721 CAUSES OF MORTALITY FOLLOWING TREATMENT FOR PROSTATE CANCER: WHAT IS THE IMPACT OF ANDROGEN DEPRIVATION?

Abstract

INTRODUCTION AND OBJECTIVES: Androgen deprivation therapy (ADT) is commonly used for prostate cancer as primary, neoadjuvant/adjuvant, and secondary. Several recent reports have drawn attention to important skeletal, metabolic, cardiovascular, and cognitive side effects of ADT. The association between ADT use and cardiovascular mortality (CVM), in particular, remains controversial. We analyzed mortality outcomes in a large national registry to elucidate the impact of treatment selection on mortality. METHODS: 8,095 men in the CaPSURE registry diagnosed between 1998 and 2005 were analyzed. Treatment was categorized as local alone, primary ADT (PADT), local treatment with adjuvant / neoadjuvant therapy (local ADT), and watchful waiting/active surveillance (WW). A competing hazards survival analysis was performed, including events defined as prostate cancer-specific mortality (CSM), CVM, and other. The model was adjusted for age, cardiovascular disease, other comorbidities, prostate cancer risk characteristics, and duration of ADT. A propensity score analysis was undertaken to adjust further for imbalances in covariate among men receiving various treatments. RESULTS: 4889 men (63.7%) received local treatment, 1044 (13.6%) PADT, 1248 (16.3%) local ADT, and 497 (6.5%) WW. 1104 (14.6%) received secondary therapy, which was ADT in 837 cases and local treatment in 267 cases. Overall, 2553 (32.2%) of men in the dataset received ADT at some point during treatment. 1068 (13.2%) men died of any cause: 132 (1.6%) with CSM, 251 (3.1%) with CVM, 593 (7.4%) of other causes, and 86 (1.1%) of unknown causes. On riskand comorbidity-adjusted survival analysis, the hazard ratios (HRs) for CSM relative to local treatment was 4.4 (2.7–7.4) for PADT, 2.8 (1.7–4.7) for local ADT, and 2.5 (1.6–3.9) for WW. For CVM, the HRs relative to local treatment were 2.1 (1.4–3.0) for PADT, 1.8 (1.0–3.2) for local ADT, and 2.5 (1.6–3.9) for WW. For overall mortality, the HRs relative to local treatment were 1.6 (1.3–2.0) for PADT, 1.3 (1.0–1.5) for local ADT, and 1.7 (1.3–2.1) for WW. In the propensity-matched analysis, receipt of ADT was associated with overall mortality, but not CVM. CONCLUSIONS: Men treated with ADT have higher likelihood of CSM, CVM, and overall mortality than those treated with local therapy onlyebut so do those managed with WW. With respect to CVM, therefore, despite adjustment for disease risk and major comorbidities, unmeasured variablesewhich influence treatment decision-making but are not recorded databaseselikely confound analyses of the impact of ADT on cardiac endpoints.