Association of behavioral factors, liver function and NAFLD: Bayesian Mendelian randomization.

Background: The causal associations between behavioral factors (BFs) and the risk of non-alcoholic fatty liver disease (NAFLD), and whether liver function mediates these associations, remain unclear. Therefore, this study aimed to assess these associations.

Methods: We performed two-sample Mendelian randomization (2SMR) and multivariable MR (MVMR) analysis using summary-level data to assess the associations between BFs and NAFLD. The linkage disequilibrium score regression (LDSC) was used for genetic correlation analysis. Additionally, we utilized NHANES database to assess dose-response relationships. Furthermore, we applied two-sample MVMR approach based on Bayesian model averaging (MR-BMA) to identify the most influential liver function index as a mediating factor, and performed mediation analysis.

Results: 2SMR results showed that leisure screen time (ST, β = 0.414, P = 5.91e-6) and smoking initiation (SI, β = 0.164, P = 0.012) were associated with NAFLD risk with no reverse causality. LDSC supported these associations (SI: rg = 0.291, P = 1.04e-8; ST: 0.518, P = 5.41e-21). However, MVMR showed that only ST was independently associated with NAFLD (β = 0.334, P = 4.6e-5). There was a linear relationship between ST and NAFLD, and NAFLD risk increased significantly after 5 h of ST. Alanine transaminase level was the most influential index (ALT, MIP = 0.452) and mediated 54% of the ST-NAFLD association.

Conclusion: ST is independently associated with an increased risk of NAFLD. It is recommended to avoid more than 5 h of ST per day. ALT is the most influential liver function index associated with NAFLD and mediates the ST-NAFLD association.