ESTRO 2021 Abstract Book

S646

ESTRO 2021

PD-0813 Metabolic and texture analysis for predicting prognosis in locally advanced squamous cervix cancer. S. Pedraza 1 , A.P. Seiffert 2 , P. Sarandeses 3 , B. Muñoz-Lopez 2 , E.J. Gómez 2 , P. Sánchez-González 2 , J.F. Perez- Regadera 1 1 h.U. 12 De Octubre, Radiation Oncology, Madrid, Spain; 2 universidad Politécnica, Biomedical Engineering And Telemedicine Centre, Madrid, Spain; 3 h.U.12 De Octubre, Nuclear Medicine, Madrid, Spain Purpose or Objective Tumoral disease is heterogeneous, so discriminating individual biologic aggressiveness at initial disease may lead to reduce relapse rate and improve overall survival (OS). To better personalize the treatment, radiomics has been used to describe intratumoral heterogeneity and to predict treatment outcome in cancer patients. The aim of this study is to investigate the prognostic value of the texture features extracted from 18 F-FDG PET/CT in locally advanced squamous cell cervix cancer (LASqCC). Materials and Methods We retrospectively analyzed 91 patients (median:50 years (25-84)) with LASqCC who received radical treatment from May 2009 to August 2017. All patients received external beam radiotherapy treatment with weekly Cisplatin and brachytherapy. Classical prognostic factors (age, FIGO stage, tumor diameter, parametrium, hydronephrosis or lymph node affection, and SCC-Ag and CYFRA levels) were analyzed. A total of 48 radiomic features consisting of conventional indexes (SUVmax, metabolic tumor volume (MTV), total lesion glycolysis (TLG)), shape indexes, histogram statistics, second-order parameters derivered from the Gray-Level Co-Occurrence Matrix (GLCM) and high-order indexes derivered from Gray-Level Size-Zone Matrix (GLSZM), Gray-Level Run Length Matrix (GLRLM) and Neighbourhood Gray-Level Difference Matrix (NGTDM) were also studied. Only primary tumor features were extracted with a semi-automatically method using LifeXâ software. Univariant survival curves were calculated using the Kaplan-Meier method, stratifying continuous variables using the median, and differences were evaluated using the log-rank-test. Univariant and multivariant analyses were studied with the Cox regression analysis. SPSS software (V19.00, IBM Corp, Armonk, NY) was used. P value <0.05 was considered statistically significant. Results The median follow-up was 56 months (range: 1-129). FIGO stages were: 1.1%-IB2, 9.9%-IB3, 2,2%-IIA, 27,5%-IIB, 6.6&-IIIB, 41,8%-IIIC1, 8,8%-IIIC2, 2.2%-IVA. Median survival was 65.7% at 5 years. A total of 82pts. had survival data available (32pts. were dead and 50pts. alive). In univariant Kaplan-Meier curves, FIGO stage (p=0.014), tumor diameter <4cm (p=0.03), hydronephrosis (p=0.079), showed higher OS. Stratifying for sphericity (p=0.015), compacity (p=0.009), 2 features from GLRLM and 3 indices from GLZLM also resulted in statistically significant differences in survival curves. After applying Cox regression model, FIGO stage, SCC-Ag and CYFRA levels, TLG, sphericity, compacity as well as 3 indices from GLZLM have significant association with OS. In multivariant analysis, SCC-Ag and CYFRA levels, sphericity and GLZLM SZE were significant predictors of OS (table 1).

Conclusion Tumor heterogeneity on pretreatment PET/CT is associated with OS in patients with LASqCC receiving definitive chemoradiotherapy and can provide prognostic information as classical prognostic factors.

PD-0814 Prediction of treatment response in patients with advanced cervical cancer using mid-PET/MRI R. Vojtíšek 1,2 , J. Baxa 3 , P. Kovářová 4 , P. Hošek 5 , E. Sukovská 4 , R. Tupý 3 , J. Ferda 3 , J. Fínek 4 1 University Hospital Pilsen, Department of Oncology and Radiotherapy , Pilsen, Czech Republic; 2 Charles University, Faculty of Medicine, Pilsen, Czech Republic; 3 University Hospital Pilsen, Department of Imaging Methods, Pilsen, Czech Republic; 4 University Hospital Pilsen, Department of Oncology and Radiotherapy, Pilsen, Czech Republic; 5 Charles University, Faculty of Medicine, Biomedical Center, Pilsen, Czech Republic Purpose or Objective We aimed to find metabolic, functional or morphological characteristics of the tumor predicting failure to achieve a complete metabolic remission (CMR) by the midtreatment PET/MRI in cervical cancer patients. Materials and Methods