2356 - Improving Plan Quality by Reducing Setup Uncertainty Using an Online Adaptive Proton Workflow

W. Liu¹, H. Zhang², J. P. Schiff³, R. Beckert⁴, B. Kalaghchi⁵, E. Laugeman⁶, H. Kim⁷, S. M. Perkins⁵, T. Zhao⁵, and A. Mo⁶; ¹University of Iowa, Iowa City, IA, ²Washington University in St. Louis, St. Louis, MO, United States, ³Tulane University School of Medicine, New Orleans, LA, ⁴Washington University School of Medicine, Saint Louis, MO, ⁵Washington University in St. Louis, St. Louis, MO, ⁶Department of Radiation Oncology, Washington University School of Medicine in St. Louis, St. Louis, MO, ⁷Department of Radiation Oncology, Washington University School of Medicine, Saint Louis, MO

Purpose/Objective(s): Proton therapy offers significant normal tissue sparing but is subject to anatomical and positional variations which can cause major dose perturbations. Typical proton therapy planning incorporates set-up and range uncertainties to improve plan robustness into a beam specific planning treatment volume (bsPTV). We hypothesized bsPTV could be significantly reduced by near elimination of set-up uncertainty if treatment delivery could be coupled with an online adaptive workflow. Materials/

Methods: A novel online adaptive therapy workflow was developed for pencil-beam scanning proton therapy using CT-on Rails (CToR) image guidance and semi-automated treatment planning with custom Monte-Carlo-based secondary quality assurance. Ten patients previously treated on a five-fraction pelvic reirradiation proton SBRT trial were used to evaluate the feasibility of significant set-up uncertainty reduction. Proton plans using a reduced set-up margin from 5mm to 1mm and 3% range uncertainty were created and were evaluated for treatment delivery based on acquired daily CToR imaging. Alignment was evaluated by a GI radiation oncologist and medical physicist and were deemed suitable if there was adequate target alignment within the 95% of prescription after six degree of freedom couch correction and minimal change in tissue density along the beam path for each gantry angle. If target alignment was insufficient, the delivered fraction would have required online proton adaptation.
Results: All reduced margin plans achieved comparable coverage to the original plan. The average beam specific planning treatment volume (bsPTV) on the original delivered proton plans was 273.3cc (sem ±95.0) and was decreased by 46% (p=0.005) in the minimal margin plans. Of the 50 delivered fractions, 7 (14%) fractions would have had insufficient alignment and required online adaptation. Reduced set-up margin proton plans also resulted in a significant integral dose reduction (15.5%, p=0.028) to large bowel.
Conclusion: The possibility of online adaptation for proton treatment can dramatically reduce set-up uncertainty requirements for plan robustness, resulting in significant reductions in PTV and improvements in OAR sparing for all proton therapy treatments. In this analysis, beam specific treatment volumes could be reduced by almost 50% with significant improvements in OAR sparing and would only require online adaptation for approximately 14% of cases. Further improvements may be possible depending on the treatment site and will be investigated with an ongoing prospective trial evaluating the feasibility of online proton adaptation.