Dietary practices, site-specific tumors, and age-at-death relationships.

When rats are given freedom of dietary choice, the quantity and composition of the diet selected varies from animal to animal, but the risk to the individual of developing a variety of neoplasms is greatly increased over that of rats fed fixed diets. The increase in frequency of tumors cannot be explained on the basis of the amount or the composition of the diet selected. However, the application of "variable selection" statistical procedures shows that there are combinations of temporal-specific interacting dietary and dietary-dependent variables that are closely linked with cancer mortality. The multi-variable statistical model for animals dying with one type of tumor is not applicable to those dying with tumors of other types or sites. In every case, the only data required to characterize the dates of death of tumor-bearing individuals are those describing the dietary practices and growth responses prior to maturity. The model for animals with pancreatic tumors is used to illustrate the extent to which dietary practices of early life are linked with the animals' survival.