MRI and 18FET-PET predict survival benefit from bevacizumab plus radiotherapy in patients with IDH wild-type glioblastoma: results from the randomized ARTE trial

PURPOSE: To explore a prognostic or predictive role of magnetic resonance imaging (MRI) and O-(2-$^{18}$F-fluoroethyl)-L-tyrosine ($^{18}$FET) positron emission tomography (PET) parameters for outcome in the randomized multi-center trial ARTE that compared bevacizumab plus radiotherapy with radiotherpay alone in elderly patients with glioblastoma. EXPERIMENTAL DESIGN: Patients with isocitrate dehydrogenase wild-type glioblastoma aged 65 years or older were included in this post-hoc analysis. Tumor volumetric and apparent diffusion coefficient (ADC) analyses of serial MRI scans from 67 patients and serial $^{18}$FET-PET tumor-to-brain intensity ratios (TBR) from 31 patients were analyzed blinded for treatment arm and outcome. Multivariate Cox regression analysis was done to account for established prognostic factors and treatment arm. RESULTS: Overall survival benefit from bevacizumab plus radiotherapy compared to radiotherapy alone was observed for larger pre-treatment MRI contrast-enhancing tumor (hazard ratio [HR] per cm$^{3}$ 0.94, 95% confidence interval [CI] 0.89-0.99) and for higher ADC (HR 0.18, CI 0.05-0.66). Higher $^{18}$FET-TBR on pre-treatment PET scans was associated with inferior overall survival in both arms. Response assessed by standard MRI-based RANO criteria was associated with overall survival in the bevacizumab plus radiotherapy arm by trend only (p=0.09). High $^{18}$FET-TBR of non-contrast-enhancing tumor portions during bevacizumab therapy was associated with inferior overall survival on multivariate analysis (HR 5.97, CI 1.16-30.8). CONCLUSION: Large pre-treatment contrast-enhancing tumor mass and higher ADC identify patients who may experience a survival benefit from bevacizumab plus radiotherapy. Persistent $^{18}$FET-PET signal of no longer contrast-enhancing tumor after concomitant bevacizumab plus radiotherapy suggests pseudoresponse and predicts poor outcome. TRIAL REGISTRATION: NCT01443676