The Evolving Nexus of Sleep and Depression

Insomnia and depression share a bidirectional relationship, in which each separate entity can influence the course of the other (2). It is rare for patients experiencing a major depressive episode not to experience some form of sleep disturbance. Insomnia is the most common sleep complaint among those with comorbid depression, occurring in approximately 80%−90% of patients (3). Importantly, meta-analysis has demonstrated that insomnia increases the odds of developing depression roughly twofold, which is on par with the risk associated with having a first-degree relative with recurrent major depressive disorder (4, 5). Additionally, twin and genome-wide association studies have demonstrated sizable genetic overlap between insomnia and major depression (6, 7).

Given the connections between insomnia and depression both cross-sectionally and longitudinally, consideration of insomnia as a contributing factor in genetics and neuroimaging results could further advance our understanding of these heterogeneous processes (8). While relatively few studies have examined the impact of insomnia on neuroimaging findings in major depression, overlapping structural and functional abnormalities in the medial prefrontal cortex and posterior cingulate cortex within the default mode network, as well as the insula, anterior cingulate cortex, and amygdala within the salience network have separately been observed in each condition (9). In terms of structural findings, the ENIGMA (Enhancing Neuroimaging Genetics Through Meta-Analysis) depression working group recently demonstrated in 1,053 persons with major depressive disorder that more severe insomnia symptoms were associated with reduced cortical surface area, primarily in the right insula, left inferior frontal gyrus pars triangularis, left frontal pole, right superior parietal cortex, right medial orbitofrontal cortex, and right supramarginal gyrus (10). Notably, these associations were not observed among healthy or clinical control subjects with bipolar disorder, suggesting that these associations between surface area in frontoparietal cortical areas and insomnia are unique to depression (10). In terms of functional neuroimaging, Cheng et al. (12) analyzed data from the Human Connectome project, demonstrating that increased functional connectivity in multiple brain areas (lateral orbitofrontal cortex, dorsolateral prefrontal cortex, anterior and posterior cingulate cortices, insula, parahippocampal gyrus, hippocampus, amygdala, temporal cortex, and precuneus) was associated with both sleep quality (measured using the Pittsburgh Sleep Quality Index [11]) and depressive symptoms, suggesting a neural basis for their overlap. Drysdale et al. (13) successfully segregated treatment-resistant depression along two dimensions—one predicting anhedonia symptoms based on frontostriatal and orbitofrontal connectivity, the other predicting anxiety and insomnia based on connectivity within the amygdala, ventral hippocampus, ventral striatum, and lateral prefrontal cortex. While patterns of clinical symptoms prior to treatment were only modestly associated with therapeutic effects of transcranial magnetic stimulation applied at the dorsomedial prefrontal cortex, subgroups that accounted for these connectivity patterns were better able to predict treatment response (13).

An important limitation of nearly all neuroimaging approaches that are used to study insomnia and depression is that they evaluate the waking brain. Thus, quantitative measures acquired during the sleep period, such as EEG, are more likely to provide insights into brain changes that may co-occur in insomnia and depression. While increased high-frequency EEG activity during sleep has been generally reported in insomnia disorder (14), such sleep EEG changes have typically not been described using sleep spectral analysis in persons with major depression and comorbid insomnia (15, 16). In this instance, the use of standard EEG montages that have limited spatial resolution may be a significant barrier in this line of inquiry, particularly since preliminary evidence has demonstrated that local alpha activity persists in sensory and sensorimotor cortical areas in persons with insomnia, even in the deepest stages of non-REM sleep (17). Thus, the use of high-density EEG arrays during sleep has significant promise to advance research on sleep and depression, as has been the case for other psychiatric disorders (18, 19).

From a more pragmatic standpoint, it is important to know whether treatment of insomnia occurring in the context of a depressive episode enables therapeutic effects beyond the improvement of sleep symptoms. Two earlier randomized controlled trials focused on nonbenzodiazepine Z-drugs (placebo versus active compound) with open-label use of specific serotonin reuptake inhibitors (SSRIs) (20, 21). While both studies demonstrated a significant benefit of the sedative-hypnotic on insomnia measures, results on depressive symptoms were mixed. One study demonstrated that the coadministration of eszopiclone and fluoxetine resulted in improvement of depressive symptoms, even beyond the impact of sleep-related changes (20). However, similar benefits from sedative-hypnotics were not observed in a study of coadministered extended-release zolpidem and escitalopram (21).

