Alzheimer’s disease disrupts deep sleep and gamma-band brain activity months before any measurable cognitive decline, according to a new preprint from researchers at VA Boston Healthcare System and Harvard Medical School. The study pinpoints two electrophysiological markers that could one day serve as early intervention targets: slow-wave power during NREM sleep and 40-Hz auditory responses linked to parvalbumin (PV) interneurons.
The sleep architecture that fails first
The team recorded longitudinal EEG and fiber photometry data from APP/PS1 mice, a model that accumulates amyloid-beta plaques, and compared them against non-AD littermates. By 3 months of age, before significant amyloid pathology had set in, the AD mice already showed reduced slow-wave power (0.5 to 4 Hz) in both hippocampus and medial prefrontal cortex. The deficit was concentrated in the delta band (1.5 to 4 Hz), the hallmark of restorative deep NREM sleep.
Beta power (15 to 30 Hz), a frequency band linked to hyperarousal and insomnia in humans, was significantly elevated across all sleep-wake states in the AD mice. The animals did not compensate for lost deep sleep by spending more time asleep; instead, NREM sleep during the dark (active) phase was slightly but significantly lower in the AD group.
Gamma responses and PV interneuron dysfunction
Beyond sleep architecture, the study examined the functional integrity of parvalbumin-positive interneurons, a class of fast-spiking cells that generate gamma-band (30 to 80 Hz) oscillations. Gamma activity is known to be disrupted early in Alzheimer’s and correlates with amyloid burden and cognitive decline.
Using auditory 40-Hz steady-state stimulation, the researchers found that evoked responses were impaired in AD mice by 3 months of age. Fiber photometry recordings of calcium activity in hippocampal PV neurons confirmed that the cellular response to 40-Hz stimulation was blunted at the same early time point.
Cognitive performance still normal
Despite these clear electrophysiological deficits, the AD mice performed no differently from controls on the Y-maze test at either 3 or 6 months of age, indicating that sleep and gamma abnormalities precede measurable cognitive impairment.
Why it matters
The finding that deep sleep loss and PV interneuron dysfunction emerge before amyloid pathology and cognitive decline opens a potential therapeutic window. Monitoring slow-wave power and 40-Hz evoked responses could enable risk stratification and early intervention in preclinical Alzheimer’s, before symptoms appear. The authors suggest these metrics may be “amenable to early intervention.”
Limits
This is a preprint posted on bioRxiv and has not yet undergone peer review. The findings come from a transgenic mouse model (APP/PS1) and may not fully translate to sporadic, late-onset Alzheimer’s in humans. The study did not test whether correcting sleep or gamma deficits alters disease progression.
Bottom line
Loss of deep NREM sleep and impaired PV-driven gamma activity are among the earliest detectable changes in a mouse model of Alzheimer’s, appearing well before amyloid plaques or cognitive symptoms. These markers could guide future early-intervention strategies.
Source
Katsuki F, McNally JM, Gerashchenko D, Uygun DS, Tyler A, McCoy JG, McKenna JT, Brown RE. Early Loss of Deep Restorative Sleep and Auditory Stimulus Evoked 40-Hz activity of Hippocampal Parvalbumin Neurons in the APP/PS1 Mouse Model of Alzheimer’s Disease. bioRxiv [Preprint]. 2026 Jul 1:2026.05.26.725476. DOI: 10.64898/2026.05.26.725476. PMID: 42427748.