Use of Blood Metals to Improve the Prediction of Death from Cardiovascular Causes

Abstract

PDS 63: Chemicals and metals: health effects, Exhibition Hall (PDS), Ground floor, August 27, 2019, 10:30 AM - 12:00 PM Background/Aim: We evaluated whether blood markers of toxic metals (lead, cadmium, and mercury) can improve risk prediction for cardiovascular disease (CVD) mortality when added individually, jointly, or as an integrative index-- the Environmental Risk Score (ERS), in a model with established risk factors. Methods: Our study sample comprised 16,030 adults aged 40 years or older who were enrolled in the NHANES 1999-2012 and followed up through December 31, 2015. Individuals were linked with the National Death Index. The study sample was randomly split into a training set for the ERS construction (n=8,044), and a testing set for evaluation of the prediction performance (n=7,986). ERS was an integrative index of health risk of exposure to multiple toxicants, which was computed for CVD mortality using elastic-net (ENET) penalized Cox’s model with 9 candidate predictors including 3 linear terms, 3 squared terms, and 3 pairwise interactions of blood lead, cadmium, and mercury concentrations. Results: During median follow-up of 7.2 years, 517 died from CVD. In training set, linear terms of cadmium and mercury, squared terms of lead and mercury, and all 3 pairwise interactions were selected by ENET for ERS construction. In testing set, addition of all 3 linear term, 3 squared term, and 3 pairwise interactions of blood metals simultaneously to established risk factors in the Cox’s models improved C-statistic from 0.843 to 0.856. The C statistic also increased to 0.853 when the ERS was added to established risk factors. The improvement in risk assessment remained significant and strong when it was estimated by other metrics including net reclassification improvement and integrated discrimination improvement. Conclusions: The use of multiple predictors of exposure to blood lead, cadmium, and mercury improves risk prediction performance over established risk factors. Our findings highlight the importance of incorporation of environmental toxicants for CVD risk assessment and prevention.