than 1\nstandard glass of alcohol/week (63.6%). Of all, 27 pa-\ntients (40.9%) had ≥ 1 h/week of physical activity which\nled to breathlessness. A minority of the patients (28.8%)\nwere living alone. More patients (62.1 vs. 51.5%) than\ncontrols (8.1 vs. 17.2%) were currently treated with hor-\nmone therapy or analgesic drugs (non-steroidal anti-\ninflammatory drugs, opioids, paracetamol) (Table 1).\nEndometriosis Characteristics\nTwenty-seven patients (40.9%) had isolated ovarian endome-\ntriosis and 18 patients (27.3%) had involvement of the gastro-\nintestinal tract (Supplementary Table 1). Of the 41 patients\nwho were currently treated with hormone therapy, 20 patients\n(48.8%) were treated with estrogen or combined oral contra-\nceptives, 19 patients (46.3%) were treated with progestin, and\n8 patients (19.5%) were treated with GnRH analogs. The ma-\njority of the patients (86.4%) had suffered from gastrointesti-\nnal symptoms over the last 2 weeks prior to inclusion in the\nstudy. There was no difference in gastrointestinal symptoms\nbetween patients with isolated ovarian lesions or spread le-\nsions ( p=0.71), gastrointestinal tract involvement or not\n(p=1.00), or with or without hormone treatment ( p=0.47).\nThere was no difference in hormone treatment between pa-\ntients with isolated ovarian lesions or spread lesions (p=0.31)\n2369Reprod. Sci. (2021) 28:2367–2377\n\nor gastrointestinal tract involvement or not (p=0.57). The me-\ndian value of plasma AXIN1 was 390.0 (357.5 –420.0) pg/ml\nand the median value of feces calprotectin was 25.00 (25.00–\n29.50) mg/kg. Of the patients, a total of 8 women (12.1%) had\nTable 1 Basal characteristics in\nendometriosis patients and\ncontrols\nVariables Controls N=198 Patients N=66 p-value\nAge, years 37.0 (32.0–44.0) 37.8 (32.8 –43.3) 0.88\nBMI, kg/m2 24.7 (22.1–27.5) 25.0 (22.0 –28.0) 0.70\nEducation level, N (%) 1.00\nMissing value 2 1\nGraduated primary 8 (4.1) 2 (3.0)\nGraduated secondary 79 (39.9) 27 (40.9)\nGraduated university 109 (55.1) 36 (54.5)\nOccupation, N (%) 0.05\nMissing value 14 1\nFull time 105 (53.0) 33 (50.0)\n51–99% 49 (24.7) 10 (15.2)\n1–50% 15 (7.6) 7 (10.6)\nSick or early retirement 3 (1.5) 5 (7.6)\nUnemployed 4 (2.0) 6 (9.1)\nStudent 8 (4.0) 5 (7.6)\nCurrent smoking, N (%) 30 (15.2) 10 (15.2) 1.00\nAlcohol intake ≥ 1 glass/week, N (%) 71 (35.9) 24 (36.4) 1.00\nPhysical activity ≥ 1 h/week, N (%) 93 (47.0) 27 (40.9) 0.48\nLives alone, N (%) 44 (22.2) 19 (28.8) 0.18\nHormone treatment, N (%) 16 (8.1) 41 (62.1) <0.001\nMissing value 1\nAntibiotic treatment last 6 months, N (%) 29 (14.6) 12 (18.2) 0.56\nMissing value 1\nAnalgesic treatment, N (%) 34 (17.2%) 34 (51.5%) <0.001\nMissing value 6\nVisual analog scale for irritable bowel syndrome\nMissing value 1\nAbdominal pain (mm) 47 (13–72)\nReference values 5 (1–15)\nConstipation (mm) 28 (2–60)\nReference values 9 (1–22)\nDiarrhea (mm) 17 (2–55)\nReference values 3 (0–10)\nBloating and flatulence 62 (20–76)\nReference values 14 (1–29)\nVomiting and nausea (mm) 15 (2–50)\nReference values 2 (0–3)\nPsychological well-being (mm) 37 (13–62)\nReference values 4 (0–16)\nIntestinal symptoms influence on daily life (mm) 52 (17–80)\nReference values 2 (0–18)\nGastrointestinal symptoms were assessed by the visual analog scale for irritable bowel syndrome, 0 –100mm,\nwhere 0 mm represents no symptoms and 100 mm maximal symptoms [ 22]. Reference values from healthy\ncontrols are shown [ 23]. Values are presented as median (interquartile range) or numbers (percentage). Mann-\nWhitney U te st or Fisher ’s exact test. p-values <0.05 were considered statistically significant\nBMI body mass index\n2370 Reprod. Sci. (2021) 28:2367–2377\n\nbeen born by caesarean section. There were no significant\ndifferences regarding endometriosis characteristics between\nthe subgroups (data not shown).\nBacterial Analysis\nThe adonis test showed a significantly higher beta diversity in the\ncontrol group compared to the endometriosis group. However,\nR2 was very small (0.02) (Fig. 1). The ANOVA test showed a\nsignificantly higher alpha diversity, p=4.9e−05, in the control\ngroup compared to the endometriosis group (Fig.2).