NEW ORLEANS—Multimodal MRI is a more accurate method of determining stroke etiology than CT-based assessment, suggested Monisha Kumar, MD, and colleagues at the 2008 International Stroke Conference. The prospective study demonstrated that MRI-based evaluation led to a change in diagnosis for approximately half of the study population who initially underwent CT, indicating overall poor positive predictive value (PPV) of CT-based stroke assessment.
“MRI is a more accurate diagnostic tool,” stated Dr. Kumar, Assistant Professor of Neurology at Stanford University School of Medicine in California. “It gives more information than a CT scan, such as vascular anomalies, congophilic angiopathy, and evidence of prior small vessel disease.”
Comparing CT- to MRI-Based Evaluation
Accurate and detailed diagnosis of stroke etiology has become increasingly important in determining goal-directed treatment strategies for the secondary prevention of stroke, and it is critical in the overall management of stroke patients. “The current work-up for stroke patients includes a wide variety of diagnostic tests,” said Dr. Kumar, “and their contribution to the final diagnosis has not previously been studied. [Our goal] was to determine whether MRI could be the pivotal study around which the stroke work-up could be conducted.”
The researchers sought to determine how often an MRI-based diagnostic algorithm altered final diagnosis in patients presenting with signs and symptoms of stroke. They examined 273 patients older than 18 with suspected cerebral disease who were able to comply with MRI and echocardiography. Within 48 hours of symptom onset, the subjects underwent a series of clinical diagnostic tests in the following order: CT, MRI, intracranial and cervical magnetic resonance angiography, and echocardiography. Modified Trial of Org 10172 in Acute Stroke Therapy (TOAST) classification was used by a stroke neurologist after assessment of history, physical examination, and CT, and following MRI. Initial diagnoses included large artery atherosclerosis; cardioembolism; small vessel disease; other determined; stroke, undetermined; not stroke; and uncertain.
Sensitivity and Positive Predictive Value of TOAST Classification
For large artery atherosclerosis, initial TOAST classification yielded a sensitivity of 60.8% and an overall PPV of 47.7%. While prevalence was estimated correctly, diagnosis was accurate only 46% of the time; 31 of the initial 66 patients were discharged with the original diagnosis. Fifty-one patients in the overall study group were discharged with a diagnosis of large artery atherosclerosis. Eleven patients originally diagnosed with large artery atherosclerosis were discharged with a diagnosis of cardioembolism.
For cardioembolism, initial TOAST classification had a sensitivity of 38.3% and PPV of 64.3%. Twenty-eight patients were initially presumed to have cardioembolism, while only 18 of them were discharged with the diagnosis. Overall, 47 patients were discharged with a diagnosis of cardioembolism. Many patients in whom this condition was initially predicted were discharged with a diagnosis of large artery atherosclerosis (n = 11) or small vessel disease (n = 8).
CT initially yielded a high sensitivity (94.0%) for small vessel disease; however, the PPV was 61.8%. The disorder was originally estimated in 102 patients, of whom only 63 were discharged with the same diagnosis. Many patients originally diagnosed with this condition were later discharged with a diagnosis of large artery atherosclerosis (n = 8) or cardioembolism (n = 8) after MRI. Overall, 67 patients were discharged with a diagnosis of small vessel disease.
Results demonstrated that slightly more than half (53.1%) of the patients had the same TOAST classification at discharge as initially identified after CT. Even with the expertise of fellowship-trained neurologists, CT is a less accurate primary neuroimaging method than MRI in determining the etiology of stroke, said the researchers.
“I think that more study needs to be done before definitive conclusions can be made,” said Dr. Kumar, “but our evidence does suggest that MRI can significantly guide the work-up of stroke and lead to perhaps a more efficient, timely, and cost-effective diagnostic strategy.”
While many stroke neurologists are already using MRI, Dr. Kumar believes this study will benefit non–stroke neurologists who believe they need to perform all aspects of diagnostic work-up “for the sake of completeness.” MRI is currently an American Stroke Association recommendation, but not a requirement, for the work-up of stroke. “With more study, this may change,” she concluded.
—Marisa Ruglio