Abstract
Tests to evaluate toxicity should be consistent and reproducible. These crucial elements have been brought into question because the controls of near-replicate experiments have significant variability in parameters such as survival and common tumor incidences critical to safety evaluation, even in some of the best controlled assays, those conducted by the National Toxicology Program (NTP). To minimize the impact of these variabilities, dietary control (DC), the control of body weight (BW) growth by nutritionally adequate dietary manipulation, has been suggested for chronic studies. The target BW growth curve for the DC should be fast enough to allow the animal to develop tumors, but slow enough to prevent too great a tumor load in controls. In order to better define these targets in Fischer 344 rats, relationships have been developed between various BWs at different times-on-test, incidences of common tumors in control animals from NTP bioassays initiated from January 1981 and reported by June 1996, as well as survival, with the adjustments for different study types used in the chronic assays. These rat studies were complemented by using a set of similar studies in B6C3F1 mice that appear to have "restricted-like" properties. Data from the experiments using dietary restriction (DR) appeared to be better explained using data derived from these experiments rather than from studies with normal BW growth curves. When the DR data are combined with these "restricted-like" experimental results, the capacity to predict the consequences of DC for tumor incidences is improved.