Abstracts

Rationale: Sudden unexpected death in epilepsy (SUDEP) is the most common category of epilepsy-related mortality with the majority of deaths occurring at night, during sleep. In many of the victims, changes in sleep before death have been reported. Advances in sleep signal analysis show that sleep homeostatic processes, a biological process that recalibrates synaptic strength and excitability in the human cortex, can be measured by the power in slow wave activity (SWA) during non-REM (NREM) sleep, which is typically high in early sleep, and then declines as the night progresses. In this work, we identify sleep-related abnormalities in premortem, multimodal polygraphed records of SUDEP cases, which can in the future serve as potential biomarkers of SUDEP risk.


Methods: We studied all night recordings of 20 SUDEP cases with age and gender-matched high SUDEP risk ( >3 recorded generalized tonic-clonic seizures (GTCs)) and low SUDEP risk (no GTCs) for a total of 60 patients. An automatic sleep staging algorithm was used for obtaining NREM sleep. The EEG signal from each of the 30 seconds of NREM sleep was processed to obtain the SWA activity (0.5 to 4 Hz). A linear model was used to fit the decline of the SWA activity through the night. An ANCOVA analysis was applied to find significant differences in each of the slopes. Sleep Fragmentation Index (SFI) was also calculated to observe differences in the sleep structure of the three groups.


Results: The linear models of the high-risk and low-risk patients displayed a global decline in SWA power across the line as reported in the literature (Figure 1A and 1B). However, the SUDEP patient's global SWA instead showed an increase across the night (Figure 1C). The group analysis performed on the slopes of the fitted linear models (Figure 2) indicated that there is no significant difference between the high-risk and low-risk groups (ANCOVA, p >0.05) but there is a significant difference between SUDEP and the two other groups (ANCOVA, p< 0.05). The SFI did not show a significant difference (KW, p >0.05) but the average value of the SUDEP group is noted as higher than the other two groups


Conclusions: There is growing support for the critical need to study sleep in people with epilepsy (PWE) and SUDEP. We have discovered an abnormality in the sleep homeostasis recovery, via the analysis of the global SWA through the night, in the SUDEP cases that is not reflected in the other two populations examined. With this study, we have shown a potential sleep feature or biomarker that might help predict SUDEP and provide insights into pathomechanisms that lead to a fatal event. Our future goal is to guide and develop future SUDEP preventive strategies in PWE who are at high-risk, ranging from sleep hygiene to medical or surgical treatments.


Funding: NIH - BreatheS - R01NS133743
NIH - CSR - U01NS090407