Obstructive Sleep Apnea and Alzheimer’s Disease Pathology: Is Sleep Architecture the Missing Key?

Obstructive sleep apnea (OSA) body of literature is growing in relation to cognitive decline and Alzheimer’s disease (AD) [1–5]. OSA can be described as a respiratory condition in which repeated episodes of partial hypopneas and complete apnea obstructions of the upper airway during sleep [6, 7]. Untreated OSA has been linked with cognitive deficits in domains of vigilance, psychomotor speed, attention, executive functioning, and smaller evidence relating to memory and language [8, 9]. These impairments may be due to disrupted sleep in individuals with OSA resulting in reduced slow wave sleep (SWS), intermittent hypoxemia, and excessive day time sleepiness which are correlated with neuronal death [10–13]. Another mechanism that may contribute to cognitive deficits is the accumulation of intracerebral senile amyloid-β (Aβ) plaques, which is a hallmark pathology of AD [14, 15]. Sleep quality and Aβ burden may be bi-directional, SWS deprivation is associated with increased accumulation of Aβ plaques and the presence of Aβ plaques is related with ensuing reduction in SWS [16, 17]. Recent human investigations have identified associations between sleep disruption, including OSA, augmented Aβ accrual in the blood, cerebrospinal fluid and brain [18–21]. However, it is unclear whether Aβ burden in middle-aged individuals with OSA is correlated with extensive deficits in cognition. Additionally, there is a dearth in data examining the associations between nocturnal hypoxemia or sleep disruptions (during SWS) and Aβ burden in individuals with OSA.

Cavuoto and associates [7] sought to examine the differential impacts of hypoxemia, SWS disruption and amyloid burden on cognition in 34 individuals with confirmed OSA and 12 healthy controls. Subjects underwent a clinical polysomnogram either in-lab or at-home, a NAV4694 positron emission tomography (PET) scan for Aβ burden, assessment of APOE 4 status and cognitive assessments. The results demonstrated that Aβ burden was associated with nocturnal hypoxemia, impaired verbal episodic memory, set shifting and information processing speed. Hypoxemia and truncated SWS were correlated with impaired autobiographical memory and verbal episodic memory, respectively [7].

We commend the authors for using objective measures to examine both sleep and cognition, most investigations use subjective measures evaluating sleep, cognition, or both [1, 22–24]. Objective measures are more reliable and valid when examining OSA and AD pathology compared to subjective measures and make the results more impactful in sleep and cognition research. It is also important to use polysomnography (PSG) and PET imaging because they are both the gold standard clinical measures for evaluating sleep and amyloid burden in the brain, respectively. This study also added to the body of literature on cortical Aβ burden being correlated with nocturnal hypoxemia and impaired cognitive functioning in middle-aged OSA preclinical population; while additionally finding relationships between hypoxemia, truncated SWS, reduced autobiographical and verbal memory.

While the strengths of this study are commendable, this investigation does have glaring deficiencies: 1) the sample size was small with 34 individuals with confirmed OSA diagnoses and only 12 controls, these findings are hard to generalize to wider populations; 2) the investigators included participants using PSG at-home as well as laboratory PSG testing to identify OSA which hinders consistency. An intraclass correlation showing strong internal reliability between in-lab versus at-home sleep metrics could have strengthened these results. While the authors may have utilized similar administration techniques in both cases, a participant may sleep unquestionably different at-home versus in a laboratory. Additionally, the participant may still use equipment incorrectly at-home versus an administrator in the laboratory; 3) Also, using cross-sectional data, while insightful, limits the impact of the findings. Examining these relationships longitudinally will provide insightful meaning behind Aβ burden, nocturnal hypoxemia, and cognitive functioning which could then help foster personalized treatments reducing OSA and AD risk.

It is vital for both cross-sectional and longitudinal studies of sleep and AD to use assessment measures that are consistent across the data, whether the assessment methods are subjective or objective. Subjective sleep measures will include self-report using validated sleep questionnaires, e.g., Pittsburgh Sleep Quality Index, Epworth Sleepiness Scale, Insomnia Severity Index, etc. Objective measures include actigraph, WatchPAT, and nocturnal PSG (in-lab and at-home). With the advent of digital health revolution, sleep wearable technology produces many novel, highly sophisticated and inexpensive consumer devices. These devices collect data from multiple sensors and claim to extract information about users’ behaviors, including sleep. However, there is currently little to no guidance within and outside the scientific sleep community for their use, therefore their validity and application are in question. Importantly, where aggregate data is necessary, appropriate methodology for harmonization as well as conducting sensitivity analysis becomes expedient for data validity and interpretation.

A recent study by Sharma and colleagues [2], showed that OSA was associated with greater amyloid burden over two years. However, this study did not examine sleep fragmentation or intermittent hypoxia like Cavuoto et al. (2023) [7]. However, identifying the relationships longitudinally provides more support for mitigation efforts if we know OSA is driving amyloid burden and cognitive deficits. Additional studies like Bubu et al. [3] and Hogan et al. [4] using the Alzheimer’s Disease Neuroimaging Initiative dataset have found that self-report OSA acts in a synergistic manner with Aβ and tau, leading to earlier progression of AD-mild cognitive impairment and dementia compared to non-OSA participants. While there is strength in these studies being longitudinal and overall large sample size, the ADNI studies were limited by self-report OSA and limited number of OSA Aβ+ participants.

Cavuoto and colleagues showed impaired SWS was associated with reduced autobiographical and verbal memory but not Aβ burden. This is an important finding with implications, as studies show ethnic and racial differences in SWS and other sleep architecture [25–27]. Black individuals typically take longer to fall asleep, have lower sleep efficiency [26–31], spend a smaller proportion of time in non-rapid eye movement SWS [28–30, 32–36]. Importantly, most studies that have examined sleep architecture in Black individuals and other minoritized groups have done so in individuals between 20–65 years old. Studies are needed that examine how race, age, and sleep architecture interact in older adults both cross-sectionally and longitudinally, and how these impact AD risk, pathology, and progression.

Notably Black subjects with OSA exhibiting reduced SWS typically have higher cerebrospinal fluid Aβ levels [10, 11]. Possible causes and mechanism to consider are: genetics, psychosocial factors, and/or vascular burden [37–43]. African ancestry has been linked to higher rates of hypertension and diabetes which is associated with deficits in SWS total time [38, 45]. Vascular dysfunction may reduce the clearance of Aβ via the blood-brain-barrier or indirectly increase Aβ deposition. Moreover stress, over the course of a lifetime, is a key facilitator in vascular risk in minorities, causing comorbidities like obesity [41]. Obesity is a risk factor for OSA and AD [19, 47]. Thus, vascular dysfunction, psychosocial factors, and genetics must be accounted for when examining SWS, AD, and Aβ accumulation.

Overall, by leveraging objective measures, the results from Cavuoto et al. [7], provide further evidence of the relationship between AB burden, nocturnal hypoxia, SWS, and cognitive deficits in pre-clinical populations. Future studies in the field need to examine these relationships longitudinally with a large sample size; additionally, studies should include PET tau, Aβ, health disparities, vascular components, multiple cognitive domain tests, along with objective sleep measures including sleep fragmentation and hypoxia. Findings will provide additional rationale for personalized intervention methods.