INFLUENCE OF LIGHT ON SLEEP AND HEALTH

Light and light pollution during the night, and insufficient exposure to light during the day, affect multiple dimensions of sleep health, said Phyllis Zee. Indeed, said Tiffany Schmidt, associate professor of neurobiology at Northwestern University, light impacts not only conscious visual perception, but subconscious perception as well. She suggested that the light detecting system responsible for these effects may be a modifiable target to improve health and well-being.

The photo entrainment of sleep–wake and activity cycles to the environmental light–dark cycle affects alertness, learning, memory, mood, and mental health, as well as reflexes such as pupillary constriction and dilation, said Schmidt. Lighting at the wrong time of the day thus strongly impacts sleep as well as physical and mental health, she said, adding that this has become a public health issue as our lighting environment has evolved. Beyond the solar day, Schmidt said light has become ubiquitous in our daily lives, with tablets and screens widely available to everyone, more people engaged in shift work, and increased light pollution at night. She added that LED [light-emitting diode] screens emit blue-shifted light, which more strongly affects sleep, circadian rhythms, and health compared to red-shifted light, for example. Reiterating what Dayna Johnson discussed in Chapter 2, Schmidt noted that poor health outcomes, resulting from mis-timed light exposure, disproportionately affects people in lower socioeconomic groups, who are more likely to live in urban areas and work later shifts. She noted that it may be possible to reduce these health disparities through a better understanding of how light impacts behavior and negative health outcomes at a cellular and circuitry level.

Prior to the late 1980s, the prevailing model of light detection was that when light falls on the rod and cone photoreceptors in the retina, the information is relayed through a network of interneurons to the retinal ganglion cells, which project to the brain, said Schmidt. However, clinicians began to notice that even patients with no conscious visual perception showed some non-image–forming subconscious responses to light (Czeisler et al., 1995). Even blind patients who lacked functional rod and cone photoreceptors, due to degeneration of the outer retina, showed normal melatonin rhythms, said Schmidt.

Researchers discovered a third class of light-sensitive cells in the retina called intrinsically photosensitive retinal ganglion cells (ipRGCs) that respond to light without any input from rod and cone photoreceptors by expressing a protein called melanopsin (Altimus et al., 2008; Berson, 2003; Hattar et al., 2002). Schmidt said that these ipRGCs relay light information to multiple non-image–forming areas of the brain.

“They are sort of the conduit through which light is getting to the brain to influence most non-image–forming functions,” she said. Thus, they are not only required for circadian photoentrainment, but also other aspects of brain health and behavior. For example, ipRGCs impact dopaminergic, glutamatergic, GABAergic, and cholinergic pathways involved in mood, attention, addiction, and aggression. This could mean that the environmental light–dark cycle impacts the dopamine reward system, suggested Schmidt.

She added that ipRGCs are well conserved between mice and humans and that melanopsin provides a genetic marker that enables assessment of cellular function, circuits, and behavior in mice, with implications for human health. Schmidt emphasized that understanding their function may elucidate how aberrant light exposure affects brain function and behavior and provide insight about a range of mental health and neurodegenerative disorders.

CENTRAL NERVOUS SYSTEM HYPERSOMNIAS AND NARCOLEPSY

Narcolepsy, a central nervous system (CNS) disorder of excessive sleepiness or hypersomnia, was first described nearly 150 years ago, but only in the past 20 years have researchers identified the fundamental cause and been able to design effective treatments, said Thomas Scammell, professor of neurology at Harvard Medical School. Narcolepsy presents in two different types. People with type 1 narcolepsy also experience cataplexy, episodes of acute muscle weakness or atonia triggered by strong emotions, while people with type 2 narcolepsy do not experience cataplexy. Other CNS hypersomnias include idiopathic hypersomnia, where people have chronic sleepiness with a tendency to sleep many hours, night after night, and then wake up with terrible sleep inertia; and the rare disorder Kleine-Levin syndrome, which typically affects adolescent males with intermittent, recurring periods of excessive sleepiness that may be accompanied by hyperphagia and hypersexuality, said Scammell.

