Functional connectivity alterations in patients with post-stroke epilepsy based on source-level EEG and graph theory

doi:10.21203/rs.3.rs-2437742/v1

Download PDF

Research Article

Functional connectivity alterations in patients with post-stroke epilepsy based on source-level EEG and graph theory

https://doi.org/10.21203/rs.3.rs-2437742/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 16 Apr, 2024

Read the published version in Brain Topography →

You are reading this latest preprint version

Purpose:We investigated the differences in functional connectivity based on the source-level electroencephalography (EEG) analysis between stroke patients with and without post-stroke epilepsy (PSE).

Methods: Thirty stroke patients with PSE and 35 stroke patients without PSE were enrolled. EEG was conducted during a resting state period. We used a Brainstorm program for source estimation and the connectivity matrix. Data were processed according to EEG frequency bands. We used a BRAPH program to apply a graph theoretical analysis.

Results: In the beta band, radius and diameter were increased in patients with PSE than in those without PSE (2.699 vs. 2.579, p=0.003; 2.261 vs. 2.171, p=0.003). In the low gamma band, radius was increased in patients with PSE than in those without PSE (2.808 vs. 2.617, p=0.004). In the high gamma band, the radius, diameter, eccentricity, and characteristic path length were increased (1.828 vs. 1.559, p=0.001; 2.653 vs. 2.306, p=0.002; 2.212 vs. 1.913, p=0.001; 1.425 vs. 1.286, p=0.002), whereas average strength, global efficiency, local efficiency, mean clustering coefficient, and transitivity were decreased in patients with PSE than in those without PSE (49.955 vs. 55.055, p=0.001; 0.756 vs. 0.827, p=0.001; 4.795 vs. 5.741, p=0.002; 0.727 vs. 0.810, p=0.001; 1.091 vs. 1.215, p=0.001). However, in the delta, theta, and alpha bands, none of the functional connectivity measures were different between groups.

Conclusion: We demonstrated significant alterations of functional connectivity in patients with PSE, who have decreased segregation and integration in brain network, compared to those without PSE.

epilepsy

ischemic stroke

electroencephalography

Newly diagnosed epilepsy has various etiologies, including idiopathic, congenital, head trauma, central nervous system infection, brain tumor, neurodegenerative disease, or cerebrovascular diseases. (Hauser et al. 1993; Pitkanen et al. 2016) Of these etiologies, cerebrovascular diseases account for approximately 11% of all cases of epilepsy, and are the most common etiologies of epilepsy in elderly patients. (Hauser et al. 1993; Pitkanen et al. 2016) Moreover, depending on the type of underlying cerebrovascular disease, between 3% and 30% of patients with stroke develop post-stroke epilepsy (PSE). (Pitkanen et al. 2016) Risk factors for PSE in patients with stroke are cortical involvement, presence of hemorrhagic transformation, early post-stroke seizures, young age at stroke onset, stroke severity as indicated by a high National Institute of Health Stroke Scale (NIHSS) score, and alcoholism. (Gasparini et al. 2020; Zhang et al. 2014)

Network science and graph theory may contribute to a better understanding of brain function. The graph metrics from the graph theory demonstrate the network’s integration and segregation properties. (Falsaperla et al. 2021) Given its ability to provide information about global and local brain organization levels, the graph theory has demonstrated promising results in various neurological diseases. In the past decade, network science and graph theory have been applied to several researches in patients with epilepsy, and they have revealed significant alterations of brain network topology in these patients compared to healthy controls. (Lee et al. 2020; Lee et al. 2022; Park et al. 2019; Shim et al. 2020) Various modalities can be used when studying brain network and connectivity using graph theory. (Farahani et al. 2019) Connectivity can be largely divided into structural and functional connectivity. For structural connectivity, diffusion tensor imaging is the most representative method, while for functional connectivity, resting-state functional magnetic resonance imaging (MRI), magnetoencephalography (MEG), or electroencephalography (EEG) are commonly used. (Farahani et al. 2019) Although brain MRI can reveal both structure and function, EEG provides complementary information on the pathological brain area much faster and has a relatively simpler assessment than neuroimaging modality. (Salvador et al. 2005) Furthermore, despite its arguably superior signal-to-noise ratio, MEG remains scarce in clinical settings around the world, and EEG continues to dominate in epilepsy diagnosis. Therefore, EEG can be easily obtained in patients with epilepsy, and it can be used well for researches.

