Gut microbiome in patients with early-stage and late-stage endometriosis

The participant recruitment process was shown in Fig.  1 . From June 23, 2021, to July 5, 2022, all women admitted to the Obstetrics and Gynecology Medical Center of Zhujiang Hospital were assessed for eligibility for the current study. Fecal specimens were collected prior to surgery to avoid potential interference with gut microbiota caused by medications administered during surgery. Since the diagnosis of endometriosis could not be confirmed before surgery, a further evaluation was conducted post-surgery to exclude patients who did not meet the inclusion criteria. Together, 75 endometriosis patients were included in the current study, and informed consent forms were completed and signed by the participants. This study was conducted in Guangzhou, a southern city in China with a consistently warm and humid climate, and minimal seasonal temperature variation. All enrolled patients were long-term residents of Guangzhou, where dietary habits are relatively similar. Previous studies have indicated that seasonal and dietary factors may influence gut microbiota [ 29 , 30 ], but the participants of the current study are expected to have good internal consistency regarding these two factors. Fig. 1 Enrollment process of the included subjects in the study Enrollment process of the included subjects in the study Basic demographic and clinical characteristics of the participants are listed in Table  1 . According to the widely used, American Society of Reproductive Medicine (ASRM) staging system, 32 patients were diagnosed with stage I endometriosis, 7 patients were diagnosed with stage II endometriosis, 16 patients were diagnosed with stage III endometriosis, and 20 patients were diagnosed with stage IV endometriosis. Patients with stage I and II endometriosis were further classified into the Early-stage Group, while patients with stage III and IV endometriosis were classified into the Late-stage Group. There was no significant difference between the Early-stage and Late-stage Groups in age, BMI, marital status, gravidity history, or parity history. Among all participants, only 3 patients were non-first episode cases, and 4 had a family history of endometriosis. Additionally, 16 patients from the Early-stage Group and 17 patients from the Late-stage Group had preoperative dysmenorrhea. Table 1 Demographic and clinical characteristics of patients with endometriosis All EMs participants n  = 75 Early_Stage n  = 39 Late_Stage n  = 36 P value Stage:  I 32 (42.7%) 32 (82.1%) 0 (0.00%)  II 7 (9.33%) 7 (17.9%) 0 (0.00%)  III 16 (21.3%) 0 (0.00%) 16 (44.4%)  IV 20 (26.7%) 0 (0.00%) 20 (55.6%) r-AFS, Median (range) 6 (1–80) 1 (1–11) 40 (20–80) CA125 26.4 [16.3;45.0] 18.0 [12.9;25.3] 44.2 [32.3;70.5] < 0.001 Age, Median (range), y 32 (18–49) 35 (18–49) 32 (21–49) 0.165 Height (cm) 160 (5.19) 160 (5.24) 160 (5.21) 0.846 Weight (kg) 53.0 [48.0;58.5] 53.0 [48.8;57.0] 52.8 [48.0;59.2] 0.803 BMI (kg*m-2) 20.8 [19.3;22.9] 20.8 [19.3;22.6] 20.8 [19.3;22.9] 0.791 Marriage 0.22  FALSE 25 (33.3%) 10 (25.6%) 15 (41.7%)  TRUE 50 (66.7%) 29 (74.4%) 21 (58.3%) Gravidity 1 [0;2] 1 [0;2] 0 [0;2] 0.452 Parity 0 [0;1] 1 [0;1] 0 [0;1] 0.501 Frst episode: 0.605  FALSE 3 (4.00%) 1 (2.56%) 2 (5.56%)  TRUE 72 (96.0%) 38 (97.4%) 34 (94.4%) Menstrual cycle, Median (range), d 29 (20–45) 29 (20–45) 29 (25–35) 0.867 Family_History: 0.048  FALSE 71 (94.7%) 39 (100%) 32 (88.9%)  TRUE 4 (5.33%) 0 (0.00%) 4 (11.1%) Dysmenorrhea (Pre-operation): 0.759  FALSE 42 (56.0%) 23 (59.0%) 19 (52.8%)  TRUE 33 (44.0%) 16 (41.0%) 17 (47.2%) Demographic and clinical characteristics of patients with endometriosis As shown in the stack bar plot, the top 3 most abundant phyla of the endometriosis patients were Firmicutes , Bacteroidota , and Proteobacteria (Fig.  2 A). The average relative abundance of Firmicutes was similar between groups, while the Late-stage Group had a higher abundance of Bacteroidota . Shannon index was calculated to evaluate alpha diversity (from phylum to species level) between the Early-stage and Late-stage Groups, with no significant difference observed (Fig.  2 B). To examine community structure, PCoA analysis was applied and a structural difference was seen along PCoA2 axis (Fig.  2 C). We next determined differential genus and species abundance between groups (Fig.  2 D; Supplement Fig.  1 A). In total, genus Gilliamella , Bartonella , Snodgrassella and 10 other genera were more abundant in the Late-stage Group, while genus Saccharofermentans and 10 other genera were enriched in the Early-stage Group. To further elucidate functional capacity, we performed predictive analysis using PICRUSt2 to infer microbial community functions from 16S taxonomic profiles. Finally, we identified 50 differential pathway abundances, 319 differential Kyoto Encyclopedia of Genes and Genomes (KEGG) Enzyme Commission (EC) numbers, and 1036 differential KEGG orthologs (KOs) (Fig.  2 E; Supplement Fig.  1 B). The findings suggest that early-stage patients harbor distinct gut microbial pattern comparing to late-stage endometriosis group. Fig. 2 Comparison of gut microbial diversity and composition of early-stage endometriosis and late-stage endometriosis patients. ( A ) The bar plot shows the top 10 most relatively abundant phyla across all microbial samples. ( B ) The line chart displays alpha-diversity results within each group at the levels of phylum, class, order, family, genus, and species. Data points represent mean Shannon index values for samples within each group, and p -values show statistical significance of between-group comparisons by t-test. ( C ) The scatter plot shows principal coordinate analysis (PCoA) results along the first and second axis. The box plot presents coordinate values along the second axis. ( D ) The ratio matchstick plot illustrates differential microbes between the early- and late-stage endometriosis groups, with purple dots representing genera enriched in the Late-stage Group, and blue dots indicating genera abundant in the Early-stage Group gut microbiota. ( E ) The heat map shows differential metabolic pathway abundances Comparison of gut microbial diversity