\nNineteen gut bacteria at genus level differed in abundance\nbetween endometriosis patients and controls. With the FDR\nset at 0.05, this number was reduced to 12 bacteria (Fig. 3,\nTable 2). These bacteria belonged to the classes Bacteroidia\n(N=4), Clostridia (N=4), Coriobacteriia (N=2), Bacilli (N=1),\nand Gammaproteobacter ( N=1). Two bacteria belonging to\nthe Bacteroidia class ( Bacteroides and Parabacteroides)a n d\ntwo belonging to the Clostridia class ( Oscillospira and\nCoprococccus) were observed in higher abundance in pa-\ntients, and two other bacteria in the Bacteroidia\n(Paraprevotella and one unidentified) and Clostridia\n(Lachnospira and one unidentified) classes were observed in\nlower abundance, compared to controls. Genus belonging to\nthe Bacilli (Turicibacter) and the Coriobacteriia (unidentified)\nclasses were found in lower abundance, whereas an unidenti-\nfied genus in the class of Gammaproteobacter was found in\nhigher abundance, compared to the controls (Fig. 3,T a b l e2).\nPatients with isolated ovarian endometriosis had a\nhigher abundance of one unidentified genus and\nLachnobacterium belonging to the Clostridia class and\nAdlercreutzia belonging to the Coriobacteriia class, com-\npared to those with spread disease (Table 3). Patients with\nendometrial involvement of th e gastrointestinal tract had a\nhigher abundance of Lactococcus belonging to the class\nBacilli compared to them without involvement (Table 4).\nEndometriosis patients with gastrointestinal symptoms\nhad a lower abundance of SMB53 in the Clostridia class,\nand lower and higher abundance of Odoribacter and\nPrevotella , respectively, belonging to the Bacteroidia\nclass, compared to those without symptoms (Table 5).\nWhen the abundance of bacteria and the degree of symp-\ntoms were compared on bacteria which differed between\npatients with and without gastrointestinal symptoms, there\nwas a correlation between Prevotella and the symptoms\nconstipation ( R=0.307, p=0.014); bloating and flatulence\n(R=0.297, p=0.016); and vomiting and nausea ( R=0.295,\np=0.017). Patients with hormonal treatment had a higher\nabundance of Blautia and Ruminococcus belonging to the\nClostridia class, and Butyricimonas in\n the Bacteroidia\nclass, compared with those without treatment (Table 6).\nThere was a correlation between levels of fecal\ncalprotectin and the abundance of Ruminococcus\n(R=0.260, p=0.038), whereas pla sma AXIN1 levels did\nnot correlate with any bacteria abundance (data not\nshown). With the FDR set at 0.05, these results lost sig-\nnificance, and there were no significant differences of\nmicrobiota abundance within the cohort depending on dis-\nease localization, symptoms, or hormone treatment\n(Tables 3, 4, 5,a n d6 ). There was no difference in gut\nbacteria between patients with and without current anal-\ngesic treatment (data not shown).\nFig. 1 Plot visualizing the beta\ndiversity (Bray-Curtis\ndissimilarity index) of gut micro-\nbiota, colored by healthy controls\n(1) and patients with endometri-\nosis (2)\n2371Reprod. Sci. (2021) 28:2367–2377\n\nSensitivity Analysis\nAfter exclusion of all participants who had received antibiotic\ntreatment in the last 6 months, 17 bacteria differed between the\ngroups in the initial calculation. After FDR adjustment, only\nthree bacteria with a significant difference in abundance be-\ntween patients and controls were detected, namely\nLachnospira, Oscillospira , and a genus in the order\nBacteroidales (Supplementary Table 2). In patients with sole\novarian involvement, difference in abundance of Prevotella\nFig. 2 Boxplot of alpha diversity\n(Shannon diversity index) for gut\nmicrobiota in healthy controls (1)\nand patients with endometriosis\n(2). p-values <0.05 were consid-\nered statistically significant\nFig. 3 Stacked bar plots of the 19 genus (mean relative abundance) from Table2 that significantly differed between healthy (group 1) and patients with\nendometriosis (group 2)\n2372 Reprod. Sci. (2021) 28:2367–2377\n\nwas gained whilst those of Lachnobacterium and\nAdlercreutzia were lost, compared to those with spread