Model for estimation of clinically achievable plasma concentrations for investigational anticancer drugs in man.

A major problem related to in vitro testing of new investigational anticancer compounds is the lack of accurate pharmacokinetic data for the drugs in man. Based on the concept of a relationship between certain toxicologic endpoints in animal systems and maximally tolerated doses in man, we hypothesized that a comparable agreement might exist between these experimental animal toxicology data and clinically achievable peak plasma concentrations (PPCs). A retrospective analysis of the known data pairs of 28 commonly used cytotoxic compounds supports the existence of a reasonably good correlation between ip LD50 values (Pearson test) in nontumored mice and PPCs in man (r2 = 0.501; P less than 0.0001). Our data suggest that by use of the resultant statistical regression model a rational starting point can be selected for in vitro screening of entirely new agents for which human PPCs are not available. Additionally, application of this approach may also prove useful in relation to selecting in vitro doses for chemosensitivity assays intended for predictive correlation with clinical response in cancer patients.