Studies that clarify the effect of sedative-hypnotic coadministration in patients with a major depressive episode are particularly relevant in patients at risk for suicide. Insomnia is independently associated with increased suicidal behaviors and ideation (22). However, suicidal ideation has also been associated with sedative-hypnotics (23), and many psychiatrists fear that prescription of sleep medications may increase the risk of overdose. To advance this area of inquiry, the Reducing Suicidal Ideation Through Insomnia Treatment (REST-IT) study applied a similar coadministration paradigm of extended-release zolpidem versus placebo with open-label SSRI in patients with major depression, insomnia, and co-occurring suicidal ideation (24). This landmark study demonstrated that coprescription of a sedative-hypnotic and an SSRI does not worsen suicidal ideation, and in fact may reduce suicidal ideation, particularly in patients with more severe insomnia (24).

The direct effects of cognitive-behavioral therapy for insomnia (CBT-I), a highly effective nonpharmacological treatment for insomnia that is generally considered a first-line therapeutic strategy (25, 26), have not been thoroughly evaluated in depressed patients with suicidal ideation. A previous study that evaluated the effects of CBT-I on veterans with insomnia demonstrated that CBT-I may improve depressive symptoms and suicidal ideation, and that the observed relationship between insomnia severity and change in suicidal ideation remained significant even when adjusted for changes in depression severity (27). A smaller pilot trial of brief CBT-I in depressed veterans with and without suicidal ideation suggested that this modality may improve sleep, but effects on depression and suicidal ideation required larger follow-up studies (28). A subsequent larger randomized controlled trial in patients with major depression and/or posttraumatic stress disorder reporting suicidal ideation demonstrated large effects of brief CBT-I on depression and insomnia, with relatively small effects of the therapy on suicidal ideation, which were moderated by insomnia, although this effect fell short of statistical significance (29).

Beyond patients with suicidal ideation, the effects of concomitant thymoleptics and CBT-I in persons with depression and comorbid insomnia have also yielded mixed results. Initial randomized controlled pilot studies examining the use of CBT-I with coadministration of escitalopram in patients with depression and insomnia improved both insomnia symptoms and rates of depression remission compared with escitalopram and active control therapy (30). However, the larger follow-up Treatment of Insomnia in Depression (TRIAD) study, which employed a similar research paradigm, did not replicate the preliminary findings (31). Post hoc analyses suggested that childhood onset of depression and insomnia, as well as evening circadian preference, were potential moderators of depression outcomes in TRIAD (32, 33). Change in depression symptoms also were not significantly different across treatment arms in a randomized controlled trial in depression and comorbid insomnia that compared three treatment conditions: escitalopram plus CBT-I, CBT-I plus placebo, and escitalopram plus sleep hygiene control (34).

Thus, taken in aggregate, studies that have examined the effects of concomitant use of insomnia treatments (i.e., medications or CBT-I) and antidepressants in persons with depression and insomnia have reliably demonstrated improvement in insomnia measures resulting from sleep-targeted therapy. However, improvement in depressive symptoms in these study paradigms has not been as consistent. While differences among study findings may be related to granular dissimilarities between investigations, from a pragmatic standpoint, specifically addressing insomnia complaints in the context of depression is likely to have a positive therapeutic effect, at least in some instances, and may improve sequelae of depression (e.g., suicidal ideation). However, from a public health perspective, it is possible that a more critical and fruitful time to target insomnia in the treatment of depression may be prior to the onset of a depressive episode.

