Abstract & Commentary
Source: Sabsevitz DS, et al. Use of preoperative functional neuroimaging to predict language deficits from epilepsy surgery. Neurology. 2003;60:1788-1792.
Functional MRI (FMRI) has made tremendous strides as a research tool in providing insight into the anatomy and physiology of higher cortical function. Sabsevitz and associates report a correlation between fMRI results and language dysfunction following anterior temporal lobectomy (ATL) in the treatment of medically refractory epilepsy. They studied 24 patients who underwent left ATL (L-ATL) and compared them to 32 patients who underwent right ATL (R-ATL). All of the patients were evaluated with preoperative fMRI employing a language task, intracarotid amobarbital test (IAT, aka Wada test), and pre- and 6-month postoperative neuropsychological testing, including the 60-item Boston Naming Test (BNT). The fMRI study analyzed 8 region-of-interest (ROI) volumes for each hemisphere. Lateralization indices (LI) were calculated (LI = [L-R]/[L+R]), reflecting the interhemispheric difference between significantly activated voxel counts. An analogous LI was computed for the IAT based upon scores assigned for language performance for left vs right carotid injection.
Sabsevitz et al reported the following results. First, the L-ATL group had a significant (P < .001) decline in BNT score relative to the R-ATL group. Second, using Pearson correlation, the fMRI LI correlated with IAT LI, in concordance with prior studies. Finally, fMRI LI (to the left hemisphere) was significantly correlated (even more so than IAT LI in this series) with decrease in BNT. In analyzing specific ROIs, temporal LI was the best predictor of decline in BNT score. Specifically, using a > 2 standard deviation decline in BNT from the R-ATL group to define "poor outcome," fMRI temporal LI showed 100% sensitivity, 73% specificity, and 81% positive predictive value (PPV) with LI threshold set at 0.25. When poor outcome was defined as at least a 10-point decrease in BNT, temporal fMRI LI demonstrated 100% sensitivity, 57% specificity, and 63% PPV.
Neurology Alert has previously reported on fMRI as a tool for evaluation of patients undergoing epilepsy surgery.1 Previously published data have been most concerned with validating noninvasive fMRI compared to the "gold standard" of the invasive IAT, in lateralization of language dominance.2 The significance of the current study is the comparison of preoperative fMRI data to postoperative language deficit. Postoperative dysnomia could be considered a "platinum" standard, since it is a clinical outcome that relates directly to the functional risk of epilepsy surgery. If these results can be replicated, this study would represent the maturation of fMRI from a research instrument to a test that affects clinical decisions regarding assessing the risk of L-ATL in patients who have an intractable seizure focus localized to this region.
There are 2 caveats, however, in the interpretation of this study, and Neurology Alert hopes that Sabsevitz et al or others will provide us with further data. First, no information is presented regarding the results of other neuropsychometric testing other than the BNT. Certainly, non-naming language deficits can affect quality of life following epilepsy surgery. Exploring other language functions would provide important details about the PPV of fMRI for nondysnomic dysphasia. Second, and more importantly, Sabsevitz et al provide no data regarding the results of intraoperative or ictal electrocorticography (EcoG), nor EcoG functional mapping and how these results affected the "tailoring" of the anatomical boundaries of the resection margin. Knowledge of these data would obviously affect the robustness of fMRI as a predictor of neuropsychological risk of ATL for epilepsy. —Andy Dean
1. Dean A. Neurology Alert. 2002;21:15.
2. Gaillard WD, et al. Neurology. 2002;59:256-265.
Dr. Dean is Assistant Professor of Neurology and Neuroscience, Director of the Epilepsy Monitoring Unit, Department of Neurology, New York Presbyterian Hospital—Cornell Campus.