Abstract
Amyotrophic Lateral Sclerosis (ALS) is one of the most severe neurodegenerative diseases, which is known to affect upper and lower motor neurons. In contrast to the classical tenet that ALS represents the outcome of extensive and progressive impairment of a fixed set of motor connections, recent neuroimaging findings suggest that the disease spreads along vast non-motor connections. Here, we hypothesised that functional network topology is perturbed in ALS, and that this reorganization is associated with disability. We tested this hypothesis in 21 patients affected by ALS at several stages of impairment using resting-state electroencephalography (EEG) and compared the results to 16 age-matched healthy controls. We estimated functional connectivity using the Phase Lag Index (PLI), and characterized the network topology using the minimum spanning tree (MST). We found a significant difference between groups in terms of MST dissimilarity and MST leaf fraction in the beta band. Moreover, some MST parameters (leaf, hierarchy and kappa) significantly correlated with disability. These findings suggest that the topology of resting-state functional networks in ALS is affected by the disease in relation to disability. EEG network analysis may be of help in monitoring and evaluating the clinical status of ALS patients.
Similar content being viewed by others
Introduction
Amyotrophic Lateral Sclerosis (ALS) is one of the most severe neurodegenerative diseases, affecting the upper and lower motor neurons. All motor functions are progressively invalidated, and with a median survival of about 3 years from the onset of symptoms1. However, in contrast to the classical tenet that ALS represents the outcome of extensive and progressive impairment of a fixed set of motor connections, recent neuroimaging findings suggest that the disease spreads along vast non-motor connections. Indeed, advanced neuroimaging techniques, which allow for the non-invasive investigation of structural and functional brain organization, have so far introduced new opportunities for the study of ALS and are currently supporting the multi-systemic pathophysiology of this disease2,3.
Recently, modern network science has aided in the understanding of the human brain as a complex system of interacting units4,5. Indeed, the organization of brain networks can be characterised by means of several metrics that allow to estimate functional integration and segregation, quantify centrality of brain regions, and test resilience to insult6. Moreover, changes in network topology have been described for a range of neurological and psychiatric disorders7,5. In this view, structural and functional network studies based on diffusion tensor imaging (DTI) and functional magnetic resonance (fMRI) have contributed in elucidating basic mechanisms related to ALS onset, spread and progression.
For instance, Verstraete et al.8 observed structural motor network degeneration and suggested a spread of disease along functional connections of the motor network. Moreover, the same group has also reported9 an increasing loss of network structure in patients with ALS, with the network of impaired connectivity expanding over time. Schmidt et al.10, have recently shown that structural and functional connectivity degeneration in ALS are coupled and that the pathogenic process strongly affects both structural and functional network organization. Other resting-state fMRI studies11,12,13 have reported alterations in specific resting-state networks. Recently, Iyer and colleagues14 have investigated the use of resting-state electroencephalographic (EEG) as a potential biomarker for ALS, suggesting that a pathologic disruption of the network can be observed in early stages of the disease. However, it still remains relevant to address methodological issues that may affect both connectivity estimation and network reconstruction15.
Although the results described above are promising, it is not yet clearly understood how whole-brain functional networks are perturbed in ALS patients, and how this relates to disability. Resting-state EEG analysis may represent a practical tool to evaluate and monitor the progression of the disease. Despite the wide use of EEG in the assessment of brain disorders5,16,17, it has not been used widely to evaluate functional network changes induced by ALS. To test our hypothesis, we reconstructed functional networks from resting-state EEG recordings in 21 ALS patients and 16 age-matched healthy controls using the phase lag index (PLI)18, a widely used and robust measure of phase synchronization that is relatively insensitive to the effects of volume conduction. The topologies of frequency specific minimum spanning trees (MSTs) were subsequently characterised and compared between groups as it has been shown19,20 that it avoids important methodological biases that would otherwise limit a meaningful comparison between the groups21. Moreover, a correlation analysis was performed between the MST parameters and disability.
