Dynamic Susceptibility Contrast MR Imaging inGlioblastomas: Correlation of Whole-TumorHistogram Analysis of the Normalized CerebralBlood Volume and Vascular Permeability withHistopathologic and Molecular Markers of TumorAggressiveness

PURPOSE To date, molecular subtypes of glioblastoma only can be identified by genetic analysis of surgical specimens. Aim of the study is to determine whether whole-tumor histogram analysis of the normalized cerebral blood volume (CBV) and vascular permeability can identify subtypes of glioblastomas (GBMs) with different histopathologic and molecular patterns of aggressiveness. MATERIALS & METHODS Dynamic susceptibility contrast (DSC) MR imaging (Avanto 1.5 T, Siemens, Erlangen, Germany) was obtained in a consecutive series of 30 patients (10 females, 20 males; mean age 59 years; age range, 32-78 years) with histologically confirmed GBMs, with a whole-brain T2*gradient-echo echo planar sequence during iv injection of a bolus of gadolinium-DTPA. Perfusion data were processed off-line with a dedicated software (NordicIce, NordicNeuroLab, Bergen, Norway) to create color-coded maps of rCBV and volume transfer coefficient (Ktrans). Regions of interest (ROIs) were manually drawn encompassing the whole contrastenhancing lesion and nonenhancing peritumoral area in normalized rCBV and Ktrans maps overlaid on axial postcontrast T1-weighted images. Data analysis was performed with the histogram analysis of normalized CBVs and Ktrans from the entire tumor volume. Brain MR examination also included DWI sequences with ADC map calculation. The following histopathologic markers were included into the statistical analysis: GBM cell proliferation (MIB), migration/invasion (Gemistocytes), angiogenesis (expression of the enhancer of Zeste 2, EZH2), epidermal growth factor receptor (EGFR), p53 mutation. RESULTS For whole-tumor histogram analysis mean rCBV (p<.001 r = .605), and rCBV maximum value (p<.05, r =.460) were found to be significantly associated with EGFR over expression. Mean and maximum rCBV values tended to be higher in patients showing increased over expression of EGFR, with significant differences between subgroups as shown by post-hoc analysis (p<.05). According to ROC curve analysis mean wholetumor CBV values helped to distinguish GBMs with EGFR over expression with 74% sensitivity and 70% specificity at a cutoff value of 1.79. Whole-tumor ADC mode (p<.05 r = -.440) and median (p<.05 r = -.430)