Epilepsy Watch After Vascular Events: Frequency, Outcomes, and Risk Markers

Overview

This research aims to determine the frequency of seizures in patients following a stroke, identify risk factors associated with post-stroke seizures, and characterize EEG changes linked to these seizures. Unexplored alterations in the aperiodic component of the EEG in post-stroke patients could potentially serve as novel biological markers for epilepsy after stroke.

This study included patients with stroke or post-stroke seizures who were admitted to the Neurology Division at the University Medical Centre Maribor, Slovenia, over a 12-month period. The anticipated number of patients was approximately 600-650. Inclusion criteria encompassed acute stroke or seizure occurring at any point after the stroke. Stroke classification was based on the World Health Organization (WHO) criteria, while post-stroke seizures and epilepsy were defined according to the International League Against Epilepsy (ILAE) definitions and classifications. All participants provided written informed consent after being thoroughly informed about the study. The study was pre-approved by the Medical Ethics Committee of the Republic of Slovenia (Approval No. 48/08/14 and Approval No. 0120-302/2024-2711-3). Patients who did not meet the inclusion criteria or could not undergo diagnostic procedures as outlined in the research protocol were excluded based on principles of good clinical practice. During the 12-month study period, demographic, imaging, laboratory, and neurophysiological data were prospectively collected from all hospitalized stroke patients. Data Collection Patient history was utilized to gather information on neurological impairments, seizure onset, and risk factors for cerebrovascular diseases. For patients who experienced seizures, the interval between stroke onset and seizure occurrence (in days) was calculated to distinguish early from late post-stroke seizures. Vital signs, including blood pressure, pulse, height, weight, and body mass index (BMI), were recorded. Neurological impairment was assessed on admission and discharge through clinical examinations and standardized scales, including the National Institutes of Health Stroke Scale (NIHSS) and the modified Rankin Scale (mRS). Laboratory and Diagnostic Assessments Within 24 hours of admission, blood samples were collected to measure urea, creatinine, electrolytes, uric acid, cholesterol, triglycerides, liver enzymes, blood glucose, cystatin C, high-sensitivity C-reactive protein (hsCRP), red blood cell count, hemoglobin concentration, and urine analysis. Within 72 hours of admission, imaging diagnostics (computed tomography \[CT\] or magnetic resonance imaging \[MRI\] of the brain) and functional diagnostics (electroencephalography \[EEG\]) were performed. The study population consisted of patients with ischemic stroke, hemorrhagic stroke, subarachnoid hemorrhage, and other rare cerebrovascular diseases (CVD). Patient Grouping Clinical Seizure Data and Timing of Onset Post-Stroke: Patients were classified into three groups: No seizures after stroke ("no EPI") Early seizures (within 7 days post-stroke; "early EPI") Late seizures (more than 7 days post-stroke; "late EPI") EEG-Based Grouping: Patients were additionally grouped based on EEG results: EEG+/EPI+: Epileptiform EEG changes with seizures EEG+/EPI-: Epileptiform EEG changes without seizures EEG-/EPI-: No epileptiform EEG changes and no seizures KON: Control group of healthy individuals Planned Analyses Demographic Analysis: Data on gender, age, cerebrovascular risk factors, stroke type, functional impairment (assessed by NIHSS and mRS), seizure prevalence, and EEG changes were analyzed for all participants. Subgroup demographic analyses were performed based on clinical and EEG data. EEG Analysis: Standard visual EEG analysis included the evaluation of spectral frequency bands and the identification of focal or generalized epileptiform abnormalities. Preprocessing involved removing segments with noise, saturation, or absence of EEG activity. Ocular artifacts, including blink-related components, were identified using independent component analysis, and the EEG signals were reconstructed without these artifacts. Using spectral parameterization (SPECPARAM 2.0 in Python), power spectral density was calculated for each patient. Aperiodic components were analyzed by extracting the exponent and offset from each frequency spectrum. Welch's t-tests were used to compare these parameters between groups. Additionally, standardized low-resolution brain electromagnetic tomography (sLORETA) was employed for signal source localization, micro-EEG potential analysis, and network distribution assessment. Statistical Data Analysis Descriptive Statistics: Basic descriptive metrics, including mean, standard deviation, median, minimum, maximum, and quartiles, were calculated for each variable to assess within-group distributions. Frequencies and relative frequencies were determined for categorical variables, with emphasis on the prevalence rates within the "no EPI," "early EPI," and "late EPI" groups. Results were presented in frequency tables. Inferential Statistics: Parametric tests (for normally distributed data): Two-group comparisons: t-tests Multi-group comparisons: ANOVA Non-parametric tests (for non-normally distributed data): Two-group comparisons: Mann-Whitney U tests Multi-group comparisons: Kruskal-Wallis tests For EEG-based groups, extracted offset and exponent values of aperiodic components were compared using Welch's t-tests. Correlation analyses (Pearson's or Spearman's) were performed based on data distribution. Post-hoc analyses used Dunn's tests for pairwise comparisons when significant differences were identified. Categorical Data Analysis: Chi-square tests evaluated differences between categorical variables among patient groups. Fisher's exact test was applied when expected frequencies were too low for the chi-square test. Survival Analysis: To examine seizure onset timing, survival analysis was conducted using the time from stroke onset as the time variable. Kaplan-Meier analysis estimated survival curves representing seizure-free intervals, and log-rank (Mantel-Cox) tests were used to compare survival distributions across groups. This analysis helped to identify factors associated with seizure onset timing among stroke patients.