Meta-Analysis of 16S rRNA Sequencing Reveals Altered Fecal but Not Vaginal Microbial Composition and Function in Women with Endometriosis

Abstract

Background and Objectives: Dysbiosis of the oral-gut axis is related to several extraintestinal inflammatory diseases, including endometriosis. This study aims to assess the microbial landscape and pathogenic potential of distinct biological niches during endometriosis. Materials and Methods: A microbiome meta-analysis was conducted on 182 metagenomic sequences (79 of fecal and 103 of vaginal origin) from women with and without endometriosis. Fecal and vaginal microbial diversity, differential abundance, and functional analysis based on disease status were assessed. Random forest, gradient boosting, and generalized linear modeling were used to predict endometriosis based on differentially enriched bacteria. Results: Only intestinal microbes displayed distinctive taxonomic and functional characteristics in women with endometriosis compared to control women. Taxonomic differences were quantified using the microbial endometriosis index (MEI), which effectively distinguished between individuals with and without the disease. The observed functional enrichment pointed to proinflammatory pathways previously related to endometriosis development. Conclusions: Dysbiosis in the oral-gut microbial community appears to play a prevalent role in endometriosis. Our findings pave the ground for future studies exploring the potential mechanistic involvement of the oral-gut axis in disease pathogenesis.

Keywords: dysbiosis; endometriosis; meta-analysis; oral–gut microbiota; vaginal microbiota.

Figure 1

Principal component analysis (PCA) identifies major sources of variation across all samples. (A) Biplot displaying variable contributions to sample variance. Dim1 (33%) is strongly influenced by sampling site and BioProject (red and orange arrows), reflecting technical and cohort-related variability. Dim2 (25.3%) is more influenced by group (endometriosis status) and microbial abundance, suggesting a biological signal partially independent of technical confounding. (B) Bar plot showing variance explained by the first four principal components. (C,D) Contribution of individual variables to Dim1 and Dim2. The red dashed line represents the expected average contribution assuming equal influence across variables. These results highlight the need to account for technical confounders in meta-analyses; however, the orthogonal contribution of Group to Dim2 supports the presence of an underlying disease signal beyond cohort effects.

Figure 2

Microbiome composition and diversity analysis. (A) Violin plots comparing the observed alpha diversity (ASV richness) of fecal and vaginal microbiomes across different groups (B) Scatter plots illustrating the relationship between observed ASV diversity and host age for fecal and vaginal microbiomes. (C) Correlation analysis between observed ASV diversity and host age in fecal and vaginal microbiomes, including Pearson correlation coefficient (R) and p-value. (D) Bar plots showing the number of DA methods (DESeq2, LEFSe, Welch’s test) that agree on the enrichment of specific microbial taxa in fecal and vaginal microbiomes. (E) Dot plots depicting the relative abundance of selected microbial taxa in fecal and vaginal microbiomes, with taxa enriched in group endometriosis and group controls highlighted. (F) Receiver operating characteristic (ROC) curves evaluating the performance of machine learning models (GBM, RF, GLM) in differentiating fecal and vaginal microbiomes based on their microbial composition. (G) Bar plot comparing the mean importance of microbial taxa as predictors in the machine learning models. Red dots indicate woman with endometriosis, while grey dots indicate controls. Significance level: * p < 0.05, ns: Not Significant.

Figure 3

Functional microbial pathways differentially enriched in the intestinal environment of women with and without endometriosis. FDR: false discovery rate. See methods for details.