Study on the Relationships between Intrinsic Functional Connectivity of the Default Mode Network and Transient Epileptic Activity

Abstract

Rationale: Simultaneous recording of electroencephalogram and functional MRI (EEG-fMRI) is a powerful tool for localizing epileptic networks via the detection of hemodynamic changes correlated with interictal epileptic discharges (IEDs). fMRI can be used to study the long-lasting effect of epileptic activity by assessing stationary functional connectivity during the resting-state period [especially, the connectivity of the default mode network (DMN)]. Temporal lobe epilepsy (TLE) and idiopathic generalized epilepsy (IGE) are associated with low responsiveness and disruption of DMN activity. A dynamic functional connectivity approach might enable us to determine the effect of IEDs on DMN connectivity and to better understand the correlation between DMN connectivity changes and altered consciousness.

Method: We studied dynamic changes in DMN intrinsic connectivity and their relation to IEDs. Six IGE patients (with generalized spike and slow-waves) and 6 TLE patients (with unilateral left temporal spikes) were included. Functional connectivity before, during, and after IEDs was estimated using a sliding window approach and compared with the baseline period.

Results: No dependence on window size was observed. The baseline DMN connectivity was decreased in the left hemisphere (ipsilateral to the epileptic focus) in TLEs and was less strong but remained bilateral in IGEs. We observed an overall increase in DMN intrinsic connectivity prior to the onset of IEDs in both IGEs and TLEs. After IEDs in TLEs, we found that DMN connectivity increased before it returned to baseline values. Most of the DMN regions with increased connectivity before and after IEDs were lateralized to the left hemisphere in TLE (i.e., ipsilateral to the epileptic focus).

Conclusion: RESULTS suggest that DMN connectivity may facilitate IED generation and may be affected at the time of the IED. However, these results need to be confirmed in a larger independent cohort.

Keywords: default mode network; dynamic; epileptic interictal event; functional connectivity; idiopathic generalized epilepsy; posterior cingulate gyrus; precuneus; temporal lobe epilepsy.

Figure 1

Definition of the WOIs. The red arrow represents the timing of the epileptic event (according to EEG data). Four types of window were defined: “before,” “during,” and “after” the epileptic event, together with the “baseline” period (i.e., with no epileptic events LTR seconds before and after the window). Tapered windows were used by convolving a rectangle (length: L TRs) with a Gaussian (σ = 2 TRs).

Figure 2

DMN functional connectivity in TLE vs. IGE groups. The DMN integration (dimensionless) was computed for each subject of the two groups. A Bayesian numerical sampling scheme was used for the inference of integration measures in a group analysis. The integration was approximated from 1000 samples. The error bars indicated the standard deviation of the 1000 samples. There was significantly less integration in the TLE group than in the IGE group (right panel). To illustrate this difference, DMN was estimated for each group using the same groupICA approach than the Section “DMN mask” (left panel). Maps were thresholded (z-score > 2).

Figure 3

Pairwise comparisons of node strength and the clustering coefficient in the TLE group (in a Wilcoxon signed-rank test). The node sizes correspond to the mean network measure being tested. Colors indicated the direction of change (the red lines mean a decrease, and the blue lines an increase, from the first condition to the second). Different line style indicated significance (thin for FDR-corrected p-values <0.1 and thick for FDR-corrected p-values <0.05).

Figure 4

Pairwise comparisons of node strength and the clustering coefficient in the IGE group (in a Wilcoxon signed-rank test) thresholded at FDR-corrected p-values <0.05. The node sizes correspond to the mean network measure being tested.

Figure 5

Maps of DMN intrinsic connectivity in the TLE group. The node sizes correspond to the number of connections for a node and the gray lines show significant connections between pairs of regions (FDR-corrected p < 0.05). The variable D represents the network density (i.e., the number of significant connections divided by the total number of connections).

Figure 6

Maps of DMN intrinsic connectivity in the IGE group. The node sizes correspond to the number of connections for a node and the gray lines show significant connections between pairs of regions (FDR-corrected p < 0.05). The variable D represents the network density (i.e., the number of significant connections divided by the total number of connections).

Figure 7

Significant differences between two WOIs in the TLE group. The node sizes correspond to the number of connections for a node and the gray lines show significant connections between pairs of regions (FDR-corrected p < 0.05).