When performing functional connectivity analysis using EEG, it can be broadly divided into sensor-level or source-level analysis. Sensor-level network analysis does not permit neuroanatomical inferences because there is no mapping between EEG channels and the underlying sources, and volume conduction can lead to erroneous estimates of functional connectivity between channels. (Brunner et al. 2016; Lai et al. 2018a; Yu et al. 2022) In addition, numerous studies have shown that meaningful network patterns are not derived directly from scalp EEG channel activities, but rather from cortical source activities, due to the reduced volume conduction in the source space. (Brunner et al. 2016; Lai et al. 2018a; Yu et al. 2022) Consequently, functional connectivity analysis based on source-level EEG provides significant insights into the neuro-mechanisms underlying PSE. However, there have been no studies to investigate the alterations of functional connectivity in patients with PSE based on the source-level EEG analysis.

In this study, we investigated the differences of the functional connectivity based on the source-level EEG analysis between the stroke patients with and without PSE. We hypothesized that there were significant alterations of functional connectivity in the patients with PSE. Most of the prior studies have shown an increased local brain network, while a decreased global brain network in patients with epilepsy compared to healthy controls. (Bernhardt et al. 2015) Thus, we expected that the patients with PSE would reveal similar results to those of the previous studies compared to those without PSE.

2.1. Participants

This study was approved by the Institutional Review Board at our hospital. We prospectively enrolled 30 stroke patients with PSE, and 35 stroke patients without PSE. We defined the patients with PSE, who had at least one unprovoked late post-stroke seizure. The temporal limit to consider late post-stroke seizure was between one week and two years after stroke onset. (Sarecka-Hujar and Kopyta 2019; Zelano et al. 2020) They were newly diagnosed epilepsy patients at our hospital and they didn’t have any seizures prior to stroke onset. The patients without PSE had also not been previously diagnosed with epilepsy or any seizures prior to stroke onset. None of the stroke patients with and without PSE had any history of psychiatric disorders, developmental disorders, or any other debilitating diseases.

We obtained the clinical characteristics in these patients, such as age at the time of EEG, sex, stroke etiology according to Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification (cardioembolism, large-artery atherosclerosis, small-vessel occlusion, other determined etiology, or undetermined etiology), (Ay et al. 2005) stroke side and location, initial NIHSS score, (Adams et al. 1993; Brott et al. 1989) presence of hemorrhagic transformation, comorbidity (hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, or others), time interval between stroke onset and EEG taken, and presence of focal to bilateral tonic-clonic seizures.

2.2. EEG acquisition

EEG was conducted on all of the patients with stroke during a resting state period with their eyes closed, but awake. All EEG recordings were performed using the same types of EEG machines (TWin® EEG software system) and methods. The recordings were performed by trained personnel using gold electrodes attached with electrode paste. The 23 electrodes (Fp1, Fp2, F7, F8, T1, T2, T3, T4, T5, T6, O1, O2, F3, F4, C3, C4, P3, P4, Cz, Pz, Oz, A1, and A2) were placed using the international 10–20 system. The impedance of the electrodes was maintained below 5kΩ. The sampling frequency was 250 Hz, and the EEG recordings lasted for at least 30 minutes.

2.3. EEG preprocessing and source modeling

The data were manually reviewed, and five second epochs were used, which showed the presence of alpha activity with voltage maximum in posterior regions without any artifacts or epileptiform discharges using a EEGLAB program. (Delorme and Makeig 2004) Then, we used a Brainstorm program (Tadel et al. 2011) for source estimation and making a connectivity matrix. We computed sources based on their scalp electrical potentials with a minimum norm imaging method, which estimated the amplitude of brain sources distributed across the brain, and the Standardized LOw REsolution brain Electromagnetic TomogrAphy (sLORETA) approach. To create a functional connectivity matrix, a Desikan-Killiany atlas was applied to define the nodes (Desikan et al. 2006) (cortical 68 regions of interest), and phase locking value (Lachaux et al. 1999) was used to measure brain synchronization as edges. A weighted connectivity matrix was built for each group (a size of 68 ⅹ 68). Data were processed according to EEG frequency bands, including alpha (8–12 Hz), beta (13–29 Hz), theta (5–7 Hz), delta (1–4 Hz), low gamma (above 30–59 Hz), and high gamma bands (above 60–90 Hz).

2.4. Graph theoretical analysis

We used a BRAPH program (Mijalkov et al. 2017) to apply graph theoretical analysis. From the connectivity matrix of the each bands, we obtained the functional connectivity measures; including average strength, radius, diameter, eccentricity, characteristic path length, global efficiency, local efficiency, mean clustering coefficient, transitivity, and small-worldness index, (Farahani et al. 2019) in two groups: stroke patients with and without PSE. We compared these functional connectivity measures between the groups.

