Authors :
Presenting Author: Nina Ghosn, BS – University of Pennsylvania
Erin Conrad, MD – University of Pennsylvania; Brian Litt, MD – Principle Investigator, University of Pennsylvania; Devin Ma, B.S – research associate, bioengineering, University of Pennsylvania
Rationale:
Reduction in epileptiform spike counts using anti-seizure medications (ASMs) has been associated with improved seizure control. However, the short-term relationship between spike rate and ASM load is unclear. Understanding how ASMs affect spike rate may enable the use of spike rate as a biomarker of ASM compliance and of disease management in response to introduction of new pharmacological therapies for patients with implantable recording devices. In this study, we seek to unravel the relationship between epileptic spike rate and ASM load and dosing, controlling for the confounders of seizures and circadian rhythm.
Methods:
We analyzed 80 patients that underwent intracranial EEG monitoring for epilepsy localization. We used an automated spike detector to detect epileptic spikes and collected medication administration times and doses from the electronic health record for all ASMs administered during the course of monitoring. We used a previously validated model to calculate estimates of ASM load in each patient. The period of ASM dosing in the EMU stay before ASM tapering began was identified for spectral analysis using a wavelet transform. We compared the dosing period of each ASM a patient was taking to the period of the spectral peaks on the wavelet periodogram.
Results:
A linear mixed model, controlling for time since the last seizure and patient variability, showed a significant negative relationship between ASM load and spike rate (coefficient = -1.96, p = 0.0002). Time-frequency analysis of ASM load and spike rate revealed a positive and statistically significant correlation between the peak period (in hours) on the spike rate periodogram and ASM dosing periods (hours) (p < 0.01, Pearson correlation). Also, patients who lacked peaks on the spike rate periodogram within 30% of ASM dosing periods exhibited more variable ASM dosing (Wilcoxon rank-sum test, p = 0.004, z = 2.9). Since most patients (n=77) were dosed with ASMs every 12 hours, we tested whether the 12 hour periodic component detected on the spike rate periodogram was a harmonic of the circadian 24-hour period and found that the occurrence of a 12 hour peak and a 24 hour peak were independent (Fisher’s test, p = 0.1, odds ratio [95% CI] = 0.32 [0.01, 1.1]). These results suggest a relationship between ASM dosing and spike rate, controlling for the circadian rhythm and seizures, with spike rates decreasing as ASM load falls between doses.