Sleep and Dementia

Abstract

Neurocognitive and sleep problems are common, underdiagnosed, and frequently co-morbid. Sleep disruption, and fatigue, predict cognitive impairment. Cognitive impairment, in turn, can worsen sleep hygiene. In dementia patients, sleep disorders are common, and dementia medications affect sleep. Emerging insights on the brain’s glymphatic system suggests that sleep may drive clearance of Aβ peptide to affect Alzheimer pathophysiology. Parkinsonian dementias are linked with REM behavior disorder, a highly treatable problem that predicts future conversion into dementia.

Introduction

Sleep problems and neurocognitive impairment have a lot in common. Both frequently occur, yet in clinical practice, are frequently missed. Sleep disorders exist in as many as 70 million people in the United States, based on epidemiological data, but a retrospective review of over 100,000 contacts with over 20,000 patients in a large primary care network found that a sleep disorder was diagnosed in under 3.5% of patients¹. Similarly, neurocognitive disorders are common, and under-diagnosed. In one study, nearly 4000 primary care patients 60 years or older underwent prospective cognitive testing. Whereas 15.7% had demonstrable cognitive impairment, it was seldom noted in a retrospective review of individual medical charts. Even for those with moderate to severe impairment, under 25% of the time had clinicians noted the problem². Perhaps unsurprisingly, in addition to being prevalent and under-recognized, sleep and neurocognitive concerns often occur together. Amongst cognitively normal elderly, daytime sleepiness is an independent risk for neurocognitive decline³. Similarly, when cognitive concerns are the index clinical marker, sleep is often impaired. In one study that queried 431 consecutive patients seen for mild cognitive impairment or dementia across 10 centers, more than 60% also met criteria for a sleep disorder⁴. Thus, there is wide overlap between the constructs of sleep disorders and neurocognitive disorders.

The purposes of this review are (1) to provide information a busy clinician may use in patients with problems relating to sleep or cognition, and (2) to highlight prominent research themes likely to dominate future work in this area.

Sleep Problems and Neurocognition

Sleep problems can cause cognitive difficulty. Thus, in a clinical setting, a sleep complaint should prompt investigation for a neurocognitive consequence, and a neurocognitive complaint should prompt concern for a possible causative sleep problem. Sleep problems can worsen cognition for nonspecific as well as more specific reasons. Nonspecific issues begin with the fatigue that a sleep problem can create. Fatigue may worsen neurocognitive performance, for example, when reduced alertness makes it harder to use one’s neurocognitive capacity. From this point of view, fatigue, whatever its cause, simply creates an additional burden for a specific neurocognitive task to overcome, in addition to the need to address the neurocognitive challenge itself. In this way fatigue from a sleep disorder might not alter one’s true neurocognitive capacity in a primary sense, but may merely make it harder to use that capacity. One way in which fatigue may have such effects is by hampering the brain’s fundamental computational functions. This idea is seen in a study that found an association between fatigue and decreased functional connectivity in the so-called “default mode network” (DMN). The concept of a DMN is that it is a widely distributed neurocognitive network of anatomically connected brain regions that tend to become active or quiescent together, by default, when a person is awake but disengaged, as when one lets one’s mind wander. The DMN is commonly assessed using such technologies as functional magnetic resonance imaging (fMRI). One study⁵ found that the more that subjects reported fatigue, the more they had reduced DMN functional connectivity on fMRI. The association of increased fatigue with worse DMN fMRI connectivity was the same in young and older individuals, suggesting that the relationship was not due to age per se. Though correlational, these data are intriguing because they suggest that fatigue affects neurocognitive performance by altering the baseline DMN, a basic neurocognitive substrate from which more specific neurocognitive modalities might draw.

