1. Home>
2. Specialties>
3. Neurology>
4. Summary and Comment

Improved Prediction of Acute Stroke Outcome with Multimodal CT

A new scoring system to evaluate multimodal CT images may improve triage of patients with acute stroke.

Unenhanced CT scanning and clinical exam alone have a restricted ability to predict functional outcome in patients with acute stroke and, therefore, have a limited value for clinicians in selecting patients for urgent thrombolytic therapy. Advanced techniques such as MR diffusion/perfusion imaging are increasingly advocated for patient triage. These authors prospectively evaluated the predictive value of a contrast-enhanced, CT-based scoring system called Multimodal Stroke Assessment Using CT (MOSAIC). The 4 components of the MOSAIC score were extent of hypodensity on unenhanced CT, degree of intracranial stenosis on CT angiography (CTA), and lesion size on 2 adjacent dynamic CT perfusion (CTP) blood-flow-weighted slices. Scores for 44 patients based on admission CT (mean time post-ictus, 3 hours) were generated by assigning a score of 0 (normal), 1 (moderately abnormal), or 2 (maximally abnormal) to each scan-component finding.

The composite MOSAIC score correlated more strongly with both final outcome (measured by modified Rankin Scale and Barthel Index 3 months after the stroke) and final infarct size (10 days post-ictus) than did any of the individual component scores. Moreover, the correlation between MOSAIC and early NIH stroke scale (NIHSS) scores was weaker than that between MOSAIC scores and clinical and tissue outcomes. A MOSAIC score <4 predicted functional independence with 89% to 96% likelihood.

Comment: Historically, CT has been used on patients with acute stroke primarily to rule out cerebral hemorrhage. However, because CT is faster, less expensive, and more available in emergency departments than MRI, a validated, relatively simple, contrast-enhanced CT scoring system may have important implications for patient triage. An advantage of the MOSAIC score is that, like widely accepted CT grading scales for subarachnoid hemorrhage, it is objective and easy to apply.

One design limitation of this pilot study is that 16% of patients received IV thrombolysis, creating a heterogeneous dataset with regard to expected outcomes. Perhaps more important, no attempt was made to optimize the MOSAIC scoring system on the basis of more detailed analysis of the available imaging data; the predictive accuracy might be even better with a slightly more complex scoring system that rated such factors as maximal degree of blood-flow reduction or lesion size on blood-volume-weighted CTA images. Nevertheless, the authors have demonstrated convincingly the superiority of their method over its individual components. We can anticipate improved predictive power, for both treated and untreated patients, with future refinements of CT-based acute-stroke models.

— Michael H. Lev, MD

Dr. Lev is Director of Emergency Neuroradiology and the Neurovascular Laboratory, Department of Radiology, Massachusetts General Hospital, and Assistant Professor of Radiology, Harvard Medical School, Boston.

Published in Journal Watch Neurology March 7, 2003

Citation(s):

Nabavi DG et al. MOSAIC: Multimodal stroke assessment using computed tomography: Novel diagnostic approach for the prediction of infarction size and clinical outcome. Stroke 2002 Dec; 33:2819-26.

• Original article (Subscription may be required)
• Medline abstract (Free)