Memory fMRI predicts verbal memory decline after anterior temporal lobe resection

Objective: To develop a clinically applicable memory functional MRI (fMRI) method of predicting postsurgical memory outcome in individual patients. Methods: In this prospective cohort study, 50 patients with temporal lobe epilepsy (23 left) and 26 controls underwent an fMRI memory encoding paradigm of words with a subsequent out-of-scanner recognition assessment. Neuropsychological assessment was performed preoperatively and 4 months after anterior temporal lobe resection, and at equal time intervals in controls. An event-related analysis was used to explore brain activations for words remembered and change in verbal memory scores 4 months after surgery was correlated with preoperative activations. Individual lateralization indices were calculated within a medial temporal and frontal region and compared with other clinical parameters (hippocampal volume, preoperative verbal memory, age at onset of epilepsy, and language lateralization) as a predictor of verbal memory outcome. Results: In left temporal lobe epilepsy patients, left frontal and anterior medial temporal activations correlated significantly with greater verbal memory decline, while bilateral posterior hippocampal activation correlated with less verbal memory decline postoperatively. In a multivariate regression model, left lateralized memory lateralization index (≥0.5) within a medial temporal and frontal mask was the best predictor of verbal memory outcome after surgery in the dominant hemisphere in individual patients. Neither clinical nor functional MRI parameters predicted verbal memory decline after nondominant temporal lobe resection. Conclusion: We propose a clinically applicable memory fMRI paradigm to predict postoperative verbal memory decline after surgery in the language-dominant hemisphere in individual patients.

Predicting memory change after temporal lobectomy for epilepsy Anterior temporal lobectomy is an accepted and effective treatment for medically resistant seizures. Anterograde verbal memory decline is a common and potentially debilitating morbidity associated with (typically language-dominant) temporal lobectomy. 1 In this issue of Neurology ® , Sidhu et al. 2 have described a functional MRI (fMRI) method that predicts memory change following temporal lobectomy.
Review of their data table shows a strong correlation between their memory Lateralization Index (LI) and regression-adjusted change in ability to learn a supraspan word list over serial presentations. Their statistical analysis demonstrated that the LI was the best predictor of change compared to volume of the resected hippocampus, age at onset, duration, and other factors. This group of investigators previously reported a similar finding for a different fMRI technique, 3 and their current fMRI method demonstrates involvement of frontal networks in verbal learning that appear useful in predicting postsurgical memory change.
Prediction of memory change following temporal lobectomy with fMRI appears to hold promise for future patient care. This methodology also needs to be considered in the context of prior research on prediction of memory change following temporal lobectomy. 4 Risks are higher for left (language-dominant) resections in patients with high preoperative neuropsychological functions (especially verbal memory and naming), no structural lesion (e.g., mesial temporal lobe sclerosis), later onset epilepsy, and lack of ipsilateral PET hypometabolism. In addition, Wada testing and quantitative MRI data predict posttemporal lobectomy memory change. Recognition scores following left and right internal carotid artery amobarbital injections predict memory outcome following language-dominant temporal lobectomy. Better recognition after the left injection (testing the right hippocampus) compared to recognition after the right injection (testing the left hippocampus) predicted memory change after left temporal lobectomy. 5 The difference between MRI-determined hippocampal volumes predicted verbal memory change after language-dominant left temporal lobectomy. 6 T2 relaxometry predicts memory change after nonlanguage-dominant temporal resections. 7,8 Overall, these results indicate that prediction of memory outcome requires that functionality of the to-be-resected hippocampus needs to be compared to the remaining hippocampus. 9 How these technologies, methods, or other data are best integrated to predict memory change is unknown. 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. For example, how much if any extra information regarding memory risk does fMRI or Wada testing add for a right-handed patient with left temporal onset seizures who has left hippocampal mesial temporal lobe sclerosis and atrophy, poor baseline verbal memory and naming, and onset of epilepsy as an adult? Decisions concerning epilepsy surgery are multifactorial, including measures related to localization of the seizure onset zone and measures related to localization of cognitive dysfunction and function. As new measures are developed, it is critical to determine their role in the battery of measures used to determine the risks and benefits of proposed surgery.
Sidhu et al., and other investigators, propose that fMRI might have the capacity to provide reassuring data and mitigate the risk suggested by the lack of MRI-detected hippocampal abnormality. However, exact fMRI methods have not been replicated by other centers and in large patient samples, and so broad experience with these methods is lacking.
In a sense, what is needed is a "Grand Prix" study of memory prediction methods to determine which methods are necessary and sufficient to determine risk to memory in temporal lobe epilepsy patients with a particular set of characteristics. To pursue such a comparison, published research needs to (1) report sufficient data and analyses to make possible detailed comparisons across studies, and (2) consistently utilize comparable methodologies and operational definitions. This would include the complete correlation table for first-order correlations among all variables in the study. Multivariate statistical analysis methods should be described in sufficient detail to allow for replication. For example, in the present study, it is unclear if the multiple regression analysis used stepwise or simultaneous entry of independent variables. This approach does not indicate if risk for memory decline would be better predicted by fMRI depending on the distributions of known risks for memory decline such as age at onset, difference between hippocampal volumes, or presurgical memory. While this study reports on memory, and the findings are important and interesting, the authors seem to operationalize memory as the number of words learned from a list over serial presentations of the list. For many investigators, this would be considered learning, or learning over trials, and memory would be considered the number of list words that could be recalled after a delay of 20 or 30 minutes. Use of a delayed recall trial might provide different results.
Despite the cautionary notes raised above, Sidhu et al. highlight the possibility of using a noninvasive methodology to evaluate risk to memory prior to temporal lobectomy, and their research group has worked diligently to enhance this approach. Their 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). The overall results seem to support Chelune's conceptualization that memory change is associated with the functional adequacy of the to-be-resected tissue, and to emphasize the role of neural networks in cognitive functions.

STUDY FUNDING
No targeted funding reported.