ESTRO 2020 Abstract book

S813 ESTRO 2020

workflow. Daily target positional offsets are corrected by shifting the treatment field apertures of a pre-defined reference plan, based on registration between daily and reference image. Due to target proximity to patient surface and lung, reference plan dosimetric robustness to patient contour and lung variations was assessed. Material and Methods Treatment planning was performed as per the PERMIT trial (NCT03727698). MRL reference plans were created in the Monaco v5.40.01 (Elekta) treatment planning system (TPS) for three right-sided partial breast patients (40.05Gy in 15 fractions). Backup VMAT plans for C-arm linac were generated in RayStation v7.99.09 (RaySearch Laboratories, Stockholm). A virtual bolus approach was utilised during optimisation for all plans to generate skin flash and avoid creation of superficial dose-boosting segments. This bolus was subsequently removed for final dose calculation. Each plan was subjected to the following perturbations: - Patient contour expansion: 5mm layer of water density material (1.00g/cm 3 ) added - Patient contour contraction: 5mm layer of patient removed by applying air density override (0.01g/cm 3 ) - Lung expansion: 5mm layer of lung density material (0.26g/cm 3 ) added - Lung contraction: 5mm layer of lung removed by applying water density override Override regions were only created within 10mm of PTV. Results Reference plan target dose metrics and perturbed plan dose differences are given in Table 1. Organ at risk constraints were achieved for all plans and are not reported. Patient contour expansions produce no remarkable differences between plan types, as anticipated given that the virtual bolus accounts for this. For MRL plans, patient contour contractions incur smaller reductions in target coverage than C-arm linac plans, which may be due to reduced depth of maximum dose in a magnetic field. For MRL plans, expansion of lung volume incurs larger increases in high dose regions (D5%: +41cGy, D2%: +69cGy) with no comparable C-arm linac plan changes (Figure 1). This may be attributed to the electron return effect. Electrons deflected by the magnetic field induce dose enhancements upstream of air cavities. This is mitigated in the reference plan geometry during plan optimisation but not in the perturbed scenario. Minor dose differences are also observed for the contracted lung scenario.

and worst error scenarios. The significance of the results was analyzed with the Wilcoxon signed-rank test. Results Figure 1 shows the relative differences in CTV D 98 , mean CTV, mean heart, and mean lung doses in the different evaluation scenarios. Robust plans show a minor decrease in CTV D 98 in the VW minimum (-0.23 ± 0.28 %, p = .028), while showing no significant difference in the VW mean (- 0.19 ± 0.45 %, p = .12) and worst (-0.05 ± 0.48 %, p = .69) scenarios. While no non-robust plan met the D 98 -criterium in the VW minimum scenario (94.49 ± 0.30 %) either, the found spread was less than in the worst (95.72 ± 0.54 %) and VW mean (96.94 ± 0.51) scenarios. As a result, the VW minimum CTV D 98 shows the largest correlation with the current PTV D 98 > 95 % criterium and seems, at a slightly lower threshold, most suitable for robust planning to achieve similar target coverage.

After scaling the robust plans to the VW minimum CTV D 98 of the non-robust plans, all evaluation scenarios show a decrease in mean heart and lung doses compared to the non-robust plans. In the VW maximum scenario, the relative differences in heart and lung doses were equal to -3.3 ± 2.9 % ( p = .0029) and -2.3 ± 2.9 % ( p = .0096). The worst (-3.5 ± 2.7 %, p = .0029; -2.4 ± 2.7 %, p = .0061) and VW mean (-3.1 ± 2.7 %, p = .0048; -2.6 ± 2.7 %, p = .0061) scenarios show similar differences. Conclusion Robust planning for left-sided breast cancer patients resulted in plans with a target coverage similar to margin based plans. At the same time, robust plans show the potential to decrease the mean heart and lung doses for some patients. PO-1437 Treatment plan robustness analysis for high field MR-linac partial breast plans R.A. Mitchell 1 , A. Dunlop 1 , J. Chick 1 , J. Mohajer 1 , E. Goodwin 1 , S. Nill 1 , R. Lawes 2 , T. Herbert 2 , A. Kirby 2 , U. Oelfke 1 1 The Institute of Cancer Research and The Royal Marsden NHS Foundation Trust, Joint Department of Physics, Sutton, United Kingdom ; 2 The Royal Marsden NHS Foundation Trust, Radiotherapy, Sutton, United Kingdom Purpose or Objective The local implementation of partial breast radiotherapy on the Unity MR-linac (MRL, Elekta AB, Stockholm) consists of fixed field IMRT delivered via an adapt-to-position (ATP)

Conclusion