EEG desynchronization during phasic REM sleep suppresses interictal epileptic activity in humans

Abstract

Objective: Rapid eye movement (REM) sleep has a suppressing effect on epileptic activity. This effect might be directly related to neuronal desynchronization mediated by cholinergic neurotransmission. We investigated whether interictal epileptiform discharges (IEDs) and high frequency oscillations-a biomarker of the epileptogenic zone-are evenly distributed across phasic and tonic REM sleep. We hypothesized that IEDs are more suppressed during phasic REM sleep because of additional cholinergic drive.

Methods: Twelve patients underwent polysomnography during long-term combined scalp-intracerebral electroencephalography (EEG) recording. After sleep staging in the scalp EEG, we identified segments of REM sleep with rapid eye movements (phasic REM) and segments of REM sleep without rapid eye movements (tonic REM). In the intracerebral EEG, we computed the power in frequencies <30 Hz and from 30 to 500 Hz, and marked IEDs, ripples (>80 Hz) and fast ripples (>250 Hz). We grouped the intracerebral channels into channels in the seizure-onset zone (SOZ), the exclusively irritative zone (EIZ), and the normal zone (NoZ).

Results: Power in frequencies <30 Hz was lower during phasic than tonic REM sleep (p < 0.001), most likely reflecting increased desynchronization. IEDs, ripples and fast ripples, were less frequent during phasic than tonic REM sleep (phasic REM sleep: 39% of spikes, 35% of ripples, 18% of fast ripples, tonic REM sleep: 61% of spikes, 65% of ripples, 82% of fast ripples; p < 0.001). In contrast to ripples in the epileptogenic zone, physiologic ripples were more abundant during phasic REM sleep (phasic REM sleep: 73% in NoZ, 30% in EIZ, 28% in SOZ, tonic REM sleep: 27% in NoZ, 70% in EIZ, 72% in SOZ; p < 0.001).

Significance: Phasic REM sleep has an enhanced suppressive effect on IEDs, corroborating the role of EEG desynchronization in the suppression of interictal epileptic activity. In contrast, physiologic ripples were increased during phasic REM sleep, possibly reflecting REM-related memory consolidation and dreaming.

Keywords: Epilepsy; High-frequency oscillations; Intracerebral EEG; Polysomnography; Sleep.

Figure 1

Mean power spectral density during phasic and tonic segments for channels in the normal zone (NoZ) with no epileptic or other abnormalities, in the irritative zone outside the seizure‐onset zone (exclusively irritative zone, EIZ), and in the seizure‐onset zone (SOZ); error bars were omitted for clarity.

Figure 2

Mean ratio of the power spectral density between phasic and tonic segments for channels in the normal zone (NoZ) with no epileptic or other abnormalities, in the irritative zone outside the seizure‐onset zone (exclusively irritative zone, EIZ), and in the seizure‐onset zone (SOZ). The broken lines correspond to one standard error of the mean. The vertical line at 30 Hz indicates that this was the limit between the predefined low and high frequency bands used in the statistical analysis. In both frequency bands and for all the zones the mean power was significantly different between phasic and tonic segments (paired t‐tests, p‐values < 0.001); larger for phasic segments only for the high frequency band in the NoZ.

Figure 3

Distribution of interictal epileptic events (spikes, ripples, fast ripples, and burst duration) across phasic (blue) and tonic (light blue) REM sleep. On the left side of the panel the proportion of all events across phasic and tonic REM sleep is given; on the right side of the panel the proportions for the individual patients are given. The numbers reflect the total number of events of the categories analyzed in this study. Note that all categories of interictal epileptic events had significantly higher rates during tonic compared to phasic REM sleep. The p‐values for the individual patients are not corrected for multiple comparisons (**p < 0.001; *p < 0.05).

Figure 4

Mean ripple rates and standard errors of the mean for channels in the normal zone (NoZ) with no epileptic activity or other abnormalities, in the exclusively irritative zone (EIZ) exhibiting epileptic activity outside the seizure‐onset zone, and in the seizure‐onset zone (SOZ). The mean ripple rate in NoZ was significantly higher in phasic REM sleep compared to tonic REM sleep (p < 0.001), whereas the opposite was true for ripples in channels with epileptic activity both outside and inside the SOZ (both p‐values < 0.001). In addition, differences between zones in phasic and tonic segments were significant except between NoZ and EIZ during phasic REM sleep (*p < 0.05; **p < 0.001).

Figure 5

Mean fast ripple rates and standard errors of the mean for channels in the normal zone (NoZ) with no epileptic activity or other abnormalities, in the exclusively irritative zone (EIZ) exhibiting epileptic activity outside the seizure‐onset zone, and in the seizure‐onset zone (SOZ). As expected, fast ripples were sparse in the NoZ during both phasic and tonic REM sleep. For EIZ and SOZ, the differences between phasic and tonic segments were statistically significant (p < 0.001). Differences between zones in phasic and tonic segments were significant for all zones, except the rate in fast ripples in NoZ and EIZ during phasic REM sleep (**p < 0.001).