Genetic epidemiology of early onset breast cancer

Risks for breast cancer when there is a family history of the disease are usually calculated using data from segregation analyses which favour a single dominant gene with high penetrance. There are, however, at least three loci known to be associated with familial breast cancer (p53, BRCA1, and an as yet unpublished locus) and the frequencies and penetrances of these genes are not likely to be the same. We have attempted to address the problem of which genetic parameters should be used to calculate risks for different patterns of familial breast cancer. Data from 384 nuclear families ascertained through a proband selected for early onset breast cancer were subjected to complex segregation analysis, correcting for ascertainment bias resulting from selection for severe phenotype. Age of onset of breast cancer, incorporated as severity, provides additional information to the segregation model over and above that given by assigning liability classes on the basis of age at observation. The use of this additional parameter in the analysis is described. There is fair agreement between estimates from this sample and previous predictions from consecutive probands and consultands. The differences suggest more than one rare dominant gene for susceptibility to breast cancer, with different penetrances. Although refinements of segregation analysis will help to delineate these different genes, perfect resolution will require identification of the mutant alleles. Methods to estimate genetic parameters under genotype specific mortality need to be developed. Meanwhile, we suggest that high and low estimates of penetrance be used in risk estimation for genetic counselling, and as a guide to candidates for entry into clinical trials of screening and chemoprevention in breast cancer.