ABSTRACT
I. INTRODUCTION Prediction of cancer risk is typically undertaken to guide decision-making with the overall aim of reducing risk. Assessments performed for these purposes necessarily take a practical tack. Despite considerable research and much insight, our understanding about mechanism of action for nearly all carcinogens is limited. It has been the practice within the United States to assume that any exposure to a carcinogen produces some risk, and in predicting risk at low doses the doseresponse relationship has been assumed to be linear. Many now believe that use of this approach for certain “nongenotoxic” carcinogens results in a substantial overstatement of risk at low doses. Thus attempts are made to identify those carcinogens that are “low-dose” nonlinear or have thresholds above environmentally relevant exposure levels. This chapter emphasizes empirical approaches to the prediction of risk of carcinogens with genotoxic activity; a separate chapter in this monograph is devoted to biologically based mathematical models of carcinogenesis. Because of limited scientific knowledge, by necessity pragmatic approaches are taken in predicting risk beyond the doses and species for which observations have been made. It is within this context that the suggested modifications given in this chapter to the standard approaches to risk prediction are made.