People with both type 1 and type 2 narcolepsy may also have sleep paralysis at the beginning or end of sleep; dreamlike hallucinations known as hypnagogic hallucinations when falling asleep or hypnopompic hallucinations when waking up; and fragmented sleep. Cataplexy in type 1 narcolepsy is thought to be similar to the paralysis that occurs normally during rapid eye movement (REM) sleep, but which intrudes into wakefulness, said Scammell. In Figure 4-1, EEG recordings from a person with narcolepsy are compared with a healthy control illustrate the disruptions in sleep and wakefulness. “They fall asleep before the designated time, immediately go into REM, have fragmented sleep across the night, and fall asleep repeatedly during the day,” said Scammell. He noted that they also have REM sleep during the day, which indicates a loss of circadian rhythmicity.

FIGURE 4-1

24-hour sleep recordings in narcolepsy. Recordings from study participants with narcolepsy type 1 show shorter sleep latencies and more disrupted sleep at night, as well as periods of daytime sleep, in comparison to control subjects without narcolepsy. (more...)

Shortly after the hypocretin/orexin neurotransmitter system was discovered by Luis de Lecea and Takeshi Sakurai and their colleagues (discussed in Chapter 3), Emmanuel Mignot at Stanford University and Masashi Yanagisawa at the University of Tsukuba in Japan independently demonstrated that narcolepsy was caused by the loss of orexins. Hao Wang, vice president and global program leader in the neuroscience therapeutic area at Takeda, noted that Mignot and Yanagisawa received the 2023 Breakthrough Prize for this discovery.

Orexin-producing neurons are found only in the lateral hypothalamus but have projections to many key brain regions where they bind to two orexin receptors, orexin-1 and orexin-2, to promote wakefulness, said Scammell. In addition, orexins suppress REM sleep in the normal active period in a circadian-mediated fashion.

The mechanisms for cataplexy are more complicated and involve multiple circuits, he said. During wakefulness, orexin neurons excite atonia-blocking regions in the brainstem to prevent cataplexy from occurring. In narcolepsy, the loss of orexin neurons removes this block on atonia. In addition, strong positive emotions signal the amygdala to inhibit the atonia blocking region, said Scammell. Oxytocin also excites this amygdala circuit and can increase cataplexy in narcoleptic mice, which could help explain why cataplexy almost only occurs under social circumstances, he said (Hasegawa et al., 2022; Mahoney et al. 2017; Uchida et al., 2021).

More than 20 years ago, Mignot and colleagues showed that decreased orexin levels were present in patients with narcolepsy plus cataplexy, but not in patients with narcolepsy but no cataplexy, said Scammell (Taheri et al., 2002). “We now tend to look at these as two different diseases,” he said. Supporting this view is that there is about a 90 percent loss of orexin producing neurons in narcolepsy type 1; narcolepsy type 2 may arise from a milder loss of the orexin neurons, though this is based on very limited neuropathology. Scammell said that an autoimmune process is thought to be the cause of this pathology. A certain human leukocyte antigen (HLA) increases the risk of narcolepsy more than 200-fold, he said. Through a process of molecular mimicry, it is thought that in people with this HLA type, T cells generated in response to a pathogen such as influenza cross-react with an antigen on orexin neurons, said Scammell (Cogswell et al., 2019).

NEURODEVELOPMENTAL DISORDERS

EEG provides rich information that is underleveraged for the study of neurodevelopment and neurobehavioral trajectories, said Ashura Buckley, director of the Sleep Disorders and Neurodevelopment Consult Service, Office of the Clinical Director, at the National Institute of Mental Health (NIMH). As a component of polysomnography (PSG), EEG provides oscillatory information from the sleeping brain that can be integrated with physiological information from other organ systems, enabling scientists to study the neural dynamics of sleep, she said.

Sleep disruption and dysregulation are ubiquitous in neurodevelopmental disorders such as autism as well as in neuropsychiatric and neurobehavioral disorders, said Buckley. Moreover, Buckley said sleep disruption is core to these disorders, not an epiphenomenon (Missig et al., 2020). She suggested that understanding the pathophysiology better may lead the field closer to the development of effective treatments. Dara Manoach, professor of psychology in the Department of Psychiatry at the Massachusetts General Hospital and Harvard Medical School, added that maternal sleep deprivation during pregnancy also leads to sleep dysregulation, which increases the risk of later psychopathology in the offspring (e.g., Benca, 2016).