2.5. Statistical analysis

We compared the clinical characteristics between the patients with and without PSE using Chi-squared tests or independent Student t-test. All of the statistical tests were performed using MedCalc® Statistical Software version 20.014 (MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org; 2021). When comparing functional connectivity, we tested the statistical significance of the differences using nonparametric permutation tests with 1,000 permutations. A p-value of less than 0.05 was considered to indicate statistical significance for all calculations. When we investigated the functional connectivity, Bonferroni correction was applied to correct multiple comparisons (p = 0.005, 0.05/10 functional connectivity measures). Categorical variables were presented in terms of both frequency and percentage. Numerical variables are presented as the mean ± standard deviation (SD).

3.1. Clinical characteristics

Table 1 shows the differences of clinical characteristics between patients with and without PSE. There were no significant differences of the clinical characteristics, including age, sex, stroke etiology, stroke side and location, NIHSS, presence of hemorrhagic transformation, and comorbidities. However, the time interval between stroke onset and EEG taken was different between the groups. The time interval in the patients with PSE was longer than that in those without PSE (68.8 vs. 5.0 months, p < 0.001).

Table 1

The differences of the clinical characteristics between patients with and without post-stroke epilepsy
	Patients with PSE (N = 30)	Patients without PSE (N = 35)	p-value
Age, years (± SD)	74.4 (± 11.2)	75.0 (± 13.8)	0.829
Male, N (%)	19 (63.3)	16 (45.7)	0.158
Stroke etiology, N (%)			0.266
Cardioembolism	12 (40.0)	16 (45.7)
Large-artery atherosclerosis	3 (10.0)	6 (17.1)
Small-vessel occlusion	0 (0.0)	1 (2.9)
Other determined etiology	7 (23.3)	2 (5.7)
Undetermined etiology	8 (26.7)	10 (28.6)
Stroke side, N (%)			0.579
Right	12 (40.0)	12 (34.3)
Left	12 (40.0)	12 (34.3)
Bilateral	6 (20.0)	11 (31.4)
Stroke location, N (%)			0.336
Frontal lobe	14 (46.7)	19 (54.3)
Temporal lobe	16 (53.3)	17 (48.6)
Parietal lobe	14 (46.7)	14 (40.0)
Occipital lobe	2 (6.7)	7 (20.0)
Subcortex or brainstem	7 (23.3)	17 (48.6)
Cerebellum	3 (10.0)	9 (25.7)
NIHSS score (± SD)	9.7 (± 5.7)	11.7 (± 7.9)	0.397
Hemorrhagic transformation, N (%)	9 (30.0)	10 (28.6)	0.979
Comorbidity, N (%)			0.928
Hypertension	16 (53.3)	24 (68.6)
Diabetes	7 (23.3)	13 (37.1)
Dyslipidemia	3 (10.0)	5 (14.3)
Atrial fibrillation	9 (30.0)	14 (40.0)
Others	6 (20.0)	15 (42.9)
Time interval between stroke onset and EEG taken, months (± SD)	68.8 (± 83.5)	5.0 (± 7.2)	< 0.001
With focal to bilateral tonic-clonic seizures, N (%)	21 (70.0)
PSE: post-stroke epilepsy, SD: standard deviation, NIHSS: national institute of health stroke scale, EEG: electroencephalography

3.2. Functional brain connectivity

Table 2 reveals the differences of functional connectivity measures between patients with and without PSE according to EEG frequency bands. In the beta band, the radius and diameter were increased in the patients with PSE than those without PSE (2.699 vs. 2.579, p = 0.003; 2.261 vs. 2.171, p = 0.003; respectively). In the low gamma band, the radius was increased in the patients with PSE than those without PSE (2.808 vs. 2.617, p = 0.004). In the high gamma band, the radius, diameter, eccentricity, and characteristic path length were increased (1.828 vs. 1.559, p = 0.001; 2.653 vs. 2.306, p = 0.002; 2.212 vs. 1.913, p = 0.001; 1.425 vs. 1.286, p = 0.002; respectively), whereas the average strength, global efficiency, local efficiency, mean clustering coefficient, and transitivity were decreased in the patients with PSE than those without PSE (49.955 vs. 55.055, p = 0.001; 0.756 vs. 0.827, p = 0.001; 4.795 vs. 5.741, p = 0.002; 0.727 vs. 0.810, p = 0.001; 1.091 vs. 1.215, p = 0.001; respectively). However, in the delta, theta, and alpha bands, none of the functional connectivity measures, including average strength, radius, diameter, eccentricity, characteristic path length, global efficiency, local efficiency, mean clustering coefficient, transitivity, and small-worldness index, were different between the groups.

