ESTRO 2020 Abstract book

S169 ESTRO 2020

each patient with 1 (anterior-posterior (AP)), 2 (AP- oblique), and 3 (AP-oblique and posterior-anterior) beams for a Varian Probeam delivery system, using a commercial treatment planning system (TPS, Eclipse v13.7). The prescription dose was 19.8 Gy (RBE) in 11 fractions. Beam arrangements were selected to reflect current clinical practice for paediatric HL treatments. Robust optimisation to the CTV, assuming an uncertainty of 3.5% in CT calibration and 5 mm in positioning was used. Single field uniform dose optimisation was used as a default and plans were optimised for a fixed RBE of 1.1. All plans were exported and recalculated with AUTOMC, an in-house Monte Carlo (MC) software based on GATE v8.1 and GEANT4 v10.3.3 (GATE RTion). Dose-averaged LET was scored and variable RBE dose was calculated using the McNamara model with an alpha/beta ratio of 2. Doses to the heart, cardiac sub-structures, lungs, breasts, oesophagus and spinal cord were recorded. A “hot spot” was defined as an overlap of regions receiving ≥ 80% prescription dose and regions with high LET (≥6 keV/μm). Results

between all strategies (TPS with fixed RBE, MC with fixed RBE and MC with variable RBE) were small (see figure 1). LET generally decreased as the number of beams increased. No “hot spot” was observed in plans using > 1 field. However, 2 plans using a 1-field arrangements revealed small “hot spots“ of 0.004 cc in muscle (Patient 1), or 0.11 cc located in the vertebra, carina, and descending aorta (Patient 3). The use of MC had a larger influence on the dose distribution than the use of a variable RBE (see figure 2). Conclusion In this study, increased LET and RBE at the end of proton range did not result in clinically meaningful hot-spots to OARs for paediatric patients with mediastinal HL, regardless of the beam arrangement. However, uncertainties remain in models of variable RBE, and further studies are needed to confirm these results. PD-0303 Plan quality assessment for rectal cancer patients using prediction of organ-at-risk dose metrics A. Vaniqui 1 , R. Canters 1 , F. Vaassen 1 , C. Hazelaar 1 , I. Lubken 1 , K. Kremer 1 , C. Wolfs 1 , W. Van Elmpt 1 1 MAASTRO, Department of Radiation Oncology MAASTRO clinic- GROW-School for Oncology and Developmental Biology- Maastricht University Medical Centre, Maastricht, The Netherlands Purpose or Objective To develop and validate an anatomy-based plan QA assessment method using organ-at-risk (OAR) dose metrics prediction for rectal cancer patients and assess treatment quality based on cohort analysis. Material and Methods Rectal cancer patients previously treated with equal institutional volumetric modulated arc therapy protocols (25 fractions of 2Gy) were selected. For a prediction model, a fully optimized training cohort of 22 patients was adopted. The average dose fall-off from the planning target volume (PTV) towards each OAR was calculated as a function of distance (Figure 1 a,b), following the method by Petit and van Elmpt [1] . The median training cohort dose– distance curve was used to predict dose-volume histogram (DVH) metrics for selected OARs. The prediction model was subsequently applied to a validation cohort planned with the clinically used manual treatment planning techniques (n=67). Furthermore, for the same cohort also a comparison with RapidPlan TM (Varian Medical Systems, Palo Alto, California) generated treatment plans was made. Differences between the predicted and achieved dose were quantified for the PTV coverage and the mean OAR dose of the bladder and bowel bag.

Results The prediction model fit showed good agreement between predicted and achieved mean doses to the bowel bag

All treatment plans had acceptable target coverage and robustness (evaluated with TPS dose). Dose differences