A retrospective analysis of the relationship between changes in serum PSA, palliative response and survival following systemic treatment in a Canadian randomized trial for symptomatic hormone-refractory prostate cancer

Background: To investigate the relationship between changes in serum PSA, palliative response and survival following systemic treatment for symptomatic hormone-refractory prostate cancer (HRPC). Patients and methods: A retrospective review of 161 patients, treated with mitoxantrone and prednisone (M + P) (n = 80), or prednisone alone (P) (n = 81) from a Canadian randomized phase III clinical trial. PSA response was defined by greater than or equal to 50% decline compared to baseline. Palliative response was defined by the primary and secondary endpoints of the trial. All responses were required to be maintained on two visits at least three weeks apart. The Cox proportional hazards model and a landmark analysis (at nine weeks) were used to evaluate survival differences between PSA responders and non-responders. Results: Using an intent-to-treat analysis in which patients with missing PSA data are considered non-responders, 34% of M + P and 11% of P patients achieved a PSA response (P = 0.0001). Nineteen of thirty-six (53%) patients with PSA response and twenty-six of ninety (29%) patients without PSA response achieved a palliative response (P= 0.001 Chi-square test, phi coefficient = 0.28). From the landmark analysis, PSA responders had longer survival than non-responders (P= 0.009). In multivariate analysis, better performance status, higher hemoglobin and PSA response (P < 0.001) predicted for survival, but palliative response did not (P = 0.11). Conclusions: There is significant but imperfect statistical association between PSA response and palliative response. PSA response was associated with longer survival. Patients treated with M + P were more likely to achieve a PSA response and a palliative response than those treated with P.