Abstract
INTRODUCTION: Currently, data on O-(2-[(18)F]fluoroethyl)-L-tyrosine (FET) PET and MRI parameters for the evaluation of response to regorafenib in glioma patients remain scarce. METHODS: Twenty patients (age range, 30-72 years) with recurrent CNS WHO grade 3 or 4 gliomas (glioblastoma, 85%) were treated according to the REGOMA trial. FET PET and MRI were performed at baseline and follow-up after the second cycle. Static FET PET parameters comprised tumor-to-brain ratios (TBR) and metabolic tumor volumes (MTV). Parameters derived from dynamic FET PET acquisition were time-to-peak and slope. MRI response assessment was based on RANO criteria. Additionally, various apparent diffusion coefficients (ADC) parameters were obtained from diffusion-weighted MRI. Thresholds derived from FET PET and ADC parameters were defined using ROC analyses to predict an overall survival (OS) of ≥6 months. The predictive values of FET PET parameters, ADC values, and RANO criteria were subsequently evaluated using univariate and multivariate survival estimates. RESULTS: Patients received a median of three regorafenib cycles (range, 2-10 cycles). After treatment initiation, the median follow-up was 8.6 months (range, 3.2-27.3 months). After two cycles of regorafenib, a reduction of mean TBR values by ≥10% predicted significantly longer OS (9.9 vs. 5.3 months; P=0.023). Additionally, absolute mean TBR values ≤2.0 at follow-up were prognostic and associated with a significantly longer OS (10.6 vs. 4.5 months; P=0.007). In contrast, MTV, dynamic PET parameters, RANO criteria, and ADC values were not predictive for response or were prognostic (all P >0.05). Multivariate survival analyses revealed that relative mean TBR changes were most potent in predicting response to regorafenib (P=0.038; HR, 0.246) and absolute mean TBR values for prognostication (P< 0.001; HR, 0.005). CONCLUSION: In contrast to MRI metrics, FET PET parameters are clinically valuable for identifying responders to regorafenib early after treatment initiation and helpful for prognostication.