ESTRO 2021 Abstract Book

S813

ESTRO 2021

PO-0980 Predicting toxicity after Head-and-Neck cancer RT: synergist role of biological markers & dosimetry? E. Orlandi 1,2 , M. Duclos 3 , N.A. Iacovelli 2 , E. Berthel 3,4 , S. Deneuve 4,5 , A. Cavallo 6 , R. Valdagni 7 , T. Rancati 1 , S. Pereira 3,8 1 Fondazione IRCCS Istituto Nazionale dei Tumori di Milano, Radiation Oncology 1, Milan, Italy; 2 Fondazione IRCCS Istituto Nazionale dei Tumori di Milano, Radiation Oncology 2, Milan, Italy; 3 Neolys Diagnostics, Neolys Diagnostics, Strasbourg-Enzheim, France; 4 Centre Regional de Lutte Contre le Cancer Léon Bérard, INSERM, U1296 Unit, Lyon, France; 5 Centre Regional de Lutte Contre le Cancer Léon-Bérard, Département de Chirurgie Oncologique, Lyon, France; 6 Fondazione IRCCS Istituto Nazionale dei Tumori di Milano, Medical Physics Unit, Milan, Italy; 7 University of Milan and Fondazione IRCCS Istituto Nazionale dei Tumori di Milano, Department of Haematology and Oncology, Milan, Italy; 8 Centre Regional de Lutte Contre le Cancer Léon-Bérard, INSERM, U1296 Unit, Lyon, France Purpose or Objective To assess and validate the predictive ability of RADIODTECT © (RDT) (based on phosphorylated ATM protein quantification in lymphocytes) for severe acute/late toxicity (tox) after head&neck cancer (HNC) radiotherapy (RT). Materials and Methods 53 consecutive HNC patients (pts) treated with radical/adjuvant RT with/without chemotherapy (CHT) and prospectively and longitudinally evaluated for tox according to CTCAEv4.0 were included in a discovery cohort to test the ability of RDT in predicting acute tox ≥grade 3 (G3). RDT was performed at least 6 months after RT end. RDT discrimination power was evaluated through AUC. The ROC curve and the Youden index were used to estimate the optimal RDT cutoff, using a bootstrap analysis on 10000 resamples. 67 consecutive HNC pts, treated with radical/adjuvant RT with/without CHT and prospectively followed for tox scoring in the same way as for the training cohort, were included in the validation cohort. These pts had also late tox evaluation till 3-year follow-up (late tox≥G3). RDT and the tox scoring in the validation population were done blindly. RDT performance was tested using the cut-off previously established in the training cohort without any adjustment. Analysis on the validation population included evaluation of the possible predictive value of the RDT when added to clinical/dosimetric variables (logistic regression models). Results 13/53 pts (24.5%) exhibited≥G3 acute tox in the training cohort. RDT distribution in this cohort is shown in fig.1a, ROC curve (AUC=0.75) in fig.1b. The optimal threshold was estimated at 46 ng/ml. Using this cutoff, 9 and 44 pts were classified as radio-resistant (RR) and radio-sensitive (RS). In the validation cohort,47/67 pts (70%) and 10/67 (14.9%) exhibited≥G3 acute tox and G3 late tox, respectively. RDT distribution in the validation cohort is in fig.2a. 16/67 pts were labelled as RS (AUC=0.56) using the same cutoff of 46 ng/mL estimated on the training cohort. Fig.2b reports the ROC curve. A combined biological/dosimetric model (AUC=0.78, fig.2b) included RDT (OR=3.3 RS vs RR), the minimum dose to the combined parotid glands (cPG)(OR=1.14 for 1 Gy increase) and concurrent CHT(OR=4.4). RDT distribution for pts with/without ≥G3 late tox is in fig.2c. Classification of pts as RS/RR was significantly associated with ≥G3 late tox (chi-squared test p-value 0.037) with OR= 4.2 (AUC=0.65). A combined biological/dosimetric model (AUC=0.76, fig.2d) included RDT (OR=4.5 RS vs RR) and the mean dose to the cPG(OR=1.04 for 1 Gy increase).