Characterization of Skull Base Lesions Using Pseudo-Continuous Arterial Spin Labeling.

PURPOSE

Pseudo-continuous arterial spin labeling (pCASL) is a non-invasive magnetic resonance (MR) perfusion technique. Our study aimed at estimating the diagnostic performance of the pCASL sequence in assessing the perfusion of skull base lesions both qualitatively and quantitatively and at providing cut-off values for differentiation of specific skull base lesions.

METHODS

In this study 99 patients with histopathologically confirmed skull base lesions were retrospectively enrolled. Based on a pathological analysis, the lesions were classified as hypervascular and non-hypervascular. Patients were divided into two subgroups according to the anatomical origin of each lesion. The MRI study included pCASL and 3D T1-weighted fat-saturated post-contrast sequences. Of the patients seven were excluded due to technical difficulties or patient movement. The lesions were classified by two raters, blinded to the diagnosis as either hyperperfused or non-hyperperfused, based on the pCASL sequence. The normalized tumor blood flow (nTBF) of each lesion was determined. Qualitative and quantitative characteristics of hypervascular and non-hypervascular lesions were compared.

RESULTS

Visual assessment enabled correct classification of 98% of the lesions to be performed. Quantitatively, we found significant differences between the nTBF values for hypervascular and non-hypervascular lesions (p < 0.001) and provided cut-off values, allowing meningioma and schwannoma to be distinguished from meningioma and adenoma. Significant differences were also found within the hypervascular group, namely, paraganglioma was more hyperperfused than meningioma (p = 0.003) or metastases (p = 0.009).

CONCLUSION

The present study demonstrates the high diagnostic performance of pCASL in characterizing skull base lesions by either visual assessment or nTBF quantification. Adding the pCASL sequence to the conventional protocol of skull base assessment can be recommended.