Background: Cases with a family history are enriched for genetic risk variants, and the power of association studies can be improved by selecting cases with a family history of disease. However, in recent genome-wide association scans utilizing familial sampling, the excess relative risk for familial cases is less than predicted when compared with unselected cases. This can be explained by incomplete linkage disequilibrium between the tested marker and the underlying causal variant.
Methods: We show that the allele frequency and effect size of the underlying causal variant can be estimated by combining marker data from studies that ascertain cases based on different family histories. This allows us to learn about the genetic architecture of a complex trait, without having identified any causal variants. We consider several validated common marker alleles for breast cancer, using our own study of high risk, predominantly bilateral cases, cases preferentially selected to have at least two affected first- or second-degree relatives, and published estimates of relative risk from standard case-control studies.
Results: To obtain realistic estimates and to accommodate some prior beliefs, we use Bayesian estimation to infer that the causal variants are probably common, with minor allele frequency >5%, and have small effects, with relative risk around 1.2.
Conclusion: These results strongly support the common disease common variant hypothesis for these specific loci associated with breast cancer.
Impact: Our results agree with recent assertions that synthetic associations of rare variants are unlikely to account for most associations seen in genome-wide studies.
©2011 AACR.