Results and Discussion
Age-matching
No significant group differences were observed in age (W = 145.5, p = 0.499).
Functional Connectivity
No significant group differences were observed for the global mean PLI in any frequency band (both with and without FDR correction for number of frequency bands). Descriptive results and statistics are summarized in Table 1. No significant correlation was observed between the patients’ global mean PLI and the disability score for any frequency band (see Table 2).
MST dissimilarity
A significant MST dissimilarity between ALS patients and healthy controls was found in the beta band using Mann-Whitney U test (W = 68.00, p = 0.008) after FDR correction.
MST topology
A significant difference between groups was observed for MST leaf fraction in the beta band (W = 87.5, p = 0.014). Results from Mann-Whitney U test statistics are summarized in Table 3. Individual values for each MST parameter in the beta band are shown in Fig. 1. Significant correlations were observed between some MST parameters (leaf, hierarchy and kappa) and disability score in the beta band (scatterplots for individual MST parameters and disability scores are reported in Fig. 2). Plots and correlations at single epoch level are also reported (see Fig. 3).
Power analysis
No significant differences were observed in any frequency band.
Discussion
In summary, by applying the PLI and the MST analysis in EEG recordings, this study shows large-scale changes in the functional brain network organization in ALS patients as identified using MST dissimilarity. Post-hoc analysis revealed that this difference in network topology between patients and controls was due to a difference in leaf fraction, and that the patients’ network organization in terms of MST parameters significantly correlated with disability, which is of clinical relevance. These results were observed in the beta band (13–30 Hz), where the MST topology was characterized by a significantly lower leaf fraction in the patients. Together with the significant negative correlation between disability score and leaf fraction, as well as a positive correlation with the diameter (even though not significant), indicates the tendency to deviate from a more centralized (star-like topology) towards a more decentralized organization (line-like topology). The negative correlation between tree hierarchy and disability score suggests that there is a sub-optimal balance between hub overload and functional integration in the network. Moreover, the negative correlation between disability and kappa, a measure that captures the broadness of the degree distribution, reflects the detrimental effect of a network topology with a reduced ease of synchronization (i.e., decreased spread of information across the tree)22. Furthermore, it is also important to note that the correlation analysis is still valid when using data from every single epoch (see Fig. 3). This further analysis suggests that even when using MST parameters extracted from single epochs it is possible to characterise the association between network topology and patients disability score.
As hypothesized, these findings suggest that ALS alters the brain network topology, which thus tends to deviate from the normal, presumably optimal, organization, and suggest that the correlation between MST parameters and the ASLFRS-R scale maybe be useful in monitoring the progression of the disease. These findings indicate that also at macroscopic scale (as measured by EEG functional networks), in accordance with previous studies on structural and functional neuroimaging2, ALS seems to affect extramotor brain regions, a result that is in line with the idea that pathological perturbations are rarely confined to a single locus23.
Moreover, it is of interest to note that a similar shift towards a more decentralized topology has been previously observed in multiple sclerosis24 and Parkinson’s disease25, and that functional networks in epilepsy patients that respond to vagal nerve stimulation re-organize towards a more centralized topology26. Together, these findings suggest that there is a possible common pathway in neurological disorders, as has been hypothesised recently5. Group differences and significant correlations between network topology and disability were found in the beta band, which may not be surprising given its link with motor function27 and that changes in beta activity can occur with ageing, sensorimotor disorders and amyotrophic lateral sclerosis28,29.
It should also be noted that the observed differences in network topology are unlikely to be due to differences in beta-band power, since no significant power differences were observed between the groups. Even though this latest finding seems in contrast with a recent MEG study30 that reported an intensified cortical beta desynchronization in ALS patients, other studies have not reported the same finding28,31. Differences in protocol design, for example the use of different tasks, or resting-state data, or differences in characteristics of the patient population, may explain these differences across studies. From a theoretical point of view, it is also unlikely that changes in the topology of PLI-based MSTs are due to power changes, because the PLI is a phase-based measure that is not directly affected by power, and the MST is based on the ranking of connections, not on their absolute values. Noteworthy in this respect is that we also observed that the mean PLI was not different between the groups.
