Autoimmune hypothyroidism and endometriosis: evidence from bidirectional Mendelian randomization and an independent clinical cohort analysis
Background Observational studies have reported comorbidity between autoimmune thyroid disease and endometriosis, but whether this reflects a direct causal relationship or shared susceptibility remains unclear. We evaluated the bidirectional relationship between autoimmune hypothyroidism (AIHT) and endometriosis using Mendelian randomization (MR) and an independent clinical cohort. Methods Genome-wide association study (GWAS) summary statistics were obtained for AIHT from a FinnGen R12–UK Biobank meta-analysis and for endometriosis from GWAS Catalog study GCST90205183. Genome-wide significant instruments underwent linkage disequilibrium clumping, genome-build liftover, and harmonization. Inverse-variance weighted (IVW) analysis was primary, with weighted median and MR-Egger analyses as complements. Sensitivity analyses included heterogeneity and MR-Egger intercept tests, Mendelian randomization pleiotropy residual sum and outlier (MR-PRESSO), leave-one-out analysis, phenome-wide association study-informed filtering, and exclusion of the human leukocyte antigen region. Clinical associations were assessed using multivariable logistic regression. Results After harmonization, 248 single-nucleotide polymorphisms (SNPs) were included for AIHT-to-endometriosis analysis and 26 for endometriosis-to-AIHT analysis. IVW showed no association of genetic liability to AIHT with endometriosis (odds ratio [OR] = 0.985, 95% confidence interval [CI]: 0.957–1.014, P = 0.315) or of genetic liability to endometriosis with AIHT (OR = 0.999, 95% CI: 0.951–1.049, P = 0.968). MR-PRESSO global tests were significant in both directions, indicating pleiotropic heterogeneity. No forward outlier was identified; removal of the reverse-direction outlier rs2433340 did not materially alter the estimate (corrected OR = 1.007, 95% CI: 0.960–1.056, P = 0.769). The cohort included 7,825 women: 853 with confirmed endometriosis and 6,972 diagnosis-negative controls. Endometriosis was not associated with AIHT after adjustment (OR = 0.753, 95% CI: 0.440–1.287, P = 0.299); findings were also non-significant for thyroid autoimmunity, thyroid peroxidase antibody positivity, thyroglobulin antibody positivity, and hypothyroidism. Conclusion This integrated genetic and clinical analysis found no robust evidence that genetic liability to AIHT materially affects overall endometriosis risk or vice versa in the datasets analyzed, and no clear clinical association between confirmed endometriosis and AIHT or thyroid autoimmunity. Small effects and sex- or subtype-specific associations cannot be excluded.