Causal relationship between nutritional assessment phenotypes and heart failure: A Mendelian randomization study

Malnutrition is strongly associated with heart failure (HF); however, the causal link remains unclear. We used Mendelian randomization (MR) to infer causal associations between different nutritional assessment phenotypes and HF and to analyze whether these associations were mediated by common HF risk factors. Two-sample bidirectional MR was used to infer causal associations between nutritional assessment phenotypes and HF. Mutual influences between different nutritional assessment phenotypes and potential correlations were estimated using multivariate MR methods. Two-step MR was used to quantify the mediating effects of common HF risk factors on the causal associations. Three phenotypes were positively associated with the development of HF: waist circumference (WC) (odds ratio [OR] = 1.74; 95% confidence interval [CI], 1.60–1.90; P = 3.95 × 10−39), body mass index (BMI) (OR = 1.70; 95%CI, 1.60–1.80; P = 1.35 × 10−73), and whole body fat mass (WBFM) (OR = 1.54; 95%CI, 1.44–1.65; P = 4.82 × 10−37). Multivariate MR indicated that WBFM remained positively associated with HF after conditioning on BMI and WC (OR = 2.05; 95%CI, 1.27–3.31; P = 0.003). Three phenotypes were negatively correlated with the development of HF: usual walking pace (UWP) (OR = 0.40; 95%CI, 0.27–0.60; P = 8.41 × 10−6), educational attainment (EA) (OR = 0.73; 95%CI, 0.67–0.79; P = 2.27 × 10−13), and total cholesterol (TC) (OR = 0.90; 95%CI, 0.84–0.96; P = 4.22 × 10−3). There was a bidirectional causality between HF and UWP (Effect estimate = −0.03; 95%CI, −0.05 to −0.01; P = 1.95 × 10−3). Mediation analysis showed that common risk factors for HF (hypertension, coronary artery disease, cardiomyopathy, and valvular heart disease) mediated these causal associations (all P