endo-\nmetriosis (Supplementary Table 3 ), whereas calculations re-\ngarding gastrointestinal i nvolvement were unaffected\n(Supplementary Table 4 ). In patients with gastrointestinal\nsymptoms, difference in the abundance of Turicibacter was\ngained whilst those of Odoribacter and Prevotella were lost,\ncompared to those without gastrointestinal symptoms\n(Supplementary Table 5 ). Patients with current hormonal\ntreatment had a difference in abundance of a genus in the\nfamily S247, compared to those without treatment, whilst dif-\nferences of Ruminococcus and Butyricimonas disappeared\n(Supplementary Table 6). The differences within the endome-\ntriosis cohort lost significance after FDR adjustment\n(Supplementary Table 3–6).\nDiscussion\nGenerally, the overall diversity of gut microbiota was signif-\nicantly higher among controls compared to patients with en-\ndometriosis. Especially, the alpha diversity differed consider-\nably whilst the beta diversity was only marginally higher in\ncontrols than in endometriosis patients. There were differ-\nences in abundance of 12 genus belonging to the classes\nBacilli, Bacteroidia, Clostr idia, Coriobacteriia, and\nGammaproteobacter between endometriosis patients and con-\ntrols, without any significant differences within the endome-\ntriosis cohort after FDR adjustments.\nA systematic review from 2020 identified 13 clinical stud-\nies that investigated the connection between endometriosis\nand the microbiome [ 28]. Out of the 13 studies, only six had\nTable 2 Bacteria with significant difference between endometriosis patients and controls\nBacteria Controls N=198 Patients N=66 p-value Q-value\ng__Paraprevotella; f__Paraprevotellaceae; o__Bacteroidales; c__Bacteroidia 0.71 (0.00 –4.70) 0.00 (0.00 –1.11) <0.001 0.00058\ng__Adlercreutzia; f__Coriobacteriaceae; o__Coriobacteriales; c__Coriobacteriia 6.76 (4.91 –8.97) 5.15 (3.10 –7.31) <0.001 0.00029\ng__f__o__Bacteroidales; c__Bacteroidia 0.63 (0.00–2.69) 0.00 (0.00 –0.50) <0.001 0.00019\ng__Lachnospira; f__Lachnospiraceae; o__Clostridiales; c__Clostridia 12.43 (11.60–13.31) 3.47 (1.34 –4.88) <0.001 0.00015\ng__Oscillospira; f__Ruminococcaceae; o__Clostridiales; c__Clostridia 10.67 (9.81–11.62) 11.79 (10.60 –12.53) <0.001 0.00012\ng__f__Coriobacteriaceae; o__Coriobacteriales; c__Coriobacteriia 8.24 (6.72–9.45) 6.95 (5.25 –8.65) 0.001 0.0096\ng__Bacteroides; f__Bacteroidaceae; o__Bacteroidales; c__Bacteroidia 15.29 (14.25–16.45) 16.08 (15.14 –17.26) 0.001 0.0083\ng__Parabacteroides; f__Porphyromonadaceae; o__Bacteroidales; c__Bacteroidia 11.27 (9.98 –12.47) 11.92 (10.95 –13.20) 0.001 0.0073\ng__f__o__SHA98; c__Clostridia 2.63 (0.00–5.70) 0.00 (0.00 –4.01) 0.004 0.026\ng__f__Enterobacteriaceae; o__Enterobacteriales; c__Gammaproteobacter 3.28 (1.06 –5.56) 4.38 (2.30 –7.16) 0.007 0.041\ng__Turicibacter; f__Turicibacteraceae; o__Turicibacterales; c__Bacilli 4.50 (2.57–6.75) 2.89 (0.00 –5.84) 0.008 0.042\ng__Coprococcus; f__Lachnospiraceae; o__Clostridiales; c__Clostridia 10.31 (9.34–11.25) 10.81 (9.95 –11.76) 0.009 0.044\ng__f__o__YS2; c__4C0d2 0.00 (0.00–3.89) 0.00 (0.00 –1.11) 0.012 0.054\ng__f__o__RF32; c__Alphaproteobacteria 3.27 (0.00–6.72) 0.00 (0.00 –5.18) 0.016 0.066\ng__f__Peptostreptococcaceae; o__Clostridiales; c__Clostridia 6.90 (5.10–8.60) 6.04 (3.74 –8.15) 0.024 0.093\ng__f__Barnesiellaceae; o__Bacteroidales; c__Bacteroidia 11.53 (9.43–12.65) 10.43 (7.45 –12.55) 0.029 0.11\ng__f__Halanaerobiaceae; o__Halanaerobiales; c__Clostridia 8.53 (7.13–9.56) 7.85 (6.12 –9.08) 0.033 0.011\ng__f__o__RF39; c__Mollicutes 2.86 (0.19–7.70) 0.54 (0.00 –6.59) 0.040 0.13\ng__f__Lachnospiraceae; o__Clostridiales; c__Clostridia 3.83 (2.37–5.11) 12.72 (12.08 –13 .57) 0.040 0.12\nValues of operational taxonomic unit (OTU) are presented as median (interquartile range). Mann-Whitney U test. The q-value is the adjusted p-value\nwith a false discovery rate (FDR) of 5% and considered the main result\nTable 3 Bacteria with significant difference between patients with isolated ovarian and spread endometriosis\nBacteria Only ovarium N=27 Spread N=38 p-value Q-value\ng__f__Christensenellaceae; o__Clostridiales; c__Clostridia 