Abstracts

Rationale: Hemodynamic response to interictal spikes often extend the area of the EEG sources, suggesting epilepsy as a network disease. Brain network can be studied using Functional Connectivity (FC) of resting state fMRI (rfMRI) in the absence of epileptic discharge, measuring temporal correlation between the signals in remote brain regions (Pittau et al., Epilepsia 2012; 53(6); 1013-1023). In FC, connector hub regions are important for integrity and communication between networks. Altered patterns of FC were observed in mesial temporal lobe epilepsy (mTLE) using rfMRI, while white-matter hubs assessed using diffusion tensor imaging were found in the meta-analysis of MRI lesions in TLE (Crossley et al., Brain 2014; 137; 2382-2395). We hypothesized that we could detect abnormal connector hub organization in patients with mTLE. Methods: We obtained EEG/rfMRI data from 14 patients with unilateral mTLE associated with mesial temporal sclerosis (5 left (L) and 9 right (R)) and 21 age- and gender-matched healthy subjects. For each subject, a 6-min rfMRI run with EEG showing less than two epileptic discharges was selected and analyzed using a SParsity-based Analysis of Reliable K-hubness (SPARK) method, identifying connector hubs by the number (k) of functionally overlapping networks in each voxel (Lee et al., Neuroimage 2016; 134; 434-449). We estimated the group average k-hubness map for each group to find a consistent spatial pattern of hub organization. To quantify the hub reorganization across the brain in each patient compared to controls, we computed the global hub disruption index (Achard et al., PNAS 2012; 109(50); 20608-20613), the degree to which the difference of k between a patient and the group of controls varies with respect to k measured in controls across the whole brain: a negative index indicates that some hubs in controls become non-hubs in patients, while some non-hubs in controls become hubs in patients. We also measured the L/R temporal hub disruption index within the L/R temporal area, including the hippocampus, amygdala, parahippocampal gyrus, superior/middle/inferior temporal gyri, superior/middle temporal poles, fusiform and lingual gyri, and insula. Results: The group average k-maps from three groups showed decreased k-hubness in the temporal lobe in both L and R mTLE patients compared to controls (Fig 1. A-C). Using the hub disruption index, we found a significant hub disruption in individual mTLE patients compared to controls (Fig 1. D-I). The mean of global hub disruption index was -0.4 in L mTLE and -0.36 in R mTLE compared to -0.2 in controls; the mean of left temporal hub disruption index was -0.39 in L mTLE and -0.47 in R mTLE compared to -0.23 ; the mean of right temporal hub disruption index was -0.37 in L mTLE and -0.47 in R mTLE compared to -0.19 in controls. Both the left and right temporal hub disruption indices significantly decreased in R mTLE patients compared to controls at p < 0.05 using Mann-Whitney test. Conclusions: We observed hub reorganization in patients with mTLE by applying the SPARK method to a 6-min rfMRI scan of each patient, in the absence of any detectable epileptic discharge. The hub reorganization included that connector hubs in parts of the temporal lobe in controls became non-hubs in individual patients with mTLE. Studying reorganization of functional connector hubs in rfMRI may provide information about patient-specific abnormal epileptogenic network in mTLE. Funding: This work was supported by the CIHR MOP-130442 and MOP-133619, NSERC Rgpin 35661, and Irma H Bauer Fellowship.