Introduction

Epilepsy represents a significant public health concern, affecting ~250 per 100,000 children in Africa.1 About 20–30% of patients with epilepsy (PWE) do not respond adequately to standard antiseizure medications (ASMs), resulting in drug-resistant epilepsy (DRE).2 DRE imposes considerable socioeconomic and psychological burdens, resulting in a significant decrease in well-being. This burden also impacts caregivers and the entire healthcare system.3 Epileptic children, especially those not in remission, have an increased incidence of sudden unexpected death in epilepsy (SUDEP).4 The relative risk of SUDEP is higher in cases of childhood-onset epilepsy compared to adult-onset epilepsy.5 Although the mechanisms underlying SUDEP are heterogeneous, cardiac dysfunction has been implicated as a critical factor.6 Cardiac abnormalities, including varying degrees of myocardial fibrosis and myocyte hypertrophy, were reported in 25% of SUDEP cases.7 Additionally, SUDEP patients showed severe autonomic dysregulation with decreased awake heart rate variability.8

Myocardial strain measurement using speckle-tracking echocardiography (STE) is currently the most widely utilized method for detecting subtle abnormalities in myocardial deformation and subclinical cardiac dysfunction.9 Although myocardial strain impairments have been reported in children with DRE,10,11 a prior study found no significant cardiac changes in adult patients with long-standing DRE.12 These divergent findings underscore the importance of pediatric-specific investigations to clarify the nature and predictors of cardiac dysfunction in this vulnerable population. Early detection of subclinical cardiac dysfunction via STE could inform interventions to mitigate SUDEP risk in children with DRE. We hypothesized that children with DRE demonstrate subclinical cardiac dysfunction detectable via STE compared to those with controlled epilepsy and healthy controls. This study aimed to assess cardiac function in children with controlled epilepsy, DRE, and healthy controls and to identify clinical predictors (e.g., seizure frequency, medication burden) of cardiac abnormalities to guide risk stratification and monitoring.

Materials and methods

Study design

This case-control study was conducted at the cardiology department and pediatric neurology outpatient clinic of Beni-Suef University Hospital between April 2023 and January 2024. The study protocol was approved by the Research Ethics Committee of Beni-Suef University’s Faculty of Medicine (Approval No. FMBSUREC/05032023/ Shaker). Written informed consent was obtained from the parents or legal guardians of all pediatric participants prior to inclusion. The study was conducted in full accordance with the ethical principles of the Declaration of Helsinki.

Participants

The study included 30 patients with DRE, 30 with drug-responsive epilepsy, and 30 healthy controls, with all groups matched for age and sex. PWE aged 3–18 years were included based on the International League Against Epilepsy (ILAE) 2014 operational definition, which defines epilepsy as “at least two unprovoked seizures >24 h apart, one unprovoked seizure with a high recurrence risk, or diagnosis of an epilepsy syndrome”.13 Patients were classified as DRE if they failed to be seizure-free after two tolerated and appropriately chosen ASMs. Patients were classified as drug-responsive if being seizure-free for one year or at least three times the longest pretreatment inter-seizure interval.14 Exclusion criteria were patients under 3 years of age or over 18 years, those without parental consent, individuals with a history of any cardiac disease (congenital or acquired), a history of hypertension, or a known metabolic disease, as well as those taking medications other than ASMs that could affect cardiac function. Patients with symptomatic epilepsy, such as those with structural or metabolic etiologies, were also excluded. Symptomatic epilepsy often presents with comorbidities or underlying factors that could independently affect cardiac function. This study protocol was designed to minimize confounding variables and isolate the impact of epilepsy and its treatment on cardiac function in children.

Evaluations and data collection

Clinical assessments

All patients underwent complete history taking, including age, sex, residence, consanguinity, perinatal and developmental history, family history of epilepsy, seizure onset, semiology, frequency, duration since the last seizure, and ASMs type, number, and response to therapy. A comprehensive clinical examination was conducted, focusing on the neurological and cardiac systems.

