To assess utilization trends and determine the effect of testosterone replacement therapy on outcomes in men who subsequently developed prostate cancer.

We used linked Surveillance, Epidemiology, and End Results–Medicare data to identify 149,354 men diagnosed with prostate cancer from 1992 to 2007. Of those, 2,237 men (1.5%) underwent testosterone replacement therapy before their prostate cancer diagnosis. Propensity scoring methods were used to assess cancer-specific outcomes of testosterone replacement vs no replacement therapy.

Testosterone replacement was associated with older age at cancer diagnosis, nonwhite race, and higher comorbidity (P <.001). No testosterone vs testosterone before the prostate cancer diagnosis was associated with higher grade (34% vs 30%, P <.0001) and more T4 (6.5% vs 4.3%, P <.0001) tumors. Mortality was decreased in men with ≥2 prostate-specific antigen (PSA) tests in the year before their cancer diagnosis. No significant difference was found between groups in overall survival, cancer-specific survival, or use of salvage androgen-deprivation therapy after initial treatment.

Through our observational study design, we show that testosterone use was low throughout the study period. Testosterone use was not associated with aggressive prostate cancer and did not affect overall or disease-specific mortality. Although our findings support growing evidence that testosterone replacement is safe with respect to prostate cancer, confirmatory prospective studies are needed.