and composition of early-stage endometriosis and late-stage endometriosis patients. ( A ) The bar plot shows the top 10 most relatively abundant phyla across all microbial samples. ( B ) The line chart displays alpha-diversity results within each group at the levels of phylum, class, order, family, genus, and species. Data points represent mean Shannon index values for samples within each group, and p -values show statistical significance of between-group comparisons by t-test. ( C ) The scatter plot shows principal coordinate analysis (PCoA) results along the first and second axis. The box plot presents coordinate values along the second axis. ( D ) The ratio matchstick plot illustrates differential microbes between the early- and late-stage endometriosis groups, with purple dots representing genera enriched in the Late-stage Group, and blue dots indicating genera abundant in the Early-stage Group gut microbiota. ( E ) The heat map shows differential metabolic pathway abundances Considering previous studies linking pain perception to the gut microbiota, we next divided the patients into subgroups based on presence of preoperative dysmenorrhea (Fig.  3 A). The stack bar plot showed the top 10 most abundant taxa across all samples, with genus Faecalibacterium ranking 1st, followed by Escherichia − Shigella , Bacteroides_uniformis , Bifidobacterium , Agathobacter , Bacteroides fragilis , Erysipelotrichaceae_UCG − 003 , Bacteroides plebeius and Subdoligranulum (Fig.  3 B). Among early-stage endometriosis patients, subjects with dysmenorrhea (Early-stage T Group) showed no differences in alpha or beta diversity compared to those without dysmenorrhea (Early-stage F Group) (Fig.  3 C, D). However, late-stage groups differed in beta diversity - PCoA analysis revealed microbial structure separation along axis PCoA2 between those with dysmenorrhea (Late-stage T Group) and without dysmenorrhea (Late-stage F Group) (Fig.  3 E, F). We then explored potential taxa associated with dysmenorrhea. As shown in Figs.  4 A, 7 and 13 differential taxa were identified between early- and late-stage endometriosis subjects with versus without dysmenorrhea, respectively. Microbial functions were further predicted by PICRUSt2 to analyze differential metabolic pathways, enzymes and KOs among subgroups (Fig.  4 B). Differential taxa and functional elements were observed within each subgroup pair, with intercepts consistently equaling zero, indicating significant gut microbiota differences between early- and late-stage endometriosis patients. Taken together, we demonstrated late-stage patients with dysmenorrhea displayed distinctly different gut profiles from other endometriosis groups. Fig. 3 Comparison of gut microbial diversity and composition of participants with and without dysmenorrhea. ( A ) The pie chart displays the number and proportion of early- and late-stage endometriosis groups. Patients were further divided into dysmenorrhea positive and negative subgroups based on preoperative dysmenorrhea symptoms. In the group labels, “T” represents patients with dysmenorrhea, and “F” represents patients without dysmenorrhea. EMs represents endometriosis. ( B ) The bar plot shows top 10 most relatively abundant species across all microbial samples. ( C-D ) Figures C and D present microbial diversity analysis results between dysmenorrhea positive (Early-Stage-T) and negative (Early-Stage-F) subgroups within the early-stage endometriosis group. The line chart displays alpha diversity results within each subgroup at the levels of domain, phylum, class, order, family, genus, and species. Data points represent mean Shannon index values for samples within each subgroup, and p -values show statistical significance of between-group comparisons by t-test. The scatter plot shows principal coordinate analysis (PCoA) results along the first and second axis. The box plot presents coordinate values along the first axis. ( E-F ) Figures E and F show microbial diversity analysis between dysmenorrhea positive (Late-Stage-T) and negative (Late-Stage-F) subgroups within the late-stage endometriosis group. The line chart displays alpha diversity results within each subgroup across taxonomic levels. Data points represent mean Shannon values. The scatter plot displays PCoA axis 1 and 2 results. The box plot presents coordinate values along the second axis Comparison of gut microbial diversity and composition of participants with and without dysmenorrhea. ( A ) The pie chart displays the number and proportion of early- and late-stage endometriosis groups. Patients were further divided into dysmenorrhea positive and negative subgroups based on preoperative dysmenorrhea symptoms. In the group labels, “T” represents patients with dysmenorrhea, and “F” represents patients without dysmenorrhea. EMs represents endometriosis. ( B ) The bar plot shows top 10 most relatively abundant species across all microbial samples. ( C-D ) Figures C and D present microbial diversity analysis results between dysmenorrhea positive (Early-Stage-T) and negative (Early-Stage-F) subgroups within the early-stage endometriosis group. The line chart displays alpha diversity results within each subgroup at the levels of domain, phylum, class, order, family, genus, and species. Data points represent mean Shannon index values for samples within each subgroup, and p -values show statistical significance of between-group comparisons by t-test. The scatter plot shows principal coordinate analysis (PCoA) results along the first and second axis. The box plot presents coordinate values along the first axis. ( E-F ) Figures E and F show microbial diversity analysis between dysmenorrhea positive (Late-Stage-T) and negative (Late-Stage-F) subgroups within the late-stage endometriosis group. The line chart displays alpha diversity results within each subgroup across taxonomic levels. Data points represent mean Shannon values. The scatter plot displays PCoA axis 1 and 2 results. The box plot presents coordinate values along the second axis Fig. 4 Comparison of gut microbial abundance and function of participants with and without dysmenorrhea. ( A ) The left ratio matchstick plot shows 7 