Linear-No-Threshold Default Assumptions for Noncancer and Nongenotoxic Cancer Risks: A Mathematical and Biological Critique

To improve U.S. Environmental Protection Agency (EPA) dose‐response (DR) assessments for noncarcinogens and for nonlinear mode of action (MOA) carcinogens, the 2009 NRC Science and Decisions Panel recommended that the adjustment‐factor approach traditionally applied to these endpoints should be replaced by a new default assumption that both endpoints have linear‐no‐threshold (LNT) population‐wide DR relationships. The panel claimed this new approach is warranted because population DR is LNT when any new dose adds to a background dose that explains background levels of risk, and/or when there is substantial interindividual heterogeneity in susceptibility in the exposed human population. Mathematically, however, the first claim is either false or effectively meaningless and the second claim is false. Any dose‐and population‐response relationship that is statistically consistent with an LNT relationship may instead be an additive mixture of just two quasi‐threshold DR relationships, which jointly exhibit low‐dose S‐shaped, quasi‐threshold nonlinearity just below the lower end of the observed ‘linear’ dose range. In this case, LNT extrapolation would necessarily overestimate increased risk by increasingly large relative magnitudes at diminishing values of above‐background dose. The fact that chemically‐induced apoptotic cell death occurs by unambiguously nonlinear, quasi‐threshold DR mechanisms is apparent from recent data concerning this quintessential toxicity endpoint. The 2009 NRC Science and Decisions Panel claims and recommendations that default LNT assumptions be applied to DR assessment for noncarcinogens and nonlinear MOA carcinogens are therefore not justified either mathematically or biologically.