Treatment margins predict biochemical outcomes after prostate brachytherapy

Due to the theoretical role of treatment margins (TMs) in cancer, we have correlated biochemical outcomes with post-implant TMs in patients treated with brachytherapy for early stage prostate cancer.

From November 1998 through September 2003, 492 of a planned total of 600 patients with 1997 AJC clinical stage T1c-T2a prostatic carcinoma (Gleason score 5 or 6, PSA 4 to 10 ng/mL) have been randomized to implantation with (125)I (144 Gy, TG-43) versus (103)Pd (125 Gy, NIST-99). This preliminary analysis included only the first 122 analyzable patients, while accrual to the trial finishes. Isotope implantation was performed by standard techniques, using a modified peripheral loading pattern. Axial CT images at 3 mm intervals were acquired within four hours postoperatively for post-implant dosimetry. The contoured images and sources were entered into Varian Variseed system 7.1 (Charlottesville, VA). After completion of standard dosimetric calculations, the 100% prescription dose TMs were measured and tabulated around the prostate periphery at the 0.0, 1.0, 2.0 and 3.0 cm planes, going distal from the bladder-prostate interface. Measurements were limited to the transverse planes. Freedom from biochemical failure was defined as a serum PSA < or = 0.5 ng/mL at last follow-up. Patients were censored at last follow-up if their serum PSA was still decreasing. Patients whose serum PSA nadired at a value >0.5 ng/mL were scored as failures at the time at which their PSA nadired. The follow-up period for non-failing patients ranged from 2.1-5.0 years (median: 3.3 years).

The average 100% prescription dose treatment margin (for individual patients) ranged from -5.0 to 8.7 mm, with an overall average of 2.6 mm (+/-3.1). In univariate analysis, the D(90) was the best predictor of biochemical control for (125)I, while the average TM was the best predictor for (103)Pd. Similarly, in multivariate analysis using the D(90), V(100), and average TM as the independent variables and biochemical control as the dependent variable, the D(90) was most closely related to biochemical control for (125)I patients, while average TM was most closely related for (103)Pd patients. In separate analysis of TM by site, the anterior TMs were the best predictors of biochemical outcomes.

V(100), D(90), and TMs all appear to have a bearing on biochemical freedom from relapse after prostate brachytherapy. Efforts to better identify and test geographic dosimetric parameters are theoretically appealing, and supported by the clinical data summarized here.