ARTICLE IN BRIEF
Researchers reported on a functional magnetic resonance imaging method that predicts memory change following temporal lobectomy.
Researchers have found that using functional magnetic resonance imaging (fMRI) with a somewhat different approach than previously reported can better predict significant verbal memory decline following epilepsy surgery — results that, if replicated elsewhere and in a larger number of patients, could aid decision-making on whether to proceed with surgery.
Researchers from the University College London found that greater activation in the left rather than the right side of the frontal and temporal lobes during word recall was the best independent predictor of verbal memory decline. The study included 24 patients with temporal lobe epilepsy — 23 of whom had left temporal lobe epilepsy — and 26 controls.
A lateralization index of at least 0.5 — indicating memory activation skewed heavily to the left hemisphere — had a sensitivity rate of 87.5 percent for significant memory decline following anterior temporal lobe resection, with a specificity of 80 percent, the researchers reported in the March 13 online edition of Neurology.
“The activation in the left temporal lobe with functional MRI on a memory test is the best predictor for individuals of what will happen to their memory,” said study author John Duncan, MD, a professor of neurology at the UCL Institute of Neurology.
The study expands on earlier work performed by the group, which looked at memory activation during word recall only in the temporal lobe. In this study, the analysis extended to the frontal lobe as well, yielding more powerful predictive capabilities.
“By taking a region of interest in both the frontal lobe that is primarily involving language and also the temporal lobe” — mainly involving verbal memory — “we're able to get both the effects of language dominance and memory dominance, and that gave us a superior sensitivity and specificity,” Dr. Duncan said.
Anterior temporal lobe resection is highly effective in treating refractory temporal lobe epilepsy, leading to remission in about 80 percent of patients, the study authors wrote. But 30 percent of patients who undergo surgery in the speech-dominant hemisphere experience significant memory loss.
That high risk can make decisions on whether to operate difficult, Dr. Duncan said.
“Let's say the patient worked as a lawyer and had to be able to summon up verbal arguments in court without any delay or hesitation,” he said. “If they had maybe one minor seizure a month, an operation that would have a big effect on their verbal memory would endanger their career and would not be appealing to them. If we could be confident that their verbal memory would not be affected, that would change the decision. So we wanted to devise a test to enable us to predict for individual patients what would happen to their verbal memory should they have surgery to the temporal lobe.”
In the study, patients were presented with 50 words on a screen while undergoing fMRI to measure brain activity. After scanning, the patients were presented with those same words again, as well as 100 other words they hadn't already been shown, and were asked whether they'd seen them already.
If a patient had a lateralization index of ≥ 0.5 on fMRI, there was a 70 percent risk of significant verbal memory decline after surgery. If the lateralization index was < 0.5, the risk of significant memory decline was 7.7 percent.
In patients with left temporal lobe epilepsy, predominantly left-sided “words remembered” activations within the amygdala, hippocampus, orbitofrontal cortex, inferior and middle frontal gyri, and anterior cingulate cortex correlated significantly with verbal memory decline following left anterior temporal lobe resection. In patients with right temporal lobe epilepsy, activations in the left inferior frontal gyrus correlated with verbal memory decline after right anterior temporal lobe resection.
Verbal memory before surgery, age at onset of epilepsy, hippocampal volume, and language lateralization did not correlate with post-operative verbal memory change, the researchers found. But Dr. Duncan said this was in part due to the small number of participants.
He added that the study, including the testing methods and the analysis, was done in a way that hopefully can be easily performed at other centers.
“I'm well aware that tests often do best in the place that designed them — that's always the case. We deliberately took quite a large region of interest in both the temporal and frontal lobes, so we've got the effect both of memory itself and of language function,” he said. “That makes it much easier for another center to put it in place and to replicate.”
If researchers at other centers arrive at similar results, the ultimate goal would be to find ways to alter surgery in the event that a patient is found to be at a high risk of verbal memory loss but is also at risk of having many seizures, Dr. Duncan said. He pointed to the way that changes to surgical methods have enabled surgeons to avoid post-operative visual field deficits by projecting the optic radiation into the operating microscope so that it can be avoided during the procedure.
Other experts in the field underscored the need for a multimodal approach in predicting memory loss. In an accompanying editorial, Max Trenerry, PhD, an associate professor of psychology at the Mayo Clinic in Rochester, MN, and Kimford J. Meador, MD, FAAN, a professor of neurology and neurosciences at Stanford University and clinical director of the Stanford Comprehensive Epilepsy Center, wrote that the “new findings need and deserve replication along with comparison to other methods for prediction of memory outcome in subgroups of patients defined by pertinent characteristics (e.g., MRI negative).
“Direct comparison of these techniques in the same patient group is needed to determine which of these data provide the most accurate, least invasive, and least expensive prediction of memory outcome,” they wrote.
“Whether we use the Wada [a more invasive neuroimaging and neuropsychological test to locate brain function prior to surgery] or other things, we don't do it in isolation,” Dr. Meador told Neurology Today. “We do it in the context of other evaluations. The more things that line up, the more likely we are to get good seizure outcomes and have low risk for memory problems.”
“A Grand Prix study of memory prediction methods” is needed, Drs. Meador and Trenerry wrote, but that would require sufficient data reporting and consistent use of comparable methodologies across studies.
The concerns regarding the need for replication notwithstanding, Dr. Meador said the sensitivity and specificity achieved in the study were “about as good as anybody has done with these types of measures.”
Michael R. Sperling, MD, FAAN, director of the Jefferson Comprehensive Epilepsy Center in Philadelphia, noted that other measures — such as hippocampal atrophy on MRI and focal hypometabolism on PET — are useful predictors as well. So “what remains to be learned is how much additional data this technique provides beyond that provided by other means,” he said.
“This study is a modest step forward — a good step — but not a leap, inasmuch as we can reasonably predict decline without this technique. It would be even more significant with a multimodality approach that should heighten sensitivity,” he said.
The technique would not be likely to alter surgical decisions in most cases, he added. “Without surgery, many patients experience significant memory decline over the years, and avoiding surgery might in some cases only delay the advent of memory decline. The small minority of patients at risk for a post-operative disabling memory deficit might find fMRI memory prediction useful and might forgo surgery, but the alternatives at present are suboptimal and only palliative.”
EXPERTS: ON fMRI TO PREDICT MEMORY LOSS AFTER EPILEPSY SURGERY