Study favors MRI over CT for acute stroke diagnosis

An influential trial has left neuroradiologists in a quandary about the best imaging approach for the initial evaluation of acute stroke. The single-center prospective trial concluded that MRI accurately diagnoses acute stroke while noncontrast CT, the old gold standard, is about as accurate as a coin flip.

The National Institute of Neurological Disease and Stroke sponsored the study, and Dr. Steven Warach, section chief of stroke diagnostics and therapeutics at NINDS, was the supervising author. It was published in The Lancet in February (Lancet 2007;369:293-298).

The findings run counter to NINDS recommendations that have guided clinical practice for a decade and revolve around tPA, the only FDA-approved thrombolytic treatment for ischemic stroke. The clot-busting agent is credited for the complete and rapid recovery of many patients treated soon after onset of stroke symptoms. Noncontrast CT has been considered essential for differentiating between ischemic and hemorrhagic stroke.

But tPA can trigger intracranial hemorrhages as well, and the probability of hemorrhage from tPA treatment increases as time passes. Based on NINDS research in the 1990s, the FDA decided the risks of tPA significantly outweighed its potential benefits three hours after onset. It ruled that the window must close after the third hour except in experimental settings.

The CT-based protocol works acceptably well for many community-based hospitals because CT, the modality of choice for diagnosing trauma, is almost always located near the emergency room. It is staffed most hours of the days, and on-call technologists are trained to respond quickly when it isn't staffed.

What may be convenient for most hospitals, however, is not at all appropriate for optimal care, according to Warach.

"If a hospital is routinely using noncontrast CT, they are missing strokes," he told Diagnostic Imaging in an interview.

Using CT to consider patients for tPA remains an important approach, but one that applies only to the single-digit percentage of stroke patients who arrive in time to potentially receive the treatment. Most stroke patients are not candidates and should be evaluated with MRI, he said.

The realities of stroke and its diagnosis favor MRI's use, Warach said. Stroke is a complex condition that often does not express classic signs and symptoms. Most strokes covered in the trial were mild, and about one in four involved signs and symptoms that are not ordinarily associated with stroke. The physicians rendering diagnosis were not neurological specialists. Most of the attending physicians were general emergency physicians or radiologists.

Functional MRI potentially expands the modality's capabilities. Researchers are experimenting with MR perfusion-diffusion mismatch to extend the therapeutic window for thrombolysis. The value of mismatch for triaging patients was demonstrated in 2005 during the Desmoteplase in Acute Ischemic Stroke (DIAS-2) phase II trial for desmoteplase, a promising thrombolytic. The findings set the stage for the much-anticipated DIAS-2 phase III trial. Those results were presented after press time at the European Stroke Conference in Glasgow in June.

Functional CT research is also moving ahead. Dr. Max Wintermark of the University of California, San Francisco and Dr. Michael Lev of Massachusetts General Hospital are refining methods that identify penumbra with perfusion CT.

Wintermark's April 2007 study of 113 stroke patients demonstrated the versatility of CT angiography for evaluating strokes (Ann Neurol 2007; Apr 12 [Epub ahead of print]). Findings were published in an online version of the Annals of Neurology.

"Collateral flow, along with the clot, is the key in understanding the pathophysiology of ischemic strokes." he said.