Causal effects of oral microbiome traits on female reproductive diseases: a two-sample Mendelian randomization study.

The MR analysis followed the Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomization (STROBE-MR) guidelines [ 25 ]. Our MR analysis was conducted based on three core instrumental variable (IV) assumptions [ 26 ]: (1) relevance, the genetic instruments are robustly associated with the exposure; (2) independence, the instruments are not associated with confounders of the exposure–outcome relationship; and (3) exclusion restriction, the instruments influence the outcome only through the exposure and not via alternative pathways (i.e., no horizontal pleiotropy). A study framework diagram is provided in Fig. 1 to depict the design of our study. As all GWAS summary statistics used in this study were publicly available and de-identified, no additional ethical approvals were required. Fig. 1 Overview of Mendelian randomization Overview of Mendelian randomization The outcome GWAS summary dataset for uterine leiomyoma, polycystic ovary syndrome, endometriosis, and tubal infertility were obtained from FinnGen (Table 1 ). Exposure GWAS summary statistics for oral microbiome traits were obtained from Stankevic et al., who conducted GWAS of salivary microbiota in 610 unrelated individuals from the Danish ADDITION-PRO cohort using 16 S rRNA gene amplicon sequencing (Table 1 ). The exposure dataset comprised 44 microbial traits (including taxon abundance traits across different taxonomic levels and one beta-diversity trait based on Bray–Curtis dissimilarity), with association models adjusted for key demographic, lifestyle, and technical covariates as described in the original study. All data used were publicly available, de-identified, and derived from studies with appropriate ethical approval and participant consent [ 27 ]. Table 1 The GWAS data for exposure and outcomes Trait GWAS ID Population case/control Decent Leiomyoma of uterus CD2_BENIGN_LEIOMYOMA_UTERI 42,107/239,957 European Polycystic ovarian syndrome E4_PCOS 2,214/267,780 European Endometriosis N14_ENDOMETRIOSIS 20,190/130,160 European Female infertility, tubal origin N14_FITUB 2,038/130,160 European Habitual aborter N14_HABITABORT 811/130,160 European Spontaneous abortion O15_ABORT_SPONTAN 23,167/199,279 European Saliva microbiota abundance (Phylum Firmicutes) GCST90429799 610 European Saliva microbiota abundance (Phylum Proteobacteria) GCST90429800 610 European Saliva microbiota abundance (Class Bacilli ) GCST90429801 610 European Saliva microbiota abundance (Order Bacteroidales) GCST90429802 610 European Saliva microbiota abundance (Order Fusobacteriales) GCST90429803 610 European Saliva microbiota abundance (Order Actinomycetales) GCST90429804 610 European Saliva microbiota abundance (Order Clostridiales) GCST90429805 610 European Saliva microbiota abundance (Family Veillonellaceae) GCST90429806 610 European Saliva microbiota abundance (Family Pasteurellaceae) GCST90429807 610 European Saliva microbiota abundance (Family Prevotellaceae) GCST90429808 610 European Saliva microbiota abundance (Family Actinomycetaceae) GCST90429809 610 European Saliva microbiota abundance (Family Lachnospiraceae_[XIV]) GCST90429810 610 European Saliva microbiota abundance (Genus Veillonella ) GCST90429811 610 European Saliva microbiota abundance (Genus Haemophilus ) GCST90429812 610 European Saliva microbiota abundance (Genus Streptococcus ) GCST90429813 610 European Saliva microbiota abundance (Genus Neisseria ) GCST90429814 610 European Saliva microbiota abundance (Genus Prevotella ) GCST90429815 610 European Saliva microbiota abundance (Genus Porphyromonas ) GCST90429816 610 European Saliva microbiota abundance (Genus Fusobacterium ) GCST90429817 610 European Saliva microbiota abundance (Genus Rothia ) GCST90429818 610 European Saliva microbiota abundance (Genus Schaalia ) GCST90429819 610 European Saliva microbiota abundance (Genus Granulicatella ) GCST90429820 610 European Saliva microbiota abundance (Genus Leptotrichia ) GCST90429821 610 European Saliva microbiota abundance (Genus Alloprevotella ) GCST90429822 610 European Saliva microbiota abundance (unknown Veillonella species (ASV0001)) GCST90429823 610 European Saliva microbiota abundance (Species parainfluenzae) GCST90429824 610 European Saliva microbiota abundance (unknown Streptococcus species (ASV0003)) GCST90429825 610 European Saliva microbiota abundance (unknown Neisseria species (ASV0004)) GCST90429826 610 European Saliva microbiota abundance (Species histicola) GCST90429827 610 European Saliva microbiota abundance (unknown Streptococcus species (ASV0006)) GCST90429828 610 European Saliva microbiota abundance (Species parvula) GCST90429829 610 European Saliva microbiota abundance (unknown Porphyromonas species (ASV0008)) GCST90429830 610 European Saliva microbiota abundance (unknown Streptococcus species (ASV0009)) GCST90429831 610 European Saliva microbiota abundance (Species periodonticum) GCST90429832 610 European Saliva microbiota abundance (Species dispar) GCST90429833 610 European Saliva microbiota abundance (unknown Rothia species (ASV0012)) GCST90429834 610 European Saliva microbiota abundance (Species micronuciformis) GCST90429835 610 European Saliva microbiota abundance (Species pallens) GCST90429836 610 European Saliva microbiota abundance ( Rothia mucilaginosa ) GCST90429837 610 European Saliva microbiota abundance (unknown Rothia species (ASV0016)) GCST90429838 610 European Saliva microbiota abundance (unknown Schaalia species (ASV0017)) GCST90429839 610 European Saliva microbiota abundance (Species rogosae) GCST90429840 610 European Saliva microbiota abundance (unknown Gemella) GCST90429841 610 European Beta diversity of salivary microbiota GCST90429842 610 European The GWAS data for exposure and outcomes Single nucleotide polymorphisms (SNPs) associated with each oral microbiome trait were used as instrumental variables (IVs) for Mendelian randomization (MR). SNPs were selected at a significance threshold of P < 5 × 10⁻⁶ to ensure sufficient instrument availability given the modest sample size of the exposure GWAS, a strategy commonly adopted in microbiome MR studies [ 28 ]. We further filtered SNPs with minor allele frequency (MAF) > 0.01. To obtain independent instruments, linkage disequilibrium (LD) clumping was performed using r² 0.8) was used [ 29 ]. Instrument strength was evaluated using the F-statistic to reduce weak instrument bias. For each SNP, the proportion of exposure variance explained (R²) was calculated as: R² = 2 × EAF × (1 − EAF) × β², where EAF is the effect allele frequency and β is the SNP–exposure effect estimate. The F-statistic was then calculated as: F = R² × ( N − 2) / (1 − R²), where N denotes the sample size of the exposure GWAS ( N = 610). SNPs with F ≤ 10 were considered weak instruments and were excluded [ 30 ]. To further minimize confounding, all candidate instruments were queried in the GWAS Catalog to identify SNPs associated with potential confounders or the outcomes themselves. Confounders of interest included body mass index/obesity, smoking, alcohol consumption, glycaemic traits/type 2 diabetes, and inflammatory markers (e.g., C-reactive protein). MR analyses were repeated after excluding such SNPs. Inverse-variance weighted (IVW) MR was used as the primary method to estimate causal effects [ 31 ]. Given the potential for heterogeneity across SNP-specific Wald ratios, we used a multiplicative random-effects IVW model as appropriate. Complementary methods (MR-Egger, weighted median, and weighted mode) were used for robustness [ 32 ]. To account for multiple comparisons across the set of tested exposure–outcome pairs, Benjamini–Hochberg FDR correction was applied to IVW results; associations with q < 0.05 were considered statistically significant, and P < 0.05 was considered nominal. Statistical power for the nominally significant MR estimates was calculated using the mRnd online calculator ( https://shiny.cnsgenomics.com/mRnd/ ), incorporating the sample size of the outcome GWAS, the proportion of variance in the exposure explained by the IVs (R2), and the observed odds ratio, with a Type-I error rate (α) set at 0.05 [ 33 ]. Sensitivity analyses were performed to assess heterogeneity and horizontal pleiotropy. Cochran’s Q statistic was used to evaluate heterogeneity [ 34 ], and the MR-Egger intercept test was used to detect directional pleiotropy [ 35 ]. MR-PRESSO was applied to identify potential outlier SNPs and provide outlier-corrected estimates [ 36 ]. Steiger directionality tests were conducted to assess whether the instruments explained more variance in the exposure than in the outcome, supporting the assumed causal direction [ 37 ].

