Large-scale Retrospective Monte Carlo Dosimetric Study for Permanent Implant Prostate Brachytherapy.

Abstract

To retrospectively compare water-based and full tissue model Monte Carlo dose calculations in a large cohort of patients undergoing ¹²⁵I permanent implant prostate brachytherapy. For 613 patients, EGSnrc BrachyDose dose calculations were performed in 2 virtual patient models: TG43sim (simulated American Association of Physicists in Medicine Task Group Report 43 conditions) and MCref (computed tomography-derived heterogeneous tissue model with interseed effects). A sensitivity analysis was performed in a patient subset (25 with and 25 without prostatic calcifications) to explore dose calculation dependence on organ-at-risk (OAR) and calcification tissue elemental compositions and modelling approach. In the target volume, the minimum radiation dose delivered to 90% of prostate (D₉₀) (volume of prostate receiving at least 100% of prescription dose [V₁₀₀]) was lower with MCref than with TG43sim by 5.9% ± 1.6% (2.6% ± 1.7%), on average. Patients with prostatic calcifications can have substantial underdosed volumes due to calcification shielding, lowering the D₉₀ by ≤25%. In the urethra, the average D₅ (D₃₀) was lower with MCref than with TG43sim by 4.4% ± 1.8% (4.7% ± 1.9%). In the rectum (bladder), the minimum dose to the hottest 0.1 cm³ (D_0.1cm³) of the contoured organ was lower (higher) with MCref than with TG43sim by 5.2% ± 1.8% (1.3% ± 1.8%). Doses to the target and OARs can increase or decrease by several percentages, depending on the assumed tissue elemental composition. In patients with calcifications, differences between approaches to model calcifications can change the target and OAR dose metrics by upward of 10%. TG43sim typically overestimates the target and OAR doses by several percentages, on average, compared with MCref. The considerable variation in the relative TG43sim and MCref doses between patients, and the larger dose differences for patients with calcification, suggests that clinical adoption of Monte Carlo dose calculations for permanent implant prostate brachytherapy should be pursued. The substantial sensitivity of the Monte Carlo dose calculations to the patient modelling approach supports the adoption of a consensus modelling scheme, such as MCref described in the present study, to ensure consistency of practice.