To determine the impact of OC on the microbiome, we recruited women undergoing hysterectomy for OC or a benign gynecologic condition requiring hysterectomy. We then proceeded to compare the microbiome of patients with and without OC. Following that analysis, we focused on patients with OC and assessed the prognostic potential of the microbiome. We examined this impact using various α- (Inverse Simpson, and Shannon indices and observed ASVs) and β- (weighted, unweighted, and generalized UniFrac distances and Bray–Curtis) measures as well as differential abundance analysis. We report the α- and β-diversity measures with the most significant results in the main text and the remaining in the supplemental material.

Patient demographics

We collected microbiome samples from a total of 64 women undergoing hysterectomy for either OC (n = 30) or a benign gynecologic condition (n = 34) at the Mayo Clinic in Rochester, MN (Table 1). Women with various benign gynecologic conditions were used as controls to characterize the microbiome specific to OC. The age (p = 0.672), menopausal state (p = 0.251) and body mass index (BMI) (p = 0.353) distributions were similar between both cohorts as shown in Table 1. We also show the results of tumor response to treatment as well as patients’ status two years and four years post-diagnosis (Table 2).

Table 1 Patient demographics. Full size table

Table 2 Patient treatment response. Full size table

The microbiome associated with ovarian cancer and benign uterine gynecologic conditions

We sampled along the female reproductive tract (vagina, cervix, uterus, Fallopian tubes, ovaries), as well as ascites or peritoneal fluid, omentum (OC cohort only), urine, and stool (OC cohort only) to characterize the microbiomes of patients with either OC or a benign gynecologic condition. While the lack of omental and stool samples in the benign cohort did not allow for the comparisons between the two cohorts in these sample types, we were able to include the analysis of these samples in the OC cohort focusing on the impact of stage, grade, histology, and treatment response. The high throughput sequencing of the V3–V5 region of the 16S rRNA gene of all the 751 samples collected, including controls, yielded a total of 7076 ASVs. Our decontamination process (filtering out microbial taxa more abundant in the negative controls as well as present in more than one negative control) resulted in the removal of potential contaminants as shown in the abundance and relative abundance plots in Supplemental Fig. S1. The results of our taxonomic analysis showed that the microbiomes from the same body site of both benign and OC cohorts are generally dominated by the same microbial taxa to varying amounts (Fig. 1). For instance, the relative abundance of Lactobacillus in the vagina is only ~ 15% in the OC cohort compared to the ~ 30% in the benign cohort (Fig. 1). Several of these microbial taxa are also dominant across body sites. For example, Lactobacillus appears to be a dominant species in OC (vagina, cervix, uterus, Fallopian tubes, ovaries, and omentum) and benign (vagina, cervix, and urine) cohorts. Ezakiella also appear to be dominant across multiple sites (uterus, Fallopian tubes, urine, stool) in both OC and benign cohorts (Fig. 1). While Peptoniphilus is dominant in the cervix and ovaries of the benign cohort, Porphyromonas is particularly dominant in the Fallopian tubes and ovaries of the benign cohort and the stool of the OC cohort (Fig. 1). We also observed dominant levels of Bacteroides in the uterus, Fallopian tubes, ovary, ascites, and stool of the OC cohort (Fig. 1). Both Prevotella and Streptococcus are also dominant in the vagina and uterus of both OC and benign cohorts (Fig. 1).

Figure 1 Genus-level microbial community composition (relative abundance) plots of patients with or without OC. (A) Vagina. (B) Cervix. (C) Uterus. (D) Fallopian tubes. (E) Ovary. (F) Ascites/Peritoneal fluid. (G) Omentum. (H) Urine. (I) Stool. Only microbial taxa present at a minimum of 1% relative frequency in at least one participant are shown for graphical clarity. Full size image

The distinguishing potential of the microbiome in ovarian cancer

Microbiome compositions of ovarian cancer patients differ significantly from those of patients with benign gynecologic conditions

