• Steven J Feigenberg, Alexandra L Hanlon, Eric M Horwitz, Robert G Uzzo, Debra Eisenberg, and Alan Pollack.
• Department of Radiation Oncology, Fox Chase Cancer Center, Philadelphia, PA 19111-2497, USA. S_Feigenberg@fccc.edu
• Am. J. Clin. Oncol. 2006 Oct 1; 29 (5): 458-62.

ObjectiveThe purpose of this report is to determine whether any specific magnitude in the prostate specific antigen (PSA) bounce predicted for a clinically poorer outcome.Methods And MaterialsBetween May 1989 and August 1999, 568 prostate cancer patients were treated with 3-dimensional conformal radiotherapy (RT). All patients had at least 5 years of follow up, 6 post-RT PSA measurements and received no hormonal therapy as part of their initial management. The median follow up was 85 months. The median RT dose was 74 Gy. A bounce was defined by a minimum rise in PSA of 0.4 ng/mL over a 6-month period, followed by a drop of PSA of any magnitude. The analysis of the optimal PSA bounce cut-point was based upon a recursive partitioning approach (RPA) for censored data using the log-rank test for nodal separation of freedom from biochemical failure (FFBF) as defined by the American Society for Therapeutic Radiology and Oncology (ASTRO) definition. Cox multivariate regression analysis (MVA) was used to confirm independent predictors of outcome among clinical and treatment related factors: PSA bounce as defined by the RPA, pretreatment PSA (continuous), Gleason score (2-6 versus 7-10), T stage (T1c/T2ab versus T2c/T3), and total radiation dose (continuous).ResultsThere were 154 patients (27%) experienced a bounce with a median magnitude of 0.6. The RPA resulted in an optimal PSA bounce cut-point of 1.4 ng/mL such that 5-year Kaplan-Meier estimates of FFBF were 71%, 59%, and 38% for nonbouncers, a bounce < or =1.4 ng/mL and >1.4 ng/mL, respectively. Twenty-one (14%) of the 154 patients who experienced a bounce had a PSA bounce magnitude >1.4 ng/mL. Stepwise MVA demonstrated that the PSA bounce grouped as above was an independent predictor of FFBF (P = 0.0013), freedom from distant metastases (P = 0.0028) and cause specific survival (P = 0.0266). Lower RT dose (P < 0.0001) was the only independent predictor of a PSA bounce >1.4 ng/mL.ConclusionsUsing recursive partitioning techniques, a clinically significant PSA bounce occurred when the magnitude of the bounce was >1.4 ng/mL. This is important information to aid clinicians in determining management after RT.

15 keywords...

hide…