Differentiation of Molecular Glioblastomas and Lower-grade Diffuse Astrocytomas Using MRI Perfusion and Diffusion Parameters.

Background And Purpose: Molecular glioblastomas are challenging to distinguish from lower-grade diffuse astrocytomas (grades 2-3) without T2/T2 FLAIR mismatch, when assessed on T1-gadolinium and T2/T2 FLAIR MRI. This study aimed to evaluate the performance of the ADC from diffusion-weighted imaging and the relative CBV from DSC perfusion imaging in differentiating molecular glioblastomas from lower-grade diffuse astrocytomas.

Materials And Methods: Fourteen patients with molecular glioblastomas (defined as isocitrate dehydrogenase wildtype (IDH-wt), exhibiting one or more of the following: telomerase reverse transcriptase (TERT) promoter mutation, epidermal growth factor receptor (EGFR) gene amplification, or +7/-10 chromosomal alterations, but without microvascular proliferation or necrosis) and thirteen patients with lower-grade diffuse astrocytomas (IDH-mutated or not elsewhere classified, grades 2-3) were included. ADC values and DSC-rCBV values were measured, and the two groups were compared using T‑test and Wilcoxon rank-sum test. Combinations of variables and tumor characteristics were analyzed using binary logistic regression, receiver operating curve (ROC) analysis, and Firth regression model.

Results: Molecular glioblastomas exhibited lower minimum ADC, mean ADC (p < 0.01), maximum ADC (p < 0.05) and higher standard deviation of ADC (p < 0.05), compared to lower-grade astrocytomas, measured in a 7 mm ROI in the lowest ADC region. Molecular glioblastoma also had higher normalized median, average, and minimum rCBV ratios (p < 0.01) in a 10 mm ROI in the highest perfused region. A combined receiver operating curve (ROC) model of ADC and rCBV achieved an area under the curve (AUC) of 0.93 (95% CI: 0.82-1.00).

Conclusions: This study demonstrates that ADC and rCBV measurements can help differentiate molecular glioblastomas from lower-grade diffuse astrocytomas lacking T2/T2 FLAIR mismatch. These findings may aid in preoperative tumor characterization, surgical planning, and prognosis.