Up to 30% of stroke patients have abnormal EEG

Abstract

AimEEG is an accessible, sensitive and informative method for recording changes in the bioelectrical activity of the brain in routine neurological clinical practice in a number of acute and chronic neurological diseases. However, the method is not used as the main one in the diagnosis of ischemic stroke due to the lack of evidence from the studies carried out to date on correlations between clinical and imaging findings with those of the EEG recording, the method is considered non-informative and non-specific.MethodsWe conducted an epidemiological study on acute stroke patients in order to find early epileptic seizures with the establishment of EEG changes in 51 patients. We also grouped the results of qualitative EEG changes as abnormal and general non-specific, and divided them by patients’ normal EEG activity.ResultsBy analyzing the results, we can conclude that up to 30% of stroke patients have abnormal EEG changes, fully consistent with the clinical and imaging pathological findings, as well as readiness for the development of early seizures.ConclusionThis creates a need for prophylaxis with antiepileptic drugs. After enrollment in therapy, none of the patients who were monitored developed structural epilepsy in the subsequent subacute period. Therefore, it can be assumed that EEG is a method with a more prognostic role in stroke patients with early onset of epileptic seizures and bioelectrical changes. This information can be used to incorporate early antiepileptic prophylaxis and prevent the complications associated with structural epilepsy.

Keywords

Introduction

EEG (Electro-encephalography) is not widely used in the diagnosis of cerebral stroke, since this diagnosis is clinical, and EEG lacks specific signs of this disease. Nevertheless, in acute and chronic cerebral ischemia, the lack of oxygen and metabolites causes EEG changes of different nature and intensity.¹ EEG remains the main diagnostic method of epilepsy. A quantitative electroencephalogram (qEEG) is a test that analyzes the electrical activity of the brain to measure and display patterns that may correspond to diagnostic information and/or cognitive deficits. qEEG involves analyzing brain wave patterns to understand brain function. In stroke, qEEG has been explored to assess brain damage, predict outcomes, and guide rehabilitation strategies. Research suggests that qEEG can detect abnormalities in stroke patients, aiding in prognosis and treatment planning.² Regarding post-stroke epilepsy, studies have investigated qEEG’s role in predicting and diagnosing seizures. Monitoring brain activity through qEEG might help identify individuals at risk for developing epilepsy after a stroke and assist in early intervention.³ Pharmacological strategies for stroke and post-stroke epilepsy continue to evolve. Medications targeting seizure prevention in post-stroke epilepsy aim to minimize neuronal hyperexcitability. Drugs for stroke management focus on preventing secondary complications and promoting recovery. Recent reviews may delve deeper into these areas, potentially highlighting advancements in qEEG technology, its application in stroke-related conditions, and emerging pharmacological approaches.⁴AimThe aim is to conduct an epidemiological study of early epileptic seizures in the acute phase of stroke finding the EEG changes.

