Insomnia Phenotypes, Cardiovascular Risk and Their Link to Brain Health

Abstract

About 30% to 40% of the general population experience insomnia symptoms of difficulty initiating and/or maintaining sleep and another 10% to 15% experiences chronic insomnia disorder. The prevalence of insomnia is disproportionally higher in people with cardiometabolic risk factors (CMR), cardiovascular diseases (CVD), cerebrovascular diseases (CBVD), and neurocognitive disorders (NCD), including vascular cognitive impairment. In fact, recent meta-analytic evidence from epidemiological studies have demonstrated that insomnia, especially when accompanied by objective short sleep duration, is a risk factor for incident hypertension, type 2 diabetes, heart failure, stroke, cognitive impairment, Alzheimer’s disease, and all-cause mortality. Insomnia should, thus, be part of the prevention and management of these adverse health outcomes. However, randomized clinical trials have not demonstrated whether treatment with cognitive-behavioral therapy for insomnia (CBT-I), the first-line guideline recommended treatment, or hypnotics/sedatives improves heart- or brain-related outcomes. Studies have also failed to consider insomnia a heterogeneous disorder consisting of distinct phenotypes that result from the relative contribution of biological vs. cognitive-behavioral perpetuating factors. Objective short sleep duration has emerged as a marker of physiologic hyperarousal in insomnia (i.e., dysregulation of the hypothalamic-pituitary axis, increased sympathetic nervous system activation, and increased inflammation), as a predictor of insomnia-related adverse heart and brain health outcomes and, potentially, poor response to CBT-I. This review summarizes the meta-analytic evidence on the association of insomnia with CMR, CVD, CBVD, and NCD, including current knowledge on the heterogeneity of the disorder. This review also summarizes the potential pathophysiologic mechanisms that lead to heart and brain morbidity, which vary across insomnia phenotypes based on objective sleep duration. This review suggests that basic and clinical sciences need to unveil the molecular, cellular and behavioral mechanisms at play across insomnia phenotypes, as the public health and clinical implications of their association with adverse heart and brain health are demanding immediate attention.

1. Introduction

Insomnia is the most prevalent sleep disorder and presents with high comorbidity in people with cardiometabolic risk factors (CMR), cardiovascular diseases (CVD), cerebrovascular diseases (CBVD), and neurocognitive disorders (NCD). This high comorbidity suggests some potentially shared pathophysiologic mechanisms. Epidemiological data reviewed below suggest that insomnia can contribute to the development of CMR, CVD, CBVD, and NCD. In addition, physiologic data reviewed herein also suggest that patients with chronic insomnia experience hypothalamic-pituitary-adrenal (HPA) axis and sympathetic nervous system (SNS) activation and reductions in cardiac autonomic modulation, which can lead to endothelial dysfunction, high blood pressure (BP), and insulin resistance and, over time, contribute to heart failure, cardiac arrhythmias and myocardial infarction. The guideline-recommended first-line treatment for patients presenting with chronic insomnia complaints is cognitive-behavioral therapy (CBT-I), which aims at reducing nocturnal wakefulness (e.g., shortening sleep onset latency and wake after sleep onset), increasing sleep efficiency (i.e., matching time in bed to time asleep), improving daytime functioning (e.g., decreasing fatigue), and restructuring sleep-related cognitions (i.e., modifying maladaptive beliefs about sleep and insomnia). CBT-I has also demonstrated efficacy in reducing mental health-related outcomes. However, as reviewed below, randomized clinical trials (RCT) have not provided solid evidence that CBT-I or hypnotics/sedatives improve heart or brain health outcomes, which has left unknown whether the association between insomnia and these health disorders is causal or coexisting and whether CBT-I or hypnotics/sedatives are effective in reducing insomnia-related heart and brain health risk. Importantly, the extant evidence reviewed herein has shown that polysomnography (PSG)-measured sleep is highly variable within individuals with chronic insomnia complaints, reflecting individual differences in objective sleep across insomnia phenotypes and impacting the ability to establish a clinically meaningful association in epidemiological studies. Improved insomnia phenotyping in recent years has stimulated interest in considering chronic insomnia a heterogeneous disorder consisting of specific phenotypes that result from the relative contribution of cognitive-behavioral vs. physiological predisposing and perpetuating factors that lead to short vs. normal objective sleep duration, that have distinct downstream effects on heart and brain health, and that may respond differently to CBT-I and hypnotics/sedatives. Objective short sleep duration, typically defined as PSG-measured total sleep time (TST) < 6 hours in adults in the evidence reviewed below, has emerged as a marker of physiologic hyperarousal and a strong predictor of insomnia-related susceptibility to adverse health outcomes and poor treatment response (Figure 1). In this Review, the author summarizes the meta-analytic evidence on the association between insomnia and CMR, CVD, CBVD and NCD, including the mechanistic pathways that result in these health risks. The author also discusses current knowledge on the heterogeneity of insomnia from an objectively-measured sleep standpoint, and discusses the pathophysiology of objective short sleep duration in chronic insomnia and its role in stratifying health risk and predicting treatment response.

