Objectives: To assess the factors effecting PSA bounce and to identify any possible relationship with biochemical control after 3-D conformal radiotherapy (3D-CRT) and total androgen deprivation (TAD) for prostate cancer by evaluating four previously described PSA bounce definitions.
Methods: Between January 1998 and January 2001, 83 consecutive patients with clinically localized prostate cancer were treated by 3D-CRT with neoadjuvant 3 months and/or 6 months adjuvant TAD. All patients had a pretreatment PSA level, at least eight post-external beam radiotherapy (EBRT) PSA and testosterone levels and minimum two years of follow-up. Total radiotherapy dose was 73.6 Gy at ICRU reference point. Four previous definitions of PSA bounce were used: Critz definition (>or=0.1 ng/mL), Cavanagh definition (>or=0.2 ng/mL), Hanlon definition (>or=0.4 ng/mL) and Rosser definition (>or=0.5 ng/mL) according to original methodology performed to report PSA bounce. Biochemical failure was defined in accordance with the ASTRO consensus guidelines.
Results: The median follow-up time was 40 months. PSA bounce was recorded as follows: Critz definition, 33 patients (40%); Cavanagh definition, 21 patients (25%); Hanlon definition, 11 patients (13%); and Rosser definition, 7 patients (8%). In multivariate analysis, pre-EBRT PSA level and the duration of TAD for Critz definition; age, pre-EBRT PSA and the duration of TAD for Cavanagh definition; age and duration of TAD for Hanlon definition; age and pre-biopsy PSA for Rosser definition were significant independent prognostic factors determining PSA bounce. A significant increase of mean testosterone level in bouncers was detected at the 6th-9th and 18th-21st months. PSA bounce did not predict for PSA failure in multivariate analysis.
Conclusions: We observed no correlation between biochemical failure and PSA bounce. The longer duration of TAD and older age were found to be inversely proportional with PSA bouncing in this cohort. Notably, recovery of testosterone might cause PSA bouncing.