Vaginal Microbiome Composition, Diversity and Dysbiosis: The ORCHiD Study

Vaginal Microbiome Composition, Diversity and Dysbiosis: The ORCHiD Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Vaginal Microbiome Composition, Diversity and Dysbiosis: The ORCHiD Study April Deveaux, Oyomoare Osazuwa-Peters, Yeon Ji Kim, Pixu Shi, and 12 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7428423/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Background Vaginal dysbiosis may contribute to ovarian cancer (OC) outcomes, but comprehensive microbiome characterization remains limited. Here, we characterize the prevalence and predictors of vaginal microbiome dysbiosis among diverse OC patients in the US. Methods We performed 16S rRNA gene sequencing on vaginal samples from 132 OC patients recruited as part of the population-based ORCHiD study. We applied topic modeling using Latent Dirichlet Allocation (LDA), a computational approach for identifying latent microbial community patterns, to identify distinct microbial signatures representing co-occurring bacterial taxa. Results Widespread dysbiosis was observed with Lactobacillus detected in only 47.7% of patients. Topic modeling identified seven distinct microbial signatures from Lactobacillus -dominated to pathogenic communities. Patients ≥ 50 years showed significant anaerobic bacterial enrichment (log₂FC = 1.31, FDR q < 0.001). Striking racial disparities emerged: Non Hispanic-Black patients had 5-fold higher Actinomycetaceae prevalence (40.9% vs 8.2%, FDR q = 0.005), while protective L. crispatus was detected exclusively in Non Hispanic-White patients (6.4% vs 0%). Conclusions This study revealed widespread vaginal microbiome dysbiosis among OC patients with clinically significant age and racial patterns that may contribute to outcome disparities. Biological sciences/Microbiology/Clinical microbiology Health sciences/Oncology/Cancer/Gynaecological cancer/Ovarian cancer microbiome ovarian cancer ORCHiD 16S rRNA sequencing topic modeling cancer disparities Vaginal microbiome topic modeling health disparities Lactobacillus Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction The vaginal microbiome, the community of microbes in the lower female genital tract, has emerged as an important factor in gynecologic health and gynecological cancers and may influence disease prognosis ( 1 , 2 ). A healthy microbiome is typically dominated by Lactobacillus species, which help maintain a low pH and protect against pathogens( 3 , 4 ). When the protective Lactobacillus -dominated balance is disrupted—a state known as vaginal dysbiosis – the microbiome shifts toward communities dominated by anaerobic bacteria( 4 ). This dysbiotic state has been linked to adverse vaginal health outcomes such as bacterial vaginosis, pelvic inflammatory disease and chronic genital inflammation( 2 , 5 , 6 ), and can contribute to immune dysregulation, epithelial dysfunction, and pro-inflammatory signaling - important mechanisms in tumor progression( 5 , 7 ). Ovarian cancer (OC) remains one of the most lethal gynecological malignancies in the United States, with a five-year survival rate of only 51%( 8 ), however few studies have characterized the role of the vaginal microbiome in driving OC prognosis. Recent studies suggest that OC is associated with a shift in the vaginal microbiome toward a dysbiotic state( 3 , 5 , 9 ). In a case-control study with ovarian cancer patients, Morikawa et al. found that the cervicovaginal microbiota of OC patients was characterized by significant loss of Lactobacillus -dominated communities and increased microbial diversity across multiple histologic subtypes compared to healthy controls( 10 ). Sipos et al. proposed the concept of an “oncobiome”, describing how pathogen-driven disruptions such as infections with Chlamydia trachomatis or Neisseria gonorrhoeae may lead to lasting alterations in the vaginal microbiome that promote chronic inflammation and carcinogenesis( 11 ). Building on this, Mehra et al. reviewed evidence of altered vaginal and gut microbial ecosystems in OC, reporting that Lactobacillus was present in only 24% of women with OC compared to 47% of healthy controls( 12 ). They also identified dysbiosis-associated microbial shifts as contributors to immune dysregulation and tumor progression, suggesting the potential for microbiome modulation as a diagnostic or therapeutic strategy( 12 ). Together, these findings describe a broader biological framework in which vaginal dysbiosis is both a marker and a possible mediator of OC pathophysiology. The vaginal microbiome varies by patient demographics and clinical factors ( 13 , 14 ). For instance, aging is associated with declining estrogen levels and shifts in immune function that reduce Lactobacillus abundance while increasing anaerobic bacterial diversity( 15 ). The vaginal microbiome composition also varies by race/ethnicity ( 16 ), with documented differences in microbial diversity, bacterial species composition, and Lactobacillus distribution patterns( 17 – 20 ). Such compositional differences might contribute to disparities in OC outcomes, including tumor aggressiveness and progression( 5 , 21 , 22 ). Notably, Black women experience significantly worse OC outcomes compared to White women, including later-stage diagnosis, more aggressive tumor biology, and reduced survival( 23 , 24 ). While socioeconomic factors and healthcare access contribute to these disparities, emerging evidence suggests that biological factors, including microbiome differences, may also play a role( 19 , 25 ). A modest number of prior studies have investigated the role of the vaginal microbiome in OC outcomes ( 3 , 26 – 28 ), however these investigations have largely evaluated alpha and beta diversity, differential abundance testing, and taxonomic profiling based on relative abundances( 29 , 30 ). These traditional analytical approaches, while informative, are unable to identify complex community-level patterns that emerge from co-occurring bacterial taxa( 30 , 31 ), insights that might reveal novel mechanistic pathways contributing to OC outcomes and disparities. In this analysis, we aimed to characterize the vaginal microbiome composition, diversity and dysbiosis among patients in the Ovarian Cancer Epidemiology, Healthcare Access and Disparities Study (ORCHiD). Specifically, we sought to: ( 1 ) describe the study population characteristics and overall microbiome diversity patterns in OC patients; ( 2 ) characterize vaginal microbiome composition using traditional differential abundance analysis across patient demographics and disease characteristics; ( 3 ) identify distinct vaginal microbial community signatures using topic modeling and examine their associations with clinical variables; and ( 4 ) validate findings through targeted species-level analysis of key Lactobacillus species. Our findings may contribute to understanding microbiome-mediated health disparities and inform targeted therapeutic interventions aimed at improving OC outcomes across demographic groups. Methods Study Design and Population An overview of the study design and framework is presented in Fig. 1 . This study included 135 ovarian cancer (OC) patients recruited as part of the ORCHiD study. Detailed methods are published elsewhere( 32 ). Briefly, this population-based study recruited patients with confirmed first primary OC between March 2021 and October 2024 from seven state cancer registries, including New York, Kentucky, California, North Carolina, Georgia, and Texas. The study inclusion criteria were patients aged 18 years and older with a pathologically confirmed diagnosis of OC (stages I-IV) at 9–12 months prior to recruitment. All patients in this cohort self-identified their race/ethnicity as non-Hispanic Black (NH-Black) or non-Hispanic White (NH-White). The study exclusion criteria were patients older than 79 years old, or patients with cognitive impairments preventing survey completion. The ORCHiD survey comprehensively captures data on patient demographics, socio-economic status, healthcare access, as well as self-reported treatment, lifestyle, and health history data. Eligible patients were contacted via phone to confirm eligibility, conduct informed consent, and determine participation interest. Patients who consent to participate complete the ORCHiD survey over the phone, online, or by mail and receive an incentive of $ 25 for survey completion. Vaginal Microbiome Sample Collection and Processing Patients who completed the ORCHiD survey were offered an opportunity to participate in the ORCHiD biospecimen sub study which required self-collection of both a vaginal swab and saliva. Once a patient consented to the sub-study, the study team mailed a biospecimen packet containing a vaginal microbiome collection kit (OMR-130, DNA Genotek, Ottawa, Canada) along with manufacturer instructions and a postage-paid sample return envelope. Each biospecimen packet also included a brief 4-question behavioral survey querying medications or microbiome-altering behaviors in the previous 30 days, including oral or IV antibiotic use or use of douching or suppository products. Patients responding yes to any of the microbiome survey questions were excluded from the analysis cohort. Patients were also instructed not to collect samples when actively menstruating or experiencing other vaginal bleeding. Upon receipt of returned kits, samples were inspected for any damage or leakage and logged electronically. The collection kit included a swab and a vial containing microbial media buffer, a proprietary stabilizing liquid to preserve samples at room temperature for up to 30 days. Any damaged kits were disposed of according to established safety regulations. Samples were processed by treating with 5 µl of Proteinase K (80 mg/ml) and incubating for one hour in a 50°C water bath, then aliquoted into 1.5 ml microcentrifuge tubes and stored at -80°C for downstream processing. Of the 764 patients recruited into the parent ORCHiD study, 435 (56.9%) consented to the biospecimen sub-study, and 230 (52.9% of consenters, 30.1% of total surveyed) returned at least one sample. Patients who completed and returned the biospecimen kits received an additional $ 20 incentive. Processed vaginal samples were sent to the Duke Microbiome Core Facility for assay. First, microbial DNA from vaginal swab samples preserved in OMNIgene vaginal kit was extracted using a DNeasy 96 PowerSoil Pro QIACube HT kit (QIAGEN, #47021) on an automated machine (QIACube HT, QIAGEN). The manufacturer’s instructions were followed with these minor deviations: 250 µl of the preserved sample was used as starting material, and during the final elution step, the elution buffer (C6) was incubated at room temperature for 10 minutes to increase DNA yield. Bacterial community composition from vaginal swab samples was characterized by amplification of the 16S rRNA V1-V3 hypervariable regions via polymerase chain reaction (PCR). Forward primer 27F (5’-GAGTTTGATCGTGGCTCAG-3’) and reverse primer 518R (5’-ATTACCGCGGCTGCTGG-3’) were used with Phusion Plus PCR Master Mix (ThermoScientific, #F631L) for a total of 25 PCR cycles. Samples were multiplexed via the reverse primers (518R) that carry unique Earth Microbiome Project ( http://www.earthmicrobiome.org/ ) barcodes ( 33 ). PCR products were purified with AMPure XP Beads (Beckman Coulter, #A63881) and quantified with a Qubit dsDNA HS assay kit (ThermoFisher, #Q32854) on a Promega GloMax plate reader. Equimolar PCR products were pooled from each sample and submitted to the Duke Sequencing and Genomic Technologies shared resource. The final pool was sequenced on an Illumina MiSeq platform using v3 chemistry (300 base pairs, paired-end). Study Variables Cohort characteristics assessed included self-reported race, age at diagnosis, OC stage, and histologic subtype. Race was self-reported and categorized as non-Hispanic Black or non-Hispanic White. Age was analyzed as a continuous variable for primary analyses and then dichotomized into two age categories (< 50 years and ≥50 years) for interpretability of topic distributions. Clinical data, such as stage at diagnosis and histological subtype, were obtained from state cancer registries. OC stage was derived from summary staging data and categorized into early-stage (summary stage 1 or 2), or late-stage (summary stage greater than 2). Histologic subtype was categorized as either type II epithelial (including high-grade serous and other aggressive subtypes) or Other (including type I epithelial, endometrioid, mucinous, clear cell subtypes, and non-epithelial tumors). Statistical Analysis Data Processing Microbiome sequencing data was processed using R programming suite (v4.3.0) ( 34 ). Sequencing reads from vaginal samples were processed using the DADA2 pipeline (v1.28.0) ( 35 ) to infer amplicon sequence variants (ASVs) with default parameters. Deviations from default parameters include a parameter for FilterAndTrim function (truncLen = c(290, 260)) to remove sequencing regions with low quality scores and only including ASVs that were 433 to 525 base pairs long in analysis. Taxonomy was assigned using the Silva v138.1 database ( 36 , 37 ) and non-bacterial taxa were removed. Samples with less than 1,000 reads were removed, which included 4 negative controls and 3 samples with insufficient sample size, and a phylogenetic tree was constructed using Randomized Axelerated Maximum Likelihood (RAxML) to account for phylogenetic relationships ( 38 ). Summary Statistics : For taxonomic composition analysis, we calculated three key metrics for each bacterial genus: ( 1 ) prevalence, defined as the percentage of samples in which the genus was detected above the minimum threshold; ( 2 ) mean relative abundance, calculated as the average percentage of total reads attributed to each genus across all samples; and ( 3 ) median relative abundance with interquartile range (IQR), providing robust measures of central tendency and spread for non-normally distributed abundance data. Topic Modeling Analysis : We employed Latent Dirichlet Allocation (LDA) to identify distinct vaginal microbial community signatures, termed "microbial topics," representing co-occurring bacterial taxa( 39 ). The range of 3 to 10 topics was selected based on established vaginal microbiome community classifications, as the foundational community state type (CST) framework identifies 5 CSTs with up to 9 CSTs with extended classification( 16 , 40 ). This biologically informed range ensures alignment with known vaginal microbiome structure while allowing for potential novel community patterns in this cancer population. Optimal topic number was determined using the FindTopicsNumber function from the "ldatuning" R package (v1.0.2)( 41 ) with multiple metrics including CaoJuan2009, Arun2010, Griffiths2004, and Deveaud2014 ( 42 – 45 ) evaluated across the topic number range. Two metrics (CaoJuan2009 and Arun2010) identify optimal models by minimization, while the other two metrics (Griffiths2004 and Deveaud2014) identify optimal models by maximization. The optimal number of topics was selected using an elbow method approach ( Supplemental Fig. 1 ). LDA was implemented using the "topicmodels" package (v0.2.17) in R( 46 ) with Gibbs sampling method, treating samples as "documents" and genera as "words." The model was run with a fixed seed (243) for reproducibility. Beta probability matrices (genus-topic associations) and gamma probability matrices (sample-topic associations) were extracted using the "tidytext" package (v0.4.3) ( 47 ). Topics were characterized by their dominant genera (highest beta probabilities) and named based on their microbial composition and established associations with vaginal health states. Differential Abundance Analysis of Topics : Topic-level differential abundance testing was performed using the LinDA (linear models for differential abundance analysis) framework( 48 ), which accounts for compositional data structure and provides robust statistical inference. Topic counts were calculated by multiplying gamma probabilities by corresponding read counts for each sample, creating a topic-level phyloseq object. LinDA analysis was conducted with the formula: ~ Race + Stage + Histology + Age, testing each clinical variable while controlling for others. The analysis used empirical Bayes shrinkage, centered log-ratio transformation, and Benjamini-Hochberg false discovery rate (FDR) correction for multiple testing (FDR q < 0.05). Species-Level Lactobacillus Validation : To validate topic-level findings, we performed targeted species-level analyses focusing on key Lactobacillus species known to be important for vaginal health. The phyloseq object was filtered to include only Lactobacillus species, and relative abundances were calculated by dividing species counts by total reads per sample. We examined both detection rates (presence/absence) and relative abundances when present for L. crispatus , L. iners , L. gasseri , and L. jensenii across demographic and clinical variables. Additional Analyses : Detection rates between groups were