Vaginal and rectal microbiome contribute to genital inflammation in chronic pelvic pain

Abstract

Background: Chronic pelvic pain (CPP) is a multifactorial syndrome that can substantially affect a patient's quality of life. Endometriosis is one cause of CPP, and alterations of the immune and microbiome profiles have been observed in patients with endometriosis. The objective of this pilot study was to investigate differences in the vaginal and gastrointestinal microbiomes and cervicovaginal immune microenvironment in patients with CPP and endometriosis diagnosis compared to those with CPP without endometriosis and no CPP.

Methods: Vaginal swabs, rectal swabs, and cervicovaginal lavages (CVL) were collected among individuals undergoing gynecologic laparoscopy. Participants were grouped based on patients seeking care for chronic pain and/or pathology results: CPP and endometriosis (CPP-Endo) (n = 35), CPP without endometriosis (n = 23), or patients without CPP or endometriosis (controls) (n = 15). Sensitivity analyses were performed on CPP with endometriosis location, stage, and co-occurring gynecologic conditions (abnormal uterine bleeding, fibroids). 16S rRNA sequencing was performed to profile the microbiome, and a panel of soluble immune mediators was quantified using a multiplex assay. Statistical analysis was conducted with SAS, R, MicrobiomeAnalyst, MetaboAnalyst, and QIIME 2.

Results: Significant differences were observed between participants with CPP alone, CPP-Endo, and surgical controls for body mass index, ethnicity, diagnosis of ovarian cysts, and diagnosis of fibroids. In rectal microbiome analysis, both CPP alone and CPP-Endo exhibited lower alpha diversity than controls, and both CPP groups revealed enrichment of irritable bowel syndrome-associated bacteria. CPP-Endo exhibited an increased abundance of vaginal Streptococcus anginosus and rectal Ruminococcus. Patients with CPP and endometrioma (s) demonstrated increased vaginal Streptococcus, Lactobacillus, and Prevotella compared to other endometriosis sites. Further, abnormal uterine bleeding was associated with an increased abundance of bacterial vaginosis-associated bacteria. Immunoproteomic profiles were distinctly clustered by CPP alone and CPP-Endo compared to controls. CPP-Endo was enriched in TNF⍺, MDC, and IL-1⍺.

Conclusions: Vaginal and rectal microbiomes were observed to differ between patients with CPP alone and CPP with endometriosis, which may be useful in personalized treatment for individuals with CPP and endometriosis from those with other causes of CPP. Further investigation is warranted in patients with additional co-occurring conditions, such as AUB/fibroids, which add additional complexity to these conditions and reveal the enrichment of distinct pathogenic bacteria in both mucosal sites. This study provides foundational microbiome-immunoproteomic knowledge related to chronic pelvic pain, endometriosis, and co-occurring gynecologic conditions that can help improve the treatment of patients seeking care for pain.

Keywords: Abnormal uterine bleeding; Chronic pelvic pain; Endometriosis; Fibroids; Gut microbiome; Vaginal microbiome; Women’s health.

Fig. 1

Diagram of cohort groupings and inclusions and exclusion. N = 81 women enrolled in the chronic pelvic pain study. Three women were excluded due to exclusion criteria of postmenopausal status or current sexually transmitted infection. Pathology evaluation confirmed diagnosis of endometriosis (n = 40) or no endometriosis (n = 38). With histological confirmation and symptom diagnosis, we had the following groups: chronic pelvic pain with endometriosis (n = 35), chronic pelvic pain without endometriosis (n = 23), and patients with no chronic pelvic pain or endometriosis (n = 15). Four women were excluded from the study due to incidental endometriosis with no chronic pelvic pain and a small sample size. Of the women diagnosed with CPP without endometriosis, ten women noted a self-report history of endometriosis. Women who were diagnosed with CPP with endometriosis were then given endometriosis staging based on the American Society for Reproductive Medicine revised classification: stage 1 (n = 15), stage II (n = 7), stage III (n = 2), and stage IV (n = 12)

Fig. 2

The vaginal and rectal microbiome in chronic pelvic pain, endometriosis, and controls. The rectal microbiome is significantly different between patients diagnosed with chronic pelvic pain and no chronic pelvic pain. A Grouped taxa bar plot of genera in the vaginal and rectal microbiomes across diagnosis groups: controls with no chronic pelvic pain (Controls), chronic pelvic pain (CPP), and chronic pelvic pain with endometriosis (CPP-Endo). B Chao1 species richness is not significantly different between Controls, CPP, and CPP-Endo in the vaginal microbiome (p-value: 0.89226; (ANOVA) F-value: 0.11418) or rectal microbiome (p-value: 0.10783; (ANOVA) F-value: 2.2996). C Differentially abundant vaginal and rectal bacterial taxa among CPP and CPP-Endo compared to controls. The differential abundance analysis was performed utilizing ANCOM-BC, visualized taxa with at least 1 log fold change. Other significant taxa are in Supplement Fig. 4, and all p-values were Bonferroni false discovery adjusted where q-value < 0.05 was significant. D Boxplots of vaginal bacteria indicated as significant differentially abundant log10 transformed, where abundance is observed by the diagnosis group. Gray indicates Controls, light pink indicates CPP, and light green indicates CPP-Endo. Additional abundance differences were performed by the Kruskal–Wallis test where * is < 0.05, ** is < 0.01, *** < 0.001, and **** < 0.0001 p-value

