Causal effects of plasma lipids on the risk of atrial fibrillation: A multivariable mendelian randomization study

Observational studies have suggested that plasma lipids contribute substantially to cardiovascular disease, but "cholesterol paradox" in atrial fibrillation (AF) remains. We sought to investigate the causal effects of lipid profiles on the risk of AF.

Two-sample Mendelian randomization (MR) framework was implemented to examine the causality of association. Summary estimations of genetic variants associated with low density lipoprotein (LDL)-cholesterol, high density lipoprotein (HDL)-cholesterol, total cholesterol, triglycerides, lipoprotein-a [Lp(a)], apolipoprotein A1 (ApoA 1), and apolipoprotein B (ApoB) were 81, 99, 96, 61, 30, 10, and 23 single nucleotide polymorphisms, respectively. Genetic association with AF were retrieved from a genome-wide association study that included 1,030,836 individuals. The complications for AF were predefined as cardioembolic stroke (CES) and heart failure (HF). In the multivariable MR, the odds ratios for AF per standard deviation (SD) increase were 1.030 (95% confidence interval (CI) 0.979-1.083; P = 0.257) for LDL-cholesterol, 0.986 (95% CI 0.931-1.044; P = 0.622) for HDL-cholesterol, 0.965 (95% CI 0.896-1.041; P = 0.359) for triglycerides, 1.001 (95% CI 1.000-1.003; P = 0.023) for Lp(a), 1.017 (95% CI 0.966-1.070; P = 0.518) for ApoA1, and 1.002 (95% CI 0.963-1.043; P = 0.923) for ApoB. There was no evidence that other lipid components were causally associated with AF, CES, or HF, other than for a marginal association between triglycerides and HF.

This MR study provides robust evidence that high Lp(a) increases the risk of AF, suggesting that interventions targeting Lp(a) may contribute to the primary prevention of AF.