Mounting evidence suggests a link between β-amyloid and disrupted sleep. Does the same hold true for tau? Yes, according to a paper published in the January 9 Science Translational Medicine. Using PET imaging, scientists led by Brendan Lucey and David Holtzman, Washington University School of Medicine, St. Louis, correlated these two hallmarks of AD pathology with sleep quality in people who were cognitively normal or very mildly impaired. The researchers found deficient slow-wave activity in deep sleep stages for those with pathology, particularly tau tangles. These synchronized neuronal oscillations are thought to be important for overnight memory consolidation. What’s more, people with tau deposits slept longer at night and took more naps during the day, indicating their sleep quality suffered. The results hint that these subtle sleep disruptions could be early biomarkers for tau deposition, ones that could help monitor patients at risk for AD.

  • People with tau tangles get less slow-wave-activity sleep.
  • Diminished SWA also correlated with Aβ pathology, but to a lesser extent.
  • More frequent daytime naps went hand-in-hand with tau pathology.

“This is an important paper because it shows for the first time how tau is related to sleep deficits,” said Bryce Mander, University of California, Irvine. “That’s going to be important in advancing our understanding of how AD pathology affects sleep.”

Accumulation of Aβ in the brain associates with excessive daytime sleepiness, more naps, and reduced sleep efficiency (Spira et al., 2013; Carvalho
et al., 2018Ju et al., 2013). Aβ plaque in the medial prefrontal cortex also correlates with less slow-wave activity (SWA) during non-REM sleep and worse memory consolidation (Mander et al., 2015). NREM SWA comprises synchronized neuronal oscillations in the 1-4.5 Hz range. A group led by Holtzman had previously reported that tau pathology disrupted NREM SWA in P301S mice (Holth et al., 2017). However, until now, no one had looked at the relationship between NREM SWA and neurofibrillary tangles in people.

Tau and Sleep Rhythms. Lateral (top) and medial (bottom) views of the brain show how flortaucipir binding (color scale) correlates with lower slow-wave activity during NREM sleep. Impoverished wave activity in the low 1-2 Hz range is most strongly correlated with tau pathology. [Courtesy of Science Translational Medicine/AAAS.]

First author Lucey and colleagues enrolled 119 participants from longitudinal studies at the Knight Alzheimer’s Disease Research Center at WashU. The volunteers averaged 74 years old and most were cognitively normal with a Clinical Dementia Rating of zero, though about 20 percent were mildly impaired with a CDR of 0.5. For up to six nights, participants used at-home devices to monitor their sleep patterns, including motion-detecting watches and headbands that recorded EEG signals from the forehead. They also kept sleep logs. Of the 119 participants, 38 subsequently underwent PET imaging with florbetapir to measure β-amyloid accumulation and flortaucipir for tau. For 104 participants, lumbar punctures yielded CSF that was tested for total tau, phosphorylated tau, and Aβ42. The researchers correlated measures of AD pathology with a host of sleep parameters, including how long it took to fall asleep, total sleep time, and wakefulness after sleep onset. They also measured SWA during NREM sleep.

Of all the sleep parameters measured, a decrease in NREM SWA most strongly reflected AD biomarkers. According to both PET and CSF data, the more tau and Aβ pathology, the less SWA during NREM sleep. The relationship was stronger for tau than Aβ. On PET imaging, tau in the orbitofrontal, entorhinal, parahippocampal, lingual, and inferior parietal regions best correlated with loss of activity in the 1-2 Hz range. This association held up even when corrected for comparisons between multiple brain regions. Aβ deposition also correlated with lower NREM SWA in the 1-2 Hz range in the frontal, temporal, and parietal lobes, but this did not hold up to correction for multiple comparisons. For CSF, Aβ42 levels did not associate with NREM SWA levels, but the ratio of total or phosphorylated tau to Aβ42 did. Lucey speculated that the decrease in SWA resulted from increased neuronal injury from tau, which disrupts the network processes generating the slow waves.

Few other sleep parameters correlated with AD pathology, but total sleep time during the night and naps during the day did, both being longer in those with neurofibrillary tangles. “Even though people were sleeping longer, this suggests they were getting less restorative sleep,” said Lucey.

“The authors are off to the races with a really great technique that would be a cost-effective way to look at what’s happening over time,” said Sigrid Veasey, University of Pennsylvania, Philadelphia. She thinks SWA could be a robust biomarker for tau deposition, because it indicates the health of astrocytes and brain connectivity, both of which are disrupted when tangles develop. However, it will be important to tease out whether slight changes in brain volume that occur with tau deposition are causing the SWA measurement to simply appear low, as opposed to really declining, she said. Veasey recently reported that early life chronic sleep deprivation in P301S mice accelerated the spread of tau pathology and neurodegeneration (Zhu et al., 2018). Longitudinal studies starting in midlife may uncover which comes first in people—tau or sleep disruption. They would also help decipher if sleep markers are clinically or diagnostically useful, said Mander.—Gwyneth Dickey Zakaib


  1. The combined methods to assess sleep in this study provide a high-quality snapshot of sleep in this population of essentially cognitively normal or only mildly impaired people. By combining biomarker methods, an equally high-quality phenotyping of presence versus absence of presumed Alzheimer pathology has been combined with this information on sleep. Previous work on sleep and Alzheimer’s has been limited to amyloid. This group has had their suspicions that tau also comes into play in this relationship, and this study provides the first clear indications that indeed it does. Given the cross-sectional nature, I agree with the authors that no causal inferences can be made from this work, however, if you read between the lines, I think this study does shed some light on causality. For example, a quick and dirty interpretation is that sleep overall shows only relatively mild changes; in other words, there is no clear indication that this study population contains people who have had a prolonged exposure to poor sleep quality or substantially impaired slow-wave sleep in their lives. Nevertheless, they have developed amyloid accumulation, and in some cases already progressed to tau pathology and now show signs of altered sleep.

