Implications of the lack of accuracy of the lifetime rodent bioassay for predicting human carcinogenicity

The NTP lifetime rodent bioassay (LRB) is the "gold standard" for predicting human carcinogenicity. Unfortunately, little attempt has been made to validate it against human carcinogenicity. Here we show that the extremely limited data available do not support either of the two common interpretations of LRB results. If a risk-avoidance interpretation is used where any positive result in a sex/species combination is considered positive, 9 of the 10 known human carcinogens tested are positive, but an implausible 22% of all chemicals are positive. If a less risk averse interpretation is used where only chemicals positive in both rats and mice are considered positive, only 3 of the 6 known human carcinogens tested are positive. In either interpretation, some known human carcinogens are not positive in the LRB, potentially allowing widespread human exposure to misidentified chemicals. Improving the predictive accuracy of the LRB and other tests for human carcinogenicity requires that test results be validated against the known human carcinogenicity of chemicals. This will require redirecting available resources from screening chemicals to validating carcinogenicity tests as well as a substantial investment in epidemiology to identify more known human carcinogens and presumed human non-carcinogens. (C) 2003 Elsevier Science (USA). All rights reserved.