Comparisons of PSA Failure Definitions Following Trimodality Therapy for Intermediate to High-Risk Prostate Cancer

Abstract

The ASTRO definition was primarily developed to assess biochemical failure following external beam irradiation (EBRT). The Phoenix (PHNX) definition was found to be a more sensitive and specific definition and takes into account patients treated with the addition of hormonal therapy or those treated with prostate brachytherapy. A recent American Urological Association consensus paper defined a biochemical failure following radical prostatectomy as a PSA > 0.2 ng/ml (AUA definition). Trimodality therapy combines hormonal therapy, external beam and brachytherapy. Data following this therapy has not been analyzed in any of the above consensus reports. It is associated with the highest biologically effective dose (BED) and so will be more ablative and perhaps better judged by the AUA definition. Since it uses both EBRT as well as brachytherapy, either of the radiation failure definitions might best be used. Biochemical failures rates were calculated using all three definitions and compared to determine which might be best applied to calculating PSA failure following this form of therapy. 523 patients underwent trimodality therapy for the treatment of non-metastastic prostate cancer. The presenting risk groups were low (large volume Gleason 6) in 2%, intermediate in 23% and high in 75%. Clinical stage was ≤ t1c in 26%, t2a in 16%, t2b in 29%, t2c in 22% and ≥ t3a in 7%. Gleason score was ≤ 6 in 21%, 7 in 50% and ≥ 8 in 29%. Presenting PSA was ≤ 10 in 52%, >10–20 in 30% and >20 in 18%. Trimodality therapy consisted of 3–9 months of hormonal therapy, a Pd-103 implant (prescription dose–100 Gy) and 45 Gy of EBRT. The BED for these treatments ranged from 148 to 280 (median–205) using an alpha/beta ratio of 2. Patients were followed from 2 to 11 years (median 4.4). PSA failure was calculated using the ASTRO, PHNX and AUA definitions. The actuarial freedom from PSA failure rates for the three definitions were similar. At 3 years the rates were: 92% (ASTRO), 94%(PHNX) and 88%(AUA). At 5 years the rates were: 89%(ASTRO), 88%(PHNX), 84%(AUA). At 8 years the rates were: 87%(ASTRO), 83%(PHNX), 83%(AUA). There were no differences between the ASTRO and PHNX actuarial curves (p = 0.9) and ASTRO and AUA curves (p = 0.058). There was a significant difference between the curves for the AUA and PHNX definitions (p = 0.04). There were more failures using the AUA (75) compared to the ASTRO (55) and the PHNX (55) definitions but this was not significantly different (p = 0.09). The last PSA levels for the 17 patients defined as an AUA failure but not an ASTRO or PHNX ranged from 0.23 to 1.79 ng/ml (median–0.34 ng/ml). Of these patients, none have developed a clinical (local or distant) recurrence nor have any been started on hormonal therapy. When applying PSA failure definitions to a high BED treatment such as trimodality therapy, the original ASTRO definition remains viable and yields similar results as the PHNX and AUA definition. Although the AUA calls more failures, these appear to be overestimates based on clinical correlations. This increase in failures is not statistically significant and does not result in a difference between the actuarial curves of the AUA and ASTRO definitions. This suggests that the AUA definition could be used in trials comparing prostatectomy to trimodality therapy.