Figure 1. Insomnia phenotypes based on objective sleep duration.

The insomnia with short sleep duration (ISSD) phenotype is associated with physiologic hyperarousal, as indexed by a dysregulated hypothalamic-pituitary-adrenal axis (e.g., cortisol levels), autonomic nervous system (e.g., norepinephrine levels, metabolic rate) and cardiac autonomic modulation (e.g., blunted heart rate variability), with increased cardiometabolic risk (e.g., hypertension, diabetes, heart disease, stroke) and neurocognitive impairment (e.g., executive deficits, memory), and its etiology is purported to carry greater biological vulnerability (e.g., genetics). Individuals with the ISSD phenotype have also been shown to have a persistent, unremitting natural course, to present with a profile characterized by somatic complaints and depressed mood, and to accurately estimate, if not overestimate, their sleep duration (i.e., lower “sleep misperception”). In contrast, the insomnia with normal sleep duration (INSD) phenotype is not associated with physiologic hyperarousal, cardiometabolic risk or neurocognitive impairment. Individuals with the INSD phenotype have been shown to present with a profile characterized by anxious-ruminative traits, greater dysfunctional sleep-related beliefs, poor coping resources, to have higher remission rates in their natural course, and to significantly underestimate their sleep duration (i.e., “sleep misperception”). Both phenotypes have been shown to present with cognitive and cortical arousal, as indexed by EEG measures of increased information processing during the pre-sleep period and while asleep, with greater psychological vulnerability (i.e., emotional reactivity) and increased psychiatric risk, albeit through different psychobiological mechanisms (e.g., poor coping resources in the INSD phenotype). Finally, the ISSD phenotype responds worse to cognitive-behavioral therapy for insomnia (CBT-I), with response rates of 48% and remission rates of 32%, while the INSD phenotype responds well to CBT-I, with response rates of 77% and remission rates of 58%. Modified with permission from Fernandez-Mendoza. ¹¹

2. Insomnia definition and current diagnostic criteria

Although obstructive sleep apnea (OSA), insufficient sleep, and excessive sleep are recognized risk factors for adverse heart and brain health outcomes, ^1–4 insomnia disorder per se is not. Yet, insomnia is the most prevalent sleep disorder and is highly comorbid with mental (e.g., depression) and physical health conditions, including CMR, CVD, CBVD and NCD. Insomnia symptoms consist of self-reports of difficulty initiating sleep (DIS), difficulty maintaining sleep (DMS), and/or early morning awakening (EMA). ^5,6 About 30-40% of the general population report insomnia symptoms at any given time (regardless of chronicity) or any given functional consequence (regardless of associated daytime impairment). ^5,6Chronic insomnia disorder which is the most frequent presentation in the general population and in clinical practice, ^7,8 consists of self-reports of at least one of the insomnia symptoms mentioned above (criterion A), associated with significant daytime functioning impairment (criterion B), despite adequate opportunity (i.e., time allotted for sleep) and circumstances (i.e., safety, darkness, quietness, and comfort) for sleep (criterion C), occurring at least 3 nights per week (criterion D), for at least 3 months (criterion E), , and not solely due to another current sleep disorder, medical disorder, mental disorder, or medication/substance use (criterion F). ⁷ About 10-15% of the general population meets criteria for a chronic insomnia disorder based on subjective complaints. ^5–7As reviewed below, not many of the large epidemiological studies on the association of insomnia with adverse heart and brain health outcomes have examined these two mutually-exclusive categories of insomnia symptoms vs. chronic insomnia disorder, separately.