To further investigate the microbiome associated with OC, we summarized the differences in the microbiome composition between patients with or without OC using various α- (within-sample richness and evenness) and β- (between-sample) diversity measures. After adjusting for batch differences where necessary (See Methods), we compared the vaginal and cervical samples, and the results showed no significant differences (unweighted UniFrac: p = 0.814) between them in agreement with results from our previous studies19,23. We therefore combined the vaginal and cervical samples (lower reproductive tract, LRT) by adding sequence reads from both body sites for each patient in the rest of the present analysis. Our results revealed statistically significantly higher α-diversity in the LRT of the OC cohort compared to the benign cohort (Fig. 2A, Observed ASVs: p = 0.049; See Supplemental Fig. S2 for other metrics) which was not seen in the other body sites (uterus: Fig. 2C, Fallopian tubes: Fig. 2E, ovaries: Fig. 2G and urine: Fig. 2K). We also observed statistically significant β-diversity differences in the uterus (Fig. 2D, unweighted UniFrac: p = 0.004; Supplemental Fig. S3, weighted UniFrac: p = 0.028), Fallopian tube (Fig. 2F, Bray–Curtis: p = 0.025) and urine (Fig. 2L, Bray–Curtis, p = 0.047) between the benign and OC cohorts. Of note, we also observed differences in the β-diversity of the LRT (Fig. 2B, unweighted UniFrac: p = 0.052) and ovarian (Fig. 2H, Bray–Curtis, p = 0.088) microbiomes between the benign and OC cohorts that aligned with other organs but did not reach statistical significance. These differences resulted in general enrichment of several taxa, including Corynebacterium tuberculostearicum, Facklamia hominis and Ruminococcus faecis in the LRT and the depletion of Microbacterium lacus in the ovaries of the OC cohort (Fig. 2M and N; Supplemental Tables 1–3).

Figure 2 Bacterial community α- and β-diversities between patients with and without OC. Both α- (Observed ASVs) and β-diversity measures were compared. For α-diversity a Wald statistical test was performed and Observed ASVs was reported. For β-diversity, Bray–Curtis (BC) and unweighted UniFrac distance metrics were reported. The most significant metric is shown in each ordination plot. Lower reproductive tract (cervix and vagina), (A) Benign vs OC, α-diversity, p = 0.049, (B) Benign vs OC, β-diversity, unweighted UniFrac, p = 0.052. Uterus, (C) Benign vs OC, α-diversity, p = 0.907, (D) Benign vs OC, β-diversity, p = 0.004. Fallopian tube, (E) Benign vs OC, α-diversity, p = 0.201, (F) Benign vs OC, β-diversity, p = 0.025. Ovaries, (G) Benign vs OC, α-diversity, p = 0.766, (H) Benign vs OC, β-diversity, p = 0.088. Ascites/Peritoneal fluid, (I) Benign vs OC, α-diversity, p = 0.882, (J) Benign vs OC, β-diversity, p = 0.007. Urine, (K) Benign vs OC, α-diversity, p = 0.382. (L) Benign vs OC, β-diversity, p = 0.047. (M–O) Heatmaps showing the effect size (Log 2 Fold Change) of the differentially abundant microbial taxa. White boxes reflect no fold change at FDR < 0.10. Analysis was adjusted for menopause status, and BMI. Samples rarefied prior to analysis. Analysis was adjusted for menopause status and BMI. *Groups are significantly different. Full size image

Ovarian cancer microbiome according to stage, grade and histology

Following the general characterization of the microbiome from both OC and benign cohorts, we focused on characterizing the microbiome associated with the stage, grade, and histology of OC (Table 2).