Neurodiagnostic assessments

EEG recordings were acquired using a 32-channel Nihon Kohden EEG-1200 system (Tokyo, Japan) with a sampling rate of 256 Hz and bandpass filtering (0.5–70 Hz), adhering to the 10–20 international electrode placement protocol. Sleep-deprived EEGs (minimum 4 h of sleep restriction) were performed for 30 min to optimize detection of epileptiform discharges. Brain MRI was performed on a 1.5 Tesla Philips Achieva scanner (Amsterdam, Netherlands), utilizing T1-weighted, T2-weighted, and fluid-attenuated inversion recovery (FLAIR) sequences to exclude structural abnormalities

Electrocardiography (ECG)

Patients and controls underwent a 3-channel 12-lead ECG examination utilizing a Schiller AT1 model ECG machine (Baar, Switzerland), during which the rhythm, PR interval, QRS complex, and QT intervals were recorded.

Standard echocardiography

A General Electric Vivid T9 ultrasound device (GE Healthcare, Chicago, IL) equipped with a phased-array transducer (frequency range: 2.5–3.5 MHz) was utilized to perform standard echocardiograms on PWE and controls. The internal diameters of the left ventricle at diastole (LVIDd) and systole (LVIDs) were assessed using the parasternal long- or short-axis view. Left ventricular end-diastolic volume (LVEDV), end-systolic volume (LVESV), and ejection fraction (LVEF) were calculated using the modified Simpson’s biplane method (apical 4- and 2-chamber views), adhering to the 2024 American Society of Echocardiography (ASE) Guidelines.15 Offline analysis was performed using EchoPAC software (v202, GE Healthcare)

Speckle tracking measurements

Longitudinal myocardial deformation parameters were assessed using STE on images acquired with the General Electric Vivid T9 system. The three standard apical views (apical 4-chamber, apical 2-chamber, and apical 3-chamber) were obtained during steady ECG recording. Three consecutive cardiac cycles were recorded and averaged, utilizing a frame rate of 60–80 frames per second. The stored images were analyzed offline using the EchoPAC software to obtain various longitudinal myocardial deformation parameters. The end-systole was defined automatically, while the operator semi-automatically delineated the region of interest by identifying one apical point at the apex endocardial border and two annular points located low and within the myocardium at the level of the mitral valve insertion. GLS was derived from a 17-segment bull’s-eye display (Fig. 1). All measurements were conducted by two pediatric cardiologists with over five years of STE experience blinded to participant status (epilepsy vs. control) and clinical data. Each operator independently obtained and analyzed three consecutive LV GLS measurements for each patient, with the results subsequently averaged. Inter-observer variability was assessed in 20% of randomly selected cases (18 patients, balanced across groups), demonstrating excellent agreement (intraclass correlation coefficient [ICC] = 0.94; 95% CI: 0.89–0.97).

Fig. 1: Left ventricular global longitudinal strain (LVGLS) assessment.
figure 1

Two-dimensional speckle-tracking echocardiography demonstrates a bull’s-eye (polar) map of the left ventricle in a child with drugresistant epilepsy. The map displays segmental longitudinal strain values across the myocardium, with numerical strain values shown in each segment; more negative values indicate better myocardial deformation.

Full size image

Statistical analysis

Statistical analyses were conducted using SPSS software (version 26; IBM Corp., Armonk, NY). The Shapiro–Wilk test was employed to assess the normality of continuous variables. Categorical data were summarized using frequencies and percentages, while continuous variables were expressed as mean ± standard deviation (SD) for normally distributed data or median (interquartile range, IQR) for non-normally distributed data. Comparisons between PWE and controls were performed using independent t-tests (normally distributed data), Mann–Whitney U tests (non-normally distributed data), and Chi-square or Fisher’s exact tests for categorical data. For comparisons across three groups (DRE, drug-responsive epilepsy, and controls), normally distributed continuous data was analyzed using One-way ANOVA with Tukey’s post-hoc test, while non-parametric data were analyzed using the Kruskal–Wallis H test followed by Dunn’s test with Bonferroni adjustment. Categorical variables were assessed with Chi-square tests, and post-hoc pairwise comparisons were conducted as necessary. Spearman’s rank correlation coefficient was used to determine the relationships between ordinal or non-normally distributed variables. Statistical significance was established with a two-tailed p-value of <0.05.

