How useful are chronic (life-span) toxicology studies in rodents in identifying pharmaceuticals that pose a carcinogenic risk to humans?

After 20 years' experience of the chronic rodent bioassay and in the light of our current understanding of the mechanisms of carcinogenesis, there is now a wide recognition that the present strategies for assessing carcinogenic risk of pharmaceuticals are in need of urgent revision. The few pharmaceuticals known to cause cancer in humans each possess at least one of four properties: genotoxicity, immunosuppression, hormonal activity or chronic irritation. These properties can be identified by genotoxicity studies in vivo and well-designed toxicology studies of up to 6 months duration in rats. About 50% of all chemicals evaluated in the bioassay are positive. The findings from non-genotoxic compounds can often be shown to lack relevance for humans because either the exposure (daily dose and duration) is excessive when compared with actual human exposure or the response to a carcinogenic challenge is qualitatively different in rodents from that in humans. Many studies of the mechanisms of carcinogenesis in rats have elucidated why the data generated in rat bioassays are often not relevant to humans. However, the relative paucity of pharmacological data in mice means that similar studies of mechanism cannot be carried out so effectively in mice. The results of mouse bioassays are dominated by the frequency with which the liver (and, to a lesser extent, the lung) is a target organ. Molecular biology is now providing evidence of why chemically induced tumours in this organ cannot be extrapolated to human risk. Examination of the pattern of results in the NTP and IARC bioassay databases and the rodent data for 18 chemicals which appear thus far not to be carcinogenic in humans, indicates that data generated in mouse bioassays do not contribute to human carcinogenic risk assessment. It is argued that, for pharmaceuticals, only compounds with a novel mode of action need be studied for longer than 6 months. For such compounds, a study in rats at pharmacodynamically relevant dose levels for 12-18 months will, in general, provide adequate information for carcinogenic risk assessment in humans.