Significant association between stage and the microbiome

The presented results showed potentially important associations between OC stage and various measures of diversity. Specifically, a significant association between OC stage and α-diversity was observed across several sampling sites (Fig. 3 and Supplemental Fig. S4). Our results showed a statistically significant association of stage with the α-diversity of the LRT microbiome (Fig. 3A, Shannon, p = 0.034; Supplemental Fig. S4). The benign cohort had significantly lower (early-stage: Shannon, p = 0.019) and higher (advanced-stage: Shannon, p = 0.019) α-diversity than the OC cohort. We also showed significant association of stage with the β-diversity in multiple organs (Fig. 3; Supplemental Fig. S5). These include benign vs. early-stage (Fig. 3D, uterus: unweighted UniFrac, p = 0.002), benign vs. advanced-stage (Fig. 3D, uterus: unweighted UniFrac, p = 0.006), and early- vs. advanced-stage (Fig. 3H, ovaries: unweighted UniFrac, p = 0.039; Fig. 3N, stool: Bray–Curtis, p = 0.042). We also observed differences trending toward significant in the LRT (benign vs. early-stage: unweighted UniFrac, p = 0.061; benign vs. advanced-stage: unweighted UniFrac, p = 0.059; early- vs. advanced-stage: unweighted UniFrac, p = 0.065), Fallopian tube (early- vs. advanced-stage: Bray–Curtis, p = 0.078) and urine (benign vs. advanced-stage: Bray–Curtis, p = 0.086). Our differential abundance analysis results revealed general enrichment of several taxa in the LRT (Peptoniphilus koenoeneniae, Facklamia hominis, Ruminococcus faecis, Fenollaria massiliensis) and urine (Dialister propionicifaciens and Anaeroglobus geminatus) of patients in both early- and advanced-stages of OC compared to the benign cohort (Fig. 3O; Supplemental Tables 1–2, and 4–5). We however observed general depletion of microbial taxa in the LRT (Corynebacterium sp. and Dialister sp.), uterus (Corynebacterium tuberculostearicum and Roseateles depolymerans), urine (Prevotella bergensis, Dialister propionicifaciens and Anaeroglobus geminatus) and stool (Peptoniphilus duerdenii, Prevotella buccalis, Mobiluncus curtisii, Porphyromonas bennonis and Alistipes shahii) of advanced-stage OC patients in comparison to early-stage OC patients (Fig. 3O; Supplemental Tables 1–2 and 6).

Figure 3 Bacterial community α- and β-diversities among patients with and without different stages of Ovarian Cancer (OC). Both α- and β-diversity measures were compared. For α-diversity a Wald statistical test was performed and Observed ASVs, Shannon Index and Inverse Simpson were reported. For β-diversity, Bray–Curtis (BC), unweighted, weighted, and generalized UniFrac distance metrics were reported. The most significant metric is shown in each ordination plot. Lower reproductive tract (cervix and vagina), (A) α-diversity: Benign vs Early stage (p = 0.019), Advanced stage (p = 0.025), (B) β-diversity: Benign vs Early stage (p = 0.061), Advanced stage (p = 0.059), Early vs Advanced stage (p = 0.065). Uterus, (C) α-diversity: Benign vs Early stage (p = 0.461), Advanced stage (p = 0.105), (D) β-diversity: Benign vs Early stage (p = 0.002), Advanced stage (p = 0.006), Early vs Advanced stage (p = 0.284). Fallopian tube, (E) α-diversity: Benign vs Early stage (p = 0.384), Advanced stage (p = 0.196), (F) β-diversity: Benign vs Early stage (p = 0.127), Advanced stage (p = 0.127), Early vs Advanced stage (p = 0.078). Ovaries, (G) α-diversity: Benign vs Early stage (p = 0.872), Advanced stage (p = 0.447), (H) β-diversity: Benign vs Early stage (p = 0.433), Advanced stage (p = 0.240), Early vs Advanced stage (p = 0.039). Ascites/Peritoneal fluid, (I) α-diversity: Benign vs Early stage (p = 0.054), Advanced stage (p = 0.010), (J) β-diversity: Benign vs Early stage (p = 0.166), Advanced stage (p = 0.028), Early vs Advanced stage (p = 0.091). Urine, (K) α-diversity: Benign vs Early stage (p = 0.310), Advanced stage (p = 0.380), (L) β-diversity: Benign vs Early stage (p = 0.175), Advanced stage (p = 0.086), Early vs Advanced stage (p = 0.566). Stool, (M) α-diversity: Benign vs Advanced stage (p = 0.302), (N) β-diversity: Early vs Advanced stage (p = 0.042). (O) Heatmaps showing the effect size (Log 2 Fold Change) of the differentially abundant microbial taxa. White boxes reflect no fold change at FDR < 0.10. Analysis was adjusted for menopause status, and BMI. Samples rarefied prior to analysis. Wald statistical test with Q value cutoff = 0.1. *Groups are significantly different. Full size image

