ESTRO 2020 Abstract book

S964 ESTRO 2020

to compare four strategies for daily online contouring: (1) manual contouring, (2) auto-deformed contours, (3) auto- deformed with manual modifications within 1 cm from PTV and (4) auto-deformed with manual modifications within 3 cm from PTV. Material and Methods Eleven fractions from three prostate cancer patients were evaluated. Patients were treated on an MR-Linac (MRIdian, ViewRay) with daily MR imaging in a non-adaptive workflow. During daily setup, planning contours were automatically propagated from the treatment plan to the daily acquired MRI using deformable image registration in the MRIdian system (strategy 2). The rectum, bladder and femoral heads were contoured manually offline on each fraction’s images and used as a golden standard reference (strategy 1). Hybrid structures were created to simulate manual adaptation of the auto-deformed contours within 1 cm (strategy 3) and 3 cm (strategy 4) from the PTV. Auto- deformed and reference contours were compared in terms of volume and pointwise distance between contours. DVH parameters of the original plan were extracted for rectum (V40Gy, V50Gy, V60Gy), bladder (V50Gy) and femoral heads (Dmax) for all four strategies. Results The auto-deformed contours were generally larger compared to the reference, with an average increase of 13.5 cm 3 (34%) for the rectum, ( p<0.001 ), 28.5 cm 3 (21%) for the bladder ( p<0.001) and 4.7 cm 3 (7%) for the femoral heads ( p=0.002 ) (Wilcoxon signed rank test). The average (standard deviation; maximum) distance between the auto-deformed and reference contours was 1.6 (1.6; 8.61) mm for the rectum, 2.5 (1.6; 22.5) mm for the bladder and 1.0 (1.1; 8.1) mm for the femoral heads. The DVH for rectum averaged across all patients and fractions is shown in upper panel of figure 1 and worst-case fraction is shown in lower panel. Differences in DVH parameters for strategy 2, 3 and 4 when compared to strategy 1 are shown in table 1.

Three PTVs were defined adding a symmetric margin of 5, 3 and 2 mm to the ITV. The dosimetric prescription was 100 % of the prescribed dose to 95 % of the PTV volume, with the correspondent dose limits to the OARs established in AAPM Task Group 101 (Lung: D1000cc < 13.5 Gy; D1500 < 12.5 Gy. Heart: Dmax < 38 Gy; V32 < 15 cc. Esophagus: Dmax < 35 Gy; V19.5 < 5 cc. Spinal cord: Dmax < 30 Gy; V23 < 0.35 cc; V14.5 < 1.2 cc). Three similar treatment plans (same geometry and arcs) were calculated for each patient, one for each PTV. We used a CIRS electronic density phantom (062M model) to calculate the calibration curve of electron density versus Hounsfield units to the CBCT. All treatments were delivered with TrueBeam STx HD120 MLC (Varian Medical System). Abdominal compression was used to reduce respiration-induced tumor motion and to make it more reproducible. Once the treatment was finished, the radiation oncologist contoured again the GTV on each acquired CBCT. Each of the three previously calculated plans were recalculated on daily CBCT and a comparison of the GTV coverage was made between those three recalculated plans. Results Average doses in GTVs for each CT- and CBCT- based treatment plans are shown in Table 1. Average doses in OARs and relative differences between plans calculated defining the PTV adding 5 and 3 mm to the ITV are displayed in Table 2. The best coverage for GTV was obtained for 5 mm margin, and a reduction of margins could reduce significantly the GTV coverage. We consider that each hospital must evaluate its better margin definition according to its immobilization systems, delivery mode (compression, gating, tracking...), and any other parameter that affect the GTV motion and definition.

Conclusion It is important to take into account the available material in terms of immobilizers and techniques used when we search the optimum margins. In our case, we consider the best PTV definition is adding a margin of 5 mm to the ITV. This definition accomplished OARs constraints, with a right GTV coverage in all patients and sessions. There are possible improvements by using another immobilizer systems, as well as gating and tracking techniques. Funded by ISCIII PI17/01735 grant (cofunded by FEDER). PO-1663 Contouring strategies for MR-guided online adaptative radiotherapy for prostate cancer M. Lundemann 1 , K. Boye 1 , I. Wahlstedt 1 , J.B. Thomsen 1 , M. Josipovic 1 , B. Smulders 1 , A.N. Pedersen 1 , K. Håkansson 1 1 Rigshospitalet, Department of Oncology- Section for Radiotherapy, København, Denmark Purpose or Objective To assess the volumetric and dosimetric impact of contour propagation in an adaptive workflow for prostate cancer patients treated on an MR-Linac. Specifically, the aim was