Objective: Germline mutations in the BRCA1 gene are associated with an increased risk of breast and ovarian cancer, but there is controversy about the true magnitude of these risks. Observed differences can arise from several sources, both genetic and methodological. To examine the efficiency and bias associated with different methods of risk calculation, we analyzed a single mutation in a large pedigree with known ascertainment.
Methods: Age-specific penetrance of breast and ovarian cancer was estimated using the Kaplan-Meier method and two likelihood-based approaches [maximum likelihood estimation, maximization of the logarithm of the odds (LOD) score]. Excess risk of other cancers due to the BRCA1 mutation was assessed.
Results: The estimated risk of breast and ovarian cancer at age 70 was 0.80 using the Kaplan-Meier approach and 0.55 using the maximum likelihood method. Both likelihood-based methods yielded similar results for the combined breast/ovarian phenotype, but using the maximum LOD score method lower estimates were obtained if only cancer at one site was considered. In the examined family, a high risk of ovarian cancer was found which might be an effect of the central location of the mutation within BRCA1. The risk of cancer at other sites than breast and ovaries was significantly elevated, but it was not possible to identify a single cancer site that could be said to be associated with the BRCA1 mutation.
Conclusion: Estimating the penetrance of a specific mutation using different approaches, we found that both the choice of study population and statistical method affect the magnitude of the estimates.