In total, GWAS summary statistics were available for 44 oral microbiome traits. After instrument selection and harmonization with outcome data, not all traits yielded eligible instruments for every outcome; therefore, the primary IVW analyses included 246 analyzable exposure–outcome pairs. Instrument strength metrics indicated generally adequate instrument strength (F-statistics > 10; Tables S1–S2). In IVW analyses, nominal associations were observed between class Bacilli and uterine leiomyoma (OR = 1.030, 95%CI: 1.001–1.060, P  = 0.041), as well as between genus Veillonella and uterine leiomyoma (OR = 1.029, 95% CI: 1.007–1.051, P  = 0.008). Conversely, nominal protective associations against tubal infertility were observed for family Veillonellaceae (OR = 0.864, 95% CI: 0.782–0.954, P  = 0.004) and genus Veillonella (OR = 0.890, 95% CI: 0.816–0.969, P  = 0.008) (Table  2 ). The individual SNP effects and pooled causal estimates for these nominal associations are visualized in scatter plots (Fig.  2 A-D) and forest plots (Fig.  3 A-D). However, after Benjamini–Hochberg FDR correction for the primary IVW analyses, none of the associations met the q < 0.05 threshold, and thus these findings should be interpreted as suggestive rather than definitive (Table  2 ). Table 2 Relationship between oral microbiota and FRDs (nominal IVW results; BH-FDR q values for primary IVW analyses) Exposure Outcome N .SNP Method OR (95% CI) P BH-FDR q Saliva microbiota abundance (Class Bacilli ) Leiomyoma of uterus 4 Inverse variance weighted 1.0303 (1.0012–1.0602) 0.0413 0.75645 Saliva microbiota abundance (Family Veillonellaceae) Female infertility, tubal origin 4 Inverse variance weighted 0.864 (0.7824–0.9541) 0.0039 0.15990 Saliva microbiota abundance (Genus Veillonella ) Female infertility, tubal origin 6 Inverse variance weighted 0.89 (0.8167–0.9699) 0.0079 0.16195 Saliva microbiota abundance (Genus Veillonella ) Leiomyoma of uterus 6 Inverse variance weighted 1.0291 (1.0075–1.0512) 0.0081 0.33210 Fig. 2 Scatter plots showing the causal associations between oral microbiota and female reproductive diseases. A , B : The plots of class Bacilli and genus Veillonella on uterine leiomyoma; ( C , D ): The plots of family Veillonellaceae and genus Veillonella on female infertility of tubal origin; ( E ): The plot of Rothia mucilaginosa on uterine leiomyoma. Each point represents a Single nucleotide polymorphism (SNP) serving as an instrumental variable for the exposure variable. The regression line’s slope, derived via inverse variance weighting and MR-Egger methods, denotes the estimated causal effect. The horizontal and vertical axes correspond respectively to the genetic association between the exposure variable (oral microbiota abundance) and the outcome variable (risk of female reproductive disorders). The shaded regions denote the 95% confidence intervals for the regression lines Relationship between oral microbiota and FRDs (nominal IVW results; BH-FDR q values for primary IVW analyses) Scatter plots showing the causal associations between oral microbiota and female reproductive diseases. A , B : The plots of class Bacilli and genus Veillonella on uterine leiomyoma; ( C , D ): The plots of family Veillonellaceae and genus Veillonella on female infertility of tubal origin; ( E ): The plot of Rothia mucilaginosa on uterine leiomyoma. Each point represents a Single nucleotide polymorphism (SNP) serving as an instrumental variable for the exposure variable. The regression line’s slope, derived via inverse variance weighting and MR-Egger methods, denotes the estimated causal effect. The horizontal and vertical axes correspond respectively to the genetic association between the exposure variable (oral microbiota abundance) and the outcome variable (risk of female reproductive disorders). The shaded regions denote the 95% confidence intervals for the regression lines Fig. 3 Forest plots of the causal associations between oral microbiota and female reproductive