Dose-Tolerance Approaches to Quantitative Risk Assessment. J. Glowa, NIDDK/ NIH, Bethesda, MD 20892; and R. MacPhail, U.S. EPA/HERL, Research Triangle Park, NC 27711
Risk assessment refers to the enterprise of estimating the chances of obtaining an adverse effect in humans following exposure to a chemical. Although it might be possible to estimate the degree of human risk if sufficient exposure- and effect-related data were available, estimates are usually derived from effects obtained in animals exposed to a range of doses. The goal is to estimate exposure levels that have negligible risks of injury, and various approaches have emerged over the years to achieve this ideal. Low-dose extrapolation models for carcinogenic and mutagenic effects were among the earliest, but estimates could vary by orders of magnitude depending upon the model. For systemic (non-cancer) toxicity, the acceptable daily intake (ADI) or reference dose (RfD) approach has been used to estimate a dose level that could be tolerated over the lifetime with negligible risks. The ADI/RfD approach has been criticized for, among other things, its failure to fully incorporate the dose-effect data and its tendency to provide disincentives for conducting experimentally rigorous investigations of potential toxic effects. Recently, several new approaches to risk assessment have emerged that can more effectively utilize dose-effect data. Most of these approaches also encourage well-controlled experiments. They can be distinguished by their focus on either variability in the dose producing a fixed effect (dose-tolerance or probabilistic approaches) or variability in the effects of a fixed dose (effect-tolerance or benchmark approaches) to define risks. In a typical dose-tolerance assessment, effects of an agent are expressed as a percentage or proportion of control, doses are transformed to natural logarithms, and a mathematical function is fit to the data. The variance in replicates is used, through Z-score transformations, to estimate exposure levels associated with the probability of obtaining a small but measurable effect. A recent comparison of results obtained with different dose-tolerance approaches (Glowa and MacPhail, in press) demonstrated a high negative correlation (r2>.94) between this variance and the levels of exposure that are estimated to result in adverse effects. Thus, these approaches clearly provide incentive for well-controlled experiments. Further studies in this regard are needed to directly compare dose-tolerance and effect tolerance approaches to quantitative risk assessment.