Results

Demographic and clinical characteristics

A total of 60 PWE and 30 age- and sex-matched controls were enrolled. A positive family history of epilepsy was significantly more common among PWE than controls (26.7% vs. 0%, p = 0.002), and rural residency was also more frequent in the PWE group (75% vs. 50%, p = 0.018). Body weight (BW) and body mass index (BMI) were significantly lower in PWE compared to controls (p < 0.001 for both) (Table 1). Notably, patients with DRE exhibited even further reductions in BW (median: 19.0 kg [IQR: 14.0–25.0] vs. 26.0 kg [19.0–31.0], p = 0.003) and BMI (14.5 kg/m² [13.5–16.0] vs. 16.0 kg/m² [15.0–17.0], p = 0.002) compared to drug-responsive patients.

Table 1 Demographic data among epilepsy patients and controls
Full size table

Twelve (20.0%) patients exhibited delayed development, whereas 48 (80.0%) demonstrated normal development. Forty-seven (78.3%) patients had generalized epilepsy, predominantly characterized by generalized tonic-clonic (GTC) seizures, while 13 patients (21.6%) presented with focal epilepsy. Fifteen (25.0%) patients had normal EEG, 23 (38.3%) had generalized epileptogenic activity, and 22 (36.7%) had focal epileptogenic activity. Twenty (66.7%) patients used polytherapy, while 20 (33.3%) used monotherapy. Sodium valproate and levetiracetam were the most commonly used medications (71.7% and 58.3%, respectively). Only two patients (3.3%) were on a ketogenic diet.

ECG

ECG findings were comparable across DRE, drug-responsive epilepsy, and control groups, with all parameters remaining within normal limits, except for one participant in the control group who displayed a short QT interval of 359 ms, identified as an isolated finding without clinical signs of arrhythmia or structural cardiac abnormalities.

Echocardiographic parameters

Echocardiographic parameters differed significantly across groups (Table 2). LVEDV was lower in children with DRE compared to those with drug-responsive epilepsy (p = 0.017) and controls (p < 0.001), and lower in drug-responsive epilepsy versus controls (p = 0.015). LVESV and fractional shortening (FS) were also lower in DRE compared to controls (p = 0.015 and p = 0.014, respectively). LVEF was significantly impaired in DRE compared to drug-responsive epilepsy (p = 0.049) and controls (p < 0.001).

Table 2 ECG, echocardiographic, and speckle tracking parameters in epilepsy patients and controls
Full size table

Speckle-tracking echocardiography

LV GLS was significantly lower in DRE than in drug-responsive epilepsy (p = 0.023) and controls (p = 0.003), while no difference was observed between drug-responsive epilepsy and controls (p = 0.783) (Table 2). The LV GLS correlated positively with the duration since the last seizure (r = 0.347, p = 0.007), and negatively with the number of ASMs (r = −0.347, p = 0.007) (Table 3).

Table 3 Correlation between LV global longitudinal strain (LV GLS) and clinical data among epilepsy patients
Full size table

Discussion

This study represents one of the few investigations evaluating cardiac function in pediatric patients with drug-responsive epilepsy and DRE. By focusing on patients with idiopathic epilepsy and excluding those with symptomatic epilepsy, we aimed to minimize confounding factors arising from the heterogeneity of underlying pathologies.

Our findings revealed a significantly higher prevalence of a positive family history of epilepsy in PWE compared to healthy controls. This finding aligns with a study conducted in Egypt, reporting that 73.9% of children with idiopathic epilepsy had a positive familial history.16 PWE exhibited a significant rural residency predominance compared to controls. One meta-analysis also revealed that the burden of epilepsy in Sub-Saharan Africa is higher in rural than in urban areas. This may be attributed to societal stigma and prevailing misconceptions about the condition.17

ECG findings were comparable among DRE, drug-responsive epilepsy, and controls, with no significant differences. This finding is consistent with the study by González et al., which showed that in adult patients with DRE, both PR and corrected QT intervals were within normal limits for all patients.12 Çelik and colleagues reported similar findings in their work on seizure-free children with epilepsy.18