compared using Fisher's exact tests, while relative abundances (when detected) were compared using Wilcoxon rank-sum tests due to non-normal distributions. For detection rate analyses, effect sizes were calculated as odds ratios with 95% confidence intervals. For abundance comparisons, effect sizes were reported as fold-changes between group medians. FDR correction was applied across all species tested within each comparison type (detection vs abundance). Statistical significance was set at FDR q < 0.05, with trends reported for FDR q < 0.10. Alpha diversity (Shannon diversity index and Faith's phylogenetic distance) was calculated using phyloseq (v1.44.0) and picante (v1.8.2) packages, while beta diversity was assessed using Jaccard, Bray-Curtis, UniFrac, and weighted UniFrac dissimilarity metrics calculated with the vegan package (v2.6.4), with principal coordinate analysis (PCoA) performed for visualization. Alpha diversity metrics were tested for normality using Shapiro-Wilk tests, and differences by demographic and clinical variables were assessed using appropriate parametric (ANOVA) or non-parametric (Kruskal-Wallis) tests, while beta diversity differences were tested using permutational analysis of variance (PERMANOVA) with FDR correction applied across multiple comparisons. Data Visualization and Statistical Software : All statistical analyses were performed in R (v4.3.0)( 49 ) using the following packages: phyloseq (v1.48.0) for microbiome data handling( 50 ), topicmodels (v0.2.17 ) for LDA implementation( 46 ), tidytext (v0.4.3) for topic model interpretation( 47 ), LinDA (v0.2.0) for differential abundance testing( 48 ), ggplot2 (v3.5.2)for visualization( 51 ), and cowplot (v1.2.0)for multi-panel figures(52). Statistical significance was set at FDR-adjusted q < 0.05 unless otherwise specified. Results Cohort Characteristics A total of 132 OC patients were included in the analysis (Table 1 ); 83.7% NH-White (n = 111) and 16.3% NH-Black (n = 22). The median age was 61.5 years (range 30–79), with 83.0% of patients 50 years or older at diagnosis. Compared to NH-White patients, NH-Black patients were more likely to report lower annual household income (22.7% vs. 7.1% reporting < $ 20,000), unemployment (13.6% vs. 3.5%), or disability (27.3% vs. 2.7%). About 72.7% of NH-Black patients widowed, divorced, separated, or never married, compared to 31.9% of NH-White patients. Most patients were diagnosed with late-stage disease, defined as FIGO stage III–IV (67.4%), although late-stage diagnosis was higher in NH-Black vs NH-White patients (77.3% vs. 65.5%). Type II epithelial was the most common histologic subtype (60.0%). Most patients received surgery (94.1%) and chemotherapy (75.6%), with no notable differences in treatment by race. Microbiome Diversity We evaluated vaginal microbiome diversity using both alpha (Shannon index) and beta (weighted UniFrac distance) diversity metrics ( Supplemental Fig. 1 ). Alpha diversity, assessed via the Shannon index which incorporates both richness and evenness, did not differ significantly by race (median: 2.37 NH-Black vs. 2.52 NH-White; p = 0.72; Figure S1 A ), OC stage (median: 2.54 early vs. 2.42 late; p = 0.49; Figure S1 C ), or histologic subtype (median: 2.39 Type I vs. 2.48 Type II; p = 0.78; Figure S1 E ). These findings suggest that within-sample microbial richness and evenness were relatively stable across demographic and clinical strata. Beta diversity, quantified using weighted UniFrac distances which incorporates relative abundance and phylogenetic information, showed no significant differences in overall microbial community composition by race (PERMANOVA p = 0.64; Figure S1 B ) or stage (PERMANOVA p = 0.76; Figure S1 D ). However, significant differences were observed by histologic subtype (PERMANOVA p = 0.025; Figure S1 F ), indicating that the structure of the vaginal microbiome varied between Type I and Type II epithelial OC. Vaginal Microbiome Composition and Relative Abundance Vaginal microbiome overall composition analysis (Fig. 2 , Supplemental File ) showed Lactobacillus as the most common genus by mean relative abundance (mean: 23.5%, median: 0% [IQR: 0-44.5%], prevalence: 47.7%), followed by Prevotella (mean: 12.3%, median: 4.9% [IQR: 0.6–23.6%], prevalence: 93.9%) and other anaerobic genera including Peptoniphilus (mean: 6.0%, prevalence: 95.5%), Anaerococcus (mean: 5.6%, prevalence: 90.9%), and Finegoldia (mean: 5.1%, prevalence: 96.2%). The difference between mean and median abundance for Lactobacillus (23.5% vs. 0%) indicated a bimodal distribution, with this genus being either absent/minimal or highly dominant across individuals. This pattern reflects a generally dysbiotic microbiome across the cohort, with Lactobacillus species present in less than half of patients and anaerobic bacteria widely prevalent (> 90% prevalence for most genera). Age-Related Differences Older patients ≥ 50 years had significantly higher prevalence of Actinotignum compared to younger patients (< 50 years: prevalence 31.8%; ≥50 years: prevalence 68.2%, FDR q = 0.007), though mean relative abundances remained low in both groups (< 50: mean 2.0%, median 0%; ≥50: mean 2.7%, median 0%). This age-related pattern suggests increased colonization by this genus in older OC patients. There were no other significant or trending associations by age group. Cancer Stage Differences Early stage patients had significantly higher prevalence of Cutibacterium compared to late-stage patients (prevalence: 61.0% vs 33.0%, FDR q = 0.048). There were no other significant or trending associations by disease stage. Racial Differences NH-Black patients had a 5-fold higher prevalence of Actinomycetaceae family bacteria compared to NH-White patients (40.9% vs 8.2%, FDR q = 0.005), representing the most robust compositional difference observed in the cohort. NH-Black patients also had higher Lactobacillus prevalence (68.2% vs 43.6%) and mean abundance (mean 34.4% vs 21.4%), though this did not reach statistical significance (FDR q = 0.271). Notably, while overall Lactobacillus abundance appeared higher in NH-Black patients, this pattern was driven primarily by different species composition (detailed in species-level analysis below). There were no other significant or trending racial differences in microbial composition. Histologic Subtype Differences There were no significant or trending associations between microbial composition and histologic subtype (all FDR q > 0.25). Topic Modeling Results Identification of Vaginal Microbial Topics LDA identified seven microbial topics using an elbow method approach, representing distinct co-occurring genera patterns that were biologically interpretable without excessive model complexity( 53 ) ( Fig. 3 , Supplemental Fig. 2). Topic 1 ( Streptococcus Mixed) was characterized by moderate dysbiosis with Streptococcus (48.1%), Atopobium (26.9%), and Sneathia (7.2%), as Atopobium and Sneathia are established bacterial vaginosis-associated taxa( 16 , 19 ). Topic 2 ( Proteus Dominated) represented severe pathogenic dysbiosis, being highly dominated by Proteus (87.9%) with minimal diversity, as Proteus species are associated with urogenital infections and pathogenic overgrowth( 54 ). Topic 3 (Diverse Community) showed intermediate dysbiosis led by Corynebacterium (27.8%), Escherichia-Shigella (27.5%), and Finegoldia (21.2%), representing the diverse, non- Lactobacillus communities characteristic of dysbiotic states( 16 ). Topic 4 (Anaerobic Dysbiosis) was dominated by Peptoniphilus (20.7%) and Anaerococcus (19.3%) with Campylobacter (9.5%), representing anaerobic dysbiosis as these taxa are associated with bacterial vaginosis and inflammatory conditions( 55 ). Topic 5 ( Lactobacillus -Dominated) was nearly exclusively composed of Lactobacillus (98.6%), which can vary significantly in protective capacity depending on the specific species composition( 16 ). Topic 6 ( Pseudomonas -dominated) was led by Pseudomonas (51.2%) with diverse anaerobes that can represent opportunistic bacterial overgrowth in severely dysbiotic vaginal environments( 56 ). Topic 7 (BV-Type) was dominated by Prevotella (70.0%), representing classic bacterial vaginosis-type dysbiosis as Prevotella is a hallmark of BV-associated microbiomes( 16 , 19 ). Topic-Clinical Variable Associations LinDA differential abundance analysis revealed significant associations between microbial topics and patient demographics (Fig. 4 ). Topic 4 (Anaerobic Dysbiosis) showed the strongest and only statistically significant association, with significantly higher abundance in patients ≥ 50 years compared to younger patients (log₂ fold-change = 1.31, FDR q < 0.001). This was the only topic-clinical variable association reaching statistical significance after multiple testing correction. Disease stage showed limited associations with topic abundance after multiple testing correction. Topic 6 (Environmental) showed enrichment in late-stage compared to early-stage disease (log₂ fold-change = -1.9, FDR q = 0.06), though this did not reach statistical significance. While no racial associations reached statistical significance after FDR correction, consistent directional patterns emerged across topics. Topic 5 ( Lactobacillus -dominated) showed the most substantial depletion in NH-White compared to NH-Black patients (log₂ fold-change = -3.0, FDR q = 0.09), and Topic 2 ( Proteus -Dominated) demonstrated depletion in NH-White versus NH-Black patients (log₂ fold-change = -1.4, FDR q = 0.09). No significant associations were observed between topics and histologic subtypes after FDR correction, suggesting that overall microbial community patterns may be more influenced by demographic factors than tumor characteristics. Species-Level Analysis of Key Lactobacillus Species To validate the topic modeling findings, particularly regarding the healthy Lactobacillus signature (Topic 5), we performed detailed species-level analyses of the four predominant vaginal Lactobacillus species ( Supplemental File ) with key findings for L. crispatus and L. iners highlighted in Fig. 5 . Given the bimodal distribution of Lactobacillus observed at the genus level, we examined both detection rates (presence/absence) and relative abundance when present for key Lactobacillus species. Detection Rates and Clinical Associations Age-related patterns revealed differential effects on Lactobacillus species colonization. L. crispatus detection was 3.75-fold higher in patients < 50 years compared to those ≥ 50 years (13.6% vs 3.6%, FDR p = 0.125), while L. iners showed a similar but less pronounced age-related decline (18.2% vs 10.9%, FDR p = 0.454). L. crispatus was detected in 12.2% of early-stage patients compared to only 2.2% of late-stage patients, representing a 5.55-fold difference (FDR p = 0.067). In contrast, L. iners showed the opposite pattern, with higher detection in late-stage patients (14.3%) compared to early-stage patients (7.3%, FDR p = 0.389). Racial differences revealed striking opposing patterns in Lactobacillus species distribution. L. crispatus was detected exclusively in NH-White patients (6.4% vs 0% in NH-Black patients, FDR p = 0.458), while L. iners showed 2.27-fold higher prevalence in NH-Black patients (22.7% vs 10.0%, FDR p = 0.295). Abundance When Present Among samples where Lactobacillus species were detected, abundance levels showed distinct patterns across clinical groups. When present, L. crispatus maintained consistently high relative abundance regardless of clinical characteristics. In early-stage patients, L. crispatus achieved a median relative abundance of approximately 95% when detected, with significantly higher abundance compared to late-stage patients (FDR p = 0.033). Age-related differences in L. crispatus abundance were minimal when the species was present (FDR p = 0.125). In contrast, L. iners demonstrated more variable abundance patterns when present. Abundance levels showed less consistent patterns across clinical groups, with median values ranging from 88–92% depending on disease stage and patient characteristics (all FDR p > 0.05). This variability may reflect the intermediate protective capacity of L. iners compared to the more robustly protective L. crispatus . Discussion This comprehensive analysis of 132 OC patients identified consistent vaginal microbiome dysbiosis characterized by Lactobacillus depletion and anaerobic bacterial overgrowth. We observed two clinically significant findings: age-related increases in anaerobic dysbiosis, with patients ≥ 50 years showing significant enrichment of Peptoniphilus and Anaerococcus (log₂ fold-change = 1.31, FDR q < 0.001), and striking racial disparities including a 5-fold higher prevalence of Actinomycetaceae family bacteria in NH-Black compared to NH-White patients (40.9% vs 8.2%, FDR q = 0.005). Species-specific Lactobacillus analyses revealed L. crispatus detected exclusively in NH-White patients and L. iners over 2-fold higher in NH-Black patients, alongside progressive loss of protective L. crispatus with advancing age and disease stage. These patterns suggest that the vaginal microbiome may play an important role in driving OC prognosis and health disparities, warranting further investigation for microbiome-based interventions. These findings align with and extend established literature on vaginal microbiome alterations in gynecological malignancies. Previous OC studies by Jacobson et al. (2021) and Asangba et al. (2023) reported Lactobacillus depletion and anaerobic overgrowth, with only 24% of OC patients showing Lactobacillus -dominated microbiomes and significantly reduced abundance compared to controls (approximately 15% versus 30%)( 27 , 57 ). Our cohort showed similar patterns with 47.7% Lactobacillus detection and 23.5% mean relative abundance, falling substantially below healthy reproductive-age women rates (70–80% dominance)( 16 ). However, our analysis also revealed age-specific patterns of dysbiosis, highlighting that the vaginal microbiome of older OC patients was significantly enriched with Peptoniphilus and Anaerococcus (log₂ fold-change = 1.31, FDR p < 0.001), a notable finding that has not been previously reported to our knowledge. The racial differences we observed also extends prior work by Ravel et al., which established that healthy Black women are more likely to have non- Lactobacillus -dominated communities, with only 61.9% exhibiting Lactobacillus -dominated communities compared to 89.7% of White women( 16 ). However, no known studies have specifically examined racial differences in vaginal microbiome composition among OC patients. We observed that Topic 5 ( Lactobacillus -dominated) showed substantial depletion in NH-White compared to NH-Black OC patients (log₂ fold-change = -3.0, FDR q = 0.09), suggesting higher overall Lactobacillus prevalence in NH-Black OC patients that is worth further exploration as a potential OC-related signature. However, species-level validation revealed that this pattern is driven mainly by the predominance of dysbiotic L. iners , which was over 2-fold higher in NH-Black vs. NH-White patients (22.7% vs 10.0%), rather than the protective L. crispatus that was detected exclusively in White patients (6.4% vs 0%). Additionally, our study is the first to document that NH-Black OC patients have a 5-fold higher Actinomycetaceae family prevalence vs. White patients, findings that can be further explored in larger cohort sizes. The age-related enrichment of Peptoniphilus -dominated signatures in our study reveals potentially important biological pathways linking aging, microbiome dysbiosis, and cancer outcomes. OC patients ≥ 50 years showed significant enrichment of Peptoniphilus and Anaerococcus , bacteria that are associated with bacterial vaginosis and chronic wound infections, where they act synergistically with other microbes to impair normal healing processes and contribute to chronic inflammatory states( 56 ). The loss of L. crispatus , which maintains vaginal health through lactic acid production( 16 ), with advancing age (3.75-fold higher in younger patients) and disease stage (5.55-fold higher in early-stage disease) suggests that protective microbiome characteristics are associated with dysbiotic shifts as both aging and cancer progress, supporting the oncobiome concept where pathogen-driven disruptions create lasting alterations promoting chronic inflammation and carcinogenesis( 11 ). The absence of L. crispatus in NH-Black patients combined with higher L. iners prevalence represents a fundamental disparity in protective microbial capacity, as L. crispatus produces predominantly D-lactic acid providing superior pathogen protection compared to L-lactate from L. iners ( 58 , 59 ). L. iners dominance is associated with intermediate vaginal health states and increased susceptibility to dysbiotic transitions( 60 ), and recent mechanistic studies show that L. iners actively promotes treatment resistance through metabolic reprogramming( 61 ), potentially explaining some of the documented treatment response disparities between racial groups. The combination of higher Actinomycetaceae prevalence