Abstract 2768: Comparison of Arterial Spin Labeling with Perfusion-weighted Imaging for Defining Perfusion-Diffusion Mismatch in Stroke Patients
Background: Reperfusion in stroke patients with a perfusion-diffusion (PWI-DWI) mismatch leads to better clinical outcomes (1). Because some patients cannot receive gadolinium contrast agents, we examined whether the ASL-DWI mismatch was equivalent to the PWI-DWI mismatch in patients with acute ischemia.
Methods: We retrospectively analyzed all patients scanned at a single institution during 2010 who met the following criteria: technically adequate DWI, ASL, and PWI images; >10 ml lesion on ADC maps; onset of symptoms within 5 days; scanned at 1.5T. Automated software (RAPID) (2) was used to determine the volume of the DWI and PWI lesions and to classify into one of three categories using revised DEFUSE criteria (3): mismatch (PWI>1.8 DWI lesion size and >10 ml absolute difference), matched (1.8 DWI>PWI>0.7 DWI lesion size or <10 ml), and reperfused (PWI<0.7 DWI lesion size and >10 ml). This was considered the gold standard read. ASL-DWI status was determined using visual observation by two radiologists blinded to clinical data and each other’s reading. After this, a consensus was determined. Agreement between the two radiologists for ASL-DWI mismatch status and for consensus ASL-DWI versus PWI-DWI status was measured using kappa statistics.
Results: 44 studies in 43 patients met the above criteria, and had the following demographic and imaging features: mean age 59±16 yrs; 22 men, mean DWI lesion size 66±56 ml (range: 10-243 ml), mean PWI Tmax>6 sec lesion size 64±64 ml (range 0-219 ml), time from last seen normal 37±24 hrs (range 2-111 hrs). Agreement between the two readers for rating the ASL-DWI mismatch was excellent (unweighted kappa 0.92 [95% CI 0.80-1.0]). Agreement between categorization of the two modalities (consensus ASL-DWI and gold-standard PWI-DWI) was moderate (unweighted kappa 0.35 [95% CI: 0.13-0.57]). Exact agreement was present in 57% of cases (25/44). Most of the time, when there was disagreement, ASL tended to overestimate the size of the perfusion lesion compared with PWI Tmax. If the groups are dichotomized between mismatch and either matched or reperfusion, the sensitivity and specificity of ASL-DWI mismatch classification is 63% and 81%, respectively. If ASL-DWI does not show a mismatch, it is almost always correct (29/32 cases, NPV 91%). If ASL-DWI does show a mismatch, a PWI-DWI mismatch is present about half the time (5/12, PPV 42%).
Conclusion: For the purposes of treatment, ASL-DWI mismatch classification using qualitative visual observation generally agrees with PWI-DWI mismatch status using Tmax>6 sec. Disagreements were typically related to ASL overestimation of the PWI lesion size. Although promising, improved methodology may be required before ASL can be used routinely in acute stroke trials.
1. Albers, et al.,Ann Neurol2006;60:508-17.
2. Straka, et al.,JMRI2010;32:1024-37.
3. Kakuda, et al.,JCBFM2008;28:887-91.
Author Disclosures: G. Zaharchuk: Other Research Support; Modest; GE Healthcare. Speakers' Bureau; Modest; Neuroradiology Advisory Board, GE Healthcare. I.S. El-Mogy: None. G.W. Albers: None.
- © 2012 by American Heart Association, Inc.