Differentiation of primary central nervous system lymphoma from high-grade glioma and brain metastases using susceptibility-weighted imaging

Abstract

Background and purpose: Conventional MRI is often difficult to distinguish between primary central nervous system lymphomas (PCNSLs), high-grade gliomas and brain metastases due to the similarity of their appearance. The aim of this study was to investigate whether the susceptibility-weighted imaging (SWI) has higher sensitivity than conventional MRI in detecting hemorrhage between PCNSLs, high-grade gliomas and brain metastases, and can be used to differentiate the diagnosis between these tumors.

Methods: The number of lesions with hemorrhage was quantified by both the conventional MR imaging and SWI. The number of micro-hemorrhage and vessels within lesions were counted on SWI.

Results: The detection rate of hemorrhage on SWI was significantly higher than that on the conventional MR imaging. The intralesional hemorrhagic burden and the number of the vessels within lesions detected by SWI were significantly higher in high-grade gliomas and brain metastases than those in PCNSLs. There was no significant difference in these two parameters between high-grade gliomas and brain metastases. The best predictor to differentiate PCNSLs from high-grade gliomas and brain metastases was intralesional vessel number that yielded the best ROC characteristics and highest classification accuracy.

Conclusions: SWI is useful in differentiating of PCNSLs from high-grade gliomas and brain metastases.

Keywords: Brain metastases; differential diagnosis of intracranial malignancies; high-grade glioma; primary CNS lymphoma; susceptibility-weighted imaging.

Figure 1

A 62-year-old male with diffuse large B cell lymphoma. One lesion was located in splenium on T2WI (A) and T1WI (B). There were no vessels and microhemorrhage shown on T1WI and T2WI. However, SWI-MinIP (C) showed microhemorrhage with sharp border in the lesion and high signal on phase image. (D) The lesion showed hyperintense signal both on DWI (E) and FLAIR (F), and demonstrated solid, intense, and homogeneous enhancement on contrast-enhanced T1-weighted image (G).

Figure 2

A 45-year-old female with glioblastoma (WHO grade IV). Three lesions were located in bilateral frontal lobe and genu corpus callosum on T2WI (A) and T1WI (B). In the lesion located in the right frontal lobe, the 2 dot-like low signals were found on T2WI. However, on SWI-MinIP (C), more hemorrhagic signals were found in lesions, which showed high signals on phase image (D), and a vessel was found in the lesion located in genu corpus callosum. The lesions showed hyperintense signal both on DWI (E) and FLAIR (F), and demonstrated ring, intense and inhomogeneous enhancement on contrast-enhanced T1-weighted image (G).

Figure 3

A 60-year-old female with metastatic neuroendocrine carcinoma (the primary sites of cancer was lung). One lesion was located in left cerebellar hemisphere on T2WI (A) and T1WI (B). There were no vessels and microhemorrhage shown on T1WI and T2WI. However, on SWI-MinIP (C), multiple hypointense signals regarded as hemorrhage and vessels were found in the lesion, which showed hyperintense signal on phase image (D). The lesion showed hyperintense signals both on DWI (E) and FLAIR (F), and demonstrated solid, intense and inhomogeneous enhancement on contrast-enhanced T1-weighted image (G).

Figure 4

(A) Receiver operating characteristic for comparison of the number of hemorrhage and intralesional vessels for differentiation of PCNSLs from non-PCNSLs. The AUC for the number of hemorrhage and intralesional vessels is 0.873 (95% CI, 0.802–0.945) and 0.914 (95% CI, 0.860–0.967) respectively. (B) ROC for comparison of the number of hemorrhage and intralesional vessels for differentiation of high-grade gliomas and brain metastases. The AUC for the number of hemorrhage and intralesional vessels is 0.495 (95% CI, 0.365–0.625) and 0.393 (95% CI, 0.265–0.521) respectively.