Abstract
Metformin, as a drug for treating type 2 diabetes, has been found to have a positive impact on mental health. This study will assess the potential causal relationship between metformin exposure and the risk of bipolar disorder (BD). We used publicly available genome-wide association studies statistical data to conduct 2-sample bidirectional Mendelian randomization (MR) analysis to determine the causal relationship between genetic susceptibility to metformin and BD risk. Genome-wide significant single nucleotide polymorphisms, that are associated with metformin use were selected as the instrumental variables. To determine the causal relationship between metformin exposure and BD, MR analysis was conducted, employing methods such as instrumental variable weighting. We applied complementary methods, including weighted median, weighted mode, simple mode, MR-Egger regression, and MR-pleiotropy residual sum and outlier to detect and correct for the effect of horizontal pleiotropy. Cochran Q statistics were used to assess instrument heterogeneity. The leave-one-out method was conducted for sensitivity analysis. The instrumental variable weighting analysis demonstrated that there was a significant causal relationship between the metformin exposure and BD risk [odds ratio [OR] (95% confidence interval [CI]) = 0.032 (0.002-0.593), P = .021] (ebi-a-GCST003724), [OR (95% CI) = 1.452E-04 (1.442E-07 to 1.463E-01), P = .012] (ieu-a-800), [OR (95% CI) = 0.33 (0.27-0.404), P = .000] (ieu-a-808). The sensitivity analysis revealed no heterogeneity in the individual results (P > .05), and no significant publication bias. This study reveals that metformin may serve as a protective factor for BD, providing new theoretical basis for the application of metabolic intervention in the prevention and treatment of mental illness.