3019 - Comparison of Delivery-Based Strategies for Reducing Voxel-Wise Dose Rate in Pencil Beam Scanning Proton Therapy

Purpose/Objective(s): Voxel-wise dose rate in pencil beam scanning (PBS) proton therapy is highly heterogeneous and has been implicated as a potential contributor to CNS toxicities not explained by dose–volume metrics alone. While methods to calculate dose rate exist, clinically practical planning strategies for dose rate reduction have not been systematically compared. This study aimed to evaluate and rank clinically available delivery-based approaches for reducing voxel-wise dose rate in PBS proton therapy.

Materials/Methods: Ten CNS proton therapy patients were retrospectively analyzed. For each patient, 5 treatment plans were generated: the clinical reference plan with a minimum monitor units (MU_min) of 3, and 4 modified plans consisting of (1) repainting of the 3 MU_min plan, (2) MU_min reduced to 2 MU, (3) MU_min reduced to 1 MU, and (4) MU_min of 1 MU with a 2-cm range shifter. Repainting effectively reduces the MU_min delivered per spot to approximately half by splitting delivery. All plans were optimized to maintain comparable dosimetric quality. Target coverage and dose to the optic chiasm and nerves differed by <1% across all plans, without significant differences in LETD.

Spot delivery times were estimated using a predictive model trained on 244 proton delivery log files, demonstrating strong agreement between predicted and measured durations (slope = 0.95, R² = 0.98). A RayStation script was used to export dose information for each spot and algorithmically matched with modeled spots to calculate dose rate and generate 3D DICOM maps of maximum voxel-wise dose rate. Dose-rate distributions were evaluated using voxel-wise near-maximum dose rate (MDR2), defined as the dose rate received by 2% of the volume.

Results: Dose-rate maps were successfully generated for all plans. Relative to 3 MU_min reference plans, reducing MU_min resulted in substantial MDR2 reductions, with median MDR2 reduced to 81.3%±9.7% for 2 MU_min and 55.2%±3.6% for 1 MU_min. Repainting of the 3 MU_min plans yielded a smaller reduction compared to 1 MU_min plan, with MDR2 reduced to 73.8%±5.4% of baseline. The largest reduction was achieved using a 1 MU_min combined with a range shifter, reducing MDR2 to 31.5%±6.2% of the original plans at the expense of a larger dose cloud and slightly increased dose to adjacent normal tissues. Across all patients, direct reduction of MU_min was consistently more effective than repainting for dose rate reduction.

Conclusion: All evaluated delivery strategies reduce voxel-wise dose rate primarily by lowering MU_min and, consequently, the dose rate. Direct reduction of MU_min was more effective and clinically efficient than repainting, which increases the number of delivered fields without significantly lowering dose rate. The addition of a range shifter enabled the greatest dose rate reduction but may compromise plan quality in selected cases. Thoughtful selection of clinically achievable delivery parameters can substantially reduce voxel-wise dose rate while maintaining acceptable dosimetric quality.