ESTRO 2021 Abstract Book

S358

ESTRO 2021

Figure 2: Dose differences between clinical plans and automatically generated plans showing mean differences (X-marks) and distributions (box-whisker plots). A negative number means that the clinical has a lower parameter value. Abbreviations: VWmin/VWmax: voxel-wise minimum/maximum. Conclusion In this study we demonstrated that using a fast iterative objective weighted robust optimization approach, it is possible to fully automatically create robust clinically acceptable IMPT plans for proton treatment of lung cancer. OC-0471 Computerized selection of non-coplanar beams to enhance coplanar VMAT for nasopharyngeal carcinoma R. Bijman 1 , J. Leitão 1 , A.W. Sharfo 1 , Y. Brus 1 , B. Heijmen 1 , L. Rossi 1 1 ErasmusMC, Radiotherapy, Rotterdam, The Netherlands Purpose or Objective Nasopharyngeal carcinoma patients are often treated with coplanar VMAT, while it is well-known that patients can benefit from using non-coplanar beams. Complexity of selecting optimal non-coplanar beam setups in treatment planning and prolonged dose delivery times compared to VMAT may hamper the use of non-coplanar treatments. In this study, we used automated planning with integrated beam angle optimization to investigate the impact of adding a few (1-5) individually optimized non-coplanar beams to VMAT, with a focus on enhanced parotid glands sparing. The hypothesis was that parotid glands could be spared better due to the added non-coplanar beams, while delivery times could be kept acceptable as VMAT is already fast, and only few non-coplanar beams would be added. Materials and Methods Delineated CT scans of 28 nasopharyngeal cancer patients were included in this study. Our in-house developed TPS for automated multi-criterial optimization of both beam profiles and (non-coplanar) beam angles was first configured for plan generation for nasopharyngeal patients with parotid glands sparing as highest priority. Automated plan quality was validated against clinical plans. Next, the system was used to generate 6 treatment plans per patient: VMAT and VMAT+1 to VMAT+5, the latter with 1-5 computer-selected non- coplanar beams. In the VMAT+ n plans, VMAT and the n non-coplanar IMRT beams were simultaneously optimized. After normalization at equal PTV coverage, VMAT+ n dose distributions were compared to VMAT, considering parotid glands, submandibular glands, swallowing muscles, larynx, oral cavity, esophagus, cochleas, brains, mandible,lenses and patient dose. Results With the developed configuration of the applied autoplanning system for nasopharyngeal cancer, generated autoplans resulted in superior quality compared to clinical plans with statistically significant differences depicted in Figure 1. For all patients, addition of individualized non-coplanar beams to VMAT did indeed result in an aimed reduction in parotid glands mean dose. Parotid glands sparing increased with the number of added non-coplanar beams, but the increase often started to level off after adding 2-3 beams (Figure 2). Observed reductions were highly patient-dependent with a maximum of 5.9 Gy on 30.7 Gy for VMAT (19.2%). For all other OARs, dosimetric differences with VMAT were small, clinically of minor importance, and always without exceeding clinical constraints. Delivery time increases caused by adding non-coplanar beams to VMAT were around 1.5 minute per added beam (e.g. +4.5 min for VMAT+3).