A case-control study of the androgen receptor gene CAG repeat polymorphismin Australian prostate carcinoma subjects

BACKGROUND. The development of prostate carcinoma is androgen-dependent. Th e coding sequence of the androgen receptor (AR) gene contains a CAG repeat polymorphism that has been shown to influence AR activity in vitro. Studies of this polymorphism as a prostate carcinoma risk factor have been conflic ting. METHODS. A matched case-control design was used in a clinic-based multicent er study of Australian prostate carcinoma subjects. Cancer subjects were ma tched by age and locality with controls, all of whom had a serum prostate s pecific antigen (PSA) level of less than 4 mg/L. Conditional logistic regre ssion was used to determine the relative risk of prostate carcinoma depende nt on AR gene CAG number. The association of disease characteristics at dia gnosis with the polymorphism also was assessed. RESULTS. Five hundred forty-five cases of prostate carcinoma and 456 matche d case-control pairs were recruited. Association studies of disease charact eristics at diagnosis showed age at diagnosis to be associated with AR CAG number by univariate (P = 0.004) and multivariate (adjusting for PSA, stage , and grade) linear regression (P = 0.018). No association was observed bet ween the polymorphism and disease stage (TNM-based categories; P = 0.277), histologic grade (P = 0.41), or PSA level at diagnosis (P = 0.48). In the p airwise case-control analysis, the odds ratio of prostate carcinoma for a c hange of 5 CAG repeats gave an odds ratio of 0.9821 (95% confidence interva l, 0.84-1.15). CONCLUSIONS. In this Australian study population, the AR CAG repeat polymor phism was not a risk factor for prostate carcinoma, but a shorter repeat se quence was associated with earlier age at diagnosis. (C) 2001 American Canc er Society.