The NIMH Intramural Research Program held a series of Sleep and Neurodevelopmental Workshops between 2014 and 2021 to identify data gaps and infrastructure needs for translational research in the field. In 2017, the Sleep and Neurodevelopment Consortium established a partnership with National Sleep Research Resource (NSSR)¹ which is funded by the National Heart, Lung, and Blood Institute to provide a platform for the collection, harmonization, and querying of large pediatric datasets focused on brain data and indexed by diagnostic group. Through a grant from the Pediatric Epilepsy Research Foundation, existing clinical data were collected into a repository for hypothesis generation. With the addition of Nationwide Children’s Hospital depository there are now about 6,000 pediatric studies in NSRR, said Buckley. In addition to these clinical data, data are needed about the normal course of development, she added. Thus, in 2021, a longitudinal, prospective study of sleep and neurodevelopment, the Electrophysiological Sleep Phenotyping Project,² was launched to collect real-time EEG oscillatory information along with behavioral, genetic, and demographic data from children with and without neurodevelopmental disorders.

These data are being used to examine cognitive and behavioral trajectories in relation to oscillatory sleep metrics, said Buckley. “We know that cognition and behavior require coordination between spatially distinct regions in the brain,” she said. The similarities between EEG recordings at any two points, or coherence, is a measure of functional connectivity between those two regions (Achermann and Borbély, 1998). Preliminary data from a previous NIMH study suggest a relationship between coherence in REM sleep and verbal scores, said Buckley. This observation raises more questions than it answers about the relationships among sleep, language delay, and memory consolidation.

To answer these questions, less siloed and more convergent research is needed, said Buckley, which requires investment in device development. For example, NIMH has been working with the Brain Electrophysiology Lab (BEL),³ using a geodesic sensor net to capture high density EEG information which optimizes signal localization (Morgan et al., 2021).

Andrew Krystal noted that coherence is also being assessed in relation to epileptiform activity. “We’ve been trying to understand why in light sleep we have greater propensity to epileptiform activity than any other state of consciousness,” he said. “It turns out that there is an increase in coherence as you move from wake to sleep,” said Krystal, adding that this increased coherence is seen in very specific circuits.

SCHIZOPHRENIA

The observation that unlike healthy controls, people with schizophrenia fail to improve their performance on a simple finger-tapping task after a night of sleep led Manoach to begin a sleep research program. Sleep disturbances have long been associated with schizophrenia beginning in its earliest stages and may even play a causal role (Cohrs, 2008; Manoach and Stickgold, 2019; Waite et al., 2020). Manoach reasoned that understanding sleep might also lead to a better understanding of the mechanisms underlying schizophrenia.

Indeed, a meta-analysis by Manoach and colleagues confirmed that the reliability of a specific sleep impairment in schizophrenia is a reduction in sleep spindles—brief bursts of neuronal activity that occur during non-REM sleep (Lai et al., 2022). Sleep spindles are key mechanisms of synaptic plasticity mediating memory consolidation during sleep. In schizophrenia, the spindle density deficit correlates with impaired sleep-dependent memory, suggesting a mechanistic role (Wamsley et. al, 2012).

Manoach and colleagues were also able to replicate their finding of sleep spindle deficits in a large ethnically and racially distinct sample in China (Kozhemiako et al., 2022). Collectively, these findings demonstrate that the spindle deficit represents a reliable and generalizable finding in people with schizophrenia that is linked to cognitive function, said Manoach. Abnormalities in spindle activity are also seen in patients with autism, according to research in Manoach’s lab (Mylonas et al., 2022; Wamsley et al., 2012).

Genetic factors likely contribute to spindle deficits in schizophrenia. Knocking out, or stopping the expression of, a risk gene for schizophrenia (CACNA1I) in mouse models gives rise to a specific spindle deficit as does knocking out a mutated form of that gene, said Manoach (Andrade et al., 2016; Astori et al., 2011; Ghoshal et al., 2020). “We have a genetic mechanism that could contribute to spindle deficits in schizophrenia,” said Manoach.

However, a not-yet-peer-reviewed study showed that another risk gene for schizophrenia, Grin2a, increases spindle density (Herzog et al., 2022), suggesting that the association of spindle density, sleep, and schizophrenia is more complex than initially thought, said Manoach. She added that along with genetic heterogeneity, there is also phenotypic heterogeneity. “Not all people with schizophrenia have spindle deficits, and the literature on spindles in autism is markedly inconsistent,” she said.

Because spindle deficits correlate with memory deficits in many patients with schizophrenia, Manoach and colleagues tested whether they could be used as a biomarker of memory in a randomized controlled clinical trial of the sleep drug eszopiclone. Interestingly, while eszopiclone increased spindles in both healthy controls and patients with schizophrenia, the increase in spindle density did not correspond to an improvement in memory (Mylonas et al., 2020). The explanation for this, according to Manoach, is that their model was too simple. “Spindles do not work in isolation to improve memory,” she said.

