Genome-wide association study of 35K men with 300K prostate specific antigen measures identifies numerous novel loci: potential for personalized screening for prostate cancer. J. S. Witte1,2, T. J. Hoffmann1,2, L. Sakoda3, E. Jorgenson3, D. S. Aaronson3, J. Shan3, L. A. Habel3, J. C. Presti3, C. Schaefer3, N. Risch1,2,3, S. K. Van Den Eeden3 1) Institute for Human Genetics, UC San Francisco, San Francisco, CA; 2) Department of Epidemiology and Biostatistics, UC San Francisco, San Francisco, CA; 3) Kaiser Permanente, Division of Research, Oakland CA.

   Using the prostate-specific antigen (PSA) test to screen for prostate cancer is controversial: it has modest predictive value and the potential to lead to over-diagnosis and over-treatment. One can improve the traditional single-cutpoint PSA screening test with more personalized thresholds to determine whether a man should be further evaluated for prostate cancer. Since there is a clear genetic component to PSA, integrating information about genetic factors that influence PSA independently of prostate cancer may improve test performance. To this end, we have undertaken a very large genome-wide association study of PSA concentrations among 35,520 men free of prostate cancer in the Kaiser Permanente Genetic Epidemiology Resource in Adult Health and Aging (GERA) cohort. These men had a total of 295,398 PSA measures from electronic health records. We evaluated the potential association between genome-wide variants and log(PSA) levels using a longitudinal generalized estimating equations model, adjusting for age, body mass index, and genetic ancestry. We detected over twenty independent genome-wide significant loci for PSA levels. Four of these have previously been reported as associated with PSA, and five with prostate cancer. The PSA replications were at KLK3 (p = 3.6x10-151), MSMB (p = 5.7x10-33), HNF1B (p = 1.1x10-16), and TBX5 (p = 3.4x10-12). The hits we detected for PSA that have previously been reported as associated with prostate cancer were at JAZF1 (p = 7.3x10-18), 8p21.2 (p = 4.6x10-23), 8q24 (p = 7.5x10-15), 10q26.12 (p = 1.6x10-36), and 11q22.2 (p = 8.5x10-9). The strongest novel locus was on chromosome 13 (p = 9.8x10-19), and we detected numerous other significant novel loci for PSA concentrations, including on chromosomes 1, 2, 6, 9, 10, 14 and X. In addition, we detected over 30 suggestive hits that merit further follow up (p < 10-5). The variants detected here may explain some of the inherent variability inand the biological basis ofserum PSA concentrations. We are presently evaluating how much incorporating these variants into the assessment of serum PSA concentrations improves the decision of whether to perform a prostate needle biopsy and the resulting prostate cancer outcome. The large number of loci we have detected substantially helps our efforts here. This is an important step toward incorporating genetic markers into PSA screening, with the ultimate goal of devising personalized PSA tests for use in the clinic.

You may contact the first author (during and after the meeting) at