Abstracts

Rationale: In previous work, we found evidence for language plasticity after epilepsy surgery with pre- and post-operative functional MRI (fMRI). We now assess effects of surgery on language dominance and task activation in a larger population to examine reorganization in detail. Methods: 36 patients with temporal lobe epilepsy (TLE, surgery age 18-61 yrs, m=36.8(11.5);20 female; seizure onset 0.25-38 yrs, m=13.25(10.5); 19 left-hemisphere resection, 17 right resection) completed structural MRI and language fMRI before and after surgery (>6 months, range 1-8 yrs). Language fMRI task is Auditory Description Decision Task (ADDT) - a semantic decision task activating IFG & Wernicke's. 32 patients completed pre- and postoperative Boston Naming Test.We computed language laterality index (LI: (left-right)/(left+right)) for IFG and Wernicke's area over the top 10% activation map, which is based on positive SPM contrast T values and increases spatial distribution reliability within subjects across runs. We automatically generated resection masks from pre/postoperative MRIs and exclude them for postoperative LI calculation to avoid bias from reduced brain volume. We used Wernicke LIs to determine language dominance, and quantified activation change by computing 3D IFG and Wernicke peak locations to determine hemispheric shifts between pre- and postoperative peaks. Fifteen healthy volunteers (HV) (age 21-25 yrs, m=23.0 (1.0); 8 female) had two ADDT runs >6 months apart. Normal data were used to calculate Z scores to identify patients with significant LI change (|z|>1.65, 90% confidence interval). Results: All HVs had Wernicke LI left language dominance; 25 patients had left language dominance (11 dominant-left, 14 nondominant-right resection), 10 bilateral (all considered dominant resection, 7 left resection, 3 right), 1 right dominance (nondominant left resection). 11 patients had significant LI change (|Z|>1.65) in IFG (9 dominant resection, 7 decreased LI), 4 in Wernicke (all dominant resection, 3 decreased LI). 14 out 36 (38.9%) patients changed IFG peak sides (11 dominant resection).  8 patients changed Wernicke peak side (6 dominant resection). Only one HV changed IFG and one Wernicke peak side. Patients had higher odds ratio for IFG peak side change than HVs (Fisher’s exact test, p=0.04).Patients had significantly reduced postoperative Wernicke LI (paired t test, t(35)=2.33, p=0.025). Within the left language dominance group (N=25) patients had increased right hemisphere Wernicke activation (decreased LI, paired t test, t(24)=3.20, p=0.004), similarly for its dominant (p=0.024) and nondominant (p=0.07) resection subgroups, while dominant subgroup also had LI reduction in IFG (N=11, p=0.08). LIs did not differ between 2 time points for HVs (paired t test, IFG: t(14) = 1.72,  p=0.11; t (14) = 0.56, Wernicke's p=0.56).Within the left language dominance group, dominant resection patients had greater BNT score reduction than those with nondominant resection (t(17.9)= -2.8, p = 0.01). Conclusions: Activation peak shift and reduced LI show resection side and language dominance affect language network reorganization in TLE.  Post-operative homologous language area recruitment may underlie functional compensation. Top 10% activation based LI calculation may be an important metric for assessing postoperative language function. Funding: NINDS NIH Clinical Epilepsy Section Division of Intramural Research; Children's National Medical Center Research Institute