Study on the bidirectional causal relationship between visceral adipose tissue and heart failure: Multivariate Mendelian randomization and mediation analysis

This genetic causal inference investigation employs Mendelian randomization (MR) to quantify the causal relationship between visceral adiposity and heart failure (HF), while applying mediation analysis to evaluate coronary artery disease (CAD) as an intermediate pathway. Triglycerides (TGs), subcutaneous fat, and low-density lipoprotein (LDL) were included in multivariate models in analyses to control for horizontal pleiotropy bias. The causal effect of visceral adipose tissue (VAT) on HF was analyzed in a European population using two-sample MR with predominantly inverse variance weighting. Multivariate models were also used to correct for the effects of TGs, subcutaneous fat, and LDL on outcomes to assess the independent causal effect of each exposure on outcomes. The proportion of the mediating effect of CAD in the causal association between VAT and HF was further tested by two-step mediation analysis. Univariate MR analyses showed a significant association between VAT and risk of HF (OR 1.65, 95% CI: 1.52–1.80; P = 1.79 × 10 −31 ), and sensitivity analyses verified the robustness of these results. Multivariate analysis showed that increased VAT mass adjusted for TGs, subcutaneous fat, and LDL remained a stable risk factor for HF (OR 1.49, 95% CI: 1.23–1.80, P = 5.73 × 10 −5 ). Mediation analyses showed that CAD partially mediated the causal association between VAT and HF risk (a 12% mediated effect and 88% direct effect). MR suggests that elevated VAT increases the risk of HF and that CAD partially mediates this causal association.