NI-65 * DIFFERENTIATING RADIATION EFFECT AND NECROSIS FROM GLIOBLASTOMA WITH DYNAMIC SUSCEPTIBILITY CONTRAST (DSC) MRI

Abstract

BACKGROUND: In glioblastoma (GBM), pseudoprogression (PsP) appears as increased enhancement on T1-weighted imaging following concurrent chemoradiotherapy (CRT), making it difficult to distinguish from true progression. This increased enhancement, due to radiation and necrotic induced changes, is associated with effective treatment response. The goal of this study was to determine the ability of measures derived from dynamic susceptibility contrast (DSC) MRI to distinguish radiation effect (RE) and necrosis from GBM. METHODS: All patients gave informed written consent according to IRB policy. Thirty-three tissue samples from 11 previously treated patients were spatially correlated to MRI with a median = 13 days between surgery and MRI. Preload and leakage-corrected metrics of DSC-MRI were co-registered with the pre-surgical Stealth exam, and positive, median values of standardized (sRCBV) or normalized (nRCBV) relative cerebral blood volume, and normalized relative cerebral blood flow (nRCBF) were obtained within 3-mm regions precisely matched to the surgically recorded biopsy locations. Statistical analysis included generalized estimating equations (to account for multiple samples from the same subject) with receiver operator characteristic curves. RESULTS: Gender and median age were similar between tissue groups. Pathologic diagnosis confirmed 11 samples with pure RE/Necrosis and 22 samples with pure GBM. All metrics showed significant differences between RE/Necrosis and GBM (p < 0.001). Interestingly, both sRCBV and nRCBV showed equally higher sensitivity and specificity (81.8%/90.9%, thresholds = 4287 and 1.32, respectively) than nRCBF (72.7%/81.8%, threshold = 1.13) for differentiating RE/Necrosis and GBM. CONCLUSIONS: Accurate assessment of treatment response using measures of DSC-MRI is promising, as these measures show high sensitivity and specificity for differentiating RE/Necrosis and GBM. Both nRCBV and sRCBV provide equally greater accuracy than nRCBF. Visualizing these differentiated regions on imaging could help clinicians better assess treatment response following CRT, with potential to impact treatment management decisions.