Kaplan-Meier tumor probability as a starting point for dose-response modeling provides accurate lifetime risk...

In rodent carcinogenicity studies the linearized multistage model for modelling the dose-response for specific tumor incidence has limitations in accuracy. This note provides an alternative basic method for analyzing the dose-response relationship. It is based on an actuarial analysis of mortality and specific tumor incidence. The survival and the Kaplan-Meier specific tumor probability are fitted to a Weibull model, in which exposure level, exposure period, and observation period are independent variables. The mortality from specific cancers at a certain time is simulated by means of the product of survival and specific tumor rate (derivative of Kaplan-Meier tumor probability) as function of exposure level, exposure duration, and observation period, integrated over the observation period. The model is demonstrated by means of fitting the mortality and tumor incidence data from the second NTP mice study on butadiene to a Weibull model and to the linear, so-called, one-hit model. It will be shown that, in the experimental exposure range, the Weibull model is far superior to the one-hit model and predicts the specific tumor incidence with a high accuracy over the total dose range. The Kaplan-Meier probability model for a specific tumor is also useful for regulatory risk estimation. It is proposed that to develop a specific tumor a risk level of 1 in 1,000 over a lifetime is about equal to 5 in 10,000 at 50% survival of the population. The Kaplan-Meier probability may be estimated at the time of 50% survival of the exposed population, which can be deduced from the all mortality data. This estimation method provides meaningful data, using exposure level, exposure duration, and observation period properly. The advantage of the actuarial analysis method for interpreting rodent studies is that allowance is made for competition between death causes, which is essential in case of considerable difference in mortality and specific mortality between dose groups. Integrating the product of survival and specific tumor rate is the proper way to predict, comparatively, mortality and specific mortality in exposed and unexposed rodent populations.