Identification of candidate target genes for prostate cancer risk-SNPs utilizing a normal prostate tissue eQTL dataset. S. N. Thibodeau1, A. J. French1, S. K. McDonnell2, J. C. Cheville1, S. Middha2, S. M. Riska2, S. Baheti2, Z. C. Fogarty2, L. S. Tillmans1, M. C. Larson2, N. B. Larson2, A. A. Nair2, D. R. O'Brien2, J. I. Davila2, Y. Zhang2, L. Wang3, J. M. Cunninghman1, D. J. Schaid2 1) Dept Lab Med & Pathology, Mayo Clinic, Rochester, MN; 2) Health Sciences Research, Mayo Clinic, Rochester, MN; 3) Medical College of Wisconsin, Milwaukee, WI.
Prostate Cancer (PC) is the most frequently diagnosed solid tumor in men in the U.S. For PC, multiple genome-wide association studies have now been performed yielding a substantial number of well-validated SNPs that are associated with an increased risk of PC. A significant problem for many of the PC risk-SNPs identified, however, is that they do not lie within or near a known gene and they have no obvious functional properties. These findings suggest that many of these risk-SNPs will be located in regulatory domains that control gene expression. However, in order to define the functional role of these non-coding risk-SNPs, the target genes must first be identified. A frequently used strategy to address this problem involves the use of expression quantitative trait loci (eQTL) analysis. In this study, we created a tissue-specific eQTL dataset and then applied this dataset to 123 well-established PC risk-SNPs in an effort to identify candidate target genes. To construct this dataset, normal prostate tissue from over 4000 men having a radical prostatectomy for PC was histologically screened in order to identify approximately 500 samples meeting a strict selection criteria: tissue localized to the posterior region of the prostate, no tumor, no high-grade PIN, no BPH, 2% lymphocytes, and 40% epithelial cells. Genome-wide genotypes and genome-wide mRNA expression levels were obtained with the use of the Illumina Human Omni 2.5M SNP array and by RNA sequencing, respectively. Of 500 processed samples, 471 samples passed stringent QC and were available for further analysis. Our primary analysis focused on identifying eQTLs for 123 PC risk-SNPs, including all SNPs in linkage disequilibrium with each risk-SNP (r2 >0.5), resulting in 78 unique risk-intervals. Furthermore, we focused on cis-acting associations only where the transcript was located within a 2Mb region (+/-1Mb) of the risk-SNP interval. Of all SNPs located in the 78 risk-intervals (N=5116 SNPs), 1002 demonstrated a significant eQTL signal after adjustment for sample histology (% lymphocytes and % epithelial cells) and meeting a Bonferroni-adjusted p-value threshold of 1.96e-7 (ranged from 1.52e-91). Of the 78 PC risk-intervals, 22 (28%) demonstrated a significant eQTL signal and these were associated with 43 genes, supporting a number of novel candidate susceptibility genes for PC. Mapping of the causative risk-SNPs and their corresponding affected regulatory elements is currently in progress.
You may contact the first author (during and after the meeting) at