Abstracts

Rationale: Beyond the brain, seizures affect systems and organs under its control, including the heart. However, peri-ictal cardiovascular changes are poorly understood and have been investigated primarily in short peri-ictal electrocardiograms (ECG), focusing predominantly on heart rate. Seizure-induced modulations of both ventricular and atrial activity, identifiable in continuous ECG during seizure evolution, remain elusive. To address this gap in knowledge, this study investigated seizure-related cardiovascular (ventricular and atrial) changes in continuous clinical ECGs from pediatric patients with medically refractory epilepsy.

Methods: All data were collected in the Epilepsy Center at Boston Children’s Hospital. Continuous ECG from 29 children (12 females, median age = 9 years, inter-quartile range (IQR) = 7.5 years) with pharmacoresistant seizures (median = 7 seizures, IQR = 7) and no reported cardiovascular anomalies were analyzed. One-/ two-lead ECGs were recorded continuously (median length = 93.9 h, IQR = 65.5 h) during inpatient neurophysiological studies, at a rate of 1024 samples/s. Data were analyzed in the Harvard Medical School High-Performance Cluster using the software MATLAB (Mathworks, Inc). A data-driven classification algorithm was used to separate ventricular and atrial activity from ECGs, using a 1-min sliding window. Patient-specific QRS/T waveform templates were adaptively matched to ECGs to extract ventricular signals and residual atrial contributions. Following separation, approximate entropy and fractal dimension (measuring signal complexity), were estimated from atrial signals, and heart rate was estimated from ventricular signals. Change-point detection identified intervals in which significant parameter changes occurred. Measures in intervals containing seizures were compared to those without seizures using non-parametric testing.

Results: Eighteen of 29 patients (62.1%) had significant changes in atrial and/or ventricular activity in intervals containing seizures. Of these,11 (61.1%) had changes in heart rate (p ≤ 0.04; 10 (90.9%) had relative tachycardia), 13 (72.2%) had changes in approximate atrial entropy (p ≤ 0.03) and/or fractal dimension (p ≤ 0.01). Of the latter, 6 (46.2%) had increased approximate entropy, i.e., higher signal irregularity, two had lower entropy (15.4%), 8 (61.5%) had higher fractal dimension and 3 (23.1%) had significant changes in both measures. Finally, 6 patients had both increased heart rate and atrial complexity. These findings were independent of age at epilepsy diagnosis, number of seizures, or origin of ictal onset.

Conclusions: This initial investigation of 29 pediatric epilepsy patients has identified frequent electrocardiographic changes during seizure evolution in continuous ECGs. Beyond changes in heart rate (mainly tachycardia), increased atrial signal complexity and irregularity was also estimated in almost half of the patients. Thus, electrocardiographic changes may be integrated with neurophysiological markers to improve seizure prediction.

Funding: Please list any funding that was received in support of this abstract.: National Institute for Neurological Disorders and Stroke (Grant # R03 NS119799).