    I do not think this study provides sufficient evidence yet to start using sleep as a marker for Alzheimer’s pathology in settings other than etiological/pathophysiological research. It does, however, provide fuel to rethink hypotheses around sleep and Alzheimer’s:

    • (Subtly) reduced slow-wave sleep could be a factor that accelerates amyloid accumulation, as a co-factor explaining individual differences in disease progression, rather than a main causal factor.
    • Similarly, reduced slow-wave sleep may accelerate the development of tau pathology in AD.
    • Conversely, sleep changes may be a consequence, not cause, of AD, such that tau pathology, more than amyloid, causes sleep alterations already in asymptomatic or mildly symptomatic AD.

    For now, I don’t see assessments of slow-wave sleep as a valid or wise marker for tau pathology to identify AD in the absence of clear evidence on diagnostic markers (sensitivity, specificity, positive predictive value, etc.), and because it will not necessarily be much cheaper or better than tau PET.

  2. This is an excellent study from the WashU group and represents an important contribution to our understanding of the link between sleep and AD. New in this study is the combined use of objective sleep measures together with tau PET imaging in humans. We are encouraged to see a convergence between these findings and previous publications. Lucey et al. found that higher CSF t-tau/Aβ42 and p-tau/Aβ42 were associated with lower non-REM (NREM) slow-wave activity (SWA) and more daytime napping. SWA contributes to feeling refreshed after sleeping, therefore these results align with our findings, where we observed that less adequate sleep was associated with higher CSF tau levels, and that higher CSF t-tau/Aβ42 and higher p-tau/Aβ42 were both associated with self-reported inadequate sleep, greater daytime sleepiness, and more sleep problems (Sprecher et al., 2017).

    An important issue raised by the results of this study is that even when it appears that people are getting enough sleep, tau and amyloid may still impair the brain’s ability to engage in the restorative neural processes that underlie SWA. The longer we are awake, the more we need to sleep, and slow-wave activity is a marker of this homeostatic need for sleep: the longer one is awake, the greater the slow-wave activity will be near the beginning of the sleep period, with slow-wave activity progressively declining across a night of sleep. If the observed SWA deficits in this study were due to too little sleep (i.e., insufficient sleep time to obtain adequate SWA), one might expect that shorter sleep duration would be associated with greater tau. Yet, Lucey and colleagues found that greater tau was associated with longer sleep duration (albeit still less than the recommended 7 hours of sleep per night). This, combined with findings indicating that greater tau and amyloid are associated with complaints of unrefreshing sleep and more daytime sleepiness (Sprecher et al. 2017),  suggests that tau and amyloid pathology may be impairing the brain’s ability to generate slow waves. SWA arises from highly coordinated neural activity that is critical for brain maintenance and cognition. Rodent studies have demonstrated mechanisms through which tau and amyloid could disrupt SWA, and as the authors state, it will now be important to investigate causality in humans using longitudinal and experimental studies.

    While the authors note that the use of single-channel EEG recording has limitations, implementing at-home recording is also a major strength of the study. At-home sleep measures may have greater ecological validity than laboratory sleep measurements. Compared with polysomnography, this approach is lower cost for monitoring several nights of sleep, and lessens the burden on patients and caregivers. With regard to implications for preventing or treating AD, it’s worth noting that several sleep conditions already have readily available treatments, and the technology for manipulating slow-wave activity is an active field of development. This study gives new impetus to conduct the longitudinal and validation studies needed to establish causality between disrupted sleep and AD, as well as test whether interventions for disrupted sleep can impact AD.


    . Poor sleep is associated with CSF biomarkers of amyloid pathology in cognitively normal adults. Neurology. 2017 Aug 1;89(5):445-453. Epub 2017 Jul 5 PubMed.

Make a Comment

To make a comment you must login or register.


Paper Citations

  1. . Self-reported Sleep and β-Amyloid Deposition in Community-Dwelling Older Adults. JAMA Neurol. 2013 Oct 21; PubMed.
  2. . Association of Excessive Daytime Sleepiness With Longitudinal β-Amyloid Accumulation in Elderly Persons Without Dementia. JAMA Neurol. 2018 Jun 1;75(6):672-680. PubMed.
  3. . Sleep quality and preclinical Alzheimer disease. JAMA Neurol. 2013 May 1;70(5):587-93. PubMed.
  4. . β-amyloid disrupts human NREM slow waves and related hippocampus-dependent memory consolidation. Nat Neurosci. 2015 Jul;18(7):1051-7. Epub 2015 Jun 1 PubMed.
  5. . Altered sleep and EEG power in the P301S Tau transgenic mouse model. Ann Clin Transl Neurol. 2017 Mar;4(3):180-190. Epub 2017 Feb 15 PubMed.
  6. . Chronic Sleep Disruption Advances the Temporal Progression of Tauopathy in P301S Mutant Mice. J Neurosci. 2018 Nov 28;38(48):10255-10270. Epub 2018 Oct 15 PubMed.

Further Reading


  1. . Slow wave sleep disruption increases cerebrospinal fluid amyloid-β levels. Brain. 2017 Aug 1;140(8):2104-2111. PubMed.
  2. . Sleep drives metabolite clearance from the adult brain. Science. 2013 Oct 18;342(6156):373-7. PubMed.

Primary Papers

  1. . Reduced non-rapid eye movement sleep is associated with tau pathology in early Alzheimer's disease. Sci Transl Med. 2019 Jan 9;11(474) PubMed.