TL;DR
This study uses graph theory to analyze EEG-based brain networks in patients with psychogenic non-epileptic seizures (PNES). They found that the frequency of PNES attacks is linked to specific network properties, particularly a weakness in local brain connectivity in the alpha band and increased network resilience in the beta band. These findings suggest that network abnormalities may play a role in PNES and could potentially be targeted for therapeutic interventions.
What Problem Does This Paper Address?
Psychogenic non-epileptic seizures (PNES) are involuntary events that mimic epileptic seizures but lack the characteristic epileptiform EEG discharges. While psychological factors are known to play a role, there's a growing interest in identifying objective neurological markers that might predispose individuals to PNES or help differentiate them from true epilepsy. Previous research has hinted at altered functional connectivity in PNES patients, but a comprehensive analysis of brain network topology using graph theory was lacking. This paper aims to fill that gap by investigating whether the structural properties of functional brain networks, as measured by EEG, are associated with the occurrence and frequency of PNES.
Methods
The researchers employed a graph theoretical approach to analyze resting-state EEG data.
- Data Preprocessing: EEG signals were processed using a Laplacian transformation to enhance source activity and then filtered into different frequency bands (delta, theta, alpha, beta).
- Network Construction: Functional brain networks were constructed by calculating cross-correlations between electrode time-series. These weighted networks were then converted into binary networks by thresholding based on network density (the proportion of connections present). This was done across a range of densities (0.05 to 0.3) to ensure comparability.
- Graph Theory Metrics: Several key graph theory metrics were calculated for these networks:
- Clustering Coefficient (C): Measures local connectivity and the tendency for nodes to form tightly knit groups.
- Modularity (Q): Quantifies the extent to which a network can be divided into distinct modules or communities.
- Global Efficiency (E): Assesses the network's ability to integrate information across distant regions.
- Small-Worldness (SW): A composite metric that captures the balance between local clustering and global integration, characteristic of efficient brain networks.
- Assortativity (r): Measures the tendency of nodes to connect to other nodes with similar degrees, reflecting network resilience.
- Statistical Analysis: Differences between PNES patients and controls were assessed using Wilcoxon's ranksum test, with corrections for multiple comparisons (Benjamini-Hochberg). Correlations between network metrics and PNES frequency were analyzed using Spearman's rank correlation coefficient.
Datasets
The study utilized resting-state EEG data from two groups:
- PNES Patients: 13 individuals diagnosed with psychogenic non-epileptic seizures.
- Healthy Controls: 13 age- and sex-matched individuals.
All data were collected at the University of Lausanne. No new benchmarks were introduced; the focus was on analyzing existing EEG data with novel graph theory methods.
Key Insights and Ideas
- Frequency Matters: The most striking finding is the direct correlation between the frequency of PNES attacks and specific network properties, rather than just the presence of PNES itself. This suggests that network dynamics might influence seizure severity or likelihood.
- Alpha Band Weakness: In the alpha frequency band, a lower clustering coefficient (indicating weaker local connectivity) and a lower small-worldness metric were associated with a higher frequency of PNES attacks. This implies that impaired local processing might contribute to seizure occurrence.
- Beta Band Resilience: Conversely, in the beta band, higher assortativity (indicating greater network resilience) was linked to a higher frequency of PNES attacks. This suggests that a more rigid or resilient network structure might paradoxically make individuals more prone to seizures.
- Local Connectivity as a Target: The findings strongly suggest that local connectivity is a critical factor. The authors propose that strengthening local connectivity could be a therapeutic target to rebalance the network and potentially prevent PNES.
- Interictal Phenotype: The observed network differences during interictal periods (between seizures) suggest that these are not transient states but rather underlying network characteristics that may predispose individuals to PNES.
Key Results
- No Group Differences in Basic Metrics: Overall, PNES patients and controls did not show significant differences in basic network metrics like clustering coefficient, modularity, or global efficiency across all frequency bands and densities.
- Correlations with PNES Frequency:
- Alpha Band: Significant inverse correlations were found between PNES frequency and clustering coefficient (mean CC = -0.64) and small-worldness (mean CC = -0.68).
- Beta Band: Significant positive correlation was found between PNES frequency and assortativity (mean CC = 0.82).
- Hub Distribution: In the alpha band, PNES patients showed a reduced number of hubs in prefrontal regions and an increased number in posterior areas compared to controls, particularly at higher network densities.
Limitations and Future Work
- Small Sample Size: The study was conducted on a relatively small cohort of 13 PNES patients and 13 controls. The authors acknowledge that these findings require further validation in larger, more diverse patient groups.
- Heterogeneity: The PNES population is known to be heterogeneous, which could influence the observed network patterns.
- Trait vs. State: Future longitudinal studies are needed to determine whether these network features are stable traits or state-dependent and if they change with treatment.
- Therapeutic Implications: While the study suggests local connectivity as a therapeutic target, further research is needed to explore selective modulation strategies and their efficacy in preventing PNES.
- Comparison with Epilepsy: The authors highlight the need for direct comparisons between PNES and epilepsy network topologies to better understand condition-specific network alterations.
Summary updated