Abstracts

Rationale: There are some reports of language decline following epilepsy surgery involving the anterior insula; however, few studies have investigated this topic especially in a pediatric epilepsy population. The anterior insular is thought to have a role in preparatory functions/coordination of language and word finding. Using historical data with our current language tasks, we explored whether the anterior insula was activated during a language fMRI task, if patterns of activation differed between pediatric epilepsy patients and controls, and, if activated, if anterior insular activation predicted language scores. Methods: Ninety-six participants (39 epilepsy patients (EPI), 28 left focus), see Table 1, completed neuropsychological evaluation and fMRI language task (Auditory Description Decision Task) that activates the temporal and frontal language network. We first investigated activation in the anterior insula between EPI and controls (TD) using the 39 EPI compared to 39 age-matched controls. Magnitude of activation, voxel count, and laterality indices were calculated in the anterior insula mask in SPM12 and compared between EPI and TD. We used linear regression to explore relationships between language scores and the following factors: magnitude of activation, voxel count, and LI in anterior insula within the whole group as well as in the EPI or TD groups separately. Results: About half of the sample across both EPI and TD activated the anterior insula during language fMRI. Looking at group contrast, TD activated more compared to EPI, p=0.001. Magnitude of insular activation regardless of hemisphere was greater in TD, p<0.05. As a group (both TD and EPI), magnitude in both left and right anterior insula predicted confrontation naming and vocabulary scores, p’s<0.05, while only magnitude on the left predicted rapid digit naming (p<0.05) and rapid letter naming at trend level, p=0.06. Voxel count and LI did not predict language scores.Looking at EPI only (28 left dominant/left seizure focus), magnitude in the left anterior insula predicted rapid letter and digit naming and vocabulary scores, p’s<0.05. Magnitude on the right predicted rapid naming scores, p<0.05. EPI who activated voxels in the anterior insula at the p=0.001 level performed significantly better on the naming task as compared to EPI with no voxels activated. In the TD group, magnitude of insular activation did not significantly predict language performance. Conclusions: About half of TD and EPI activate the anterior insula during a language fMRI task. Despite modest EPI activation, this exploratory study provides evidence that activation in the anterior insula is predictive of language performance for patients. These findings are consistent with studies that have found insular involvement in speech planning/fluency and word finding and deficits following resection of the insula. A fMRI task that requires overt fluency and naming might optimize anterior insula activation; however, these initial results suggest further exploration is warranted. Funding: Children’s National Medical Center Research Institute