Abstracts

Rationale: Functional imaging of brain connectivity may provide insights into the neural basis for language and memory impairments in left temporal lobe epilepsy. Field theory allows a means on assessing the broad impact of epilepsy on organization of fundamental brain networks. Methods: We used 3T EPI BOLD fMRI during a language (auditory word definition decision) task and independent component analysis to compute functional network connectivity networks for 62 healthy controls (mean age 26.9 yrs; range 18-56) and 83 patients (mean age 24.7; range 8-57) with left temporal lobe epilepsy. Data were analyzed in SPM2 for regional laterality indices (IFG and Wernicke s area) and independent component analysis (ICA). Patient adjacency matrices following ICA were reshaped into feature vectors and clustered using the Euclidian distance k-means clustering algorithm. Small-world indices to quantify information transfer efficiency were calculated for each individual using clustering coefficients and average path lengths. Results: Patient cluster 1 (n=20, (24%)) showed increased connectivity and small-world indices, while patient cluster 2 (n=63, (76%)) showed decreased connectivity and normal small-world indices in relation to healthy controls (Kruskal-Wallis test (K = 17.82, p = 0.0001)). There were no differences between the patient clusters for age of seizure onset, study age, Verbal IQ or regional language laterality indices (Mann-Whitney U Tests). Network visualizations indicated a two-module architecture in healthy controls, and confirmed higher connectivity in patients in cluster 1 and reduced connectivity in patients in cluster 2. Conclusions: These clusters likely represent different compensatory strategies necessary to support language performance and are independent of language dominance. Our results show that functional network connectivity patterns of patients with left temporal lobe epilepsy are not homogenous and that distinct subgroups exist.