differential species between dysmenorrhea positive and negative subgroups within the early-stage endometriosis group. The right ratio matchstick plot displays 13 differential species between dysmenorrhea positive and negative subgroups in the late-stage endometriosis group. The Venn diagram presents the intersecting differential species between these two comparisons. In the group labels, “T” represents patients with dysmenorrhea, and “F” represents patients without dysmenorrhea. EMs represents endometriosis. ( B ) The Venn diagrams show the intersections of differential Metabolic Pathways, Enzymes, and KEGG Orthologs predicted by PICRUSt2 among the subgroups, respectively Comparison of gut microbial abundance and function of participants with and without dysmenorrhea. ( A ) The left ratio matchstick plot shows 7 differential species between dysmenorrhea positive and negative subgroups within the early-stage endometriosis group. The right ratio matchstick plot displays 13 differential species between dysmenorrhea positive and negative subgroups in the late-stage endometriosis group. The Venn diagram presents the intersecting differential species between these two comparisons. In the group labels, “T” represents patients with dysmenorrhea, and “F” represents patients without dysmenorrhea. EMs represents endometriosis. ( B ) The Venn diagrams show the intersections of differential Metabolic Pathways, Enzymes, and KEGG Orthologs predicted by PICRUSt2 among the subgroups, respectively To better understand the gut microbiome functionality in the endometriosis population, we further calculated the top 10 functional variables based on their relative abundance and presented the results in the form of stacked plots (Fig.  5 ). In the PICRUSt2 functional prediction results of this batch of data, 489 Metabolic Pathways, 2903 Enzymes, and 10,491 KEGG Orthologs were obtained. By calculating the top 10 functional variables in all samples, we found that the relative abundance of the top ten Metabolic Pathways accounted for less than 10%, and the same situation was observed in the Enzymes and KEGG Orthologs results. However, due to the large base number, the top 10 functional variables accounting for 5% of the total suggests that they play an important role in the gut. Among them, the Metabolic Pathways prediction results in the first row of Fig.  5 show that the highest relative abundance results are: NONOXIPENT-PWY (pentose phosphate pathway), PWY-7111 (isobutanol biosynthesis), PWY-6737 (starch degradation), PWY-5101 (L-isoleucine biosynthesis II), PWY-7220 (adenosine deoxyribonucleotides de novo biosynthesis), PWY-7222 (guanosine deoxyribonucleotides de novo biosynthesis), PWY-7663 (gondoate biosynthesis), PWY-5104 (L-isoleucine biosynthesis IV), CALVIN-PWY (Calvin-Benson-Bassham cycle), and PWY-5973 (cis-vaccenate biosynthesis). The above metabolic pathways involve multiple aspects such as carbohydrates, fatty acids, amino acids, and nucleotides, playing crucial roles in cellular energy supply, material synthesis, and signal transduction processes. The intermediate products of multiple pathways, such as cAMP, cGMP, and 3-isopropylmalate, have signal molecule functions and participate in metabolic regulation. Fig. 5 The top ten functional variables in relative abundance among all samples for Metabolic Pathways, Enzymes, and KEGG Orthologs. The first row of results shows the distribution of the mean values within subgroups for the top ten Metabolic Pathways in relative abundance and the proportion of relative abundance for each sample; the second row of results shows the top ten Enzymes in relative abundance; the third row of results shows the top ten KEGG Orthologs in relative abundance. Apart from the top ten functional variables, the relative abundances of the remaining variables are summed up and represented in dark gray. In the group labels, “T” represents patients with dysmenorrhea, and “F” represents patients without dysmenorrhea. EMs represents endometriosis The top ten functional variables in relative abundance among all samples for Metabolic Pathways, Enzymes, and KEGG Orthologs. The first row of results shows the distribution of the mean values within subgroups for the top ten Metabolic Pathways in relative abundance and the proportion of relative abundance for each sample; the second row of results shows the top ten Enzymes in relative abundance; the third row of results shows the top ten KEGG Orthologs in relative abundance. Apart from the top ten functional variables, the relative abundances of the remaining variables are summed up and represented in dark gray. In the group labels, “T” represents patients with dysmenorrhea, and “F” represents patients without dysmenorrhea. EMs represents endometriosis The Enzymes prediction results in the second row of Fig.  5 show that the highest relative abundance results are: EC:3.6.4.12 (DNA helicase), EC:2.7.7.7 (DNA-directed DNA polymerase), EC:2.7.13.3 (Histidine kinase), EC:1.6.5.3 (NADH dehydrogenase), EC:5.2.1.8 (Peptidylprolyl isomerase), EC:2.1.1.72 (Histone-arginine N-methyltransferase), EC:3.2.1.21 (beta-glucosidase), EC:3.4.16.4 (Angiotensin-converting enzyme), EC:2.7.7.6 (DNA-directed RNA polymerase), and EC:1.97.1.4 (Heparan-sulfate glucosaminyl 3-O-sulfotransferase). Although these enzymes catalyze different reactions and participate in various biological processes, they play key roles in important life activities such as gene expression, signal transduction, energy metabolism, protein folding, epigenetic regulation, coagulation, and inflammation. Their abnormalities are often associated with multiple diseases and are therefore important drug targets. The KEGG Orthologs prediction results in the third row of Fig.  5 show that the highest relative abundance results are: K03088 , K01990 , K06147, K02004 , K01992 , K02003 , K07024, K02529 , K05349, and K03497 . These KOs play important roles in cellular material transport, energy metabolism, signal transduction, and ionic homeostasis maintenance. Some of these genes are related to diseases and drug responses, and have important physiological significance and clinical application value. For example, K01990 (ABC-2.A) involves encoding a member of the ATP-binding cassette (ABC) transporter subfamily, participating in lipid and cholesterol transport, and is associated with Alzheimer’s disease and pulmonary surfactant secretion. Based on the microbial diversity analysis between dysmenorrhea-positive (Late-Stage-T) and dysmenorrhea-negative (Late-Stage-F) subgroups within the late-stage endometriosis group, as well as the comparison of gut microbial abundance and function between participants with and without dysmenorrhea, we discovered an association between preoperative dysmenorrhea and the gut microbiota. We therefore further analyzed the gut microbiota between early- and late-stage endometriosis groups. PCoA analysis revealed no significant structural difference between early- and late-stage groups among patients without preoperative dysmenorrhea (Fig.  6 A). However, a significant difference was observed between early- and late-stages in patients with preoperative dysmenorrhea (Fig.  6 B). Combined with Fig.  2 , we inferred that the between-group difference was driven by gut microbiota differences in dysmenorrhea patients between early- and late-stage group. Taken together, these findings show the preoperative dysmenorrhea positive late-stage endometriosis subgroup has a distinct gut microbiota composition compared to the other 3 subgroups, suggesting it represents an independent endometriosis subtype with unique gut microbial characteristics. Fig. 6 Comparison of gut microbial composition and function of early- and late-stage participants with and without dysmenorrhea. ( A ) Figures A present microbial diversity analysis results between dysmenorrhea negative subgroups between the early-stage endometriosis group (Early-Stage-F) and late-stage endometriosis group (Late-Stage-F). ( B ) Figures B present microbial diversity analysis results between dysmenorrhea positive subgroups between the early-stage endometriosis group (Early-Stage-T) and late-stage endometriosis group (Late-Stage-T). ( C ) The Venn diagram presents the intersecting differential species between these two comparisons. ( D ) The boxplots display the relative abundance levels of 10 differential species consistently enriched between early-stage and late-stage groups. The heatmap shows correlations of the differential species with r-AFS score and CA125. Heatmap colors represent Spearman correlation coefficient r values, with red indicating positive correlation and green denoting negative correlation. The asterisks indicate p -values from correlation analysis. ( E ) The Venn diagrams present intersections of differential metabolic pathways, Enzymes, and KEGG Orthologs predicted by PICRUSt2 among the subgroups, respectively Comparison of gut microbial composition and function of early- and late-stage participants with and without dysmenorrhea. ( A ) Figures A present microbial diversity analysis results between dysmenorrhea negative subgroups between the early-stage endometriosis group (Early-Stage-F) and late-stage endometriosis group (Late-Stage-F). ( B ) Figures B present microbial diversity analysis results between dysmenorrhea positive subgroups between the early-stage endometriosis group (Early-Stage-T) and late-stage endometriosis group (Late-Stage-T). ( C ) The Venn diagram presents the intersecting differential species between these two comparisons. ( D ) The boxplots display the relative abundance levels of 10 differential species consistently enriched between early-stage and late-stage groups. The heatmap shows correlations of the differential species with r-AFS score and CA125. Heatmap colors represent Spearman correlation coefficient r values, with red indicating positive correlation and green denoting negative correlation. The asterisks indicate p -values from correlation analysis. ( E ) The Venn diagrams present intersections of differential metabolic pathways, Enzymes, and KEGG Orthologs predicted by PICRUSt2 among the subgroups, respectively Further differential analysis between subgroups identified 28 differential species between early- and late-stage endometriosis patients without preoperative dysmenorrhea, and 32 differential species between early- and late-stage groups with preoperative dysmenorrhea (Fig.  6 C). Among these, 10 species were common across both comparisons. These 10 differential species were then selected for further analysis. Genus Bartonella , Snodgrassella , Bombella , and Commensalibacter were enriched in the late-stage group and demonstrated positive correlations with r-AFS and CA125 levels, suggesting their association with endometriosis progression. The correlations were assessed using Spearman correlation analysis to account for the non-parametric nature of the data (Fig.  6 D). Genus Bacteroidales , F082.5 , Succiniclasticum , Rikenellaceae , and taxa Prevotella ruminicola and Bacteroides caecimuris were increased in early-stage endometriosis and negatively correlated with r-AFS and CA125 level. The variable progression rate of endometriosis between individuals clinically suggests these microbes may relate to milder disease. Differential analysis of predicted functions revealed only 2 differential Enzymes (EC:2.1.1.41; EC:1.1.1.21) and 3 differential KEGG Orthologs ( K00559 ; K00011 ; K12688) consistently differing between early- and late-stage groups (Fig.  6 E). Interestingly, K00559 encodes sterol 24-C-methyltransferase (EC:2.1.1.41). K00559 is also involved in steroid biosynthesis (map00100), metabolic pathways (map01100), and biosynthesis of secondary metabolites (map01110). K00011 encodes aldehyde reductase (EC:1.1.1.21) and participates in pentose and glucuronate interconversions (map00040), fructose and mannose metabolism (map00051), galactose metabolism (map00052), glycerolipid metabolism (map00561), folate biosynthesis (map00790), and metabolic pathways (map01100). The higher abundance of enzymes EC:2.1.1.41 and EC:1.1.1.21 in the early-stage group indicates more active functions of the associated pathways occurring within the gut microbiome of early-stage endometriosis. In summary the findings revealed suppressed steroid biosynthesis pathways in the late-stage endometriosis gut microbiome.

