PCN8 Primary Androgen Deprivation Therapy Versus Radical Prostatectomy in Seer-Medicare Dataset: A Comparative Effectiveness Analysis of Retrospective Cohorts

Abstract

To examine the comparative effectiveness of primary androgen deprivation therapy (PADT) and radical prostatectomy (RP). Male patients with localized prostate cancer (T1-T2, N0, M0) were identified in the SEER-Medicare database from January 1998 to December 2007. Patients were 66-74 years old, without other documented cancers, and had PADT initiation or RP within 6 months after the first recorded diagnosis of prostate cancer in the dataset. PADT-treated patients were 1:1 matched to the RP-treated patients via propensity score (PS) matching. The overall survival from diagnosis to death was analyzed using Cox proportional hazard models; prostate cancer specific survival examined using Fine and Gray competing risk modeling. The independent regression variables included age at diagnosis, race, marital status, census regions, urban residence, clinical cancer stage, Gleason score, prostate specific antigen level, Charlson comorbidity index, calendar year at diagnosis, and hospitalization (yes or no), surgery (yes or no) and outpatient visit (yes or no) during one year baseline period. The PS-matched sample size was 3432 with mean age of 66.0 years. The baseline characteristics were comparable between the two cohorts (all p-values>0.05). During median follow-up of 2.95 years, the cumulative incidence of death was 304 (17.72%) among 1716 PADT patients and 94 (5.48%) among 1716 RP patients; 66 (3.85%) and 2 (0.12%) for prostate cancer specific deaths, respectively. The PADT group had nearly 4 times higher overall mortality risk compared to those using RP (odds ratio (OR)=3.534, 95% confidence interval (CI)=2.801-4.464, p<0.001). Furthermore, patients who received PADT had significantly higher prostate cancer specific mortality compared to those using RP (OR=30.875, 95% CI=7.535-126.506, p<0.001). Overall mortality and prostate cancer specific mortality following PADT were significantly higher compared to those following RP among localized prostate cancer patients. These data do not support the use of PADT in men with clinically localized prostate cancer.