Thirty studies imaged the brain before and after CPAP. None of it replicates cleanly.

Obstructive sleep apnoea is associated with structural, microstructural, functional, vascular and metabolic alterations on brain imaging [s1]. Continuous positive airway pressure is the first-line treatment. The obvious question — whether treating the apnoea reverses the imaging findings — has been asked repeatedly, and a systematic review published in SLEEP on August 29 collects the answers [s1].

The answer is that the literature does not currently support one.

What was reviewed

The authors searched PubMed/MEDLINE, Scopus and Web of Science from database inception to December 2025 [s1]. Eligible studies enrolled adults with sleep apnoea who had brain MRI before and after CPAP, or who were compared against untreated, sham-treated, best-supportive-care or healthy control groups [s1].

The MRI outcomes covered were broad: structural, diffusion, functional, perfusion, cerebrovascular reactivity, magnetic resonance spectroscopy, susceptibility-related and perivascular-space measures [s1]. Risk of bias was assessed with an adapted Newcastle–Ottawa Scale [s1].

Thirty studies met criteria [s1].

The finding is the heterogeneity

Reported post-CPAP MRI changes were heterogeneous [s1]. They included grey matter or cortical alterations, white matter microstructural changes, modified functional connectivity and regional activity, perfusion or cerebrovascular reactivity changes, and preliminary metabolic, susceptibility-related and perivascular-space findings [s1].

That list reads like a summary of positive results, and this is exactly where such reviews are misread. Results varied across MRI techniques, anatomical targets, CPAP duration, adherence definitions, comparator groups and analytical methods [s1]. Because of that heterogeneity, the authors synthesised findings qualitatively rather than pooling them [s1].

Their conclusion is that the available longitudinal MRI literature suggests possible domain-specific brain changes after CPAP, but the evidence is largely exploratory, because most studies were small and non-randomised and used heterogeneous imaging and analytical approaches [s1]. The findings do not establish reproducible MRI treatment effects [s1].

Why "no reproducible effect" is not "no effect"

The distinction is worth holding precisely, because it cuts both ways.

Thirty small studies, each free to choose its imaging modality, its region of interest, its definition of adequate CPAP adherence and its comparison group, will produce scattered positive findings whether or not there is a real effect. That is a property of the design space, not evidence about the brain. Nothing in this review shows CPAP does not change the brain; it shows that the existing studies cannot tell you whether it does.

What the review implicitly specifies is what a study capable of answering would look like: prespecified modality and region, a randomised or sham comparator, an objective adherence threshold fixed in advance, and enough participants that a null is informative. None of the thirty appears to have combined all of those.

What is known about the clinical stakes

The imaging question is a proxy for a clinical one — whether untreated or poorly treated sleep apnoea damages cognition over time. On that, the observational evidence is stronger than the imaging evidence, and a cohort study published in the same journal four days earlier illustrates the level at which the field can currently detect something.

Using electronic health records from the TriNetX US Collaborative Network, researchers identified adults with a first diagnosis of obstructive sleep apnoea between 2016 and 2019 and followed them for up to five years [s2]. Patients with comorbid insomnia and sleep apnoea — insomnia diagnosed within 90 days of the apnoea index date — were matched 1:1 by propensity score to patients with apnoea alone, yielding 51,043 in each group [s2].

Comorbid insomnia was associated with higher risk of all-cause dementia over five years (hazard ratio 1.34; 95% CI 1.24–1.46) [s2]. Risk elevations were consistent across subtypes: vascular dementia (HR 1.35; 95% CI 1.13–1.60), dementia with Lewy bodies (HR 1.85; 95% CI 1.22–2.81) and Alzheimer's disease (HR 1.19; 95% CI 1.01–1.40) [s2]. Risk separation appeared within the first year and persisted [s2].

Separation inside the first year is a caution as much as a finding. Dementia has a long preclinical phase, and insomnia is among its early symptoms, so some of that gap is plausibly reverse causation rather than harm caused by the insomnia. The authors' claim is limited to the clinical relevance of recognising the comorbid presentation as a high-risk one [s2].

What to watch

Whether any adequately powered randomised imaging trial is registered. Until then, claims that CPAP "restores" brain structure rest on a literature its own systematic reviewers describe as exploratory.

This article is informational and is not medical advice. Decisions about sleep apnoea treatment belong with a clinician who knows the individual case.

Sources

• Brain MRI Changes Following Continuous Positive Airway Pressure Therapy in Adults with Obstructive Sleep Apnea: A Systematic Review — SLEEP, 2026-08-29
• Dementia Risk in Patients with Comorbid Insomnia and Obstructive Sleep Apnea: A Propensity Score-Matched Cohort Study — SLEEP, 2026-08-25