Abstract
Introduction
Prior research has reported NREM spectral EEG differences between individuals with insomnia and good-sleeper controls, including elevated high-frequency EEG power (beta/gamma bands, ~16-50Hz) and, to a lesser extent, elevations in sleep spindle parameters. However, the mechanisms driving these differences remain unclear. Harmonics have been observed in EEG data as spectral peaks at multiples of a fundamental frequency associated with an event (e.g., for a 14Hz spindle, the 2nd harmonic is expected to be a peak at 28Hz). Thus far, there has been very limited application of this idea of spectral harmonics to sleep spindles, even though these patterns can indeed be seen in some existing literature. We sought to build on this literature to apply spectral harmonic analysis to better understand differences between insomnia and good sleepers.
Methods
15 individuals with insomnia disorder (DSM-5 criteria, 13 female, age 18–32 years) and 15 good-sleeper controls (matched for sex, age, and BMI) completed an overnight polysomnography recording in the laboratory and subsequent daytime testing. Insomnia diagnosis was determined by a board-certified sleep specialist, and exclusion criteria included psychiatric history within past 6 months, other sleep disorders, significant medical conditions, and medications with significant effects on inflammation, autonomic function, or other psychotropic effects.
Results
Consistent with prior studies, we found elevated sleep spindle density and fast sigma power (14-16Hz). Despite no difference in beta or gamma band power when averaged across NREM sleep, time-frequency analysis centered on the peaks of detected spindles revealed a phasic elevation in spectral power surrounding the 28Hz harmonic peak in the insomnia group, especially for spindles coupled with slow waves. We also observed an overall pattern of time-locked delay in the 28Hz harmonic peak, occurring approximately 40 msec after spindle peaks. Furthermore, we observed a 42Hz ‘3rd harmonic’ peak, not yet predicted by the existing modeling work, which was also elevated for insomnia.
Conclusion
In conjunction with existing mathematical modeling work that has linked sleep spindle harmonic peaks with thalamic relay nuclei as the primary generators of this EEG signature, these findings may enable novel insights into specific thalamocortical mechanisms of insomnia and non-restorative sleep.
Support (if any)
NIH 5T32HL007901-22
Endogenous memory reactivation during sleep in humans is clocked by slow oscillation-spindle complexes