Materials and Methods

Prospectively acute stroke patients, who were treated in our clinic and met the inclusion and exclusion criteria (Table 1) were enrolled. Forty-three biological, clinical, laboratory and instrumental variables were evaluated (Table 2). Each patient underwent an instrumental Electroencephalography (EEG) examination on the second day of the onset of the stroke. EEG was done with Neuron-Spectrum-64 Neurosoft EEG/EP device, with the following characteristics: Channels: 25; ADC: 24 bit; Frequency range for EEG channels: 0 Hz-600 Hz; Sensitivity for EEG channels: 0.01-10000000 jaV/mm; Sampling frequency: 1000 Hz. The research is performed in an electrophysiological laboratory with preliminary preparation according to the approved standards and the international system 10/20 (Available at: https://www.mh.government.bg/media/filer_public/2015/11/18/nervni-bolesti.pdf). Applying a NOTCH filter was done using a specific type of filter designed to eliminate or reduce a particular frequency or frequencies, typically 50 or 60 Hz. These frequencies often correspond to electrical interference or noise originating from power sources like mains electricity. When a NOTCH filter is applied during EEG analysis, it indicates that the filtering process aimed to remove or significantly reduce the interference caused by these specific frequencies. This filtering helps to clean the EEG signal by attenuating or eliminating the noise from power lines or other environmental sources, allowing a clearer view of the brain’s electrical activity. The application of a NOTCH filter is a common practice in EEG analysis to improve the quality of the recorded brain signals by reducing unwanted external interference, ensuring that the focus remains on the brain’s electrical activity without contamination from environmental noise at these specific frequencies.⁵ A common standard for EEG recordings is to maintain electrode impedances below a certain threshold to ensure accurate and reliable signals. The impedance levels used for this study were within the range of 5 to 10 kΩ (kiloohms). Maintaining electrode impedances below this range helps to minimize noise, ensure good signal quality, and enhance the accuracy of the recorded brain activity.⁵
The data was analyzed manually by a neurophysiology researcher, who processed the recorded data without relying extensively on automated or computer-based algorithms. Analyzing EEG data manually involves visual inspection, in which the expert examines the raw EEG signals, looking for specific patterns, anomalies, or events of interest. This visual inspection might involve identifying characteristic waveforms (such as alpha, beta, theta, or delta waves), spikes, sharp waves, or other abnormalities. The event marking process includes noting and marking specific events or occurrences within the EEG recording, such as the onset of seizures, artifacts, or changes in brain activity. Artifact rejection is a very important step to be done by human to identify and exclude artifacts caused by external interference (movement, electrical noise, etc.) or physiological sources (muscle activity, eye blinks, etc.) from the analysis. Data segmentation was done by dividing the EEG recording into segments based on different experimental conditions, time frames, or specific activities for further analysis. Quantification and interpretation is the final step of the manual data analysis which involves measuring and quantifying specific EEG features or parameters manually, such as frequency bands, amplitudes, or durations of certain brain activities. This process might involve using rulers or digital tools to make precise measurements. Manual analysis of EEG data allows for a detailed and nuanced understanding of the recorded brain activity. It permits the expert to identify subtle patterns or abnormalities that automated algorithms might miss, especially in complex or atypical cases.⁶
To achieve the goal of the study, we determined and grouped the qualitative changes in the EEG, considering as abnormal only the focal or generalized epileptic graph elements such as sharps, spikes and sharp-slow complexes or focuses of theta and delta rhythms (Table 3). All other changes in the type of oscillations, phase amplitude coupling, phase to phase coupling, amplitude to amplitude coupling, cross-frequency phase coupling and inter-hemisphere bands, are defined as general non-specific changes. The normal EEG results are according to the reference values of the software analysis, reported by us earlier.⁷ Understanding the functional significance involves considering the implications. While general nonspecific changes and moderate disorganization in brain bioelectrical activity might not directly indicate epilepsy or focal seizure tendencies, they could reflect generalized disturbances in brain function. Conversely, epileptic graph elements like spikes, sharp waves, or specific rhythmic patterns often carry more diagnostic and prognostic significance, indicating a higher likelihood of epilepsy or focal brain abnormalities.⁸
Finally, each evaluated patient was counted in excel by determined type of EEG findings. Statistical processing is in accordance with the developed criteria.⁹ Although medication effects and acute stroke treatments were controlled and unified across the entire patient sample, the study may still encounter potential confounding factors and unaddressed influences on EEG patterns. While the standardized medication regimen and therapy offer a level of control, variations in individual responses to medications, potential interactions between drugs, or underlying comorbidities might introduce subtle yet impactful differences in EEG manifestations.¹⁰Ethical ApprovalThis study protocol was reviewed and approved by Local Ethics Committee of Trakia University - Stara Zagora city, Bulgaria (Date: 02.01.2020, Decision No:14). All patients voluntarily signed an informed consent form prior to inclusion in the study.