Table 2

The differences of functional connectivity measures between patients with and without post-stroke epilepsy according to EEG frequency bands
	Patients with PSE	Patients without PSE	Difference	CI lower	CI upper	p-value
Delta band
Average strength	58.176	58.667	0.491	-0.730	0.737	0.138
Radius	1.293	1.288	-0.005	-0.034	0.033	0.432
Diameter	2.008	1.949	-0.059	-0.077	0.082	0.115
Eccentricity	1.583	1.549	-0.034	-0.052	0.052	0.148
Characteristic path length	1.172	1.160	-0.012	-0.016	0.018	0.114
Global efficiency	0.868	0.876	0.007	-0.011	0.011	0.140
Local efficiency	6.216	6.317	0.101	-0.151	0.153	0.159
Mean clustering coefficient	0.864	0.872	0.008	-0.011	0.012	0.128
Transitivity	1.297	1.308	0.011	-0.016	0.018	0.138
Small-worldness index	1.000	1.000	0.000	0.000	0.000	0.316
Theta band
Average strength	57.784	58.354	0.571	-0.714	0.719	0.083
Radius	1.326	1.302	-0.024	-0.032	0.037	0.117
Diameter	2.014	1.998	-0.016	-0.074	0.074	0.374
Eccentricity	1.612	1.582	-0.031	-0.046	0.051	0.167
Characteristic path length	1.181	1.168	-0.014	-0.016	0.016	0.079
Global efficiency	0.863	0.871	0.009	-0.010	0.011	0.090
Local efficiency	6.136	6.253	0.117	-0.149	0.137	0.085
Mean clustering coefficient	0.858	0.867	0.009	-0.010	0.011	0.081
Transitivity	1.287	1.301	0.013	-0.015	0.016	0.080
Small-worldness index	1.000	1.000	0.000	0.000	0.000	0.196
Alpha band
Average strength	55.649	54.539	-1.109	-1.147	1.153	0.056
Radius	1.435	1.499	0.064	-0.067	0.067	0.058
Diameter	2.189	2.284	0.095	-0.114	0.116	0.093
Eccentricity	1.802	1.898	0.096	-0.096	0.103	0.061
Characteristic path length	1.245	1.276	0.031	-0.031	0.033	0.054
Global efficiency	0.831	0.814	-0.017	-0.017	0.017	0.066
Local efficiency	5.730	5.514	-0.217	-0.241	0.233	0.073
Mean clustering coefficient	0.824	0.806	-0.018	-0.019	0.020	0.054
Transitivity	1.236	1.209	-0.027	-0.029	0.027	0.069
Small-worldness index	0.999	0.999	-0.001	0.000	0.000	0.036
Beta band
Average strength	49.407	50.840	1.433	-1.077	1.049	0.012
Radius	1.840	1.732	-0.108	-0.078	0.077	0.012
Diameter	2.699	2.579	-0.120	-0.074	0.073	*0.003
Eccentricity	2.261	2.171	-0.090	-0.057	0.059	*0.003
Characteristic path length	1.441	1.397	-0.044	-0.032	0.035	0.012
Global efficiency	0.744	0.764	0.020	-0.016	0.014	0.010
Local efficiency	4.635	4.883	0.249	-0.167	0.178	0.007
Mean clustering coefficient	0.721	0.744	0.023	-0.016	0.017	0.011
Transitivity	1.081	1.116	0.035	-0.026	0.024	0.008
Small-worldness index	0.989	0.991	0.003	-0.002	0.003	0.044
Low gamma band
Average strength	48.045	50.047	2.002	-1.726	1.593	0.020
Radius	1.918	1.792	-0.125	-0.105	0.105	0.028
Diameter	2.808	2.617	-0.192	-0.135	0.123	*0.004
Eccentricity	2.317	2.198	-0.120	-0.096	0.098	0.018
Characteristic path length	1.483	1.423	-0.059	-0.048	0.048	0.024
Global efficiency	0.730	0.757	0.027	-0.022	0.022	0.016
Local efficiency	4.466	4.814	0.348	-0.287	0.259	0.017
Mean clustering coefficient	0.696	0.729	0.033	-0.027	0.028	0.025
Transitivity	1.044	1.093	0.049	-0.039	0.042	0.022
Small-worldness index	0.983	0.986	0.003	-0.004	0.004	0.116
High gamma band
Average strength	49.955	55.055	5.100	-2.599	2.490	*0.001
Radius	1.828	1.559	-0.269	-0.153	0.152	*0.001
Diameter	2.653	2.306	-0.347	-0.225	0.221	*0.002
Eccentricity	2.212	1.913	-0.298	-0.166	0.161	*0.001
Characteristic path length	1.425	1.286	-0.139	-0.078	0.074	*0.002
Global efficiency	0.756	0.827	0.071	-0.035	0.037	*0.001
Local efficiency	4.795	5.741	0.946	-0.474	0.509	*0.002
Mean clustering coefficient	0.727	0.810	0.083	-0.046	0.046	*0.001
Transitivity	1.091	1.215	0.125	-0.064	0.063	*0.001
Small-worldness index	0.987	0.992	0.006	-0.004	0.004	0.011
PSE: post-stroke epilepsy, CI: 95% confidence interval of the difference between the groups
*p < 0.005