and differential Lactobacillus species composition may also create a dual biological disadvantage for NH-Black patients that may contribute to worse OC prognosis. Actinomycetaceae have been associated with pelvic inflammatory disease and chronic genital inflammation( 62 ), and recent work demonstrates that certain Actinobacteria species can metabolize estrogen and steroid hormones( 63 ), potentially altering the local hormonal microenvironment in ways that could influence cancer biology. Taken together, our findings highlight a potentially important role for vaginal microbiome patterns in driving OC prognosis, an area that deserves further study. Importantly, given that Black women experience later-stage diagnosis( 64 ), more aggressive tumor biology( 65 ), and reduced survival rates( 14 ), differential microbiome patterns may be a potential biological mechanism that drives these disparities. Efforts to more comprehensively characterize the microbiome and identify targetable pathways may provide valuable intervention opportunities. While there is currently insufficient evidence to support routine probiotic recommendations for OC patients( 66 ), additional studies building on our findings can provide important insights to inform species-specific rather than generic interventions to alter the microbiome. Limitations Several important limitations should be considered. The cross-sectional design prevents determination of causality or temporal relationships between microbiome changes and disease characteristics. We cannot distinguish whether observed patterns reflect cancer-induced changes, treatment effects, aging-related changes independent of cancer, or pre-existing differences influencing cancer outcomes, although the majority of patients in this cohort had received treatment. The sample size was relatively small for detecting subtle associations, particularly for NH-Black patients (n = 22), limiting statistical power for subgroup analyses. Our 16S rRNA sequencing approach provides taxonomic composition but limited functional insight into microbiome activity. Methodological considerations such as the exclusion of patients with recent antibiotic use, and focusing on patients at least nine months post-diagnosis, potentially resulted in selecting for more stable microbiome profiles, and missing acute changes associated with initial diagnosis and treatment. The inclusion of patients across different treatment stages likely introduced heterogeneity that could have masked some associations. Nevertheless, our study has several notable strengths that support the validity of our findings. The population-based recruitment from multiple state cancer registries enhances generalizability, while the multi-scale analytical approach combining traditional composition analysis, topic modeling, and species-level validation provided convergent evidence for our key findings. The self-collection methodology improved feasibility and reduced selection bias, and our rigorous exclusion criteria for antibiotic and suppository use minimized confounding factors. Conclusions We identified consistent vaginal microbiome dysbiosis among OC patients with notable age- and race-associated patterns that may have important implications for prognosis. These findings establish a strong foundation for future research investigating specific vaginal microbiome patterns beyond the species level in understanding and addressing OC progression and outcome disparities. If validated in larger longitudinal studies, vaginal microbiome profiles could serve as biomarkers for disease monitoring or therapeutic targets to improve outcomes, particularly for older patients and NH-Black patients who demonstrate the most concerning microbiome signatures. The integration of microbiome assessment into clinical care protocols may ultimately contribute to more personalized and effective OC management strategies, though species-specific interventions will be essential for addressing the biological disadvantages revealed by our analysis. Declarations Funding This research was funded by the National Institutes of Health/National Cancer Institute (Grant Numbers R37CA233777 and K01CA255408) Institutional Review Board Statement The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board (or Ethics Committee) of Duke University (Pro00101872). Informed Consent Statement This study was approved by Duke University and the participating registries’ Institutional Review Boards (Pro00101872). All participants included provided informed consent. Acknowledgments The authors thank all the study participants who participated in the ORCHiD study for their vital contribution in advancing the science of cancer in the United States. Texas Cancer Registry (TCR), Texas Department of State Health Services; Maryland Cancer Registry (MCR), Maryland Department of Health; California Cancer Registry (CCR), California Department of Public Health; New York State Cancer Registry (NYSCR), New York State Department of Health; Kentucky Cancer Registry (KCR); Emory University Rollins School of Public Health: Department of Epidemiology, North Carolina Cancer Registry (NCCR). The findings and conclusions in this publication are those of the author(s) and do not necessarily represent the views of the above-mentioned cancer registries. The authors also thank Duke School of Medicine for use of Duke Microbiome Core Facility and Sequencing and Genomic Technologies shared resource, which provided DNA extraction, library preparation, and sequencing services. Disclosures Kentucky cancer data have been provided by the Kentucky Cancer Registry, 2365 Harrodsburg Road, Lexington, KY 40504 (www.kcr.uky.edu). Data from the Kentucky Cancer Registry is supported by the following: Cooperative Agreement #NU58DP007144 from the Centers for Disease Control and Prevention and Contract #HHSN261201800013I from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program. The ideas and opinions expressed herein are those of the author(s) and do not necessarily reflect the opinions of the National Cancer Institute, the Centers for Disease Control and Prevention or their Contractors and Subcontractors, and the Commonwealth of Kentucky. The collection of cancer incidence data in Georgia was supported by contract HHSN261201800003I, Task Order HHSN26100001 from the NCI and cooperative agreement 6NU58DP006352-05-01 from the CDC. This work was supported in part by the Centers for Disease Control and Prevention’s National Program of Cancer Registries through cooperative agreement NU58DP007218 awarded to the New York State Department of Health and by Contract HHSN261201800005I (Task Order HHSN26100001) from the National Cancer Institute, National Institutes of Health. Cancer data was used by the Maryland Cancer Registry, Center for Cancer Prevention and Control, Maryland Department of Health, with funding from the State of Maryland and the Maryland Cigarette Restitution Fund to provide assistance in patient recruitment. The collection and availability of cancer registry data is also supported by the Cooperative Agreement NU58DP006333, funded by the Centers for Disease Control and Prevention. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control and Prevention or the Department of Health and Human Services. The collection of cancer incidence data used in this study was supported by the California Department of Public Health pursuant to California Health and Safety Code Section 103885; Centers for Disease Control and Prevention’s (CDC) National Program of Cancer Registries, under cooperative agreement 5NU58DP006344; the National Cancer Institute’s Surveillance, Epidemiology and End Results Program under contract HHSN261201800032I awarded to the University of California, San Francisco, contract HHSN261201800015I awarded to the University of Southern California, and contract HHSN261201800009I awarded to the Public Health Institute. The ideas and opinions expressed herein are those of the author(s) and do not necessarily reflect the opinions of the State of California, Department of Public Health, the National Cancer Institute, and the Centers for Disease Control and Prevention or their Contractors and Subcontractors. Texas cancer data have been provided by the Texas Cancer Registry, Cancer Epidemiology and Surveillance Branch, Texas Department of State Health Services, 1100 West 49th Street, Austin, TX 78756 (www.dshs.texas.gov/tcr). Data from the Texas Cancer Registry is supported by the following: Cooperative Agreement #1NU58DP007140 from the Centers for Disease Control and Prevention, Contract #75N91021D00011 from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program, and the Cancer Prevention and Research Institute of Texas. The findings and conclusions in this publication are those of the author(s) and do not necessarily represent the views of the North Carolina Department of Health and Human Services, Division of Public Health. References Chen X, Lu Y, Chen T, Li R. The Female Vaginal Microbiome in Health and Bacterial Vaginosis. Front Cell Infect Microbiol. 2021;11:631972. Miller EA, Beasley DE, Dunn RR, Archie EA. Lactobacilli Dominance and Vaginal pH: Why Is the Human Vaginal Microbiome Unique? Front Microbiol. 2016;7:1936. Nene NR, Reisel D, Leimbach A, Franchi D, Jones A, Evans I, et al. Association between the cervicovaginal microbiome, BRCA1 mutation status, and risk of ovarian cancer: a case-control study. Lancet Oncol. 2019;20(8):1171-82. Saraf VS, Sheikh SA, Ahmad A, Gillevet PM, Bokhari H, Javed S. Vaginal microbiome: normalcy vs dysbiosis. Arch Microbiol. 2021;203(7):3793-802. Laniewski P, Ilhan ZE, Herbst-Kralovetz MM. The microbiome and gynaecological cancer development, prevention and therapy. Nat Rev Urol. 2020;17(4):232-50. Li W, Ma ZS. Dominance network analysis of the healthy human vaginal microbiome not dominated by Lactobacillus species. Comput Struct Biotechnol J. 2020;18:3447-56. Hakimjavadi H, George SH, Taub M, Dodds LV, Sanchez-Covarrubias AP, Huang M, et al. The vaginal microbiome is associated with endometrial cancer grade and histology. Cancer Res Commun. 2022;2(6):447-55. Siegel RL, Miller KD, Fuchs HE, Jemal A. Cancer statistics, 2022. CA Cancer J Clin. 2022;72(1):7-33. Zhou B, Sun C, Huang J, Xia M, Guo E, Li N, et al. The biodiversity Composition of Microbiome in Ovarian Carcinoma Patients. Sci Rep. 2019;9(1):1691. Morikawa A, Kawabata A, Shirahige K, Akiyama T, Okamoto A, Sutani T. Altered cervicovaginal microbiota in premenopausal ovarian cancer patients. Gene. 2022;811:146083. Sipos A, Ujlaki G, Miko E, Maka E, Szabo J, Uray K, et al. The role of the microbiome in ovarian cancer: mechanistic insights into oncobiosis and to bacterial metabolite signaling. Mol Med. 2021;27(1):33. Mehra Y, Chalif J, Mensah-Bonsu C, Spakowicz D, O'Malley DM, Chambers L. The microbiome and ovarian cancer: insights, implications, and therapeutic opportunities. J Cancer Metas Treat. 2023;9. Chan JK, Kapp DS, Shin JY, Husain A, Teng NN, Berek JS, et al. Influence of the gynecologic oncologist on the survival of ovarian cancer patients. Obstet Gynecol. 2007;109(6):1342-50. Surveillance E, and End Results Program. Cancer Stat Facts: Ovarian Cancer [Available from: https://seer.cancer.gov/statfacts/html/ovary.html. Brotman RM, Shardell MD, Gajer P, Fadrosh D, Chang K, Silver MI, et al. Association between the vaginal microbiota, menopause status, and signs of vulvovaginal atrophy. Menopause. 2014;21(5):450-8. Ravel J, Gajer P, Abdo Z, Schneider GM, Koenig SS, McCulle SL, et al. Vaginal microbiome of reproductive-age women. Proc Natl Acad Sci U S A. 2011;108 Suppl 1(Suppl 1):4680-7. Fettweis JM, Serrano MG, Brooks JP, Edwards DJ, Girerd PH, Parikh HI, et al. The vaginal microbiome and preterm birth. Nat Med. 2019;25(6):1012-21. Sun S, Serrano MG, Fettweis JM, Basta P, Rosen E, Ludwig K, et al. Race, the Vaginal Microbiome, and Spontaneous Preterm Birth. mSystems. 2022;7(3):e0001722. Fettweis JM, Brooks JP, Serrano MG, Sheth NU, Girerd PH, Edwards DJ, et al. Differences in vaginal microbiome in African American women versus women of European ancestry. Microbiology (Reading). 2014;160(Pt 10):2272-82. Zhou X, Brown CJ, Abdo Z, Davis CC, Hansmann MA, Joyce P, et al. Differences in the composition of vaginal microbial communities found in healthy Caucasian and black women. ISME J. 2007;1(2):121-33. Li C, Feng Y, Yang C, Wang D, Zhang D, Luo X, et al. Association between vaginal microbiota and the progression of ovarian cancer. J Med Virol. 2023;95(7):e28898. Xu J, Peng JJ, Yang W, Fu K, Zhang Y. Vaginal microbiomes and ovarian cancer: a review. Am J Cancer Res. 2020;10(3):743-56. Stewart SL, Rim SH, Richards TB. Gynecologic oncologists and ovarian cancer treatment: avenues for improved survival. J Womens Health (Larchmt). 2011;20(9):1257-60. Terplan M, Schluterman N, McNamara EJ, Tracy JK, Temkin SM. Have racial disparities in ovarian cancer increased over time? An analysis of SEER data. Gynecol Oncol. 2012;125(1):19-24. Borgdorff H, van der Veer C, van Houdt R, Alberts CJ, de Vries HJ, Bruisten SM, et al. The association between ethnicity and vaginal microbiota composition in Amsterdam, the Netherlands. PLoS One. 2017;12(7):e0181135. Banerjee S, Tian T, Wei Z, Shih N, Feldman MD, Alwine JC, et al. The ovarian cancer oncobiome. Oncotarget. 2017;8(22):36225-45. Jacobson D, Moore K, Gunderson C, Rowland M, Austin R, Honap TP, et al. Shifts in gut and vaginal microbiomes are associated with cancer recurrence time in women with ovarian cancer. PeerJ. 2021;9:e11574. Yu B, Liu C, Proll SC, Manhardt E, Liang S, Srinivasan S, et al. Identification of fallopian tube microbiota and its association with ovarian cancer. Elife. 2024;12. Schloss PD. Identifying and Overcoming Threats to Reproducibility, Replicability, Robustness, and Generalizability in Microbiome Research. mBio. 2018;9(3). Weiss S, Xu ZZ, Peddada S, Amir A, Bittinger K, Gonzalez A, et al. Normalization and microbial differential abundance strategies depend upon data characteristics. Microbiome. 2017;5(1):27. Nearing JT, Douglas GM, Hayes MG, MacDonald J, Desai DK, Allward N, et al. Microbiome differential abundance methods produce different results across 38 datasets. Nat Commun. 2022;13(1):342. Akinyemiju T, Deveaux A, Wilson L, Gupta A, Joshi A, Bevel M, et al. Ovarian Cancer Epidemiology, Healthcare Access and Disparities (ORCHiD): methodology for a population-based study of black, Hispanic and white patients with ovarian cancer. BMJ Open. 2021;11(10):e052808. Caporaso JG, Ackermann G, Apprill A, Bauer M, Berg-Lyons D, Betley J, et al. EMP 16S Illumina amplicon protocol. See http://www earthmicrobiome org/protocols-and-standards/16s. 2018. Team RC. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2024. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: high-resolution sample inference from Illumina amplicon data. Nature methods. 2016;13(7):581-3. Yilmaz P, Parfrey LW, Yarza P, Gerken J, Pruesse E, Quast C, et al. The SILVA and “All-species Living Tree Project (LTP)” taxonomic frameworks. Nucleic Acids Research. 2014;42(D1):D643-D8. Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic acids research. 2012;41(D1):D590-D6. Stamatakis A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics. 2014;30(9):1312-3. Shrode RL, Ollberding NJ, Mangalam AK. Looking at the Full Picture: Utilizing Topic Modeling to Determine Disease-Associated Microbiome Communities. bioRxiv. 2023. Brooks JP, Buck GA, Chen G, Diao L, Edwards DJ, Fettweis JM, et al. Changes in vaginal community state types reflect major shifts in the microbiome. Microb Ecol Health Dis. 2017;28(1):1303265. Nikita Μ CN. ldatuning: Tuning of the Latent Dirichlet Allocation Models Parameters. 2020. Arun R, Suresh V, Madhavan CEV, Murty MN. On Finding the Natural Number of Topics with Latent Dirichlet Allocation: Some Observations. Lect Notes Artif Int. 2010;6118:391-402. Cao J, Xia T, Li JT, Zhang YD, Tang S. A density-based method for adaptive LDA model selection. Neurocomputing. 2009;72(7-9):1775-81. Griffiths TL, Steyvers M. Finding scientific topics. P Natl Acad Sci USA. 2004;101:5228-35. Deveaud R, SanJuan E, Bellot P. Accurate and effective Latent Concept Modeling for ad hoc information retrieval. Document Numerique. 2014;17(1):61-84. Grün B, Hornik K. Topicmodels: An R Package for Fitting Topic Models. J Stat Softw. 2011;40(13):1-30. Robinson S, Silge J. tidytext: Text Mining and Analysis Using Tidy Data Principles in R. JOSS. 2016;1(3). Zhou H, He K, Chen J, Zhang X. LinDA: linear models for differential abundance analysis of microbiome compositional data. Genome Biol. 2022;23(1):95. Kassambara A. rstatix: Pipe-friendly framework for basic statistical tests. CRAN: Contributed Packages. 2019. McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PloS one. 2013;8(4):e61217. Wickham H. ggplot2: Elegant Graphics for Data Analysis. Use R. 2009:1-212. Wilke C. cowplot: Streamlined Plot Theme and Plot Annotations for ‘ggplot2’. 1.2.0 ed2025. Sankaran K, Holmes SP. Latent variable modeling for the microbiome. Biostatistics. 2019;20(4):599-614. Schaffer JN, Pearson MM. Proteus mirabilis and Urinary Tract Infections. Microbiol Spectr. 