LAUSR

BACKGROUND To examine the causal association of device-measured physical activity with Bipolar disorder (BIP) and schizophrenia (SCZ) in adults by performing 2-sample MR analysis. METHODS Summary-level data for exposure variable including overall physical activity, and three different types of physical activities(sedentary, moderate intensity activity, sleep duration) were derived from UK Biobank(n = 91,105). Summary-level data for outcomes were obtained from Psychiatric Genomics Consortium (PGC) including BIP (n = 51,710) and SCZ(n = 65,967). Cochran's Q statistics, MR-Egger regression, and MR-PRESSO global test were used to identify pleiotropy and MR-PRESSO outlier test for removing the pleiotropy of the genetic instruments. For assessing causality, proven MR techniques including one main analysis(IVW)and four sensitive analysis(MLE,SME,WME, MR-PRESSO)were applied for robust results. RESULTS Five MR methods provided consistent genetically evidence that overall physical activity was associated with lower risk of BIP(OR, 0.491; 95% CI: 0.314-0.767; p = 0.002), there was no significant relationship between overall physical activity and SCZ(IVW OR; 1.133; 95% CI: 0.636-2.020; p = 0.672). In addition, no significant causal association was observed between any other types of physical activities and both mental disorders. LIMITATIONS The nonlinear association of physical activities on BIP and SCZ were unaccessible through MR analysis. CONCLUSIONS This study provides genetic evidence that overall physical activity is an effective preventive factor for BIP but not for SCZ. Together with evidence from observational studies, our finding provides further rationale for individuals at risk for BIP to maintain physical activity.