Abstracts

Rationale: Wearable devices can detect changes in autonomic nervous system (ANS) activity during seizures. Furthermore, anti-seizure medications (ASMs) may affect ANS activity in patients with epilepsy (PWE). We aimed to assess ASM effects on ANS signals from wearables before and after ASM intake in patients on no ASM, 1 ASM and >1 ASM.


Methods: We included patients 1 month to 21 years old admitted for long-term video-EEG monitoring at Boston Children’s Hospital between 2/2015 and 2/2021 who wore at least one wearable (Empatica E4, Milan, Italy) on their wrist and/or ankle. We evaluated electrodermal activity (EDA), heart rate (HR), and root mean square of successive differences of heart rate variability (RMSSD) in patients on no ASM (control group), 1 ASM, and >1 ASM at 2 timepoints: baseline (1 hour before ASM intake) and at ASM peak window (30 min before and after ASM peak concentration) (Figure 1a, Table 1). In the >1 ASM group, the ASM peak window was calculated as the average of all ASM peak concentrations per patient. For the control group without ASMs, we used the same mean ASM intake time and mean ASM peak times of both ASM groups for comparison. Both ASM groups were composed of PWE, while the control group contained PWE and without epilepsy, without ASMs. We analyzed 1 device recording per patient and assessed group-level mean EDA, HR, and RMSSD. We conducted a repeated measures analysis of variance with a 3 cohort design (two ASM groups and a control group) across 2 time points (baseline and mean ASM peak time). We excluded patients on continuous infusion, neurostimulation devices, adrenocorticotropic hormone treatment, corticosteroids, and patients with seizures within 2 hours of ASM intake.


Results: We included 92 patients, 7 patients on no ASM (57.1% female, median age: 6.12 yrs), 18 patients on 1 ASM (55.5% female, median age: 10.67 yrs), and 67 patients on > 1 ASM (53.7% female, median age: 9.05 yrs) (Table 1). Mean EDA was higher at mean ASM peak time than at baseline (F (1,89) = 8.80, p < 0.01, η² = 0.09), with a trend to a group by time point interaction (F (2,89) = 2.54, p =0.08, η² = 0. 0.5). Post-hoc analysis showed higher EDA at peak time compared to baseline for control (p < 0.01) and >1 ASM (p = 0.01) groups, but not for 1 ASM group. HR was lower at peak time compared to baseline (F (1,89) = XY, p = 0.03, η² = 0.05) for all groups. For RMSSD, there was a group effect (F (2,89) = 3.30, p = 0.04, η² = 0. 0.07). Bonferroni posthoc comparisons showed RMSSD was lower for >1 ASM group compared to 1 ASM group (p = 0.04, Figure 1).


Conclusions: ASMs may impact EDA, HR and RMSSD as recorded from wearables and offer potential biomarkers for non-invasive ASM monitoring, given the potential relation between ASM effects on the ANS and seizure susceptibility. Specific ASMs may reduce sweat gland activity and thus research is needed to consider ASM mechanism of action.


Funding: This study was funded by Epilepsy Research Fund.