ESTRO 2020 Abstract Book

S158 ESTRO 2020

PD-0306 Breathing motion robustness of 4D-CT and ITV based treatment plans in lung cancer IMPT O. Pastor-Serrano 1 , D. Lathouwers 1 , S. Habraken 2 , M. Hoogeman 2 , Z. Perko 1 1 TU Delft, Radiation- Science and Technology, Delft, The Netherlands ; 2 Erasmus MC, Radiotherapy, Rotterdan, The Netherlands Purpose or Objective In Pencil Beam Scanning (PBS) proton therapy treatments, breathing interplay effects arise from motion of the organs during delivery. This study aims to investigate the robustness of clinically planned Internal Target Volume (ITV) based plans against breathing motion uncertainties (represented by a breathing signal depicting movement of chest markers) in PBS lung patients, and to compare them Two different treatment plans were made using 4D-CT and ITV optimization for the same lung cancer patient. The dataset consists of the reference 50% exhale delineated CT set, and the delineated CT sets for 4 auxiliary phases: 0%, 25%, 75% and 100% inhale. The ITV optimization is conducted on the ITV of the 50% reference phase, with the ITV being defined as the union of the Gross Target Volumes (GTV + nodes) of all present phases. A 2 mm margin expansion, 5 mm setup errors and 5% range errors are used in the robust planning based on clinical practice. Conversely, the 4D-CT plan is performed optimizing on the Clinical Target Volumes (CTV + nodes) of all 5 phases. Robustness of treatment plans against breathing motion uncertainties is evaluated by calculating dose distributions using an in-house developed interplay calculation engine that distributes spots over different phases, for each breathing signal. The input to this engine are either the recorded signal (at different starting points for different fractions) or statistically generated breathing signals (1 per fraction) representing motion uncertainty. The artificial signals are obtained through sampling a reduced order model (ROM) based on Principal Component Analysis (PCA) and a multilayer perceptron neural network (NN). The ROM accuracy in the reconstruction/generation of signals was controlled by the number of modes included, and it was evaluated through comparison of Dose Volume Histogram (DVH) distributions. To evaluate interplay effects, DVH distributions of both plans were compared. Results The Mean Absolute Error (MAE) between DVH distributions using 1000 fractions from the real signal and signals generated from the ROM using 3 modes (parameters) is 0.1194 Gy and 0.1079 Gy for the 4D-CT and ITV plans, respectively. The low MAEs indicate that the ROM model is valid to quantify motion uncertainty. Regarding interplay effects on GTV + nodes coverage, the average D 98 for the 4D-CT plan is 56.36 Gy, with 95% confidence interval [53.88, 58.64] Gy. For the ITV plan, the average D 98 is 57.34 [54.97, 59.71] Gy. ITV planning yielded only slightly higher D 98 and narrower DVH distribution, and both approaches share similarly degraded robustness. with 4D-CT based plans. Material and Methods

radiation dose from the CtE constraints. PBT plans were prescribed to the CTV and assessed under uncertainty conditions of ±5mm shifts in all axes, and a nominal ±3.5% stopping power uncertainty. The CTV and OAR dose statistics from the nominal PBT plans were then compared to those from the original clinically delivered photon SABR plans. Results Dosimetric comparison of relevant OAR statistics seen in Figure 1 shows a marked decrease in OAR dose using PBT. Only the small bowel saw any sizeable increase in dose in the PBT plan, seen in only 3 patients. A smaller increase by PBT was seen in the cauda equina, vessels, and colon, for a 4 th patient. The largest dose reductions were seen for the sacral plexus, up to 18Gy.

Conclusion PBT has the potential to make significant dose reductions for OARS in the pelvic reRT setting. These savings are most pronounced for OARs that abut the target on one side. Less impact is seen when OARs surround the target, as with small bowel. This offers the possibility of increased eligibility within this patient group, to include those who due to OAR doses may have been ineligible for photon SABR, and may even provide room for dose escalation. Whilst this study has made initial steps in the development of the PBT planning technique to align with practical deliverability, further work is required to investigate robustness under inter- and intra-fraction patient variation to determine effective set up and imaging strategies for a PBT technique for this patient group.