13-year outcomes following treatment for clinically localized prostate cancer in a population based cohort

Purpose: Because data from randomized trials initiated after the introduction of prostate specific antigen testing are unavailable, we performed a retrospective, population based study to estimate prostate cancer specific survival and overall survival after surgery, radiation or observation to manage clinically localized prostate cancer. Materials and Methods: From the Connecticut Tumor Registry we identified Connecticut residents 75 years or younger diagnosed with clinically localized prostate cancer between January 1, 1990 and December 31, 1992. We obtained information from physician offices concerning treatments received by 1,618 patients who underwent surgery (802), external beam radiation therapy (702) or no initial therapy (114) and subsequent medical outcomes. Treatment comparisons were adjusted for pretreatment Gleason score, prostate specific antigen and clinical stage along with age at diagnosis and comorbidities using 3 methods, including categorization by risk, a proportional hazards model and a propensity score. Results: At an average followup of 13.3 years 13% of patients had died of prostate cancer, 5% had died of other cancers and 24% had died other noncancer causes. Patients undergoing surgery were younger, and had more favorable histology and lower pretreatment prostate specific antigen compared to patients undergoing radiation. Patients who elected observation had significantly worse cause specific survival than those who elected surgery. They also fared worse than men who received radiation therapy but the difference was not statistically significant, possibly because of the small number of prostate cancer deaths to date. Conclusions: Our findings suggest that patients undergoing surgery for clinically localized prostate cancer may have a cancer specific survival advantage compared to those electing radiation or observation. However, only a randomized trial can control for the many known and unknown confounding factors that can affect long-term outcomes.