Pretreatment ADC Histogram Analysis Is a Predictive Imaging Biomarker for Bevacizumab Treatment but Not Chemotherapy in Recurrent Glioblastoma

BACKGROUND AND PURPOSE: Pre-treatment ADC characteristics have been shown to predict response to bevacizumab in recurrent glioblastoma multiforme. However, no studies have examined whether ADC characteristics are specific to this particular treatment. The purpose of the current study was to determine whether ADC histogram analysis is a bevacizumab-specific or treatment-independent biomarker of treatment response in recurrent glioblastoma multiforme. MATERIALS AND METHODS: Eighty-nine bevacizumab-treated and 43 chemotherapy-treated recurrent glioblastoma multiformes never exposed to bevacizumab were included in this study. In all patients, ADC values in contrast-enhancing ROIs from MR imaging examinations performed at the time of recurrence, immediately before commencement of treatment for recurrence, were extracted and the resulting histogram was fitted to a mixed model with a double Gaussian distribution. Mean ADC in the lower Gaussian curve was used as the primary biomarker of interest. The Cox proportional hazards model and log-rank tests were used for survival analysis. RESULTS: Cox multivariate regression analysis accounting for the interaction between bevacizumab- and non-bevacizumab-treated patients suggested that the ability of the lower Gaussian curve to predict survival is dependent on treatment (progression-free survival, P = .045; overall survival, P = .003). Patients with bevacizumab-treated recurrent glioblastoma multiforme with a pretreatment lower Gaussian curve > 1.2 mu m(2)/ms had a significantly longer progression-free survival and overall survival compared with bevacizumab-treated patients with a lower Gaussian curve < 1.2 mu m(2)/ms. No differences in progression-free survival or overall survival were observed in the chemotherapy-treated cohort. Bevacizumab-treated patients with a mean lower Gaussian curve > 1.2 mu m(2)/ms had a significantly longer progression-free survival and overall survival compared with chemotherapy-treated patients. CONCLUSIONS: The mean lower Gaussian curve from ADC histogram analysis is a predictive imaging biomarker for bevacizumab-treated, not chemotherapy-treated, recurrent glioblastoma multiforme. Patients with recurrent glioblastoma multiforme with a mean lower Gaussian curve > 1.2 mu m(2)/ms have a survival advantage when treated with bevacizumab.