Absolute vs. Relative Risk as a Basis for Cross-Species Prediction of Carcinogen Potency. L. R. Rhomberg, Harvard Center for Risk Analysis, Boston, MA 02115; I. Linkov, A. Shagiahmetov, and R. Wilson, Department of Physics, Harvard University, Cambridge, MA 02138; and G. M. Gray, Harvard Center for Risk Analysis, Boston, MA 02115
Although epidemiologists have long considered both absolute and relative risk models, it has been the practice when extrapolating animal bioassay risks to humans to project only measures of absolute risk above background. Since the stronger animal tumor responses are frequently at sites with a high spontaneous rate, and since humans typically have lower spontaneous tumor rates at these sites, there is concern that absolute risk elevation may overstate human risks by failing to account for (a) especial sensitivity to induction of certain tumors among experimental rodent strains or (b) interaction of agents with different, species-specific rates of the processes responsible for spontaneous carcinogenesis. If such quantitative differences are indeed critical, risk elevation measured as a fraction of background rate (i.e., relative risk) should be a better predictor of carcinogen potency across species than is absolute risk elevation. We investigate this proposition through analysis of site-specific carcinogenic potencies for a large number of agents as revealed in bioassays from the National Toxicology Program.
Go to . . .