Abstract
Dose-response relationships for chemical tumor induction can be described with stochastic carcinogenesis models by incorporation of dose-dependent parameters. We used a two-stage mutational model with clonal expansion of intermediate cells to mimic the observed U-shaped (more precisely: J-shaped) dose response for cell division in the forestomach of rats treated with caffeic acid. The question was whether biologically acceptable model parameters for the rates of mutation and cell turnover could be found to fit the experimental data, including tumor incidence in controls and at high dose. Using rates of cell turnover in proportion to the experimental labeling indices for DNA replication, model parameters could fit the tumor incidences within the 95% confidence limits of the experimental data, and the J-shape was retained quantitatively. However, when the smaller confidence limits derived from historical data on the spontaneous forestomach tumor incidence were taken into account, the original model could no longer accommodate all data and had to be expanded. For that, mutation rates were considered to be proportional to both the level of DNA damage and the rate of cell division, and both factors were allowed to be dose dependent. For the DNA damage, an increment over background was used at the higher, cytotoxic dose levels. This model improvement resulted in biologically acceptable parameters to fit all available data, demonstrating its usefulness for further analyses of hormetic dose response.