From June 23, 2021, to July 5, 2022, a total of 8,399 patients admitted for surgical treatment at the Obstetrics and Gynecology Medical Center of Zhujiang Hospital, Southern Medical University, Guangzhou, China, were assessed for eligibility. Of these, 8,287 patients either did not meet the inclusion criteria or declined to participate. Ultimately, 112 patients consented to participate, and samples along with necessary data were collected from them. During data analysis, 37 patients were excluded due to incomplete data, leaving 75 patients whose data were included in the final analysis for this study. The inclusion criteria of current study include: (1) Female ≥ 18 years of age; (2) Subject has adequate understanding of study rationale; (3) Signed informed consent; (4) No sexual activity in the week prior to surgery; (5) No history of acute or chronic pelvic inflammatory disease; (6) Diagnosis of pelvic endometriosis was confirmed by laparoscopy and pathology; (7) No systemic or local antibiotics used in the 6 months prior to surgery. The exclusion criteria include: (1) Pregnancy; (2) Malignancy suggested by intraoperative or postoperative pathology; (3) Severe pelvic adhesions or anatomical abnormalities discovered during surgery; (4) History of gene therapy, blood transfusion, stem cell therapy, or bone marrow transplantation; (5) Psychiatric, personality disorders, or substance abuse; (6) Immunodeficiency, allergies, or autoimmune diseases; (7) Contraindications for endotracheal intubation and anesthesia; (8) Absolute or relative contraindications for laparoscopic surgery. In addition, participants were excluded if their demographic data were incomplete or if sequencing results did not pass quality control. To ensure accurate disease diagnosis, the revised American Fertility Society (r-AFS) score and staging of endometriosis were independently assessed by two experienced gynecologists for all patients. In cases where the two gynecologists’ assessments were inconsistent, a third senior gynecologist was consulted to provide a final diagnosis. Prior to surgery, all gynecological inpatients were assessed for eligibility based on the inclusion criteria. For eligible patients with suspected endometriosis, the rationale and clinical significance of this study were explained and informed consent was obtained before enrollment. Despite the lack of a standardized protocol for sample collection and processing in microbiome research, we followed widely accepted methodologies in the field to ensure the reliability of sample handling, processing, and storage [ 23 ]. Fecal samples were collected from enrolled patients 1 day before surgery. All fecal samples were collected by trained physicians at patient bedsides after disinfecting environments with 75% alcohol and using sterile disposable containers. Patients emptied bladders before defecation to prevent sample contamination. After collection, samples were immediately sealed, and transported to the laboratory. To prevent contamination from host-derived cells, fecal samples were aliquoted under sterile conditions using autoclaved spatulas to collect at least 3 mL from the core of each sample, which was then sealed in sterile centrifuge tubes and stored at -80 °C. After the completion of patient enrollment, all fecal samples from eligible patients underwent 16S rRNA amplicon sequencing at the same time to avoid batch effects. Given the high technical variability in DNA extraction and sequencing, we used the same reagent kits throughout the entire process to ensure consistency. Genomic DNA was isolated from each fecal sample utilizing the E.Z.N.A. ® Stool DNA Kit (D4015, Omega, Inc., USA) following the manufacturer’s protocols. The V3-V4 hypervariable region of the 16S rRNA gene was polymerase chain reaction (PCR) amplified using 341F (5’-CCTACGGGNGGCWGCAG-3’) and 805R (5’-GACTACHVGGGTATCTAATCC-3’) primers. Amplicons were sequenced on an Illumina NovaSeq platform per Illumina’s guidelines by LC-Bio. FastQC, a quality control tool for high-throughput sequence data, was used to assess the raw sequence data. The interquartile range of the quality score for all samples met Q30 quality control standards, ensuring high-quality sequencing data across the entire dataset. Demultiplexing assigned paired-end reads to samples based on unique barcodes. Primers and barcodes were truncated. Read pairs were merged with FLASH software. Raw reads underwent quality control filtering according to fqtrim (v0.94) parameters to generate high-quality clean tags. QIIME2 bioinformatics software conducted fecal microbiome profiling [ 24 ]. PICRUSt2 inferred functional metagenomes from marker gene surveys [ 25 ]. DADA2 filtered sequences and constructed feature Table [ 26 ], and amplicon sequence variants (ASVs) were generated. Silva 16S classifiers taxonomically characterized samples [ 27 ]. Alpha diversity, beta diversity, principal coordinate analysis (PCoA), data decontamination, and other microbiome downstream analyses and data visualizations were performed in the EasyMicroPlot R package [ 28 ]. Clinical and pathological characteristics of the patient cohorts were summarized using descriptive statistics. Continuous variables were reported as median and range. Categorical variables were presented as frequencies and percentages. All data analyses were conducted in the R version 4.2.1. The selection of statistical tests was based on the distribution of the clinical data. For normally distributed continuous data, comparisons between two groups were performed using Student’s t-test, as it assumes normality and homogeneity of variance. In cases where the data did not meet the assumption of normality, the Wilcoxon rank-sum test, a non-parametric alternative, was applied. For categorical data, the chi-squared test or Fisher’s exact test (for smaller sample sizes) was used. Comparisons among multiple groups utilized analysis of variance (ANOVA), Kruskal-Wallis test, and chi-squared test. Microbiome data between two groups were compared by Student’s t-test, while ANOVA and least significant difference (LSD) post hoc tests were used for more than two groups. For the PICRUSt2-predicted functional data, which is often sparse and non-normally distributed, we applied the Kruskal-Wallis test for group comparisons and Duncan’s post hoc test to further investigate significant differences among the four groups.

