Abstracts

Rationale: A spatiotemporal evaluation of pathological hypersynchrony marks the onset and propagation of focal seizures in temporal lobe epilepsy. The seizures originate in dysfunctional brain networks and spread to surrounding tissues, disrupting function in otherwise relatively quiescent brain. Studies have performed meticulous time-varying and static analysis of functional networks, to identify brain regions whose topological role was to synchronize or desynchronize the epileptic network, but these were limited to associations at cortical level (1-2). Current evidence points to a very critical role of a sub-cortical hub in predicting the treatment outcomes in TLE, i.e., the thalamus (3). We hypothesize that the anterior nucleus of thalamus (ANT) has a central role in the early organization of seizure network in temporal lobe epilepsy (TLE). We aim to look at the influence of thalamus (Th) and its cortical connectivity on the initiation and propagation of seizures in a time-varying targeted attack analysis. Methods: Patients with suspected TLE with a localized SOZ were selected for the study, a total of 9 patients and 34 seizures were analyzed. Epileptogenecity Index was used to select channels located in the seizure onset zone, seizure generation network and the limbic system. Line Length was used to determine seizure onset and time of thalamic involvement. Interchannel intact connectivity matrices (Th+) was estimated using Generalized Partial Directed Coherence (GPDC) in the baseline period, then during onset, propagation, termination and post-ictal periods. This was done for frequency bands ranging from delta (1-4Hz) to high gamma (45-95Hz). Next, we performed Targetted Attack Analysis focused on Thalamus using two approaches: (1) Node resection (ThNR) simulating the absence of Thalamus in the entire network and (2) Edge resection (ThER) simulating the removal of influence of thalamus on the remaining network. Network resilience was estimated using betweenness centrality and clustering coefficient. * FDR corrected results were considered significant. Results: We found that seizure networks were vulnerable and showed decreased centrality to targeted node and edge attack on the thalamus. Edge resection caused a greater decrease in centrality compared to node resection alone for all frequencies (Th+ vs. Thnr FDR-p’s<0.048, Th+ vs. ThER FDR-p’s<0.047). This post-attack re-organization was influenced by the seizure stages in theta and alpha frequencies as noted by the significant interaction statistics (FDR-p’s<0.007). Post-attack Clustering coefficient showed a similar decreasing significantly following thalamic edge resection. Conclusions: The current results shed light on the crucial concept of how to modulate thalamus to cause the network to disintegrate, resulting in termination of the seizure. A disruption of centrality of the seizure network is likely to prevent seizure spread and eventually result in seizure termination. A fall in clustering coefficient in the seizure network can prevent irreversible pathological neuronal plasticity (i.e., kindling) from occurring within the seizure network that lower seizure threshold and epileptogenesis. We also emphasize that perturbing the thalamic connectivity is more crucial to such therapeutic targets. Funding: No funding