A critical appraisal of the sensitivity of in vivo genotoxicity assays in detecting human carcinogens.

Abstract

Genotoxicity testing is an important part of standard safety testing strategies. Animal studies have always been a key component, either as a mandatory part of the regulatory test battery, or to follow-up questionable in vitro findings. The strengths and weaknesses of in vivo assays is a continuous matter of debate, including their capacity to predict (human) carcinogenicity. We have therefore analysed the sensitivity of five routinely used in vivo tests to determine, in addition to other aspects, which tests or combination of tests best identify 73 chemicals classified as IARC Group 1 and 2A carcinogens. The in vivo tests included the micronucleus (MN), unscheduled DNA synthesis (UDS), comet, Pig-a and transgenic rodent assays (TGR). The individual assays detect 74.2% (49/66, MN), 64.3% (9/14, UDS), 92.1% (35/38, comet), 82.4% (14/17, Pig-a) and 90.3% (28/31, TGR) of the probable and confirmed human carcinogens that were tested in these assays. Combining assays that cover different genotoxicity endpoints and multiple tissues, e.g. the bone marrow MN and the liver comet assays, increases the sensitivity further (to 94%). Correlations in terms of organ-specificity for these assays with human cancer target organs revealed only a limited correlation for the hematopoietic system but not for other organs. The data supports the use of the comet and TGR assays for detection of 'site-of-first-contact' genotoxicants, but these chemicals were generally also detected in assays that measure genotoxicity in tissues not directly exposed, e.g. liver and the hematopoietic system. In conclusion, our evaluation confirmed a high sensitivity of the five in vivo genotoxicity assays for prediction of human carcinogens, which can be further increased by combining assays. Moreover, the addition of the comet to the in vivo MN test would identify all DNA reactive human carcinogens. Importantly, integration of some of the study readouts into one experiment is an animal-saving alternative to performing separate experiments.