Eszopiclone also interferes with the coupling of spindles with the two other cardinal neuronal oscillations of non-REM sleep: cortical slow oscillations and hippocampal ripples. The hippocampus is a deep structure, and its activity is difficult to measure without invasive recording techniques, but spindle coupling with cortical slow oscillations may be a better biomarker of memory, she said.

Manoach noted that since non-REM sleep oscillation abnormalities are seen in both schizophrenia and autism, cross-disciplinary research provides the opportunity to forge empirical links in the causal chains that go from genes to diagnosis and can lead to mechanistically guided treatment, including drugs and neurostimulation (see Figure 4-2). “Sleep may be the new frontier for treating these circuitopathies,” she said.

FIGURE 4-2

Forging links in causal chains: Genes to diagnosis to treatment. The causal chain linking genes to sleep disorders and their treatment starts with the premise that the discovery of genes will lead to mechanisms, then to biomarkers and diagnosis, and finally, (more...)

MOOD DISORDERS

Disrupted sleep has long been linked to mood disorders (Peterson and Benca, 2006). Indeed, insomnia and hypersomnia are among the symptoms listed as criteria for major depressive disorder in the Diagnostic and Statistical Manual of Mental Disorders (APA, 2022); 90 percent of people diagnosed with major depressive disorder report sleep problems, said Andrew Krystal.

For many years, insomnia was thought to be a symptom of depression that would improve with treatment for depression, said Krystal. More recently, however, Krystal said that evidence has emerged suggesting a bidirectional relationship between sleep disturbance and depression. For example, insomnia in the absence of depression increases the risk for depression emerging in the future (e.g., Fang et al., 2019). In patients being treated for depression, insomnia decreases the antidepressant response, increases the risk of relapse, and often presents as a residual symptom of therapy, said Krystal.

In more than 40 studies, insomnia has been shown to predict suicidality, said Krystal. In fact, he said, having insomnia is a better predictor of whether a patient will attempt suicide than assessing suicidality by asking a patient whether they have planned or thought of harming or killing themselves (Agargun et al., 1997, 2007; Chellappa and Araujo, 2007; Fawcett et al., 1990; Sjöström et al., 2007; Turvey et al., 2002). This suggests that diagnosing insomnia could be an important opportunity for identifying people at risk of suicide, said Krystal.

Krystal cited several studies suggesting that insomnia is not just correlative, but causative. Studies in three different populations—pregnant or postpartum women, older adults, and younger adults—have shown that delivering cognitive behavioral therapy for insomnia (CBT-I), either digitally or in person, to people who have insomnia, but not current depression, decreases subsequent depression (Cheng et al., 2022; Felder et al., 2022; Irwin et al., 2022).

In another study, suicidal ideation also decreased to a greater degree in people who received insomnia therapy (controlled-release zolpidem) in conjunction with an antidepressant (selective serotonin reuptake inhibitor, or SSRI) compared with those who received a placebo plus the SSRI (McCall et al., 2019). Whether suicide attempts can also be decreased by treating insomnia remains an important, but unanswered, question, said Krystal.

The mechanisms underlying these effects also remain poorly understood, he said. Preliminary evidence from intracranial brain stimulation research in patients with severe depression have shown that direct brain stimulation that improves depression often leads to alterations in sleep–wake function, suggesting overlapping circuitry, said Krystal. Genes that regulate sleep have also been shown to modulate resilience, which has been found to mediate the depression preventative effects of CBT-I

NEURODEGENERATIVE DISORDERS

The co-occurrence of sleep or circadian dysfunction in people with Alzheimer’s disease and other neurodegenerative diseases has been known for many years, said Erik Musiek, the Charlotte and Paul Hagemann professor of neurology at Washington University in St. Louis (WUSTL) and co-director of the Center on Biological Rhythms and Sleep. A wide range of problems may present in these individuals: poor sleep efficiency, increased napping, decreased slow-wave sleep and REM sleep, and fragmented sleep, said Musiek. They may also have what is known as a phase delay, where their activity level peaks a few hours later than normal, he said, noting that this may contribute to the phenomenon of sundowning, in which people become more confused later in the day.

