Abstracts

Rationale: In patients with focal epilepsy, brain connectivity alterations in the absence of visible interictal epileptiform discharges (IED) could reflect profound network alterations and explain cognitive changes. Using Granger-causal modeling applied to EEG source signals, we investigated connectivity alterations in left and right temporal lobe epilepsy (LTLE and RTLE) compared to controls in the absence of scalp interictal epileptiform discharges (IED).Methods: Twenty LTLE, 20 RTLE patients and 20 healthy controls underwent a resting-state high-density EEG. Source activity free of IED was obtained for 82 regions of interest (ROI) using an individual head model and a distributed linear inverse solution. Granger-causal modeling (weighted Partial Directed Coherence) was applied to the source signal of all ROIs in usual frequency bands to estimate the summed outflow (SO) from each ROI and ROI-to-ROI directed connectivity between ROIs with high SO. Correlations with duration of the disease, age, cognitive impairments and mood impairment were computed. A non-parametric statistical test assessed significant difference between groups.Results: In all regions and groups, SO peaked in the alpha band and subsequent analysis were restricted to this frequency band. The highest SO occurred in the medial temporal lobe and cingulate cortex, concordant with regions of the Default Mode Network (DMN). The strongest connections in controls were from the Posterior Cingulate Cortex, while in both patient groups, these were from the ipsilateral hippocampus. RTLE was more contralaterally affected than LTLE compared to controls. In RTLE, disease duration was negatively correlated to SO in bilateral amygdala and contralateral hippocampus, olfactory and rectus, while in LTLE it was positively correlated to SO in contralateral hippocampus. In both patient groups, SO was significantly decreased with age in regions concordant with the piriform cortex. We found a trend for more bilateral cognitive deficits and significant impairment of prefrontal regions in RTLE than LTLE. SO from the Anterior Cingulate Cortex was lower in patients with depressive disorders compared to patients with normal mood, who had lower values than controls.Conclusions: Directed connectivity based on high density EEG and electric source imaging was able to identify consistent patterns of resting-state connectivity in TLE, connectivity was impaired in regions of the DMN and the strongest driving was shifted to the ipsilateral hippocampus. These network abnormalities, independent of scalp IED were more marked with older age in regions concordant with the piriform cortex, while the cingulate gyrus connectivity was related to memory and mood changes. This could represent an important biomarker of TLE for the severity of the disease as well as cognitive and psychiatric comorbidities.