Purpose: We determined whether prostate specific antigen (PSA) velocity can serve as surrogate end point for prostate cancer specific mortality (PCSM) in patients with nonmetastatic, hormone refractory prostate cancer.
Materials and methods: The study cohort comprised 919 men treated from 1988 to 2002 at 1 of 44 institutions with surgery (560) or radiation therapy (359) for clinical stages T1c-4NxMo prostate cancer, followed by salvage hormonal therapy for PSA failure. All patients experienced PSA defined recurrence while on hormonal therapy. Prentice criteria require that the surrogate should be a prognostic factor and the treatment used did not alter time to PCSM following achievement of the surrogate end point. These criteria were tested using Cox regression. All statistical tests were 2-sided.
Results: PSA velocity greater than 1.5 ng/ml yearly was statistically significantly associated with time to PCSM and all cause mortality following PSA defined recurrence while undergoing hormonal therapy (Cox p <0.0001). While initial treatment was statistically associated with time to PCSM and all cause mortality (Cox p = 0.001 and 0.01), this association became insignificant when PSA velocity and potential confounding variables were included in the Cox model (p = 0.22 and 0.93, respectively). The adjusted HR for PCSM in patients who experienced a greater than 1.5 ng/ml increase in PSA within 1 year while on hormonal therapy was 239 (95% CI 10 to 5,549).
Conclusions: These data provide evidence to support PSA velocity greater than 1.5 ng/ml yearly as a surrogate end point for PCSM in patients with nonmetastatic, hormone refractory prostate cancer. Enrolling these men onto clinical trials evaluating the impact of chemotherapy on time to bone metastases and PCSM is warranted.