The micronucleus test and NTP rodent carcinogens: not so many false negatives.

In the study by Shelby et al. (1993) on 49 chemicals, the results of the micronucleus (MN) test in mouse bone marrow were compared with the results of the 2 year rodent carcinogenicity assays. Seven of the 25 rodent carcinogens were considered positive in the MN test, 5 following a protocol in which chemicals were given in three daily doses, and a further 2 when the chemical was administered only once. This low rate of positive results has led to disappointment in the MN test as a screen for carcinogens, but a careful examination of the data and of its analysis by Shelby et al. (1993) shows that many of the negative results are appropriate because: of the 18 carcinogens that were negative in the MN test, 1 has been retested and found to be non-carcinogenic, 9 were non-genotoxic and at least 2 were site-of-contact carcinogens not expected to be detected in the bone marrow. Two others were clearly positive in the MN test in other labs. Thus, the MN test 'missed' not 18 carcinogens, but 4 genotoxic carcinogens. The significance of these 4 needs further assessment, since three were liver specific carcinogens and the fourth was a very weak inducer of hemangiosarcomas in female mice only. Overall, the results of Shelby et al. (1993) do not cast such a shadow on the micronucleus test as many feared, and must be examined in the context of all the information available on each chemical. As Ashby and Tinwell emphasize in the accompanying article and in Tinwell and Ashby (1994), the data show that the MN test is capable of identifying human carcinogens and rodent germ cell mutagens, and remains a useful part of genotoxicity evaluation of chemicals.