Evaluation of 18F-FDG PET and DWI Datasets for Predicting Therapy Response of Soft-Tissue Sarcomas Under Neoadjuvant Isolated Limb Perfusion

Purpose: To evaluate and compare the clinical utility of simultaneously obtained quantitative 18F-fluorodeoxyglucose positron-emission-tomography (18F-FDG PET) and diffusion-weighted imaging (DWI) datasets for the prediction of histopathological therapy response of soft tissue sarcomas (STS) under neoadjuvant isolated limb perfusion (ILP). Methods: A total of 37 patients with confirmation of a STS of the extremities underwent an 18F-FDG PET/magnetic resonance imaging (MRI) examination before (1st scan) and after (2nd scan) ILP with melphalan and TNF-α. For each patient, the maximum tumor size, metabolic activity (standardized uptake values, SUV) and diffusion-restriction (apparent diffusion restriction values, ADC) were determined in pre- and posttherapeutic examinations and percentage changes during treatment were calculated. A Mann-Whitney U test was used and receiver operating characteristic (ROC) analysis was performed to compare the results of the different quantitative parameters to predict histopathological therapy response. Results from histopathological analysis after subsequent tumor resection served as reference standard and patients were defined as responders/non-responders based on the grading scale by Salzer-Kuntschik. Results: Histopathological analysis categorized 22 (59%) patients as therapy responders (Grade I-III) and 15 (41%) patients as non-responders (Grade IV-VI). Tumors in the responder group showed a mean reduction in size of -9.7% and metabolic activity (SUVpeak: -51.9%; SUVmean: -43.8%) as well as an increase of the ADC values (ADCmin: +29.4% and ADCmean: +32.8%) under treatment. Percentage changes in the non-responder group amounted to: tumor size -6.2%; SUVpeak: -17.3%; SUVmean: -13.9%; ADCmin: +15.3% and ADCmean: +14.6%. Changes of the SUVs and ADCmean values between responders and non-responders were significantly different ( 0.05). The corresponding AUCs were 0.63 (tumor size), 0.87 (SUVpeak), 0.82 (SUVmean), 0.63 (ADCmin), 0.84 (ADCmean) and 0.89 (ratio: ADCmean/SUVpeak), respectively. Conclusion:18F-FDG PET and MR-derived quantitative imaging parameters (SUVs and ADCmean) and their combination reveal a good performance for the prediction of histopathological therapy response of STS under neoadjuvant ILP. Therefore, integrated PET/MRI could serve as a valuable tool for pretherapeutic assessment as well as monitoring of neoadjuvant treatment strategies of STS.