Primary central nervous system lymphoma and atypical glioblastoma: differentiation using the initial area under the curve derived from dynamic contrast-enhanced MR and the apparent diffusion coefficient

Abstract

OBJECTIVES: To evaluate the ability of the initial area under the curve (IAUC) derived from dynamic contrast-enhanced MR imaging (DCE-MRI) and apparent diffusion coefficient (ADC) in differentiating between primary central nervous system lymphoma (PCNSL) and atypical glioblastoma (GBM).

METHODS: We retrospectively identified 19 patients with atypical GBM (less than 13 % necrosis of the enhancing tumour), and 23 patients with PCNSL. The histogram parameters of IAUC at 30, 60, 90 s (IAUC30, IAUC60, and IAUC90), and ADC were compared between PCNSL and GBM. The diagnostic performances and added values of the IAUC and ADC for differentiating between PCNSL and GBM were evaluated. Interobserver agreement was assessed via intraclass correlation coefficient (ICC).

RESULTS: The IAUC and ADC parameters were higher in GBM than in PCNSL. The 90th percentile (p90) of IAUC30 and 10th percentile (p10) of ADC showed the best diagnostic performance. Adding p90 of IAUC30 to p10 of ADC improved the differentiation between PCNSL and GBM (area under the ROC curve [AUC] = 0.886), compared to IAUC30 or ADC alone (AUC = 0.789 and 0.744; P < 0.05 for all). The ICC was 0.96 for p90 of IAUC30.

CONCLUSIONS: The IAUC may be a useful parameter together with ADC for differentiating between PCNSL and atypical GBM.

KEY POINTS: • High reproducibility is essential for practical implementation of advanced MRI parameters. • IAUC and ADC are highly reproducible parameters. • IAUC values were higher in atypical GBM than in PCNSL. • Adding IAUC to ADC improved the differentiation between PCNSL and GBM. • IAUC with ADC are useful for differentiating PCNSL from GBM.