Mechanistic basis for nonlinearities and thresholds in rat liver carcinogenesis by the DNA-reactive carcinogens 2-acetylaminofluorene and diethylnitrosamine.

To explore differences in mechanisms of carcinogenicity at low and high exposures, we have conducted a series of exposure-response studies of hepatocarcinogenesis in rats using 2 well-studied DNA-reactive carcinogens, 2-acetylaminofluorene and diethylnitrosamine. In these studies, we have used intraperitoneal injection or intragastric instillation to deliver exact doses during an initiation segment followed by phenobarbital as a liver tumor promoter to enhance manifestation of initiation. This protocol results in carcinogenicity comparable to that produced by lifetime exposure to the carcinogens. Our findings in these experiments provide evidence for the following: (a) formation of DNA adducts can be nonlinear, with a plateau at higher exposures; (b) cytotoxicity shows no-effect levels and is related to exposure; (c) compensatory hepatocyte proliferation shows no-effect levels and can be supralinear at high exposures; (d) formation of preneoplastic hepatocellular altered foci can show no-effect levels and appears supralinear at high exposures; (e) no-effect levels can exist for tumor development, and the exposure response can be supralinear. We interpret these findings to reflect thresholds for hepatocellular initiating effects of these carcinogens and exaggerated responses at high exposures attributable to cytotoxicity and compensatory hepatocyte proliferation. Such enhanced proliferation of hepatocytes harboring DNA damage likely results in an exaggerated yield of mutations in critical genes, leading to supralinear initiation of carcinogenesis. Thus, mechanisms differ between low and high exposures. Based on these observations, we suggest that linear extrapolation from high toxic exposures to postulated low-exposure effects of DNA-reactive carcinogens can yield overestimates. Such extrapolation must be supported by mechanistic information. The finding of no-effect levels provides a basis for understanding why low-level environmental exposures of humans to even DNA-reactive carcinogens may convey no cancer risk.