diseases. A , B : The plots of class Bacilli and genus Veillonella on uterine leiomyoma; ( C , D ): The plots of family Veillonellaceae and genus Veillonella on female infertility of tubal origin; ( E ): The plot of Rothia mucilaginosa on uterine leiomyoma. Odds ratios (ORs) and 95% Confidence intervals (CIs) were estimated using the Inverse-variance weighted (IVW) method. The size of each square reflects the weight of the corresponding SNP in the meta-analysis. Horizontal lines represent 95% CIs. The diamond at the bottom represents the pooled causal estimate Forest plots of the causal associations between oral microbiota and female reproductive diseases. A , B : The plots of class Bacilli and genus Veillonella on uterine leiomyoma; ( C , D ): The plots of family Veillonellaceae and genus Veillonella on female infertility of tubal origin; ( E ): The plot of Rothia mucilaginosa on uterine leiomyoma. Odds ratios (ORs) and 95% Confidence intervals (CIs) were estimated using the Inverse-variance weighted (IVW) method. The size of each square reflects the weight of the corresponding SNP in the meta-analysis. Horizontal lines represent 95% CIs. The diamond at the bottom represents the pooled causal estimate Sensitivity analyses were conducted to assess heterogeneity and horizontal pleiotropy. Cochran’s Q test and the MR-Egger intercept were used to evaluate heterogeneity and directional pleiotropy, respectively (Table S3). MR-PRESSO was applied to identify potential outlier SNPs and to obtain outlier-corrected estimates when outliers were detected (Table S4), and the corresponding outlier-removed results are provided in Table S5. MR-PRESSO identified outlier SNPs in a subset of exposure–outcome pairs. After outlier correction, a nominal association between the species Rothia mucilaginosa and uterine leiomyoma was observed (OR = 1.0228, 95% CI 1.0069–1.0391; P  = 0.0202; outlier SNP: rs953559; Table S4). Across the full set of analyses, no associations remained significant after FDR correction, and the overall pattern should be interpreted as exploratory. In addition to the primary IVW results, several nominal associations were observed in the weighted median analyses; however, these did not remain significant after multiple-testing correction and were not consistently supported across complementary MR methods (Table S6). Steiger directionality testing supported the assumed direction from oral microbiome traits to FRDs in the analyzable pairs (Table S8). Overall, sensitivity analyses demonstrated the robustness of the suggestive findings. Funnel plots indicated no observable horizontal pleiotropy (Fig.  4 ), and leave-one-out sensitivity analyses confirmed that no single SNP disproportionately drove the causal associations (Fig.  5 ). Finally, post-hoc power calculations indicated that the statistical power to detect these modest causal effects was relatively limited (ranging from 57.94% to 71.40%) (Table S9), which likely explains the loss of statistical significance following rigorous FDR correction. Fig. 4 Funnel plots of the causal associations between oral microbiota and female reproductive diseases. A , B : The plots of class Bacilli and genus Veillonella on uterine leiomyoma; ( C , D ): The plots of family Veillonellaceae and genus Veillonella on female infertility of tubal origin; ( E ): The plot of Rothia mucilaginosa on uterine leiomyoma Funnel plots of the causal associations between oral microbiota and female reproductive diseases. A , B : The plots of class Bacilli and genus Veillonella on uterine leiomyoma; ( C , D ): The plots of family Veillonellaceae and genus Veillonella on female infertility of tubal origin; ( E ): The plot of Rothia mucilaginosa on uterine leiomyoma Fig. 5 Loo plots of the causal associations between oral microbiota and female reproductive diseases. A , B : The plots of class Bacilli and genus Veillonella on uterine leiomyoma; ( C , D ): The plots of family Veillonellaceae and genus Veillonella on female infertility of tubal origin; ( E ): The plot of Rothia mucilaginosa on uterine leiomyoma Loo plots of the causal associations between oral microbiota and female reproductive diseases. A , B : The plots of class Bacilli and genus Veillonella on uterine leiomyoma; ( C , D ): The plots of family Veillonellaceae and genus Veillonella on female infertility of tubal origin; ( E ): The plot of Rothia mucilaginosa on uterine leiomyoma

