JS03.5.A ASSESSMENT OF 18F-FET PET-BASED RESPONSE TO BEVACIZUMAB-BASED REGIMENS IN PATIENTS WITH GLIOBLASTOMA AT RELAPSE USING THE PET RANO 1.0 CRITERIA

Abstract BACKGROUND We evaluated the recently proposed amino acid PET-based response assessment criteria (PET RANO 1.0) for their proficiency in predicting a significantly longer overall survival (OS) in patients with glioblastoma treated with bevacizumab-based regimens at relapse. MATERIAL AND METHODS Thirty-eight adult patients with IDH-wildtype glioblastoma were identified from three previously published studies (PMID 29982845, 39659835, 23053325). All patients (i) received bevacizumab in combination with a second agent (irinotecan, lomustine, or nivolumab), (ii) underwent MRI- and O-(2-[18F]fluoroethyl)-L-tyrosine (FET) PET imaging at baseline and at a median time of 8 weeks (range, 4-12 weeks) after treatment initiation, and (iii) had available PET-derived parameters, including metabolic tumor volume (MTV) as well as maximum and mean tumor-to-brain ratios (TBRmax and TBRmean). A post-hoc analysis was performed to evaluate the prediction of response using the PET RANO 1.0 criteria. In addition, ROC analyses were performed to define PET parameter thresholds for predicting an OS≥9 months. Furthermore, response prediction using the RANO 2.0 criteria for MRI was compared with the PET RANO 1.0 criteria. RESULTS According to the PET RANO 1.0 criteria, patients fulfilling the criterion Stable Disease (n=10), Partial Response (n=19), or Complete Response (n=1) had a significantly longer OS than patients with Progressive Disease (n=8) (9.0 vs. 4.0 months; P=0.001). Among the suggested thresholds by the PET RANO 1.0 criteria, only a MTV reduction ≥40% was significantly predictive of response (7.3 vs. 4.0 months; P=0.008). Optimal thresholds identified by ROC analysis differed from those proposed by the PET RANO 1.0 criteria. A reduction in MTV by ≥30% and in TBRmean by ≥4% were both predictive of longer OS (7.3 vs. 4.0 months; P=0.008, and 10.6 vs. 6.0 months; P=0.010, respectively). The RANO 2.0 criteria for MRI were less significant to predict a longer OS (9.0 vs. 6.0 months; P=0.015). CONCLUSION Our data suggest that the PET RANO 1.0 criteria can predict response to bevacizumab-based therapy in patients with glioblastoma at relapse.