Fig. 3
Results of stepwise MR mediation analysis between NAFLD, glycemic-related traits, sex hormones, and PCOS. a Direct causal effects of NAFLD, glycemic-related traits, and sex hormones on PCOS. b Direct causal effects of NAFLD, glycemic-related traits, and SHBG on serum BT levels. c Direct causal effects of NAFLD and glycemic-related traits on serum SHBG levels. d Causal effects of NAFLD on glycemic-related traits. MR-PRESSO analysis was not applicable to estimate the causal effect of NAFLD on fasting glucose levels due to the small number of genetic instruments used. θ1: direct causal effect of NAFLD on PCOS; θ2: direct causal effect of fasting insulin levels on PCOS; θ3: direct causal effect of serum BT levels on PCOS; θ4: direct causal effect of SHBG on serum BT levels; θ5: direct causal effect of fasting insulin levels on SHBG; θ6: causal effect of NAFLD on fasting insulin; θ2×θ6 indirect causal effect of NAFLD on PCOS via fasting insulin levels only; θ3×θ4×θ5×θ6: indirect causal effect of NAFLD on PCOS via fasting insulin and sex hormone levels. a: A secondary IVW analysis was conducted after excluding rs2068834 due to its genome-wide significant association with obesity. b: Outlying genetic instruments were excluded in the corrected MR-PRESSO analysis. Abbreviations: BT, bioavailable testosterone; CI, confidence interval; FG, fasting glucose; FI, fasting insulin; IVs, instrumental variables; MVMR, multivariable Mendelian randomization; NAFLD, non-alcoholic fatty liver disease; PCOS, polycystic ovary syndrome; SHBG, sex hormone-binding globulin