Abstract book - ESTRO meets Asia

S83 ESTRO meets Asia 2018

Purpose or Objective Pulmonary large cell neuroendocrine carcinoma (LCNEC) is a rare histological subtype of lung cancer, and optimal treatment is uncertain. The objective was to investigate the prognostic factors and optimal treatment strategy. Material and Methods For identifying patients of LCNEC, the SEER database was used, which stands for Surveillance, Epidemiology, and End Results. The search included those diagnosed between 2000 and 2014. Additionally, the overall survival (OS) rates were determined based on the Kaplan-Meier analysis, along with the disease-specific survival (DSS) rates. Meanwhile, a log-rank test was used to execute the univariate survival analysis. Finally, the Cox proportional hazard model was utilized for identifying independent variables in connection with the survival rates. Results The search found 760,065 patients diagnosed with bronchus and lung cancer, out of these only 0.51% (3,911) were diagnosed with LCNEC. In addition, 2,810 of these cases were identified as eligible patients. Meanwhile, the 5-year OS rates were 14.7%, while the 5-year DSS rate was 18.2%. The multivariate analysis discovered those £ 65 years old (P<0.001), earlier summary stage (P<0.001), female(P<0.001), surgery(P<0.001), chemotherapy (P<0.001), and radiotherapy (P<0.001) showed to be independent favorable prognostic factors. After screening, a total of 1855 patients were eligible for further treatment analyze. These patients were divided into three groups: surgery (SX) (N=743, 40%), radiotherapy and surgery (RT/SX) (N=185, 10%), radiotherapy (RT) (N=927, 50%). The survival curves adjusted for age, primary site, and chemotherapy. Within distant and localized disease, the SX group had better prognosis compared to RT in OS (HR: 2.174, 95% CI: 1.467-3.211, P<0.001) and DSS (HR: 2.508, 95% CI: 1.617-3.891, P<0.001). However, no difference was observed between SX and SX/RT. Meanwhile, with regional diseases, the SX group showed the best prognosis in OS and DSS compared with SX/RT (P<0.001) and RT (P<0.001). Conclusion Primary surgery should be the first option in operable patients. The combination therapy of radiotherapy and chemotherapy can also benefit patients, especially for those with advanced stage. PO-206 Comprehensive strategy for daily adaptive planning for lung SBRT patients with pleural effusion J. Barbiere 1 , G. Beninati 1 , A. Ndlovu 1 1 Hackensack University Medical Center, Medical Physics, Hackensack, USA Purpose or Objective Pleural effusion results in significant geometrical changes that affect lung SBRT plan quality, target coverage and normal tissue constraints. This work presents a comprehensive strategy to quickly adapt an existing plan for each of the small number of treatment fractions. Pertinent structures, plan parameters and goals are transferred onto the daily patient CT scan and an acceptable revised plan including quality assurance created for treatment within an allocated time frame. Material and Methods An acceptable treatment plan (Pln_0) was first created immediately after a case study patient 4D CT data set (Img_0) simulation with stage T1b N0 M0 IA2 non-small cell cancer complicated by pleural effusion. For simplicity and reproducibility the only optimization parameters allowed are for the physician drawn GTV_0, PTV_0 expansion, and planning ring structures. Since the first patient treatment was scheduled for two weeks later there was sufficient time to create on optimal plan using multiple iterations of routine interactive inverse planning and then record he

R3D<5mm. (d) MI_IGTV10_ DEF R3D<5mm with optimal MI displayed. (b)MI_IGTV10_ DEF R3D: 5~10mm. (e) MI_IGTV10_ DEF R3D: 5~10mm with optimal MI displayed. (c) MI_IGTV10_ DEF R3D>10mm. (f) MI_IGTV10_ DEF R3D>10mm with optimal MI displayed. Conclusion For patients after simple breathing training by keeping the respiratory frequency in 10 BMP, the four phase combinations of 0%, 30%, 50% and 90% can be considered as a potential candidate for an optimal combination to synthesis IGTV in all respiration amplitudes. PO-204 Exploring the predictive value of feature changes in PET radiotherapy of NSCLC patients C. Ma 1 1 Shandong Cancer Hospital, radiotherpy oncology physics, Jinan, China Purpose or Objective There are many studies investigating the applications of texture features in the clinical practice, but there is still not a gold standard points which kind of texture feature fit for all clinical applications. The purpose of the study is to explore: the predictive value of feature changes in PET radiotherapy of NSCLC patients. Material and Methods To explore the highest predictive value of features for radiotherapeutic efficacy in non-small cell lung cancer (NSCLC), 132 patients with histologically proven NSCLC were retrospectively investigated. All patients had both pre-treatment and post-treatment 18F-FDG PET/CT scans. Five expert physicians were invited to contour and review the tumor targets manually. We used the MATLAB program to reconstruct and segment the tumor targets. Five traditional features, including the maximum SUV, minimum SUV, mean SUV, SUV kurtosis, and tumor volume, and five texture features, including second angular moment, contrast, inverse different moment, entropy and correlation, were extracted from the tumor targets of PET scans on pre-treatment and post- treatment. The change ratios of features were calculated by features of pre-treatment and post-treatment. The cohort of patients was divided into a training data set and a validation data set. Both data sets were each divided into three groups according to the change ratio of tumor volume. Finally, the statistical significance among change ratios of features of three groups in the training data set was determined by the Mann-Whitney U test. The features extracted from PET scans on pre-treatment in the validation data set were used to validate the predictive performances of features which have statistical significance by ROC curves. Results The Mann-Whitney U test results showed that inverse different moment, contrast, correlation, and maximum SUV had statistical significance. The predictive performance of contrast is better than the other features in identification of Group 1, and the predictive performance of inverse different moment is better than the other features in identification of Group 2. Conclusion One feature cannot be used to identify all groups. We believe this method can also be applied to other types of cancers and other clinical practices. PO-205 Surgery is essential for pulmonary large cell neuroendocrine carcinoma L. Cao 1 1 Ji Lin Cancer Hospital, Radiation, Changchun- Jilin, China