Current criteria to establish human carcinogens

Several national and international health agencies worldwide have established carcinogen identification programs with the aim of identifying the agents and exposures that contribute to the global burden of cancer. These programs have many features in common. IARC's program is described in some detail, with an emphasis on how evaluations can be changed by mechanistic data. Recently, several programs have expanded on the guidance they provide in assessing mechanistic data. The most comprehensive example is EPA's recent draft final Guidelines for Carcinogen Risk Assessment. In all programs, however, the principal role of mechanistic information has been to support the positive results observed in epidemiological studies or to discount the relevance of positive results observed in experimental animal bioassays. An alternative paradigm for carcinogen identification is proposed, one where mechanistic studies have a central role, rather than a supporting one. Under this paradigm, potentially carcinogenic agents would be identified by (1) identifying the key precursor events and processes involved in human cancer and (2) testing to see whether an agent can affect these events and processes. Under this paradigm, which is consistent with a multi-factorial view of carcinogenesis, it might be possible to identify carcinogens through mechanistic understanding alone, without waiting for epidemiological studies or 2-year carcinogenesis bioassays in rats and mice. This paradigm asks the question, "What is a human carcinogen? Is it an agent that we observe to induce tumors, or more generally, an agent with a clear role in tumor development?"