Purpose: The purpose of the present report is to describe the relationship between two dosimetric quantifiers (V(100) and D(90)) and freedom from biochemical recurrence (FFBR) in a cohort of men treated with low-dose-rate prostate brachytherapy (LDRPB) alone.
Methods and materials: One hundred three men were treated with LDRPB alone between September 1997 and December 1999. All men had histologically confirmed clinically localized prostate cancer. Fifty-nine percent of the cohort had low-risk disease (defined as PSA<10, Gleason <7, and T stage <T2b), and 41% had intermediate-risk disease (defined as PSA>/=10, Gleason>/=7, or T stage T2b). The prescription dose was 144Gy according to the Task Group 43 formalism. LDRPB was performed jointly by a radiation oncologist and a urologist. Dosimetric quantifiers (D(90), V(100)) were calculated from a CT scan performed 1 month after LDRPB. Biochemical recurrence was defined according to the ASTRO Consensus Definition. FFBR was estimated using the product-limit method. Disease-specific and treatment variables were examined as putative covariates for FFBR using the proportional hazards regression method. Univariate and multivariate methods were used. All p values are two sided.
Results: The median followup for the entire cohort is 61 months. The median followup of patients at risk for biochemical failure is 66 months. The median D(90) is 129Gy (range 47-221Gy), and the median V(100) is 86% (range 51-99%). Thirteen men have developed evidence of biochemical relapse at a median of 25 months (range 6-42 months). The 5-year estimate of FFBR for the entire cohort is 87% (95% CI 80-94%). On univariate analysis, disease-specific variables found to be significantly associated with FFBR included pretreatment PSA and percent positive biopsies. When considered as a continuous variable, each of the dosimetric quantifiers was associated with FFBR (V(100): p=0.007; D(90): p=0.05). D'Amico risk group classification is highly predictive of FFBR after LDRPB (HR 5.68, p=0.003). Multivariate analysis indicated that each dosimetric quantifier was independently associated with FFBR, but due to the high degree of correlation (Pearson correlation coefficient 0.94, p<0.0001) between the dosimetric quantifiers both could not be included simultaneously in the model. In the two models explored, V(100) was at least as good as D(90) in predicting FFBR.
Conclusions: Dosimetric quantifiers (V(100) and D(90)) are independent predictors of FFBR after treatment with LDRPB alone. In our experience, V(100) seems to be at least equivalent (and perhaps superior) to D(90) for predicting FFBR.