Differentiation between Glioblastoma Multiforme and Primary Cerebral Lymphoma: Additional Benefits of Quantitative Diffusion-Weighted MR Imaging

Abstract

The differentiation between glioblastoma multiforme (GBM) and primary cerebral lymphoma (PCL) is important because the treatments are substantially different. The purpose of this article is to describe the MR imaging characteristics of GBM and PCL with emphasis on the quantitative ADC analysis in the tumor necrosis, the most strongly-enhanced tumor area, and the peritumoral edema. This retrospective cohort study collected 104 GBM (WHO grade IV) patients and 22 immune-competent PCL (diffuse large B cell lymphoma) patients. All these patients had pretreatment brain MR DWI and ADC imaging. Analysis of conventional MR imaging and quantitative ADC measurement including the tumor necrosis (ADCn), the most strongly-enhanced tumor area (ADCt), and the peritumoral edema (ADCe) were done. ROC analysis with optimal cut-off values and area-under-the ROC curve (AUC) was performed. For conventional MR imaging, there are statistical differences in tumor size, tumor location, tumor margin, and the presence of tumor necrosis between GBM and PCL. Quantitative ADC analysis shows that GBM tended to have significantly (P<0.05) higher ADC in the most strongly-enhanced area (ADCt) and lower ADC in the peritumoral edema (ADCe) as compared with PCL. Excellent AUC (0.94) with optimal sensitivity of 90% and specificity of 86% for differentiating between GBM and PCL was obtained by combination of ADC in the tumor necrosis (ADCn), the most strongly-enhanced tumor area (ADCt), and the peritumoral edema (ADCe). Besides, there are positive ADC gradients in the peritumoral edema in a subset of GBMs but not in the PCLs. Quantitative ADC analysis in these three areas can thus be implemented to improve diagnostic accuracy for these two brain tumor types. The histological correlation of the ADC difference deserves further investigation.

MRI as a diagnostic biomarker for differentiating primary central nervous system lymphoma from glioblastoma: A systematic review and meta‐analysis

“…Detailed information of excluded studies is described in the Supplementary Materials. One further relevant study was also identified, resulting in a final total of 22 studies investigating the diagnostic performance of MRI for differentiating PCNSL from glioblastoma, with these studies covering 1182 patients.…”

“…Seven studies enrolled patients in a consecutive manner, while the enrollment was not explicit in the others. Nineteen studies used only histopathology as the reference standard, while two studies used histopathology and clinicoradiologic diagnosis for patients with PCNSL …”

“…A variety of MRI sequences were used in various combinations (Table ), with the most commonly used sequences being DWI in 14 studies, DSC in nine studies, and DCE in four studies . The other less‐commonly used sequences were ASL in four studies, SWI in four studies, MRS in three studies, and IVIM in two studies .…”

“…The other less‐commonly used sequences were ASL in four studies, SWI in four studies, MRS in three studies, and IVIM in two studies . In terms of the ROI placement, the entire enhancing tumor was selected in 14 studies, and both the entire enhancing tumor and peritumoral area were selected in seven studies . A variety of different cutoff values were used, with multiparametric MRI sequences being used to determine optimal cutoff values in 11 studies .…”

“…During the last decade, multiple MRI sequences have been introduced for differentiating PCNSL from glioblastoma . These sequences, which include diffusion‐weighted imaging (DWI), dynamic susceptibility‐weighted contrast‐enhanced imaging (DSC), dynamic contrast‐enhanced imaging (DCE), arterial spin labeling (ASL), susceptibility‐weighted imaging (SWI), intravoxel incoherent motion (IVIM), and magnetic resonance spectroscopy (MRS), have demonstrated variable diagnostic performance.…”