Results

The enrolled patients are 51. The men/women ratio is 25/26 with age from 41 to 88 (average of 67) with Gaussian-Laplace symmetric distribution shape. NIHSS score is from 2 to 20 (average 7 - moderate stroke) with right skewness exponential distribution shape. The provided information describes the distribution of cases based on EEG findings and suggests a statistically significant difference among the groups (Figure 1). The majority of patients, comprising 56.9% (29 patients), exhibited general nonspecific changes and moderate disorganization in the bioelectrical brain activity (2nd column). Following this group, 29.4% (15 patients) displayed abnormal EEG results (3rd column). The smallest proportion, 13.7% (7 patients), showed normal EEG values (1st column). The variable “Type of EEG” is characterized as a qualitative variable with three levels (normal, nonspecific changes with moderate disorganization, and abnormal results) and two degrees of freedom, indicating that it contains three categories and provides two contrasts within the groups. The statement suggesting a statistically significant difference between groups at the 0.05 significance level implies that, based on the analysis conducted, there is strong evidence to support that the distribution of patients among the three EEG categories is not occurring purely by chance. Instead, it suggests that there’s a likely association or difference among the groups regarding their EEG findings that is beyond random variation. This finding is statistically significant, meaning it’s likely not due to random fluctuations in the data but rather reflects a meaningful difference between the categories in terms of EEG patterns.
In examining the collective findings from the entire cohort of 15 patients exhibiting abnormal EEG results, a notable trend emerges in their combined clinical and laboratory profiles. It appears that the severity of the stroke, irrespective of the scale employed, emerges as a pivotal determinant in the onset of post-stroke epilepsy. Specifically, a heightened susceptibility to epileptogenesis seems linked to certain factors: infarction within the anterior cerebral artery territory, notably in the dominant cerebral hemisphere and in close proximity to the cortex; a lower age correlating with an increased risk; INR levels at or below 1.11; and cholesterol levels equal to or greater than 4 mmol/l.

Discussion

In this article, we reported our results related to patients diagnosed with stroke and their EEG changes in the post-stroke recovery period. EEG, as a predictor of post-stroke recovery, is currently largely based on the initial behavioral changes, although the recovery outcomes could differ significantly from the initial clinical similarities. A personalized and healing strategy pointed to the maximal efficient recovery.¹¹
In their 2017 study, Philips et al.¹² proposed an EEG beta band-based network model of biomarkers aimed at evaluating stroke rehabilitation. Current studies show that oscillation in the infra-frequency bands could be reliable for assessing brain abnormality after stroke. Cross-frequency coupling (CFC) presents new perspective to understanding brain activity after stroke. Many researchers performed tests related to phase amplitude coupling (PAC), phase to phase coupling (PPC), and amplitude to amplitude coupling (AAC).¹³ Some combined these with EMG signals. So far, it has been reported that CFC abnormalities were registered in the whole brain area scale (divided into low- and high-frequency and low- and low-frequency). According to Bin Ren et al.¹⁴ abnormal cross-frequency phase coupling could be highly related to the delta-band since delta band can reflect injury and recovery of neurons. According to Yan et al., 2013¹⁵ brain’s internal communication of the affected hemisphere and inter-hemisphere in inter-frequency bands (like beta) is also disturbed. The CFC analyses could further reveal that the information exchange in the non-lesion hemisphere is affected (which could not be seen in intra-frequency bands). CFC shows advances in understanding the impact of numerous neurological diseases such as Alzheimer’s disease,¹⁶ epilepsy and multiple sclerosis.¹⁷

Limitations

Potential limitations within the study: The study included a relatively small sample size of acute stroke patients (up to 51). While the results provide insights, the small sample might limit the generalizability of findings to broader stroke populations. A larger, more diverse sample might offer a more comprehensive understanding of EEG changes post-stroke. The research was conducted in a single clinic, potentially limiting the variability and representation of stroke patient populations. Multi-center studies with diverse patient demographics could offer a more comprehensive perspective on post-stroke EEG changes. While manual EEG analysis offers detailed insights, it’s subject to interpretation biases and might lack the objectivity of automated algorithms. Providing inter-rater reliability measures or comparisons with automated analyses could strengthen the study’s methodology. The study primarily focused on classifying EEG changes into abnormal, nonspecific, and normal categories based on specific elements (spikes, sharp waves, theta/delta rhythms). This classification might overlook nuanced EEG patterns or miss subtleties that automated or alternative analyses could capture. While mentioning CFC and its potential, the study did not directly employ these advanced analyses. Incorporating CFC or other advanced EEG analysis methods could further enrich the understanding of brain activity post-stroke and reveal additional insights beyond the traditional EEG classification used in this study.

