Estimating genetic influence on disease from population-based case-control data: application to cancers of the breast...

We describe genetic mixture models and goodness-of-fit statistics for evaluating the joint effects of genetic and environmental factors on the risk of chronic diseases. We focus particularly on situations wherein the gene(s) of interest play roles in several diseases, and death due to one disease can censor the occurrence of others. We use the methods to investigate the risks of cancers of the breast and ovary associated with germline mutations of BRCA1, using data pooled from three population-based U.S. case-control studies of ovarian cancer. We evaluate the goodness-of-fit of the genetic models by comparing the predicted numbers of diseased mother-daughter and sister-sister pairs to the numbers observed. We also use simulations to examine the performance of estimates obtained from such complex mixture models, and the contribution of control families to the precision of parameter estimates.