Animal toxicology for early clinical trials with anticancer agents.

Abstract

This analysis was carried out to assess quantitative and qualitative relationships between animal and human toxicology data with anticancer drugs. Among 21 chemotherapeutic agents, one-sixth LD10 in the mouse or one-third toxic dose low (TDL) in the dog corresponded to acceptable doses in man when experimental and clinical data were obtained at identical schedules and compared on a mg/m2 basis. The mouse and the dog largely differed in their tolerance to individual drugs. One-tenth LD10 in the mouse seemed always tolerated in the dog. On the average, these species were equally relevant for establishing the initial dose in man. A similar number of dose escalation steps would have been required in phase I clinical trials if the starting dose had been based on one-tenth LD10 in the mouse or the lowest value of one-sixth LD10 in the mouse and one-third TDL in the dog. These observations indicate that the starting dose in phase I clinical trials could be safely and efficiently based on one-tenth LD10 in the mouse. Prior verification that the resulting dose is not lethal or life-threatening in the dog could add further safety to this procedure. The predictive value (PV+) in man of organ system toxicity in animals depends upon the prevalence of this toxicity in man. In our study, PV+ was high (greater than 0.85) for common toxic effects in man, i.e., gastrointestinal intolerance and myelosuppression. PV+ declined dramatically (0.05 to 0.54) with rarer toxic manifestations. There was no clear superiority of animal findings over the mere knowledge of the prevalence of these findings in man. Thus, it would appear that routine and undiscerned investigation of organ system toxicity in animals is of questionable usefulness for the clinician experienced in early clinical trials with chemotherapeutic agents.