Methodological aspects of investigating gene-nutrient interactions.

Most diseases arise through the joint action of genetic susceptibilities and environmental factors. This is particularly true for cancers in which both types of factors have already been identified or strongly suspected. There may, however, be other as yet unidentified factors. Recent advances in molecular genetics will make it easier to study the gene– environment interactions in these diseases. Synergism (i.e. assessment of whether the effect of the genotype on risk is modified by the exposure level or whether the effect of exposure on risk is modified by the genotype) has important scientific and public health implications. Indeed, it could help to elucidate the mechanisms underlying these interactions and to evaluate the risks for individuals. The definition of interaction is dependent on a model and on how the effects are measured. When the effects are measured in terms of relative risks of disease, which is usually the case, interaction means that the joint effects of genetic and environmental factors cannot simply be multiplied. There are a number of potential situations: (1) a genetic factor and an environmental factor have already been evidenced and the aim is to specify their joint effect; (2) some genetic and environmental factors are only suspected of playing a role in the aetiology of the disease and some kind of interaction is thought to be hiding the main effects, which could be overcome by taking it into account; and (3) oblivious to potential causes, a search is made for a large number of factors in the hope of being lucky enough to find something. In the case of gene–nutrient interactions, all these situations are applicable. Although some genetic factors (Bonaı̈ti-Pellié, 1999; Olschwang, 1999) and the role of some nutritional factors (Franceschi, 1999; Giacosa et al., 1999) have already been identified, several uncertainties remain and there are numerous inconsistencies across studies on possible interactions between these factors (Little and Faivre, 1999). The objective of this paper is to review some available methods for evidencing genetic factors and assessing gene–environment interactions, to compare their respective advantages and limitations, and to consider the possible applications to the study of gene–nutrient interactions in colorectal cancer.