Functional partitioning of Prostate Cancer heritability in European Americans and African Americans from AAPC and BPC3 consortia reveals tissue specific regulation. B. Pasaniuc1,2, A. Gusev3, F. R. Schumacher4, S. Lindstrom3, M. Pomerantz5, F. Li5, H. Long5, P. Kraft3,6, A. L. Price3,6, M. Freedman5,7, C. A. Haiman4, The BPC3 Consortium, The AAPC Consortium 1) Pathology and Laboratory Medicine, Geffen School of Medicine at UCLA, Los Angeles, CA; 2) Human Genetics, Geffen School of Medicine at UCLA, Los Angeles, CA; 3) Program in Genetic Epidemiology and Statistical Genetics, Harvard School of Public Health. Boston MA; 4) Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA; 5) Medical Oncology, Dana-Farber Cancer Institute, Boston MA; 6) Dept. Epidemiology, Harvard School of Public Health. Boston MA; 7) Department of Medicine, Harvard Medical School, Boston, MA.
Although GWAS have identified over 90 genetic loci associated with prostate cancer, they jointly explain a fraction of the overall genetic component of the risk, leaving much of heritability missing. Here we use variance components models to dissect the genetic contribution to risk of prostate cancer across various functional categories identified in the ENCODE project using genome-wide array SNP data across ~20,000 prostate cancer cases and controls of European and African-American ancestry from the BPC3 and AAPC consortia. We find that 1000 Genomes imputed SNPs explain a heritability of h2g=0.19 (s.e. 0.04) in Europeans, a significant increase from heritability explained by the known risk variants in this data (0.08, s.e. 0.02). Interestingly, we find that typed and imputed SNPs explain similar amount of variance in prostate cancer in African Americans (h2g=0.23, s.e. 0.04) suggesting a similar genetic architecture to the risk of prostate cancer across the two ethnicities (meta h2g=0.21, s.e. 0.03, 36% of total total h2=0.58 [Hjelmborg et al 2014]). We partition heritability explained by SNPs across ENCODE functional elements and find that variants in DNaseI Hypersensitivity Sites (DHS) specific to Prostate Adenocarcinoma LNCaP cell lines that span only 1.1% of the genome explain 29% of h2g (s.e. 13%, 25-fold enrichment). We also explored the role of Androgen Receptor (AR) binding sites in prostate tissue, (which overlap by 33% with the LNCaP DHS) and found that variants in these regions (1.0% of genome) explain 37% of h2g (s.e. 11%, 39-fold enrichment). In comparison, coding variants that account for 0.7% of the genome only explain 5% of h2g (s.e. 8%, 8-fold enrichment). Neither LNCaP DHS nor AR-binding variants were significantly enriched in a meta-analysis of 11 common non-cancer traits from the WTCCC, serving as a negative control and supporting the cancer specificity of the tested functional annotations. Analyses of admixed populations present complexities and we show by simulation that our estimates of enrichment are robust to population structure and population-specific selection. Overall, our results demonstrate a significant contribution of phenotype-specific regulatory variants to the genetic risk for prostate cancer across different ethnic groups (no significant differences were observed between the two ethnicities in any analysis above).