Generalizability of established prostate cancer risk variants in men of African ancestry.
Han Y, Signorello LB, Strom SS, Kittles RA, Rybicki BA, Stanford JL, Goodman PJ, Berndt SI, Carpten J, Casey G, Chu L, Conti DV, Rand KA, Diver WR, Hennis AJ, John EM, Kibel AS, Klein EA, Kolb S, Le Marchand L, Leske MC, Murphy AB, Neslund-Dudas C, Park JY, Pettaway C, Rebbeck TR, Gapstur SM, Zheng SL, Wu SY, Witte JS, Xu J, Isaacs W, Ingles SA, Hsing A, PRACTICAL Consortium, ELLIPSE GAME-ON Consortium, Easton DF, Eeles RA, Schumacher FR, Chanock S, Nemesure B, Blot WJ, Stram DO, Henderson BE, Haiman CA
Genome-wide association studies have identified more than 80 risk variants for prostate cancer, mainly in European or Asian populations. The generalizability of these variants in other racial/ethnic populations needs to be understood before the loci can be used widely in risk modeling. In our study, we examined 82 previously reported risk variants in 4,853 prostate cancer cases and 4,678 controls of African ancestry. We performed association testing for each variant using logistic regression adjusted for age, study and global ancestry. Of the 82 known risk variants, 68 (83%) had effects that were directionally consistent in their association with prostate cancer risk and 30 (37%) were significantly associated with risk at p < 0.05, with the most statistically significant variants being rs116041037 (p = 3.7 × 10(-26) ) and rs6983561 (p = 1.1 × 10(-16) ) at 8q24, as well as rs7210100 (p = 5.4 × 10(-8) ) at 17q21. By exploring each locus in search of better markers, the number of variants that captured risk in men of African ancestry (p < 0.05) increased from 30 (37%) to 44 (54%). An aggregate score comprised of these 44 markers was strongly associated with prostate cancer risk [per-allele odds ratio (OR) = 1.12, p = 7.3 × 10(-98) ]. In summary, the consistent directions of effects for the vast majority of variants in men of African ancestry indicate common functional alleles that are shared across populations. Further exploration of these susceptibility loci is needed to identify the underlying biologically relevant variants to improve prostate cancer risk modeling in populations of African ancestry.