ESTRO 2021 Abstract Book

S438

ESTRO 2021

Conclusion CONCLUSION: mEHT improves local disease control as well as three year survival and disease free survival at three years. These results provide strong motivation for the incorporation of mEHT into clinical practices as a radiosensitiser for locally advanced cervical cancer, especially in resource constrained settings. PH-0552 Imaging-based patient selection in MR-guided hyperthermia for locally advanced cervix cancer I. VilasBoas-Ribeiro 1 , S. Curto 1 , G.C. van Rhoon 1,2 , M. Franckena 1 , M.M. Paulides 1,3 1 Erasmus MC Cancer Institute, University Medical Center Rotterdam, Department of Radiotherapy, Rotterdam, The Netherlands; 2 Faculty of Applied Sciences, Delft University of Technology, Department of Radiation Science and Technology, Delft, The Netherlands; 3 Center for Care and Cure Technologies Eindhoven (C3Te), Eindhoven University of Technology, Department of Electrical Engineering, Eindhoven, The Netherlands Purpose or Objective Randomized clinical studies have shown the benefit of hyperthermia, heating tumors to 40-45°C, as a sensitizing agent for radiotherapy and chemotherapy. Monitoring the temperature distribution combined with precise control of energy delivery are key components for treatment success. Magnetic Resonance (MR) imaging has been clinically used to non-invasive monitor the temperature of the treated volume. However, for deep pelvic hyperthermia, there are no comprehensive systematic studies on MR thermometry accuracy, while the investigational studies lacked an objective cut-off data criteria resulting in a high probability of bias. In this study, we performed the first systematic analysis, and identified an imaging-based selection criteria that can serve as tool for prospective patient selection for MR-guided hyperthermia, and as a benchmark to indicate the resulting quality of MR thermometry assessment. Materials and Methods A total of 14 patients diagnosed with locally advanced cervix cancer were treated using hyperthermia as adjunct to radiotherapy at our institution. All patients were treated in the MR-compatible BSD-2000-3D hyperthermia system integrated in a 1.5T GE Signa Excite scanner. During hyperthermia treatment, two temperature monitoring techniques were employed: MR- and intraluminal thermometry. Intraluminal thermometry was acquired using Bowman probes, which location was identified in the anatomic images acquired prior to the start of treatment. MR thermometry accuracy was quantified with respect to the intraluminal measurements. For prospective patient selection, we formulated three imaging-based parameters to enable associating them with MR thermometry accuracy, i.e. fat volume, gastrointestinal air volume and gastrointestinal air motion. The gastrointestinal air motion was quantified using the Jaccard coefficient. We used a receiver operating characteristic (ROC) curve to analyze how predictive each feature is for acceptable MR thermometry (≤1°C). Results Figure 1 shows the ROC analysis for the patient features with regard to the predictive value for acceptable MR thermometry. Jaccard coefficient performed best in terms of AUC value and significance, and had an optimal cut-off value of 0.91. Fat volume presented an AUC≈0.5 indicating that it is not a suitable predictor for acceptable MR thermometry