Abstract
Previous epidemiological evidence has suggested associations between psychiatric disorders and gynecologic tumors, but the causal relationships have not been fully characterized. This study aims to investigate the potential causal relationships using genetic instrumental variables. A 2-sample Mendelian randomization (MR) analysis was conducted utilizing 3 large-scale genome-wide association study databases: FinnGen, UK Biobank, and the Integrative Epidemiology Unit. The exposures included anxiety disorder, obsessive-compulsive disorder, schizophrenia, major depressive disorder, and bipolar disorder. The outcomes comprised cervical cancer (CC), ovarian cancer (OC), endometrial cancer, and uterine fibroids. Causal estimates were primarily derived using inverse variance weighting, with robustness assessed via MR-Egger regression, weighted median, and pleiotropy-robust methods. Heterogeneity and horizontal pleiotropy were evaluated using Cochran’s
Q
statistic, MR-Egger intercept test, and sensitivity analyses. False discovery rate (FDR) correction was applied for multiple testing across 20 exposure-outcome pairs. The data included CC (909 cases and 2,38,249 controls), OC (2188 cases and 2,37,839 controls), endometrial cancer (1218 cases and 1,98,523 controls), and uterine fibroids (21,024 cases and 2,37,694 controls). No data loss occurred. Among 20 exposure-outcome pairs, only schizophrenia showed a statistically significant and robust association with OC after FDR correction (OR = 1.0464, 95% CI: 1.0136–1.0801,
P
= .0052; FDR-adjusted
P
= .0258). Sensitivity analyses confirmed no significant heterogeneity (inverse variance weighting
Q P
= .068) or horizontal pleiotropy (MR-Egger intercept
P
= .116). Mendelian Randomization-Pleiotropy RESidual Sum and Outlier (MR-PRESSO) was performed with 5000 permutations; the global test revealed no evidence of horizontal pleiotropy (RSSobs = 328.65,
P
= .058) and no outlier Single nucleotide polymorphisms were detected. Although major depressive disorder was nominally associated with CC (
P
= .0056, FDR = 0.0258), leave-one-out sensitivity analysis revealed this association was highly unstable and not robust. No other causal associations were observed. The suggestive genetic association between schizophrenia and OC merits attention in future etiological research. The null findings for anxiety disorder, bipolar disorder, and obsessive-compulsive disorder with gynecologic tumors may help alleviate unnecessary clinical concerns.