1988). For example, application of the MF may be regarded as a professional judgment that each test group contained too few animals.
The quantitative estimation of risks associated with low levels of exposure to carcinogens present in the environment is an important part of the regulatory process. At present, risk estimation methods are usually based on the assumption that the dose-response curve for carcinogenesis is linear in the low-dose region. This position is reflected in the principles proposed by the Office of Science and Technology Policy (1986), which stated:
When data and information are limited … and when much uncertainty exists regarding the mechanism of carcinogenic action, models or procedures which incorporate low-dose linearity are preferred. (OSTP, 1986)
This position is also reflected in the EPA's (1986) Carcinogen Risk Assessment Guidelines.
Although the Guidelines emphasize risk estimates derived using some form of linear extrapolation, they are based on the assumption that such estimates may be more appropriately viewed as plausible upper limits on risk and that the lower limit may well be effectively zero. The Guidelines also state that procedures for obtaining a best estimate lying somewhere between these two extremes generally do not exist at present, but that it may be possible to move away from a linearized extrapolation to obtain an upper limit in some circumstances and instead, use a threshold approach. For example, the EPA has recently taken a step in this direction by considering the possibility of a threshold for the induction of thyroid tumors (Paynter et al., 1988).
The EPA now uses the linearized multistage model for low-dose cancer risk estimation. The most important aspect of this practice is not the choice of the multistage model itself for risk estimation purposes but, rather, the linearized form of the model. Because the model is constrained to be linear at low doses, it is expected to yield risk estimates comparable to those based on other linear extrapolation procedures, including those proposed by Gaylor et al. (1987) and Krewski et al. (1990).
Additivity to background is often cited in support of the assumption of low-dose linearity in carcinogenic risk assessment. In the additive background model proposed by Crump et al. (1976) and Peto (1978),