Significant association between grade and the microbiome

While we did not observe any significant association of grade with α-diversity, our results revealed significant association of grade with β-diversity in the uterine and ovarian microbiomes (Fig. 4; Supplemental Fig. S6 and S7). These include significant differences between benign vs. low-grade (Fig. 4D, uterus: generalized UniFrac, p = 0.023), benign vs. high-grade (Fig. 4D, uterus: generalized UniFrac, p = 0.014), and low- vs. high-grade (Fig. 4D, uterus: generalized UniFrac, p = 0.019; Fig. 4F, ovaries: Bray–Curtis, p = 0.045) OC patients. Differences which were not quite significant were also observed between benign and high-grade OC patients in the LRT (Fig. 4B, unweighted UniFrac, p = 0.087), ovaries (Fig. 4F, Bray–Curtis, p = 0.067) and urine (Fig. 4J, Bray–Curtis, p = 0.056). The results of the differential abundance analysis revealed general enrichment of several taxa in the LRT of both low- and high-grade patients compared to the benign cohort (Fig. 4K; Supplemental Tables 1–2, and 7–8). The enriched taxa include Streptococcus infantis, Fusobacterium nucleatum, Varibaculum cambriense, Escherichia coli, Faecalibacterium prausnitzii, and Bacteroides fragilis. Comparing LRT microbiome of low-grade OC patients to that of high-grade OC patients, however, results in the depletion of these microbial taxa in the high-grade OC patients (Fig. 4K; Supplemental Tables 1–2, and 9). We also observed similar trends in the urinary microbiome with general depletion of microbial taxa in high-grade OC patients compared to the low-grade OC patients. A few examples of the depleted taxa include Peptostreptococcus anaerobius, Mobiluncus curtisii, Dialister propionicifaciens, Peptoniphilus. bennonis, and Atopobium deltae (Fig. 4K; Supplemental Tables 1–2, and 9).

Figure 4 Bacterial community α- and β-diversity among patients with and without different grades of OC (OC). Both α- and β-diversities measures were compared. For α-diversity a Wald statistical test was performed and Observed ASVs was reported. For β-diversity, Bray–Curtis (BC), unweighted, and generalized UniFrac distance metrics were reported. The most significant metric is shown in each ordination plot. Lower reproductive tract (cervix and vagina), (A) α-diversity: Benign vs Low grade (p = 0.221), High grade (p = 0.997), (B) β-diversity: Benign vs Low grade (p = 0.195), High grade (p = 0.087), Low grade vs High grade (p = 0.406). Uterus, (C) α-diversity: Benign vs Low grade (p = 0.400), High grade (p = 0.159), (D) β-diversity: Benign vs Low grade (p = 0.023), High grade (p = 0.014), Low grade vs High grade (p = 0.019). Ovaries, (E) α-diversity: Benign vs Low grade (p = 0.972), High grade (p = 0.552), (F) β-diversity: Benign vs Low grade (p = 0.350), High grade (p = 0.067), Low grade vs High grade (p = 0.045). Ascites/Peritoneal fluid, (G) α-diversity: Benign vs High grade (p = 0.536), (H) β-diversity: Benign vs High grade (p = 0.016). Urine, (I) α-diversity: Benign vs Low grade (p = 0.490), High grade (p = 0.534), (J) β-diversity: Benign vs Low grade (p = 0.615), High grade (p = 0.056), Low grade vs High grade (p = 0.717). (K) Heatmaps showing the effect size (Log 2 Fold Change) of the differentially abundant microbial taxa. White boxes reflect no fold change at FDR < 0.10. Analysis was adjusted for menopause status, and BMI. Samples rarefied prior to analysis. Wald statistical test with Q value cutoff = 0.1. *Groups are significantly different. Full size image

