Differentiating Giant Cell Glioblastoma from Classic Glioblastoma With Diffusion-Weighted Imaging.

Abstract

Differential diagnosis of giant cell glioblastoma (GC) and classic glioblastoma (GBM) using conventional radiological modalities is difficult. This study aimed to use diffusion-weighted imaging (DWI) to distinguish GC from GBM and thereby improve the accuracy of preoperative assessment of patients with GB. The clinical, magnetic resonance imaging, and pathologic data of 12 patients with GC and 21 patients with GBM were retrospectively analyzed. Independent sample t tests were used to compare the minimum apparent diffusion coefficient (ADC_min) and the normalized apparent diffusion coefficients (nADC) of the 2 tumor types. Receiver operating curve (ROC) analysis was used to assess the diagnostic efficacy of ADC_min and nADC values. Compared with that of the classic GBM group, the ADC_min (0.98 ± 0.14 vs. 0.80 ± 0.19×10^-3 mm²/second, P = 0.007) and nADC (1.42 ± 0.25 vs. 1.17 ± 0.25, P = 0.011) of the GC group were significantly higher. ROC curve analysis showed that the maximum area under the curve of ADC_min and nADC were 0.800 ± 0.080 and 0.778 ± 0.082, respectively. The sensitivity, specificity, and accuracy distinguishing GC and classic GBM was best (83.33%, 76.19%, and 78.79%, respectively) when ADC_min = 0.84×10^-3 mm²/second (maximum area under the ROC, 0.800). Its positive and negative predictive values under this condition were 88.89% and 66.67%, respectively. By distinguishing GC from classic GBM, the ADC_min parameter of DWI can improve the accuracy of the preoperative differential diagnosis of the 2 tumor types.