Dosimetric impact of androgen deprivation therapy (ADT)-induced prostate deformation in low-dose-rate prostate brachytherapy

Abstract

PURPOSE: Androgen deprivation therapy (ADT) for prostate cancer typically reduces prostate volume. This study investigated how the timing of ADT relative to low-dose-rate (LDR) brachytherapy influences the resulting dosimetric outcomes. METHODS: Using deformable image registration and principal component analysis on pre-and post-ADT magnetic resonance images of 34 patients, we developed a statistical model of ADTinduced prostate deformation. We applied this model to 30 low-dose-rate (LDR) brachytherapy plans (prescription dose: 145 Gy) to simulate seed displacement and dose distribution changes from prostate shrinkage. Prostate deformation over time post-ADT was categorized into early, linear, and late response phases. Dosimetric outcomes were analyzed across scenarios with varying prostate volume reduction magnitudes and different intervals between ADT initiation and brachytherapy. RESULTS: When ADT was initiated concurrently with brachytherapy, the model predicted substantial dose escalation if prostate shrinkage occurred early (prostate D90 '206 Gy; urethral V200 '47.7%) compared to late shrinkage response (D90 '183Gy; V200 '7.1%). In contrast, separating the treatments in time greatly mitigated this effect. For example, assuming early response, performing brachytherapy 3 months before ADT yielded a prostate D90 of '183.0 Gy (urethral V200 '7.7%), while delaying brachytherapy to 3 months after ADT yielded a D90 of '177.8 Gy (V200 '1.2%). CONCLUSIONS: Dosimetric analysis showed that greater prostate volume reduction, early ADT response, and concurrent ADT all corresponded to substantially increased doses to the prostate and urethra. Clinically, these findings underscore the importance of carefully planning the timing of ADT relative to brachytherapy to optimize target coverage and minimize unintended dose escalation to normal tissues. (c) 2025 American Brachytherapy Society. Published by Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.