In addition to nonspecific matters, some sleep disorders may disrupt neurocognition in a more specific, or at least nonrandom, manner. One example of such a relationship was shown in a systematic literature review of 65 published studies of obstructive sleep apnea⁶. The review concluded that across the published reports identified, though some “controversies” existed, neurocognitive function was commonly impaired. Certain neuropsychological domains were likely to be affected (e.g. attention, memory), some were unlikely to be affected (e.g. language, psychomotor speed), and some were inconsistently affected (e.g. executive function). Thus, a sleep disorder may not affect all neurocognitive domains to the same degree.

Sleep problems may harm waking neurocognition in a manner separate from the degree to which fatigue is simply imposed on neurocognitive performance. Disrupted sleep itself may directly hamper neurocognition. One reason is that the components of the normal “architecture” of sleep, with repeated cycling through rapid eye movement (REM) and non-REM sleep, may make specific contributions to waking cognition. In an animal model of this relationship, when experimenters forced fragmented sleep architecture onto animals, doing so caused disrupted hippocampal long-term potentiation and worsened a behavioral learning task, outcomes apparently not attributable to fatigue alone since animals did not sleep more when allowed to do so⁷. Such findings raise the suggestion that human sleep disorders could disrupt waking cognition via primary effects on sleep architecture itself. Indeed, such appears to be the case in sleep disordered breathing, for example, where cognitive impairment may be associated with more time in non-REM stage 1 sleep and less time in REM sleep⁸.

Given these considerations, it is unsurprising that in a large sample of elderly patients, when people reported more fatigue or worse sleep quality, and when they were observed to have inefficient or reduced sleep, these variables predicted observed cognitive decline⁹. Thus, when a patient has neurocognitive trouble, sleep is a good place to look for a possible cause.

Sleep Problems in Neurocognitive Disorders

People with dementias of all types are more likely to have a protean variety of sleep problems¹⁰. This linkage may be independent of the specific brain neurocognitive disease. The idea that a neurocognitive disorder could impair sleep makes sense. Since a person with dementia would be expected to make less good decisions about behavior and self-care, such a person may also make worse decisions regarding sleep hygiene, too (e.g. planning and sticking to regular times to go to bed or wake up; avoiding fluids, caffeine or alcohol before bed; remembering to turn off lights and eliminating sources of noise). Another reason to expect sleep problems in dementia patients is because misinterpretations about the meaning of night time experiences may make for a tormenting exercise – fatigue related cognitive decompensation, dreams being mistaken for reality, failure to recognize a sleeping partner, increased risk of visual agnosia or visual hallucinations in low lighting conditions, for example, could collectively conspire to make a night’s sleep a challenging experience.

Commonly, people with dementia have impairments in diurnal sleep-wake variations. Because the neurological diseases that cause dementia may also affect brain structures such as the suprachaismatic nucleus, a region that helps maintain circadian rhythmicity, chronobiological difficulties occur in many individuals with a neurodegenerative dementia¹¹. Such a finding may explain why even subtle changes in circadian behavioral activity may more than double the conversion rate into a clinical state of mild cognitive impairment (MCI) or dementia¹². One type of chronobiological problem may be particularly vexing in those who suffer from dementia. When advanced sleep phase problems ensue, sleep onset tends to occur earlier than may be usual based on environmental cues. Such problems could contribute to the “sundown” phenomenon, wherein a dementia patient becomes more agitated or worsens neurocognitively in the late afternoon or early evening. Advanced phase problems might also result in a patient waking up early, feeling a need to dress and take on the day while others are still asleep, thus resulting in the common caregiver’s report that a patient is “mixing up day and night.” An otherwise well-intended clinical response to prescribe sedative-hypnotics at bedtime does not address the underlying causal chronobiological problem per se, and may risk causing more unwanted side effects than may be worth the effort. By contrast, judicious use of such modalities as melatonin, light therapy, encouraging daytime activity to delay sleep, and other caregiver-assisted sleep hygiene maintenance may be better choices with more efficacy, tolerability, and less cost¹³.

