A population-based analysis of clustering identifies a strong genetic contribution to lethal prostate cancer

Background: Prostate cancer is a common and often deadly cancer. Decades of study have yet to identify genes that explain much familial prostate cancer. Traditional linkage analysis of pedigrees has yielded results that are rarely validated. We hypothesize that there are rare segregating variants responsible for high-risk prostate cancer pedigrees, but recognize that within-pedigree heterogeneity is responsible for significant noise that overwhelms signal. Here we introduce a method to identify homogeneous subsets of prostate cancer, based on cancer characteristics, which show the best evidence for an inherited contribution. Methods: We have modified an existing method, the Genealogical Index of Familiality (GIF) used to show evidence for significant familial clustering. The modification allows a test for excess familial clustering of a subset of prostate cancer cases when compared to all prostate cancer cases. Results: Consideration of the familial clustering of eight clinical subsets of prostate cancer cases compared to the expected familial clustering of all prostate cancer cases identified three subsets of prostate cancer cases with evidence for familial clustering significantly in excess of expected. These subsets include prostate cancer cases diagnosed before age 50 years, prostate cancer cases with body mass index (BMI) greater than or equal to 30, and prostate cancer cases for whom prostate cancer contributed to death. Conclusions: This analysis identified several subsets of prostate cancer cases that cluster significantly more than expected when compared to all prostate cancer familial clustering. A focus on high-risk prostate cancer cases or pedigrees with these characteristics will reduce noise and could allow identification of the rare predisposition genes or variants responsible.