Relation of preclinical toxicology to findings in early clinical trials.

The evolution of the Division of Cancer Treatment's preclinical toxicology protocols over the last decade, a description of the current protocol, and the progress of this protocol regarding quantitative and qualitative relationships to current clinical findings are presented in this report. Data are reviewed for seven experimental antineoplastic drugs. Preliminarily, the following conclusions can be made from preclinical and clinical toxicology experience with these first seven drugs. The 1/10 MELD10 establishes a safe human starting dose. Six of seven drugs could have started at 1/10 MELD10 with no adverse effects. Toxicity data from the beagle dog did not effectively predict whether the 1/10 MELD10 dose level represents a hazard to humans. Clinical dose escalation procedures can not be efficiently predicted from currently acquired preclinical information. Toxicity studies in dogs were effective in disclosing the human dose-limiting toxic effects and approximated the maximally tolerated dose for six of seven drugs. In the exceptional case, the dose level predicted for human toxicity was grossly underestimated in both experimental species. We conclude that the preclinical toxicity data from the beagle dog are valuable in predicting the potential risk to humans, even though the safety of the entry-level doses is occasionally underpredicted.