ESTRO 2021 Abstract Book

S962

ESTRO 2021

Our model corroborates previous findings regarding the possible role of the lower right lung regarding the development of APT and successfully evaluates its robustness in a prospective cohort treated by VMAT. Regional radiosensitivity should be considered in usual lung dose contraints.

PO-1159 Prognostic role of pro-inflammatory cytokines in multimodal treatment of inoperable stage III NSCLC L. Käsmann 1 , N. Cabeza-Boeddinghaus 2 , J. Taugner 1 , C. Eze 1 , B. Flörsch 1 , T. Hofer 3 , C. Pelikan 3 , C. Belka 1 , E. Noessner 3 , C. Staab-Weijnitz 4 , F. Manapov 5 1 LMU University hospital, Radiation Oncology, Munich, Germany; 2 Helmholtz Zentrum München, Institute of Lung Biology and Disease, Munich, Germany; 3 Helmholtz Center Munich, Immunoanalytics Research Group Tissue Control of Immunocytes, Munich, Germany; 4 Helmholtz Zentrum München, Institute of Lung Biology and Disease , Munich, Germany; 5 LMU University, Radiation Oncology, Munich, Germany Purpose or Objective Inflammatory cytokines produced by tumor cells or inflammatory cells in the tumor microenvironment can promote tumor progression and treatment resistance. The prognostic role of dynamic changes of proinflammatory cytokines in inoperable stage III NSCLC patients treated with (chemo)-radiotherapy ± immune checkpoint inhibition is unknown. We aimed to evaluate inflammatory cytokines, including interleukin 2, 6, and 8, for their impact regarding overall survival (OS) and progression-free survival (PFS). Materials and Methods Twenty patients were prospectively enrolled in this study. Eighteen (90%) patients received platinum-based concurrent chemoradiotherapy (CRT); seven (35%) patients additional concurrent and/or sequential immune checkpoint inhibition. PD-1 and PD-L1 inhibition was performed in four and three patients, respectively. All patients treated with PD-1 inhibition (nivolumab) received induction chemotherapy. Thoracic irradiation (TRT) was applied in all patients with a median cumulative dose in equivalent 2Gy fractions (EQD2) of 64Gy (range: 52-65Gy). Blood samples were collected 5-10 days before treatment start (T1), on the last day of thoracic radiotherapy (T2), and 3 weeks after radiation (T3). Results Median follow-up was 25 (range 14-30) months, median OS was not reached and median PFS was 11.8 (95%CI 5.2-18.4) months. Cut-off points were established by median values to assess the predictive value of proinflammatory cytokines. Higher IL6 levels (≥9.75pg/ml) before irradiation (T1) were associated with inferior OS (median 11 months vs. not reached; p < 0.001) and PFS (median 7.0 months vs. not reached; p = 0.041). Higher IL8 levels (≥4.62pg/ml) were also associated with shorter OS (median 16 months vs. not reached; p = 0.009) and PFS (median 7 months vs. not reached p = 0.040). Patients with a decline of ≥10% of IL8 level between T1 and T2 had a significantly shorter PFS (11.8 months vs. not reached; p = 0.028). Conclusion High proinflammatory cytokine levels were significantly associated with deterioration of outcome regarding OS and PFS in patients with inoperable stage III NSCLC receiving multimodal treatment. A decline of ≥10% of IL8 level during TRT was associated with inferior PFS. PO-1160 An automated segmentation algorithm for GTV delineation in SBRT of the lung, a proof of concept. A.A. Mohamed 1 , K. Gester 1 , L. Schmitz 2 , M. Schlenter 1 , A. Chughtai 1 , M. Ivanciu 1 , M. Eble 1 1 RWTH Aachen University, Radiation Oncology department, Aachen, Germany; 2 RWTH Aachen university, Radiation Oncology department , Aachen, Germany Purpose or Objective Despite the rapid evolution of the ablative role of stereotactic body radiotherapy (SBRT) over the last years in both primary and secondary lung tumors, it still faces some challenges in the everyday-application. Interobserver varliability of the gross target volume (GTV) represents one of those challenges. We developed an easily applicable automatic delineation algorithm for GTV for either planning computer tomography (P-CT) as well as cone-beam CT (CBCT), based on the difference in Hounsfield units between the lung and region of interest. Materials and Methods Treatment Planning from 15 Patients with 18 peripheral lung lesions, treated with SBRT in our department between 1 st May 2016 till 31 st May 2017, were included in the retrospective analysis. GTVs from the 18 lesions were delineated in 15 P-CTs and 90 CBCTs by 5 radiation oncologists (experience 3-11 years) and an expert in thoracic radiation oncology (experience more than 20 years), yielding the ground truth labels for further evaluation. Delineation was based on the RTOG-0813 Study protocol. Auto-contours for the 15-CTs and 90- CBCTs were generated using the delineation algorithm. In addition 2 physicians modified another set of auto- contours 3 months later to remove any adjacent vessel or chest-wall based. Manual contours (from the 5 physicians), auto-contours as well as the modified auto contours were evaluated for their interobserver variability by comparing the spatial overlap with ground truth labels using the Dice similarity coefficient (DSC), where an spatial agreement > 0.7 with the ground truth labels considered acceptable. The mean from all DSCs from manual delineation in P-CTs as well as CBCTs were used as a reference to be compared with the auto-contours and the modified auto-contours. Finally, we compared the time that was utilized by the every physician or the algorithm-operator to delineate the GTV in each P-CT and CBCT Results The Median DSC for the auto-contours was 0.83 in P-CTs and 0.80 in CBCTs, which was improved by the manual modification of the auto-contours in modified auto contours 0.86 in P-CTs and 0.80 in CBCTs. The reference