A decision tree approach for carcinogen risk assessment.

When presented with the information that an environmental chemical produces cancer in animals, the default approach is to apply the linearized multistage risk model with no mechanistic information. The decision tree presented here outlines a straightforward strategy in the collection of information relevant to risk assessment. Three situations are noted where a different approach to risk assessment should be considered: (1) a tumor response that is not relevant to the human situation; (2) chemicals acting through a nongenotoxic/cytotoxic mode of action; and (3) chemicals acting through a nongenotoxic/mitogenic mode of action. Usually upon learning that a chemical produced a rodent tumor response, no additional research is done because of the high costs in money, time, and personnel to conduct such studies. Compounding this situation is a widely-held view that no amount of mechanistic information will be sufficient to convince regulators to depart from the default risk assessment. Hence, there are very few data sets to confirm and refine the above suggestions. There should be incentives to conduct the research required to obtain the type of information outlined in the decision tree approach presented here. All involved with these issues agree that the gathering and use of scientific information should be encouraged. Hopefully, these suggestions will provide a means of furthering that goal.