5.76 (3.86–6.94) 3.77 (0.82 –6.10) 0.014 0.868\ng__Lachnobacterium; f__Lachnospiraceae; o__Clostridiales; c__Clostridia 6.68 (5.99 –7.80) 6.62 (5.21 –7.70) 0.036 1.000\ng__Adlercreutzia; f__Coriobacteriaceae; o__Coriobacteriales; c__Coriobacteriia 6.30 (4.61 –7.30) 4.64 (2.47 –6.76) 0.046 0.951\nValues of operational taxonomic unit (OTU) are presented as median (interquartile range). Mann-Whitney U test. The q-value is the adjusted p-value\nwith a false discovery rate (FDR) of 5% and our main results\n2373Reprod. Sci. (2021) 28:2367–2377\n\nstudied the gut microbiota, and only a single study had been\nperformed in human gut microbiota [28]. The main finding in\nAta et al. 2019 [ 29] was that two women with stage 3 –4\nendometriosis had an Escherichia/Shigella- dominant gut\nmicrobiome at genus level, whereas none of the\nendometriosis-free controls exhibited this dominance.\nEscherichia and Shigella belong to the family\nEnterobacteriaceae. We found a non-significant enrichment\nof Enterobacteriaceae in endometriosis patients, although we\ndid not identifyEscherichia/Shigellaat genus level using 16 S\nrRNA sequencing. Since the previous study only included as\nfew as 14 patients [29], and our findings lost significance due\nto correction for multiple testing, further studies are required\nto determine the true significance of Enterobacteriaceae be-\ntween subjects with and without endometriosis. No differ-\nences between endometriosis patients and controls in micro-\nbiota composition could be found in a recent study examining\nrectal swab samples [ 30]. A few studies have examined and\nfound altered microbiota composition in the human reproduc-\ntive tract in endometriosis, but the low number of studies\nmakes it impossible to estimate the associations between mi-\ncrobiota alterations in the reproductive tract and the gastroin-\ntestinal tract [28, 31].\nSeveral animal studies support the hypothesis that endome-\ntriosis has an impact on the gut microbiota. Rhesus monkeys\nwith endometriosis had a significantly altered gut microbiota\nprofile compared to healthy controls, and endometriosis was\nassociated with higher concentrations of Gram-negative bac-\nteria and lower concentrations of Lactobacilli [32]. When en-\ndometriosis was induced in mice, a higher beta diversity of gut\nmicrobiota was developed first after 42 days, with similar\nalpha diversity, among the endometriosis mice compared with\ncontrol mice [ 33]. In contrast, other mouse studies with\nendometriosis induction have described effects on the gut mi-\ncrobiota already after 21 days; one study found decreased\ndiversity, richness, and abundance of gut microbiota [ 34]\nand one found increased alpha and beta diversity [35], where-\nas a third study could not identify any effect at all [ 36].\nInterestingly, antibiotic treatment to mice reduced the endo-\nmetriosis lesions and inflammatory responses, changes which\nwere restored after oral feces gavage [35]. Although different\nmodels were used to induce and evaluate endometriosis in\nmice, the results point to an association between endometri-\nosis and gut microbiota and suggest that gut bacteria promote\nendometriotic lesion progression [ 31, 34–36]. However, the\nmicrobiota alterations may also depend on other functions,\ne.g., subclinical infections [37]. Our results indicated that the\nbeta diversity was slightly higher in the general population,\nwhilst the alpha diversity was significantly higher, compared\nto the endometriosis patients. Even though present and previ-\nous results are contradicting, which might be explained by\ndifferent species, our results indicate that long-term exposure\nto endometriotic tissue may affect the gut microbiota in\nhuman.