Abstracts

Rationale: The organization of the brain is nonrandom and highly economical. Balancing the cost and efficiency of this system requires a network connecting highly modular communities with topographically important central nodes or hubs. A highly promising but largely untested theory suggests that epilepsy affects complex brain networks, leading to abnormal systems rearrangement with implications for epileptogenesis and cognitive development. In this study, we aimed to evaluate the role of regional brain structures within the brain network of pediatric patients with new-onset epilepsy. We hypothesized that pediatric epilepsy is associated with a disruption of the normal pattern of structural network organization. Methods: We studied a group of 67 subjects: 28 healthy controls (mean age 13.3 ± 3.28 years, 11 males), 39 children with new-onset epilepsy (21 LRE (11.6 ± 2.68 years, 12 males) and 18 IGE (15 ± 3.3 years, 7 males)). Patients and controls were similar in age (p=0.6) and gender distribution (p=0.9). Subjects underwent MRI scanning yielding SPGR images (1.5 Tesla GE Signa MR scanner, TR = 24 ms, TE = 5 ms, flip angle = 40°, Slice thickness 1.5mm), which underwent automatic volumetric segmentation utilizing FreeSurfer. Data from 171 regions of interest were utilized for the construction of group-wise adjacency matrices (171x171 entries representing the partial correlation between the volumes from each pair of regions, controlled for age). Graphical metric properties (Betweenness Centrality and Clustering Coefficient) were calculated through the use of the Brain Connectivity Toolbox. Betweenness centrality represents the fraction of all the shortest paths that run through a node, and nodes with high betweenness centrality have high influence on the network and function as hubs. Clustering coefficient is a measure of "cliquishness" of a local community and represents the number of triangular connections between neighborhoods of 3 nodes. The regional properties were classified according to their distribution in percentiles ranging from 0 to 80th percentile, and the classified data was plotted according to the stereotaxic coordinates of the structural region on interest. Results: We observed a significant rearrangement of the regional network role of structures of interest between patients and controls. The results are summarized in Figure 1. Patients with epilepsy demonstrated a significant decrement in betweenness centrality within paracentral regions, with a concurrent segregation of these regions, as demonstrated by the diffuse lack of decrement in clustering coefficient. Conclusions: Children with new-onset epilepsy have a decreased number of hub regions while the local neighborhoods remained highly connected. These findings suggest that the normal developmental process is disrupted in pediatric epilepsy resulting in highly segregated and poorly integrated brain networks. Thus pediatric epilepsy is not only associated with regional structural changes, but also a global network restructuring, which may have direct impact on epileptogenesis, treatment and global cognitive development.