To the best of our knowledge, this is the first study to investigate the brain functional connectivity in patients with PSE using source-level EEG analysis and graph theory. The main finding of this study was that there were significant alterations of functional connectivity in stroke patients with PSE compared to those without PSE. In addition, the differences of functional connectivity measures between patients with and without PSE were dependent on EEG frequency bands. There were no differences in the functional connectivity between the groups, especially in the delta, theta, and alpha bands. In the beta band, the radius and diameter were increased, and in the low gamma band, the radius was increased in patients with PSE compared to those without PSE. In the high gamma band, the radius, diameter, eccentricity, and characteristic path length were increased, whereas the average strength, global efficiency, local efficiency, mean clustering coefficient, and transitivity were decreased in the patients with PSE than in those without PSE.

Several cases have demonstrated functional connectivity alterations in patients with stroke. (Caliandro et al. 2017; Vecchio et al. 2019) Graph characteristics based on the EEG in 30 patients with acute stroke were investigated in a previous study. It found that brain network rearrangements, mainly decreased small-worldness, were detected in patients with stroke compared with healthy subjects. (Caliandro et al. 2017) Another study examined whether acute cerebellar stroke could cause alterations in brain network architecture as defined by cortical EEG rhythm. The authors demonstrated that the network remodeling characteristics were independent of the size of the ischemic lesion, and that cerebellar acute stroke induced significant changes in the balance of local segregation and global integration. (Vecchio et al. 2019) However, there was no research investigating the alterations of function connectivity in patients with PSE.

We demonstrated the alterations of functional connectivity in patients with PSE compared to stroke patients without PSE. The radius, diameter, eccentricity, and characteristic path length were increased, whereas the average strength, global efficiency, local efficiency, mean clustering coefficient, and transitivity were decreased in the patients with PSE than those without PSE, especially in the beta and gamma bands. The radius of a graph represents the minimum eccentricity of any graph vertex, while the diameter of a graph represents the maximum eccentricity of any graph vertex. (Collantoni et al. 2022; Falsaperla et al. 2021; Farahani et al. 2019; Mijalkov et al. 2017; Thomas et al. 2016; Zenil et al. 2018) The eccentricity is defined as the greatest distance between two vertices. The maximum distance between a vertex and every other vertex is regarded as the vertex's eccentricity. The average distance between pairs of vertices is known as the characteristic path length of the graph. The maximum number of edges that connect any two adjacent vertices in a multigraph. The global efficiency is a measure of the effectiveness of distant information transfer in a network and is defined as the inverse of the typical path length between all nodes. The local efficiency is a measure of the average efficiency of information transfer within local subgraphs or neighborhoods. It is defined as the inverse of the shortest average path length between all neighbors of a given node. The clustering coefficient quantifies the degree to which graph nodes tend to cluster together. The transitivity is the probability that adjacent network nodes are interconnected, revealing the existence of densely interconnected communities. (Collantoni et al. 2022; Falsaperla et al. 2021; Farahani et al. 2019; Mijalkov et al. 2017; Thomas et al. 2016; Zenil et al. 2018) Therefore, all these results suggest that functional segregation and integration are decreased in patients with PSE compared to those without PSE. Our finding in the present study suggests that post-stroke epileptogenesis is associated with alterations of the functional brain connectivity. This is consistent with a previous study with a rat model demonstrating that post-stroke epileptogenesis was related with a rearrangement of the intrinsic excitability across neurons. (Vera and Lippmann 2021)

Interestingly, the difference in functional connectivity between the two groups varied according to the EEG frequency bands. The differences of functional connectivity between the groups were revealed, especially in the beta and gamma bands. In the delta, theta, and alpha bands, all of the functional connectivity measures were not different between the groups. Brain oscillation has distinct function at different frequencies in various neural processing such as perceptual, sensorimotor and cognitive function. (Wang et al. 2017) Low frequency oscillations (delta, theta and alpha frequency bands) are associated with the integration across wide spatially large-scale networks, while high frequency oscillations (beta and gamma frequency bands) are linked to a more limited topographic area. (Wang et al. 2017) Beta and gamma frequency oscillations are known to occur in the inhibitory gamma-aminobutyric acid (GABA) interneuron network, and they are verified to relate to cognitive and executive functions. (Missonnier et al. 2010; Rossiter et al. 2014; Wang et al. 2017) The GABAergic interneurons receive excitatory inputs through N-methyl-D-aspartate (NMDA)- receptors and dysfunction of NMDA neurotransmission may result in abnormalities of gamma oscillation. (González-Burgos and Lewis 2012) The alterations of brain functional connectivity in patients with PSE in beta and gamma frequency bands might reflect the imbalance of GABAergic interneuron network induced by NMDA-receptor over-activation after stroke. (Nicolo et al. 2019)

