10 Sep 2021

People who log seven to eight hours of sleep per night have less amyloid and perform better on memory and other tests than do people who reported sleeping for six hours or less. This is the main finding from a study published August 30 in JAMA Neurology. Those who spent the most time under the covers—longer than nine hours per night—had different problems, doing poorly on tests of executive function and on daily tasks. At 4,417 participants, the study was the largest yet to relate sleep duration to both amyloid burden and cognition.

In A4 trial, more than 4,400 people reported their sleep habits.
Short sleep meant more plaques and memory deficits.
Excessive sleep came with poor executive function.

Although the findings hint at links between sleep patterns and Alzheimer's pathophysiology, this cross-sectional study cannot disentangle cause and effect. Led by Joseph Winer and Elizabeth Mormino at Stanford University in Palo Alto, California, the study also found that race or ethnicity as well as health and lifestyle factors influenced sleep patterns, complicating interpretation of these association data.

Prior epidemiological research has supported the idea of a sweet spot for shut-eye. Both too much and too little sleep have been tied to cognitive impairment, producing a classic U-shaped curve where just the right amount of zzz’s appears protective (Kronholm et al., 2009; Ma et al., 2020). Understanding the mechanisms underlying these patterns has been difficult, although some small studies have tied insufficient sleep to elevated Aβ in the brain (Spira et al., 2013; Sep 2009 news; Jan 2018 news).

Winer and colleagues wanted to measure in a larger sample how sleep duration relates to both cognitive performance and amyloid burden. To do that, they tapped screening data from A4—an ongoing secondary prevention trial evaluating the anti-Aβ antibody solanezumab in cognitively unimpaired people who had signs of brain amyloid (Jan 2013 news; Jun 2017 news). As part of the massive recruiting effort for the trial, people whose cognition was normal were screened with an amyloid-PET scan and filled out health questionnaires that included questions about their sleep habits. Specifically, participants were asked how long they slept each night, and whether, and for how long, they napped during the day.

The researchers categorized participants into three groups: short duration, who reported getting on average six or fewer hours sleep per night; normal, i.e., seven to eight hours; and long, at longer than nine hours. The scientists acknowledge that, as opposed to objective measures such as actigraphy or EEG, self-reported sleep duration is not highly accurate, does not reflect sleep quality or disruption, and can change over time.

First, the researchers analyzed how self-reported sleep duration related to amyloid burden among these participants. They found that the fewer hours people reported sleeping each night, the more brain amyloid they had. Broken down by group, short sleepers, but not long sleepers, had slightly more amyloid than did normal-length sleepers (see image below).

**Less Sleep More Plaque.** Amyloid burden correlated with hours of sleep per night. *[Courtesy of Winer et al., JAMA Neurology, 2021.]*

What about cognition? Compared to people who reportedly got the recommended seven to eight hours of sleep each night, those who snoozed less than six hours scored slightly lower both on the mini mental state examination (MMSE) and on the logical memory delayed recall test. Short sleepers were also less likely to achieve the maximum score on the free and cued selective reminding test, a measure of episodic memory. Long sleepers, on the other hand, fell short on the digit symbol substitution test, a measure of executive function. Scores on the cognitive function instrument, a self- and partner-reported measure of everyday functioning, were lower in both short and long sleepers than in normal sleepers (Mar 2015 news).

Health and lifestyle factors also appeared to play into how much people reported sleeping. Both short and long sleepers were more likely to be depressed, to have a higher body mass index, and to take daytime naps than normal-length sleepers. The association between alcohol consumption and sleep duration was particularly robust. People who slept more than nine hours a night drank more than normal or short sleepers. This supports the idea that different etiologies may underlie the cognitive slips among long- and short-duration sleepers.

Winer emphasized that this data cannot prove what causes what. That said, the findings do imply that short sleep duration aligns with both the amyloid pathology and the subtle memory deficits that crop up in the preclinical stage of AD. In contrast, the reasons why people spend excessive amounts of time in bed at night appear unrelated to amyloid; they may reflect other health issues.

The scientists also found that sleep duration differed markedly by race and ethnicity. Compared to non-Hispanic white people, participants who self-identified as non-Hispanic black, African American, Asian, Latino, or Hispanic white all slept less. The difference was largest for non-Hispanic black people and African Americans, who reported sleeping 38 minutes less than did non-Hispanic white people per night. The findings jibe with previous studies that have tied racial and ethnic differences to sleep patterns (Kingsbury et al., 2013; Carnethon et al., 2016; Grandner et al., 2016).

Sigrid Veasey of the University of Pennsylvania School of Medicine in Philadelphia pointed out that short and long sleep are not two distinct conditions or behaviors; rather, they are symptoms of diverse sleep behaviors and disorders, such as sleep apnea. “The higher body mass indices and depression scores, and the presence of daytime napping for both short and long sleepers, are consistent with obstructive sleep apnea,” Veasey said. She thinks it will be important to understand the underlying causes of different sleep durations, and then identify how those underlying causes relate to amyloid burden and cognitive function.