The ability to effectively prevent the onset of mental illness is the holy grail of psychiatry. Unlike the majority of fields of medicine, which have successfully developed and championed preventive strategies for a variety of chronic conditions, psychiatry still largely relies on diagnosis and management after symptoms are well established. Given the large societal and economic costs of mood disorders (35), the development of effective strategies to prevent the onset of depressive episodes should be a high priority for the field. The robust longitudinal connections between insomnia and risk of subsequent depression (4, 25, 26), as well as effective pharmacological and nonpharmacological methods to treat insomnia, place insomnia as a key target for depression prevention. While the possibility that insomnia may represent a unique opportunity for prevention of mental illness has been proposed for several decades (36), the development of Internet-based CBT-I therapies that are both effective and can be widely disseminated has begun to advance this critical area of inquiry.

Large-scale studies of digital CBT-I (dCBT-I) in persons with insomnia have generally demonstrated improvements in both insomnia and depressive symptoms. The GoodNight Study was a randomized controlled trial that examined the use of dCBT-I in adults with insomnia and subthreshold depression symptoms (37). The study demonstrated that dCBT-I significantly reduced depressive symptoms but did not significantly reduce the number of participants with major depressive disorder at 6 months. Freeman et al. (38) examined the effects of dCBT-I in a large study of university students with insomnia and demonstrated large reductions in depression scores at 10- and 22-week follow-ups in dCBT-I compared with usual care, with significant reductions in the number of persons developing a major depressive episode. Large-scale randomized controlled trials of the effects of dCBT-I have replicated findings of reductions in depressive rating scales resulting from the therapy (39, 40).

More recently, Cheng et al. (41) conducted a randomized controlled trial of dCBT-I in a large group of persons with insomnia disorder, approximately half of whom had depressive symptoms rated as moderate or worse at baseline. dCBT-I was associated with significantly increased rates of depression remission at 1-year follow-up compared with the control condition (treatment as usual plus online sleep education). Baseline severity of insomnia, lack of insomnia response or remission to dCBT-I, and decreased durability of insomnia improvement with dCBT-I were conversely all associated with increased moderate to severe depression at 1-year follow-up. Notably, among individuals with minimal or no depression at baseline, the risk of developing moderate to severe depression was reduced by half in the dCBT-I relative to the control condition. When considering participants without depression at baseline, the number needed to treat to prevent depression at 1 year was 11 (41). Because of the high prevalence of insomnia in the general population (42), this finding suggests that treatment of insomnia with dCBT-I has the potential to dramatically decrease the incidence of depression if broadly available and applied. Clearly, such promising results demonstrate the need for further large-scale longitudinal research trials that focus on the ability to prevent major depressive episodes through the treatment of insomnia as a modifiable risk factor.

While the definition of treatment-resistant depression may vary across the literature and is not without some controversy, generally it is operationally defined as depression that fails to respond to at least two adequate therapeutic trials of an antidepressant medication (43, 44). Treatment-resistant depression has been estimated to occur in up to half of cases of major depressive disorder (43). It is a significant cause of morbidity and mortality, and its optimal treatment relies on careful history taking and a tailored approach to treatment that balances comorbidities with the risks and benefits of available treatment options. Sleep disorders, particularly obstructive sleep apnea, may play an underappreciated role in treatment-resistant depression. The overarching pathophysiology of obstructive sleep apnea consists of cycles of apneas and hypopneas—intermittent upper airway obstruction due to pharyngeal collapse during sleep, resultant drops in blood oxygenation, and arousals from sleep to temporarily resolve upper airway obstruction. Obstructive sleep apnea is associated with several negative health outcomes, including hypertension, cardiovascular disease, arrhythmia, stroke, impaired glucose regulation, cognitive impairment, and mortality (45).

The prevalence of obstructive sleep apnea among patients with major depression is higher than in the general population, with approximately 20% of persons with major depression having obstructive sleep apnea and vice versa (46). Recent data suggest that such elevated rates of obstructive sleep apnea occur in persons with major depression even in the absence of symptoms suggestive of sleep-disordered breathing (47). In addition, persons with major depression and comorbid obstructive sleep apnea may have higher rates of suicidal ideation and acts compared with those without sleep apnea (48, 49). Obstructive sleep apnea has been associated with nonresponse to antidepressant pharmacotherapy in adolescents and adults (50, 51). Randomized controlled trials have also demonstrated that positive airway pressure therapy decreases both symptoms and cases of depression in persons with obstructive sleep apnea (52). Thus, as supported by the APA practice guidelines, “Clinicians should be alert to the possibility of sleep apnea in patients with depression, particularly those who present with daytime sleepiness, fatigue, or treatment-resistant symptoms” (53).

