Biobank-scale exposome-wide risk factors in cardiometabolic disease: observational, predictive, and causal evidence

Cardiovascular disease and diabetes are intricately related and influenced by factors within the “exposome”. Distinguishing between correlational and causal risk associations is challenging, especially at exposome scale. Here, we triangulate observational Exposure-Wide Association Study ( ExWAS ) evidence with “randomized” evidence for the exposome using mendelian randomization (MR) for almost 500 exposures. First, the ExWAS identified 144 significant factors for coronary artery disease (CAD) and 237 for type 2 diabetes (T2D), with 120 shared between both. These factors had modest predictive ability (variance explained) for both phenotypes. However, genetic-based causality was deduced for only 14 factors in CAD and 16 in T2D, with seven implicated in both. Additionally, we found strong concordance of MR-validated findings between prevalent and incident disease associations (85.7% [12/14] for CAD and 87.5% [14/16] for T2D). Most correlational findings pertain to lifestyle factors (particularly diet), but social educational factors are more prominently highlighted among those with causal support.