Predictive Accuracy of MRI Prognosticators in the Early Detection of Malignant MCA Infarcts

Abstracts & Commentary

Sources: Thomalla GJ, et al. Prediction of malignant middle cerebral artery infarction by early perfusion- and diffusion-weighted magnetic resonance imaging. Stroke. 2003;34:1892-1899; Wijman CA [editorial]. Can we predict massive space-occupying edema in large hemispheric infarctions? Stroke. 2003;34:1899-1900.

Patients with massive infarcts in the middle cerebral artery (MCA) territory may develop brain edema leading to midline shift, raised intracranial pressure, and downward herniation. Their clinical course is characterized first by a deterioration in the level of consciousness and then an orderly rostrocaudal brainstem failure as herniation progresses, usually 2-5 days after the ictus. This subgroup of MCA infarcts has been labeled malignant MCA infarction (MMI).1 At present, there is no universally accepted treatment modality for patients who deteriorate as a result of MMI. Hemicraniectomy with dural patch enlargement has been proposed as life saving, and the measure is being studied as a therapeutic option. In 2 nonrandomized controlled studies from the same group of investigators,2,3 mortality was reduced and functional outcome in the survivors in the hemicraniectomy group was improved compared with historic controls. Nevertheless, at present, without a randomized control study, it remains uncertain as to whether the quality of life in survivors is acceptable enough to advocate surgical treatment for patients with acute strokes, particularly those who are elderly and those with dominant hemisphere infarctions. Nevertheless, if hemicraniectomy proves to be useful in patients with life-threatening edema from stroke, it is obvious that early surgery will result in the best clinical outcomes. Therefore, early and accurate prediction of MMI will be crucial in the management of such patients.

In the present article, Thomalla and colleagues report on the predictive value of early perfusion-weighted magnetic resonance imaging (PWI) and diffusion-weighted magnetic resonance imaging (DWI) within 6 hours of stroke onset in 37 patients with MCA stroke and a proximal vessel occlusion, either the carotid-T or MCA stem. Patients developed MMI defined by a decline in consciousness demonstrated as a loss of at least 1 point on the level of consciousness item on the NIH Stroke Scale and radiologic signs of space-occupying brain edema following a large infarct with compression of the ventricles or a midline shift. Thomalla et al found that an apparent diffusion coefficient < 80% (ADC < 80%) > 82 mL was the most accurate MRI prognosticator predicting MMI with 87% sensitivity and 91% specificity. Admission NIH Stroke Scale score had a higher sensitivity than any single MRI prognosticator but had only moderate (72%) specificity. Three patients, however, were misclassified as MMI with the use of the 82 mL (ADC < 80%) cut off. Thomalla et al conclude that MRI can help in the selection of patients for early craniectomy or other aggressive therapeutic approaches before the onset of clinical deterioration.


Previous studies have used clinical signs to predict the development of MMI. Coma on admission and early nausea and vomiting have been found to correlate with fatal brain edema. The initial NIH Stroke Scale score at admission was higher in patients who died or were dependent at 1 month after stroke. In the present study, Thomalla et al found that an NIH Stroke Scale score > 19 was highly sensitive to the prediction of MMI; however, the specificity of this clinical score was low. In other studies, the site of vessel occlusion has been reported to predict fatal brain swelling with high specificity of greater than 80% but with low sensitivity of about 50%. Therefore, neither clinical assessment nor site of vessel occlusion allows for the reliable prediction of MMI.

Thomalla et al focused on the 6-hour time window because at this early stage of stroke development, therapeutic decisions must be made concerning hemicraniectomy. They found that MRI-derived parameters resulted in good prediction of MMI, but, nevertheless, 3 patients who had large infarctions with edema and midline shift who remained clinically stable were misclassified as MMI. Therefore, if this study had been a therapeutic trial of hemicraniectomies, these patients would have been subjected to an unnecessary surgical treatment.

In addition, as Thomalla et al admit, there were limitations to their study. It used a retrospective design, was not a community-based study, and enrolled only a small sample of patients. Therefore, validation of these results will depend on a multinational prospective study of MRI prognosticators in MMI patients. — John J. Caronna, MD, Vice-Chairman, Department of Neurology, Cornell University Medical Center; Professor of Clinical Neurology, New York Hospital; Associate Editor, Neurology Alert.


1. Hache W, et al. Arch Neurol. 1996;53:309-315.

2. Rieke K, et al. Crit Care Med. 1995;23:1576-1587.

3. Schwab S, et al. Stroke. 1998;29:1888-1893.