Genetic Evidence for Causal Effects of Domain-Specific Physical Activity and Sedentary Behavior on Osteoporosis and Inflammatory Arthritis: A Bidirectional Mendelian Randomization Study

Abstract

Objective: Genetic factors contribute to physical activity (PA) and sedentary behavior (SB), two complex behavioral traits. They may be associated with musculoskeletal and inflammatory diseases. However, the relationships of domain-specific PA and SB traits with osteoporosis (OS), psoriatic arthritis (PsA), and rheumatoid arthritis (RA) remain unclear. A two-sample Mendelian randomization (MR) analysis assessed potential bidirectional associations between these behavioral traits and the three diseases. Methods: GWAS summary statistics were used to examine five domain-specific PA traits, three SB traits, and OS, RA, and PsA. Primary causal estimates were obtained using the inverse-variance weighted (IVW) method. Complementary analyses used MR-Egger regression, the weighted median method, and the weighted mode method. Sensitivity analyses were performed using Cochran’s Q test, the MR-Egger intercept test, MR-PRESSO, and leave-one-out analysis. Multiple testing was addressed using false discovery rate (FDR) correction. Results: In the forward MR analyses, genetically predicted liability to Light DIY was associated with lower odds of PsA after FDR correction (OR = 0.006, 95% CI = 0.0002–0.236, FDR-adjusted p = 0.018). Genetically predicted longer television viewing was linked to increased odds of PsA (OR = 2.205, 95% CI = 1.089–4.463, FDR-adjusted p = 0.028) and RA (OR = 1.006, 95% CI = 1.002–1.010, FDR-adjusted p = 0.004). Walking for pleasure showed a nominal inverse association with PsA, and computer use showed a nominal inverse association with RA. Neither association remained significant after FDR correction. No evidence of associations with the broad self-reported OS diagnosis was observed. Reverse MR analyses showed lower odds of Walking for pleasure (OR = 0.662, 95% CI = 0.484–0.906, FDR-adjusted p = 0.013) and Other exercises (OR = 0.651, 95% CI = 0.490–0.864, FDR-adjusted p = 0.006) for genetic liability to RA. RA liability was also associated with longer television viewing time (β = 0.708, 95% CI = 0.251–1.165, FDR-adjusted p = 0.006). No reverse associations were observed for PsA. Conclusions: This bidirectional MR study identified several potential genetically predicted associations of domain-specific PA and SB traits with PsA and RA. The findings for television viewing and RA may suggest a potential bidirectional association. However, the results should be considered hypothesis-generating. These findings require independent replication and further validation before causal or clinical conclusions can be drawn.