Prognostic value of relative cerebral blood volume in patients with recurrent glioblastoma multiforme treated with bevacizumab

BACKGROUND: The aim of this study was to assess whether the early monitoring of the effects of bevacizumab in patients with recurrent glioblastoma multiforme (GBM) using perfusional dynamic susceptibility contrast (DSC) magnetic resonance imaging (MRI) before and after the beginning of antiangiogenic therapy is predictive of treatment response. METHODS: Thirteen patients with recurrent GBM underwent perfusion MRI with relative cerebral blood volume (rCBV) mapping before (T0) and after the beginning (T1) of bevacizumab treatment. Recurrence Regions of Interest (RoIs) were positioned on the enhancing component of tumoral tissue revealed by postcontrast T1-weighted images. The rCBV measurements on the corresponding maps were made before and after the start of the antiangiogenic therapy. The Cox proportional hazards model and the Kaplan-Meier method were used with the log-Rank Test to establish whether pre- and postbevacizumab rCBV predicted progression-free survival (PFS). We tried to assess if there was a correlation between rCBV at T0 and rCBV at T1 using the Pearson's correlation coefficient. RESULTS: In the univariable analysis, rCBV was significantly predictive of PFS at T0 (HR=5.3, P=0.003) and at T1 (HR=4.14, P=0.04). Similarly, in the multivariate Cox model analysis, rCBV was predictive of PFS at T0 (HR=4.4, P=0.04) and T1 (HR=4.2, P=0.02). PFS was longer in patients whose rCBV was less than 4.50 mL/100g at T0 and less than 1.83 mL/100g at T1 than in patients with higher rCBV values. There was a moderate positive correlation between rCBV at T0 and rCBV at T1 (P=0.032, R=0.546). CONCLUSIONS: Despite the limited number of enrolled patients, rCBV assessed using DSC-MRI through the parameter rCBV is proved reliable in predicting the effects of antiangiogenic treatment in patients with recurrent GBM.