Abstract 2747: Application of a novel multi-locus test for genetic association incorporating gene-gene interaction suggests functionality for multiple susceptibility loci for prostate cancer

The discovery of gene-gene interactions in genome-wide association studies (GWAS) can generate hypotheses about the biologic process through which multiple susceptibility loci influence the pathogenesis of disease. We present an extension of a previously proposed score-test for disease association that allows for interaction between sets of biologically similar single nucleotide polymorphism (SNPs). The score-test gains power over alternative methods by incorporating a parsimonious one-degree-of-freedom model for gene-gene interaction. Our proposed extension allows for additional power gains in case-control studies via use of a retrospective likelihood that can exploit an assumption of gene-gene independence for the underlying population between unlinked sets of SNPs. Simulation studies were used to validate the performance of this method. It was found to maintain type I error and to gain power over alternative methods. We used it in a prostate cancer (PRCA) application with data from Stage II of the Cancer Genetic Markers of Susceptibility GWAS that includes 27,383 SNPs in 3941 cases and 3964 controls. We performed two genome scans, conditioning on established susceptibility SNPs for PRCA. Both analyses identified a high-ranking SNP notable for its biology and worthy of follow-up study. The first scan involved a sub-region of 8q24 that contains independent susceptibility loci specific to PRCA, as well as androgen-receptor (AR) binding sites and androgen-responsive enhancer elements. The analysis highlights rs748120 intronic to NR2C2 which produces an AR co-regulator such that, when NR2C2 binds AR, expression of their target genes decreases. The observed interaction suggests NR2C2 may contribute to the androgen sensitivity of this 8q24 gene desert in the etiology of PRCA. The second scan involved the JAZF1 and HNF1B genes regions that are associated with both PRCA and type 2 diabetes, which itself is associated with PRCA risk. The analysis highlights rs4810671 intronic to SULF2 which has demonstrated oncogenic properties in several cancers. It is regulated by insulin, the hormone thought to mediate the association of type 2 diabetes with PRCA. The observed interaction suggests abnormal insulin function may affect PRCA risk through dysregulation of SULF2. In summary, we propose an innovative multi-locus test for disease association that gains power over alternative methods and we describe promising results from a PRCA GWAS that suggest functionality for susceptibility regions. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 2747. doi:10.1158/1538-7445.AM2011-2747