Genetically proxied sex hormones, Sjögren’s disease, and broad dry eye disease: A bidirectional two-sample mendelian randomization study

Objective To evaluate bidirectional associations of genetically proxied total testosterone and estradiol with Sjögren’s disease (SjD) and broad dry eye disease (DED) using female-derived, male-derived, and sex-combined hormone GWASs. Methods We conducted bidirectional two-sample Mendelian randomization using European-ancestry genome-wide association studies (GWASs). Forward analyses evaluated total-testosterone and estradiol traits from female-derived, male-derived, and sex-combined GWASs against FinnGen SjD (3,309 cases and 484,260 controls) and Million Veteran Program broad DED (84,262 cases and 314,374 controls); reverse analyses used disease instruments to estimate disease-to-hormone associations. Primary instruments met P < 5 × 10 -8 and F > 10. Secondary forward SjD analyses used bioavailable-testosterone and testosterone-cluster instruments. Inverse-variance-weighted (IVW) estimates were complemented by weighted median, pleiotropy diagnostics, major histocompatibility complex (MHC) exclusion, Steiger orientation, and direction-specific LDtrait analyses. Results Sex-combined total testosterone was inverse for SjD by IVW (26 variants; odds ratio [OR], 0.41; 95% confidence interval [CI], 0.23–0.72; P = 0.002) and weighted median (OR, 0.30; 95% CI, 0.14–0.68; P = 0.004). The IVW estimate remained inverse after LDtrait exclusion. Male- and female-derived total-testosterone estimates were null and directionally discordant. Sex-combined bioavailable testosterone was also inverse (73 variants; OR, 0.50; 95% CI, 0.33–0.76; P = 0.001); the other four secondary IVW estimates did not meet correction, and all five secondary sets were heterogeneous. No estradiol–SjD, hormone–DED, or reverse IVW estimate met correction. Reverse sensitivity signals were not robust to MHC exclusion. Conclusions Sex-combined total- and bioavailable-testosterone estimates were concordantly inverse for SjD, but sex-specific estimates were null. The data establish neither a common effect across sex strata, independence from sex hormone-binding globulin, nor a disease-to-hormone association. Sparse estradiol and reverse instruments constrain inference. Larger sex-stratified SjD datasets are needed; these findings do not support hormone testing or treatment.