Epigenetic carcinogens: evaluation and risk assessment.

Regulatory policies in the U.S. have been developed based upon a single model of cancer causation, which assumes chemical-induced genetic alterations. Such a model predicts some degree of cancer risk even at extremely low exposure levels. Many chemicals that produce tumors in experimental animals have been shown to act by epigenetic mechanisms that do not involve an attack by the chemical on DNA leading to subsequent genetic alteration. Such indirect mechanisms require prolonged exposures to high levels of chemicals for the production of tumors. For chemicals that are carcinogenic in this manner, the cancer mechanism would not be operative at exposures below a threshold at which the relevant cellular effect does not occur. Also, in contrast to DNA-reactive mechanisms, epigenetic effects may be unique to the rodent species used for testing. Certain chemical tumorigens have been well studied and provide examples for the use of mechanistic information in risk assessment. Butylated hydroxyanisole and saccharin are nongenotoxic food additives for which no risk to humans is predicted based upon low exposure levels and the likelihood that humans are either insensitive or much less sensitive to the tumorigenic effects found in rodent test species. For another non-genotoxic food additive d-limonene, the mechanism that underlies kidney tumor development in male rats is not expected to be operative in humans at all. The pharmaceutical phenobarbital represents a large group of non-genotoxic liver microsome enzyme inducers, which produce liver cancer in mice at levels that are near to therapeutic doses in humans. Epidemiology studies have not shown phenobarbital-related tumors in humans, indicating that humans may be less sensitive to the effects of phenobarbital. The mechanistic considerations involved in the risk assessment of these agents demonstrate that humans are not at risk from current exposure levels of many epigenetic carcinogens.