Perfusion MRI in the Evaluation of Suspected Glioblastoma Recurrence

PURPOSETreatment-related changes (TRC) often imitate tumor progression in glioblastomas. Increased regional cerebral blood volume (rCBV) can differentiate tumor progression from TRC after the standardized first-line radiochemotherapy, but information about diagnostic accuracy of rCBV for patients without any clinical selection criteria is limited. Therefore, we aimed to evaluate if rCBV can differentiate between TRC and tumor progression irrespective of preceding therapies and number of tumor progressions. METHODSWe analyzed mean and maximum rCBV from the enhancing areas normalized to the contralateral white matter in 44 pretreated glioblastomas with MR-morphological tumor progression. The diagnosis (real progression vs. TRC) was determined by histopathology or by clinical/MRI-follow-up. We performed nonparametric tests, receiver operating characteristics (ROC), and Kaplan-Meier analysis. RESULTSSignificant differences between tumor progression (N = 37) and TRC (N = 7) were found for rCBV(mean) (2.44 1.05 vs. 1.69 +/- .56, P < .03) and rCBV(max) (3.40 +/- 1.25 vs. 2.21 +/- .62, P < .0007). A rCBV(max) of 2.6 had 78% sensitivity and 86% specificity to detect tumor progression. Neither rCBV(mean) nor rCBV(max) was predictive for the patient overall survival (OS). There were no statistically different rCBV(mean) and rCBV(max) between the first and further tumor progressions. CONCLUSIONSThe rCBV(max) differentiates tumor progression from TRC in unselected recurrent glioblastomas, but it is not predictive for the OS.