ESTRO 2022 - Abstract Book

S738

Abstract book

ESTRO 2022

While osteopontin plasma concentrations increased during chemoradiation ( p =0.07), there was a trend towards decreasing CTGF plasma over the course of chemoradiation ( p =0.08). Baseline osteopontin (r=0.579, p =0.002) as well as galectin-3 (r=0.429, p =0.032) moderately correlated with the HSV prior to chemoradiation, whereas VEGF (r=0.196, p =0.357) and CTGF (r=0.314, p =0.118) did not. Patients with non-resolving tumor hypoxia in week 2 of treatment, as determined by [ 18 F]FMISO PET, were found to exhibit significantly higher VEGF (451.1 versus 221.7 ng/mL, p <0.05) and CTGF (27.8 versus 17.3 ng/mL, p <0.05) plasma concentrations in treatment week 5. Baseline osteopontin plasma concentration was higher in patients with residual PET hypoxia at the end of treatment (38.4 versus 14.8 ng/mL, p <0.01) and predicted residual hypoxia with an AUC=0.821 (95% CI 0.612-1.000, p <0.05). Conclusion Baseline osteopontin and galectin-3 plasma levels moderately correlated with the pretherapeutic HSV and therefore indicated hypoxic tumors prior to radiotherapy. Pretherapeutic osteopontin was also associated with residual tumor hypoxia at the end of chemoradiation, providing a rationale to investigate hypoxic modification based on osteopontin plasma levels. However, as plasma hypoxia proteins do not convey any spatial distribution of tumor hypoxia, they rather add to than replace [ 18 F]FMISO PET/CT-imaging for hypoxia-based radiotherapy dose escalation.

PD-0819 Long term safety of further surveillance in HPV+ OPSCC with equivocal response on 12-week PET-CT

Withdrawn

PD-0820 Explainability of deep learning-based HPV status prediction in oropharyngeal cancer

A. La Greca 1,2 , C. Marchiori 3 , M. Bogowicz 1 , J. Barranco-García 1 , E. Konukoglu 4 , O. Riesterer 5,1 , P. Balermpas 1 , C. Malossi 3 , M. Guckenberger 1 , J.E. van Timmeren 1 , S. Tanadini-Lang 1 1 University Hospital Zurich, University of Zurich, Department of Radiation Oncology, Zurich, Switzerland; 2 ETH Zurich, Department of Information Technology and Electrical Engineering, Computer Vision Laboratory , Zürich, Switzerland; 3 IBM Research Zurich, AI Automation, Zurich, Switzerland; 4 ETH Zurich, Department of Information Technology and Electrical Engineering, Computer Vision Laboratory, Zurich, Switzerland; 5 Cantonal Hospital Aarau, Center for Radiation Oncology KSA-KSB, Aarau, Switzerland Purpose or Objective Patients with human papilloma virus (HPV)-positive oropharyngeal tumors are characterized by a more favorable prognosis when compared to their negative counterparts and, thus, hold the potential for treatment de-escalation. In clinical practice, HPV diagnosis requires the analysis of biopsy samples, while medical image analysis tools have been proposed in literature as complementary non-invasive methods. In this study, we aimed to assess the diagnostic accuracy and explainability of deep learning (DL) for HPV status prediction in computed tomography (CT) images of oropharyngeal cancer (OPC) patients. Materials and Methods One internal (n 1 =96) and two public cohorts (n 2 =498; n 3 =146) of OPC patients were employed. The dataset was split in a stratified fashion based on HPV status into training (60%), validation (20%) and test (20%) sets. All CT scans were resampled to a cubic resolution of 2 mm 3 and a sub-volume of 96x96x96 pixels was cropped. In the axial direction, the sub-volume spanned from the nasal columella to 96 pixels below, i.e., approximately the start of the lungs. On the axial plane, the crop was centered around the center of mass of the first cranial slice. ModelsGenesis, a publicly available 3D model pre- trained on lung CT, was fine-tuned to perform the classification task. The model with the highest F1-score on the validation set was selected and applied to the test set. Class activation maps (CAMs) of those test subjects belonging to the internal dataset (n=25) were obtained post-hoc by means of two explainability methods, Grad-CAM and Score-CAM. CAMs were posteriorly thresholded using the 70 th and 90 th percentile values to select the most important regions (CAM 70th and CAM 90th ) and their volumetric overlap with the gross tumor volume (GTV) was calculated using Szymkiewicz–Simpson formula for the primary tumor (GTV pt ) and the affected lymph nodes (GTV ln ), separately and together (GTV all ). Results The model achieved an AUC/accuracy/F1-score of 0.89/0.82/0.78, 0.83/0.77/0.70 and 0.87/0.79/0.74 on the training, validation, and test cohorts, respectively. Figure 1 shows the visual explanation obtained after applying Grad-CAM for two test subjects. Among the 25 internal test cases, 19 were correctly classified. An overlap between GTV all and Grad-CAM 70th of at least 0.8 was observed in 21 cases, while the same was true for 24 cases using Score-CAM 70th . The overlap coefficients of GTV all with Grad-CAM 90th and Score-CAM 90th were at least 0.5 for 13 subjects. The mean overlap coefficients of the GTV pt , GTV ln and GTV all with the different CAMs are shown in Table 1.