Fig. 3

Cervicovaginal lavage immune markers for chronic pelvic pain and chronic pelvic pain with endometriosis. Cervicovaginal levels of protein biomarkers in patients vary based on disease group. A volcano plot analysis was used to assess differences in the immune protein levels among patients based on the diagnosis group. Statistical significance was determined using a two-sample t-test with the false discovery rate (FDR) correction. Proteins with q < 0.05 were considered significant. A Proteins G-CSF, RANTES, and VEGF were significantly downregulated, and IL-1Ra was upregulated across diagnosis groups: controls with no chronic pelvic pain (Ctrl) in gray and chronic pelvic pain (CPP) in pink, B proteins MDC, IL-8, GRO, and VEGF were significantly downregulated across diagnosis groups: controls with no chronic pelvic pain (Ctrl) in gray and chronic pelvic pain with endometriosis (EC) in green. C Protein IL-1alpha was significantly downregulated, and RANTES was upregulated across diagnosis groups when p-value analysis was performed. FDR correction showed no significant markers: chronic pelvic pain (CPP) in pink and chronic pelvic pain with endometriosis (EC) in green. D A principal component analysis (PCA) of the diagnosis groups shows distinct differences in diagnosis groups. PC 1 shows a significant difference between controls with no chronic pelvic pain in gray and chronic pelvic pain in pink. PC 2 shows a significant difference between controls with no chronic pelvic pain in gray and chronic pelvic pain with endometriosis in green. Gray indicates controls, pink indicates CPP, and green indicates CPP-Endo. Additional differences between disease groups were performed by ANOVA test where * is denoted as < 0.05, ** is denoted as < 0.01, *** < 0.001, and **** < 0.0001 p-value. E A Venn diagram showcasing shared and unique markers that were identified with the volcano plot analysis between the diagnosis groups

Fig. 4

Hierarchical clusters of immune and microbiome profiles indicate unique global profiles from patients with chronic pelvic pain with and without endometriosis. Cervicovaginal protein levels are associated with the disease groups. A heatmap reflects relative levels of proteins in cervicovaginal lavages (CVL) across all the samples (n = 72). Data were mean-centered and log-transformed. Hierarchical clustering was based on Euclidean distance and Ward linkage. Two immune markers were removed due to being constant, IL-3 and IL-5. The analysis revealed two distinct clusters based on the immune profiles of this cohort. Pie charts show significant differences observed in distribution between immune clusters 1 and 2 for B disease groups, C CPP status, D vaginal pH, and E blood in the sample were significantly different between the clusters. p-values were calculated using Fisher’s exact test or chi-square test. p-values < 0.05 were significant, where * is denoted as < 0.05, ** is denoted as < 0.01, *** < 0.001, and **** < 0.0001 p-value

Fig. 5

Endometriosis characteristics reveal distinct microbial and immune profiles. Endometriosis characteristics such as stage and location can be differentiated by vaginal and rectal microbiome. A Differentially abundant vaginal and rectal bacterial taxa diverging bar plot among ASRM revised classification stages 3/4 (blue) compared to 1/2. B Differentially abundant vaginal and rectal bacterial taxa diverging bar plot of the site of endometriosis detected during surgery of peritoneum (pink); other (purple) which included locations of the bladder, vagina, bowel, and lymph node; and multiple locations (green) which we defined as finding endometriosis in more than one determined location during surgery. These locations were compared to the most prevalent location site of the ovary, sometimes referred to as endometrioma. The differential abundance analysis was performed utilizing ANCOM-BC, visualized taxa with at least 1 log fold change; other significant taxa are in supplement Fig. 12, and all p-values were Bonferroni false discovery adjusted where q-value < 0.05 was significant. C Cervicovaginal levels of protein biomarkers in patients vary. A volcano plot analysis was used to assess differences in the immune protein levels among patients based on the diagnosis group. Statistical significance was determined using a two-sample t-test with the false discovery rate correction. Proteins with q < 0.05 were considered significant. Protein biomarker diverging bar plot among ASRM revised classification stages 3/4 (blue) compared to 1/2. D Protein biomarker diverging bar plot of the site of endometriosis detected during surgery of peritoneum (pink); other (purple); and multiple locations (green), which we defined as finding endometriosis in more than one determined location during surgery. These locations were compared to the site of the ovary. The protein biomarker analysis was performed utilizing a t-test, visualized markers where q-value < 0.05 was significant. E PCA plot of immune markers by endometriosis location sites. Utilizing the Kruskal–Wallis test, no significant difference between endometrial locations was found