Unfortunately, screening for sleep-disordered breathing as a contributing factor for both depression and treatment resistance is limited among practicing psychiatrists. This is likely due to several factors, including limited education about sleep disorders among mental health providers while in training (54), absence of simple blood-based laboratory testing to evaluate for the presence or absence of obstructive sleep apnea, and the high cost associated with in-laboratory polysomnography, which has long been the gold standard for the diagnosis of sleep-disordered breathing. However, with the advent of widely accessible and lower-cost methods to assess for obstructive sleep apnea in the home environment (55), barriers to sleep testing are substantially lower than in previous decades. In the case of depression, unique opportunities exist to develop models of care that more deliberately consider sleep-disordered breathing in the diagnostic and therapeutic algorithms of care provision in ambulatory and inpatient treatment settings. While the identification of obstructive sleep apnea will not likely cure depression in most patients, identifying and treating this key comorbidity may improve the efficacy of existing treatments and decrease the likelihood of treatment resistance. Clearly, further research that clarifies the impact of unidentified sleep-disordered breathing on the presentation and treatment of treatment-resistant depression is warranted.

In addition to the role of occult sleep disorders such as sleep apnea contributing to treatment resistance, data suggest that sleep itself may be related to rapid antidepressant effects of emerging therapies. Sleep deprivation has long been known to rapidly improve depressive symptoms, but its effects are generally short-lived, with return of depression after a subsequent bout of sleep (56). While sleep deprivation is thus unlikely to be a clinically useful treatment strategy for patients, it does afford the opportunity to compare its effects with other rapidly acting antidepressants to understand potential underlying mechanisms of action.

The synaptic homeostasis hypothesis posits that extended wakefulness increases net synaptic strength, with subsequent renormalization during sleep reflected by increases in sleep slow-wave activity (57). Recently, this hypothesis has been combined with the synaptic plasticity hypothesis of major depression (58, 59), positing that the enhanced homeostatic plasticity of extended wakefulness during sleep deprivation allows persons with major depression to temporarily move into a window of long-term potentiation inducibility/associative plasticity, which under normal circumstances would be unattainable in depression (60). Notably, increases in slow-wave activity and slope during non-REM sleep are observed after a single infusion of ketamine (61). Thus, hypothetically, these findings during sleep suggest that ketamine may acutely increase synaptic strength in persons with depression, which could contribute to its antidepressant properties.

However, just as the potential mechanisms underlying ketamine’s rapid antidepressant activity may be many, the effects of ketamine on sleep may extend beyond changes in sleep slow waves and include changes in sleep efficiency, continuity, and REM sleep (62). Notably, decreased nighttime wakefulness and other aspects of improved sleep have been associated with reductions in suicidal ideation resulting from ketamine treatment (63, 64). Additionally, the effects of ketamine have been hypothesized also to have an impact on circadian rhythms, which along with sleep homeostatic processes, govern the overt sleep-wake rhythm (65, 66). Further research that clarifies the connections between sleep, circadian rhythms, and rapid responses to ketamine may both elucidate underlying mechanisms of action and lead to new areas of inquiry that leverage alterations in sleep and chronobiology to optimize the antidepressant response for patients with treatment-resistant depression.