\nEndometriosis is an estrogen-dependent disease and high\nlevels of estrogen have been linked to the pathogenesis of\nendometriosis [13]. The symptoms related to the disease are\ncommonly treated with estrogen, combined oral contracep-\ntives, progestin, or GnRH analogs, which abolish ovulation\nby lowering of the systemic levels of estrogen [ 38]. Previous\nstudies have shown that the gut microbiota has an impact on\nestrogen levels and a bi-directional relationship in estrogen-\ndependent diseases [ 11, 12]. Thus, gut microbiota could be\nin\nvolved in the development and symptomology of endome-\ntriosis, but endometriosis and its treatment may also affect the\ncomposition of gut microbiota. In the present study, we could\nTable 4 Bacteria with significant differences between endometriosis patients with and without involvement of the gastrointestinal (GI) tract\nBacteria GI tract not involved N=47 GI tract involved N=18 p-value Q-value\ng__Lactococcus; f__Streptococcaceae; o__Lactobacillales; c__Bacilli 2.30 (1.05 –4.16) 3.90 (1.81–6.11) 0.034 1.000\nValues of operational taxonomic unit (OTU) are presented as median (interquartile range). Mann-Whitney U test. The q-value is the adjusted p-value\nwith a false discovery rate (FDR) of 5% and our main results\nTable 5 Bacteria with significant difference between patients with and without gastrointestinal (GI) symptoms\nBacteria No GI symptoms N=8 GI symptoms N=57 p-value Q-value\ng__SMB53; f__Clostridiaceae; o__Clostridiales; c__Clostridia 6.96 (5.81–8.52) 4.72 (2.54 –6.92) 0.011 0.682\ng__Odoribacter; f__Odoribacteraceae; o__Bacteroidales; c__Bacteroidia 3.06 (0.54 –6.08) 0.00 (0.00 –0.82) 0.028 0.868\ng__Prevotella; f__Prevotellaceae; o__Bacteroidales; c__Bacteroidia 0.00 (0.00 –1.63) 4.96 (2.90 –10.44) 0.030 0.620\nValues of operational taxonomic unit (OTU) are presented as median (interquartile range). Mann-Whitney U test. The q-value is the adjusted p-value\nwith a false discovery rate (FDR) of 5% and our main results\n2374 Reprod. Sci. (2021) 28:2367–2377\n\nnot prove that changes in gut microbiota would cause or con-\ntribute to gastrointestinal symptoms.\nA previous study has described an association between the\nprotein AXIN1 and endometriosis in humans [17]. The study\nshowed that plasma levels of AXIN1 were higher in patients\nwith current hormone treatment, positively correlated with\nboth duration and degree of gastrointestinal symptoms, and\nnegatively correlated with levels of fecal calprotectin.\nHowever, we could not find any correlation between AXIN1\nlevels and abundance of bacteria, which differed between\nthose with or without gastrointestinal symptoms or hormonal\ntreatment. The correlation be tween fecal calprotectin and\nRuminococcus could possibly reflect more inflammation in\nthe group treated with hormones [ 17].\nThe results of the current study did not suggest that the\nlocalization of the endometriosis lesions is related to an altered\nprofile of gut microbiota, in accordance with a recent study\nconcerning potential plasma biomarkers for endometriosis,\nsuch as AXIN1, ST1A1, CXCL9, and OSM [16]. On the other\nhand, patients with pelvic endometriosis, with or without\novarian involvement, may have a higher prevalence of\ntenascin-C autoantibodies than patients with isolated ovarian\nendometriosis [39].\nAntibiotic treatment has beenshown to affect the gut micro-\nbiota for up to 6 months [40]. In the FDR-controlled sensitivity\nanalysis, only three bacteria at genus level differed significantly\nbetween patients and controls. It is unknown whether the reduced\nnumber of significant differences in bacteria abundance was due\nto the lower number of participants in the sensitivity analysis, or\nif antibiotic treatment had an impact on the results.\nThe strength of the present study is the examination of a\nhuman cohort and matched controls. However, some of the\ncontrols could theoretically also suffer from endometriosis.\nSince all participants from the MOS who suffered from gas-\ntrointestinal symptoms were excluded, this risk was mini-\nmized. Another limitation is that only 16S rRNA has been\nexamined, and not the whole microbiota genome.\nFurthermore, adjustments for food and several other con-\nfounders were not possible to perform. Examination of biopsy\nsamples instead of feces may be more representative. Also,\nlarge numbers of statistical calculations of several bacteria\nwere performed. By using FDR, we have tried to reduce the\neffect of multiple testing. On the other hand, by using FDR,\nthere is a risk to reduce the significance of a true association,\nwhich had been found if a smaller number of bacteria had\nbeen measured. Therefore, we have chosen to show also the\nresults before FDR-adjustment.