Since the process of information in brain is encoded at a very short time scale (milliseconds to seconds), EEG is known as a good tool for evaluating the resting function with good time resolution. (Van Diessen et al. 2015) In addition, EEG is more advantageous than functional MRI in demonstrating mechanism of signal process among neuronal assemblies which do not modify energy consumption such as synchronization, coherence or phase locking-unlocking without any change of firing frequency, which reflect resting function of brain. (Rossini et al. 2019) However, there are some problems when analyzing sensor-level in performing connectivity analysis with EEG: field spread and volume conduction. (Hatlestad-Hall et al. 2021; Van Diessen et al. 2015) In addition, it is difficult to ensure resolution in terms of neuroanatomy in sensor level EEG analysis, as access to the source-level is required. (Lai et al. 2018b) Several solutions including Minimum Norm(MN), Dynamical Statistical Parametric Mapping(dSPM), Standard Maximum Entropy on the Mean(cMEM), Local AutoRegressive Average(LAURA) and LORETA, are introduced to solve the inverse problem that occurs during the reconstruction of EEG from the sensor-level to the source-level analysis. (Michel and Brunet 2019) These source reconstruction solutions assume the concept of a lead field that each signal of the scalp electrodes is produced by a source of unitary strength. (Van Diessen et al. 2015)

In the present study, we chose the combination of sLORETA for inverse solution and phase locking value for connectivity measure to analyze source-level EEG connectivity. In a recent study comparing source-level functional connectivity analysis and classical source localization in patients with epilepsy, source-level functional connectivity analysis showed high similarity to real interictal epileptic networks. (Hassan et al. 2017) In addition, it is important to select the optimal combination of inverse solution and connectivity measure. (Rizkallah et al. 2019) As a feature of the epileptic network, functionally connected brain regions have hyper-synchronization phenomena, and phase locking value is one of the viable methods because it has the advantage of nonlinearity network analysis. (Hatlestad-Hall et al. 2021; Rizkallah et al. 2019; Van Diessen et al. 2015)

This study has some limitations. First, the sample size was relatively small, and the study was conducted at a single center. Second, this study was a cross-sectional study. It is necessary to confirm whether there is a longitudinal difference in functional connectivity in patients with or without PSE, respectively. Third, time interval between stroke onset and EEG taken was different between the two groups. EEG were performed after more than one late-onset seizure was identified in patients with PSE, whereas in most of the patients without PSE, EEG were performed within a week after the stroke diagnosis. However, the effects of anti-seizure medications on EEG could be excluded because EEG were performed before anti-seizure medications were administered. Lastly, we only used 23 electrodes were placed using the international 10–20 system. In general, high density EEG is advantageous considering the forward problem from source to sensor. (Van Diessen et al. 2015)

In this study, we firstly investigate the brain functional connectivity in patients with PSE using source-level EEG analysis and graph theory. We demonstrate that there are significant alterations of functional connectivity in stroke patients with PSE, who have decreased segregation and integration in brain network, compared to those without PSE. These functional network changes could be related with epileptogenesis in PSE.

Funding

None.

Conflicts of interest

Neither of the authors has any conflict of interest to disclose.

Availability of data and material

Data that support the findings of this study are available upon reasonable request

Ethics approval and Consent

This retrospective chart review study involving human participants was in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The Human Investigation Committee (IRB) of Haeundae Paik hospital approved this study.

Code availability

Not applicable

Authors’ contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Dong Ah Lee, Taeik Jang, Jaeho Kang, Seongho Park, and Kang Min Park^.The first draft of the manuscript was written by Kang Min Park and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Acknowledgments

This work was supported by the Ministry of Science and ICT of the Republic of Korea (NRF-2021R1F1A1049605).