2015;3(5). Diop K, Diop A, Michelle C, Richez M, Rathored J, Bretelle F, et al. Description of three new Peptoniphilus species cultured in the vaginal fluid of a woman diagnosed with bacterial vaginosis: Peptoniphilus pacaensis sp. nov., Peptoniphilus raoultii sp. nov., and Peptoniphilus vaginalis sp. nov. Microbiologyopen. 2019;8(3):e00661. Murphy EC, Frick IM. Gram-positive anaerobic cocci--commensals and opportunistic pathogens. FEMS Microbiol Rev. 2013;37(4):520-53. Asangba AE, Chen J, Goergen KM, Larson MC, Oberg AL, Casarin J, et al. Diagnostic and prognostic potential of the microbiome in ovarian cancer treatment response. Sci Rep. 2023;13(1):730. Witkin SS, Mendes-Soares H, Linhares IM, Jayaram A, Ledger WJ, Forney LJ. Influence of vaginal bacteria and D- and L-lactic acid isomers on vaginal extracellular matrix metalloproteinase inducer: implications for protection against upper genital tract infections. mBio. 2013;4(4). France MT, Mendes-Soares H, Forney LJ. Genomic Comparisons of Lactobacillus crispatus and Lactobacillus iners Reveal Potential Ecological Drivers of Community Composition in the Vagina. Appl Environ Microbiol. 2016;82(24):7063-73. Zheng N, Guo R, Wang J, Zhou W, Ling Z. Contribution of Lactobacillus iners to Vaginal Health and Diseases: A Systematic Review. Front Cell Infect Microbiol. 2021;11:792787. Colbert LE, El Alam MB, Wang R, Karpinets T, Lo D, Lynn EJ, et al. Tumor-resident Lactobacillus iners confer chemoradiation resistance through lactate-induced metabolic rewiring. Cancer Cell. 2023;41(11):1945-62 e11. Ferjaoui MA, Arfaoui R, Khedhri S, Hannechi MA, Abdessamia K, Samaali K, et al. Pelvic actinomycosis: A confusing diagnosis. Int J Surg Case Rep. 2021;86:106387. Garcia-Gomez E, Gonzalez-Pedrajo B, Camacho-Arroyo I. Role of sex steroid hormones in bacterial-host interactions. Biomed Res Int. 2013;2013:928290. Mei S, Chelmow D, Gecsi K, Barkley J, Barrows E, Brooks R, et al. Health Disparities in Ovarian Cancer: Report From the Ovarian Cancer Evidence Review Conference. Obstet Gynecol. 2023;142(1):196-210. Srivastava SK, Ahmad A, Miree O, Patel GK, Singh S, Rocconi RP, et al. Racial health disparities in ovarian cancer: not just black and white. J Ovarian Res. 2017;10(1):58. Mitra A, Gultekin M, Burney Ellis L, Bizzarri N, Bowden S, Taumberger N, et al. Genital tract microbiota composition profiles and use of prebiotics and probiotics in gynaecological cancer prevention: review of the current evidence, the European Society of Gynaecological Oncology prevention committee statement. Lancet Microbe. 2024;5(3):e291-e300. Table Table 1 is available in the Supplementary Files section. Additional Declarations There is NO Competing Interest. Supplementary Files SupplementalFileMicrobiome.xlsx Supplemental Data Table ORCHiDK01DybiosisManuscriptSupplementalNature20250821AED.pdf Supplemental Figures Tables.docx Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7428423","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":505892667,"identity":"8e468c66-f467-493c-8c1f-7f9cd8abafdb","order_by":0,"name":"April Deveaux","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA8UlEQVRIiWNgGAWjYJCCAw8YmKHMCmYGCRDNQ0hLAlzLGSK1MMC1MLYRoYV/du9DoC3WcgbHm59J886zTpzZ3sD44G0bbi0Sd44bALWkGxucOWYmOXNbeuJsngPMhnPxaGG4kQbyy+HEmTMSzCQ+bjucOE8igU2aF48WeaiW+pnzn3+TSJwD1CL/gP03Pi0GUC0J/BI8QFsaDifOlmBgY8anxfDOMaAWg3TDfp6cYssZx9KNZ/YkNkvOOYdbi9ztNuYPHyqs5dnYj2+8zVNjLTvj+OGDH96U4fE+OBoMUIQYG/Coh2kZBaNgFIyCUYAPAABumVIDP609BwAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0002-1185-7750","institution":"Duke University School of Medicine","correspondingAuthor":true,"prefix":"","firstName":"April","middleName":"","lastName":"Deveaux","suffix":""},{"id":505892668,"identity":"4b08495f-a175-400e-9fc8-fbda10ebe897","order_by":1,"name":"Oyomoare Osazuwa-Peters","email":"","orcid":"","institution":"Duke University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Oyomoare","middleName":"","lastName":"Osazuwa-Peters","suffix":""},{"id":505892669,"identity":"6f7838dc-722e-4a76-bed1-d3ff81d563bc","order_by":2,"name":"Yeon Ji Kim","email":"","orcid":"https://orcid.org/0000-0002-1118-5026","institution":"Duke University","correspondingAuthor":false,"prefix":"","firstName":"Yeon","middleName":"Ji","lastName":"Kim","suffix":""},{"id":505892670,"identity":"e8c5615e-66f4-4c5f-bfda-b8406022b7f7","order_by":3,"name":"Pixu Shi","email":"","orcid":"","institution":"Duke University","correspondingAuthor":false,"prefix":"","firstName":"Pixu","middleName":"","lastName":"Shi","suffix":""},{"id":505892671,"identity":"9370fb7f-268e-4416-9118-8626136e248e","order_by":4,"name":"Drew Neish","email":"","orcid":"","institution":"Duke Cancer Institute","correspondingAuthor":false,"prefix":"","firstName":"Drew","middleName":"","lastName":"Neish","suffix":""},{"id":505892672,"identity":"7d9804b2-d9d1-4705-9817-cc3968e1f771","order_by":5,"name":"Ashwini Joshi","email":"","orcid":"","institution":"Duke University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Ashwini","middleName":"","lastName":"Joshi","suffix":""},{"id":505892673,"identity":"a48d1ef9-b2af-458d-b8bc-a8b1b101421e","order_by":6,"name":"Veronica Duck","email":"","orcid":"","institution":"Duke University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Veronica","middleName":"","lastName":"Duck","suffix":""},{"id":505892674,"identity":"17f9aa51-b62f-4adc-8be4-600384ebd922","order_by":7,"name":"Ashliegh Williams","email":"","orcid":"","institution":"Duke Human Vaccine Institute","correspondingAuthor":false,"prefix":"","firstName":"Ashliegh","middleName":"","lastName":"Williams","suffix":""},{"id":505892675,"identity":"3556fb51-c605-4dfc-8ea7-eacacfc49c4f","order_by":8,"name":"Margaret Gates Kuliszewski","email":"","orcid":"","institution":"University at Albany","correspondingAuthor":false,"prefix":"","firstName":"Margaret","middleName":"Gates","lastName":"Kuliszewski","suffix":""},{"id":505892676,"identity":"d12a6ca3-db85-481e-b134-2c524cf59ec8","order_by":9,"name":"Bin Huang","email":"","orcid":"","institution":"University of Kentucky","correspondingAuthor":false,"prefix":"","firstName":"Bin","middleName":"","lastName":"Huang","suffix":""},{"id":505892677,"identity":"2af05a14-8fda-4126-af57-7d373f9e90df","order_by":10,"name":"Kevin Ward","email":"","orcid":"","institution":"Emory University, Winship Cancer Institute","correspondingAuthor":false,"prefix":"","firstName":"Kevin","middleName":"","lastName":"Ward","suffix":""},{"id":505892678,"identity":"2607823b-593b-4f4f-a9be-ca88e4efab8d","order_by":11,"name":"Maria Pisu","email":"","orcid":"","institution":"University of Alabama at Birmingham","correspondingAuthor":false,"prefix":"","firstName":"Maria","middleName":"","lastName":"Pisu","suffix":""},{"id":505892679,"identity":"474efd93-4289-4c83-a7fd-912d16c3662e","order_by":12,"name":"Thomas Tucker","email":"","orcid":"","institution":"University of Kentucky, Markey Cancer Center","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"","lastName":"Tucker","suffix":""},{"id":505892680,"identity":"c91bebbc-b0d1-4cab-8301-1ffbdd4b0243","order_by":13,"name":"Rebecca Previs","email":"","orcid":"","institution":"Labcorp","correspondingAuthor":false,"prefix":"","firstName":"Rebecca","middleName":"","lastName":"Previs","suffix":""},{"id":505892681,"identity":"e7e6e394-cc53-4277-bb65-320eaf565d13","order_by":14,"name":"Andrew Berchuck","email":"","orcid":"","institution":"Duke University","correspondingAuthor":false,"prefix":"","firstName":"Andrew","middleName":"","lastName":"Berchuck","suffix":""},{"id":505892682,"identity":"78155ed7-62bb-4d5e-b8e0-11ea56d645ad","order_by":15,"name":"Tomi Akinyemiju","email":"","orcid":"","institution":"Duke University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Tomi","middleName":"","lastName":"Akinyemiju","suffix":""}],"badges":[],"createdAt":"2025-08-21 17:35:22","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7428423/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7428423/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":91074789,"identity":"6b92d5e4-498c-449a-83b5-94b822332823","added_by":"auto","created_at":"2025-09-11 11:04:17","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":148770,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStudy Design and Multi-Scale Analytical Framework for Dysbiotic Populations. Flow chart illustrating the analytical approach used in the ORCHiD study. A total of 132 OC patients provided vaginal samples for 16S rRNA gene sequencing (V1-V3 region). Three complementary analytical approaches were employed: traditional genus-level analysis, topic modeling using Latent Dirichlet Allocation (LDA) for unsupervised community detection, and species-level analysis focused on \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eL. crispatus, L. iners, L. gasseri, \u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003eand\u003c/strong\u003e\u003cem\u003e\u003cstrong\u003e L. jensenii\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e. All analyses were stratified by self-reported race (NH-Black/NH-White), age (\u0026lt;50/≥50 years), cancer stage (Early/Late), and histology Type I + Other/Type II).\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7428423/v1/06f038e7b51250ea24a86819.png"},{"id":91073027,"identity":"e4a6dfce-3027-40fb-8bc8-9904bfad5d77","added_by":"auto","created_at":"2025-09-11 10:56:17","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":231746,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eVaginal Microbiome Composition Across Demographic and Clinical Variables. Stacked bar charts showing relative abundance of major bacterial genera in OC patients. (A) Overall cohort composition, (B) Composition by age group (\u0026lt;50 vs ≥50 years, (C) Composition by cancer stage (Early vs Late), (D) Composition by race (NH-Black vs NH-White), and (E) Composition by histologic subtype (Type I epithelial + Other vs Type II epithelial).\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7428423/v1/8ada7df84135136c1f427df5.png"},{"id":91074790,"identity":"f0192ba1-2611-4a56-ae7e-ed9d028db9a2","added_by":"auto","created_at":"2025-09-11 11:04:17","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":169491,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eVaginal Microbiome Topic Signatures Identified by Latent Dirichlet Allocation. Bar chart displaying \u003c/strong\u003ethe top 3 genera per microbial topic based on beta probability values. Seven distinct topics were identified: Topic 1 (\u003cem\u003eStreptococcus\u003c/em\u003eMixed) - moderate dysbiosis characterized by \u003cem\u003eStreptococcus\u003c/em\u003e (48.1%), \u003cem\u003eAtopobium\u003c/em\u003e(26.9%), and \u003cem\u003eSneathia\u003c/em\u003e (7.2%); Topic 2 (\u003cem\u003eProteus\u003c/em\u003e Dominated) - severe pathogenic dysbiosis dominated by \u003cem\u003eProteus\u003c/em\u003e (87.9%); Topic 3 (Diverse Community) - intermediate dysbiosis with \u003cem\u003eCorynebacterium\u003c/em\u003e(27.8%), \u003cem\u003eEscherichia-Shigella\u003c/em\u003e (27.5%), and \u003cem\u003eFinegoldia\u003c/em\u003e (21.2%); Topic 4 (Anaerobic Dysbiosis) - anaerobic dysbiosis led by \u003cem\u003ePeptoniphilus\u003c/em\u003e(20.7%) and \u003cem\u003eAnaerococcus\u003c/em\u003e (19.3%); Topic 5 (\u003cem\u003eLactobacillus\u003c/em\u003eDominated) - healthy signature nearly exclusively composed of \u003cem\u003eLactobacillus\u003c/em\u003e(98.6%); Topic 6 (\u003cem\u003ePseudomonas \u003c/em\u003eDominated) - potentially representing opportunistic growth with \u003cem\u003ePseudomonas\u003c/em\u003e (51.2%); Topic 7 (BV-Type) - bacterial vaginosis-type dysbiosis dominated by \u003cem\u003ePrevotella\u003c/em\u003e (70.0%).\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7428423/v1/478c947d04de2605f2a06d79.png"},{"id":91073029,"identity":"d372d973-ddb4-43d0-809b-77bce0929032","added_by":"auto","created_at":"2025-09-11 10:56:17","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":221520,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTopic-Clinical Variable Associations Using LinDA Differential Abundance Analysis. \u003c/strong\u003eForest plots showing log₂fold-changes for microbial topics across clinical variables. (A) Age comparison (≥50 vs \u0026lt;50 years) revealing Topic 4 (Transitional/Anaerobic Dysbiosis) as significantly enriched in older patients (log₂ FC = 1.31, FDR p = 0.001, highlighted in blue). (B) Disease stage comparison (Late vs Early) demonstrating Topic 7 (BV-Type) enrichment in late-stage disease (log₂ FC = -1.5). (C) Race comparison (White vs Black) showing Topic 5 (\u003cem\u003eLactobacillus\u003c/em\u003e-dominated) depletion in White patients (log₂FC = -2.5). (D) Histologic subtype comparison (Type II vs Type I) showing variable topic distributions across cancer types. Gray bars indicate non-significant associations, while the blue bar represents the only statistically significant finding after FDR correction.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7428423/v1/037e2b1251936bea6bb07e27.png"},{"id":91073043,"identity":"0658916e-5ec2-44d7-9a47-ca4e1ecc46b5","added_by":"auto","created_at":"2025-09-11 10:56:17","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":196614,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSpecies-Level Lactobacillus Validation of Topic Modeling Findings.\u003c/strong\u003eAnalysis of key \u003cem\u003eLactobacillus\u003c/em\u003e species supporting topic-level results. Top panels show detection rates (prevalence) for \u003cem\u003eL. crispatus\u003c/em\u003e (teal) and \u003cem\u003eL. iners\u003c/em\u003e (purple) by age category, cancer stage, and race. \u003cem\u003eL. crispatus\u003c/em\u003e detection was 3.75-fold higher in younger vs older patients (13.6% vs 3.6%, FDR p = 0.125) and 5.55-fold higher in early-stage vs late-stage patients (12.2% vs 2.2%, FDR p = 0.067), while being detected exclusively in White patients (6.4% vs 0% in Black patients, FDR p = 0.458). Conversely, \u003cem\u003eL. iners\u003c/em\u003e showed 2.27-fold higher prevalence in Black vs White patients (22.7% vs 10%, FDR p = 0.295). Bottom panels display relative abundance when species were detected, showing that \u003cem\u003eL. crispatus\u003c/em\u003e maintains high abundance levels when present, while \u003cem\u003eL. iners\u003c/em\u003e shows more variable patterns.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7428423/v1/deca1513efe785b24580613f.png"},{"id":91079986,"identity":"b8d1fbed-c9e2-4094-8c86-ace50568051b","added_by":"auto","created_at":"2025-09-11 11:28:18","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2390694,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7428423/v1/1a7e8975-4909-4d12-8df1-67e39731ccb4.pdf"},{"id":91073024,"identity":"a8688aef-af08-42d7-b98b-88a905a72a7e","added_by":"auto","created_at":"2025-09-11 10:56:17","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":35656,"visible":true,"origin":"","legend":"Supplemental Data Table","description":"","filename":"SupplementalFileMicrobiome.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7428423/v1/53f68c089f1a991e14b1b1ac.xlsx"},{"id":91074792,"identity":"e22085f5-519d-4c9f-9909-86ba41c0d95c","added_by":"auto","created_at":"2025-09-11 11:04:17","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":947666,"visible":true,"origin":"","legend":"Supplemental Figures","description":"","filename":"ORCHiDK01DybiosisManuscriptSupplementalNature20250821AED.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7428423/v1/1f2a7890bcc5d51962959d24.pdf"},{"id":91073026,"identity":"685b6ee6-de30-4cb5-937d-ce401052cfbf","added_by":"auto","created_at":"2025-09-11 10:56:17","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":18174,"visible":true,"origin":"","legend":"","description":"","filename":"Tables.docx","url":"https://assets-eu.researchsquare.com/files/rs-7428423/v1/88b12e7f9d79da4633facbdc.