Analysis of 75 endometriosis microbiomes revealed distinct gut bacterial and functional signatures stratifying early- versus late-stage group. Dysmenorrhea-positive late-stage cases displayed uniquely altered microbial patterns constituting an independent subtype. Differential microbes correlated with validated clinical severity biomarkers: Bartonella , Snodgrassella and others were enriched in more advanced disease while Bacteroides , Prevotella decreased. Functional analysis implicated suppressed steroid biosynthesis pathways in severe microbiomes. Our findings provide microbiota-based disease subclassification, highlight gut community members linked to symptoms and progression, and nominate microbial activities underpinning pathogenesis. This study has some limitations. One limitation is that we only conducted recruitment for one year and set strict exclusion criteria, resulting in a relatively small sample size. Since all samples were collected prior to surgery during the enrollment period, great effort was made in screening participants and collecting specimens. We will continue related studies in the future to expand the sample size and further analyze the gut microbiota of endometriosis patients. Another limitation is the lack of a healthy control group. In our previous study, non-endometriosis patients with benign gynecological diseases served as controls, with all study participants undergoing surgery to confirm or exclude endometriosis. In future studies, we will attempt to enroll healthy individuals without disease or frailty, having intact physiological, psychological, and social adaptation capabilities as controls using clinical manifestations and noninvasive detection approaches. However, this method still cannot rule out asymptomatic endometriosis patients.

Through 16S rRNA gene amplicon sequencing and comprehensive bioinformatic analysis, we discovered significant differences in gut microbial profiles between early- and late-stage endometriosis patients. Notably, the gut microbiota of late-stage patients with preoperative dysmenorrhea differed distinctly from other endometriosis cohorts. Our study identified 10 microbes consistently enriched when comparing late- versus early-stage in both dysmenorrhea positive and negative groups, 4 of which correlated with disease severity. Additionally, we found 2 corresponding enzymes and KEGG Orthologs combinations enriched in the early-stage endometriosis gut microbiome. The experimental designs of previous studies focused on comparing gut and reproductive tract microbiota between endometriosis patients and non-endometriosis controls, or analyzing associations between female reproductive tract microbiota across different sites [ 9 – 11 , 31 – 35 ]. Existing human studies investigating the gut microbiota have reported inconsistent conclusions regarding whether endometriosis patients exhibit gut dysbiosis. Some researchers discovered gut microbial disturbances in endometriosis patients [ 9 , 11 ], while others found no overall differences from non-endometriosis controls [ 5 , 31 ]. Notably, endometriosis patients included in two of these studies comprised exclusively late-stage disease cases. No study has yet stratified early- versus late-stage endometriosis in analyses. Akiyama et al. only enrolled late-stage endometriosis to highlight differences from controls [ 6 ]. Shan et al. also only recruited late-stage patients without explaining their rationale [ 33 ]. Both studies indicate late-stage endometriosis can serve as an independent subgroup for microbiome analyses. Our study systematically explored differences and associations between early- and late-stage endometriosis by stratifying them as independent groups. Although no significant differences were observed in alpha diversity, distinct variations were evident in overall community structure. Differential gut microbes and functional pathways between early- and late-disease further demonstrate distinct gut microbial traits exist between these endometriosis stages. The relationship between intestinal diseases and the gut microbiota has been extensively studied. Within the human gut, the microbiota conducts complex metabolic activities, not only providing energy and nutrition required for human growth and reproduction, but also generating abundant metabolites affecting the host. As chemical messengers, these microbial metabolites mediate interactions between microbes and the host, exerting bifacial roles in human health [ 36 ]. They associate with diseases like IBS, IBD, cancer, autoimmunity, allergy and neurodegeneration. Of high interest is recent research on the gut microbiome-gut-brain axis in pain modulation. The gut microbiota can directly or indirectly regulate neuronal excitability in the peripheral nervous system by activating TLRs, GABA receptors, transient receptor potential (TRP) and acid-sensing ion channels (ASICs) [ 37 ]. As a chronic inflammatory condition, some endometriosis patients experience lower abdominal pain, sagging, waist soreness or other discomfort before, during or after menstruation, severely impacting quality of life [ 38 ]. First-line therapies for dysmenorrhea include NSAIDs, acetaminophen and/or hormonal contraceptives. In recent