Why preclinical models for cancer drug development fail

It is well established that the majority of anti-cancer agents identified and developed through preclinical studies in cell culture and animal models do not prove to be sufficiently effective in the clinic to move into later stage clinical trials. The simple explanations are that tumor cells in culture or implanted in mice cannot predict what will occur in patients, due in large part to the lack of pharmacokinetics in cell culture and because a mouse is not a miniature human being. Factors such as drug absorption, distribution, metabolism, and excretion (ADME) are likely to be markedly different in mice and humans, and cross-species ADME good laboratory practice (GLP) toxicology findings are often not fully incorporated into later investigational rodent studies. Furthermore, the frequent use of immune-deficient mice to host human tumors eliminates the critical involvement of the immune system. A colleague once remarked that we can cure virtually all cancers in mice. While this is certainly hyperbole, it is true that drug efficacy often appears significantly greater in rodent experiments than in humans. This article attempts to highlight and place in perspective many of the issues that limit the utility of preclinical models in common use for the development of antitumor drugs. We further identify factors that could and should be modified to improve their ultimate translation to the clinic, particularly given current efforts to replace the use of animal models with human cell-based and computer-based assays for drug development.