Adams HP, Jr., Bendixen BH, Kappelle LJ, Biller J, Love BB, Gordon DL, Marsh EE, 3rd (1993) Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment Stroke 24:35–41 doi:10.1161/01.str.24.1.35
Ay H, Furie KL, Singhal A, Smith WS, Sorensen AG, Koroshetz WJ (2005) An evidence-based causative classification system for acute ischemic stroke Ann Neurol 58:688–697 doi:10.1002/ana.20617
Bernhardt BC, Bonilha L, Gross DW (2015) Network analysis for a network disorder: the emerging role of graph theory in the study of epilepsy Epilepsy & Behavior 50:162–170
Brott T et al. (1989) Measurements of acute cerebral infarction: a clinical examination scale Stroke 20:864–870 doi:10.1161/01.str.20.7.864
Brunner C, Billinger M, Seeber M, Mullen TR, Makeig S (2016) Volume Conduction Influences Scalp-Based Connectivity Estimates Front Comput Neurosci 10:121 doi:10.3389/fncom.2016.00121
Caliandro P et al. (2017) Small-World Characteristics of Cortical Connectivity Changes in Acute Stroke Neurorehabil Neural Repair 31:81–94 doi:10.1177/1545968316662525
Collantoni E, Alberti F, Meregalli V, Meneguzzo P, Tenconi E, Favaro A (2022) Brain networks in eating disorders: a systematic review of graph theory studies Eat Weight Disord 27:69–83 doi:10.1007/s40519-021-01172-x
Delorme A, Makeig S (2004) EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis J Neurosci Methods 134:9–21 doi:10.1016/j.jneumeth.2003.10.009
Desikan RS et al. (2006) An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest Neuroimage 31:968–980 doi:10.1016/j.neuroimage.2006.01.021
Falsaperla R et al. (2021) Graph theory in paediatric epilepsy: A systematic review Dialogues Clin Neurosci 23:3–13 doi:10.1080/19585969.2022.2043128
Farahani FV, Karwowski W, Lighthall NR (2019) Application of Graph Theory for Identifying Connectivity Patterns in Human Brain Networks: A Systematic Review Front Neurosci 13:585 doi:10.3389/fnins.2019.00585
Gasparini S et al. (2020) Younger age at stroke onset but not thrombolytic treatment predicts poststroke epilepsy: An updated meta-analysis Epilepsy Behav 104:106540 doi:10.1016/j.yebeh.2019.106540
González-Burgos G, Lewis DA (2012) NMDA receptor hypofunction, parvalbumin-positive neurons, and cortical gamma oscillations in schizophrenia Schizophrenia bulletin 38 5:950–957
Hassan M, Merlet I, Mheich A, Kabbara A, Biraben A, Nica A, Wendling F (2017) Identification of interictal epileptic networks from dense-EEG Brain topography 30:60–76
Hatlestad-Hall C et al. (2021) Source-level EEG and graph theory reveal widespread functional network alterations in focal epilepsy Clinical Neurophysiology 132:1663–1676
Hauser WA, Annegers JF, Kurland LT (1993) Incidence of epilepsy and unprovoked seizures in Rochester, Minnesota: 1935–1984 Epilepsia 34:453–468 doi:10.1111/j.1528-1157.1993.tb02586.x
Lachaux JP, Rodriguez E, Martinerie J, Varela FJ (1999) Measuring phase synchrony in brain signals Hum Brain Mapp 8:194–208 doi:10.1002/(sici)1097-0193(1999)8:4<194::aid-hbm4>3.0.co;2-c
Lai M, Demuru M, Hillebrand A, Fraschini M (2018a) A comparison between scalp- and source-reconstructed EEG networks Sci Rep 8:12269 doi:10.1038/s41598-018-30869-w
Lai M, Demuru M, Hillebrand A, Fraschini M (2018b) A comparison between scalp-and source-reconstructed EEG networks Scientific reports 8:1–8
Lee DA, Kim BJ, Lee HJ, Kim SE, Park KM (2020) Network characteristics of genetic generalized epilepsy: Are the syndromes distinct? Seizure 82:91–98 doi:10.1016/j.seizure.2020.09.022
Lee DA, Lee HJ, Kim HC, Park KM (2022) Network differences based on arterial spin labeling related to anti-seizure medication response in focal epilepsy Neuroradiology 64:313–321 doi:10.1007/s00234-021-02741-8
Michel CM, Brunet D (2019) EEG source imaging: a practical review of the analysis steps Frontiers in neurology 10:325
Mijalkov M, Kakaei E, Pereira JB, Westman E, Volpe G, Alzheimer's Disease Neuroimaging I (2017) BRAPH: A graph theory software for the analysis of brain connectivity PLoS One 12:e0178798 doi:10.1371/journal.pone.0178798
Missonnier P, Herrmann FR, Michon A, Fazio-Costa L, Gold G, Giannakopoulos P (2010) Early disturbances of gamma band dynamics in mild cognitive impairment Journal of Neural Transmission 117:489–498
Nicolo J-P, O'Brien TJ, Kwan P (2019) Role of cerebral glutamate in post-stroke epileptogenesis NeuroImage: Clinical 24:102069
