Approaches to the prediction of human cancer risk

Epidemiological studies provide the only direct means of measuring the risk of cancer in humans and for investigating effects directly avoiding interspecies extrapolation. Epidemiology is, however, too insensitive to detect small increases in risk. Furthermore, traditional epidemiology cannot be used prospectively to indicate future problems nor to explore the current exposure situation. Ideally, cancer hazards should be identified before humans are exposed. For the foreseeable future, therefore, identification of carcinogens will necessarily depend heavily on data derived from studies in experimental animals. Emerging molecular biological techniques, if used critically, hold great promise for resolving many contemporary puzzles in cancer risk prediction.