A Standardized Benchmark Approach To the Use of Cancer Epidemiology Data for Risk Assessment

Clarity and scientific validity are two criteria for assessing the quality of communications between scientists and risk managers. Regulating permissible exposures on the basis of very-low-dose risk extrapolation uses scientific information that may not meet either criterion. With regard to clarity, it is difficult for an individual to conceive of the meaning of risks on the order of 1 in 1,000,000 lifetime excess. With regard to scientific validity, the uncertainties of extrapolating risks at very low doses are evident in the wide variation in results produced depending on the statistical and pharmacokinetic assumptions made. An alternative approach is to fix benchmarks from which safety factors are chosen (Gaylor, 1983). For epidemiological data, we propose a benchmark of that exposure which would cause 1% cancer excess with 10 yr of exposurefollowed by 30 yr of further followup. These values have been chosen at this stage of development since they are close to observable values in many cancer epidemiology studies. In the first stage, excess cancer risk versus duration of exposure is plotted and the excess risk is estimated for a exposure duration of 10 yr. The next step involves a short linear extrapolation from the exposure levels linked with this excess risk to the exposure levels that would cause a 1% excess risk over a duration of 10 yr. We have established preliminary benchmarks from published data for benzene and radon daughter exposure. Permissible exposure levels might then be set by deciding on safety factors. Although low-dose extrapolation will aid in selecting a safety factor for carcinogenic exposures, it should be acknowledged that the actual risk at the permissible level is not known. The risk might be zero, it might be between zero and a linear extrapolation value, or it might possibly be a little higher than the value from a linear projection (Gaylor and Kodell, 1980; Gaylor, 1983).