Park KM, Lee BI, Shin KJ, Ha SY, Park J, Kim SE, Kim SE (2019) Pivotal Role of Subcortical Structures as a Network Hub in Focal Epilepsy: Evidence from Graph Theoretical Analysis Based on Diffusion-Tensor Imaging J Clin Neurol 15:68–76 doi:10.3988/jcn.2019.15.1.68
Pitkanen A, Roivainen R, Lukasiuk K (2016) Development of epilepsy after ischaemic stroke Lancet Neurol 15:185–197 doi:10.1016/S1474-4422(15)00248-3
Rizkallah J et al. (2019) Decreased integration of EEG source-space networks in disorders of consciousness NeuroImage: Clinical 23:101841
Rossini P et al. (2019) Methods for analysis of brain connectivity: An IFCN-sponsored review Clinical Neurophysiology 130:1833–1858
Rossiter HE, Davis EM, Clark EV, Boudrias M-H, Ward NS (2014) Beta oscillations reflect changes in motor cortex inhibition in healthy ageing Neuroimage 91:360–365
Salvador R, Suckling J, Coleman MR, Pickard JD, Menon D, Bullmore E (2005) Neurophysiological architecture of functional magnetic resonance images of human brain Cereb Cortex 15:1332–1342 doi:10.1093/cercor/bhi016
Sarecka-Hujar B, Kopyta I (2019) Poststroke epilepsy: current perspectives on diagnosis and treatment Neuropsychiatr Dis Treat 15:95–103 doi:10.2147/NDT.S169579
Shim HK, Lee HJ, Kim SE, Lee BI, Park S, Park KM (2020) Alterations in the metabolic networks of temporal lobe epilepsy patients: A graph theoretical analysis using FDG-PET Neuroimage Clin 27:102349 doi:10.1016/j.nicl.2020.102349
Tadel F, Baillet S, Mosher JC, Pantazis D, Leahy RM (2011) Brainstorm: a user-friendly application for MEG/EEG analysis Comput Intell Neurosci 2011:879716 doi:10.1155/2011/879716
Thomas J, Seo D, Sael L (2016) Review on Graph Clustering and Subgraph Similarity Based Analysis of Neurological Disorders Int J Mol Sci 17 doi:10.3390/ijms17060862
Van Diessen E et al. (2015) Opportunities and methodological challenges in EEG and MEG resting state functional brain network research Clinical Neurophysiology 126:1468–1481
Vecchio F et al. (2019) Acute cerebellar stroke and middle cerebral artery stroke exert distinctive modifications on functional cortical connectivity: A comparative study via EEG graph theory Clin Neurophysiol 130:997–1007 doi:10.1016/j.clinph.2019.03.017
Vera J, Lippmann K (2021) Post-stroke epileptogenesis is associated with altered intrinsic properties of hippocampal pyramidal neurons leading to increased theta resonance Neurobiol Dis 156:105425 doi:10.1016/j.nbd.2021.105425
Wang J, Fang Y, Wang X, Yang H, Yu X, Wang H (2017) Enhanced gamma activity and cross-frequency interaction of resting-state electroencephalographic oscillations in patients with Alzheimer’s disease Frontiers in aging neuroscience 9:243
Yu M, Xiao S, Tian F, Li Y (2022) Frontal-occipital network alterations while viewing 2D & 3D movies: a source-level EEG and graph theory approach Biomed Tech (Berl) 67:161–172 doi:10.1515/bmt-2021-0300
Zelano J, Holtkamp M, Agarwal N, Lattanzi S, Trinka E, Brigo F (2020) How to diagnose and treat post-stroke seizures and epilepsy Epileptic Disord 22:252–263 doi:10.1684/epd.2020.1159
Zenil H, Kiani NA, Tegner J (2018) A Review of Graph and Network Complexity from an Algorithmic Information Perspective Entropy (Basel) 20 doi:10.3390/e20080551
Zhang C et al. (2014) Risk factors for post-stroke seizures: a systematic review and meta-analysis Epilepsy Res 108:1806–1816 doi:10.1016/j.eplepsyres.2014.09.030

No competing interests reported.

Download PDF

Journal Publication

published 16 Apr, 2024

Read the published version in Brain Topography →

Editorial decision: Major revision
04 Apr, 2023
Reviewers agreed at journal
27 Feb, 2023
Reviewers invited by journal
16 Jan, 2023
Editor assigned by journal
03 Jan, 2023
Submission checks completed at journal
03 Jan, 2023
First submitted to journal
03 Jan, 2023

You are reading this latest preprint version

Functional connectivity alterations in patients with post-stroke epilepsy based on source-level EEG and graph theory

Status:

Journal Publication

Version 1

Abstract

1. Introduction

2. Methods

2.1. Participants

2.2. EEG acquisition

2.3. EEG preprocessing and source modeling

2.4. Graph theoretical analysis

2.5. Statistical analysis

3. Results

3.1. Clinical characteristics

3.2. Functional brain connectivity

4. Discussion

5. Conclusion

Declarations

References

Additional Declarations

Status:

Journal Publication

Version 1