Significant association between histology and the microbiome

Consistent with results obtained from stage, histological features of OC are significantly associated with both α-diversity and β-diversity in multiple body sites (Fig. 5; Supplemental Figs. S8 and S9). There was a significant overall association of histology with the α-diversity of the LRT microbiome (Fig. 5A, Shannon, p = 0.045), with a significantly lower α-diversity in the benign cohort compared to other OC histologies (Shannon, p = 0.015). The β-diversity analysis results showed significant differences between the patients with benign lesions and serous OC (Fig. 5B, LRT: unweighted UniFrac, p = 0.038; Fig. 5D, uterus: unweighted UniFrac; p = 0.002), between patients with benign conditions and other histologies (Fig. 5L, urine: generalized UniFrac, p = 0.048) as well as between serous and other histologies (Fig. 5J, omentum: generalized UniFrac). We also observed differences between the microbiota of ovaries from patients with benign lesions vs serous OC that aligned with other organs but did not reach statistical significance (Bray–Curtis, p = 0.051). Results from the differential abundance analysis revealed general enrichment of several microbial taxa in the LRT, Fallopian tube, omentum, and urine of OC patients with serous and other histologies compared to the benign cohort (Fig. 5M; Supplemental Tables 1–2, and 10–11). These enriched microbial taxa include Facklamia hominis, Anaerococcus senegalensis, Lactobacillus iners, and Actinomyces turicensis. Within the OC patient cohort, the results of the differential abundant analysis also showed enrichment of microbial taxa including Lactobacillus iners, Fusobacterium nucleatum, Prevotella buccalis, and Dialister propionicifaciens, in patients with other OC histologies in comparison to the serous OC patients (Fig. 5M; Supplemental Tables 1–2, and 12).

Figure 5 Bacterial community α- and β-diversities among patients with and without different histology of OC (OC). Both α- and β-diversities measures were compared. For α-diversity a Wald statistical test was performed and Observed ASVs was reported. For β-diversity, Bray–Curtis (BC), unweighted, and generalized UniFrac distance metrics were reported. The most significant metric is shown in each ordination plot. Lower reproductive tract (cervix and vagina), (A) α-diversity: Benign vs serous (p = 0.021), others (p = 0.859), (B) β-diversity: Benign vs serous (p = 0.038), others (p = 0.238), serous vs others (p = 0.275). Uterus, (C) α-diversity: Benign vs serous (p = 0.767), others (p = 0.459), (D) β-diversity: Benign vs serous (p = 0.002), others (p = 0.123), serous vs others (p = 0.400). Ovaries, (E) α-diversity: Benign vs serous (p = 0.918), others (p = 0.234), (F) β-diversity: Benign vs serous (p = 0.051), others (p = 0.433), serous vs others (p = 0.138). Ascites/Peritoneal fluid, (G) α-diversity: Benign vs serous (p = 0.636), others (p = 0.807), (H) β-diversity: Benign vs serous (p = 0.019), others (p = 0.667), serous vs others (p = 0.571). Omentum, (I) α-diversity: Serous vs others (p = 0.377), (J) β-diversity: serous vs others (p = 0.003). Urine, (K) α-diversity: Benign vs serous (p = 0.360), others (p = 0.911), (L) β-diversity: Benign vs serous (p = 0.329), others (p = 0.048), serous vs others (p = 0.129). (M) Heatmaps showing the effect size (Log 2 Fold Change) of the differentially abundant microbial taxa. White boxes reflect no fold change at FDR < 0.10. Analysis was adjusted for menopause status, and BMI. Samples rarefied prior to analysis. Wald statistical test with Q value cutoff = 0.1. *Groups are significantly different. Full size image