Another important aspect to briefly note is that insomnia research has not necessarily included objective sleep measures in the past few decades. ⁸ Insomnia symptoms are, by definition, subjective complaints and, thus, chronic insomnia disorder is a diagnosis based on self-reports, including the associated daytime functioning impairment (e.g., difficulties with attention, concentration or memory). ⁷ While there are suggested thresholds to determine clinically significant DIS or DMS (i.e., > 30 minutes), these quantitative criteria are not required for the diagnosis of insomnia disorder as shown above and are not used in the absence of a complaint to identify the presence of insomnia. ⁷ Furthermore, although individuals with chronic insomnia show shorter objective nighttime sleep, as measured by polysomnography (PSG) or actigraphy (ACT), and higher objective daytime alertness, as measured by the multiple sleep latency test (MSLT), ⁷ absolute mean differences are small and not as large as expected for the severity of complaints reported by those with chronic insomnia when compared to good sleepers. Together, these issues led the field to not recommend or require PSG, ACT or MSLT for the diagnosis or severity assessment of chronic insomnia disorder. ⁷ This is at odds with the other most widely diagnosed sleep-wake disorders, where PSG or home sleep apnea testing is indicated for the diagnosis and severity assessment of OSA, PSG coupled with MSLT is indicated for hypersomnia disorders, including narcolepsy, and ACT is recommended for circadian rhythm sleep-wake disorders. ⁷ Insomnia research and clinical practice has fallen behind in the use of objective sleep measures for multiple reasons; three are potentially the most relevant ones: 1) lack of clear sleep continuity or duration thresholds in prior nosology, despite assuming objective sleep differences among the proposed subtypes at the time; 2) demonstrated poor reliability and validity of prior nosology subtypes when evaluated across clinicians and sites; and 3) understanding small-to-null differences in objective sleep testing as invalid for insomnia research and clinical management. ⁸ In this Review, we will demonstrate how these small differences are rather indicative of high heterogeneity in physiologic sleep amongst individuals with chronic insomnia complaints, of the relative contribution of biological vs. behavioral factors to the pathophysiology of distinct phenotypes, and of their diverse association with heart and brain health outcomes vs. mental health outcomes (Figure 1). ^9–13

The aspects reviewed above are critical for understanding the state of the science, as revealed by extant systematic reviews and meta-analyses, for the association of insomnia and its phenotypes with cardiovascular, metabolic and neurocognitive morbidity and mortality.

3. Insomnia and Cardiometabolic Health

The association of insomnia with CMR, CVD or CBVD was largely ignored until about a decade ago. ¹⁴ Ignited by previous studies on pathophysiological findings in patients with chronic insomnia and studies on the cardiometabolic health consequences of chronic sleep restriction, studies begun to provide better insight onto the cardiometabolic consequences of chronic insomnia. This section reviews the adverse consequences of chronic insomnia on blood pressure, glucose regulation, CVD, CBVD, and their potential underlying stress and immune mechanisms. It also reviews evidence on how improved insomnia phenotyping based on objective sleep measures can provide stronger insight onto such associations and identify which phenotype has greatest cardiometabolic risk. As shown in Table 1, a total of 18 systematic reviews and meta-analyses have examined the association of insomnia symptoms, chronic insomnia and its phenotyping with adverse cardiometabolic health outcomes. ^15–32

Table 1.

Meta-analyses on the association between insomnia and adverse cardiometabolic health outcomes

