Modeling uncertainties in the delivery and evaluation of prostate brachytherapy implants

Abstract

Trans-rectal ultrasound guided prostate brachytherapy is used for treating early stage prostate cancer. Uncertainties in all stages of the implant process limit the ability to deliver planned dose distributions, interpret clinical outcomes, and design novel techniques. We hypothesize that assessing the impact of uncertainties on the dosimetric and radiobiological indices of implant quality will allow identification of the most influential sources of uncertainty and aid development of new strategies to minimize their effect. Three specific issues are addressed: (1) dose anisotropy of individual brachytherapy seeds; (2) imaging uncertainties in the visualization and localization of seeds and in delineating the prostate boundary; and (3) implant execution uncertainties, including needle placement, seed migration, and changes in prostate shape and volume between treatment planning and postimplant evaluation. Additionally, we investigate factors affecting estimates of α/β, an important radiobiological parameter. Incorporation of seed anisotropy, inability to visualize up to 20% of the implanted seeds, typical seed localization uncertainties, and random contouring uncertainties each led to minimal changes in D90 (the minimum dose to 90% of the target volume). However, random contouring uncertainties led to a variability of 9% in D90 (quantified as the average standard deviation), and systematic contouring uncertainties of 2 mm led to >10% change in D90. Actual achieved D90 values for 35 clinical cases were >25% lower than planned D90 values. We present an analysis of these differences and the details of a model to predict the expected dosimetric indices taking needle placement, seed migration, and differences in prostate shape into consideration. We also assess how variability in dose distributions between different patients, the relative biological effectiveness of the radiation from I-125 seeds, and uncertainties (due to seed localization, contouring, and edema) affect the determination of α/β. We conclude that accurate estimates of α/β are difficult to derive if these variables are not taken into consideration. The knowledge of which sources of uncertainty are most significant and how they impact the dosimetric and radiobiological outcome will allow future research to focus on techniques to reduce the impact of these uncertainties and improve implant quality, thereby improving patient outcome.