Microbiome prognostic potential for ovarian cancer treatment

The ovarian cancer microbiome is prognostic of treatment response

Because the microbiome samples were collected from treatment naïve patients, we also investigated the role of microbiome in treatment response to better understand the prognostic potential of the microbiome at the time of hysterectomy. We explored outcome data including tumor response, patients’ status two years and four years post-diagnosis (Table 2). Our results showed significant association of the tumor response with both α-diversity and β-diversity in multiple body sites (Fig. 6A–D; Supplemental Figs. S10–S11). We found a significantly lower α-diversity (Inverse Simpson, p = 0.044) in the omental microbiome of patients who had chemotherapy sensitive OCs in comparison to those who did not (Supplemental Fig. S10). Our β-diversity results also showed significant differences between patients with chemotherapy sensitive OCs (Fallopian tube: unweighted UniFrac, p = 0.003; urine: unweighted UniFrac, p = 0.015) compared to refractory/resistant (other) OCs (Fig. 6A and D; Supplemental Fig. S11). These differences however did not result in differentially abundant microbial taxa between the two groups (Fig. 6; Supplemental Tables 1–2, and 13). We further analyzed the potential of the microbiome to predict patients’ status two years and four years post-diagnosis (Table 2). Our results showed significant differences in both α-diversity and β-diversity in multiple body sites (Fig. 6E–J; Supplemental Figs. S12–S15). We report a significantly higher α-diversity (uterus: Shannon, p = 0.038) in patients who were alive with no adverse events after two years compared to those who were deceased (Supplemental Fig. S12). These differences are also seen in the β-diversity results (Fig. 6; Supplemental Fig. S13) with significant differences between patients who were alive but experienced adverse events and those who were deceased two years post-diagnosis (Fig. 6G, omentum: unweighted UniFrac, p = 0.010; Fig. 6I, stool: unweighted UniFrac, p = 0.050). We also observed significant differences in the β-diversity of patients who were alive with no adverse events and those who experienced adverse events (Fig. 6J, LRT: unweighted UniFrac, p = 0.017; Supplemental Figs. S14–S15) 4 years post-diagnosis. While not statistically significant, we also observed differences between patients who were alive with adverse events and those who were deceased (Fig. 6J, LRT: unweighted UniFrac, p = 0.058). Our differential abundance analysis identified several differentially enriched microbial taxa in the urine and stool of patients who were alive but experienced adverse events and those who were deceased compared to those who were alive with no adverse events two years post-diagnosis (Fig. 6K Supplemental Tables 1–2, and 14–16). Examples include Lactobacillus gasseri, Diasliter invisus., Blautia pseudococcoides, Veillonella nakazawae, Bacteroides ovatus, Butyricicoccus faecihominis and Sutterella wadsworthensis, (Fig. 6K; Supplemental Tables 1–2, and 14–16). The LRT microbiomes of patients who were alive with adverse events had generally enriched taxa (Lactobacillus iners, Fenollaria massiliensis, Ezakiella coagulans, and Campylobacter ureolyticus, and Actinomyces urogenitalis) in comparison to those who were alive without event four years post-diagnosis (Fig. 6L; Supplemental Tables 1–2, and 17). We further observed general depletion of Prevotella bucalis in patients who were deceased compared to those who were alive with or without adverse events (Fig. 6L; Supplemental Tables 1–2, and 18–19).

