Abstracts

Rationale: Epilepsy surgery is an effective method for seizure control, but post-surgical language deficits can occur. FMRI can be used to identify brain regions that process language, but few studies have used fMRI to investigate post-operative language outcomes in epilepsy patients. We sought to determine whether fMRI lateralization in language regions changes following epilepsy surgery. Methods: Twenty-one patients (mean age 32.2 +/- 12 years) referred to the Clinical Epilepsy Section, NINDS, NIH for evaluation of intractable epilepsy completed fMRI language mapping before and after temporal resection using an auditory description decision task (ADDT). Six were excluded due to motion artifacts. Sixteen healthy volunteers (mean age 22.1 +/- 3.1 years) completed repeated fMRI language mapping (one year apart) using ADDT. Patients and controls completed Boston Naming and Controlled Oral Word Association tests with repeated tests after surgery. Regional laterality indices (LI) for each task were calculated for three language areas: inferior frontal gyrus (IFG), midfrontal gyrus (MFG), and Wernicke's Area (WA). The hemisphere of language dominance, categorized by LI, was compared pre- and post-operatively (for patients) or between repeated scans (for controls). Regional LI > 0.20 on ADDT were classified as left dominant, LI < -0.20 were classified as right dominant, and -0.20 < LI < 0.20 were classified as bilateral. Tissue map differences between pre- and post-surgical anatomy scans were used to create a resection mask and calculate resection volume. Results: Controls had no significant change to language dominance in any region between repeated scans. There was no difference between repeated Boston Naming or Controlled Oral Word Association tests for patients or controls. 13 patients had temporal lobe resections (6 left, 7 right), one had a left amygdalohippocampectomy, and one had a right parietal resection. Mean resection volume was 18.2+/- 9.6 CC and, of the top 10 percent of functional activation, a mean of 0.52 +/- 0.01 CC was within the resection volume. Resection area that overlapped with WA was 3.59 +/- 1.37 CC, and with IFG was 0.01 +/- 0.02 CC. There was no overlap with MFG. However, after surgery, 10 patients had significant LI changes in MFG: three patients went from left to bilateral (one left temporal, two right temporal resection); four went from bilateral to left (two right temporal, one right parietal, one left temporal resection); two went from right to left (both right temporal resections); and one patient went from left to right dominant (left temporal resection. The amount of overlap with Wernicke's area predicted the level of LI change (post-pre surgery): r=-0.66, p < 0.01. Two patients had significant LI changes in WA: one went from left to bilateral (left temporal resection), and one went from bilateral to left dominant (right temporal resection). Two patients had significant LI changes in IFG: one patient went from bilateral to left dominant (left temporal resection), and one went from right to left dominant (right temporal resection). Conclusions: These results may represent language processing compensation and reorganization to the contralateral side to resection, and may be related to the resection volume and overlap with functional activation, or post-operative seizure control, and warrant further study. Funding: n/a