Researchers working with mouse models of Alzheimer’s disease found that depleting the brain’s immune cells, called microglia, restored NREM sleep to near-normal levels even though amyloid plaques remained intact. The finding, reported in the journal Alzheimer’s and Dementia using APP/PS1 transgenic mice, separates sleep rescue from plaque clearance and points to microglial overactivity, not plaques alone, as a direct driver of the sleep disruption that accompanies the disease. A parallel line of evidence from 5XFAD mice treated with a dual orexin receptor antagonist showed a similar split: sleep improved while plaque density and neuroinflammation stayed unchanged.
Why restoring sleep without clearing plaques changes the research calculus
For more than a decade, the dominant assumption in Alzheimer’s sleep research has been that amyloid buildup causes poor sleep, and that removing amyloid should fix it. A small NIH human study reinforced that link by showing that acute sleep loss raises beta-amyloid levels in the brain. Earlier mouse work demonstrated that the sleep-wake cycle regulates interstitial fluid amyloid-beta dynamics, and that chronic sleep restriction can accelerate plaque formation, according to research in Science. Those findings cemented a feedback-loop model: poor sleep drives amyloid accumulation, which in turn worsens sleep.
The new APP/PS1 data complicate that picture. NREM sleep loss appeared when plaques first emerged, but it did not worsen as plaque burden climbed higher over time. That plateau suggests plaques trigger sleep disruption at an early threshold and that something else sustains it. The researchers traced that sustained disruption to microglia, the brain’s resident immune cells, which expanded well beyond the regions where plaques had formed. When the team used CSF1R inhibition to deplete those microglia, sleep recovered, yet plaques remained in place.
This dissociation matters because every major anti-amyloid therapy approved or in trials, from lecanemab to donanemab, targets plaque removal. If sleep loss is driven primarily by microglial behavior rather than by plaques themselves, then clearing amyloid may not be enough to restore the sleep that patients and caregivers rank among their most disabling symptoms. It also suggests that sleep disturbance could persist or even progress in patients whose amyloid burden has been substantially lowered, unless parallel strategies address the inflammatory and circuit-level components of the disease.
Microglial depletion and orexin blockade both rescue sleep but leave plaques untouched
Two independent experimental strategies now converge on the same conclusion. In the APP/PS1 model, CSF1R-mediated microglial depletion restored NREM sleep without reducing amyloid pathology. Separately, in 5XFAD mice, the dual orexin receptor antagonist DORA-22 increased light-phase sleep while leaving amyloid-beta levels, plaque density, and neuroinflammatory markers unchanged, according to findings in Alzheimer’s Research & Therapy.
The two approaches act through different mechanisms. CSF1R inhibition removes microglia from the equation entirely, eliminating the inflammatory signaling that appears to disrupt sleep-regulating circuits. In the APP/PS1 study, microglial depletion was associated with more consolidated NREM episodes and fewer spontaneous awakenings, even though histological analysis showed persistent plaque deposits. That pattern implies that the presence of amyloid is not sufficient to fragment sleep once the inflammatory milieu is dialed down.
DORA-22 works downstream, blocking orexin receptors that promote wakefulness and thereby shifting the sleep-wake balance pharmacologically. In 5XFAD mice, the drug increased total sleep time during the light phase, improved bout length, and did so without measurable changes in amyloid-beta biochemistry or glial activation. The fact that sleep improved while standard neuropathological readouts remained static reinforces the idea that sleep circuitry can be modulated independently of plaque burden.
Neither strategy touches plaques, yet both recover sleep. That convergence strengthens the case that sleep disruption in Alzheimer’s models is not a passive byproduct of plaque accumulation but an active process maintained by cellular and molecular pathways that can be targeted independently. It also raises the prospect that different nodes in this network-immune signaling, orexinergic tone, perhaps synaptic plasticity in thalamocortical loops-could be leveraged in combination to stabilize sleep architecture even in the continued presence of amyloid pathology.
A natural question follows: would combining the two produce additive gains? If microglial depletion addresses the inflammatory driver and orexin blockade adjusts the arousal circuit, a combined protocol could theoretically push NREM sleep continuity closer to wild-type levels. No published data test that combination in 5XFAD or APP/PS1 mice, but the mechanistic logic is straightforward: one intervention removes a major upstream source of cytokines and synaptic pruning, while the other directly counters hyperarousal. Whether such a dual approach would yield better sleep without compromising safety remains an open experimental question.
Gaps between mouse sleep rescue and human treatment
Several unresolved questions limit how far these results can travel toward the clinic. The primary APP/PS1 study does not report exact hours of restored sleep or detailed statistical effect sizes for the CSF1R depletion intervention, making it difficult to gauge the magnitude of the benefit. Without those metrics, it is unclear whether the change corresponds to a modest shift in sleep efficiency or a near-complete normalization of NREM structure. Long-term cognitive or survival outcomes after sustained microglial depletion in amyloid models have not been reported in the cited work, and depleting the brain’s immune cells for extended periods carries obvious safety concerns that mouse lifespans may not fully reveal.
Microglia play essential roles in synaptic maintenance, debris clearance, and host defense. Chronic CSF1R inhibition could plausibly increase vulnerability to infections, impair synaptic remodeling, or introduce subtle neurodevelopmental effects if used preventively. Translating such an approach to humans would likely require more nuanced modulation-tamping down specific inflammatory pathways or reprogramming microglial states-rather than broadly eliminating the cells.
The DORA-22 results also carry a notable tension. While that compound failed to reduce plaque density or neuroinflammation in 5XFAD mice, earlier work in a different amyloid model reported that dual orexin receptor antagonism decreased plaque formation. The discrepancy may reflect differences in mouse strain, drug dosing, treatment timing, or sensitivity of the assays used to quantify amyloid. It is also possible that orexin signaling influences amyloid dynamics only within a narrow temporal window, such as the pre-plaque phase, and that once pathology is established, blocking orexin primarily affects sleep without reshaping plaque trajectories.
For clinicians, the most immediate translational link lies with existing dual orexin receptor antagonists already approved for insomnia. These drugs have established safety profiles in older adults, but they have not been systematically evaluated in people with Alzheimer’s disease, who may have altered pharmacokinetics, polypharmacy, and heightened fall risk. The mouse data suggest that improving sleep via orexin blockade is unlikely to worsen amyloid pathology and might provide symptomatic relief, yet whether such interventions meaningfully alter cognitive decline, neuropsychiatric symptoms, or caregiver burden remains unknown.
On the research side, the new findings argue for more granular sleep phenotyping in Alzheimer’s trials. Rather than treating sleep as a secondary or exploratory endpoint, future studies could incorporate polysomnography, spectral analysis of NREM slow waves, and longitudinal tracking of sleep fragmentation alongside biomarkers of amyloid, tau, and neuroinflammation. Such designs would help clarify whether modulating microglia or orexinergic tone produces sleep gains that translate into better daytime function or slower neurodegeneration.
Ultimately, the separation of sleep rescue from plaque clearance reframes how investigators think about symptoms versus pathology in Alzheimer’s disease. Amyloid may set the stage, but ongoing dysfunction in immune and arousal circuits appears to sustain one of the condition’s most debilitating features. That distinction opens a therapeutic window: even if disease-modifying agents only partially curb amyloid, targeted interventions on microglia and orexin pathways could still deliver meaningful improvements in how patients sleep-and, by extension, how they live with the disease.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
