Improved biochemical relapse-free survival with increased external radiation doses in patients with localized prostate cancer: The combined experience of nine institutions in patients treated in 1994 and 1995

Purpose: To study the radiation dose-response as determined by Kaplan-Meier prostate-specific antigen (PSA) disease-free survival (PSA-DFS) estimates in patients with stage T1-T2 prostate cancer treated within a 2-year period (1994-1995). Methods: Nine institutions combined data on 4839 patients with stage T1 and T2 adenocarcinoma of the prostate who received greater than or equal to60 Gy external beam radiation therapy (RT) as sole treatment. No patient received neoadjuvant androgen deprivation or planned adjuvant androgen deprivation. Of the 4839 patients, 1325 were treated in 1994 and 1995; 1061 were treated with <72 Gy and 264 with greater than or equal to72 Gy. The median RT doses for the <72 Gy and the greater than or equal to72 Gy groups were 68.4 Gy and 75.6 Gy, respectively. The median follow-up for the <72 Gy and the greater than or equal to72 Gy groups were 5.8 and 5.7 years, respectively. Risk groups, defined on the basis of T stage, pretherapy PSA level, and biopsy Gleason score (GS), were as follows: low risk-T1b, T1c, T2a, GS less than or equal to6 and PSA less than or equal to10 ng/mL; intermediate risk-T1b, T1c, T2a, GS less than or equal to6 and PSA >10 ng/mL but less than or equal to20 ng/mL or T2b, GS less than or equal to6 and PSA less than or equal to20 ng/mL or GS 7 and PSA less than or equal to20 ng/mL; high risk-GS 8-10 or PSA >20 ng/mL. The endpoint for outcome analysis was PSA-DFS at 5 years after therapy using the American Society for Therapeutic Radiology and Oncology failure definition. Results: Patients receiving greater than or equal to72 Gy had significantly more advanced cancers. The proportion of stage T2b/T2c cancers in the greater than or equal to72 Gy group was 42% compared with 32% in the <72 Gy group (P = 0.027). The mean pretherapy PSA was 11.4 ng/mL in the greater than or equal to72 Gy group compared with 10.7 ng/mL in the <72 Gy group (p = 0.001). The proportion of GS greater than or equal to8 cancers in the greater than or equal to72 Gy group was 9% compared with 7% in the <72 Gy group (p = 0.309). Overall, 15% of patients receiving <72 Gy had high-risk disease, compared with 22% of patients receiving greater than or equal to72 Gy (p = 0.034). The greater than or equal to72 Gy group had a greater number of follow-up PSA levels (mean 10.6/patient) compared with the <72 Gy group (mean 9.6/patient) (p = 0.007). For all 1325 patients, the 5- and 8-year PSA-DFS estimates were 64% and 62%, respectively. The 5-year PSA-DFS estimates for <72 Gy vs. greater than or equal to72 Gy were 63% vs. 69%, respectively (p = 0.046). Multivariate analysis for factors affecting PSA-DFS was performed for all cases using the following variables: pretherapy PSA (continuous), biopsy GS (continuous), stage (T1 vs. T2), radiation dose (continuous), and radiation technique (three-dimensional conformal vs. conventional). Pretreatment PSA (p < 0.001, chi-square 112.2), GS (p < 0.001, chi-square 12.8), radiation dose (p < 0.001, chi-square 13.5), and stage (p = 0.007, chi-square 7.2) were independent predictors of outcome. Radiotherapy technique was not (p = 0.50). Conclusion: Differences in PSA-DFS estimates observed in multiple retrospective series have been attributed to differences in follow-up duration between patients treated to conventional doses (longer follow-up intervals) and those treated to higher doses (shorter follow-up intervals). In this report, the median follow-up duration in the greater than or equal to72 Gy group was essentially identical to the <72 Gy group, because the study included a large number of patients treated consecutively during a narrow time range (1994-1995). With similar follow-up duration, higher than conventional radiotherapy doses were associated with improved PSA-DFS when controlled for the influence of pretreatment PSA levels, biopsy GS, and clinical T stage. (C) 2005 Elsevier Inc.