Bayesian methods for a three-state model for rodent carcinogenicity studies

The objective of a chronic rodent bioassay is to assess the impact of a chemical compound on the development of tumors. However, most tumor types are not observable prior to necropsy, making direct estimation of the tumor incidence rate problematic. In such cases, estimation can proceed only if the study incorporates multiple interim sacrifices or we make use of simplified parametric or nonparametric models. In addition, it is widely accepted that other factors, such as weight, can be related to both dose level and tumor onset, confounding the association of interest. However, there is not typically enough information in the current study to assess such effects. The addition of historical data can help alleviate this problem. In this article, we propose a novel Bayesian semiparametric model for the analysis of data from rodent carcinogenicity studies. We develop informative prior distributions for covariate effects through the use of historical control data and outline a Gibbs sampling scheme. We implement the model by analyzing data from a National Toxicology Program chronic rodent bioassay.

The effect of different tumor groupings on findings of anticarcinogenic responses in long-term rodent bioassays

Many investigators have found that there is a decrease in tumor rates at some sites when rodents are exposed to some chemicals. The generality of this finding of anticarcinogenicity has been questioned. In this study, we evaluate the effect of several alternative ways of grouping the 3000 tumor types in the Cancer Bioassay Data System (CBDS) database of carcinogenesis test results into a limited number of classes on findings of anticarcinogenicity. We also study the influence of random variation of tumor rate on the apparent anticarcinogenic effects of specific chemicals. We compare the numbers of chemicals classified as anticarcinogenic in (1) our "standard" classification system, (2) a modification of that system to correct some deficiencies in the CBDS database pointed out by Dr. J. Haseman, (3) an alternative classification system developed by Dr. K. S. Crump and colleagues, (4) the number of animals displaying at least one tumor, or (5) the total number of tumors appearing in all animals in control and dosed groups. Although there is a difference in the number of chemicals classified as anticarcinogens by these alternative classification schemes, all of them show a statistically significant increase in the number of anticarcinogenic responses above the random rate predicted by a Monte Carlo simulation of the rodent bioassay. The number of anticarcinogenic responses is similar in our standard classification, our modified classification, and a classification scheme developed by Crump, though specific site/organs may differ. This arises because these schemes use approximately the same number of tumor groups (about 100). If the number of tumor groups decreases (for example, in the total tumors or tumor-bearing animal schemes), the number of anticarcinogenic responses decreases because of the decrease in overall sensitivity of the test. A scheme which combines tumor types can be useful by integrating carcinogenic and anticarcinogenic responses to exposure. At the same time, combination may hide important biological information. We urge consideration of both increases and decreases when evaluating the likely human effects of exposure.