Quantitative analysis of dynamic contrast-enhanced T1-weightedperfusion MR imaging identifies glioblastoma molecular phenotypes viatumoral and peritumoral approach: Preliminary results with majorgenomic biomarkers

Glioblastoma (GB) is the most common and highly lethal primary malignant brain tumor in adults. Pretherapeutic non-invasive characterization of major genomic and molecular profiles using imaging could assist in understanding disease subtypes, as well as in risk stratification and treatment planning of GB. Our purpose was to evaluate the utility of dynamic contrastenhanced T1-weighted perfusion MR imaging (DCE-pMRI) as a potential noninvasive method to predict certain genomic or molecular alterations in GB. We retrospectively reviewed thirty-six patients (M/F: 25/11; mean age: 53, age range: 29-85 years) who had pretreatment DCE-pMRI with molecular studies of their excised GBs in this review board-approved study. ROIs of the enhancing tumor (ER) and non-enhancing peritumoral T2-hyperintense region (NER) were used to calculate DCE-pMRI derived quantitative parameters of Ktrans (volume transfer constant), Kep (backflux constant), and Ve (volume of the extravascular extracellular space) as well as semiquantitative parameters including the area under the curve (iAUC) and maximum slope (MaxSlope) of the signal intensity time series curve. Molecular characteristics determined included Ki-67 labeling index, epidermal growth factor receptor (EGFR) amplification, oligodendrocyte transcription factor 2 (Olig2) expression, isocitrate dehydrogenase 1 (IDH1) mutation, and p53 status. The imaging metrics of GB with different genetic profiles were compared using KruskalWallis test with receiver operating characteristic curve (ROC) analysis, and the Spearman correlation analysis was used for identifying imaging-molecular associations. Among 30 patients with available IDH1 mutation status, 23 (76.6%) patients presented with IDH1-mutation. EGFR amplification was present in 28/36 (77.7%) patients. Mean Ki-67 labeling index was 29% (range:1.5-80%). P53 expression was present in 20/36 (55%), while Olig2 expression was negative (7/36) and positive (29/36) in all tumors. The Kruskall-Wallis test indicates that various quantitative and semi-quantitative DCE-pMRI parameters of the ER and NER were statistically different in GBs with major genomic profiles of IDH1 mutation, EGFR amplification, and Olig2 expression (p < 0.05). The VeNER exhibited the highest sensitivity and specificity for evaluating the IDH1 mutation, MaxSlope for EGFR amplification, Kep for Olig2 expression status. Ki-67 labeling index indicated significant positive correlation with all DCE-pMRI parameters (p < 0.05). However, P53 expression status (all p > 0.05) did not present a significant correlation with any of DCE-pMRI values. DCE-pMRI may have a role in identifying major genomic alterations in GB as potential targets for individualized treatment protocols. Moreover, the imaging characteristics suggested that subtype-specific treatment of NER might be substantial because the imaging-molecular associations were different in the ER and NER of GB. Further investigation with a larger cohort is necessary.