Conclusion

Our findings suggest that about 30% of stroke patients exhibit abnormal EEG results, signifying a predisposition to early seizures. Consequently, early administration of antiepileptic drugs becomes crucial for this subgroup. It is advisable to screen a larger proportion of stroke patients using EEG to promptly identify bioelectrical alterations, initiate early antiepileptic prophylaxis, and prevent the development of structural epilepsy-related complications. The findings underscore a critical association emergence of post-stroke epilepsy and certain factors such as infarction location within the anterior cerebral artery territory, proximity to the dominant cerebral hemisphere cortex, younger age, lower INR levels, and elevated cholesterol appear pivotal in influencing the risk of post-stroke epileptogenesis.Clinical ImplicationsThe observed distribution of EEG patterns in post-stroke patients, characterized by a predominant presence of nonspecific changes and moderate disorganization in bioelectrical brain activity, highlights the complexity of post-stroke neurophysiological alterations. These findings underscore the need for nuanced clinical evaluations beyond initial assessments, acknowledging that a substantial proportion of patients exhibit non-specific EEG changes that might not align directly with classical epileptic elements.Diagnostic ConsiderationsIdentifying a significant percentage of patients displaying abnormal EEG results and a smaller subset demonstrating normal EEG values suggests the variability in post-stroke brain activity presentations. This calls for a comprehensive diagnostic approach that considers the diverse EEG patterns seen in stroke recovery, emphasizing the necessity of precise and individualized diagnostic interpretations beyond a binary classification of abnormal or normal EEG findings.Treatment Strategies and PrognosisUnderstanding the distribution of EEG patterns following stroke could have implications for prognosis and treatment planning. The prevalence of nonspecific EEG changes might signal a need for tailored rehabilitation strategies targeting these particular neurophysiological alterations. Moreover, the subset displaying abnormal EEG results could prompt closer monitoring for potential epileptic tendencies or secondary complications.Research DirectionsFurther research should delve into the long-term implications of these observed EEG distributions in post-stroke patients. Investigating how these varied EEG patterns correlate with functional outcomes, cognitive recovery, or predisposition to long-term neurological conditions could offer deeper insights into prognostic markers and guide more refined therapeutic interventions in stroke rehabilitation.Overall ImpactThese findings contribute to the evolving understanding of post-stroke neurophysiological changes, emphasizing the necessity of nuanced EEG assessments. Recognizing the spectrum of EEG alterations in post-stroke populations is pivotal in advancing our comprehension of recovery trajectories and tailoring interventions to optimize neurological outcomes in these patients.

