Purpose: In this era of dose escalation, the benefit of higher radiation doses for low-risk prostate cancer remains controversial. For intermediate-risk patients, the data suggest a benefit from higher doses. However, the quantitative characterization of the benefit for these patients is scarce. We investigated the radiation dose-response relation of tumor control probability in low-risk and intermediate-risk prostate cancer patients treated with radiotherapy alone. We also investigated the differences in the dose-response characteristics using the American Society for Therapeutic Radiology and Oncology (ASTRO) definition vs. an alternative biochemical failure definition.
Methods and materials: This study included 235 low-risk and 387 intermediate-risk prostate cancer patients treated with external beam radiotherapy without hormonal treatment between 1987 and 1998. The low-risk patients had 1992 American Joint Committee on Cancer Stage T2a or less disease as determined by digital rectal examination, prostate-specific antigen (PSA) levels of < or =10 ng/mL, and biopsy Gleason scores of < or =6. The intermediate-risk patients had one or more of the following: Stage T2b-c, PSA level of < or =20 ng/mL but >10 ng/mL, and/or Gleason score of 7, without any of the following high-risk features: Stage T3 or greater, PSA >20 ng/mL, or Gleason score > or =8. The logistic models were fitted to the data at varying points after treatment, and the dose-response parameters were estimated. We used two biochemical failure definitions. The ASTRO PSA failure was defined as three consecutive PSA rises, with the time to failure backdated to the mid-point between the nadir and the first rise. The second biochemical failure definition used was a PSA rise of > or =2 ng/mL above the current PSA nadir (CN + 2). The failure date was defined as the time at which the event occurred. Local, nodal, and distant relapses and the use of salvage hormonal therapy were also failures.
Results: On the basis of the ASTRO definition, at 5 years after radiotherapy, the dose required for 50% tumor control (TCD(50)) for low-risk patients was 57.3 Gy (95% confidence interval [CI], 47.6-67.0). The gamma50 was 1.4 (95% CI, -0.1 to 2.9) around 57 Gy. A statistically significant dose-response relation was found using the ASTRO definition. However, no dose-response relation was noted using the CN + 2 definition for these low-risk patients. For the intermediate-risk patients, using the ASTRO definition, the TCD(50) was 67.5 Gy (95% CI, 65.5-69.5) Gy and the gamma50 was 2.2 (95% CI, 1.1-3.2) around TCD(50). Using the CN + 2 definition, the TCD(50) was 57.8 Gy (95% CI, 49.8-65.9) and the gamma50 was 1.4 (95% CI, 0.2-2.5). Recursive partitioning analysis identified two subgroups within the low-risk group, as well as the intermediate-risk group: PSA level <7.5 vs. > or =7.5 ng/mL. Most of the benefit from the higher doses for the low- and intermediate-risk group was derived from the patients with the higher PSA values. For the low-risk group, the dose-response curves essentially plateaued at 78 Gy.
Conclusion: A dose-response relation was found using the ASTRO definition for low-risk prostate cancer. However, we found only marginal or no dose-response relation when the CN + 2 definition was used. Most of the benefit from the higher doses derived from low-risk patients with higher PSA levels. In all cases, little projected gain appears to exist at doses >78 Gy for these patients. A dose-response relation was noted for the intermediate-risk patients using either the CN + 2 or ASTRO definition. Most of the benefit from the higher doses also derived from the intermediate-risk patients with higher PSA levels. Some room for improvement appears to exist with additional dose increases in this group.