Abstract Book

S767

ESTRO 37

metrics. The contours by Clinician involved in development of contouring atlas with input from cardiologist and cardiac radiographer was used as standard contour for comparison. DICE coefficents were calculated to compare the delineated structures of each operator with those of these standard contours. Results

is to report on the incidence and predictive variables of chest wall pain following lung SBRT with the same dose and fractionation in a single institution. Material and Methods Clinical, tumour and dosimetric parameters were collated and analysed in patients treated with lung SBRT with 48Gy in 3 fractions. Toxicity was graded by two physicians according to Common Terminology Criteria for Adverse Events v3.0. Results A total of 135 number of patients were treated between 2007 and 2016 with a median follow up of 20 months. A total of 12/135 (9%) patients developed CW pain, of whom 8 (6%) had grade 1, 3 (2%) had grade 2, 1 (0.7%) had grade 3 reactions. Univariate analysis demonstrated that V30, V40, V50 and PTV volume significantly predicted chest wall pain but were later proved non- significant on multi-variate calculations. Dosimetric variables such as D1cc, D2cc, D5cc and overlap volume as well as clinical parameters such as age, osteoporosis and diabetes all failed to show significance. Conclusion CW pain is an important late toxicity causing significant morbidity following lung SBRT in certain subsets of patients. Our analysis could not find significant correlations between any of the DVH variables and the incidence of CW pain. Further multi-institutional analysis is required to develop clearer dosimetric constraints. EP-1408 An investigation of hybrid planning with multi criteria optimization for mesothelioma T. Mitchell 1 , A. Cascales 1 , W. Poon 1 , R. Valentine 1 , S. Currie 1 , S. Harrow 1 1 Beatson Cancer Center, Clinical Oncology, Glasgow, United Kingdom Purpose or Objective To evaluate the treatment plan quality produced by two treatment planning approaches for mesothelioma; the use of hybrid IMRT and VMAT techniques, with and without multi-criteria optimization (Trade-Offs). Material and Methods Seven palliative patients, previously treated as part of the System Clinical trial were planned retrospectively using Eclipse TPS v15.5 [Varian Medical Systems, Palo Alto, CA, USA] with a hybrid IMRT-VMAT plan. This consisted of either a 2 or 4 field parallel opposed IMRT plan combined with two VMAT fields (P_ HYB ). These plans were then optimised with Trade-Offs (P_ TO ), a multi- criteria optimisation planning method available in Eclipse v15.5. This cohort of patients had extensive disease with one of either lungs included within the PTV. The bulky PTV was prescribed to receive 55Gy in 25 fractions and the remaining pleural cavity of the involved lung treated to 45Gy in 25 fractions. Individual PTV volumes varied from 764 to 4924cc. Plans consisted of up to three isocentres and were normalized to ensure that 95% of the bulky PTV received 95% of the prescribed dose. Additional, Trade- Off optimisation was applied to both the IMRT and VMAT plans creating a second hybrid plan that was normalised by the same method. Plan quality was evaluated by: volume of the lung receiving 5Gy (V5Gy<75%), 20Gy (V20Gy<10%) and a mean dose <10%, volume of heart receiving 35Gy (V35Gy<35%) and mean dose <26Gy, mean dose to liver <30Gy, volume of ipsi-lateral kidney receiving 30Gy (V30Gy<50%), dose to 0.1cc of esophagus <55Gy, spinal cord <50Gy, small bowel <45Gy and stomach <45Gy. Results Bulky target volume coverage was achieved by normalisation in each case. Both plan types were able to achieve OAR objectives for lung V20Gy and mean lung dose, mean heart dose, volume of ipsi-lateral kidney receiving 30Gy, dose to

Structures outlined were sinoatrial node (SA), atrioventricular node(AV), SVC, pulmonary artery (PA), aortic valve(A valve), mitral valve(M valve ), pulmonary valve(Pulm valve), triscuspid valve (T valve), left and right chambers – LV, RV, LA, RA. The left ventricular muscle was contoured into 17 segments, which were then combined to form the left anterior descending artery (LAD), left circumflex artery (LCX), and right coronary artery (RCA) circulation territories. The developed contouring atlas was used to contour five patients (Figure 1 – page 2 of atlas). The heart contour volumes for the five patients ranged from 524-1118 cm 3 , AV node 2.6-8.5 cm 3 , SA node 0.79- 4.94 cm 3 , A valve 6.1-12.8 cm 3 , Pulm valve 3.9-11.5 cm 3 , M valve 1.8-6.1 cm 3 , LAD 7.1-180.7 cm 3 , RCA 9.3-15.6 cm 3 , and LCX 13.2-18.7 cm 3. The great vessels (SVC and PA) were contoured within the vicinity of the heart volume, and the intention would be to use the maximum dose DVH parameter. The contour overlap was better for the heart chambers (Figure 2 mean Dice coefficient for contours), the mean DICE coefficient for left ventricle was 0.9 (SD 0.05) and ranged from 0.78-0.9 for the 4 heart chambers. The heart substructures demonstrated less overlap mean DICE coefficient ranged from 0.35-0.6 and the coronary artery territories showed less overlap whereby mean DICE coefficient ranged from 0.26-0.36. Conclusion The atlas facilitated heart substructure contouring and the variation in structures is acceptable given the smaller volumes. The atlas will be validated in a cohort of lung cancer patients. EP-1407 Dosimetric Variables for Chest Wall Pain following Lung Stereotactic Body Radiotherapy C. Pembroke 1 , J. Albers 2 , J. Kildea 3 , W. Parker 4 , S. Faria 3 1 Velindre Cancer Centre, Clinical Oncology, cardiff, United Kingdom 2 McGill University, Radiation Oncology, Montreal, Canada 3 McGill University Health Centre, Radiation Oncology, Montreal, Canada 4 McGill University Health Centre, McGill University, Montreal, Canada Purpose or Objective Chest wall (CW) pain is a recognized late complication of stereotactic body radiotherapy (SBRT) to the lung. Despite multiple retrospective case series investigating the influence of dose volumes histogram (DVH) metrics and CW pain, there remains no clear consensus on what the constraints should be. Furthermore, differing dose and fractionation schedules could influence the incidence of the pain and interpretation of these metrics. Our aim