ESTRO 2021 Abstract Book

S834

ESTRO 2021

Conclusion Compared to completing two scheduled cycles of CDDP, omission of a second cycle of chemotherapy is associated with poor survival. Although not powered to confirm non-inferiority, our data suggest switching to a carboplatin-containing regimen for cycle two may attenuate detriment to outcome. This approach may be considered for patients due a second cycle of concurrent chemotherapy who remain fit but with specific contra-indications to CDDP. PO-1002 Variability of pretherapeutic FDG PET/CT parameters in patients with head and neck cancers F. Lucia 1 , A. Vu 2 , V. Bourbonne 1 , O. Pradier 1 , R. Abgral 3 , U. Schick 1 1 University Hospital of Brest, Radiation Oncology Department, Brest, France; 2 University Hospital of Brest, Radiation Oncology Department, Brest, France; 3 University Hospital of Brest, Nuclear Medicine Department, Brest, France Purpose or Objective The aim of this study was to assess inter-segmentation method variability in order to define robust imaging biomarkers extracted from (18F)-fluorodeoxyglucose (FDG) positron emission tomography with computed tomography (PET/CT) in patients treated with reirradiation for a HNSCC. Materials and Methods Thirty-six patients with a local or local-regional HNSCC recurrence receiving a second course of radiotherapy using intensity modulated radiotherapy (IMRT) from 01/2012 to 05/2020 were included. All patients had a 18FDG PET/CT performed before re-irradiation. A gross tumor volume (GTV) T and/or a GTV N were defined as well as an overall volume GTV TN (GTV T +/- GTV N). These regions of interest (ROI) were delineated using 3 different segmentation methods: a relative standardized uptake value threshold (40%SUVmax), an image gradient-based method (PET-EDGE) and a manual one, and then compared using the Hausdorff distance (HD), Dice coefficient (DSC) and Jaccard index (JI). Then, for each ROI, 11 FDG PET features (SUVmax, SUVmean, SUVpeak, TLG, MTV, total SUVbw, TLRmax, TBRmax, TLRmean, TBRmean, number of voxel) were extracted and analysed using the MIM software®. To estimate their robustness and reproducibility, a comparison of these parameters according to segmentation method was conducted using the Wilcoxon and Friedman tests, and the intraclass correlation coefficient (ICC). Results 36 GTV TN, 33 GTV T and 14 GTV N were delineated and analysed. The lowest HD and the highest mean DSC and mean JI values were found between PET-EDGE and manual methods for the GTV TN, GTV T and GTVN (HD 16,5; 9,7 ;10,8mm – mean DSC 0,791; 0,78; 0,706- mean JI 0,657; 0,646; 0,555, for 40%SUVmax, PET- EDGE and manual methods, respectively). There was no significant difference according to Friedman test for