ICHNO-ECHNO 2022 - Abstract Book

S32

ICHNO-ECHNO 2022

Conclusion We demonstrate a 3D ResUnet using model ensembling offers reasonable performance in auto-segmenting GTVp regions of interest across multiple independent OPC PET/CT datasets from Europe and Canada. However, degradations in performance exist for external validation sets from the United States, particularly with respect to surface distance accuracy, potentially indicating the need for additional training or model fine-tuning when applying models to datasets with different patient populations.

PO-0064 Prognostic value of CT-Determined Sarcopenia in Patients with Head and Neck Squamous Cell Carcinoma

N. Coccia 1 , N. Giannini 1 , G. Gadducci 1 , T. Fuentes 1 , R. Mattioni 1 , G. Malfatti 1 , R. Morganti 2 , F. Paiar 1

1 Azienda Ospedaliera Universitaria Pisana, Radiation Oncology, Pisa, Italy; 2 Azienda Ospedaliera Universitaria Pisana, Statistics, Pisa, Italy Purpose or Objective The aim of the study is to evaluate the impact of sarcopenia as prognostic factor in HNSCC (Head and Neck Squamous Cell Carcinoma) patients treated with definitive or adjuvant CRT at the Division of Radiotherapy, University of Pisa. Materials and Methods This single-center retrospective analysis includes 134 HNSCC patients treated with CRT from January 2015 to June 2020. Sarcopenia was evaluated by contouring skeletal muscles at level C3 on the radiation planning CT: the volume delineated was divided by the thickness of the CT-slide, in order to obtain the CSA ( cross -sectional muscle area) at the level of C3. By applying the validated algorithm described by Swartz et al. in 2016, CSA at C3 was used to estimate the CSA at L3 an then adjusted for patient height (m ² ) resulting in the Skeletal Muscle Index (SMI cm ² /m ² ). The impact of sarcopenia on OS and DFS was evaluated using Kaplan -Meier method. Results Median follow-up of the cohort was 26 months (range 1-71 months) with 71 deaths, 37 local recurrences and 15 distant recurrences. Mean SMI was 32,46 cm ² /m ² , significantly higher in men than in women (35,4 cm ² /m ² vs 25,56 cm ² /m). Cut- off values for sarcopenia, corresponding to the lowest gender-specific quartile, were set at 32,2 cm ² /m ² for men and 23,15 cm ² /m ² for women. Patient with sarcopenia (n=33) were generally older (median age 69 years vs 59 years). Although not statistically significant, there was a trend which suggested a worse OS in the sarcopenic patients [HR 1,654 (0,916-2,991), p=0,096]. In the univariate analysis, we found a statistically significant association between sarcopenia and a worse DDFS (distant disease-free survival) [HR 2,732 (1,015-7,351), p= 0,047].