Which Are the Most Reliable Sleep Parameters that Predict Cognitive Decline and Alzheimer’s Disease?

In the last years, a bidirectional relationship between sleep and neurodegenerative disorders has been underlined by a large number of studies [1]. It is well known that sleep disorders are frequent and can be the first manifestation of underlying neurodegenerative diseases. Moreover, sleep changes or disorders can represent an independent risk factor for neurodegeneration. A meta-analysis published in 2018 that included 18 prospective studies (with an average follow up of 9.5 years in about 250,000 subjects) showed that those with sleep disturbance at baseline had a higher risk of incident all-cause dementia compared with those not reporting disturbed sleep [2]. A complaint of insomnia was associated only with incident Alzheimer’s disease (AD), while sleep disordered breathing was a risk factor for all cause dementia. A more recent nationwide cohort study on 1,491,276 people evaluated the association between sleep disorders diagnosed after age 50 years in the secondary healthcare sector and subsequent late onset dementia development. The authors found that subjects with any sleep disorder had a 17% higher risk of dementia compared to those with no sleep disorder [3]. These findings suggest that sleep disorders could be considered a modifiable risk factor for dementia, and be added to the other well-known factors, as depression, physical inactivity, untreated hypertension, diabetes, tobacco use, and obesity [4].

Excluding specific sleep disorders, several studies showed that some changes of sleep macrostructure (i.e., sleep duration, NREM and REM stages) and sleep microstructure (i.e., arousals, cyclic alternating pattern, spindles, and K-complex) may be observed in patients with neurodegenerative disorders compared to age-matched controls. Unfortunately, the majority of data have been reported in cross-sectional investigations.

In their systematic review [5], Carpi and colleagues reported the results of longitudinal studies investigating if objective sleep parameters measured by polysomnography or actigraphy can predict dementia or AD-related neurodegeneration. From the initial sample of 933 identified records, a total of 17 records were included in the review.

Concerning sleep duration, a medium amount of sleep of seven hours seems to be a protective factor against the cognitive decline. This finding has been confirmed in the 7,959 participants of the Whitehall II study, with a 25-year follow up: measurement of sleep duration measured by an accelerometer at age 50, 60, and 70 years showed a higher risk of dementia with a sleep duration of six hours or less compared to seven hours [6]. In the same study, in contrast to some previous studies, the authors did not find evidence to support the hypothesis that long sleep duration is a risk factor for dementia.

In the review of Carpi and colleagues [5], also findings of sleep fragmentation, as increased wake after sleep onset or increased night-time activity seem to be linked to higher dementia risk. Interestingly, these aspects were observed especially in studies using actigraphy rather than polysomnography. This could be related to the possibility of multiple consecutive night evaluation by actigraphy in contrast to one-night polysomnography. Can we consider the polysomnographic findings of just one night reliable? Concerning sleep architecture, it is known that a glymphatic system clears the brain of protein waste products, especially during slow wave sleep (SWS) [7]. Specific disruption of SWS has been shown to increase amyloid-β levels and neuroinflammation in AD animal models [8].

No associations between baseline SWS percentage and dementia at follow-up have been observed in the longitudinal studies. Moreover, some authors found that a lower percentage of REM sleep [9, 10], as well a higher REM latency in the baseline condition [8] was associated to an increased risk of dementia or more pronounced worsening of cognitive functioning. Cholinergic innervation is fundamental for promoting REM sleep; cholinergic dysfunction has clearly been associated with cognitive deficits; and reduced cholinergic innervation has been found in patients with early stages of AD [11].

In order to define if cognitive aging or dementia are associated with levels of SWS or REM sleep, a more extended evaluation of sleep macrostructure with a polysomnographic recording of consecutive nights is needed. An example of enhanced duration of sleep data collection (7 nights of homebased multi-channel EEG sleep recording) has been recently reported in a protocol study publication [12].

Concerning sleep microstructure, Carpi and colleagues [5] found three studies that used EEG spectral power analysis across conventional frequency bands: higher alpha and theta power in NREM sleep, higher alpha and sigma power in REM sleep resulted associated with cognitive decline, however with some inconsistencies across studies. Probably, future prospective studies should evaluate other parameters, such as spindle activity, coupling between spindles and slow wave oscillations, and multiscale fluctuation dispersion entropy that provides a sensitive non-linear measure of EEG information content across a range of biologically relevant time-scales [13].

Carpi and co-workers [5] also evaluated the indexes of sleep-disordered-breathing as possible biomarkers. Some studies reported associations between baseline apnea-hypopnea index or oxygen desaturation index and subsequent cognitive decline, while others did not find such relation. A possible explanation could be the additional impact of EEG arousals that can contribute to metabolic derangements, inflammation, and oxidative stress. Moreover, the concomitant presence of depression or anxiety in sleep apnea patients may potentially influence cognitive functions [14], but these psychiatric conditions are usually neglected as comorbidities.

Carpi et al [5] concluded that some studies support the link between abnormal objectively-measure sleep parameters and risk of subsequent cognitive impairment, but the full evidence of specific sleep biomarkers is still limited. On the other hand, the high heterogeneity in the methodology of the studies included in their review may explain some discrepancies. Actigraphic monitoring is useful for the evaluation of sleep over many nights, as well as of rhythmic variations in sleep-wake cycles. This is crucial since association between evening-preference chronotype and dementia risk has been reported [15]. However, actigraphy does not allow to identify NREM and REM stages, and microstructural sleep parameters that may have a role as biomarkers. The combination of actigraphic recordings with sleep EEG monitoring for some consecutive days should be considered in future research. In-lab polysomnography has some limits including costs and unnatural settings, but over recent years, comparable performance has been demonstrated by portable EEG devices [16]. At-home single-channel sleep EEG could be used for the early identification of individuals at risk for cognitive decline [17].

The data already available suggest that sleep disorders may be a red flag for possible early phase dementia disorders, and they may contribute to the early identification of individuals at risk for cognitive decline [18]. However, further research is necessary to identify reliable sleep parameters for predicting cognitive impairment and AD.