Declarations

References

1. Shah AK, Agarwal R, Carhuapoma JR, Loeb JA. Compressed EEG pattern analysis for critically ill neurological-neurosurgical patients. Neurocrit Care. 2006;5(2):124-133. doi:10.1385/ncc:5:2:124
   Article · PubMed · Google Scholar
2. Zhang N, Chen F, Xie X, et al. Application of quantitative EEG in acute ischemic stroke patients who underwent thrombectomy: a comparison with CT perfusion. Clin Neurophysiol. 2022;141:24-33. doi:10.1016/j.clinph.2022.06.007
   Article · PubMed · Google Scholar
3. Sheorajpanday RV, Nagels G, Weeren AJ, van Putten MJ, De Deyn PP. Quantitative EEG in ischemic stroke: correlation with functional status after 6 months. Clin Neurophysiol. 2011;122(5):874-883. doi:10.1016/j.clinph.2010.07.028
   Article · PubMed · Google Scholar
4. Johannessen LC, Johannessen SI. Pharmacological management of epilepsy: recent advances and future prospects. Drugs. 2008;68(14):1925-1939.
   PubMed · Google Scholar
5. Schomer DL, Lopes da Silva FH, eds. Niedermeyer’s Electroencephalography: Basic Principles, Clinical Applications, and Related Fields. Oxford University Press; 2017.
   PubMed · Google Scholar
6. Babiloni C, Barry RJ, Başar E, et al. International Federation of Clinical Neurophysiology EEG research workgroup: recommendations on frequency and topographic analysis of resting state EEG rhythms, part 1: applications in clinical research studies. Clin Neurophysiol. 2020;131(1):285-307. doi:10.1016/j.clinph.2019.06.234
   Article · PubMed · Google Scholar
7. Naydenov C, Yordanova A, Mancheva V. Methodology for EEG and reference values of the software analysis. Open Access Maced J Med Sci. 2022;10(B):2351-2354. doi:10.3889/oamjms.2022.10751
   Article · PubMed · Google Scholar
8. Maganti RK, Rutecki P. EEG and epilepsy monitoring. Continuum (Minneap Minn). 2013;19(3):598-622. doi:10.1212/01.con.0000431378.51935.d8
   Article · PubMed · Google Scholar
9. Yordanova L, Kiryakova G, Veleva P, Angelova N, Yordanova A. Criteria for selection of statistical data processing software. IOP Conf Ser Mater Sci Eng. 2021;1031:1-8. doi:10.1088/1757-899x/1031/1/012067
   Article · PubMed · Google Scholar
10. Meschia JF. Pharmacogenetics and stroke. Stroke. 2009;40(11):3641-3645. doi:10.1161/strokeaha.109.562231
    Article · PubMed · Google Scholar
11. Vatinno A, Ramakrishnan V, Simpson A, Bonilha H, Seo NJ. EEG as a predictor of poststroke recovery: a systematic review and meta-analysis. J Clin Transl Sci. 2020;4(suppl 1):71. doi:10.1017/cts.2020.233
    Article · PubMed · Google Scholar
12. Philips GR, Daly JJ, Principe JC. Topographical measures of functional connectivity as biomarkers for poststroke motor recovery. J Neuroeng Rehabil. 2017;14:1-16.
    PubMed · Google Scholar
13. Dimitriadis SI, Sun Y, Kwok K, Laskaris NA, Thakor N, Bezerianos A. Cognitive workload assessment based on the tensorial treatment of EEG estimates of cross-frequency phase interactions. Ann Biomed Eng. 2015;43:977-989. doi:10.1007/s10439-014-1143-0
    Article · PubMed · Google Scholar
14. Ren B, Yang K, Zhu L, et al. Multigranularity analysis of brain networks assembled with intrafrequency and cross-frequency phase coupling for human EEG after stroke. Front Comput Neurosci. 2022;16:785397. doi:10.3389/fncom.2022.785397
    Article · PubMed · Google Scholar
15. Yan J, Sun J, Guo X. Motor imagery cognitive network after left ischemic stroke: study of the patients during mental rotation task. PLoS One. 2013;8(10). doi:10.1371/journal.pone.0077325
    Article · PubMed · Google Scholar
16. Cai L, Wei X, Wang J, Yu H, Deng B, Wang R. Reconstruction of functional brain network in Alzheimer disease via cross-frequency phase synchronization. Neurocomputing. 2018;314:490-500. doi:10.1016/j.neucom.2018.07.019
    Article · PubMed · Google Scholar
17. Jacobs D, Hilton T, Del Campo M, Carlen PL, Bardakjian BL. Classification of preclinical seizure states using scalp EEG cross-frequency coupling features. IEEE Trans Biomed Eng. 2018;65(11):2440-2449. doi:10.1109/tbme.2018.2797919
    Article · PubMed · Google Scholar

Tables

Additional Information

Publisher’s Note
Bayrakol MP remains neutral with regard to jurisdictional and institutional claims.

Rights and Permissions

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0). To view a copy of the license, visit https://creativecommons.org/licenses/by-nc/4.0/

View full license details →