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Vaginal Microbiome Composition, Diversity and Dysbiosis: The ORCHiD Study","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThe vaginal microbiome, the community of microbes in the lower female genital tract, has emerged as an important factor in gynecologic health and gynecological cancers and may influence disease prognosis (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). A healthy microbiome is typically dominated by \u003cem\u003eLactobacillus\u003c/em\u003e species, which help maintain a low pH and protect against pathogens(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). When the protective \u003cem\u003eLactobacillus\u003c/em\u003e-dominated balance is disrupted\u0026mdash;a state known as vaginal dysbiosis \u0026ndash; the microbiome shifts toward communities dominated by anaerobic bacteria(\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). This dysbiotic state has been linked to adverse vaginal health outcomes such as bacterial vaginosis, pelvic inflammatory disease and chronic genital inflammation(\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e), and can contribute to immune dysregulation, epithelial dysfunction, and pro-inflammatory signaling - important mechanisms in tumor progression(\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eOvarian cancer (OC) remains one of the most lethal gynecological malignancies in the United States, with a five-year survival rate of only 51%(\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e), however few studies have characterized the role of the vaginal microbiome in driving OC prognosis. Recent studies suggest that OC is associated with a shift in the vaginal microbiome toward a dysbiotic state(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). In a case-control study with ovarian cancer patients, Morikawa et al. found that the cervicovaginal microbiota of OC patients was characterized by significant loss of \u003cem\u003eLactobacillus\u003c/em\u003e-dominated communities and increased microbial diversity across multiple histologic subtypes compared to healthy controls(\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). Sipos et al. proposed the concept of an \u0026ldquo;oncobiome\u0026rdquo;, describing how pathogen-driven disruptions such as infections with \u003cem\u003eChlamydia trachomatis\u003c/em\u003e or \u003cem\u003eNeisseria gonorrhoeae\u003c/em\u003e may lead to lasting alterations in the vaginal microbiome that promote chronic inflammation and carcinogenesis(\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Building on this, Mehra et al. reviewed evidence of altered vaginal and gut microbial ecosystems in OC, reporting that \u003cem\u003eLactobacillus\u003c/em\u003e was present in only 24% of women with OC compared to 47% of healthy controls(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). They also identified dysbiosis-associated microbial shifts as contributors to immune dysregulation and tumor progression, suggesting the potential for microbiome modulation as a diagnostic or therapeutic strategy(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). Together, these findings describe a broader biological framework in which vaginal dysbiosis is both a marker and a possible mediator of OC pathophysiology.\u003c/p\u003e\u003cp\u003eThe vaginal microbiome varies by patient demographics and clinical factors (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). For instance, aging is associated with declining estrogen levels and shifts in immune function that reduce \u003cem\u003eLactobacillus\u003c/em\u003e abundance while increasing anaerobic bacterial diversity(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). The vaginal microbiome composition also varies by race/ethnicity (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e), with documented differences in microbial diversity, bacterial species composition, and \u003cem\u003eLactobacillus\u003c/em\u003e distribution patterns(\u003cspan additionalcitationids=\"CR18 CR19\" citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). Such compositional differences might contribute to disparities in OC outcomes, including tumor aggressiveness and progression(\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). Notably, Black women experience significantly worse OC outcomes compared to White women, including later-stage diagnosis, more aggressive tumor biology, and reduced survival(\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). While socioeconomic factors and healthcare access contribute to these disparities, emerging evidence suggests that biological factors, including microbiome differences, may also play a role(\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). A modest number of prior studies have investigated the role of the vaginal microbiome in OC outcomes (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan additionalcitationids=\"CR27\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e), however these investigations have largely evaluated alpha and beta diversity, differential abundance testing, and taxonomic profiling based on relative abundances(\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). These traditional analytical approaches, while informative, are unable to identify complex community-level patterns that emerge from co-occurring bacterial taxa(\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e), insights that might reveal novel mechanistic pathways contributing to OC outcomes and disparities.\u003c/p\u003e\u003cp\u003eIn this analysis, we aimed to characterize the vaginal microbiome composition, diversity and dysbiosis among patients in the Ovarian Cancer Epidemiology, Healthcare Access and Disparities Study (ORCHiD). Specifically, we sought to: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) describe the study population characteristics and overall microbiome diversity patterns in OC patients; (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) characterize vaginal microbiome composition using traditional differential abundance analysis across patient demographics and disease characteristics; (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) identify distinct vaginal microbial community signatures using topic modeling and examine their associations with clinical variables; and (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) validate findings through targeted species-level analysis of key \u003cem\u003eLactobacillus\u003c/em\u003e species. Our findings may contribute to understanding microbiome-mediated health disparities and inform targeted therapeutic interventions aimed at improving OC outcomes across demographic groups.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eStudy Design and Population\u003c/h2\u003e\u003cp\u003eAn overview of the study design and framework is presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. This study included 135 ovarian cancer (OC) patients recruited as part of the ORCHiD study. Detailed methods are published elsewhere(\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e). Briefly, this population-based study recruited patients with confirmed first primary OC between March 2021 and October 2024 from seven state cancer registries, including New York, Kentucky, California, North Carolina, Georgia, and Texas. The study inclusion criteria were patients aged 18 years and older with a pathologically confirmed diagnosis of OC (stages I-IV) at 9\u0026ndash;12 months prior to recruitment. All patients in this cohort self-identified their race/ethnicity as non-Hispanic Black (NH-Black) or non-Hispanic White (NH-White). The study exclusion criteria were patients older than 79 years old, or patients with cognitive impairments preventing survey completion. The ORCHiD survey comprehensively captures data on patient demographics, socio-economic status, healthcare access, as well as self-reported treatment, lifestyle, and health history data. Eligible patients were contacted via phone to confirm eligibility, conduct informed consent, and determine participation interest. Patients who consent to participate complete the ORCHiD survey over the phone, online, or by mail and receive an incentive of \u003cspan\u003e$\u003c/span\u003e25 for survey completion.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eVaginal Microbiome Sample Collection and Processing\u003c/h3\u003e\n\u003cp\u003ePatients who completed the ORCHiD survey were offered an opportunity to participate in the ORCHiD biospecimen sub study which required self-collection of both a vaginal swab and saliva. Once a patient consented to the sub-study, the study team mailed a biospecimen packet containing a vaginal microbiome collection kit (OMR-130, DNA Genotek, Ottawa, Canada) along with manufacturer instructions and a postage-paid sample return envelope. Each biospecimen packet also included a brief 4-question behavioral survey querying medications or microbiome-altering behaviors in the previous 30 days, including oral or IV antibiotic use or use of douching or suppository products. Patients responding yes to any of the microbiome survey questions were excluded from the analysis cohort. Patients were also instructed not to collect samples when actively menstruating or experiencing other vaginal bleeding. Upon receipt of returned kits, samples were inspected for any damage or leakage and logged electronically. The collection kit included a swab and a vial containing microbial media buffer, a proprietary stabilizing liquid to preserve samples at room temperature for up to 30 days. Any damaged kits were disposed of according to established safety regulations. Samples were processed by treating with 5 \u0026micro;l of Proteinase K (80 mg/ml) and incubating for one hour in a 50\u0026deg;C water bath, then aliquoted into 1.5 ml microcentrifuge tubes and stored at -80\u0026deg;C for downstream processing. Of the 764 patients recruited into the parent ORCHiD study, 435 (56.9%) consented to the biospecimen sub-study, and 230 (52.9% of consenters, 30.1% of total surveyed) returned at least one sample. Patients who completed and returned the biospecimen kits received an additional \u003cspan\u003e$\u003c/span\u003e20 incentive.\u003c/p\u003e\u003cp\u003eProcessed vaginal samples were sent to the Duke Microbiome Core Facility for assay. First, microbial DNA from vaginal swab samples preserved in OMNIgene vaginal kit was extracted using a DNeasy 96 PowerSoil Pro QIACube HT kit (QIAGEN, #47021) on an automated machine (QIACube HT, QIAGEN). The manufacturer\u0026rsquo;s instructions were followed with these minor deviations: 250 \u0026micro;l of the preserved sample was used as starting material, and during the final elution step, the elution buffer (C6) was incubated at room temperature for 10 minutes to increase DNA yield. Bacterial community composition from vaginal swab samples was characterized by amplification of the 16S rRNA V1-V3 hypervariable regions via polymerase chain reaction (PCR). Forward primer 27F (5\u0026rsquo;-GAGTTTGATCGTGGCTCAG-3\u0026rsquo;) and reverse primer 518R (5\u0026rsquo;-ATTACCGCGGCTGCTGG-3\u0026rsquo;) were used with Phusion Plus PCR Master Mix (ThermoScientific, #F631L) for a total of 25 PCR cycles. Samples were multiplexed via the reverse primers (518R) that carry unique Earth Microbiome Project (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.earthmicrobiome.org/\u003c/span\u003e\u003cspan address=\"http://www.earthmicrobiome.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) barcodes (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e). PCR products were purified with AMPure XP Beads (Beckman Coulter, #A63881) and quantified with a Qubit dsDNA HS assay kit (ThermoFisher, #Q32854) on a Promega GloMax plate reader. Equimolar PCR products were pooled from each sample and submitted to the Duke Sequencing and Genomic Technologies shared resource. The final pool was sequenced on an Illumina MiSeq platform using v3 chemistry (300 base pairs, paired-end).\u003c/p\u003e\n\u003ch3\u003eStudy Variables\u003c/h3\u003e\n\u003cp\u003eCohort characteristics assessed included self-reported race, age at diagnosis, OC stage, and histologic subtype. Race was self-reported and categorized as non-Hispanic Black or non-Hispanic White. Age was analyzed as a continuous variable for primary analyses and then dichotomized into two age categories (\u0026lt;\u0026thinsp;50 years and \u0026ge;50 years) for interpretability of topic distributions. Clinical data, such as stage at diagnosis and histological subtype, were obtained from state cancer registries. OC stage was derived from summary staging data and categorized into early-stage (summary stage 1 or 2), or late-stage (summary stage greater than 2). Histologic subtype was categorized as either type II epithelial (including high-grade serous and other aggressive subtypes) or Other (including type I epithelial, endometrioid, mucinous, clear cell subtypes, and non-epithelial tumors).\u003c/p\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003e\u003cstrong\u003eData Processing\u003c/strong\u003e\u003cp\u003eMicrobiome sequencing data was processed using R programming suite (v4.3.0) (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). Sequencing reads from vaginal samples were processed using the DADA2 pipeline (v1.28.0) (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e) to infer amplicon sequence variants (ASVs) with default parameters. Deviations from default parameters include a parameter for FilterAndTrim function (truncLen\u0026thinsp;=\u0026thinsp;c(290, 260)) to remove sequencing regions with low quality scores and only including ASVs that were 433 to 525 base pairs long in analysis. Taxonomy was assigned using the Silva v138.1 database (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e) and non-bacterial taxa were removed. Samples with less than 1,000 reads were removed, which included 4 negative controls and 3 samples with insufficient sample size, and a phylogenetic tree was constructed using Randomized Axelerated Maximum Likelihood (RAxML) to account for phylogenetic relationships (\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e).\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eSummary Statistics\u003c/b\u003e: For taxonomic composition analysis, we calculated three key metrics for each bacterial genus: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) prevalence, defined as the percentage of samples in which the genus was detected above the minimum threshold; (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) mean relative abundance, calculated as the average percentage of total reads attributed to each genus across all samples; and (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) median relative abundance with interquartile range (IQR), providing robust measures of central tendency and spread for non-normally distributed abundance data.\u003c/p\u003e\u003cp\u003e\u003cb\u003eTopic Modeling Analysis\u003c/b\u003e: We employed Latent Dirichlet Allocation (LDA) to identify distinct vaginal microbial community signatures, termed \"microbial topics,\" representing co-occurring bacterial taxa(\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e). The range of 3 to 10 topics was selected based on established vaginal microbiome community classifications, as the foundational community state type (CST) framework identifies 5 CSTs with up to 9 CSTs with extended classification(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e). This biologically informed range ensures alignment with known vaginal microbiome structure while allowing for potential novel community patterns in this cancer population. Optimal topic number was determined using the FindTopicsNumber function from the \"ldatuning\" R package (v1.0.2)(\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e) with multiple metrics including CaoJuan2009, Arun2010, Griffiths2004, and Deveaud2014 (\u003cspan additionalcitationids=\"CR43 CR44\" citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e) evaluated across the topic number range. Two metrics (CaoJuan2009 and Arun2010) identify optimal models by minimization, while the other two metrics (Griffiths2004 and Deveaud2014) identify optimal models by maximization. The optimal number of topics was selected using an elbow method approach (\u003cb\u003eSupplemental Fig.\u0026nbsp;1\u003c/b\u003e). LDA was implemented using the \"topicmodels\" package (v0.2.17) in R(\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e) with Gibbs sampling method, treating samples as \"documents\" and genera as \"words.\" The model was run with a fixed seed (243) for reproducibility. Beta probability matrices (genus-topic associations) and gamma probability matrices (sample-topic associations) were extracted using the \"tidytext\" package (v0.4.3) (\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e). Topics were characterized by their dominant genera (highest beta probabilities) and named based on their microbial composition and established associations with vaginal health states.\u003c/p\u003e\u003cp\u003e\u003cb\u003eDifferential Abundance Analysis of Topics\u003c/b\u003e: Topic-level differential abundance testing was performed using the LinDA (linear models for \u003cem\u003edifferential\u003c/em\u003e abundance analysis) framework(\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e), which accounts for compositional data structure and provides robust statistical inference. Topic counts were calculated by multiplying gamma probabilities by corresponding read counts for each sample, creating a topic-level phyloseq object. LinDA analysis was conducted with the formula: ~ Race\u0026thinsp;+\u0026thinsp;Stage\u0026thinsp;+\u0026thinsp;Histology\u0026thinsp;+\u0026thinsp;Age, testing each clinical variable while controlling for others. The analysis used empirical Bayes shrinkage, centered log-ratio transformation, and Benjamini-Hochberg false discovery rate (FDR) correction for multiple testing (FDR q\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e\u003cp\u003e\u003cb\u003eSpecies-Level\u003c/b\u003e \u003cb\u003eLactobacillus\u003c/b\u003e \u003cb\u003eValidation\u003c/b\u003e: To validate topic-level findings, we performed targeted species-level analyses focusing on key \u003cem\u003eLactobacillus\u003c/em\u003e species known to be important for vaginal health. The phyloseq object was filtered to include only \u003cem\u003eLactobacillus\u003c/em\u003e species, and relative abundances were calculated by dividing species counts by total reads per sample. We examined both detection rates (presence/absence) and relative abundances when present for \u003cem\u003eL. crispatus\u003c/em\u003e, \u003cem\u003eL. iners\u003c/em\u003e, \u003cem\u003eL. gasseri\u003c/em\u003e, and \u003cem\u003eL. jensenii\u003c/em\u003e across demographic and clinical variables.