PubMed1989
At present, there is a whole array of risk factors, such as industrial chemicals, drugs, pesticides, complex chemical mixtures, physical and biological agents, for which there is a proven causal relationship with human cancer. The epidemiological studies are essential in providing proof for a causal relationship in humans. However, the epidemiological approach to the identification of the etiology of cancer is limited for two main reasons: only relatively high risks can be detected; and, secondly, the epidemiological surveys are based on observations of the effects as a consequence of an exposure that took place many years before. It can be estimated that epidemiological cancer research on common types of tumours is able to detect risk levels ranging from 5 x 10(-2) to 10(-3). This is, of course, far from commonly defined "acceptable risk" of 10(-5) - 10(-6). Frequently, negative epidemiological studies are thought to provide evidence of non-carcinogenicity, even though the exposure levels were so low that no such effect could even be expected. While the epidemiological studies have their greatest limitation in not being able to detect a low risk, that is to produce "false negative" results, carcinogenicity tests in experimental animals have been criticized in producing "false positive" results, i.e., in picking up cancer risks at very high exposure levels. Some investigators hold that the relatively high exposure levels given to experimental animals often results in carcinogenicity because of compensatory cell proliferation from toxic effects of the chemical, and accordingly, they have questioned the use of the Maximum Tolerated Dose (MTD). However, most of the chemicals determined to be carcinogenic would exhibit carcinogenic effects at exposure levels below the level of target organ toxicity. All epidemiologically proven human carcinogens which have been adequately studied, are also carcinogenic in laboratory animals. This does not necessarily mean that all animal carcinogens are human carcinogens. A more comprehensive evaluation of interspecies correlation between humans and rodents is limited by the lack of known human non-carcinogens. Nevertheless, the high interspecies correlation shown between rats and mice supports the view that extrapolation of carcinogenicity outcomes to other species, including humans, is appropriate. There is now a range of short-term laboratory tests which have been used to predict the results of long-term tests of carcinogenicity in mammalian species. Prediction is not entirely accurate, mainly because a fraction of carcinogens is active via a mechanism not employed by the short-term tests.(ABSTRACT TRUNCATED AT 400 WORDS)