Despite the observed differences between healthy controls and ALS patients in terms of MST dissimilarity and leaf fraction, and clinically relevant correlations between disability and network topology, we found that the detection of distinctive EEG network properties still remains a difficult task during the early stages of the disease. This is in contrast with the study by Iyer and colleagues14, who used a set of connectivity metrics in combination with network analysis. In contrast to their work, we used different methods of functional connectivity (PLI) and network reconstruction (MST), as well as a conservative statistical approach. Indeed, the PLI has been shown to be an index of phase synchronization that is robust to biases introduced by volume conduction and field spread. Moreover, the MST represents a network approach that, although still providing network characteristics that can be related to conventional graph measures20, is not biased by common methodological issues arising when reconstructing and comparing networks using traditional approaches (i.e. the use of arbitrary thresholds)21.
Previous studies have shown a direct relation between disability and disease progression32. Interestingly, the observed correlation between network organization and disease disability suggests that it might be possible to track disease progression on the basis of EEG network analysis. However, a longitudinal study is needed to confirm this idea. The difference between groups in terms of overall MST topology (i.e. the MST dissimilarity results), in combination with the change in leaf fraction, suggests that ALS affects the brain networks at a global level. However, it could be that the networks were reconstructed at a level that was too coarse, and/or without enough anatomical precision. The study of source-reconstructed time-series would be of help in investigating the role of specific brain regions, and in identifying whether certain regions are more affected than others.
Despite the limits of EEG analysis, the proposed approach has several advantages over other techniques. In particular, in comparison with fMRI, EEG is a direct measure of neuronal activity, allows time-frequency analysis of this activity, is less affected by head motion, can more readily be used with challenging patient populations (such as ALS), and the hardware costs are significantly lower. Furthermore, the high temporal resolution enables the study of dynamic processes involved in brain diseases such as ALS. Even though MEG has comparable temporal and spectral resolutions, and typically a higher spatial resolution, the costs of installment and maintenance, as well as the limited number of installed systems, still represent important obstacles. EEG has therefore still an important role in the study of ALS.
In conclusion, this study shows that EEG functional network re-organization in ALS patients, as computed by the PLI and MST approach, is associated with the patient disability. This finding suggests that resting-state EEG networks analysis may play an important role in evaluating the status of ALS patients and monitoring disease progression.
Methods
Subjects
Twenty-one patients (7 female; mean age 66, standard deviation 9 years) diagnosed with ALS according to the revised El Escorial criteria33, who attended the ALS Centre of the Azienda Ospedaliera Universitaria of Cagliari (Italy), were included in the study. A control group, consisting of sixteen age- and gender- matched healthy subjects (9 female; mean age 65, standard deviation 7 years), was also included. Informed consent was obtained from all participants. Local ethics committee from University of Cagliari approved the study protocol (NP/2013/1496) and all procedures were carried out in accordance with the approved protocol. The clinical ALSFRS-R score32, a validated rating instrument for monitoring the progression of disability in patients with ALS, was evaluated at the time of EEG recording. This score was converted to a disability score by subtracting it from the maximum obtainable score, i.e. 48 - ALSFRS-R. That is, a disability score of 0 means that you are healthy.
Recordings
Five minutes EEG signals were recorded using a 61 EEG channels system (Brain QuickSystem, Micromed, Italy) during an eyes-closed resting-state condition. The reference electrode was placed in close proximity of the electrode POz. Signals were digitized with a sampling frequency of 256 Hz and offline re-referenced to the common average reference (excluding channels Fp1, AF3, AF7, Fp2, AF4 and AF8). For each subject the first four epochs (avoiding as possible contaminations from eye blinks, eye-movements, muscle activity, ECG, as well as systems- and environmental artifacts) of 2048 samples (8 s)34 were selected and band-pass filtered in the classical frequency bands: delta (1–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), and beta (13–30 Hz).