Erik Musiek of Washington University in St. Louis noted that the large numbers of people with amyloid PET imaging was a major strength of the study, though it was limited by its assessments at a single time point. “This study provides more strong evidence that brain amyloid is associated with short sleep in cognitively normal individuals, even though directionality cannot be proven,” he wrote. “The integration of objective sleep measurements with multimodal AD biomarkers in longitudinal studies of preclinical AD is the ultimate goal, and this study is an important step in that direction.”—Jessica Shugart

Comments

- Sigrid Veasey
  University of Pennsylvania
- Posted: 10 Sep 2021
This large cross-sectional study does a nice job of extending our understanding of sleep duration and Alzheimer's risk. The data suggest that the steeper cognitive decline associated with long sleep may be an Aβ-independent process, as long sleep in this study did not predict Aβ burden.

It would be tempting to conclude that short sleep contributes to the observed associated increased amyloid burden, but it is important to understand that short and long sleep are not two distinct conditions or behaviors, but rather are symptoms of diverse sleep behaviors and disorders. For example, both short sleep and long sleep frequently occur in untreated obstructive sleep apnea. The former occurs when sympathetic activity is high and the latter when sleep is disrupted without significant surges in catecholamines. The higher body mass indices and depression scores and the presence of daytime napping for both short and long sleepers are consistent with obstructive sleep apnea risk for both short and long sleepers. Similarly, sleep duration, as confirmed here, is increased in women and with higher education.

This being a cross-sectional study, it is also possible that the increased Aβ is increasing brain excitation and thereby disturbing sleep.

It is critical now to further develop the sleep/wake phenotypes of short and long sleepers, to identify underlying causes, and then to analyze identified causes for amyloid burden risk and cognitive decline rather than the symptom of sleep duration.

View all comments by Sigrid Veasey

- Erik Musiek
  Washington University School of Medicine
- Posted: 10 Sep 2021
This nice study is consistent with several smaller previous studies linking self-reported sleep issues (in this case, short sleep duration) with brain amyloid level as assessed by amyloid PET. This study has a very large number of people with amyloid PET imaging, and also accounts for other lifestyle factors, which are its major strengths.

Several previous studies have suggested that reported short sleep duration is associated with increased risk of future dementia, and some have shown that long sleep durations also have increased risk. This paper does a nice job of noting that short and long sleep are associated with other health and lifestyle factors that might also influence amyloid deposition and AD risk (such as increased BMI, alcohol use, etc.).

It is very difficult to disentangle these factors. Self-report of sleep duration can only go so far, as objective measures such as actigraphy (better) or EEG (best) are really needed to get the detailed sleep quality data needed to fully understand the relationship between sleep and AD.

The question of which comes first, sleep problems or amyloidosis, remains unanswered by this study, as it is cross-sectional. There is no marker of tau in this study, though brain tau deposition is associated with amyloid. Thus, we don’t know if it is amyloid or tau (or another amyloid-related neuropathology) that drives sleep changes, or vice versa.

However, this study provides more strong evidence that brain amyloid is associated with short sleep in cognitively normal individuals, even though directionality cannot be proven. The integration of objective sleep measurements with multimodal AD biomarkers in longitudinal studies of preclinical AD is the ultimate goal, and this study is an important step toward that goal.

View all comments by Erik Musiek

- Barbara Bendlin
  University of Wisonsin
- Posted: 10 Sep 2021
Of the many potentially modifiable factors that putatively impact dementia risk, sleep is one of the more biologically plausible ones to be linked with AD pathology specifically. Rodent studies indicate that Aβ levels rise during wakefulness, and Aβ appears to be cleared during NREM sleep. Several human studies have also linked sleep characteristics, and disrupted sleep, with biomarkers of amyloid and tau.

Our research group previously found that self-reported sleep quality among cognitively unimpaired individuals is associated with amyloid pathology, but these studies, using CSF and amyloid PET, were rather small. The size of the dataset included here is impressive, and a major strength of the current analysis.

The caveat of course, is that we can't infer from this analysis whether shorter sleep duration is exacerbating amyloid accumulation, or whether individuals who harbor preclinical amyloid are sleeping less due to their accumulating AD pathology. Animal studies suggest that the sleep/amyloid relationship is bidirectional.

It's also important to consider that there are other pathologies that co-occur with amyloid that could be playing a role. Longitudinal human studies where sleep is well-characterized, and which incorporate comprehensive AD biomarkers, could shed further light on the mechanisms at play.

Sleep intervention studies will also tell us whether sleep can be modified to reduce dementia risk.

View all comments by Barbara Bendlin

- Brendan Lucey
  Washington University School of Medicine
- Posted: 14 Sep 2021
This study is a very important contribution to the literature and builds on previous research showing a non-linear, or U-shaped, relationship between sleep and cognitive performance.

Although short or long sleep duration likely represent poor sleep quality, multiple factors may lead to disturbed sleep in older adults, and this cross-sectional study is unable to sort out the potential cause(s) of sleep disturbance. Several of the factors identified in this study, such as depressive symptoms as well as sleep, are potential targets for intervention trials. However, longitudinal studies and more objective sleep measures in well-characterized cohorts are first needed to sort out these complex relationships.

View all comments by Brendan Lucey