Traditional approaches and study design in cancer epidemiology have not been very successful in identifying and evaluating adequately the potential risk and/or protective factors associated with the disease. The main reasons for the failure are often due the small study sample size, and inadequate exposure information. In this paper, issues and approaches relevant to these two challenges are discussed. Multicentre study is proposed as a way to increase study size and to mitigate criticism about meta-analysis of independent studies. A multicentre study of large cohort or case-control studies also offer an exciting opportunity to study the contribution of epigenetic events that may be associated with lifestyle and environmental risk factors for human health. Optimizing methods for exposure assessment and how to reduce exposure to misclassification represent a difficult component in epidemiological studies. A potentially useful approach for improving exposure estimation is to rely on biomarkers of exposures. An example is provided to demonstrate how biomarkers of exposures could provide valuable information in addition to exposure measurements in traditional epidemiological studies. Finally, it is argued that risk assessment and the precautionary principle should not be viewed as conflicting paradigms but, rather, as a complementary approach for developing appropriate policies to address risks posed by exposure to carcinogens and a wide spectrum of other health hazards.
Keywords: cancer epidemiology; carcinogenicity; epidemiological data; epigenomics; genomics of cancer; multicentre studies; precautionary principle; risk assessment.