Quadrant dosimetry as a predictor of biochemical relapse in 125I prostate brachytherapy

Abstract

Listen

Translate article

Purpose To evaluate region-specific dose metrics as predictors of biochemical relapse in prostate brachytherapy patients. Methods and Materials In a cohort of 1006 low-risk and “low-tier” intermediate-risk prostate brachytherapy patients treated to a planned dose of 144 Gy mPD (minimal peripheral dose), 30 of 32 with biochemical relapse (nadir + 2 ng/mL definition) had postimplant CT scans available for retrospective analysis. These were matched to nonrelapsing controls from the same era. Three copies of each CT were created and, after randomization and deletion of identifiers and original contours, were re-contoured by three radiation oncologists. Prostate contours were then divided into quadrants: Anterior-Superior (ASQ), Posterior-Superior (PSQ), Anterior-Inferior (AIQ), and Posterior-Inferior (PIQ), and dosimetric parameters calculated. Results were analyzed using mixed-effects linear regression and multivariate logistic regression. Results Whole prostate volume of the prostate receiving at least 100% of the prescribed dose ( V 100) and minimum dose, as a percentage of the prescribed dose, received by 90% of the prostate volume ( D 90) were similar for relapses and controls ( p = 0.40 and 0.48, respectively). Among the quadrants, the largest differences between relapses and controls were seen for the AIQ. Mean AIQ V 100s were 91.2% (relapses) and 95.5% (controls) ( p = 0.096), and D 90s were 112.8% (relapses) and 119.3% (controls) ( p = 0.145). Overall, the lowest doses were in the ASQ, but were not very different for relapses and controls ( V 100 = 76.5% and 78.5%, respectively) ( p = 0.54). On multivariate analysis along with various clinical parameters, AIQ metrics approached significance at the p ≤ 0.05 level in models that also included initial prostate-specific antigen, androgen suppression, and risk group. Conclusions Although whole prostate dose metrics did not predict for biochemical relapse in our data set, dose to the AIQ was predictive in multivariate analysis.