ESTRO 2021 Abstract Book

S991

ESTRO 2021

treatments (antiblastics, surgery, radiotherapy) were collected. The cohort of pts of the first Institution (107) were used to build the model. Cox regression model and neural networks were applied. The cohort of pts of the secont Institution was used as external validation. The Kaplan-Meier curves was elaborated for the survival analysis and the Log Rank test was used for the internal comparison. Results Results: In the first cohort the median OS was 47 months while it was not reached in the second series due to the shorter follow-up (29 vs 14 months). All the variables considered were well balnced in the 2 cohorts. At multivariate analysis PS and ONeo were independently correlated to OS (PS HR=0.28 95% - CI 0.10-0.51, p=0.000; ON HR=0.45 95% - CI 0, 25-0.83, p=0.001).The outcome of the neural network analysis showed a different degree of importance for the variables considered: ONeo=100%, GOLD=91.4%, PS=54.4%, CCI=46%, age=44,1%, Pack years=35.6%, number of lesions=9.8%. The model was built combining findings of regression and neural networks. The variables found to be statistically significant in the Cox regression or with importance to the neural network greater than 50% were assigned a maximum score of 2, variables with neural network importance ranging between 10% and 50% were assigned a score of 1, for all other variables a score of 0 wes assigned. Pts of the first Institution were divided in two prognostic groups related to the total score: Group 1 for score 0-2 and Group 2 for score 3-9. This grouping was found to be close to statistical significance (p=0.081) in terms of OS. One, 2- and 4-years OS resulted 93%, 85% and 65% for Group 1 and 87%, 65% and 42% for the Group 2. The survival analysis were performed for the external cohort after the same grouping. The trend in OS was confirmed with a 1- and 2 years OS of 100% and 100% for Group 1 and 95% and 75% for Group 2. Conclusion Conclusions: These results show the role of differnet factors affecting the prognosis of early stage NSCLC after SBRT. This predictive model is applicable in the clinical practice to stratify SBRT candidate patients and identify different prognostic groups. Neural networks demostrated to be an addictional tool to provide useful data for the construction of a prognostic model that need to be validated in a prospective cohort PO-1195 Residual MTV after chemoradiotherapy ± immune checkpoint inhibition for inoperable stage III NSCLC J. Taugner 1 , M. Unterrainer 2 , L. Käsmann 1 , C. Eze 1 , W. Kunz 3 , A. Tufman 4 , N. Reinmuth 5 , C. Belka 1 , F. Manapov 1 1 LMU Munich, radiation oncology, Munich, Germany; 2 LMU Munich, department of nuclear medicine and department of radiology, Munich, Germany; 3 LMU Munich, department of radiology and department of nclear medicine, Munich, Germany; 4 LMU Munich, pulmology, Munich, Germany; 5 Asklepios Lungenfachklinik Müchen Gauting, thoracic oncology, Munich, Germany Purpose or Objective Metabolic-tumor-volume (MTV) of non-small cell lung cancer (NSCLC) on FDG-PET is an independent prognosticator for patients treated with conventional chemoradiotherapy (CRT). We compared the impact of residual MTV after CRT for inoperable stage III NSCLC in patients treated with vs. without immune check-point inhibition (ICI). Materials and Methods Fifty six inoperable stage III NSCLC patients (16 female, median age: 65 years) underwent 18 F-FDG PET/CT at the same institution before and after completion of CRT. MTV was delineated on 18 F-FDG PET/CT using a standard threshold (hepatic SUV mean + 2 x standard-deviation). Patients were distributed into volumetric subgroups using median split dichotomization (residual MTV ≤4.0 ml & >4.0 ml). Residual MTV, clinical features and ICI maintenance (RCT-IO; 21/56 (37.5%) patients) were correlated with clinical outcome (progression free survival (PFS), local PFS (LPFS), metastasis-free survival (MFS), and overall survival (OS). Results Median follow-up was 52.0 months. 52 (93%) patients were treated with CRT, 12 (21%) patients with CRT followed by durvalumab, and 9 (16%) patients treated with CRT plus nivolumab (concurrent and sequential). Smaller residual MTV was associated with longer PFS in the entire cohort (median 29.3 vs. 10.5 months, p=0.015); in patients treated with CRT+ICI PFS was significantly longer compared to the CRT-only subgroup (median 29.3 vs. 11.2 months, p=0.001). However, residual MTV was only predictive for longer PFS in CRT (median 33.5 vs. 8.6 months, p=0.001), but not for the CRT-ICI patients (p=0.909). Analogously, patients with smaller MTV had a longer LPFS in the entire cohort (median 49.9 vs. 16.3 months, p=0.002); CRT-ICI patients showed a significantly longer LPFS compared to CRT-only patients (median not reached vs. 16.9 months, p=0.016). Residual MTV remained a significant prognosticator for LPFS in the CRT- only (median 49.9 vs. 10.1 months, p=0.001), but not in CRT-ICI patients (p=0.291). For OS, smaller residual MTV remained a significant prognosticator in the CRT-only subgroup (median 63.0 vs. 16.3 months, p=0.004), but not in CRT-ICI patients (p=0.720). Even in patients with larger residual MTV, the application of ICI significantly improved OS compared to CRT-only subgroup (median not reached vs. 22.9 months, p=0.004). Conclusion In patients with inoperable stage III NSCLC undergoing CRT with concurrent and/or sequential ICI clinical outcome was independent of residual MTV. In contrast, larger residual MTV is associated with impaired outcome in patients undergoing CRT-only PO-1196 Role of prophylactic cranial irradiation in limited and extensive-stage small cell lung cancer Y.J. Lim 1 , C. Song 2 , H.J. Kim 2 1 Kyung Hee University School of Medicine, Radiation Oncology, Seoul, Korea Republic of; 2 Seoul National University College of Medicine, Radiation Oncology, Seoul, Korea Republic of