ESTRO 38 Abstract book

S550 ESTRO 38

Medical Center, Radiation Oncology, Leiden, The Netherlands

Purpose or Objective To determine setup and range robustness settings for IMPT planning required to achieve a D98%-CTV of 95% of the prescribed dose in fractionated treatments in 98% of skull- base meningioma patients, for given systematic and random setup errors and systematic relative stopping- power prediction (range) errors. Material and Methods Robust treatment plans to 50.4 Gy in 28 fractions were made for 9 patients using in-house developed software for multi-criteria optimization of treatment plans. Setup robustness settings from 0 to 5mm and relative range robustness settings from 0 to 5% were used in minimax robust optimization with 9 error scenarios: the nominal scenario, 6 shifted setup scenarios and 2 scenarios with range errors. Polynomial Chaos Expansion (PCE) was applied to model the impact of setup and range uncertainties on dose. PCE provides a computationally efficient metamodel of the voxel doses. Normal distributions were assumed, and setup and range uncertainties were modeled as rigid shifts and scaling of CT values respectively. PCE was validated for beam and couch angles used in clinical IMPT planning on an independent dataset for the same treatment site. Required range and setup robustness settings were determined in two steps. First, the range uncertainty leading to adequate target and population coverage was determined for various range robustness settings and no setup robustness. To this end, a PCE for range uncertainties was constructed to simulate and evaluate 100.000 complete fractionated treatments. Subsequently, random and systematic setup uncertainties for given setup and range robustness settings were determined in the same fashion. All plans were scaled such that, in the worst optimization scenario, the D98%-CTV was equal to 95% of the prescribed dose. Final results were based on the worst performing patient and validated on the others. Results Scaling all plans to the same target dose in the worst optimization scenario reduced the population variation of the D98%-CTV due to inter-patient variability from 2% to 0.4%. We found a linear relationship between range robustness (RR) and the actual range uncertainty (ρ), see Fig. 1. The relationship between random (σ) and systematic (Σ) setup errors that just can be dealt with for given setup robustness (SR, different colors) is shown in Fig. 2. Nearby points of the same color correspond to different range uncertainties (and the corresponding range robustness from Fig. 1). The dashed line and curves are fits to the data points.

Figure 2: Random and systematic setup and range errors leading to adequate target and population coverage. Conclusion With a population coverage of 98%, the required range robustness for range errors from 0 to 5% is a factor 2 higher than the actual range error. Interference of setup and range error is negligible, and the required setup robustness does not depend on range robustness.

Poster: Physics track: Quantitative functional and biological imaging

PO-0999 Functional Avoidance planning allows for lung dose reduction in radiotherapy of lung cancer K. Farr 1 , K. West 1 , R. Yeghiaian-Alvandi 1 , D. Farlow 2 , R. Stensmyr 3 , A. Chicco 2 , E. Hau 1,4,5 1 Crown Princess Mary Cancer Centre- Westmead Hospital, Department of Radiation Oncology, Sydney West Radiation Oncology Network- Sydney, Australia ; 2 Westmead Hospital, Department of Nuclear Medicine- PET and Ultrasound, Sydney, Australia ; 3 Crown Princess Mary Cancer Centre- Westmead Hospital, Department of Medical Physics, Sydney West Radiation Oncology Network- Sydney, Australia ; 4 University of Sydney, Faculty of Medicine and Health, Sydney, Australia ; 5 Blacktown Hospital- Sydney West Radiation Oncology Network, Department of Radiation Oncology, Sydney, Australia Purpose or Objective To evaluate the feasibility of functional avoidance radiotherapy planning for the definitive treatment of lung cancer. To characterize the achievable dosimetry of target and normal tissues of functional image-guided dose redistribution. Material and Methods In all 15 consecutive patients with locally advanced non- small cell lung cancer (NSCLC) were included prospectively. Patients were planned to receive definitive (chemo)-radiation therapy (RT) of minimum 60 Gy. Perfusion SPECT/CT were performed prior to RT commencing. Functional lung, identified as 20-80% subvolumes of the maximum perfusion count (FL20-80), was segmented on SPECT/CT, and registered to planning CT. Two plans were optimized: 1) a reference CT-plan, blinded for functional structures, and 2) functional avoidance SPECT-plan, imposing higher priority on functional levels. The objective was to reduce dose to the highly perfused lung subvolumes without compromising PTV coverage, and respecting dose to other organs at risk (OAR) within the predefined constraints. Based on our previous study, dose constraint of 16 Gy to the FL40 sub volume, was used in functional dose planning to reduce the risk of radiation pneumonitis. For each patient a 3D- conformal, intensity modulated (IMRT) and volumetric arc (VMAT) plans were created for both reference and functional avoidance. In all six plans were produced per patient. Anatomical versus functional dose-volume parameters for functional lung subvolumes, and other OAR

Figure 1: Required range robustness for varying range error.