In this two-sample MR study, we evaluated the causal effects of genetically predicted oral microbiome traits on six female reproductive diseases using publicly available GWAS summary statistics. We identified several nominal IVW associations, primarily involving uterine leiomyoma and tubal infertility; however, none of these signals survived FDR correction. Accordingly, the results should be viewed as exploratory and hypothesis-generating, especially considering the modest exposure GWAS sample size and the multiple-testing burden. In sensitivity analyses, MR-PRESSO outlier correction provided an additional nominal signal for Rothia mucilaginosa in relation to uterine leiomyoma; this finding should be interpreted cautiously as exploratory and requires independent confirmation. The nominal association between class Bacilli and uterine leiomyoma is biologically plausible in so far as oral dysbiosis and periodontal inflammation can contribute to systemic inflammatory tone [ 5 – 7 ]. Importantly, MR estimates effects of genetically predicted microbial traits; it does not demonstrate translocation of live bacteria into reproductive tissues. A more conservative interpretation is that oral dysbiosis may increase circulating microbial components and metabolites (e.g., LPS–TLR4 signaling; SCFAs acting on FFAR2/FFAR3) that modulate immune cell activation and cytokine production [ 7 – 9 ], which are relevant to leiomyoma biology (e.g., uterine smooth muscle cells, fibroblasts, macrophages, and extracellular matrix remodeling) [ 4 ]. Similarly, Veillonellaceae/Veillonella are common oral anaerobes that may co-vary with oral ecological states linked to inflammation or metabolic profiles [ 6 , 38 ]; the observed nominal protective association with tubal infertility may reflect host-mediated pathways rather than a direct microbial effect. From a mechanistic perspective, members of class Bacilli can produce bioactive metabolites such as polyamines (e.g., spermidine) [ 39 ]. Veillonella species can generate short-chain fatty acids and other metabolites under specific ecological conditions, and Veillonella–Lactobacillus interactions have been reported to modulate intestinal inflammation in experimental models [ 40 , 41 ]. With respect to Rothia mucilaginosa , genome-scale metabolic modeling has characterized its metabolic capabilities, which may inform hypothesis-driven follow-up studies [ 42 ]. Although our MR results do not demonstrate translocation of live bacteria into reproductive tissues, microbial translocation has been documented in other clinical contexts involving mucosal barrier injury [ 43 ]. More broadly, microbiome–inflammation links are supported by evidence from chronic inflammatory disorders such as inflammatory bowel disease [ 44 ]. We observed opposite directions of effect estimates for genus Veillonella in tubal infertility versus uterine leiomyoma. Given the small effect sizes, lack of FDR significance, and potential heterogeneity, this pattern should not be over-interpreted as tissue-specific mechanisms. Instead, it may arise from differences in outcome definitions, distinct causal architectures, residual pleiotropy, or statistical fluctuation. Future studies using larger oral microbiome GWAS, colocalization analyses, and functional experiments will be needed to clarify whether this reflects true biological heterogeneity. Uterine leiomyoma and tubal infertility represent distinct etiologic pathways to infertility: leiomyoma may affect fertility through distortion of the uterine cavity, altered uterine