Study Suggests That Primary Androgen-Deprivation Therapy Does Not Improve Long-Term Survival in Older Patients With Localized Prostate Cancer


Key Points

  • Cox multivariate analysis showed significantly worse prostate cancer–specific and overall survival in patients receiving primary androgen-deprivation therapy.
  • Instrumental variable analysis comparing primary androgen-deprivation therapy high-use vs low-use areas showed no significant differences in prostate cancer-specific or overall survival.

Primary androgen-deprivation therapy has been widely used in localized prostate cancer, despite absence of definitive evidence of benefit in early-stage disease. In a large population-based cohort study reported in JAMA Internal Medicine, Lu-Yao et al found that primary androgen-deprivation therapy was not associated with improved 15-year prostate cancer–specific survival or overall survival following diagnosis of localized prostate cancer in Medicare patients aged ≥ 66 years.

Study Details

The study included 66,717 men aged ≥ 66 years with localized prostate cancer (T1/T2) diagnosed between 1992 and 2009 who received no definitive local therapy within 180 days of prostate cancer diagnosis. Data on these patients were obtained from the Surveillance, Epidemiology, and End Results (SEER) program and linked Medicare files. The primary outcome measures were cancer-specific survival and overall survival.

A total of 25,125 patients (37.7%) received primary androgen-deprivation therapy and 41,592 received conservative management. Patients receiving primary androgen-deprivation therapy were older (median, 79 vs 77 years), sicker (15.2% vs 12.5% with a Charlson score ≥ 2), and more likely to have high-risk disease (47.7% vs 23.0%) and had a higher mean prostate-specific antigen level (19.5 vs 11.1 ng/mL).

Instrumental Variable Analysis

Outcomes were assessed using both a Cox multivariate analysis and a stratified instrumental variable analysis. The Cox multivariate analysis included age, race, comorbidity status, cancer stage, zip code income quartiles, zip code education quartiles, urban area, marital status, year of diagnosis, SEER region, state buy-in status, and cancer grade.

The main analysis of the study used a stratified instrumental variable approach, which is intended to account for both measured and unmeasured confounders. Use of primary androgen-deprivation therapy varied widely across the 122 health service areas included in the study.

In the main instrumental variable analysis, treatment and region variables were replaced by a single covariate indicating high or low use of primary androgen-deprivation therapy, with the comparison being between the top and bottom use tertiles. Covariates in the instrumental variable analysis consisted of age, race, comorbidity status, cancer stage, zip code income quartiles, zip code education quartiles, urban area, marital status, year of diagnosis, state buy-in status, and cancer grade.

Outcome on Cox Multivariate Analysis

Median follow-up was 110 months. On the Cox multivariate analysis, primary androgen-deprivation therapy use was associated with significantly increased risk of prostate cancer–specific mortality (adjusted hazard ratio [HR] = 1.53, 95% confidence interval [CI] = 1.45–1.63) and overall mortality (adjusted HR = 1.19, 95% CI = 1.17–1.22), including significantly increased risks of both cancer-specific and overall mortality among both patients with moderately differentiated disease (adjusted HRs = 1.65 and 1.20) and poorly differentiated disease (adjusted HRs = 1.37 and 1.16).

No Differences on Instrumental Variable Analysis

However, no significant differences in survival were observed on instrumental variable analysis. On the instrumental variable analysis, for high- vs low-use areas, 15-year prostate cancer–specific survival was 85.4% vs 85.4% (adjusted HR = 1.01, 95% CI = 0.90–1.14), with survival rates of 90.6% vs 90.6% (adjusted HR = 1.00, 95% CI = 0.85–1.18) among patients with moderately differentiated disease and 78.6% vs 78.5% (adjusted HR = 0.99, 95% CI = 0.84–1.17) among those with poorly differentiated disease. Death due to other cause was treated as a competing risk in calculating prostate cancer–specific survival.

Fifteen-year overall survival was 15.9% vs 16.8% (HR = 1.04, 95% CI = 0.99–1.09), with rates of 20.0% vs 20.8% (HR = 1.03, 95% CI = 0.96–1.10) in patients with moderately differentiated disease and 8.6% vs 9.2% (HR = 1.03, 95% CI = 0.96–1.10) among those with poorly differentiated disease. Most patients were followed for < 10 years; thus, the observed 15-year overall survival rates are lower than the proportions of patients remaining alive at the end of the study (40%). 

Regarding the use of the instrumental variable analysis, the investigators explained:

Results obtained from a Cox model that adjusted only for measured confounding factors differed significantly from those derived from the [instrumental variable] approach. Results from the Cox model showed less favorable outcomes among men receiving primary [androgen-deprivation therapy], but this most likely reflects a selection bias toward men with higher risk disease…. Differences in outcomes based on the Cox model vanished when an [instrumental variable]-analysis approach was used. These observations suggest that there is considerable unaccounted residual bias associated with the Cox model analytic methods and demonstrate an important advantage to the [instrumental variable] approach.

The investigators concluded, “Primary [androgen-deprivation therapy] is not associated with improved long-term overall or disease-specific survival for men with localized prostate cancer. Primary [androgen-deprivation therapy] should be used only to palliate symptoms of disease or prevent imminent symptoms associated with disease progression.” They noted, “Since the study was limited to men 66 years or older, the results could differ for younger men.”

Grace L. Lu-Yao, MPH, PhD, of Rutgers Cancer Institute of New Jersey and Robert Wood Johnson Medical School, is the corresponding author for the JAMA Internal Medicine article.

The study was supported by the National Cancer Institute and Cancer Institute of New Jersey. For full disclosures of the study authors, visit

The content in this post has not been reviewed by the American Society of Clinical Oncology, Inc. (ASCO®) and does not necessarily reflect the ideas and opinions of ASCO®.