Functional connectivity
The phase lag index (PLI)18, which evaluates the asymmetry of the distribution of instantaneous phase differences between pairs of channels, was used to estimate functional connectivity (FC). Computing FC between all pair wise combinations of EEG time-series resulted in a weighted adjacency matrix of 58 × 58 entries (after excluding bad channels form both patients and healthy subjects) for each epoch. Mean PLI was also computed across epochs and channels.
MST reconstruction
The MST is an acyclic sub-network, which connects all nodes minimizing the link weights (for the computation of the MST, the link weight is defined as 1–PLI). The MST was obtained using Kruskal algorithm35. The topology of the MST was characterised using several measures: the diameter, the normalized leaf fraction, kappa (or degree divergence) and the tree hierarchy. The diameter, which is the largest distance between any two nodes, is defined as the longest shortest path in the network. The normalized leaf fraction is calculated as the number of nodes with degree of 1 divided by the total number of nodes, and is a measure of integration in the network19,20. It has been shown that both these MST descriptors are important network parameters during development36, that they capture network alterations in multiple sclerosis24, in Parkinson disease25, in epileptic patients implanted with VNS26 and in aging37. Kappa, or degree divergence, is a measure of the broadness of the degree distribution. It is related to resilience against attack, epidemic spreading and synchronizability20. Furthermore, it has been shown that Kappa is high in networks with a scale-free degree distribution4. The tree hierarchy captures the balance between a small diameter and overload of hub nodes. This MST parameter ranges from 0 to 1 and an optimal tree configuration is thought to correspond to hierarchy values around 0.5 (intermediate between a line-like and star-like topology, i.e. a compromise between hub-overload and weak integration)19.
MST dissimilarity
MST dissimilarity, which assesses the overlap between MSTs38, was estimated between MSTs of ALS patients and healthy controls. In this study the MST reconstructed from the average connectivity matrix of all healthy subjects was used as reference in order to compute MST dissimilarities for both patients and controls24.
Power analysis
Power analysis was performed using Welch’s averaged modified periodogram method. The power spectral density (PSD) was computed separately for each epoch and subject. The relative power for each frequency band was successively evaluated and averaged over the epochs.
Statistical analysis
Statistical differences in age, mean PLI, and MST dissimilarity between groups were evaluated using the non-parametric Mann-Whitney test. The value used for significance was set to p < 0.05 and a correction for multiple comparisons was performed by the false discovery rate (FDR), correcting for the four frequency bands39. In case we found significant MST dissimilarity, post-hoc analysis was performed to find out which MST parameters were different. Moreover, a Spearman’s rank correlation coefficient was computed to assess the relationship between the network topology (in terms of mean PLI and MST parameters) and disease severity (in terms of the disability score). Statistical analysis was performed using JASP (version 0.7.5 beta 2 for Mac OS X)40.
Additional Information
How to cite this article: Fraschini, M. et al. EEG functional network topology is associated with disability in patients with amyotrophic lateral sclerosis. Sci. Rep. 6, 38653; doi: 10.1038/srep38653 (2016).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
Del Aguila, M., Longstreth, W., McGuire, V., Koepsell, T. & Van Belle, G. Prognosis in amyotrophic lateral sclerosis a population-based study. Neurology 60, 813–819 (2003).
Chiò, A. et al. Neuroimaging in amyotrophic lateral sclerosis: insights into structural and functional changes. The Lancet Neurology 13, 1228–1240 (2014).
Verstraete, E. & Foerster, B. R. Neuroimaging as a new diagnostic modality in amyotrophic lateral sclerosis. Neurotherapeutics 12, 403–416 (2015).