Actigraphy studies in patients with Alzheimer’s disease show a breakdown in normal day–night patterns, said Musiek. This has been linked to degeneration of neurons in the suprachiasmatic nucleus (SCN) that help regulate circadian rhythms (Wang et al., 2015). One of the major questions to be answered, said Musiek, is whether this is the chicken or the egg—is circadian dysfunction causing the disease or is it simply a result of brain degeneration? “It may be some of both,” he said.

Epidemiological studies indicate that people who report low amounts of sleep each night have an elevated risk of developing Alzheimer’s disease, said Musiek. To address the question of whether sleep disruption in midlife increases the risk of Alzheimer’s disease in late life, one study followed people for more than 25 years. The study showed that those who got fewer than 6 hours of sleep at age 50 had a 1.28 relative risk of developing dementia 25 to 30 years later (Sabia et al., 2021).

Because Alzheimer’s pathology—amyloid beta plaques and tau tangles—accumulates 15 to 20 years prior to the appearance of symptoms, this may suggest a causal relationship between the pathology and the sleep problems, said Musiek. Indeed, other studies show that people with positive Alzheimer’s disease biomarkers indicating the preclinical phase of the disease have sleep abnormalities that include increased sleep latency, sleep fragmentation, waking after sleep onset, decreased sleep efficiency, and increased napping. Cognitively normal individuals with positive biomarkers also are more likely to have disrupted circadian rhythms, and these individuals with fragmentation of circadian rhythms are more likely to develop dementia, said Musiek.

Clifford Saper, the James Jackson Putnam Professor of Neurology and Neuroscience at Harvard Medical School, mentioned another condition called REM sleep behavior disorder, in which people act out their dreams during sleep. More than 90 percent of people with this condition are diagnosed many years later with a synucleinopathy such as Lewy body dementia, Parkinson’s disease, or multiple systems atrophy, he said. He suggested that age-related “normal” changes in sleep may reflect extremely early insults to the brain’s sleep–wake circuitry.

Indeed, added Brian Fiske, co-chief scientific officer for The Michael J. Fox Foundation for Parkinson’s Research, REM sleep behavior disorder is a very early sign of Parkinson’s disease and is commonly used as a means of identifying people at risk of Parkinson’s. Insomnia and excessive daytime sleepiness are also commonly seen early in Parkinson’s disease, he said. At the onset of motor symptoms, 44 percent of patients have at least one of these sleep disturbances, and this increases to about 70 percent within 5 years (Xu et al., 2021).

Understanding how sleep impacts the development of Alzheimer’s disease and other neurodegenerative diseases could have therapeutic implications, according to Musiek. “We don’t know if the sleep problem is driving the pathology or the pathology is driving the sleep problem,” he said.

Nathaniel Watson added that there are some studies indicating that treatments for insomnia, such as benzodiazepines and z-drugs, are associated with the development of Alzheimer’s disease; although, there are mixed findings in the field. “Is that simply confounding by indication, where the drugs are at the scene of the crime, the sleep disturbance is the problem, and by attempting to treat it, the drugs look like they’re causing difficulties?” he said. “Are we missing a huge opportunity to optimize or improve sleep in order to bend the curve of neurodegeneration over time?”

David Holtzman and colleagues at WUSTL have demonstrated circadian rhythms in levels of amyloid beta in the brain’s extracellular fluid and increased plaque development in sleep-deprived mice (Kang et al., 2009). Using lumbar catheters in humans, other colleagues at WUSTL have shown that sleep deprivation causes an increase in amyloid beta levels in spinal fluid (Lucey et al., 2018). This suggests that the brain can deal with amyloid beta under normal sleep conditions, but not during sleep deprivation, said Musiek.

Musiek also cited another theory promoted by Maiken Nedegaard and colleagues at University of Rochester involving the movement of fluid through the brain via the glymphatic system, which removes waste products and toxic metabolites such as beta amyloid, tau, and synuclein during sleep (Xie et al., 2013).

“So there’s a debate about whether sleep causes clearance or decreases production,” said Musiek. “It could do both.” In favor of the idea that sleep decreases production is the notion that amyloid beta is released as a product of neuronal activity, which is greater during wake. In support of the clearance hypothesis, orexin antagonists, which are used to treat insomnia, seem to be effective at removing plaque formation in mice, and thus may have a protective effect against neurodegenerative disease, he added. However, increasing sleep with oxybate (a medication used to decrease daytime sleepiness in people with narcolepsy) in people who were already sleeping well did not seem to affect amyloid beta levels.