Figure 6 β-diversities measures were compared. For β-diversity, Bray–Curtis (BC), unweighted, and weighted, UniFrac distance metrics were reported. The most significant metric is shown in each ordination plot. (A–D) Bacterial community β-diversity between OC patients with sensitive vs other (resistant/refractory) tumor response. Fallopian tube, (A) β-diversity: sensitive vs other (p = 0.003). Ovaries, (B) β-diversity: sensitive vs other (p = 0.073). Ascites, (C) β-diversity: sensitive vs other (p = 0.021). Urine, (D) β-diversity: sensitive vs other (p = 0.015). (E–I) Bacterial community β-diversity among OC patients with different status two years post-diagnosis. Ovaries, (E) β-diversity: Alive, no event vs alive, event (p = 0.073), Alive, no event vs dead (p = 0.761), Alive, event vs dead. (p = 0.437). Ascites, (F) β-diversity: Alive, no event vs alive, event (p = 0.573), Alive, no event vs dead (p = 0.029), Alive, event vs dead (p = 0.250). Omentum, (G) β-diversity: Alive, no event vs alive, event (p = 0.273), Alive, no event vs dead (p = 0.350), Alive, event vs dead (p = 0.010). Urine, (H) β-diversity: Alive, no event vs alive, event (p = 0.088), Alive, no event vs dead (p = 0.347), Alive, event vs dead (p = 0.356). Stool, (I) β-diversity: Alive, no event vs alive, event (p = 0.063), Alive, no event vs dead (p = 0.601), Alive, event vs dead (p = 0.050). (J) Bacterial community β-diversity among OC patients with different status four years post-diagnosis. Lower reproductive tract (cervix and vagina), (J) β-diversity: Alive, no event vs alive, event (p = 0.017), Alive, no event vs dead (p = 0.568), Alive, event vs dead (p = 0.058). (K–L) Heatmaps showing the effect size (Log 2 Fold Change) of the differentially abundant microbial taxa. White boxes reflect no fold change at FDR < 0.10. Analysis was adjusted for menopause status, and BMI. Samples rarefied prior to analysis. Wald statistical test with Q value cutoff = 0.1. *Groups are significantly different. Full size image

Microbiome composition of malignant versus benign peritoneal fluid

The volume of ascites at initial surgery of epithelial OC has been shown to be an important clinical parameter in the prognosis of the disease24. We therefore compared the peritoneal fluid microbiome of patients with or without OC to characterize the microbiome composition associated with ascites. Here we compared the properties of peritoneal fluid from patients with OC vs. those without OC. Our taxonomic analysis results showed that in addition to both the benign cohort and OC cohort having peritoneal fluid microbiomes dominated by Methylobacterium, Anaerococcus, and Stenotrophomonas, the OC cohort was also dominated by Bacteroides, Finegoldia, Lactobacillus and Peptoniphilus; and the benign cohort by Tumebacillus, Micrococcus and Prevotella (Fig. 1F). While this did not result in significant differences in the α-diversity (Fig. 2I) between these two cohorts, our results showed significant differences in β-diversity in the peritoneal fluid between patients with OC and those without (Fig. 2J, Bray–Curtis, p = 0.007). These reflected the enrichment of Methylorubrum extorquens in the OC cohort (Fig. 2, Supplemental Tables 1–3). Following these analyses, we also characterized the peritoneal fluid microbiomes associated with the stage, grade, and histology of OC compared to the benign conditions at the time of hysterectomy (Table 2). We observed significant differences in both α-diversity and β-diversity in malignant ascites versus peritoneal fluid from patients with benign conditions. These include significant differences in α-diversity between samples from patients with benign conditions vs. advanced-stage OC patients (Fig. 3I; Bray–Curtis, p = 0.014) with enriched M. extorquens in the OC patients (Supplemental Tables 1–2, and 5). Our results also showed significant differences in β-diversity between the patients with benign conditions and high-grade OC patients (Fig. 4H; Bray–Curtis, p = 0.016). We also observed significant differences in the in β-diversity between the patients with benign conditions vs. serous OC patients (Fig. 4; Bray–Curtis, p = 0.019), with enriched M. extorquens in the OC patients (Supplemental Tables 1–2, and 10). Finally, we also explored the prognostic potential of the peritoneal fluid in treatment response (Fig. 6F; Supplemental Tables 1–2, and 13–14). Our results showed significant differences in the β-diversity of patients with sensitive tumor response compared to others (Fig. 6F; unweighted UniFrac, p = 0.022), with enriched Anaerococcus tetradius in patients who did not experience sensitive tumor response (Fig. 6M, Supplemental Tables 1–2 and 13). We also showed significant differences in the β-diversity of the patients who were alive without adverse events and those who were deceased two years post-diagnosis (Fig. 6F; Bray–Curtis, p = 0.029). A few microbial taxa including A. tetradius, Peptoniphilus harei, Methylobacterium radiotolerans, and Lactobacullus gasseri were also found enriched in patients who were alive with adverse events compared those who were alive with no adverse events two years post-diagnosis (Fig. 6M, Supplemental Tables 1–2, and 14).