Another matter is that many dementia patients are prescribed cholinesterase inhibitors – donepezil, galantamine, or rivastigmine¹⁴. Though all three agents are FDA approved for Alzheimer disease, and one (rivastigmine) is FDA approved for Parkinson disease dementia, off-label use is common. Though these agents provide clinical benefit to the degree that future decline in cognition, behavior, and function may be less for individuals on medications as compared to those not on medications, their cholinomimetic effects are not without possible additional effects. Cholinesterase inhibitors may have a multitude of sleep consequences. Double-blind placebo-controlled trials show that a cholinesterase inhibitor reduces sleep apnea and increases night time oxygen saturation, and as would be expected for a pro-cholinergic medication, there is a marked stimulatory effect on REM sleep¹⁵. The latter consequence may explain why some patients report “wild” or “vivid” dreams on cholinesterase inhibitors. While we have anecdotally noted that restricting oral administration of such medications to a morning dose may minimize the complaint of overly vivid dreams, in some people the complaint persists and remains disruptive to sleep quality, if not quantity. Thus, the clinician may have to make a patient based judgment about how best to balance the functional costs of cholinesterase inhibitor use versus their theoretic clinical drug benefits.

Sleep and Alzheimer Disease

An intriguing emerging notion posits that sleep issues may play a direct role in causing Alzheimer disease, the most common disease leading to dementia. This argument rests on three main conclusions. First, the argument linking sleep to Alzheimer disease begins with accepting the “amyloid cascade” model of Alzheimer etiology¹⁶. First proposed in the early 1990s, and revised substantially since, the current instantiation of the amyloid cascade hypothesis posits that Alzheimer disease is caused by a series of molecular and cellular events that ultimately result in the formation of extracellular amyloid plaques. Key to this model is the notion that pathological splicing of trans-membranous and extracellular portions of the amyloid precursor protein (APP) yields a potentially toxic soluble extracellular peptide, Aβ, the primary molecular species that accumulates in amyloid plaques. Considerable evidence has established that such an amyloid cascade does occur in people with Alzheimer disease. It has been hypothesized that events which tend to oppose soluble Aβ peptide should tend to prevent or delay the symptomatic brain damage of Alzheimer disease, whereas events that tend to promote soluble Aβ peptide may accelerate Alzheimer related brain dysfunction, though it remains unclear which step between generating soluble Aβ peptide and amyloid plaque formation is most critical for brain dysfunction itself. While there is some debate about the extent to which the amyloid series of events is a cause of Alzheimer disease, or simply a marker of it¹⁷, the model remains influential and widely accepted as involved in Alzheimer etiology.

The second pillar to the argument linking sleep to a cause of Alzheimer disease rests on the functioning of the recently proposed “glymphatic” system, a system thought to be the brain’s analog to the body’s lymphatic system¹⁸. The notion of glymphatic flow is that brain astrocytes create movement of extracellular fluid, a process that may help clearance of waste substances and other materials out of the central nervous system. It stands to reason, then, that an extension of the amyloid cascade hypothesis and the concept of a glymphatic system is the speculation that glymphatic flow may help remove soluble extracellular Aβ peptide from the brain.

The third key point in the argument linking sleep to Alzheimer etiology is that sleep modulates the glymphatic system, and that this modulation helps control Aβ clearance from the brain. Supporting such a view are several findings of interest: (1) Aβ concentrations in the brain’s extracellular milieu fall during sleep and rise during wakefulness¹⁹; (2) mice that are genetically programmed to create Aβ show more Aβ deposition when sleep deprived as compared to when the mice are not sleep deprived¹⁹; (3) exogenously infused Aβ peptide is largely cleared during slow wave sleep (SWS) in a mouse model²⁰; and (4) in humans, non-REM SWS impairment is correlated with increased amyloid²¹.

By implication, then, an emerging hypothesis is that sleep, perhaps SWS in particular, may help clear Aβ peptide out of the brain via glymphatic flow, and thus impaired SWS could contribute to the processes that directly cause Alzheimer disease, or at least it may be that impaired SWS could influence ongoing Alzheimer pathophysiology. This line of inquiry promises to be a topic of future research as it carries potential for therapeutic intervention if a way can be found to promote more adaptive sleep patterns to help slow Alzheimer pathology. Also, sleep related modulation of glympahtic clearance of Aβ may serve as a template for understanding underlying etiology of other neurodegenerative diseases where a pathologic accumulation of a toxic molecular species may be hypothesized to occur.