contractility, or endometrial receptivity, whereas tubal infertility is often related to chronic infection, pelvic inflammatory disease, endometriosis, or surgical injury. In many settings, infectious etiologies—particularly pelvic inflammatory disease (including Chlamydia trachomatis )—are major contributors to tubal infertility, whereas leiomyoma-related subfertility is more often mediated by uterine anatomy and endometrial receptivity. Therefore, any shared “inflammatory background” should be interpreted as a broad upstream context rather than evidence of a unified disease pathway. Although both conditions may share upstream inflammatory risk factors, our results do not imply a direct clinical link between them. The current findings, if confirmed, would suggest that host genetic determinants of oral microbial traits could be modestly related to specific infertility phenotypes rather than broadly supporting an oral–reproductive axis across FRDs. Notably, the nominal signals were confined to uterine leiomyoma and tubal infertility, while we did not observe consistent evidence across the other FRD outcomes, arguing against a generalized oral–reproductive axis [ 4 , 45 ]. The principal strength of this study is the MR framework, which reduces confounding and reverse causation compared with conventional observational studies. We applied multiple sensitivity analyses (MR-Egger intercept, heterogeneity tests, MR-PRESSO, and Steiger directionality) to probe robustness. Several limitations warrant emphasis. First, the oral microbiome GWAS sample size (n = 610) is small, and instrument selection used a relaxed P-value threshold, which may increase the risk of weak-instrument bias and winner’s curse. Such a modest exposure GWAS sample size may limit statistical power and increase the risk of false-negative findings. Therefore, the present results should be interpreted as exploratory and require confirmation in larger oral microbiome GWAS and independent outcome datasets. Second, the exposure GWAS likely included both males and females, whereas FinnGen outcomes were female-specific; sex-related pathways could introduce residual pleiotropy, and female-stratified oral microbiome GWAS are needed. Third, not all predefined oral traits yielded eligible instruments across outcomes, and MR-PRESSO could not be applied to all pairs. Fourth, the effect sizes were small and are not directly applicable for clinical prediction or biomarker development. Furthermore, our analysis was restricted to the 44 microbial traits available in the primary GWAS dataset. Highly specific clinical periodontal pathogens, such as the established ‘red complex’ bacteria (e.g., Porphyromonas gingivalis), were not adequately captured or did not yield robust summary statistics in the original study. Consequently, we were unable to evaluate the specific causal roles of these well-known disease-associated oral microbes [ 17 , 23 , 31 ]. Larger and sex-stratified oral microbiome GWAS, independent replication outcome datasets, and integrative analyses will be valuable to refine causal pathways. Where suitable mediator GWAS are available, multivariable MR and mediation MR (e.g., incorporating inflammatory biomarkers or hormonal traits) could help separate direct microbial-trait effects from host-physiology pathways. Functional studies are needed to evaluate how host genetic determinants of oral microbial traits may influence systemic inflammation and reproductive tissue biology, including the roles of bacterial load, host immunity, and tissue microenvironment [ 32 ].

In conclusion, this two-sample MR analysis provides suggestive genetic evidence that certain oral microbiome traits may be modestly associated with uterine leiomyoma and tubal infertility. Because the associations did not remain significant after FDR correction and effect sizes were small, the findings should be interpreted cautiously and validated in larger, independent studies before any clinical implications are considered.

