ESTRO 37 Abstract book

S14

ESTRO 37

PV-0038 Esophagus toxicity after stereotactic radiotherapy of central lung tumor: NTCP modelling M. Duijm 1 , H. Tekatli 2 , E. Oomen-de Hoop 1 , W. Verbakel 2 , W. Schillemans 1 , B.J. Slotman 2 , S. Senan 2 , J.J. Nuyttens 1 1 Erasmus MC Cancer Institute, Radiation Oncology, Rotterdam, The Netherlands 2 Cancer Center Amsterdam VU University Medical Center, Radiation Oncology, Amsterdam, The Netherlands Purpose or Objective The esophagus is an organ at risk in stereotactic radiotherapy for central lung tumors. We studied the correlation between esophagus toxicity and dosimetric/volumetric parameters in order to assess risks, and to derive a Normal Tissue Complication Probability (NTCP) model. Material and Methods Patients with a central lung tumor from 2 centers who received stereotactic or hypofractionated radiotherapy (≤ 12 fractions) were retrospectively analyzed. Doses were recalculated to an equivalent dose of 2 Gy (EQD 2 ) with an α/β ratio of 10 for acute toxicity (within 3 months). The esophagus was manually delineated and dose-volume histogram (DVH) parameters (D max , D 1cc , D 2cc , D 5cc ) were evaluated. The primary endpoint was esophagus toxicity as scored by CTCAE version 4.0. NTCP was calculated based on a logistic regression model and significant parameters (p-value < 0.05) were plotted into the logistic model curve. Overall survival was calculated using Kaplan-Meier analysis and groups were compared with the log-rank test. Results Two-hundred-and-thirty-one patients (with 252 tumors) were eligible with median follow-up of 16 months (range 0.3 – 84.3). No acute or late grade 3-5 esophagus toxicity was reported. Acute grade 1-2 toxicity was recorded in 38 patients (16.5%). All DVH parameters differed significantly between patients with toxicity versus those without toxicity; median D max of 62.2 Gy 10 versus 31.6 Gy 10 , median D 1cc of 40.8 Gy 10 versus 18.8 Gy 10 , median D 2cc of 35.6 Gy 10 versus 16.3 Gy 10 and median D 5cc of 24.1 Gy 10 versus 11.1 Gy 10 , respectively (p < 0.001 for all parameters, based on Mann-Whitney U test). Out of 27 patients who received a D max ≥ 54.7 Gy 10 (equivalent of 5 x 7.5 Gy), 37% suffered from grade 1-2 esophagus toxicity. A D max ≥ 60.0 (equivalent of 5 x 8 Gy) resulted in toxicity in 1 out of 14 patients. A D 1cc ≥ 44.7 Gy 10 (equivalent of 5 x 6.5 Gy) resulted in toxicity in 6 out of 12 patients. A D 1cc ≥ 49.6 Gy 10 (equivalent of 5 x 7 Gy) resulted in toxicity in 5 out of 8 patients. Logistic regression showed significant correlations between all analyzed DVH parameters and toxicity, therefore NTCP-curves were calculated for all parameters (Figure 1 a-b). A 50% probability of acute grade 1-2 esophagus toxicity was found at a D max of 67 Gy 10 , D 1cc of 42 Gy 10 , D 2cc of 38 Gy 10 and D 5cc of 30 Gy 10 . No significant differences in overall survival were found between patients with and without toxicity: 2-year survival rate was 44% versus 51%, respectively ( p = 0.428).

Purpose or Objective It has been demonstrated that irradiating the base of the heart is linked to poorer overall survival (OS) in both early stage and locally advanced non-small cell lung cancer (NSCLC) patients 1,2 . In this study, we hypothesised that the origin of both coronary arteries are the dose sensitive structure driving this increased mortality. We therefore investigated the correlation between OS and the dose to the origin of the left and right coronary arteries (LCA and RCA) in a large, single- institution cohort. Material and Methods Two observers identified the origin of the LCA and RCA on contrast enhanced CT scans from a total of 804 patients treated between 2010 and 2013 with curative-intent radiotherapy (55Gy in 20 fractions). For 167 of 804 patients, LCA and RCA were identified by both observers, allowing intra-observer variation to be calculated. The mean lung dose (MLD) and dose to the origin of RCA and LCA (D RCA , D LCA ) were extracted from the radiotherapy plan. These were use in a multivariate survival analysis including patient and tumour characteristics (age, sex, tumour size, TNM stage, induction chemotherapy and performance status). Smoking status was available for less than one third of the cohort and was not included in the analysis. Data on co-morbidities and cause of deaths were not available, so only overall survival was analysed. Results Median follow up was 18 months and 623 deaths were recorded. Observers deviated in their identification of the LCA and RCA (average vector length 8 mm). Tumour size, sex, MLD, TNM stage, performance status as well as D RCA and D LCA were significant (p <0.05) on univariate Cox- regression. However, D RCA and D LCA did not remain significant when included in a multivariate analysis (Table 1).

Conclusion Even though the dose to the base of the heart has been linked with survival, in this cohort, the dose to the origin of the coronary arteries was not an independent predictor of OS. However, the inter-observer variation in localising the origin of the LCA and RCA was substantial, suggesting that manual identification of cardiac substructures on planning CT scans is challenging. Future work in our institution will include automatic voxel-based methods to identify the sensitive cardiac substructures in NSCLC patients to explain previous observations. 1. McWilliam et al EJC 2017 2. Stam et al R&O 2017