#7222 Modelling NMDA receptor activity in focal epilepsy: a resting state study of EEG effective connectivity, towards a marker for epileptic seizures from unremarkable EEG traces

<h3></h3> The role of N-methyl-D-aspartate (NMDA) receptors in the pathophysiology of epilepsy and seizure generation has been established in pathological and animal models. Recent research has investigated how NMDA receptors modulate brain dynamics in patients with NMDAR autoimmune encephalitis using Dynamic Causal Modelling (DCM) for effective connectivity analysis. With this work, we first aim to study regional hippocampal NMDA-dependent effective connectivity in patients with medial temporal lobe epilepsy in resting state EEG recording without any clinical evidence of epileptogenic activity. Moreover, if validate, this approach could be a useful tool to differentiate between epileptic and non-epileptic seizures in clinically unremarkable resting state EEG traces. Resting state clinical EEG data and MRI structural brain scans were collected from a cohort of 20 patients with diagnosis of epilepsy and unilateral hippocampal sclerosis. The EEG data were pre- processed to obtain 5-seconds epochs of source-localised electrophysiological activity arising from left and right hippocampus. MRI brain scans were used as the anatomical reference for patient-specific source localisation analysis. We studied individual resting state microcircuitry in the affected (sclerotic) versus non-affected hippocampus using a neural mass model of regional network dynamics. We then performed group-level DCM analysis validating the model with empirical priors. Our effective connectivity model was able reproduce patient and area-specific changes in the NMDA- dependent effective connectivity of the hippocampal microcircuitry. We were able to identify individual differences between the affected and unaffected hippocampus, and to obtain a group effect across the patient population with the use of shared empirical priors. In the next step of this work, we are now validating this approach running the same network analysis on control resting state EEG to validate the results. This work validates DCM as a valuable approach to investigate individual and group resting state EEG network dynamics, in particular in identifying abnormal NMDAR activity using clinical EEG when no obvious interictal epileptiform abnormalities are found.