Background/Objectives: Contrast enhancement (CE) on T1-weighted MRI is routinely used to guide therapy in the management of glioblastoma, although adjacent non-contrast-enhancing (non-CE) T2/FLAIR abnormalities can also harbor viable tumor tissue. The differences between these radiographic compartments remain incompletely characterized beyond conventional structural imaging. We therefore compared CE and non-CE compartments in untreated IDH-wildtype glioblastoma using dynamic susceptibility contrast (DSC) and diffusion-weighted MRI derived indices. Methods: Adults with untreated glioblastoma imaged preoperatively between January 2021 and September 2024 on multi-vendor 1.5 T and 3 T scanners were retrospectively included. Regions of interest were placed in CE tumor, adjacent non-CE T2/FLAIR hyperintense tissue, and contralateral normal-appearing white matter (NAWM). Mean apparent diffusion coefficient (rADC), cerebral blood volume (rCBV), capillary transit time heterogeneity (rCTH), oxygen extraction fraction (rOEF), and a cerebral metabolic rate of oxygen index (rCMRO2) were extracted and harmonized for scanner effects and normalized to NAWM. Paired CE–non-CE differences were tested using Wilcoxon signed-rank tests and summarized by Hodges–Lehmann differences with bootstrap 95% confidence intervals. Spearman correlations were used to assess coupling within contrast-enhancing tumor regions. Results: Seventy-two participants were analyzed (median age 67 years; 34 women); 66 had paired CE and non-CE data. rCMRO2 and rCBV were higher in CE than non-CE (both p < 0.001), while rADC was lower (p = 0.003). rOEF (p = 0.12) and rCTH (p = 0.52) did not differ significantly between compartments. Conclusions: CE in untreated IDH-wildtype glioblastoma predominantly reflects higher perfusion capacity (rCBV) along with a higher model-derived rCMRO2 index, while capillary-function indices (rCTH and rOEF) are not consistently compartment-restricted. These findings may refine the physiological interpretation of CE in glioblastoma and support further validation of DSC-derived indices.
Reis et al. (Wed,) studied this question.