Abstracts

Rationale: Sudden Unexpected Death in Epilepsy (SUDEP) is the most frequent epilepsy-related cause of death, with an annual prevalence of one in every 150 patients with drug-resistant epilepsies. Currently, predictors of SUDEP are determined based on population studies that are personalized by modeling demographics, neuroimaging, and electrophysiological data collected at fixed time intervals. However, seizure burden and the response to the treatment change with time, thereby changing the assigned risks. Thus, these static risk assessments are fraught with uncertainty when applied prospectively to identify the increased susceptible periods to SUDEP. Addressing this knowledge gap would enable the development of timely interventions delivered during the susceptible periods to prevent SUDEP. Chronic ambulatory electrocorticography (ECoG) allows quantification of seizure burden, thereby automating temporally dynamic risk assessment of SUDEP. Here for the first time, we demonstrate temporal trends of cortical electrophysiological parameters that identified SUDEP susceptible period months preceding probable SUDEP.

Methods: A single-center retrospective study involving a 38-year-old woman with probable SUDEP and two age, sex, and disease-matched controls. All three patients had bi-temporal epilepsy with bilateral amygdala-hippocampal RNS. The ECoG was analyzed for power spectral density (PSD) between 1-70 Hz and Teager energy for the following epochs- a) baseline (scheduled ECoG); b) entire 90 secs ECoG detected as long episodes (LE), and c) 6 seconds postictal epoch curated from visually confirmed seizures. A ratio of beta/theta power was analyzed for postictal epochs as a surrogate for postictal arousal. Temporal trends were determined for mean power values by fitting a Gaussian distribution over a moving window of 80 days with a 79-day overlap between successive windows. Inter-seizure intervals were computed for all three subjects, and a moving average was computed to analyze the frequency of seizure occurrence.

Results: A total of 2315 RNS ECoG recorded over 1857 patient days were analyzed. For Probable SUDEP case: The temporal trend analysis revealed an increase in the inter-seizure interval (Fig1C) and increased mean power spectra of LE (Fig1A) that was consistent with increased patient-reported convulsive seizures. There was a progressive decrease in postictal mean power and beta/theta ratio in the last 80 days preceding SUDEP, thereby implying a heavy postictal EEG suppression and lack of homeostatic arousal response (Fig1A). The graphical display in the temporal trends mapping a) increased mean power spectra during seizures, and b) decrease postictally can be defined as cortical physiological markers of SUDEP susceptible periods that preceded over weeks before probable SUDEP. Similar trends mapped over two years were absent from the controls (Fig 1B, D).

Conclusions: Chronic ambulatory electrocorticography from seizure focus provides a unique opportunity to quantify the burden of epileptiform paroxysms and identify personalized SUDEP susceptible periods through dynamic risk assessment.

Funding: Please list any funding that was received in support of this abstract.: The authors received no specific funding for this work.