We propose to perform case-control studies of cancer within sets of monozygotic and dizygotic like-sexed twins discordant for specific neoplasms. Affected twins will be ascertained using record-linkage between population-based twin cohort rosters and population-based cancer registries in the Nordic countries (Denmark, Finland, Norway, Sweden) and using a registry based on newspaper solicitation of volunteer twins in North America. For the initial study of female breast cancer, over 2000 discordant pairs will be identified. Each twin will be asked to independently respond to a detailed questionnaire. The instrument will be applied in person when feasible and otherwise by mail. It will ask about the demographic and environmental characteristics of the families, the medical and environmental characteristics of the twinships, and medical histories and traits portinent to risk from the specific cancer. In the case od breast cancer, the latter will include measures of growth, development, obesity, reproductive history, and past consumption of various foodstuffs during youth and adulthood. In addition to absolute questions, the unique capacity of twins to provide information of a comparative (intra-twin) nature will be exploited, and it is expected that this will not only greatly reduce the effects of misclassification, but that it will provide information about mundane exposures from the distant past that non-twins cannot provide. Further reduction of misclassification will be achieved by confirmatory telephone interviews with mail respondants and with third family members. For twins interviewed in person, and for other twins of questionable zygosity for whom appropriate arrangements can be made, blood will be obtained for antigents, blood lipids, and female hormone studies. Standard methods of discordant matched pair analysis will be used. We will compute risk ratios, evaluate bias and confounding, and search for effect modification by age at exposure, age at diagnosis, histologic type, stage and bilateriality, presence of other risk factors such as family history and benign breast disease, zygosity, degree of lifetime closeness between twins, and other characteristics of twins and their lifetime environment. Similar studies of other sites will be planned.
Showing the most recent 10 out of 12 publications
