¹⁸F-FDG PET/MRI for therapy response assessment of isolated limp perfusion in patients with soft-tissue sarcomas.

Abstract

Purpose: To assess the diagnostic potential of simultaneously acquired ¹⁸F-FDG PET- and MR datasets for therapy response assessment of isolated limp perfusion (ILP) in patients with soft-tissue sarcomas (STS). Methods: A total of 45 patients with histopathologically verified STS were prospectively enrolled for an integrated ¹⁸F-FDG PET/MR examination before and after ILP. Therapy response was assessed based on different MR- and PET-derived morphological (RECIST, MR-adapted Choi criteria) and metabolic (PERCIST) criteria. In addition, a regression model was used combining relative changes of quantitative variables to predict treatment response under ILP. Histopathological results after subsequent tumor resection served as reference standard and patients were categorized as responders/non-responders based on the six-stage regression scale by Salzer-Kuntschik. Results: Histopathological analysis categorized 27 patients as therapy responders (Grade I-III) and 18 patients as non-responders (Grade IV-VI). Calculated sensitivities, specificities, positive and negative predictive values and diagnostic accuracies were 22%, 89%, 75%, 43% and 49% for RECIST, 70%, 44%, 66%, 50% and 60% for MR-adapted Choi and 85%, 78%, 85%, 78% and 82% for the PERCIST criteria. ROC analysis revealed AUC values of 0.56 (RECIST), 0.57 (MR-adapted CHOI) and 0.81 (PERCIST), respectively. The combined regression model revealed higher values (AUC: 0.90) than for the stand-alone analysis, however, differences to metabolic parameters did not reached significance (p-value: 0.067). Conclusion: Our study demonstrates the superiority of ¹⁸F-FDG PET over MR-datasets for response assessment of STS under neoadjuvant ILP. In a clinical setting combined MRI delivers valuable information for treatment planning. However, combining the information of ¹⁸F-FDG PET and MRI may enable more accurate prediction of histopathological tumor response.