Relationship between the shape of dose-response curves and background tumor rates

Various authors have argued that a chemical which augments a carcinogenic process, which is already producing tumors spontaneously, will produce an increase in tumors no matter how small the dose. For this situation, no threshold dose exists whether or not the chemical is genotoxic. Under such conditions, it is expected that the dose-response will contain a linear term. A large database of animal bioassays (Gold et al., 1984) was examined to study the relationship between the shape of dose-response curves and the background tumor rate. The multistage model was fit to 143 data sets from 75 different chemicals. As expected, the presence of the linear term was correlated with the background tumor rate and the presence of higher degree terms in dose (cubic or greater) decreased as the background rate increased. This examination of a large number of chronic bioassay results appears to support the premise that low dose linearity is generally expected for tumor sites where background tumors occur, even for carcinogens which were negative by the Salmonella genotoxicity test.