Why are pretreatment prostate-specific antigen levels and biochemical recurrence poor predictors of prostate cancer survival?

Abstract

BACKGROUND: The value of pretreatment (initial) prostate-specific antigen (iPSA) and biochemical recurrence (BR) as prognostic factors for survival remains unclear. The authors sought to determine why using randomized trial data with 7-year minimum follow-up. METHODS: In the Trans-Tasman Radiation Oncology Group 96.01 trial, 802 men with T2b, T2c, T3, or T4 N0 prostate cancer (PC) were randomized to radiotherapy alone or with 3 or 6 months neoadjuvant androgen deprivation between 1996 and 2000. Cox modeling was used to identify outcome predictors at follow-up landmark points. RESULTS: Higher iPSA was found to be a potent predictor of BR-free survival (P < .01) but was not prognostic for prostate cancer-specific survival (PCSS) from randomization. Patients experiencing BR had unfavorable initial prognostic factors compared with patients who did not. After BR, these factors were not prognostic for PC death in models adjusted for time to BR (TTBR). In these models, TTBR predicted PCSS more satisfactorily than the occurrence of BR itself. Survival probability 5 years after BR exceeded 90% for men with TTBR ≥4 years; however, it dropped to 44% ± 6% for men with TTBR <1 year. After BR, rapid PSA doubling time (DT), low iPSA, and short TTBR were identified as the most important predictors of inferior PCSS. CONCLUSIONS: When BR occurs, prognostic factors for survival change. Low iPSA, short TTBR, and rapid PSA DT take over at this point, providing reasons why iPSA and occurrence of BR alone predict PCSS unsatisfactorily. © 2009 American Cancer Society.