SDDRO-PBS: Simultaneous Dose and Dose Rate Optimization for FLASH Proton SBRT Based on Pencil Beam Scanning Dose Rate Modeling.

Purpose/Objective(s) Although the FLASH therapy (FLASH-RT) can reduce radiation-induced normal tissue toxicities while maintaining tumor response, its effectiveness for general cancer patients is to be further validated through clinical trials. The proton therapy is a natural choice for clinical FLASH-RT, as isochronous cyclotron-based proton systems are the only commercially available devices that can deliver ultra-high dose rates (e.g., 40 Gy/s or more) required for general-purpose clinical FLASH-RT. However, the state-of-the-art treatment planning method via intensity modulated proton therapy (IMPT) only optimizes the dose distribution and does not directly optimize the dose rate. This work aims to develop a new treatment planning method with FLASH-dose-rate optimization capability that is compatible with clinical FLASH-RT. Materials/Methods The dose rate modeling in this work is based on pencil beam scanning (PBS) dose rate, which accounts for the spatiotemporal delivery patterns of PBS and the beam-off time between multiple spots. The planning objectives for SDDRO-PBS include both dose-rate-volume and dose-volume constraints. Various machine constraints are enforced during plan optimization for plan deliverability and efficiently solved via proximal operators, including maximum beam intensity constraint, minimum constraint for spot weights per energy layer, overall machine minimum constraint for spot weight thresholds. Setup and range uncertainties are modelled by probabilistic formulation and solved by robust optimization. Mathematically, SDDRO-PBS solves an inverse optimization problem for deliverable PBS spot weights and per-energy-layer minimum-spot thresholds. Effective optimization algorithms are developed to deal with the nonconvexity and nonlinearity of SDDRO-PBS, using iterative convex relaxations powered by alternating direction method of multipliers. Results Three hypofractionated or SBRT lung cases were considered to validate SDDRO-PBS in comparison with IMPT. For fair comparison, all six plans under comparison were optimized with the same dose-volume objectives, optimization algorithms, and plan normalization. Owning to its dose rate optimization capability, SDDRO-PBS had additional dose-rate-volume objectives and aimed to achieve FLASH dose rate in the high-dose regions of interest (ROI) surrounding tumor targets. The results suggest that SDDRO-PBS had comparable dose distribution and substantially improved FLASH dose rate coverage from IMPT. In particular, SDDRO-PBS achieved 95% or higher FLASH-dose-rate coverage in high-dose ROI. Moreover, the increase of dose per fraction further improved the plan quality in terms of both dose and dose rate distributions. Conclusion A new and compatible treatment planning method SDDRO-PBS for clinical FLASH-RT is developed with FLASH-dose-rate optimization capability. Author Disclosure H. Gao: None. M.M. Folkerts: None. Y. Lin: None. E. Abel: None. J.D. Bradley: Honoraria; Genentech, Inc, Mevion Medical Systems. Consultant; AstraZeneca, Inc, Varian Medical Systems, Inc. Advisory Board; Genentech, Inc, Mevion Medical Systems; American Radium Society Executive Committee. Organize NRG Oncology research agenda on lung cancer; American College of Radiology.