Transient Sleep EEG Oscillation Dynamics Reveal Novel Changes in Brain State Across the Lifespan

2589 

Abstract Submission 

Michael Prerau¹, Thomas Possidente², Habiba Noamany², Mingjian He^2,3,4, Sophie Saremsky²

¹Brigham and Women's Hospital/Harvard Medical School, Boston, MA, ²Division of Sleep and Circadian Disorders, Brigham and Women’s Hospital, Boston, MA, ³MIT-HST Program, Cambridge, MA, ⁴Department of Anesthesia, Critical Care, and Pain Management, Massachusetts General Hospital, Boston, MA

Michael Prerau, Ph.D.  
Brigham and Women's Hospital/Harvard Medical School
Boston, MA

Thomas Possidente  
Division of Sleep and Circadian Disorders, Brigham and Women’s Hospital
Boston, MA

Habiba Noamany  
Division of Sleep and Circadian Disorders, Brigham and Women’s Hospital
Boston, MA

Mingjian He  
Division of Sleep and Circadian Disorders, Brigham and Women’s Hospital|MIT-HST Program|Department of Anesthesia, Critical Care, and Pain Management, Massachusetts General Hospital
Boston, MA|Cambridge, MA|Boston, MA

Sophie Saremsky  
Division of Sleep and Circadian Disorders, Brigham and Women’s Hospital
Boston, MA

Transient oscillatory events in the sleep electroencephalogram (EEG), such as sleep spindles, have been linked to memory consolidation, and changes in spindle activity have been associated with numerous psychiatric, neurodevelopmental, and neurodegenerative disorders. However, our recent studies have shown that traditional sleep spindles represent only a subset of a larger class of transient oscillatory events and that classes of spindle-like events at other frequencies, such as those occurring at low-alpha and theta ranges, can act as a biomarker of disorders such as schizophrenia. Moreover, we have shown that transient oscillation dynamics are fingerprint-like: highly heterogeneous between individuals, yet strongly consistent night-to-night. While sleep spindles change throughout the lifespan, there is little characterization of general classes of transient oscillations and their dynamics. In this study, we describe oscillation activity as a function of slow oscillation (SO) power (a continuous depth-of-sleep metric), and SO-coupled phase (a correlate of cortical up/down states) to expand our view of potential biomarkers in neurodevelopment and aging.

Transient oscillations were detected as time-frequency peaks (TF-peaks) in the sleep EEG spectrogram from the Cleveland Family Study (N=735, ages 7-89) using the DYNAM-O toolbox (sleepEEG.org), an extended implementation of methods described in detail in Stokes et. al 2022. In short: we first identify time-frequency peaks in the topography of the multitaper spectrogram of the sleep EEG using a variant of the watershed method. We then summarize the dynamics of the tens of thousands of detected TF-peaks per subject in visualizations called SO-power and SO-phase histograms, which plot TF-peak rate as a function of continuous depth of sleep (SO-power) and SO-phase. SO-power and phase histograms were computed for all CFS subjects and stratified by age. Statistical comparisons across population-averaged age groups were computed using FDR-corrected two-sample t-tests.

Significant differences in transient oscillation activity mode strength, frequency, SO-power, and SO-phase were observed across the lifespan. We identified stereotyped dynamics in sleep depth/frequency/amplitude in traditional "fast" spindle activity, rising from ~11 to 15 Hz, as well in a class of rising 6 to 10 Hz transient oscillation events during neurodevelopment, and the appearance of slow spindle activity at approximately 25 years of age. Moreover, we observe an initial mismatch between the frequencies of maximal SO-phase coupling and maximal transient oscillation density, which converge during development, suggesting the gradual emergence of coordination between networks essential for memory consolidation.

By characterizing general classes of transient oscillation activity across frequency, we identify unreported features of the sleep EEG and their associated dynamics throughout the lifespan. The framework constitutes a dramatic shift in approach for population analysis of sleep EEG activity-characterizing the distributional properties of general event classes rather than attempting to detect specific waveforms. These modal dynamics raise important mechanistic questions and act as new windows into the intersection of sleep physiology and the pathophysiology of related disorders. Finally, these dynamics provide a powerful new basis for the development of EEG biomarkers for tracking and detection of neurological disorders.

Early life, Adolescence, Aging

EEG/MEG Modeling and Analysis ²

Sleep and Wakefulness ¹

Aging

Computational Neuroscience

Data analysis

Development

Electroencephaolography (EEG)

Modeling

Sleep