\nThe large number of different bacteria in the gut in combi-\nnation with varying lifestyle habits and several confounders\nfor the microbiota composition may lead to different results in\ndifferent studies, maybe only by chance. One of the great\nchallenges for the future is to standardize sample collection\nand analysis, with appropriate adjustments for confounders, to\nbe able to compare different studies. Sample collection prior\nto any endometriosis treatment is important. In addition, asso-\nciations between the microbiota composition of the gastroin-\ntestinal tract and the reproductive tract and estrogen levels\nwould be of interest to evaluate. Furthermore, the microbiota\ncomposition should be related to fecal metabolites, to better\nunderstand the functional role of the composition [ 34].\nOur results indicate that the overall gut microbial diversity\nis significantly higher in controls compared to patients with\nendometriosis. Although the analyses showed significant re-\nsults for a number of bacteria at the genus level, the differ-\nences could be a coincidence depending on multiple compar-\nisons. Based on the cross-sectional study design, it is not pos-\nsible to decide whether the gut microbiota has any major im-\npact on endometriosis development and the related symptoms\nor whether endometriosis affects the gut microbiota.\nHowever, our study suggests that the gut microbiota may be\naltered to some extent in patients with endometriosis. The\nfindings put a perspective on microbiota profiling in endome-\ntriosis and provides a basis for further research on the patho-\nphysiology, diagnosis, and treatment of endometriosis.\nSupplementary Information The online version contains supplementary\nmaterial available at https://doi.org/10.1007/s43032-021-00506-5.\nAcknowledgements We would like to thank Malin Ek and the staff at the\nClinical Research Unit at the Department of Internal Medicine, Skåne\nUniversity Hospital, Malmö, for collecting all data, and Johan Hultman\nfor setting up the microbiota pipeline.\nTable 6 Bacteria with significant difference between patients with and without current hormonal treatment\nBacteria No treatment N=24 Treatment N=41 p-value Q-value\ng__Blautia; f__Lachnospiraceae; o__Clostridiales; c__Clostridia 10.80 (9.85–12.15) 12.12 (11.05 –13.30) 0.009 0.558\ng__Ruminococcus; f__Lachnospiraceae; o__Clostridiales; c__Clostridia 8.54 (8.12 –10.33) 9.75 (6.82 –10.74) 0.019 0.589\ng__Butyricimonas; f__Odoribacteraceae; o__Bacteroidales; c__Bacteroidia 8.91 (6.04 –10.62) 9.31 (4.27 –12.47) 0.034 0.703\nValues of operational taxonomic unit (OTU) are presented as median (interquartile range). Mann-Whitney U test. The q-value is the adjusted p-value\nwith a false discovery rate (FDR) of 5% and our main results\n2375Reprod. Sci. (2021) 28:2367–2377\n\nCode Availability Code is not available due to European laws.\nAuthor’s Contribution Conceptualization, B.R. and B.O.; methodology,\nA.S., B.R., and B.O.; software, L.B., A.S., and B.O.; validation, A.S.,\nB.R., L.B., and B.O.; formal analysis, A.S., L.B., and B.O.; investigation,\nB.R.; resources, B.O.; data curation, A.S. and B.O.; writing —original\ndraft preparation, A.S.; writing —review and editi ng, B.R., L.B.,\nM.O.M., and B.O.; visualization, A.S. and L.B.; supervision, M.O.M\nand B.O.; project administration, B.O.; funding acquisition, B.R. and\nB.O. All authors have read and agreed to the published version of the\nmanuscript.\nFunding Open access funding provided by Lund University. This re-\nsearch was funded by grants from Bengt Ihre Foundation, Dir Albert\nPåhlsson’s Foundation, and Development Foundation of Region Skåne.\nData Availability Data can be provided from the authors upon request.\nDeclarations\nEthics Approval and Consent to Participate This study was ap-\nproved by the Ethics Review Board of Lund University, No 2012/594,\n2012/564, and 2016/56.\nAll subjects gave written, informed consent before inclusion in the\nstudy.