\u003c/p\u003e\u003cp\u003e\u003cb\u003eAdditional Analyses\u003c/b\u003e: Detection rates between groups were compared using Fisher's exact tests, while relative abundances (when detected) were compared using Wilcoxon rank-sum tests due to non-normal distributions. For detection rate analyses, effect sizes were calculated as odds ratios with 95% confidence intervals. For abundance comparisons, effect sizes were reported as fold-changes between group medians. FDR correction was applied across all species tested within each comparison type (detection vs abundance). Statistical significance was set at FDR q\u0026thinsp;\u0026lt;\u0026thinsp;0.05, with trends reported for FDR q\u0026thinsp;\u0026lt;\u0026thinsp;0.10. Alpha diversity (Shannon diversity index and Faith's phylogenetic distance) was calculated using phyloseq (v1.44.0) and picante (v1.8.2) packages, while beta diversity was assessed using Jaccard, Bray-Curtis, UniFrac, and weighted UniFrac dissimilarity metrics calculated with the vegan package (v2.6.4), with principal coordinate analysis (PCoA) performed for visualization. Alpha diversity metrics were tested for normality using Shapiro-Wilk tests, and differences by demographic and clinical variables were assessed using appropriate parametric (ANOVA) or non-parametric (Kruskal-Wallis) tests, while beta diversity differences were tested using permutational analysis of variance (PERMANOVA) with FDR correction applied across multiple comparisons.\u003c/p\u003e\u003cp\u003e\u003cb\u003eData Visualization and Statistical Software\u003c/b\u003e: All statistical analyses were performed in R (v4.3.0)(\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e) using the following packages: \u003cem\u003ephyloseq\u003c/em\u003e (v1.48.0) for microbiome data handling(\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e), \u003cem\u003etopicmodels\u003c/em\u003e (v0.2.17\u003cem\u003e)\u003c/em\u003e for LDA implementation(\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e), \u003cem\u003etidytext\u003c/em\u003e (v0.4.3) for topic model interpretation(\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e), \u003cem\u003eLinDA\u003c/em\u003e (v0.2.0) for differential abundance testing(\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e), \u003cem\u003eggplot2\u003c/em\u003e (v3.5.2)for visualization(\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e), and \u003cem\u003ecowplot\u003c/em\u003e (v1.2.0)for multi-panel figures(52). Statistical significance was set at FDR-adjusted q\u0026thinsp;\u0026lt;\u0026thinsp;0.05 unless otherwise specified.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eCohort Characteristics\u003c/h2\u003e\u003cp\u003eA total of 132 OC patients were included in the analysis (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e); 83.7% NH-White (n\u0026thinsp;=\u0026thinsp;111) and 16.3% NH-Black (n\u0026thinsp;=\u0026thinsp;22). The median age was 61.5 years (range 30\u0026ndash;79), with 83.0% of patients 50 years or older at diagnosis. Compared to NH-White patients, NH-Black patients were more likely to report lower annual household income (22.7% vs. 7.1% reporting \u0026lt;\u003cspan\u003e$\u003c/span\u003e20,000), unemployment (13.6% vs. 3.5%), or disability (27.3% vs. 2.7%). About 72.7% of NH-Black patients widowed, divorced, separated, or never married, compared to 31.9% of NH-White patients. Most patients were diagnosed with late-stage disease, defined as FIGO stage III\u0026ndash;IV (67.4%), although late-stage diagnosis was higher in NH-Black vs NH-White patients (77.3% vs. 65.5%). Type II epithelial was the most common histologic subtype (60.0%). Most patients received surgery (94.1%) and chemotherapy (75.6%), with no notable differences in treatment by race.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eMicrobiome Diversity\u003c/h3\u003e\n\u003cp\u003eWe evaluated vaginal microbiome diversity using both alpha (Shannon index) and beta (weighted UniFrac distance) diversity metrics (\u003cb\u003eSupplemental Fig.\u0026nbsp;1\u003c/b\u003e). Alpha diversity, assessed via the Shannon index which incorporates both richness and evenness, did not differ significantly by race (median: 2.37 NH-Black vs. 2.52 NH-White; p\u0026thinsp;=\u0026thinsp;0.72; \u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eA\u003c/b\u003e), OC stage (median: 2.54 early vs. 2.42 late; p\u0026thinsp;=\u0026thinsp;0.49; \u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eC\u003c/b\u003e), or histologic subtype (median: 2.39 Type I vs. 2.48 Type II; p\u0026thinsp;=\u0026thinsp;0.78; \u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eE\u003c/b\u003e). These findings suggest that within-sample microbial richness and evenness were relatively stable across demographic and clinical strata. Beta diversity, quantified using weighted UniFrac distances which incorporates relative abundance and phylogenetic information, showed no significant differences in overall microbial community composition by race (PERMANOVA p\u0026thinsp;=\u0026thinsp;0.64; \u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eB\u003c/b\u003e) or stage (PERMANOVA p\u0026thinsp;=\u0026thinsp;0.76; \u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eD\u003c/b\u003e). However, significant differences were observed by histologic subtype (PERMANOVA p\u0026thinsp;=\u0026thinsp;0.025; \u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eF\u003c/b\u003e), indicating that the structure of the vaginal microbiome varied between Type I and Type II epithelial OC.\u003c/p\u003e\n\u003ch3\u003eVaginal Microbiome Composition and Relative Abundance\u003c/h3\u003e\n\u003cp\u003eVaginal microbiome overall composition analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, \u003cb\u003eSupplemental File\u003c/b\u003e) showed \u003cem\u003eLactobacillus\u003c/em\u003e as the most common genus by mean relative abundance (mean: 23.5%, median: 0% [IQR: 0-44.5%], prevalence: 47.7%), followed by \u003cem\u003ePrevotella\u003c/em\u003e (mean: 12.3%, median: 4.9% [IQR: 0.6\u0026ndash;23.6%], prevalence: 93.9%) and other anaerobic genera including \u003cem\u003ePeptoniphilus\u003c/em\u003e (mean: 6.0%, prevalence: 95.5%), \u003cem\u003eAnaerococcus\u003c/em\u003e (mean: 5.6%, prevalence: 90.9%), and \u003cem\u003eFinegoldia\u003c/em\u003e (mean: 5.1%, prevalence: 96.2%). The difference between mean and median abundance for \u003cem\u003eLactobacillus\u003c/em\u003e (23.5% vs. 0%) indicated a bimodal distribution, with this genus being either absent/minimal or highly dominant across individuals. This pattern reflects a generally dysbiotic microbiome across the cohort, with \u003cem\u003eLactobacillus\u003c/em\u003e species present in less than half of patients and anaerobic bacteria widely prevalent (\u0026gt;\u0026thinsp;90% prevalence for most genera).\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eAge-Related Differences\u003c/h2\u003e\u003cp\u003eOlder patients\u0026thinsp;\u0026ge;\u0026thinsp;50 years had significantly higher prevalence of \u003cem\u003eActinotignum\u003c/em\u003e compared to younger patients (\u0026lt;\u0026thinsp;50 years: prevalence 31.8%; \u0026ge;50 years: prevalence 68.2%, FDR q\u0026thinsp;=\u0026thinsp;0.007), though mean relative abundances remained low in both groups (\u0026lt;\u0026thinsp;50: mean 2.0%, median 0%; \u0026ge;50: mean 2.7%, median 0%). This age-related pattern suggests increased colonization by this genus in older OC patients. There were no other significant or trending associations by age group.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003eCancer Stage Differences\u003c/h2\u003e\u003cp\u003eEarly stage patients had significantly higher prevalence of \u003cem\u003eCutibacterium\u003c/em\u003e compared to late-stage patients (prevalence: 61.0% vs 33.0%, FDR q\u0026thinsp;=\u0026thinsp;0.048). There were no other significant or trending associations by disease stage.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eRacial Differences\u003c/h2\u003e\u003cp\u003eNH-Black patients had a 5-fold higher prevalence of Actinomycetaceae family bacteria compared to NH-White patients (40.9% vs 8.2%, FDR q\u0026thinsp;=\u0026thinsp;0.005), representing the most robust compositional difference observed in the cohort. NH-Black patients also had higher \u003cem\u003eLactobacillus\u003c/em\u003e prevalence (68.2% vs 43.6%) and mean abundance (mean 34.4% vs 21.4%), though this did not reach statistical significance (FDR q\u0026thinsp;=\u0026thinsp;0.271). Notably, while overall \u003cem\u003eLactobacillus\u003c/em\u003e abundance appeared higher in NH-Black patients, this pattern was driven primarily by different species composition (detailed in species-level analysis below). There were no other significant or trending racial differences in microbial composition.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003eHistologic Subtype Differences\u003c/h2\u003e\u003cp\u003eThere were no significant or trending associations between microbial composition and histologic subtype (all FDR q\u0026thinsp;\u0026gt;\u0026thinsp;0.25).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003eTopic Modeling Results\u003c/h2\u003e\u003cdiv id=\"Sec16\" class=\"Section3\"\u003e\u003ch2\u003eIdentification of Vaginal Microbial Topics\u003c/h2\u003e\u003cp\u003eLDA identified seven microbial topics using an elbow method approach, representing distinct co-occurring genera patterns that were biologically interpretable without excessive model complexity(\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e) \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, \u003cb\u003eSupplemental Fig.\u0026nbsp;2).\u003c/b\u003e Topic 1 (\u003cem\u003eStreptococcus\u003c/em\u003e Mixed) was characterized by moderate dysbiosis with \u003cem\u003eStreptococcus\u003c/em\u003e (48.1%), \u003cem\u003eAtopobium\u003c/em\u003e (26.9%), and \u003cem\u003eSneathia\u003c/em\u003e (7.2%), as \u003cem\u003eAtopobium\u003c/em\u003e and \u003cem\u003eSneathia\u003c/em\u003e are established bacterial vaginosis-associated taxa(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). Topic 2 (\u003cem\u003eProteus\u003c/em\u003e Dominated) represented severe pathogenic dysbiosis, being highly dominated by \u003cem\u003eProteus\u003c/em\u003e (87.9%) with minimal diversity, as \u003cem\u003eProteus\u003c/em\u003e species are associated with urogenital infections and pathogenic overgrowth(\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e). Topic 3 (Diverse Community) showed intermediate dysbiosis led by \u003cem\u003eCorynebacterium\u003c/em\u003e (27.8%), \u003cem\u003eEscherichia-Shigella\u003c/em\u003e (27.5%), and \u003cem\u003eFinegoldia\u003c/em\u003e (21.2%), representing the diverse, non-\u003cem\u003eLactobacillus\u003c/em\u003e communities characteristic of dysbiotic states(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). Topic 4 (Anaerobic Dysbiosis) was dominated by \u003cem\u003ePeptoniphilus\u003c/em\u003e (20.7%) and \u003cem\u003eAnaerococcus\u003c/em\u003e (19.3%) with \u003cem\u003eCampylobacter\u003c/em\u003e (9.5%), representing anaerobic dysbiosis as these taxa are associated with bacterial vaginosis and inflammatory conditions(\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e). Topic 5 (\u003cem\u003eLactobacillus\u003c/em\u003e-Dominated) was nearly exclusively composed of \u003cem\u003eLactobacillus\u003c/em\u003e (98.6%), which can vary significantly in protective capacity depending on the specific species composition(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). Topic 6 (\u003cem\u003ePseudomonas\u003c/em\u003e-dominated) was led by \u003cem\u003ePseudomonas\u003c/em\u003e (51.2%) with diverse anaerobes that can represent opportunistic bacterial overgrowth in severely dysbiotic vaginal environments(\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e). Topic 7 (BV-Type) was dominated by \u003cem\u003ePrevotella\u003c/em\u003e (70.0%), representing classic bacterial vaginosis-type dysbiosis as \u003cem\u003ePrevotella\u003c/em\u003e is a hallmark of BV-associated microbiomes(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003eTopic-Clinical Variable Associations\u003c/h2\u003e\u003cp\u003eLinDA differential abundance analysis revealed significant associations between microbial topics and patient demographics (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Topic 4 (Anaerobic Dysbiosis) showed the strongest and only statistically significant association, with significantly higher abundance in patients\u0026thinsp;\u0026ge;\u0026thinsp;50 years compared to younger patients (log₂ fold-change\u0026thinsp;=\u0026thinsp;1.31, FDR q\u0026thinsp;\u0026lt;\u0026thinsp;0.001). This was the only topic-clinical variable association reaching statistical significance after multiple testing correction. Disease stage showed limited associations with topic abundance after multiple testing correction. Topic 6 (Environmental) showed enrichment in late-stage compared to early-stage disease (log₂ fold-change = -1.9, FDR q\u0026thinsp;=\u0026thinsp;0.06), though this did not reach statistical significance. While no racial associations reached statistical significance after FDR correction, consistent directional patterns emerged across topics. Topic 5 (\u003cem\u003eLactobacillus\u003c/em\u003e-dominated) showed the most substantial depletion in NH-White compared to NH-Black patients (log₂ fold-change = -3.0, FDR q\u0026thinsp;=\u0026thinsp;0.09), and Topic 2 (\u003cem\u003eProteus\u003c/em\u003e-Dominated) demonstrated depletion in NH-White versus NH-Black patients (log₂ fold-change = -1.4, FDR q\u0026thinsp;=\u0026thinsp;0.09). No significant associations were observed between topics and histologic subtypes after FDR correction, suggesting that overall microbial community patterns may be more influenced by demographic factors than tumor characteristics.\u003c/p\u003e\u003cp\u003e\u003cb\u003eSpecies-Level Analysis of Key\u003c/b\u003e \u003cb\u003eLactobacillus\u003c/b\u003e \u003cb\u003eSpecies\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo validate the topic modeling findings, particularly regarding the healthy \u003cem\u003eLactobacillus\u003c/em\u003e signature (Topic 5), we performed detailed species-level analyses of the four predominant vaginal \u003cem\u003eLactobacillus\u003c/em\u003e species (\u003cb\u003eSupplemental File\u003c/b\u003e) with key findings for \u003cem\u003eL. crispatus\u003c/em\u003e and \u003cem\u003eL. iners\u003c/em\u003e highlighted in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e. Given the bimodal distribution of \u003cem\u003eLactobacillus\u003c/em\u003e observed at the genus level, we examined both detection rates (presence/absence) and relative abundance when present for key \u003cem\u003eLactobacillus\u003c/em\u003e species.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\u003ch2\u003eDetection Rates and Clinical Associations\u003c/h2\u003e\u003cp\u003eAge-related patterns revealed differential effects on \u003cem\u003eLactobacillus\u003c/em\u003e species colonization. \u003cem\u003eL. crispatus\u003c/em\u003e detection was 3.75-fold higher in patients\u0026thinsp;\u0026lt;\u0026thinsp;50 years compared to those\u0026thinsp;\u0026ge;\u0026thinsp;50 years (13.6% vs 3.6%, FDR p\u0026thinsp;=\u0026thinsp;0.125), while \u003cem\u003eL. iners\u003c/em\u003e showed a similar but less pronounced age-related decline (18.2% vs 10.9%, FDR p\u0026thinsp;=\u0026thinsp;0.454). \u003cem\u003eL. crispatus\u003c/em\u003e was detected in 12.2% of early-stage patients compared to only 2.2% of late-stage patients, representing a 5.55-fold difference (FDR p\u0026thinsp;=\u0026thinsp;0.067). In contrast, \u003cem\u003eL. iners\u003c/em\u003e showed the opposite pattern, with higher detection in late-stage patients (14.3%) compared to early-stage patients (7.3%, FDR p\u0026thinsp;=\u0026thinsp;0.389). Racial differences revealed striking opposing patterns in Lactobacillus species distribution. \u003cem\u003eL. crispatus\u003c/em\u003e was detected exclusively in NH-White patients (6.4% vs 0% in NH-Black patients, FDR p\u0026thinsp;=\u0026thinsp;0.458), while \u003cem\u003eL. iners\u003c/em\u003e showed 2.27-fold higher prevalence in NH-Black patients (22.7% vs 10.0%, FDR p\u0026thinsp;=\u0026thinsp;0.295).