Oral, Vaginal, and Stool Microbial Signatures in Patients With Endometriosis as Potential Diagnostic Non-Invasive Biomarkers: A Prospective Cohort Study

Abstract

Objective: To identify a microbial signature for endometriosis for use as a diagnostic non-invasive biomarker.

Design: Prospective cohort pilot study.

Setting: Nepean Hospital and UNSW Microbiome Research Centre, Australia.

Population: Sixty-four age- and sex-matched subjects (n = 19 healthy control (HC); n = 24 non-endometriosis (N-ENDO) and n = 21 confirmed endometriosis (ENDO)). All study participants, besides healthy controls, underwent laparoscopic surgical assessment for endometriosis, and histology was performed on excised lesions.

Methods: Oral, stool and, vaginal samples were self-collected at a single time point for healthy controls, and preoperatively for patients undergoing laparoscopy. Samples underwent 16S rRNA amplicon sequencing, followed by bioinformatics analysis.

Main outcome measures: Compositional differences between cohorts as identified by diversity analyses, and differentially abundant microbial taxa, as identified by LEfSE analysis.

Results: The composition of the oral (adjusted p = 0.003), and stool (adjusted p = 0.042) microbiota is different between the three cohorts. Differentially abundant taxa are present within each cohort as identified by LEfSE analysis. Particularly, Fusobacterium was enriched in the oral samples of patients with moderate/severe endometriosis.

Conclusions: Taxonomic and compositional differences were found between the microbiota in the mouth, gut and, vagina of patients with and without endometriosis and healthy controls. Fusobacterium was enriched in patients with moderate/severe endometriosis. Fusobacterium is noted as a key pathogen in periodontal disease, a common comorbidity in endometriosis. These findings suggest a role for the oral, stool and, vaginal microbiome in endometriosis, and present potential for microbial-based treatments and the design of a diagnostic swab.

Keywords: bacteria; biomarker; diagnostic; dysbiosis; endometriosis; inflammation; microbiome; microbiota; oral.

FIGURE 1

Overview of sample collection and study cohorts. (A) All participants were provided with oral, stool and vaginal sample collection kits. (B) Healthy controls (HC) collected samples at a single time point. Endometriosis (ENDO) and non‐endometriosis (N‐ENDO) cohorts collected samples preoperatively, at a single time point. N‐ENDO and ENDO participants underwent laparoscopy and were surgically investigated for the presence of endometriosis, and histological diagnosis was performed on any lesion findings at the hospital pathology laboratory and staged according to the revised American Society of Reproductive Medicine (rASRM) classification of endometriosis. n = 10 rASRM, stage 1; n = 1, stage 2; n = 2, stage 3; and n = 7, stage 4. Stage 1 and 2 = minimal/mild, stage 3 and 4 = moderate/severe.

FIGURE 2

Relative abundances of the top seven most abundance phyla sequenced using 16S rRNA. Samples are charted horizontally by body site: (A) oral, (B) stool and (C) vaginal samples and separated in vertical panels according to study cohorts: Healthy controls (HC), patients with confirmed no‐endometriosis (N‐ENDO) and patients with confirmed endometriosis (ENDO). All other detected phyla not in the top seven are cohorted together and labelled as ‘Remaining’.

FIGURE 3

Comparison of alpha diversity, measured by Shannon diversity index (based on features at genus rank), between study cohorts of healthy controls (HC), patients with confirmed no‐endometriosis (N‐ENDO) and confirmed endometriosis (ENDO). Comparisons were assessed using Kruskal–Wallis test and then followed by Wilcoxon rank‐sum tests for pairwise comparisons adjusted using the FDR approach (text annotation). There were borderline differences (adjusted p < 0.1, estimate = 0.11, 95% CI 7.4e−5–0.22) between HC and ENDO group observed in oral samples (A) and significant differences in stool (adjusted p < 0.05, estimate = 0.27 and 95% CI 0.09–0.44) (B) samples. There was no difference in alpha diversity based on the vaginal samples (C).

FIGURE 4

Principal coordinate analysis (PCoA) plots of beta diversity measured using Bray–Curtis dissimilarity between study cohorts of oral samples (A), stool samples (B) and vaginal samples (C). Results from PERMANOVA tests are annotated in the top right corner where significant differences between cohorts were observed in stool and oral samples, and there were no significant differences observed in the vaginal samples. HC, Healthy controls; N‐ENDO, Patients with confirmed no endometriosis; ENDO, Patients with confirmed endometriosis.

FIGURE 5

Overview of taxonomy features detected as differentially abundant using LEfSe analysis. Colours denote the cohorts the corresponding taxon is higher in relative abundance. Samples from the three body sites are separated into vertical panels. HC, Healthy controls; N‐ENDO, Patients with confirmed no endometriosis; ENDO, Patients with confirmed endometriosis.

FIGURE 6

Overview of taxonomy features detected as differentially abundant using LEfSe analysis between patients (n = 12) with minimal/mild stage endometriosis (pink) and patients (n = 9) with moderate/severe endometriosis (purple). Samples from the three body sites are separated into vertical panels.