\nConsent for Publication All subjects gave written, informed con-\nsent before inclusion in the study.\nConflict of Interest The authors declare no competing interests.\nOpen Access This article is licensed under a Creative Commons\nAttribution 4.0 International License, which permits use, sharing, adap-\ntation, distribution and reproduction in any medium or format, as long as\nyou give appropriate credit to the original author(s) and the source, pro-\nvide a link to the Creative Commons licence, and indicate if changes were\nmade. The images or other third party material in this article are included\nin the article's Creative Commons licence, unless indicated otherwise in a\ncredit line to the material. If material is not included in the article's\nCreative Commons licence and your intended use is not permitted by\nstatutory regulation or exceeds the permitted use, you will need to obtain\npermission directly from the copyright holder. To view a copy of this\nlicence, visit http://creativecommons.org/licenses/by/4.0/.\nReferences\n1. Giudice LC, Kao LC. Endometriosis. Lancet. 2004;364:1789 –99.\n2. Ahn SH, Monsanto SP, Miller C, et al. Pathophysiology and\nImmune dysfunction in endometriosis. Biomed Res Int.\n2015;2015:795976.\n3. DiVasta AD, Vitonis AF, Laufer MR, Missmer SA. Spectrum of\nsymptoms in women diagnosed with endometriosis during adoles-\ncence vs adulthood. Am J Obstet Gynecol. 2018;218:324e1 –\n324e11.\n4. Bloski T, Pierson R. Endometri osis and chronic pelvic pain:\nunraveling the mystery behind this complex condition. Nurs\nWomen Health. 2008;12:382–95.\n5. Maroun P, Cooper MJW, Reid GD, Keirse MJNC. Relevance of\ngastrointestinal symptoms in endometriosis. Aust N Z J Obstet\nGynaecol. 2009;49:411–4.\n6. Luscombe GM, Markham R, Judio M, Grigoriu A, Fraser IS.\nAbdominal bloating: an under-recognized endometriosis symptom.\nJ Obstet Gynaecol Can. 2009;31:1159–71.\n7. Neish AS. Microbes in gastrointestinal health and disease.\nGastroenterology. 2009;136:65–80.\n8. Dieterich W, Schink M, and Zopf Y. Microbiota in the gastrointes-\ntinal tract. Med Sci. 2018;6(4).\n9. Arvonen M, Berntson L, Pokka T, Karttunen TJ, Vähäsalo P, Stoll\nML. Gut microbiota-host interactions and juvenile idiopathic arthri-\ntis. Pediatr Rheumatol Online J. 2016;14:44.\n10. Huipeng W, Lifeng G, Chuang G, Jiaying Z, Yuankun C. The\ndifferences in colonic mucosal microbiota between normal individ-\nual and colon cancer patients by polymerase chain reaction-\ndenaturing gradient gel electr ophoresis. J Clin Gastroenterol.\n2014;48:138–4.\n11. Flores R, Shi J, Fuhrman B, Xu X, Veenstra TD, Gail MH, et al.\nFecal microbial determinants of fecal and systemic estrogens and\nestrogen metabolites: a cross-sectional study. J Transl Med.\n2012;10:253.\n12. Baker JM, Al-Nakkash L, Herbst-Kralovetz MM. Estrogen-gut\nmicrobiome axis: physiological and clinical implications.\nMaturitas. 2017;103:45–53.\n13. Zhang Q, Shen Q, Celestino J, et al. Enhanced estrogen-induced\nproliferation in obese rat endometrium. Am J Obstet Gynecol.\n2009;200:186–e1-8.\n14. Karmarkar D, Rock KL. Microbiota signalling through MyD88 is\nnecessary for a systemic neutro philic inflammatory response.\nImmunology. 2013;140:483–92.\n15. Zhang X, Xu H, Lin J, Qian Y, Deng L. Peritoneal fluid concen-\ntrations of interleukin-17 correlate with the severity of endometri-\nosis and infertility of this disorder. Bjog. 2005;112:1153–5.\n16. Ek M, Roth B, Engström G, Ohlsson B. AXIN1 in plasma or serum\nis a potential new biomarker for endometriosis. Int J Mol Sci.\n2019;20.\n17. Dihm K, Ek M, Roth B, Ohlsson B. Plasma AXIN1 expression\nexhibit negative correlations with inflammatory biomarkers and is\nassociated with gastrointestinal symptoms in endometriosis.\nBiomed Rep. 2020;12:211–21.\n18. Laschke MW, Menger MD. The gut microbiota: a puppet master in\nthe pathogenesis of endometriosis? Am J Obstet Gynecol.\n2016;215(1):68–e1-4.\n19. Hamrefors V, Fedorowski A, Ohlsson B. Susceptibility to diarrhea\nis related to hemodynamic markers of sympathetic activation in the\ngeneral population. Scand J Gastroenterol. 2019;54:1426–32.