The 100-day PSA: usefulness as surrogate end point for biochemical disease-free survival after definitive radiotherapy of prostate cancer

Overall and biochemical disease-free (bNED) survival data after definitive radiotherapy (RT) for prostate cancer (CaP) requires decades of patient follow-up. Surrogates involving dynamics of prostate-specific antigen (PSA) decline, PSA nadir and time thereto have been unrewarding. This study investigated the metric of the PSA value 100 days after RT (PSA(100)), analyzed with respect to 8-y bNED survival. A total of 214 patients with T1-3 CaP were treated with definitive RT (defined as dose >66 Gy) in our institution between 1/1/1988 and 12/31/2000. All were subject to continuous follow-up with routine PSA levels. Biochemical failure ( 77 patients) was defined by the ASTRO criteria ( n = 67) or by the date of first hormonal therapy for a rising PSA, which did not meet the ASTRO criteria ( n = 10). No patients were included if they received postoperative radiation, or if hormones were administered prior to bNED recurrence, if any. Patients were stratified by PSA(100) values less than or equal to or >4.0 ng/ml, and less than or equal to or <2.5 ng/ml. Median follow-up was 64.3 months: follow-up data were calculated as of time to last PSA, with data collection as of 12/31/02. Patients with PSA(100) <= 4.0 ng/ml had 62% 8-y bNED survival, and those with PSA(100) >4.0 ng/ml had 20% 8-y bNED survival ( P<0.001). Use of a PSA(100) cutoff of 2.5 ng/ml yielded no significant difference in 8-y bNED survival ( P = 0.229). Cox proportional analysis revealed that initial PSA ( P = 0.006), stage ( P = 0.001) and PSA(100) <= 4.0 ng/ml ( P = 0.002) were significantly related to bNED survival, but that age ( P = 0.887), race ( P = 0.500), RT dose ( P = 0.669), Gleason sum ( P = 0.091), and PSA(100) <= 2.5 ng/ml ( P = 0.128) were not. In conclusion, PSA(100) using a cutoff of 4 ng/ml is a valuable and reliable surrogate for bNED survival after definitive RT, requiring less follow-up than other metrics. Patients with less values will have only about a 1 in 3 chance of bNED failure at 8 y. Patients with greater values will have a 4 in 5 chance of bNED failure at 8 y.