In Epilepsy, Not All MRI Scans are Created Equal

Abstract & Commentary

Source: Von Oertzen J, et al. Standard magnetic resonance imaging is inadequate for patients with refractory focal epilepsy. J Neurol Neurosurg Psychiatry. 2002;73:643-647.

Among patients undergoing epilepsy surgery for pharmacologically refractory seizures, detection and excision of a structural lesion congruent with the electroencephalographic ictal onset is a strong predictor of a seizure-free outcome. For cases of hippocampal sclerosis (HS), the seizure-free rate can be as high as 85-90% vs a rate of about 70% for all forms of temporal lobe epilepsy. Similarly, in extratemporal cases, successful outcome from epilepsy surgery is 50% in nonlesional cases but rises to 70% when a structural epileptogenic focus is identified.

In analyzing the interpretation of "standard" cerebral MRI vs MRI performed according to epilepsy-dedicated protocols, Von Oertzen and colleagues conclude that the former scans do not yield sufficient diagnostic information to guide prognosis for surgical treatment. In addition to classifying MRI scans as either standard or epilepsy-dedicated, they identified the readers as either nonexpert or expert, where the latter were defined as the neuroradiologists in their epilepsy center, all of whom had greater than 3 years of "epileptological experience." Nonexperts interpreting standard MRI scans in 123 patients reported 61% as normal. Of the 39% of scans reported as showing structural lesions, 7% (n = 8) were said to show HS. By contrast, when the standard MRI scans were re-interpreted by the expert readers, 28% were deemed technically inadequate, 22% were read as normal, and 50% were reported as having structural lesions. Eighteen percent (n = 22) of all scans showed HS. Leaving aside the possibility that the nonexpert radiologists may have read some "false-positive" HS (particularly if some studies were technically inadequate), the expert readers picked up an additional 14 patients whose imaging findings may have placed them in a more favorable outcome category. In fact, if the structural lesions were identified sooner, an earlier referral to an epilepsy center may have been made, possibly resulting in reduced seizure-related morbidity for these patients. Even more dramatically, when expert readers interpreted epilepsy-dedicated MRI scans, 85% of patients with reportedly normal standard MRI were found to have structural lesions. Of those patients who underwent resection, neuropathological diagnosis was correctly predicted by 89% of dedicated MRI reports, but only 22% of "nonexpert" standard MRIs.

Commentary

By 1999 (when this study closed) routine cerebral MRI scans done in the community demonstrated poor sensitivity in diagnosing lesional epilepsy. There is no reason for this to be the case, since an epilepsy-dedicated MRI can be performed on standard 1.5 Tesla scanners with software already installed. The problem of inadequate MRI protocols is compounded by a shortage of neuroradiologists with sufficient experience with epilepsy. Unfortunately, Von Oertzen et al were not able to address the sensitivity and specificity of nonexpert interpretation of dedicated MRI. I suspect that the sensitivity of identifying structural lesions would increase, but there may still be sufficient concerns regarding decreased specificity to warrant systematic investigation of this issue.

More alarmingly, the data in this paper raise the possibility that patients presenting with a first seizure and a subtle (but epileptogenic) structural lesion may not be properly diagnosed. While prudent practice follows the rule of treating the patient, not the scan or EEG in isolation, the threshold for antiepileptic drug prophylaxis is potentially lower if there is knowledge of imaging or EEG abnormalities predisposing to recurrent seizures. A higher neuroimaging standard is needed in evaluating first-time seizures, as well as intractable epilepsy. Dedicated epilepsy MRIs for a first-time seizure could decrease potential morbidity associated with delaying treatment until the epilepsy declares itself. — Andy Dean

Dr. Dean is Assistant Professor of Neurology and Neuroscience, Director of the Epilepsy Monitoring Unit, Department of Neurology, New York Presbyterian Hospital—Cornell Campus.