Essential concepts for interpreting the dose-response of low-level arsenic exposure in epidemiological studies

Abstract

Scientifically robust selections of epidemiological studies and assessments of the dose-response of inorganic arsenic in the low-dose range must consider key issues specific to arsenic in order to reduce risk of bias. The abundance of toxicological, mechanistic, and epidemiological evidence on arsenic enables a nuanced assessment of risk of bias in epidemiological studies of low-level arsenic, as opposed to a generic evaluation based only on standard principles. Important concepts in this context include 1) arsenic metabolism and mode of action for toxicity and carcinogenicity; 2) effects of confounding factors such as diet, health status including nutritional deficiencies, use of tobacco and other substances, and body composition; 3) strengths and limitations of various metrics for assessing relevant exposures consistent with the mode of action; and 4) the potential for bias in the positive direction for the observed dose-response relationship as exposure increases in the low-dose range. As an example, evaluation of a recent dose-response modeling using eight epidemiological studies of inorganic arsenic and bladder cancer demonstrated that the pooled risk estimate was markedly affected by the single study that was ranked as having a high risk of bias, based on the above factors. The other seven studies were also affected by these factors to varying, albeit lesser, degrees that can influence the apparent dose-response in the low-dose range (i.e., drinking water concentration of 65 µg/L or dose of approximately ≤1 µg/kg-day). These issues are relevant considerations for assessing health risks of oral exposures to inorganic arsenic in the U.S. population, and setting evidence-based regulatory limits to protect human health.