The ovaries and uterus are essential reproductive and endocrine organs. Female reproductive diseases (FRDs)—including uterine leiomyoma, endometriosis, polycystic ovary syndrome (PCOS), premature ovarian insufficiency, pelvic inflammatory sequelae, and infertility—affect quality of life and contribute to a substantial clinical and societal burden [ 1 ]. Inflammatory signaling and endocrine dysregulation (notably estrogen and progesterone pathways) are implicated in the pathogenesis of several FRDs, while tubal infertility is frequently related to chronic pelvic inflammation and infection-driven tissue damage [ 2 – 4 ]. Accumulating evidence suggests that disturbances in the oral microbial community (oral dysbiosis) are linked to systemic diseases beyond the oral cavity [ 5 , 6 ]. Oral inflammation and dysbiosis may promote low-grade systemic inflammation through transient bacteremia and circulation of microbial components and metabolites. For example, lipopolysaccharide (LPS) can activate Toll-like receptor 4 (TLR4) on innate immune cells (e.g., monocytes/macrophages), promoting cytokine release such as IL-6, IL-1β, and TNF-α and increasing inflammatory biomarkers (e.g., C-reactive protein) [ 7 ]. Microbial metabolites, including short-chain fatty acids (SCFAs), can signal via G-protein–coupled receptors (e.g., FFAR2/GPR43 and FFAR3/GPR41) and may also modulate histone deacetylase activity [ 8 , 9 ]. These systemic immune and metabolic changes may intersect with endocrine regulation via the hypothalamic–pituitary–gonadal axis and tissue steroid responsiveness, offering plausible routes by which oral dysbiosis could influence reproductive health. Nevertheless, observational associations are vulnerable to confounding (e.g., socioeconomic status, smoking, obesity, and health behaviors) and reverse causation. In addition to microbial components and metabolites, oral bacteria can produce other bioactive molecules such as bacteriocins, which may contribute to host–microbe interactions [ 10 ]. Systemic immunometabolic regulation in immune cells, including solute carrier transporters, provides another interface by which microbial signals may shape inflammatory responses [ 11 ]. Furthermore, oral streptococcal infections have been linked to autoimmune phenomena, and trained immunity has been implicated in autoimmune responses [ 12 , 13 ]. Evidence from pregnancy research also suggests that placental Toll-like receptor recognition of salivary and subgingival microbiota is associated with pregnancy complications [ 14 ]. Recent reviews further highlight age-related oral dysbiosis and systemic comorbidities, the dynamic host interactions of the oral microbiota, and potential viral contributions in periodontitis [ 15 – 17 ]. A growing body of research has explored microbiome–reproductive links, mostly focusing on gut, vaginal, and endometrial microbiota [ 18 – 21 ]. Compared with these niches, the oral microbiome is an accessible microbial ecosystem and a potential upstream contributor to systemic immune and metabolic states [ 5 , 6 ]. However, whether variation in oral microbial traits has a causal role in FRDs remains unclear. Mendelian randomization (MR) is a genetic epidemiologic approach that uses germline variants associated with an exposure as instrumental variables to estimate the causal effect of that exposure on an outcome, mitigating confounding and reverse causation under key assumptions [ 22 – 24 ]. Because MR estimates the effect of genetically predicted microbial traits rather than the direct effect of microbial exposure itself, interpretation should be framed at the level of host genetic determinants of microbial composition and their downstream consequences. In this study, we performed a two-sample MR analysis to evaluate the causal effects of oral microbiome traits on six FRDs using publicly available GWAS summary statistics. We further applied extensive sensitivity analyses and multiple-testing correction to provide a transparent and conservative assessment of the evidence.

Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3. Supplementary Material 4. Supplementary Material 4. Supplementary Material 5. Supplementary Material 5. Supplementary Material 6. Supplementary Material 6. Supplementary Material 7. Supplementary Material 7. Supplementary Material 8. Supplementary Material 8. Supplementary Material 9. Supplementary Material 9.