Low-level Arsenic Exposure and Cardiovascular Disease in Multi-Ethnic Adults (MESA As)

PROJECT SUMMARY Epidemiologic and experimental evidence supports that exposure to moderate-to-high arsenic (As) is a cardiovascular disease (CVD) risk factor. Little is known, however, on the cardiovascular effects of low As exposure through diet, particularly rice. We aim to investigate the prospective association of As exposure and metabolism with clinical and subclinical CVD (Aim 1), as well as effect modification (Aim 2), and potential mediators (Aim 3) of these associations in the Multi-Ethnic Study of Atherosclerosis (MESA), a cohort study of White, Black, Hispanic, and Chinese-American adults 45-84 years of age from 6 US cities. In a pilot study in MESA (n=310), median (percentile 90) urine As not derived from seafood was 4.9 (11.7), 3.8 (11.3), 2.2 (8.5), 2.2 (6.1) µg/L for Chinese Americans, Hispanics, Blacks and Whites; higher rice intake was associated with higher urine As; and a candidate SNP in AS3MT (gene encoding As(III) methyltransferase), was non- significantly associated with urine As metabolism profile. In the proposed study, we will include 6,725 MESA participants free of CVD at baseline who have been followed for a mean of 12.1 years with up to 5 exams. Urine samples, cardiovascular risk factors, genetic and epigenetic data, early markers of CVD, and subclinical and clinical CVD outcomes are available. We will measure As species and total As and other metals in spot urine samples and estimate non seafood As using a previously validated method. We will evaluate all CVD, coronary heart disease and stroke mortality and incidence (as clinical outcomes) and carotid intima-media thickness, coronary artery calcification, arterial distensibility and peripheral arterial disease (as subclinical outcomes). We will examine smoking status and secondhand smoke, genome-wide and Metabochip genetic variants, urine As metabolism patterns, and plasma homocysteine as a marker of one-carbon metabolism. We will also evaluate markers of inflammation (C-reactive protein, interleukin(IL)-2, and IL-6), endothelial function (soluble cell adhesion molecules), thrombosis (fibrinogen and plasminogen activator inhibitor-1, electrophysiology (QT interval), hypertension, diabetes, and epigenetic modifications (DNA methylation). In Aim 1 we will conduct Cox proportional hazards models for time to event clinical outcomes, generalized linear models (GLM) with logit or identity link for binary or continuous subclinical outcomes at baseline, and marginal linear models with repeated subclinical outcomes longitudinally. Arsenic levels will be modeled as continuous, categorical or flexible splines, separately. Models will be progressively adjusted for potential confounders including sociodemographic and traditional CVD risk factors. Effect modification will be estimated by adding interaction terms to the regression models. For mediation analyses, we will first estimate the association between As and the potential mediators using similar strategies as in Aim 1, then formal statistical mediation analysis will be applied. This study can inform on the cardiovascular role, or lack thereof, of low-level As exposure and contribute to As risk assessment and to CVD prevention and control in general populations.