Genetic analysis of human breast cancer: implications for family study designs.

Genetic analysis of human breast cancer, as with many common diseases, raises several problems including sampling strategies, genetic heterogeneity, and gene-environment interactions. A reanalysis of 200 Danish breast cancer pedigrees, under the unified mixed model, was conducted to investigate more specifically these three points. We found that use of different sampling schemes leads to similar conclusions: familial transmission of breast cancer in this whole Danish sample cannot be accounted for by the Mendelian segregation of a dominant gene. Homogeneity tests, based on an a priori subdivision of the sample, were all nonsignificant under a given genetic model. However, it was possible to isolate a particular subgroup of pedigrees displaying only breast cancer, which was compatible with the segregation of a dominant gene. We have also shown that correct specification of a liability indicator according to epidemiological factors is of major importance to detect a major effect under the mixed model. Our results emphasize the need to design family studies including various types of information in the probands and family members to permit some progress in the understanding of complex diseases.