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\u003ch2\u003eAbundance When Present\u003c/h2\u003e\u003cp\u003eAmong samples where \u003cem\u003eLactobacillus\u003c/em\u003e species were detected, abundance levels showed distinct patterns across clinical groups. When present, \u003cem\u003eL. crispatus\u003c/em\u003e maintained consistently high relative abundance regardless of clinical characteristics. In early-stage patients, \u003cem\u003eL. crispatus\u003c/em\u003e achieved a median relative abundance of approximately 95% when detected, with significantly higher abundance compared to late-stage patients (FDR p\u0026thinsp;=\u0026thinsp;0.033). Age-related differences in \u003cem\u003eL. crispatus\u003c/em\u003e abundance were minimal when the species was present (FDR p\u0026thinsp;=\u0026thinsp;0.125). In contrast, \u003cem\u003eL. iners\u003c/em\u003e demonstrated more variable abundance patterns when present. Abundance levels showed less consistent patterns across clinical groups, with median values ranging from 88\u0026ndash;92% depending on disease stage and patient characteristics (all FDR p\u0026thinsp;\u0026gt;\u0026thinsp;0.05). This variability may reflect the intermediate protective capacity of \u003cem\u003eL. iners\u003c/em\u003e compared to the more robustly protective \u003cem\u003eL. crispatus\u003c/em\u003e.\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis comprehensive analysis of 132 OC patients identified consistent vaginal microbiome dysbiosis characterized by \u003cem\u003eLactobacillus\u003c/em\u003e depletion and anaerobic bacterial overgrowth. We observed two clinically significant findings: age-related increases in anaerobic dysbiosis, with patients\u0026thinsp;\u0026ge;\u0026thinsp;50 years showing significant enrichment of \u003cem\u003ePeptoniphilus\u003c/em\u003e and \u003cem\u003eAnaerococcus\u003c/em\u003e (log₂ fold-change\u0026thinsp;=\u0026thinsp;1.31, FDR q\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and striking racial disparities including a 5-fold higher prevalence of Actinomycetaceae family bacteria in NH-Black compared to NH-White patients (40.9% vs 8.2%, FDR q\u0026thinsp;=\u0026thinsp;0.005). Species-specific \u003cem\u003eLactobacillus\u003c/em\u003e analyses revealed \u003cem\u003eL. crispatus\u003c/em\u003e detected exclusively in NH-White patients and \u003cem\u003eL. iners\u003c/em\u003e over 2-fold higher in NH-Black patients, alongside progressive loss of protective \u003cem\u003eL. crispatus\u003c/em\u003e with advancing age and disease stage. These patterns suggest that the vaginal microbiome may play an important role in driving OC prognosis and health disparities, warranting further investigation for microbiome-based interventions.\u003c/p\u003e\u003cp\u003eThese findings align with and extend established literature on vaginal microbiome alterations in gynecological malignancies. Previous OC studies by Jacobson et al. (2021) and Asangba et al. (2023) reported \u003cem\u003eLactobacillus\u003c/em\u003e depletion and anaerobic overgrowth, with only 24% of OC patients showing \u003cem\u003eLactobacillus\u003c/em\u003e-dominated microbiomes and significantly reduced abundance compared to controls (approximately 15% versus 30%)(\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e). Our cohort showed similar patterns with 47.7% \u003cem\u003eLactobacillus\u003c/em\u003e detection and 23.5% mean relative abundance, falling substantially below healthy reproductive-age women rates (70\u0026ndash;80% dominance)(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). However, our analysis also revealed age-specific patterns of dysbiosis, highlighting that the vaginal microbiome of older OC patients was significantly enriched with \u003cem\u003ePeptoniphilus\u003c/em\u003e and \u003cem\u003eAnaerococcus\u003c/em\u003e (log₂ fold-change\u0026thinsp;=\u0026thinsp;1.31, FDR p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), a notable finding that has not been previously reported to our knowledge. The racial differences we observed also extends prior work by Ravel et al., which established that healthy Black women are more likely to have non-\u003cem\u003eLactobacillus\u003c/em\u003e-dominated communities, with only 61.9% exhibiting \u003cem\u003eLactobacillus\u003c/em\u003e-dominated communities compared to 89.7% of White women(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). However, no known studies have specifically examined racial differences in vaginal microbiome composition among OC patients. We observed that Topic 5 (\u003cem\u003eLactobacillus\u003c/em\u003e-dominated) showed substantial depletion in NH-White compared to NH-Black OC patients (log₂ fold-change = -3.0, FDR q\u0026thinsp;=\u0026thinsp;0.09), suggesting higher overall \u003cem\u003eLactobacillus\u003c/em\u003e prevalence in NH-Black OC patients that is worth further exploration as a potential OC-related signature. However, species-level validation revealed that this pattern is driven mainly by the predominance of dysbiotic \u003cem\u003eL. iners\u003c/em\u003e, which was over 2-fold higher in NH-Black vs. NH-White patients (22.7% vs 10.0%), rather than the protective \u003cem\u003eL. crispatus\u003c/em\u003e that was detected exclusively in White patients (6.4% vs 0%). Additionally, our study is the first to document that NH-Black OC patients have a 5-fold higher Actinomycetaceae family prevalence vs. White patients, findings that can be further explored in larger cohort sizes.\u003c/p\u003e\u003cp\u003eThe age-related enrichment of \u003cem\u003ePeptoniphilus\u003c/em\u003e-dominated signatures in our study reveals potentially important biological pathways linking aging, microbiome dysbiosis, and cancer outcomes. OC patients\u0026thinsp;\u0026ge;\u0026thinsp;50 years showed significant enrichment of \u003cem\u003ePeptoniphilus\u003c/em\u003e and \u003cem\u003eAnaerococcus\u003c/em\u003e, bacteria that are associated with bacterial vaginosis and chronic wound infections, where they act synergistically with other microbes to impair normal healing processes and contribute to chronic inflammatory states(\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e). The loss of \u003cem\u003eL. crispatus\u003c/em\u003e, which maintains vaginal health through lactic acid production(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e), with advancing age (3.75-fold higher in younger patients) and disease stage (5.55-fold higher in early-stage disease) suggests that protective microbiome characteristics are associated with dysbiotic shifts as both aging and cancer progress, supporting the oncobiome concept where pathogen-driven disruptions create lasting alterations promoting chronic inflammation and carcinogenesis(\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). The absence of \u003cem\u003eL. crispatus\u003c/em\u003e in NH-Black patients combined with higher \u003cem\u003eL. iners\u003c/em\u003e prevalence represents a fundamental disparity in protective microbial capacity, as \u003cem\u003eL. crispatus\u003c/em\u003e produces predominantly D-lactic acid providing superior pathogen protection compared to L-lactate from \u003cem\u003eL. iners\u003c/em\u003e(\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e). \u003cem\u003eL. iners\u003c/em\u003e dominance is associated with intermediate vaginal health states and increased susceptibility to dysbiotic transitions(\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e), and recent mechanistic studies show that \u003cem\u003eL. iners\u003c/em\u003e actively promotes treatment resistance through metabolic reprogramming(\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e), potentially explaining some of the documented treatment response disparities between racial groups. The combination of higher Actinomycetaceae prevalence and differential \u003cem\u003eLactobacillus\u003c/em\u003e species composition may also create a dual biological disadvantage for NH-Black patients that may contribute to worse OC prognosis. Actinomycetaceae have been associated with pelvic inflammatory disease and chronic genital inflammation(\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e), and recent work demonstrates that certain Actinobacteria species can metabolize estrogen and steroid hormones(\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e), potentially altering the local hormonal microenvironment in ways that could influence cancer biology.\u003c/p\u003e\u003cp\u003eTaken together, our findings highlight a potentially important role for vaginal microbiome patterns in driving OC prognosis, an area that deserves further study. Importantly, given that Black women experience later-stage diagnosis(\u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e), more aggressive tumor biology(\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e), and reduced survival rates(\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e), differential microbiome patterns may be a potential biological mechanism that drives these disparities. Efforts to more comprehensively characterize the microbiome and identify targetable pathways may provide valuable intervention opportunities. While there is currently insufficient evidence to support routine probiotic recommendations for OC patients(\u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e), additional studies building on our findings can provide important insights to inform species-specific rather than generic interventions to alter the microbiome.\u003c/p\u003e\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e\u003ch2\u003eLimitations\u003c/h2\u003e\u003cp\u003eSeveral important limitations should be considered. The cross-sectional design prevents determination of causality or temporal relationships between microbiome changes and disease characteristics. We cannot distinguish whether observed patterns reflect cancer-induced changes, treatment effects, aging-related changes independent of cancer, or pre-existing differences influencing cancer outcomes, although the majority of patients in this cohort had received treatment.\u003c/p\u003e\u003cp\u003eThe sample size was relatively small for detecting subtle associations, particularly for NH-Black patients (n\u0026thinsp;=\u0026thinsp;22), limiting statistical power for subgroup analyses. Our 16S rRNA sequencing approach provides taxonomic composition but limited functional insight into microbiome activity. Methodological considerations such as the exclusion of patients with recent antibiotic use, and focusing on patients at least nine months post-diagnosis, potentially resulted in selecting for more stable microbiome profiles, and missing acute changes associated with initial diagnosis and treatment. The inclusion of patients across different treatment stages likely introduced heterogeneity that could have masked some associations.\u003c/p\u003e\u003cp\u003eNevertheless, our study has several notable strengths that support the validity of our findings. The population-based recruitment from multiple state cancer registries enhances generalizability, while the multi-scale analytical approach combining traditional composition analysis, topic modeling, and species-level validation provided convergent evidence for our key findings. The self-collection methodology improved feasibility and reduced selection bias, and our rigorous exclusion criteria for antibiotic and suppository use minimized confounding factors.\u003c/p\u003e\u003c/div\u003e"},{"header":"Conclusions","content":"\u003cp\u003eWe identified consistent vaginal microbiome dysbiosis among OC patients with notable age- and race-associated patterns that may have important implications for prognosis. These findings establish a strong foundation for future research investigating specific vaginal microbiome patterns beyond the species level in understanding and addressing OC progression and outcome disparities. If validated in larger longitudinal studies, vaginal microbiome profiles could serve as biomarkers for disease monitoring or therapeutic targets to improve outcomes, particularly for older patients and NH-Black patients who demonstrate the most concerning microbiome signatures. The integration of microbiome assessment into clinical care protocols may ultimately contribute to more personalized and effective OC management strategies, though species-specific interventions will be essential for addressing the biological disadvantages revealed by our analysis.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was funded by the National Institutes of Health/National Cancer Institute (Grant Numbers R37CA233777 and K01CA255408)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInstitutional Review Board Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board (or Ethics Committee) of Duke University (Pro00101872).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInformed Consent Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by Duke University and the participating registries\u0026rsquo; Institutional Review Boards (Pro00101872). All participants included provided informed consent.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors thank all the study participants who participated in the ORCHiD study for their vital contribution in advancing the science of cancer in the United States. Texas Cancer Registry (TCR), Texas Department of State Health Services; Maryland Cancer Registry (MCR), Maryland Department of Health; California Cancer Registry (CCR), California Department of Public Health; New York State Cancer Registry (NYSCR), New York State Department of Health; Kentucky Cancer Registry (KCR); Emory University Rollins School of Public Health: Department of Epidemiology, North Carolina Cancer Registry (NCCR). The findings and conclusions in this publication are those of the author(s) and do not necessarily represent the views of the above-mentioned cancer registries. The authors also thank Duke School of Medicine for use of Duke Microbiome Core Facility and Sequencing and Genomic Technologies shared resource, which provided DNA extraction, library preparation, and sequencing services. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDisclosures\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eKentucky cancer data have been provided by the Kentucky Cancer Registry, 2365 Harrodsburg Road, Lexington, KY 40504 (www.kcr.uky.edu). Data from the Kentucky Cancer Registry is supported by the following: Cooperative Agreement #NU58DP007144 from the Centers for Disease Control and Prevention and Contract #HHSN261201800013I from the National Cancer Institute\u0026rsquo;s Surveillance, Epidemiology, and End Results (SEER) Program. The ideas and opinions expressed herein are those of the author(s) and do not necessarily reflect the opinions of the National Cancer Institute, the Centers for Disease Control and Prevention or their Contractors and Subcontractors, and the Commonwealth of Kentucky.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe collection of cancer incidence data in Georgia was supported by contract HHSN261201800003I, Task Order HHSN26100001 from the NCI and cooperative agreement 6NU58DP006352-05-01 from the CDC.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis work was supported in part by the Centers for Disease Control and Prevention\u0026rsquo;s National Program of Cancer Registries through cooperative agreement NU58DP007218 awarded to the New York State Department of Health and by Contract HHSN261201800005I (Task Order HHSN26100001) from the National Cancer Institute, National Institutes of Health.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eCancer data was used by the Maryland Cancer Registry, Center for Cancer Prevention and Control, Maryland Department of Health, with funding from the State of Maryland and the Maryland Cigarette Restitution Fund to provide assistance in patient recruitment. The collection and availability of cancer registry data is also supported by the Cooperative Agreement NU58DP006333, funded by the Centers for Disease Control and Prevention. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control and Prevention or the Department of Health and Human Services.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe collection of cancer incidence data used in this study was supported by the California Department of Public Health pursuant to California Health and Safety Code Section 103885; Centers for Disease Control and Prevention\u0026rsquo;s (CDC) National Program of Cancer Registries, under cooperative agreement 5NU58DP006344; the National Cancer Institute\u0026rsquo;s Surveillance, Epidemiology and End Results Program under contract HHSN261201800032I awarded to the University of California, San Francisco, contract HHSN261201800015I awarded to the University of Southern California, and contract HHSN261201800009I awarded to the Public Health Institute. The ideas and opinions expressed herein are those of the author(s) and do not necessarily reflect the opinions of the State of California, Department of Public Health, the National Cancer Institute, and the Centers for Disease Control and Prevention or their Contractors and Subcontractors.