The fact that depression is a heterogeneous construct is evidenced by the myriad combinations of symptoms allowable under current diagnostic criteria (67). The challenge of understanding the underlying biology of depression under such circumstances has been highlighted by the National Institute of Mental Health’s Research Domain Criteria initiative (RDoC), which emphasizes the study of measurable traits that occur across psychiatric disorders rather than nosological distinctions that may have limited validity (68, 69). Just as depression is heterogeneous, so too is sleep, and too often the consideration of sleep complaints in major depression in both clinical and research contexts occurs at a superficial level. Sleep disturbances are not all the same, and reliance on brief self-report measures alone is unlikely to meaningfully advance research or clinical care at the interface between sleep and depression. Within the RDoC framework, sleep falls under “Arousal and Regulatory Systems,” with three separate constructs (arousal, circadian rhythms, and sleep-wakefulness) available to more fully characterize sleep disturbances. The need for such an expansive tool kit to study sleep highlights the fact that sleep is a complex behavior generated by the brain and affects the entire organism and its functioning during wakefulness. During our lifespan, sleep plays a key role in development, and we spend more time sleeping than any other behavior during the course of our lives. In this context, it is crucial to more carefully phenotype the nature of a given sleep complaint in depressive illness to advance scientific progress at their nexus (70).

While the preponderance of research on the connections between sleep and depression have focused on insomnia, the relationships between hypersomnolence and depression highlight the importance of multiple levels of investigation and objective phenotyping of specific sleep complaints in clarifying relationships between sleep and psychiatric disorders. Hypersomnolence, broadly defined as excessive daytime sleepiness and/or excessive sleep duration, occurs in a minority of patients with major depression but, like insomnia, is associated with treatment resistance, symptomatic relapse, increased risk of suicide, and functional impairment (71, 72). Like insomnia, hypersomnolence is also associated with longitudinal risk of developing depression (73). From a physiological standpoint, the presence or absence of hypersomnolence in major depression also has key effects on differences in regional slow-wave activity during sleep when quantified using high-density EEG (74). Like all sleep complaints, hypersomnolence has multiple facets and quantifiable aspects that must be considered to fully understand their relationships with mood disorders. In fact, these different measurable aspects of hypersomnolence are particularly critical to consider in the context of depression because different objective tests used to quantify somnolence and daytime vigilance can exhibit divergent relationships with depressive symptoms (75, 76). These more discrete measurable aspects of hypersomnolence have also proven to be critical to identifying epigenetic changes associated with sleep disturbances, which would otherwise be missed by nonspecific self-report measures (77). Thus, more thorough consideration and delineation of the sleep phenotype is just as important to understanding the connections between sleep and depression as careful consideration of the multiple symptom domains that constitute depressive illness. From a broader transdiagnostic perspective, these discrete and measurable sleep-related traits may be particularly relevant for furthering scientific inquiry across the full spectrum of psychiatric disorders.

The ability to more accurately and objectively capture key sleep measures and phenotypes has long been hindered by low accessibility and high cost. Even beyond the use of home sleep testing for obstructive sleep apnea, emerging technologies offer the opportunity to dramatically lower barriers to the objective measurement of sleep in the ambulatory environment. Such methods have great promise to be adapted to tailor personalized sleep-related therapies early in the course of treatment and potentially reduce the burden of psychiatric illness. However, consumer-grade wearables are not yet accurate enough to define sleep stages, particularly REM sleep that may occur earlier than typical, as is often observed in depression (78 –81). The development, production, and release of wearable products into the marketplace far outpaces their validation in persons with mood disorders, as these devices are generally designed to estimate sleep in healthy populations. In the coming years, advances in these technologies will likely lead to tools that can be leveraged at the interface of sleep and depression, but it will be critical to fully validate these products in patients with depression before they can be applied broadly in research or clinical care settings.

Sleep and depression are deeply intertwined on numerous levels. Careful consideration, evaluation, and treatment of sleep disturbances will inform research and clinical care in depressive disorders. Recent advances and scalability in the evaluation and treatment of sleep disorders has great promise for future advances in this area. The close longitudinal relationship between insomnia and depression offers the tantalizing opportunity to prevent depressive episodes using evidence-based treatments for insomnia, which would represent a paradigm shift for psychiatry. Recognition and clarification of the role of disordered sleep in treatment-resistant depression may improve the efficacy and effectiveness of rapid-acting antidepressants and lead to other novel therapeutics. Careful consideration and measurement of the type and nature of sleep disturbance will also lead to transdiagnostic discoveries that can be applied and studied across the full spectrum of psychiatric disorders to continue to advance our understanding of the biological bases of complex psychopathologies and their connections with sleep.