Patients with DRE exhibited significantly lower LVEDV compared to those with drug-responsive epilepsy and controls, as well as significantly lower LVESV compared to controls. The lower LVESV and LVEDV reported in our cohort in DRE can be attributed to the autonomic dysfunction previously reported in PWE, characterized by sympathetic predominance.19 During seizures, sympathetic activity often increases, leading to tachycardia and increased cardiac workload.20 Another potential explanation for the reduced volumes is that the LVESV and LVEDV are dependent on body surface area in the pediatric population.21 The weight of PWE was significantly lower than that of the controls. Additionally, the weight was significantly lower in DRE compared to drug-responsive epilepsy. Reduced body weight in epilepsy may indicate underlying disease mechanisms, such as metabolic alterations, activity limitations due to seizure burden, or medication effects (e.g., appetite suppression) like topiramate and zonisamide.22,23 Pfeifer et al. demonstrated a positive correlation between generalized epilepsy and underweight status.24 However, the relationship between body weight and epilepsy remains controversial and needs further investigation.25

Furthermore, patients with DRE exhibited significantly lower EF compared to those with drug-responsive epilepsy and controls, as well as lower FS compared to controls. Schreiber and colleagues observed that FS was significantly higher in children with refractory epilepsy compared to controls, while EF was comparable between the two groups.10 Ibrahim et al.19 and Sridech et al.11 found no statistically significant differences in FS and EF between children with DRE and controls. These discrepancies might be due to the limitations of EF measurement, which can be highly operator-dependent and influenced by loading conditions, thereby reducing its reliability as a sole indicator of ventricular function.7,26

Patients with DRE exhibited significantly lower LV GLS compared to drug-responsive epilepsy and controls. Similarly, Schreiber and colleagues found impaired longitudinal and circumferential strain in children with refractory epilepsy than in healthy controls.10 Sridech and colleagues also demonstrated that LV GLS was significantly lower in DRE children than in healthy controls.11 In contrast to these findings, González and colleagues, using both echocardiography and cardiac MRI (CMR), demonstrated no significant cardiac changes in adult patients with DRE compared to healthy controls.12 The divergence between our results and those of González et al. may stem from a combination of multiple factors. Their study included a small cohort of adults with long-standing DRE (n = 21, with CMR performed in only 10), excluded patients with a history of status epilepticus, which is linked to cardiac dysfunction in pediatric populations, and included both lesional and non-lesional cases. In contrast, our pediatric cohort examined a specific population characterized by unique vulnerabilities. Pediatric DRE may encompass genetic channelopathies, such as SCN1A, which could have cardiac implications.4 The observed differences highlight the necessity for age- and etiology-specific strategies in cardiac surveillance for epilepsy, especially in pediatric DRE, where early identification of subclinical dysfunction may guide monitoring strategies.

LV GLS was positively correlated with the duration since the last seizure (r = 0.347). Conversely, it was negatively correlated with the number of ASMs (r = −0.347). The negative correlation between the number of ASMs and LV GLS may reflect cumulative cardiotoxicity from polytherapy or inherent DRE severity necessitating aggressive treatment, or both. Supporting this, Kaman et al. demonstrated that valproate therapy was associated with subclinical ventricular dysfunction in children with epilepsy,27 while Mayer et al. found that carbamazepine and valproate were associated with an increased risk of cardiovascular events compared to lamotrigine in adults.28 Importantly, children with DRE typically receive higher cumulative doses and combinations of ASMs than drug-responsive children, some of which possess documented cardiotoxic potential. This differential exposure creates a fundamental interpretive challenge. The impairment in cardiac function observed in DRE patients may be substantially influenced by this medication burden and cannot be fully distinguished from epilepsy-related mechanisms based on our cross-sectional design. Future studies should enroll drug-naive pediatric cohorts for longitudinal cardiac assessments pre- and post-ASM initiation and stratify analyses by ASM class to clarify these contributions.

No significant relationship was found between LV GLS and the age of seizure onset, seizure frequency, number of hospital admissions due to seizures, or duration of epilepsy. Çelik et al. found no correlation between STE values and the duration of treatment, frequency, or duration of seizures.18 Similarly, Schreiber and colleagues found no associations between myocardial strain parameters and patient age; in age-adjusted analyses, there was no association between echocardiographic measurements and either the duration of epilepsy or the age at onset.10

Patients with drug-responsive epilepsy showed comparable values of LVESV, FS, EF, and LV GLS with controls. Çelik and colleagues similarly reported that seizure-free children with epilepsy had similar FS and EF values to healthy controls, though they observed significantly lower LV GLS in seizure-free patients compared to controls, which contrasts with our findings.18 The observed differences in LV GLS outcomes between our results and those reported by Çelik et al. may reflect differences in cohort characteristics, particularly treatment regimens. Specifically, sodium valproate (28.3%), oxcarbazepine (26.6%), and carbamazepine (23.3%) were the most frequently prescribed ASMs in Çelik’s cohort, whereas sodium valproate and levetiracetam predominated in ours. Such pharmacological heterogeneity could contribute to divergent strain patterns. Further studies are needed to delineate ASM-specific effects on myocardial deformation in pediatric epilepsy.

