The case-control study is being used increasingly to evaluate cancer risks of common genetic variants. It is commonplace for detailed information on family history of disease to be collected in these studies. The rationale for this investigation is the belief that these data on family history are not utilized efficiently by investigators. Data on the family history of first degree relatives can be assembled in a retrospective """"""""kin-cohort"""""""", and analyzed using cohort-study methodology, whereby the incident rates of the cancer under study in relatives of gene """"""""carriers"""""""" are compared with the rates in relatives of """"""""non-carriers"""""""". In this application we show that the rate ratio from a kin-cohort analysis of this nature can be expressed as a function of the rate ratio of interest (i.e. the parameter that is directly estimated in the case-control study). Thus the kin-cohort analysis can be used, in principle, as an independent verification of any genetic associations observed in the case-control study. We present evidence to suggest that the kin-cohort might have good statistical power for this purpose for the study of common cancers. In this application we plan to study in detail the statistical power of the kin-cohort analysis relative to that of the case-control study from which the data for the kin-cohort are derived. We will examine the relative efficiency of the kin-cohort analysis in relation to the underlying incidence rate of the cancer under study, the population frequency and the relative risk of the genotype. The method will also be extended to the study of the joint effect of two genotypes, and the evaluation of haplotypes. It is anticipated that the project will demonstrate that for commonly occurring cancers the kin-cohort analysis can provide an immediate independent confirmation (or refutation) of genetic associations that are observed serendipitously in case-control studies.