Fig. 6

Co-occurring conditions of chronic pelvic pain reveal unique microbial and immune signatures. Abnormal uterine bleeding and fibroids had differences in bacterial vaginosis-associated bacteria and depletion of cytokines and growth factors. A Venn diagram of vaginal and rectal bacteria that were most differentially abundant of co-occurring conditions: AUB and no AUB (blue circle), fibroids and no fibroids (purple circle), ovarian cysts and no ovarian cysts (orange circle). Bacterial names in the co-occurring condition group were identified as enriched (red) or depleted (blue). Labeled were taxa with at least 1 log fold change, other significant taxa are in additional data file 1, and all p-values were FDR adjusted where q-value < 0.05 was significant. No bacteria that fit these criteria were identified for diagnosis of ovarian cysts compared to no ovarian cysts. B Grouped relative abundance stacked bar plots for vaginal and rectal differentially abundant bacteria in no AUB vs. AUB and no fibroids vs. fibroids. Rectal stacked bar plots were zoomed in to 92% and above relative abundance due to the large diversity of the rectal microbiome to visualize differences between co-occurring condition groups. C Venn diagram of immune proteins that were significantly upregulated or downregulated of histologically confirmed co-occurring conditions groups. Proteins in the co-occurring condition group were identified as enriched (red) or depleted (blue). The analysis was performed utilizing volcano plot analysis labeled proteins with at least 1 log fold change. p-values were Bonferroni adjusted where q-value < 0.05 was significant are indicated by an asterisk (*). D Graphs of the relative concentration levels of proteins between the groups of histologically confirmed co-occurring conditions groups of AUB (blue) and no AUB (yellow), fibroids (purple) and no fibroids (yellow), ovarian cysts (orange) and no ovarian cysts (yellow). p-values that were not Bonferroni adjusted are shown where p-value < 0.05 was significant

Fig. 7

Correlations of differentially abundant microbes and cervicovaginal immune markers. A correlation analysis between cervicovaginal lavage levels of 41 immune proteins with significant differentially abundant vaginal taxa of subjects from analysis groups: A chronic pelvic pain (CPP), B fibroids, C abnormal uterine bleeding (AUB). Correlation coefficients (r) were calculated using Spearman’s rank correlation, where positive (red) and negative (blue) correlations are depicted as a heatmap. p-values < 0.05 were significant, where * is denoted as < 0.05, ** is denoted as < 0.01, *** < 0.001, and **** < 0.0001 p-value. Taxa labels in red signify that a bacterial species was enriched in the samples, and taxa in blue signifies that a bacterial species was depleted in the samples from each analysis group

Fig. 8

Summary of findings of microbiome and cervicovaginal immune markers for CPP and CPP with endometriosis. Dysbiotic vaginal and rectal bacteria might affect the etiology of chronic pelvic pain with and without endometriosis. An increase in bacterial-vaginosis-associated bacteria such as Megasphaera, Lachnospiraceae, Fannyhessea, Mobiluncus, Bifidobacterium/Gardnerella were associated with peritoneal endometriosis, fibroids, and abnormal uterine bleeding. These microorganisms also revealed a positive correlation with proinflammatory immune mediators and growth factors. Our cervicovaginal lavage samples also revealed a decrease in IL-1a in CPP-Endo and RANTES and G-CSF in CPP, revealing a dysregulation in inflammatory pathways for chronic pelvic pain with or without endometriosis. Decreased MDC and MCP-1 were also observed in endometriosis stage 3/4, which may link to neurogenesis, which is observed more in endometriosis stage 1/2. This linkage may also interplay with the increase of Lactobacillus jensenii observed in endometriosis stage 3/4. Increased Prevotella was observed in both CPP and CPP-Endo, while Streptococcus anginosus was more abundant in CPP-Endo than CPP. Our rectal samples revealed an increase in bacteria previously associated with rectal dysbiosis in colitis and irritable bowel disorder, which has also been associated with chronic pelvic pain; some of these bacteria, including Stoquefichus and Ruminococcus that were increased in both CPP and CPP-Endo. A shared depletion of Desulfovibrio was also observed in CPP and CPP-Endo. Finally, a sharing of vaginal taxa such as Fannyhessea and Winkia in the rectal samples was observed, which has previously been linked to systemic dysbiosis and sharing of microbes between sites