In our cohort, LVEDV was the only parameter significantly lower in drug-responsive epilepsy compared to controls. This isolated reduction in LVEDV may reflect autonomic dysregulation for instance, reduced preload due to a sympathetic-parasympathetic imbalance. Specifically, sympathetic overactivity during seizures may reduce LVEDV by decreasing venous return or shortening diastolic filling time. However, the absence of dedicated diastolic function assessment in our work, such as tissue Doppler Imaging (TDI), precludes definitive conclusions about diastolic function. Further studies integrating comprehensive diastolic evaluations in drug-responsive epilepsy are needed to clarify whether the reduced LVEDV reflects autonomic-mediated preload reduction, intrinsic myocardial abnormalities (such as early diastolic dysfunction), or compensatory hemodynamic mechanisms.

Limitations of this study include the single-center design and the small sample size, which may limit the generalizability of the findings. Second, the lack of TDI and CMR precludes comprehensive assessment of diastolic function and myocardial tissue characterization. Third, all patients were on ASMs at the time of enrollment, and the lack of a pretreatment group in our cohort prevents definitive attribution of the observed cardiac changes solely to epilepsy rather than to treatment effects. Children with DRE typically receive polytherapy and higher cumulative ASM exposure, introducing a significant confounding effect. Thus, the myocardial changes detected in this study cannot be attributed to epilepsy independently of medication effects. Despite these limitations, our study provides valuable insights into cardiac function in pediatric epilepsy. First, it is one of the few works combining ECG, M-mode echocardiography, and STE to navigate the cardiac changes in pediatric epileptic patients, both drug-resistant and responsive after the exclusion of symptomatic epilepsy, ensuring the homogeneity of the study population and excluding other confounding morbidities of symptomatic epilepsy. Second, while prior studies have explored cardiac strain in DRE, our work uniquely compares three distinct groups: DRE, drug-responsive epilepsy, and healthy controls. To our knowledge, this is the first study to offer such a comparative analysis, contributing new evidence regarding the spectrum of cardiac dysfunction in pediatric epilepsy. The observation of impaired LV GLS in DRE patients, contrasted with preserved strain in those with drug-responsive epilepsy, indicates that subclinical cardiac abnormalities may be specifically linked to drug resistance rather than to epilepsy itself. Third, our study contributes to the emerging normative data on LV GLS in the pediatric population, providing a basis for future research and clinical applications. While larger cohorts are required to establish definitive reference ranges, these initial findings regarding strain patterns in children are especially significant due to the scarcity of existing data. CMR is regarded as the gold standard for myocardial tissue characterization and strain analysis; however, its high costs, limited availability, and the requirement for sedation in younger children restrict its application in pediatric epilepsy. In contrast, STE offers a rapid, non-invasive, and widely accessible screening tool for detecting early cardiac dysfunction. Our findings underscore the clinical utility of STE as a first-line modality for cardiac surveillance in pediatric epilepsy, with CMR advised for specific cases requiring advanced tissue characterization.

Conclusions

Pediatric patients with DRE demonstrate significant impairments in cardiac function, including reduced LVEDV, LVESV, FS, EF, and LV GLS. Pediatric patients with drug-responsive epilepsy exhibit lower LVEDV while maintaining systolic function. These findings highlight the potential cardiovascular burden associated with epilepsy. LV GLS measurement by STE should be incorporated into the follow-up of patients with DRE to enable early detection of subclinical cardiac abnormalities. Further longitudinal studies utilizing multimodal imaging like TDI, CMR, and diastolic strain rate with enrollment of newly diagnosed, treatment-naive patients are needed to elucidate the cardiovascular burden associated with epilepsy and to differentiate the effects of epilepsy from those of ASMs.