years, the gut microbiome and gut-brain axis have been closely linked to various chronic pain types, and the gut flora also affects opioid tolerance. Opioid is also known as painkillers. Studies show germ-free mice exhibit visceral hypersensitivity at birth, alongside heightened spinal Toll-like receptor expression and cytokines, alleviated after conventional microbiota colonization. This suggests a regulatory role of commensal gut microbes in maintaining equilibrium of colonic sensory neuronal excitability [ 39 ]. Since only some endometriosis patients present dysmenorrhea, we speculated the gut microbiota was involved. Our subgroup analyses categorizing early-stage and late-stage cohorts by dysmenorrhea history revealed that late-stage endometriosis patients with dysmenorrhea showed significantly different gut microbial traits compared to the other three subgroups. This finding supports our hypothesis of an association between endometriosis-related dysmenorrhea and the gut microbiome. There are over 1,000 varieties of steroids that have been reported in nature, including sterols, steroid hormones, and bile acids. Previous studies have shown steroid hormones are a group of hormones, including glucocorticoids, mineralocorticoids, androgens, estrogens and progestogens, and endometriosis patients exhibit hormonal-dependent characteristics. The latest research suggests endometriosis associates closely with factors like progesterone resistance. Microbiome-related studies indicate the gut microbiota can regulate host estrogen and androgen levels by generating, reactivating and degrading sex hormones [ 40 ]. These modulations are quantitatively sufficient to impact host physiological states. Microbially-derived changes in estradiol and testosterone levels correlate with diseases like endometriosis, prostate cancer and depression [ 41 ]. Endometriosis is an estrogen-dependent disease, and clinical medications like combined oral contraceptives (COC) and gonadotropin releasing hormone agonists (GnRH-a) control endometriosis progression by generating a low estrogen internal environment. The gut microbiota collectively can activate steroids and convert dietary polyphenols into estrogen mimics [ 42 ]. Conversely, hormones also impact bacteria. For example, Agrobacterium tumefaciens and Pseudomonas associate with estriol and estradiol [ 43 ]; Clostridium scindens transforms glucocorticoids into androgens [ 44 ]; while estradiol and progesterone promote the growth of Platysaurus intermedius [ 45 ]. Germ-free animal models demonstrate gut microbes are essential for maintaining normal bodily sex hormone levels [ 46 – 48 ]. The current study found suppressed steroid biosynthesis pathways in the late-stage endometriosis gut microbiome. We speculate gut microbes may mediate endometriosis progression through steroids. Many mysteries still exist surrounding the microbial endocrinology field. Our findings suggest future studies could devote more efforts towards investigating relationships between steroids, the gut microbiota and endometriosis. The World Health Organization (WHO) indicates that in addition to early diagnosis and appropriate medications, a key factor for successful endometriosis treatment is patient adherence. During diagnosis and treatment, doctors should focus more on patient health education. As endometriosis is incurable, women of childbearing age require ongoing treatment after diagnosis until menopause or planned pregnancy. Commonly used drugs clinically include NSAIDs and hormonal modalities. Such traditional regimens have clear limitations—NSAIDs can cause gastrointestinal reactions, progesterone can lead to breast tenderness, pregnancy cannot be planned during treatment periods. Thus, safe and effective alternative or adjuvant therapies are urgently needed. In recent years, development of probiotics and engineered bacterial strains has surged. Numerous studies prove the efficacy of probiotics in treating various diseases [ 49 – 52 ]. Probiotics have good palatability and dietary therapy is easily accepted and adhered to by the public. However, our study did not identify differentially abundant probiotic species between early- and late-stage endometriosis, possibly due to limitations of 16S amplicon sequencing. As our results showed, many differential microbes exist between early and late-stage groups with complex constituent bacteria lacking systematic characterization methods currently. Hence, some scholars propose microbiome research based on functional units may better meet actual demands. For example, butyrate-producing bacteria generate anti-inflammatory, immunomodulatory butyrate [ 53 ]; gut microbes like Bifidobacterium , Clostridium and Lactobacillus produce enzymes regulating intestinal estrogen metabolism [ 54 , 55 ]. This study found suppressed steroid biosynthesis pathways in late-stage endometriosis, suggesting enteric bacteria (s.g., the genera Bifidobacterium and Lactobacillus ) supplementation may modulate hormonal metabolism thereby delaying endometriosis progression. However, larger cohorts and mechanistic studies are still necessary to further elucidate relationships between microbes involved in steroid synthesis and endometriosis advancement.