Sleep and Parkinsonsonian Dementias

One of the most robust connections between a sleep disorder and a dementia-causing disease is seen in the association between REM behavior disorder (RBD) and neurodegenerative Parkinsonism, a topic substantially reviewed in its own right²². In normal REM sleep, a stage of sleep in which most dreams occur, normally most of the body’s skeletal muscles are paralyzed (sparing muscles that move the eyes, and muscles of respiration). By contrast, in RBD, the expected skeletal muscle atonia that should have occurred throughout the body in REM sleep does not occur. As a result, when the sleeper experiences a dream, movements may occur, such that it may appear that the person is acting out the dream. A caregiver may report that the patient makes subtle movements or dramatic and violent ones, and the matter can thus pose a safety risk for both sleeper and sleep partner, as well as create greater sleep fragmentation. The condition is often quite treatable with melatonin or low dose clonazepam. But, a diagnosis based on history alone can be difficult, since motor activity during non-REM sleep is not pathological, and some dreams and other mental activity can occur in non-REM sleep. Thus, even a caregiver report of observing possible dream enactment movements during sleep may be non-definitive. An RBD diagnosis can only be made when it is known that a sleeper is moving during REM sleep. Thus a polysomnogram is often needed for a definitive diagnosis.

Interestingly, longitudinal studies of people with RBD show that in those who have this diagnosis, virtually all go on to develop a neurodegenerative Parkinsonian disease, though in some cases, such may emerge only after many years’ delay. Approximately 50% of people with RBD show neurological signs of a Parkinsonian disorder (e.g. Dementia with Lewy Bodies, or Parkinson Disease) within a decade, and as much as 90% do so at some future point thereafter²². In a review of over 5000 studies that evaluated risk factors for Parkinson disease dementia (PDD), of the studies that met criteria for further analysis, the average relative risk for PDD was estimated to be 8.38 in individuals with RBD, higher by far than for any other factor²³. Thus, the finding of RBD in a patient should particularly prompt close clinical attention for the potential future emergence of a Parkinsonian condition, a complication of which is highly likely to include dementia at some point. Similarly, in a patient with a dementia, careful attention to clinical history or a formal sleep study to rule in RBD should prompt high suspicion for concluding that the cause of the dementia might be a Parkinsonian spectrum condition, even if not otherwise clinically apparent at that moment.

Conclusion

Sleep disorders and neurocognitive disorders are both frequent categories of disease, but they often are missed in regular clinical practice. In this brief review we have shown that sleep and neurocognitive disorders often occur together. The relationship between problems with sleep and impaired neurocognition is likely complex and bidirectional. In some instances the co-occurrence of sleep and neurocognitive difficulty may be because sleep problems cause cognitive difficulty, in both nonspecific ways and specific ones. In other cases a disease causing a neurocognitive disorder, or some of its symptomatic or iatrogenic sequlae, may lead to impaired sleep. Alterations in sleep may be a fundamental substrate or major relative risk for some neurodegenerative causes of dementia. As some of the involved conditions are treatable, and most can be highly symptomatic with risks that patients and support systems alike are well advised to be aware of, it behooves the clinician to be proactive and aggressive in diagnosing both sleep disorders and neurocognitive ones. As we have argued in this review, then, the finding of a sleep disorder should prompt careful clinical attention to the possibility for an occult neurocognitive impairment, and vice versa.

Biography

Joel I. Shenker, MD, PhD, is an Associate Professor of Clinical Neurology, and Director, Neurology Residency Program at the University of Missouri - Columbia School of Medicine. Gurtej Singh, MD, is a Neurologist with University of Missouri Health Care.

Contact: shenkerj@health.missouri.edu