Authors :
Presenting Author: Hunki Kwon, PhD – MGH/HMS
Dhinakaran Chinappen, PhD Student – MGH/BU; Anirudh Wodeyar, PhD – MGH/HMS; Erin Berja, Clinical Research Coordinator – MGH/HMS; Wen Shi, PhD – MGH/HMS; Katherine Walsh, Clinical Research Coordinator – MGH/HMS; Bryan Baxter, PhD – MGH/HMS; Dara Manoach, Professor – MGH/HMS; Mark Kramer, Professor – BU; Catherine Chu, Professor – MGH/HMS
Rationale:
Auditory stimulation timed to the upstate of slow oscillations (SO, 0.5-2 Hz) during non-rapid eye movement sleep enhances both SOs and sleep spindles (9-15 Hz) in healthy adults. Rolandic epilepsy (RE) is the most common focal developmental epilepsy in childhood and characterized by a focal sleep spindle deficit that correlates with cognitive deficits. We evaluated the impact of auditory stimulation on SOs and sleep spindles in children with RE and age-matched controls.
Methods:
We recruited six children with RE and six age-matched controls for two sessions of electroencephalogram (EEG) recordings during a 90-min nap opportunity. During one of the sessions, auditory stimulation (50 ms pink noise burst) was randomly presented throughout the nap; the other session had no auditory stimulation (sham). Offline SOs were detected at the FZ channel referenced to the mastoid following bandpass filtering (0.5-4 Hz). SOs were detected if two consecutive positive-to-negative zero crossings occurred within a range of 0.5 to 2 s, corresponding to 0.5–2 Hz, and the negative peak during that time window was below -40 μV. Some random stimulations were delivered during background activity and some at the time of SOs. In the latter case, the EEG signal was bandpass filtered (0.5-2 Hz) and the smoothed phase estimated after removing waveform distortions due to a narrowband filter. The proportion of evoked SOs was computed as the number of SOs within 1s after an auditory stimulation. This proportion was computed for stimuli delivered in the absence of a SO as well as for each of 12 phase bins (30 degree increments) when a SO was present at the time of stimulation. For sleep spindle measures, spindles were detected at the FZ channel referenced to the average across channels using a validated, automated spindle detector. To evaluate for differences after sham or auditory stimulation and for differences in the proportion of evoked SOs and spindles, we used two-tailed paired t-tests corrected for multiple comparisons using a cluster-based permutation method.
Results:
In all subjects regardless of group, and in the RE and control groups, auditory stimulation evoked SOs compared to sham stimulation (p< 0.029, for all groups). When a SO was present at the time of stimulation, stimuli delivered in the upstate of the SO evoked a higher proportion of SOs, and stimuli delivered in the downstate of the SO evoked a lower proportion of SOs, compared to stimuli delivered in the absence of a SO (p< 0.05, corrected,
Figure 1). In control subjects, auditory stimuli that evoked SOs trended to decreased sleep spindles coupled to SOs at the time of stimulation compared to sham (p< 0.05, uncorrected), whereas in RE subjects, there was no difference in spindles coupled to SOs at the time of auditory stimulation compared to sham and stimulation trended to increase spindles coupled to evoked SOs compared to sham (p< 0.05, uncorrected,
Figure 2).
Conclusions:
Auditory stimulation delivered during the upstate of SOs evokes SOs and sleep spindles in RE. Thus, this non-invasive intervention has the potential to improve sleep-dependent memory consolidation associated with these sleep oscillations.
Funding:
NIH NINDS R01NS115868