Stam, C. v. & Van Straaten, E. The organization of physiological brain networks. Clinical neurophysiology 123, 1067–1087 (2012).
Stam, C. J. Modern network science of neurological disorders. Nature Reviews Neuroscience 15, 683–695 (2014).
Rubinov, M. & Sporns, O. Complex network measures of brain connectivity: uses and interpretations. Neuroimage 52, 1059–1069 (2010).
Fornito, A. & Harrison, B. J. Brain connectivity and mental illness. Psychiatry 23, 239–249 (2012).
Verstraete, E. et al. Motor network degeneration in amyotrophic lateral sclerosis: a structural and functional connectivity study. PloS one 5, e13664 (2010).
Verstraete, E., Veldink, J. H., den Berg, L. H. & den Heuvel, M. P. Structural brain network imaging shows expanding disconnection of the motor system in amyotrophic lateral sclerosis. Human brain mapping 35, 1351–1361 (2014).
Schmidt, R. et al. Correlation between structural and functional connectivity impairment in amyotrophic lateral sclerosis. Human brain mapping 35, 4386–4395 (2014).
Agosta, F. et al. Divergent brain network connectivity in amyotrophic lateral sclerosis. Neurobiology of aging 34, 419–427 (2013).
Luo, C. et al. Patterns of spontaneous brain activity in amyotrophic lateral sclerosis: a resting-state fmri study. PloS one 7, e45470 (2012).
Mohammadi, B. et al. Changes of resting state brain networks in amyotrophic lateral sclerosis. Experimental neurology 217, 147–153 (2009).
Iyer, P. M. et al. Functional connectivity changes in resting-state eeg as potential biomarker for amyotrophic lateral sclerosis. PloS one 10, e0128682 (2015).
Van Diessen, E. et al. Opportunities and methodological challenges in eeg and meg resting state functional brain network research. Clinical Neurophysiology 126, 1468–1481 (2015).
Diessen, E., Diederen, S. J., Braun, K. P., Jansen, F. E. & Stam, C. J. Functional and structural brain networks in epilepsy: what have we learned? Epilepsia 54, 1855–1865 (2013).
van Straaten, E. C., Scheltens, P., Gouw, A. A. & Stam, C. J. Eyes-closed task-free electroencephalography in clinical trials for alzheimer’s disease: an emerging method based upon brain dynamics. Alzheimer’s research & therapy 6, 1 (2014).
Stam, C. J., Nolte, G. & Daffertshofer, A. Phase lag index: assessment of functional connectivity from multi channel eeg and meg with diminished bias from common sources. Human brain mapping 28, 1178–1193 (2007).
Stam, C. et al. The trees and the forest: characterization of complex brain networks with minimum spanning trees. International Journal of Psychophysiology 92, 129–138 (2014).
Tewarie, P., Van Dellen, E., Hillebrand, A. & Stam, C. The minimum spanning tree: an unbiased method for brain network analysis. Neuroimage 104, 177–188 (2015).
Van Wijk, B. C., Stam, C. J. & Daffertshofer, A. Comparing brain networks of different size and connectivity density using graph theory. PloS one 5, e13701 (2010).
Otte, W. M. et al. Aging alterations in whole-brain networks during adulthood mapped with the minimum spanning tree indices: The interplay of density, connectivity cost and life-time trajectory. NeuroImage 109, 171–189 (2015).
Fornito, A., Zalesky, A. & Breakspear, M. The connectomics of brain disorders. Nature Reviews Neuroscience 16, 159–172 (2015).
Tewarie, P. et al. Functional brain network analysis using minimum spanning trees in multiple sclerosis: an meg source-space study. Neuroimage 88, 308–318 (2014).
Dubbelink, K. T. O. et al. Disrupted brain network topology in parkinson’s disease: a longitudinal magnetoencephalography study. Brain awt316 (2013).
Fraschini, M. et al. The re-organization of functional brain networks in pharmaco-resistant epileptic patients who respond to vns. Neuroscience letters 580, 153–157 (2014).
