78th Annual Meeting Abstracts

BACKGROUND: Steroid 5α reductase type 2 (SRD5a2) is a critical enzyme in the metabolism of androgen. Two SRD5a2 gene polymorphisms, V89L and A49T, have been postulated to be associated with prostate cancer risk. However, genetic association studies have conflicting results. Thus, we conducted a meta-analysis to reexamine these associations and intend to incorporate findings from genome wide association (GWA) studies of prostate cancer.
METHODS: We conducted a meta-analysis and chose random effects model to take into account between-study heterogeneity. Meta-analyses by genotype included models corresponding to homozygote, recessive and dominant models. Meta-analyses by alleles included overall analysis and subgroup analysis. We assessed selection bias, information bias, and confounders for each study. We searched and reviewed GWA studies on prostate cancer risk and analyzed data from Cancer Genetic Markers of Susceptibility (CGEMS) database.
RESULTS: Our meta-analysis included 25 case-control studies: 10,088 cases and 10,120 controls for V89L and 5,386 cases and 5,912 controls for A49T. The V89L polymorphism was not associated with prostate cancer risk (L versus V allele: odds ratio (OR) = 0.99; 95% confidence interval (CI): 0.94-1.05). T allele of the A49T had small effect (OR=1.23; 95% CI: 0.94-1.61), but non-significant. Spectrum of disease bias, convenience sampling of controls, genotype misclassification, and confounding effects may influence these conclusions.
CONCLUSIONS: Neither V89L nor A49T were examined by GWA studies because of their absence in the selected microarray chips. Future candidate gene studies of prostate cancer should take a systematic approach to reduce bias. Analysis of well-designed population-based studies with pathway-based arrays containing common genetic variants could be useful for identifying genetic factors in prostate cancer.