Familial heterogeneity in breast cancer predisposition: a study of 22 Utah families

The problem of “missing heritability” in genome-wide analyses of complex diseases is thought to be attributable to some combination of: rare variants of moderate to large effect, common variants of very small effect, and epigenetic, epistatic, or shared environmental effects. Rare variants do not affect large numbers of people by definition, but identified genes and pathways frequently lead to important insights into pathogenesis, and become targets of chemoprevention or therapy. Family studies remain an efficient way to identify rare variants with sizable effects on disease risk. We present a genome-wide study of breast cancer in 22 large high-risk families including 154 women diagnosed with breast cancer. Appropriate marker spacing was achieved by simulation studies of founder haplotypes to reduce the chance that linkage disequilibrium produced spurious linkage peaks. For each family, we generated 100 simulations of null linkage genome-wide to estimate the probability that individual results were due to chance. We identified a total of 12 putative susceptibility regions with per-family genome-wide probability < 0.05. These regions were located on 10 chromosomes; 10 of the 22 families showed linkage at these locations; two or more families showed linkage to 6 regions on 5 chromosomes (4q, 5q, 6p, 14q, 18p, and 18q). These results indicate that there is considerable heterogeneity among families in genomic regions and thus variants predisposing to breast cancer. Moreover, they suggest that uncommon high– or medium-risk genetic variants remain to be found, and that family designs can be an efficient way to identify them.