Recurrent glioblastoma multiforme: ADC histogram analysis predicts response to bevacizumab treatment.

PURPOSE To determine if apparent diffusion coefficient (ADC) histogram analysis can stratify progression-free survival in patients with recurrent glioblastoma multiforme (GBM) prior to bevacizumab treatment. MATERIALS AND METHODS The study was approved by the institutional review board and was HIPAA compliant; informed consent was obtained. Bevacizumab-treated and control patients (41 per cohort) diagnosed with recurrent GBM were analyzed by using whole enhancing tumor ADC histograms with a two normal distribution mixture fitting curve on baseline (pretreatment) magnetic resonance (MR) images to generate ADC classifiers, including the overall mean ADC as well as the mean ADC from the lower curve (ADC(L)). Overall and 6-month progression-free survival (as defined by the Macdonald criteria) was determined by using Cox proportional hazard ratios and the Kaplan-Meier method with log-rank test. RESULTS For bevacizumab-treated patients, the hazard ratio for progression by 6 months in patients with less than versus greater than mean ADC(L) was 4.1 (95% confidence interval: 1.6, 10.4), and there was a 2.75-fold reduction in the median time to progression. For the control patients, there was no significant difference in median time to progression for the patients with low versus high ADC(L) (hazard ratio, 1.8; 95% confidence interval: 0.9, 3.7). For bevacizumab-treated patients, pretreatment ADC more accurately stratified 6-month progression-free survival than did change in enhancing tumor volume at first follow-up (73% vs 58% accuracy, P = .034). CONCLUSION Pretreatment ADC histogram analysis can stratify progression-free survival in bevacizumab-treated patients with recurrent GBM.