Shared Inherited Genetics of Benign Prostatic Hyperplasia and Prostate Cancer
AbstractBackgroundThe association between benign prostatic hyperplasia (BPH) and prostate cancer (PCa) remains controversial, largely due to inherent detection bias in traditional observational studies. The objective of this study is to assess their association using inherited SNPs.MethodsSubjects were White men from the large population-based UK Biobank (UKB). Association between BPH and PCa was tested: 1) phenotypical correlation using chi-square test, 2) genetic correlation (rg) based on 1,126,841 polymorphic SNPs across the genome using linkage disequilibrium score regression (LDSR), and 3) cross-disease genetic associations based on known risk-associated SNPs (15 for BPH and 239 for PCa), individually and cumulatively as measured by genetic risk score (GRS).FindingsAmong 214,717 White men in the UKB, 24,623 (11.47%) and 14,311 (6.67%) had a diagnosis of BPH and PCa, respectively. Diagnoses of these two diseases were significantly correlated, χ2=1862.80, P<1E-299. A significant genetic correlation was found, rg (95% confidence interval (CI))=0.27 (0.15-0.39), P=9.17E-06. In addition, significant cross-disease genetic associations for established risk-associated SNPs were also found. Among the 250 established GWAS-significant SNPs of PCa or BPH, 51 were significantly associated with risk of the other disease at P<0.05, significantly more than expected by chance (N=12), P=3.04E-7 (χ2-test). Furthermore, significant cross-disease GRS associations were also found; GRSBPH was significantly associated with PCa risk (odds ratio (OR)=1.26 (1.18-1.36), P=1.62E-10), and GRSPCa was significantly associated with BPH risk (OR=1.03 (1.02-1.04), P=8.57E-06). Moreover, GRSBPH was significantly and inversely associated with lethal PCa risk in a PCa case-case analysis (OR=0.58 (0.41-0.81), P=1.57E-03). In contrast, GRSPCa was not significantly associated with lethal PCa (OR=0.99 (0.94-1.04), P=0.79).InterpretationBPH and PCa share common inherited genetics which suggests the phenotypical association of these two diseases in observational studies is not entirely caused by detection bias. This novel finding may have implications in disease etiology and risk stratification.FundingNone.
