MRI is more accurate than CT in identifying acute ischemic strokes, according to a blinded comparison of the two techniques in an emergency room setting by researchers at the NINDS. However, most emergency rooms are not equipped or staffed to provide immediate screening, experts told Neurology Today.
Non-contrast MRI was about five times more sensitive, and two times more specific, than non-contrast CT in detecting acute ischemic stroke (AIS), but the procedures were equally effective in diagnosing acute intracranial hemorrhage, according to the study, published in the Jan. 27 issue of The Lancet (2007;369:293–298).
“Our idea was to compare the two techniques in a real-world setting that would estimate typical patients seen in emergency rooms with suspicion of stroke — to take all comers,” said Steven Warach, MD, PhD, director of the NINDS Stroke Diagnostics and Therapeutic Section and senior author of the study.
“Twenty-five percent of patients seen in emergency rooms with possible stroke symptoms are actually diagnosed with something else,” he told Neurology Today in a telephone interview. “We wanted to survey a ‘real time’ cross section of consecutive patients and have blinded readers look at the evidence on CT and MRI, both prospectively and after diagnosis.”
The study included 356 consecutive patients treated at the NINDS Stroke Center at Suburban Hospital in Bethesda, MD. Stroke specialists conducted emergency clinical assessments of all patients, including evaluation using the NIH Stroke Scale to gauge severity. MRI was done before CT in 304 patients, and all scans were completed within two hours; the median difference was 34 minutes.
Each patient's images were sorted randomly and examined independently by two neuroradiologists and two stroke neurologists, Dr. Warach explained. The viewers were unanimous about the presence or absence of acute stroke in 80 percent of patients with non-contrast MRI, compared to 58 percent using non-contrast CT.
“Our results show that overall, MRI is twice as accurate in distinguishing stroke from non-stroke. Based on these results, MRI should become the preferred imaging technique for diagnosing patients with acute stroke,” he said.
“Although CT scanning has been the criterion that is standard for diagnosis of acute stroke, our study shows that use of CT is no longer justifiable on the basis of diagnostic accuracy alone.”
Of an estimated 700,000 strokes each year in the US, about 85 percent are ischemic. Neurons die rapidly within a few minutes after a vessel is blocked, so rapid assessment and early intervention are important. The mortality rate from strokes is between 20 percent and 50 percent.
Injury may be avoided or reduced in some patients by quickly re-opening a blocked vessel with tissue plasminogen activator (tPA), but the technique works only if administered within three hours of symptom onset. About 25 percent of stroke patients have no clinical signs of stroke when they reach the ER, and are therefore much less likely to receive tPA Of equal importance, the drug cannot be given to patients with hemorrhagic strokes.
“One issue is being able to identify and treat patients with tPA within three hours, but these patients are already a minority — the vast majority aren't eligible for tPA,” he said. “The larger issue is the best way to diagnose stroke types. If you talk to ER personnel, the reason they don't use tPA is because of uncertainty. If every ER used MRI, I predict that tPA use would go up. But most hospitals don't have MRIs set up for ER rooms. Non-contrast CT is now the first-line diagnostic procedure, even though most hospitals have MRI. One reason we did this study is to provide data for these decisions,” he said.
Most possible stroke victims are first evaluated by non-specialists who are often reluctant to treat a patient with tPA without more confidence in the accuracy of diagnosis, he said.
“Although MRI is remarkably accurate in detecting early stroke damage, it can't substitute for a doctor's clinical judgment in making a stroke diagnosis and deciding upon treatment,” said Dr. Warach.
“I want to emphasize that the implications of this study are across the board for all patients, including those who are not candidates for tPA. The stroke is mild in most patients admitted to emergency rooms. Larger strokes are not difficult to diagnose – so we're talking about milder strokes and transient ischemic attacks or TIAs which are difficult to diagnose and where CT is not informative.”
The two major causes of delay in treatment are failure by patients or family members to recognize stroke symptoms, and failure to contact the medical system efficiently, according to the American Heart Association (AHA).
In a June 2006 Scientific Statement, the organization said that the delay in seeking treatment after ischemic stroke was two or three times longer than after a heart attack, largely because pain is not likely, and cognitive impairment is more prevalent after a stroke. Community education and outreach efforts have failed to address the problem, the AHA panel noted.
According to the NINDS study, however, many stroke patients seen in emergency rooms are misdiagnosed.
“I think the study adds to other data we have that MRI might be better in some ways than CT, but I don't think this will change the way things are done,” said Larry B. Goldstein, MD, director of the Stroke Center at Duke University in Durham, NC.
“The problem is that MRI is not available in real time in most emergency situations. This study was prospective and analysis was after the fact, but even though these were consecutive patients, we still don't know if treatment decisions would have been any different if they were evaluated by CT versus MRI,” he told Neurology Today in a telephone interview.
Also, he noted that the study was conducted at only one medical center, with experienced neurologists and neuroradiologists. Whether the findings apply to medical facilities with less experienced staff is questionable, Dr. Goldstein said.
“Everyone agrees that MRI is more accurate, but does it change treatment or diagnosis? The answer is probably not. The question is whether you need to have that information to make a treatment decision, and the major decision is whether to use tPA or cardiovascular therapy. For the most part you don't need MRI to make that decision.”
Claiborne Johnston, MD, associate professor of neurology at the University of California-San Francisco (UCSF), and director of the UCSF Stroke Service, also questioned whether the results will have much immediate impact.
“I don't think these results are surprising to neurologists working with stroke patients – we already know that MRI is more specific in differentiating ischemic strokes from hemorrhagic strokes, and most of us tend to believe in the accuracy of either CT or MRI in identifying hemorrhagic stroke. The study's conclusion is that MRIs should be used for all patients, but this all depends on the feasibility in ER settings,” he said.
“Some facilities have MRIs right in their emergency rooms, but not many. This could change with results like these, but it will take time.”
But even if MRI becomes more widely used, he said, there are other problems with these patients. “The patient must remain still during an MRI, and that can be difficult. Also, many neurologists do not have experience with MRI.”
Dr. Johnston continued, “What I've seen is that most neurologists are confident in their ability to diagnose a stroke by the symptoms. If MRI were immediately available there's absolutely no question that it would be the preferred technique, and we might eventually get there. I would love to be there now.
“But we also hope that other kinds of CT, such as CT angiography and perfusion flow imaging, will speed the diagnostic imaging process. Most hospitals already have the ability to install these technologies without adding equipment.”
ARTICLE IN BRIEF
A new study reports that non-contrast MRI was about five times more sensitive, and two times more specific, than non-contrast CT in detecting acute ischemic stroke.