Stroke Imaging: Whats Best vs. Whats Practical
 CT images in a patient with left hemiparesis. Admission CT perfusion images show A) cerebral blood flow (CBF), B) cerebral blood volume (CBV), and C) mean transit time (MTT). The presence of a matched CBF/CBV perfusion deficit suggests irreversibly ischemic infarct “core,” likely to correlate with MRI diffusion-weighted imaging findings, and not a target for reperfusion therapies. Image courtesy of Michael H. Lev, MD, and Angelos Konstas, MD, Massachusetts General Hospital, Boston.
The solid evidence points to MRI as the first-line test for acute stroke assessment. Limited scanner availability, however, has prompted many to rely on CT, despite the lack of rigorous clinical studies to support its use.

MRI is better, but…

Recently, the American Academy of Neurology (AAN) reviewed the evidence for MR diffusion-weighted imaging (DWI) in the assessment of acute ischemic stroke and concluded that it should be the first-line test (Neurology 2010;75;177-185). While many neuroradiologists agree with this conclusion, many emergency departments (EDs) do not have easy access to an MRI scanner. Consequently, the first-line test for suspected stroke is a non-contrast brain CT to rule out hemorrhage.  

“No radiology or neurology specialists would question whether MRI is more sensitive to the presence of acute stroke, although CT imaging properly performed and read also serves as a very good test,” says Lawrence N. Tanenbaum, MD, director of MRI, CT and outpatient/advanced development at Mount Sinai School of Medicine in New York City. However, there are practical considerations with MRI as a first-line test, including scanner availability, substantial delays between ordering and obtaining an MR exam and imaging times up to 40 minutes—“a long time in the life of a stroke.”

“It is possible to design an efficient MR exam that takes less than 10 minutes and includes MR angiography, diffusion-weighted imaging and perfusion-weighted imaging,” Tanenbaum says. “Performing the exam quickly is more practical than it was 10 years ago, but MR scanners are still less accessible than CT scanners—and that is the crux of the debate.”

Steven Warach, MD, PhD, senior author of the AAN guidelines, would like to see stronger lobbying from ED physicians and neurologists to make MRI scanners more readily available for the assessment of suspected stroke.

“Emergency physicians, as the main gatekeepers, are making decisions about which test has to be ordered. Hopefully, these guidelines will encourage them to take a more active role in making sure these patients get the best test,” says Warach, chief of stroke diagnostics and therapeutics at the National Institutes of Neurological Disorders and Stroke, part of the National Institutes of Health, in Bethesda, Md. He notes that the guidelines do not advocate for what hospitals should do regarding the availability of MRI scanners.

In a review of the literature up to 2008, Warach and colleagues concluded that DWI is the most sensitive and specific technique to determine at admission how much brain is likely to be irreversibly infarcted (Level A). The next conclusion is that the amount of brain infarcted at admission does have a reasonably good correlation with not only current clinical status but with a patient’s outcome (Level B). This evidence holds true for anterior territory strokes, while evidence for DWI in vertebral basilar territory stroke is not as strong (Level C). They found insufficient evidence to offer a conclusion regarding MR perfusion’s role in assessing acute ischemic stroke.

Bleed or no bleed

In the seminal study showing the utility of IV tissue plasminogen activator (tPA) to thrombolize ischemic stroke within three hours of onset, one of the eligibility criteria was “a baseline CT scan of the brain that showed no evidence of intracranial hemorrhage” (N Engl J Med 1995; 333:1581-1588). When the FDA subsequently approved tPA, the indications included a non-contrast head CT to rule out hemorrhage.

“That is still the standard today,” says Michael H. Lev, MD, director of emergency neuroradiology and the neurovascular lab at Massachusetts General Hospital (MGH) in Boston. “That is evidence-based medicine for IV-tPA administration that no one will dispute. What the AAN guidelines are saying is that if you want to diagnose the presence of acute ischemic stroke, or if you are trying to determine the extent of the stroke at admission, there is no better way to do that in a typical ED setting than with diffusion-weighted imaging.”

The problem, according to Warach, is that even when the three-hour window for tPA is long gone, ED docs are still ordering non-contrast CT exams for suspected stroke. “For those patients who are not tPA-eligible and who have no contraindication to MRI, it’s hard to justify—based on the data—why we would give just a non-contrast CT.”

Interestingly, the European Acute Stroke Study III (ECASS III) in 2008 concluded that IV tPA can be safely administered up to 4.5 hours after stroke onset (N Engl J Med 2008; 359:1317-1329). Following these results, the American Heart Association and American Stroke Association recommended extending the three-hour window for tPA treatment; however, neither the FDA nor the European Medicines Agency have amended their indications for IV tPA beyond three hours. A recent study found that tPA administration between three and 4.5 hours tripled after the release of the ECASS III results, with no concomitant increase in time-to-treatment (Lancet Neurol 2010;9[9]:866-874).

“If hemorrhage is the only question, and it is rarely the only question, non-contrast CT or MRI can equally make the assessment,” Warach says. He notes that the prescribing information offered by the FDA for tPA and the practice guidelines from the AHA endorse either MRI or CT for stroke triage. “However, a non-contrast CT is the minimum standard requirement,” he says, adding that the use of MRI provides more valuable information to better guide therapeutic decisions.

“There is a lot of inertia to changing hospital routine,” he says. “Usually best practices lag behind the clinical data. Our hope is that these guidelines will help inform that discussion.”

The situation may not be much different in Europe. A study by Balucani et al of 25 European countries found that less than one-third of ischemic stroke patients were admitted to hospitals that offer all three imaging techniques to detect intracranial atherosclerosis: CT angiography (CTA), MR angiography (MRA), and transcranial color Doppler (Stroke 2009;40:726-729). Only half of the 886 hospitals offered at least one imaging technique, and less than 20 percent offered none. CTA was the most frequently available technique, followed by MRA.

While Warach and colleagues were not charged with evaluating MRA, and the evidence was too scant to make a definitive conclusion about MR perfusion, Warach offers this opinion: “The evidence is beginning to accumulate showing multimodality MRI is more cost effective and clinically superior than non-contrast CT to triage suspected stroke patients.”

Will CTA, CT perfusion have a role?

Warach acknowledges that many centers are now performing CTA and CT perfusion scans along with the initial non-contrast brain CT. His team, however, was not charged with evaluating CTA or CT perfusion. Nevertheless, it’s hard to disregard the promising data on CT perfusion, albeit most of it from small studies.

“To ignore CT perfusion in 2010, some 15 years after it has become available, does not give the most thorough evaluation of today’s technology,” says Tanenbaum. “CT perfusion is as accurate, perhaps even more accurate, and more quantitative than MR perfusion.” He adds that a disadvantage of CT perfusion—its inability to cover the entire brain—has been rectified with newer software.

However, the lack of standardized vendor software and acquisition protocols is an issue with CT perfusion, according to Lev. “At MGH, we have validated hundreds of cases where we’ve correlated CT thresholds with MR diffusion scanning,” he says. Yet, those thresholds may be different for sites using other vendors’ software or performing a different type of acquisition for a different amount of time.

“Even given that validation, you have to realize that perfusion is measuring something different than diffusion. Consequently, at its best, CT perfusion will never be as accurate as MR diffusion for predicting acutely, irreversibly infarcted brain,” Lev says. “But, if MR diffusion is not available, we can get very good estimates with CT of what the diffusion scan would look like.”