Arsenic-Induced Carcinogenesis—Oxidative Stress as a Possible Mode of Action and Future Research Needs for More Biologically Based Risk Assessment

Abstract

Exposure to inorganic arsenic (iAs) induces cancer in human lungs, urinary bladder, skin, kidney, and liver, with the majority of deaths from lung and bladder cancer. To date, cancer risk assessments for iAs have not relied on mechanistic data, as we have lacked sufficient understanding of arsenic's pharmacokinetics and mode(s) of carcinogenic action (MOA). Furthermore, while there are vast amounts of toxicological data on iAs, relatively little of it has been collected using experimental designs that efficiently support development of biologically based dose-response (BBDR) models and subsequently risk assessment. This review outlines an efficient approach to the development of a BBDR model for iAs that would reduce uncertainties in its cancer risk assessment. This BBDR-based approach is illustrated by using oxidative stress as the carcinogenic MOA for iAs but would be generically applicable to other MOAs. Six major research needs that will facilitate BBDR model development for arsenic-induced cancer are (1) MOA research, which is needed to reduce the uncertainty in risk assessment; (2) development and integration of the pharmacodynamic component (MOA) of the BBDR model; (3) dose-response and extrapolation model selection; (4) the determination of internal human speciated arsenical concentrations to improve physiologically based pharmacokinetic (PBPK) models; (5) animal models of arsenic carcinogenesis; and (6) the determination of the low dose human relationship for death from cancer, particularly in lungs and urinary bladder. The major parts of the BBDR model are arsenic exposure, a physiologically based pharmacokinetic model, reactive species, antioxidant defenses, oxidative stress, cytotoxicity, growth factors, transcription factors, DNA damage, chromosome damage, cell proliferation, mutation accumulation, and cancer. The BBDR model will need to be developed concurrently with data collection so that model uncertainties can be identified and addressed through an iterative process of targeted additional research.