Engel, A. K. & Fries, P. Beta-band oscillations—signalling the status quo? Current opinion in neurobiology 20, 156–165 (2010).
Bizovičar, N., Dreo, J., Koritnik, B. & Zidar, J. Decreased movement-related beta desynchronization and impaired post-movement beta rebound in amyotrophic lateral sclerosis. Clinical neurophysiology 125, 1689–1699 (2014).
Kasahara, T. et al. The correlation between motor impairments and event-related desynchronization during motor imagery in als patients. BMC neuroscience 13, 1 (2012).
Proudfoot, M. et al. Altered cortical beta-band oscillations reflect motor system degeneration in amyotrophic lateral sclerosis. Human Brain Mapping (2016).
Riva, N. et al. Cortical activation to voluntary movement in amyotrophic lateral sclerosis is related to corticospinal damage: electrophysiological evidence. Clinical Neurophysiology 123, 1586–1592 (2012).
Cedarbaum, J. M. et al. The alsfrs-r: a revised als functional rating scale that incorporates assessments of respiratory function. Journal of the neurological sciences 169, 13–21 (1999).
Brooks, B. R., Miller, R. G., Swash, M. & Munsat, T. L. El escorial revisited: revised criteria for the diagnosis of amyotrophic lateral sclerosis. Amyotrophic lateral sclerosis and other motor neuron disorders 1, 293–299 (2000).
Fraschini, M. et al. The effect of epoch length on estimated eeg functional connectivity and brain network organisation. Journal of neural engineering 13, 036015 (2016).
Kruskal, J. B. On the shortest spanning subtree of a graph and the traveling salesman problem. Proceedings of the American Mathematical society 7, 48–50 (1956).
Boersma, M. et al. Growing trees in child brains: graph theoretical analysis of electroencephalography-derived minimum spanning tree in 5-and 7-year-old children reflects brain maturation. Brain connectivity 3, 50–60 (2013).
Smit, D. J., de Geus, E. J., Boersma, M., Boomsma, D. I. & Stam, C. J. Life-span development of brain network integration assessed with phase lag index connectivity and minimum spanning tree graphs. Brain connectivity 6, 312–325 (2016).
Lee, U., Kim, S. & Jung, K.-Y. Classification of epilepsy types through global network analysis of scalp electroencephalograms. Physical Review E 73, 041920 (2006).
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the royal statistical society. Series B (Methodological) 289–300 (1995).
Love, J. et al. Jasp (version 0.7)[computer software]. Amsterdam, the netherlands: Jasp project (2015).
Acknowledgements
M. Fraschini and M. Demuru were in part supported by the Fondazione Banco di Sardegna (Prot.U7989.2013/AI.713.MGB. Prat.2013.1237).
Author information
Authors and Affiliations
Contributions
M.F., A. H., G.F., G.B. and F.M. conceived the study, S.P., F.D., M.P., M.D. and M.F. conducted the experiment, M.D., L.C. and M.F. analysed the data, A.H., M.F. and F.M. wrote the draft of the paper. All authors reviewed the manuscript.
Ethics declarations
Competing interests
The authors declare no competing financial interests.
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Fraschini, M., Demuru, M., Hillebrand, A. et al. EEG functional network topology is associated with disability in patients with amyotrophic lateral sclerosis. Sci Rep 6, 38653 (2016). https://doi.org/10.1038/srep38653
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/srep38653
This article is cited by
-
Decreased resting-state alpha-band activation and functional connectivity after sleep deprivation
Scientific Reports (2021)
-
Uncertainty in Functional Network Representations of Brain Activity of Alcoholic Patients
Brain Topography (2021)
-
EEG based emotion recognition using minimum spanning tree
Physical and Engineering Sciences in Medicine (2020)
-
A comparison between scalp- and source-reconstructed EEG networks
Scientific Reports (2018)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.