\n20. Manjer J, Carlsson S, Elmstahl S, et al. The Malmo Diet and Cancer\nStudy: representativity, cancer incidence and mortality in partici-\npants and non-participants. Eur J Cancer Prev. 2001;10:489–99.\n21. Brunkwall L, Jönsson D, Ericson U, Hellstrand S, Kennbäck C,\nÖstling G, et al. The Malmö Offspring Study (MOS): design,\nmethods and first results. Eur J Epidemiol. 2020;36:103 –16.\nhttps://doi.org/10.1007/s10654-020-00695-4.\n22. Bengtsson M, Ohlsson B, Ulander K. Development and psycho-\nmetric testing of the Visual Analogue Scale for Irritable Bowel\nSyndrome (VAS-IBS). BMC Gastroenterol. 2007;7:16.\n23. Bengtsson M, Hammar O, Mandl T, Ohlsson B. Evaluation of\ngastrointestinal symptoms in different patient groups using the vi-\nsual analogue scale for irritable bowel syndrome (VAS-IBS). BMC\nGastroenterol. 2011;11:122.\n24. Magoc T, Salzberg SL. FLASH: fast length adjustment of short\nreads to improve genome assemblies. Bioinformatics. 2011;27:\n2957–63.\n25. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman\nFD, Costello EK, et al. QIIME allows analysis of high-throughput\ncommunity sequencing data. Nat Methods. 2010;7:335–6.\n26\n. Brunkwall L, Ericson U, Nilsson PM, et al. Self-reported bowel\nsymptoms are associated with differences in overall gut microbiota\n2376 Reprod. Sci. (2021) 28:2367–2377\n\ncomposition and enrichment of Blautia in a population-based co-\nhort. J Gastroenterol Hepatol. 2020; https://doi.org/10.1111/jgh.\n15104.\n27. McDonald JH. Handbook of Biological Statistics 3rd ed., editor.\nSparky House Publishing. Maryland: Baltimore; 2014.\n28. Leonardi M, Hicks C, El-Asseed F, et al. Endometriosis and the\nmicrobiome: a systematic review. Bjog. 2020;127:239–49.\n29. Ata B, Yildiz S, Turkgeldi E, Brocal VP, Dinleyici EC, Moya A,\net al. The endobiota study: comparison of vaginal, cervical and gut\nmicrobiota between women with stage 3/4 endometriosis and\nhealthy controls. Sci Rep. 2019;9:2204.\n30. Perrotta A, Borrelli GM, Martins CO, et al. The vaginal microbiome\nas a tool to predict rASRM stage of disease in endometriosis: a pilot\nstudy. Reprod Sci. 2020;27:1064–73.\n31. Wei W, Zhang X, Tang H, Zeng L, Wu R. Microbiota composition\nand distribution along the female reproductive tract of women with\nendometriosis. Ann Clin Microbiol Antimicrob. 2020;19:15.\n32. Bailey M, Coe C. Endometriosis is associated with an altered pro-\nfile of intestinal cimroflora in female Rhesus monkeys. Hum\nReprod. 2002;17:1704–8.\n33. Yuan M, Dong L, Zhang Z, et al. Endometriosis induces gut mi-\ncrobiota alterations in mice. Hum Reprod. 2018;33:607–16.\n34. Ni Z, Sun S, Bi Y, et al. Correlation of fecal metabolomics and gut\nmicrobiota in mice with endometriosis. Am J Reprod Immunol.\n2020;84:e13307.\n35. Chadchan SB, Cheng M, Parnell LA, Yin Y, Schriefer A,\nMysorekar IU, et al. Antibiotic therapy with metronidazole reduces\nendometriosis disease progression in mice: a potential role for gut\nmicrobiota. Hum Reprod. 2019;34:1106–16.\n36. Hantschel J, Weis S, Schäfer KH, Menger MD, Kohl M, Egert M,\net al. Effect of endometriosis on the fecal bacteriota composition of\nmice during the acute phase of lesion formation. PLoS One.\n2019;14:e0226835.\n37. Chen S, Gu Z, Zhang W, Jia S, Wu Y, Zheng P, et al. Microbiome\nof the lower genital tract in Chinese women with endometriosis by\n16s-rRNA sequencing technique: a pilot study. Ann Transl Med.\n2020;8:1440.\n38. Vercellini P, Vigano P, Somigliana E, et al. Endometriosis: patho-\ngenesis and treatment. Nat Rev Endocrinol. 2014;10:261–75.\n39. Ek M, Roth B, Valentin L, et al. Autoantibodies common in patients\nwith gastrointestinal diseases are not found in patients with endo-\nmetriosis: a cross-sectional study. Eur J Obstet Gynecol Reprod\nBiol. 2019;240.\n40. Albert Palleja KH. Recovery of gut microbiota of healthy adults\nfollowing antibiotic exposure. Nat Microbiol. 2018;3:1255–65.\nPublisher’sN o t eSpringer Nature remains neutral with regard to jurisdic-\ntional claims in published maps and institutional affiliations.\n2377Reprod. Sci. (2021) 28:2367–2377","source_license":"CC0","license_restricted":false}