Endometriosis is a prevalent gynecological disorder that affects approximately 10% of women of reproductive age worldwide. The condition is characterized by the presence of endometrial-like tissue outside the uterus, leading to severe pain, infertility, and a plethora of gastrointestinal issues. Despite its widespread impact, the etiology of endometriosis remains enigmatic, resulting in delayed diagnosis and a lack of curative treatments [ 1 , 2 ]. Furthermore, the disease is marked by a multifactorial cause, suggesting that multiple factors may contribute to its development and progression. As a result, researchers must continue to delve deeper into the underlying mechanisms of endometriosis to unravel its intricate pathophysiology and develop more effective, safe, and low-recurrence treatment options. Emerging research in recent years has demonstrated interactions between the gut microbiota and multiple human organs, and the critical roles of the gut microbiota in disease initiation and progression [ 3 , 4 ]. Recent scientific and clinical findings have begun to unveil the complex relationship between endometriosis and the microbiota in different body sites, including the female reproductive tract, gastrointestinal tract, and peritoneal region [ 5 – 11 ]. The human gut microbiota is an intricate ecosystem composed of trillions of microorganisms that play a crucial role in health and disease. In clinical practice, the use of antibiotics in endometriosis patients can alleviate dysmenorrhea symptoms, and animal experiments have demonstrated that antibiotics can reduce inflammation, congestion, and adhesion in ectopic endometrial lesions [ 12 ]. Some researchers have also proven through animal experiments that antibiotics can reduce inflammatory factors, thereby alleviating endometriosis [ 12 ]. In our previous research, we collected paired samples of feces, cervical mucus, and peritoneal lavage fluid for 16S rRNA amplicon sequencing [ 11 ]. Bioinformatic analysis revealed that endometriosis patients exhibited dysbiosis in their microbial composition, with the gut microbial structure being significantly different from that of non-endometriosis patients. Furthermore, by establishing a disease predictive model, we found that the gut microbiota is more meaningful than the cervical mucus microbiota in the early diagnosis of endometriosis. In a meta-analysis [ 13 ], we also discovered that the Shannon index of the gut microbiota was significantly reduced in endometriosis patients, while the α-diversity indices of the vaginal and cervical microbiota did not differ significantly compared to the control group. Therefore, we hypothesize that the gut microbiota plays an important role in the progression of endometriosis, with microbial diversity being a key factor. However, the specific gut microbiota states in early-stage versus late-stage endometriosis remain unclear. Studies indicate that stage I/II endometriosis is primarily a pro-inflammatory state, while stage III/IV tends towards immune tolerance [ 14 ]. A meta-analysis revealed that the cellular microenvironment and immune cell profiles of the eutopic endometrium differ between women with stage I-II and stage III-IV endometriosis [ 15 ]. M1 macrophages are more prevalent in stage I-II, while M2 macrophages are more common in stage III-IV endometriosis, suggesting that M1 to M2 polarization may be crucial for disease progression [ 16 ]. A case-control study collected peripheral blood and peritoneal fluid from 31 Stage I-II cases, 39 Stage III-IV cases, and 36 controls [ 17 ]. Flow cytometry was used to determine percentages of Treg and Th17 cells. Results showed significantly higher peritoneal fluid Treg percentages in Stage III-IV patients compared to Stage I-II cases and controls. The authors propose an imbalance between Treg and Th17 cells may be involved in endometriosis onset and progression, promoting survival and implantation of ectopic endometrial tissue. Comparison studies between ectopic lesions and normal endometrium in endometriosis patients have found increased macrophages in both eutopic endometrium and lesions among Stage I-II cases [ 18 ]. Animal studies have shown that C57BL/6 mice tend to produce Th1/M1 macrophage-dominated immune responses, while BALB/c mice produce Th2/M2 macrophage-dominated responses [ 19 ]. C57BL/6 mice (Th1/M1) predominantly develop small, compact lesions (similar to early-stage endometriosis), while BALB/c mice (Th2/M2) mostly develop large, cystic lesions (similar to late-stage endometriosis). In addition to endometriosis biomarkers existing in peripheral blood and peritoneal fluid, our previous study also suggested an association between microorganisms in the pelvic cavity and endometriosis [ 11 ]. In the early stages of the disease, pro-inflammatory factors dominate the local microenvironment. In the late stages, immunity tends towards a tolerant state, and IL-27 accumulated in the microenvironment of ectopic lesions inhibits Th17 differentiation and promotes IL-10 production by Th17 cells through the c-Maf/RORC/Blimp-1 complex, participating in the formation of an immune tolerance pattern in late-stage endometriosis [ 20 ]. Together, existing evidence indicates differences in peripheral blood, peritoneal fluid, lesions and more between early- (Stage I-II) and late-stage (Stage III-IV) endometriosis. However, no studies have yet focused on characterizing gut microbial profiles specific to early- versus late-stage endometriosis. Additionally, previous studies have linked pain perception to the gut microbiota; for example, sex differences in pain sensitivity and tolerance may be partially mediated by specific gut microbes like Prevotella and Staphylococcus species [ 21 , 22 ]. We therefore questioned whether the gut microbiota may also associate with dysmenorrhea symptoms in endometriosis patients. The current study enrolled eligible patients who underwent surgery at the Obstetrics and Gynecology Medical Center, Zhujiang Hospital between June 2021 to July 2022. The fecal samples were collected prior to surgery and underwent 16S rRNA amplicon sequencing. And the sequencing data was analyzed using bioinformatic techniques. The findings of this study suggest that the gut microbiota of early-stage and late-stage endometriosis patients have distinctly different characteristics, with the gut microbiome of late-stage patients with concomitant dysmenorrhea especially distinct. By exploring the role of the gut microbiota in the onset and progression of endometriosis and identifying potential biomarkers to monitor disease progression, this study contributes to the discovery of methods to intervene in disease progression, reduce pain, and prevent recurrence.

Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1