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTexas cancer data have been provided by the Texas Cancer Registry, Cancer Epidemiology and Surveillance Branch, Texas Department of State Health Services, 1100 West 49th Street, Austin, TX 78756 (www.dshs.texas.gov/tcr). Data from the Texas Cancer Registry is supported by the following: Cooperative Agreement #1NU58DP007140 from the Centers for Disease Control and Prevention, Contract #75N91021D00011 from the National Cancer Institute\u0026rsquo;s Surveillance, Epidemiology, and End Results (SEER) Program, and the Cancer Prevention and Research Institute of Texas.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe findings and conclusions in this publication are those of the author(s) and do not necessarily represent the views of the North Carolina Department of Health and Human Services, Division of Public Health.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eChen X, Lu Y, Chen T, Li R. The Female Vaginal Microbiome in Health and Bacterial Vaginosis. Front Cell Infect Microbiol. 2021;11:631972.\u003c/li\u003e\n\u003cli\u003eMiller EA, Beasley DE, Dunn RR, Archie EA. Lactobacilli Dominance and Vaginal pH: Why Is the Human Vaginal Microbiome Unique? Front Microbiol. 2016;7:1936.\u003c/li\u003e\n\u003cli\u003eNene NR, Reisel D, Leimbach A, Franchi D, Jones A, Evans I, et al. Association between the cervicovaginal microbiome, BRCA1 mutation status, and risk of ovarian cancer: a case-control study. Lancet Oncol. 2019;20(8):1171-82.\u003c/li\u003e\n\u003cli\u003eSaraf VS, Sheikh SA, Ahmad A, Gillevet PM, Bokhari H, Javed S. Vaginal microbiome: normalcy vs dysbiosis. Arch Microbiol. 2021;203(7):3793-802.\u003c/li\u003e\n\u003cli\u003eLaniewski P, Ilhan ZE, Herbst-Kralovetz MM. The microbiome and gynaecological cancer development, prevention and therapy. Nat Rev Urol. 2020;17(4):232-50.\u003c/li\u003e\n\u003cli\u003eLi W, Ma ZS. Dominance network analysis of the healthy human vaginal microbiome not dominated by Lactobacillus species. Comput Struct Biotechnol J. 2020;18:3447-56.\u003c/li\u003e\n\u003cli\u003eHakimjavadi H, George SH, Taub M, Dodds LV, Sanchez-Covarrubias AP, Huang M, et al. The vaginal microbiome is associated with endometrial cancer grade and histology. Cancer Res Commun. 2022;2(6):447-55.\u003c/li\u003e\n\u003cli\u003eSiegel RL, Miller KD, Fuchs HE, Jemal A. Cancer statistics, 2022. CA Cancer J Clin. 2022;72(1):7-33.\u003c/li\u003e\n\u003cli\u003eZhou B, Sun C, Huang J, Xia M, Guo E, Li N, et al. The biodiversity Composition of Microbiome in Ovarian Carcinoma Patients. Sci Rep. 2019;9(1):1691.\u003c/li\u003e\n\u003cli\u003eMorikawa A, Kawabata A, Shirahige K, Akiyama T, Okamoto A, Sutani T. Altered cervicovaginal microbiota in premenopausal ovarian cancer patients. Gene. 2022;811:146083.\u003c/li\u003e\n\u003cli\u003eSipos A, Ujlaki G, Miko E, Maka E, Szabo J, Uray K, et al. The role of the microbiome in ovarian cancer: mechanistic insights into oncobiosis and to bacterial metabolite signaling. Mol Med. 2021;27(1):33.\u003c/li\u003e\n\u003cli\u003eMehra Y, Chalif J, Mensah-Bonsu C, Spakowicz D, O\u0026apos;Malley DM, Chambers L. The microbiome and ovarian cancer: insights, implications, and therapeutic opportunities. J Cancer Metas Treat. 2023;9.\u003c/li\u003e\n\u003cli\u003eChan JK, Kapp DS, Shin JY, Husain A, Teng NN, Berek JS, et al. Influence of the gynecologic oncologist on the survival of ovarian cancer patients. Obstet Gynecol. 2007;109(6):1342-50.\u003c/li\u003e\n\u003cli\u003eSurveillance E, and End Results Program. Cancer Stat Facts: Ovarian Cancer [Available from: https://seer.cancer.gov/statfacts/html/ovary.html.\u003c/li\u003e\n\u003cli\u003eBrotman RM, Shardell MD, Gajer P, Fadrosh D, Chang K, Silver MI, et al. Association between the vaginal microbiota, menopause status, and signs of vulvovaginal atrophy. Menopause. 2014;21(5):450-8.\u003c/li\u003e\n\u003cli\u003eRavel J, Gajer P, Abdo Z, Schneider GM, Koenig SS, McCulle SL, et al. Vaginal microbiome of reproductive-age women. Proc Natl Acad Sci U S A. 2011;108 Suppl 1(Suppl 1):4680-7.\u003c/li\u003e\n\u003cli\u003eFettweis JM, Serrano MG, Brooks JP, Edwards DJ, Girerd PH, Parikh HI, et al. The vaginal microbiome and preterm birth. Nat Med. 2019;25(6):1012-21.\u003c/li\u003e\n\u003cli\u003eSun S, Serrano MG, Fettweis JM, Basta P, Rosen E, Ludwig K, et al. Race, the Vaginal Microbiome, and Spontaneous Preterm Birth. mSystems. 2022;7(3):e0001722.\u003c/li\u003e\n\u003cli\u003eFettweis JM, Brooks JP, Serrano MG, Sheth NU, Girerd PH, Edwards DJ, et al. Differences in vaginal microbiome in African American women versus women of European ancestry. Microbiology (Reading). 2014;160(Pt 10):2272-82.\u003c/li\u003e\n\u003cli\u003eZhou X, Brown CJ, Abdo Z, Davis CC, Hansmann MA, Joyce P, et al. Differences in the composition of vaginal microbial communities found in healthy Caucasian and black women. ISME J. 2007;1(2):121-33.\u003c/li\u003e\n\u003cli\u003eLi C, Feng Y, Yang C, Wang D, Zhang D, Luo X, et al. Association between vaginal microbiota and the progression of ovarian cancer. J Med Virol. 2023;95(7):e28898.\u003c/li\u003e\n\u003cli\u003eXu J, Peng JJ, Yang W, Fu K, Zhang Y. Vaginal microbiomes and ovarian cancer: a review. Am J Cancer Res. 2020;10(3):743-56.\u003c/li\u003e\n\u003cli\u003eStewart SL, Rim SH, Richards TB. Gynecologic oncologists and ovarian cancer treatment: avenues for improved survival. J Womens Health (Larchmt). 2011;20(9):1257-60.\u003c/li\u003e\n\u003cli\u003eTerplan M, Schluterman N, McNamara EJ, Tracy JK, Temkin SM. Have racial disparities in ovarian cancer increased over time? An analysis of SEER data. Gynecol Oncol. 2012;125(1):19-24.\u003c/li\u003e\n\u003cli\u003eBorgdorff H, van der Veer C, van Houdt R, Alberts CJ, de Vries HJ, Bruisten SM, et al. The association between ethnicity and vaginal microbiota composition in Amsterdam, the Netherlands. PLoS One. 2017;12(7):e0181135.\u003c/li\u003e\n\u003cli\u003eBanerjee S, Tian T, Wei Z, Shih N, Feldman MD, Alwine JC, et al. The ovarian cancer oncobiome. Oncotarget. 2017;8(22):36225-45.\u003c/li\u003e\n\u003cli\u003eJacobson D, Moore K, Gunderson C, Rowland M, Austin R, Honap TP, et al. Shifts in gut and vaginal microbiomes are associated with cancer recurrence time in women with ovarian cancer. PeerJ. 2021;9:e11574.\u003c/li\u003e\n\u003cli\u003eYu B, Liu C, Proll SC, Manhardt E, Liang S, Srinivasan S, et al. Identification of fallopian tube microbiota and its association with ovarian cancer. Elife. 2024;12.\u003c/li\u003e\n\u003cli\u003eSchloss PD. Identifying and Overcoming Threats to Reproducibility, Replicability, Robustness, and Generalizability in Microbiome Research. mBio. 2018;9(3).\u003c/li\u003e\n\u003cli\u003eWeiss S, Xu ZZ, Peddada S, Amir A, Bittinger K, Gonzalez A, et al. Normalization and microbial differential abundance strategies depend upon data characteristics. Microbiome. 2017;5(1):27.\u003c/li\u003e\n\u003cli\u003eNearing JT, Douglas GM, Hayes MG, MacDonald J, Desai DK, Allward N, et al. Microbiome differential abundance methods produce different results across 38 datasets. Nat Commun. 2022;13(1):342.\u003c/li\u003e\n\u003cli\u003eAkinyemiju T, Deveaux A, Wilson L, Gupta A, Joshi A, Bevel M, et al. Ovarian Cancer Epidemiology, Healthcare Access and Disparities (ORCHiD): methodology for a population-based study of black, Hispanic and white patients with ovarian cancer. BMJ Open. 2021;11(10):e052808.\u003c/li\u003e\n\u003cli\u003eCaporaso JG, Ackermann G, Apprill A, Bauer M, Berg-Lyons D, Betley J, et al. EMP 16S Illumina amplicon protocol. See http://www earthmicrobiome org/protocols-and-standards/16s. 2018.\u003c/li\u003e\n\u003cli\u003eTeam RC. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2024.\u003c/li\u003e\n\u003cli\u003eCallahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: high-resolution sample inference from Illumina amplicon data. Nature methods. 2016;13(7):581-3.\u003c/li\u003e\n\u003cli\u003eYilmaz P, Parfrey LW, Yarza P, Gerken J, Pruesse E, Quast C, et al. The SILVA and \u0026ldquo;All-species Living Tree Project (LTP)\u0026rdquo; taxonomic frameworks. Nucleic Acids Research. 2014;42(D1):D643-D8.\u003c/li\u003e\n\u003cli\u003eQuast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic acids research. 2012;41(D1):D590-D6.\u003c/li\u003e\n\u003cli\u003eStamatakis A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics. 2014;30(9):1312-3.\u003c/li\u003e\n\u003cli\u003eShrode RL, Ollberding NJ, Mangalam AK. Looking at the Full Picture: Utilizing Topic Modeling to Determine Disease-Associated Microbiome Communities. bioRxiv. 2023.\u003c/li\u003e\n\u003cli\u003eBrooks JP, Buck GA, Chen G, Diao L, Edwards DJ, Fettweis JM, et al. Changes in vaginal community state types reflect major shifts in the microbiome. Microb Ecol Health Dis. 2017;28(1):1303265.\u003c/li\u003e\n\u003cli\u003eNikita \u0026Mu; CN. ldatuning: Tuning of the Latent Dirichlet Allocation Models Parameters. 2020.\u003c/li\u003e\n\u003cli\u003eArun R, Suresh V, Madhavan CEV, Murty MN. On Finding the Natural Number of Topics with Latent Dirichlet Allocation: Some Observations. Lect Notes Artif Int. 2010;6118:391-402.\u003c/li\u003e\n\u003cli\u003eCao J, Xia T, Li JT, Zhang YD, Tang S. A density-based method for adaptive LDA model selection. Neurocomputing. 2009;72(7-9):1775-81.\u003c/li\u003e\n\u003cli\u003eGriffiths TL, Steyvers M. Finding scientific topics. P Natl Acad Sci USA. 2004;101:5228-35.\u003c/li\u003e\n\u003cli\u003eDeveaud R, SanJuan E, Bellot P. Accurate and effective Latent Concept Modeling for ad hoc information retrieval. Document Numerique. 2014;17(1):61-84.\u003c/li\u003e\n\u003cli\u003eGr\u0026uuml;n B, Hornik K. Topicmodels: An R Package for Fitting Topic Models. J Stat Softw. 2011;40(13):1-30.\u003c/li\u003e\n\u003cli\u003eRobinson S, Silge J. tidytext: Text Mining and Analysis Using Tidy Data Principles in R. JOSS. 2016;1(3).\u003c/li\u003e\n\u003cli\u003eZhou H, He K, Chen J, Zhang X. LinDA: linear models for differential abundance analysis of microbiome compositional data. Genome Biol. 2022;23(1):95.\u003c/li\u003e\n\u003cli\u003eKassambara A. rstatix: Pipe-friendly framework for basic statistical tests. CRAN: Contributed Packages. 2019.\u003c/li\u003e\n\u003cli\u003eMcMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PloS one. 2013;8(4):e61217.\u003c/li\u003e\n\u003cli\u003eWickham H. ggplot2: Elegant Graphics for Data Analysis. Use R. 2009:1-212.\u003c/li\u003e\n\u003cli\u003eWilke C. cowplot: Streamlined Plot Theme and Plot Annotations for \u0026lsquo;ggplot2\u0026rsquo;. 1.2.0 ed2025.\u003c/li\u003e\n\u003cli\u003eSankaran K, Holmes SP. Latent variable modeling for the microbiome. Biostatistics. 2019;20(4):599-614.\u003c/li\u003e\n\u003cli\u003eSchaffer JN, Pearson MM. Proteus mirabilis and Urinary Tract Infections. Microbiol Spectr. 2015;3(5).\u003c/li\u003e\n\u003cli\u003eDiop K, Diop A, Michelle C, Richez M, Rathored J, Bretelle F, et al. Description of three new Peptoniphilus species cultured in the vaginal fluid of a woman diagnosed with bacterial vaginosis: Peptoniphilus pacaensis sp. nov., Peptoniphilus raoultii sp. nov., and Peptoniphilus vaginalis sp. nov. Microbiologyopen. 2019;8(3):e00661.\u003c/li\u003e\n\u003cli\u003eMurphy EC, Frick IM. Gram-positive anaerobic cocci--commensals and opportunistic pathogens. FEMS Microbiol Rev. 2013;37(4):520-53.\u003c/li\u003e\n\u003cli\u003eAsangba AE, Chen J, Goergen KM, Larson MC, Oberg AL, Casarin J, et al. Diagnostic and prognostic potential of the microbiome in ovarian cancer treatment response. Sci Rep. 2023;13(1):730.\u003c/li\u003e\n\u003cli\u003eWitkin SS, Mendes-Soares H, Linhares IM, Jayaram A, Ledger WJ, Forney LJ. Influence of vaginal bacteria and D- and L-lactic acid isomers on vaginal extracellular matrix metalloproteinase inducer: implications for protection against upper genital tract infections. mBio. 2013;4(4).\u003c/li\u003e\n\u003cli\u003eFrance MT, Mendes-Soares H, Forney LJ. Genomic Comparisons of Lactobacillus crispatus and Lactobacillus iners Reveal Potential Ecological Drivers of Community Composition in the Vagina. Appl Environ Microbiol. 2016;82(24):7063-73.\u003c/li\u003e\n\u003cli\u003eZheng N, Guo R, Wang J, Zhou W, Ling Z. Contribution of Lactobacillus iners to Vaginal Health and Diseases: A Systematic Review. Front Cell Infect Microbiol. 2021;11:792787.\u003c/li\u003e\n\u003cli\u003eColbert LE, El Alam MB, Wang R, Karpinets T, Lo D, Lynn EJ, et al. Tumor-resident Lactobacillus iners confer chemoradiation resistance through lactate-induced metabolic rewiring. Cancer Cell. 2023;41(11):1945-62 e11.\u003c/li\u003e\n\u003cli\u003eFerjaoui MA, Arfaoui R, Khedhri S, Hannechi MA, Abdessamia K, Samaali K, et al. Pelvic actinomycosis: A confusing diagnosis. Int J Surg Case Rep. 2021;86:106387.\u003c/li\u003e\n\u003cli\u003eGarcia-Gomez E, Gonzalez-Pedrajo B, Camacho-Arroyo I. Role of sex steroid hormones in bacterial-host interactions. Biomed Res Int. 2013;2013:928290.\u003c/li\u003e\n\u003cli\u003eMei S, Chelmow D, Gecsi K, Barkley J, Barrows E, Brooks R, et al. Health Disparities in Ovarian Cancer: Report From the Ovarian Cancer Evidence Review Conference. Obstet Gynecol. 2023;142(1):196-210.\u003c/li\u003e\n\u003cli\u003eSrivastava SK, Ahmad A, Miree O, Patel GK, Singh S, Rocconi RP, et al. Racial health disparities in ovarian cancer: not just black and white. J Ovarian Res. 2017;10(1):58.\u003c/li\u003e\n\u003cli\u003eMitra A, Gultekin M, Burney Ellis L, Bizzarri N, Bowden S, Taumberger N, et al. Genital tract microbiota composition profiles and use of prebiotics and probiotics in gynaecological cancer prevention: review of the current evidence, the European Society of Gynaecological Oncology prevention committee statement. Lancet Microbe. 2024;5(3):e291-e300. \u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Table","content":"\u003cp\u003eTable 1 is available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"microbiome, ovarian cancer, ORCHiD, 16S rRNA sequencing, topic modeling, cancer disparities, Vaginal microbiome, topic modeling, health disparities, Lactobacillus","lastPublishedDoi":"10.21203/rs.3.rs-7428423/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7428423/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eVaginal dysbiosis may contribute to ovarian cancer (OC) outcomes, but comprehensive microbiome characterization remains limited. Here, we characterize the prevalence and predictors of vaginal microbiome dysbiosis among diverse OC patients in the US.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eWe performed 16S rRNA gene sequencing on vaginal samples from 132 OC patients recruited as part of the population-based ORCHiD study. We applied topic modeling using Latent Dirichlet Allocation (LDA), a computational approach for identifying latent microbial community patterns, to identify distinct microbial signatures representing co-occurring bacterial taxa.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eWidespread dysbiosis was observed with \u003cem\u003eLactobacillus\u003c/em\u003e detected in only 47.7% of patients. Topic modeling identified seven distinct microbial signatures from \u003cem\u003eLactobacillus\u003c/em\u003e-dominated to pathogenic communities. Patients\u0026thinsp;\u0026ge;\u0026thinsp;50 years showed significant anaerobic bacterial enrichment (log₂FC\u0026thinsp;=\u0026thinsp;1.31, FDR q\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Striking racial disparities emerged: Non Hispanic-Black patients had 5-fold higher Actinomycetaceae prevalence (40.9% vs 8.2%, FDR q\u0026thinsp;=\u0026thinsp;0.005), while protective \u003cem\u003eL. crispatus\u003c/em\u003e was detected exclusively in Non Hispanic-White patients (6.4% vs 0%).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e\u003cp\u003eThis study revealed widespread vaginal microbiome dysbiosis among OC patients with clinically significant age and racial patterns that may contribute to outcome disparities.\u003c/p\u003e","manuscriptTitle":"Vaginal Microbiome Composition, Diversity and Dysbiosis: The ORCHiD Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-11 10:56:12","doi":"10.21203/rs.3.rs-7428423/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"communications-medicine","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"commsmed","sideBox":"Learn more about [Communications Medicine](http://www.nature.com/commsmed)","snPcode":"43856","submissionUrl":"https://mts-commsmed.nature.com/cgi-bin/main.plex","title":"Communications Medicine","twitterHandle":"@commsmedicine","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Communications Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ec3cf025-62ea-4834-acce-b8bdc081bd21","owner":[],"postedDate":"September 11th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":53736235,"name":"Biological sciences/Microbiology/Clinical microbiology"},{"id":53736236,"name":"Health sciences/Oncology/Cancer/Gynaecological cancer/Ovarian cancer"}],"tags":[],"updatedAt":"2025-09-11T10:56:12+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-11 10:56:12","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7428423","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7428423","identity":"rs-7428423","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}