Genetic evidence for independent causal relationships between metabolic biomarkers and risk of coronary artery diseases

Abstract

Decades of epidemiological research have identified numerous risk factors and biomarkers that are associated with risk of coronary artery disease (CAD) and myocardial infarction. The most well recognized of these are circulating levels of total cholesterol, LDLcholesterol, HDL-cholesterol, and triglycerides, as well as metabolic syndrome-related traits, such as obesity, hypertension, and T2D (1). However, the association of a biomarker with CAD in observational studies does not necessarily prove a causal relationship. Furthermore, inferring causality based on epidemiology alone can be confounded by the biomarkers themselves often being associated with each other (i.e., obesity and blood triglyceride levels). In this regard, the gold standard approach for establishing causality is through rigorous and appropriately designed randomized clinical trials. For example, therapeutic interventions that lower LDL levels or blood pressure have demonstrated clear protective effects on risk of CAD, thus validating their causal roles in development and progression of atherosclerosis (2). By comparison, clinical trials aimed at raising HDL levels have failed to demonstrate therapeutic benefits for CAD outcomes (3–7). These latter observations cast doubt on the causal role of HDL in CAD despite its strong inverse clinical association with CAD. Another complementary and efficient strategy for inferring causality relies on human genetics and is termed Mendelian randomization (MR) (8). The premise behind this approach is that genetic variants affecting a biomarker should also yield a directionally consistent association with risk of CAD if the biomarker is driving disease. Thus, the random segregation of alleles associated with a biomarker and CAD during gametogenesis mimics the random assignment of subjects to either treatment or placebo in clinical interventions. Given large publicly available genomewide association study (GWAS) summary data sets for