Abstract

Introduction: MR dynamic susceptibility contrast-enhanced perfusion (DSC) has demonstrated utility in the evaluation of follow-up malignant CNS neoplasms. Preliminary studies suggest Arterial Spin Labeling (ASL) perfusion may be equivalent for this purpose. Using a retroprospective, blinded test design, we compared interpretations of ASL and DSC in tumor follow-up. Methods: Follow-up studies from fifty patients with primary CNS tumors (76 total studies) were evaluated: 75 studies with ASL and 28 with DSC. Of these, 27 were synchronous studies. Three trained neuroradiologists evaluated and graded the studies using a 3-point scale: 1 = increased perfusion (persistent, progressive or recurrent tumor), 2 = equivocal perfusion (inconclusive), 3 = low perfusion (treatment necrosis or persistent low grade). Observer and technique concordances were calculated, along with observer accuracy. Radiologic grading was compared to the final clinical or pathologic diagnosis. Results: In the synchronous cohort, inter-observer agreement was 78% for ASL and 70% for DSC, with Fleiss' free marginal kappa 0.67 for ASL (substantial agreement) and 0.56 (moderate agreement) for DSC. Individual observer concordances between ASL and DSC, using Cohen's kappa, ranged from 0.58 (moderate) to 0.86 (near perfect). There was one complete discordant score set (1 vs. 3 ratings) between readers for the DSC study limb and none for the ASL limb. Accuracy for 21 unequivocal synchronous cases was 92% for ASL and 88% for DSC. The extended cohort accuracy for ASL was 86% (60 unequivocal cases). Conclusions: ASL perfusion is a promising alternative to DSC in primary malignant CNS neoplasm follow-up.

Authors and Affiliations