First author, year (studies’ design)	N (# of studies)	Insomnia Definition	Outcome	Findings
Cappuccio, 2010 15 (Longitudinal)	24,812 (6)	Difficulty initiating sleep Difficulty maintaining sleep	T2D	RR = 1.57* RR = 1.84*
Meng, 2013 16 (Longitudinal)	42,636 (7)	Difficulty initiating sleep Difficulty maintaining sleep Early morning awakening Insomnia symptoms	HTN	RR = 1.17 RR = 1.20* RR = 1.14* RR = 1.05*
Li M, 2014 17 (Longitudinal)	311,260 (17)	Insomnia symptoms	MI CHD CBVD	RR = 1.41* RR = 1.28* RR = 1.55*
Sofi, 2014 18 (Longitudinal)	122,501 (10)	Insomnia symptoms	CVD/CBVD	RR = 1.45*
Li Y, 2014 19 (Longitudinal)	110,530 (10)	Difficulty initiating sleep Difficulty maintaining sleep Early morning awakening Nonrestorative sleep	CVD/CBVD	RR = 1.45* RR = 1.02 RR = 1.00 RR = 1.30*
Anothaisintawee, 2016 20 (Longitudinal)	289,588 (11)	Difficulty initiating sleep Difficulty maintaining sleep Insomnia symptoms	T2D	RR = 1.55* RR = 1.74* RR = 1.40*
Irwin, 2016 21 (Cross-sectional & longitudinal)	34,943 (31)	Sleep disturbance	CRP IL-6 TNF-α	ES = 0.12* ES = 0.20* ES = 0.07
He Q, 2017 22 (Longitudinal)	160,867 (15)	Difficulty initiating sleep Difficulty maintaining sleep Early morning awakening Nonrestorative sleep	CVD/CBVD	RR = 1.27* RR = 1.11* RR = 1.02 RR = 1.18*
Ge, 2019 23 (Longitudinal)	1,598,628 (29)	Difficulty initiating sleep Difficulty maintaining sleep Early morning awakening Nonrestorative sleep Insomnia disorder	CVD/CBVD Mortality	HR = 1.20* HR = 1.03 HR = 0.93 HR = 1.48* HR = 1.66
Li L, 2021 24 (Longitudinal)	395,641 (14)	Difficulty initiating sleep Difficulty maintaining sleep Early morning awakening Insomnia symptoms	HTN	RR = 1.14 RR = 1.27* RR = 1.14* RR = 1.21*
Zhang, 2021 25 (Cross-sectional)	151,299 (12)	Insomnia symptoms	HTN HPG HPL OBE	OR = 1.41* OR = 1.29* OR = 1.12 OR = 1.31*
Maiolino, 2021 26 (Cross-sectional)	924 (4)	Insomnia symptoms or disorder	SBPD DBPD	Percent = −2.00 Percent = −1.58
Ali, 2023 27 (Longitudinal)	2,583,117 (21)	Insomnia symptoms	MI CVD/CBVD Mortality	RR = 1.48* RR = 1.31* RR = 1.53*
Dean, 2023 28 (Longitudinal)	1,184,256 (9)	Difficulty initiating or maintaining sleep Nonrestorative sleep Insomnia symptoms	MI	RR = 1.13* RR = 1.06 RR = 1.69*
Wu, 2023 29 (Longitudinal)	>1 Million (25)	Insomnia symptoms	AFib CVD CHD MI CBVD HPL	RR = 1.30* RR = 1.45* RR = 1.28* RR = 1.42* RR = 1.55* RR = 1.64*
Zhao, Jiang, 2023 30 (Cross-sectional)	921 (17)	Insomnia disorder	HRV	R_equiv = 0.19
Johnson, 2021 31 (Cross-sectional)	3,034 (7)	Insomnia disorder < 6h of sleep	HTN T2D	RR = 1.54* RR = 1.63*
Dai, 2024 32 (Cross-sectional)	5914 (6)	Insomnia disorder < 6h of sleep	HTN	OR = 2.67*
Dai, 2024 32 (Longitudinal)	1963 (2)	Insomnia disorder < 6h of sleep	HTN	RR = 1.95*

AFib = atrial fibrillation. CVD = incident or death from cardiovascular disease, including myocardial infarction (MI), coronary heart disease (CHD) and/or heart failure (HF). CBVD = incident or death from cerebrovascular disease, including ischemic stroke. DBPD = diastolic blood pressure dipping. ES = effect size. HPG = hyperglycemia. HPL = hyperlipidemia. HTN = hypertension. MetS = metabolic syndrome. NS = not statistically significant. OBE = obesity. RR = relative risk. SBPD = systolic blood pressure dipping. T2D = incident type 2 diabetes.

^* = Statistically significant.

3.1. Elevated and dysregulated blood pressure and hypertension

The high comorbidity between insomnia and hypertension (HTN) was observed early in the 1970s. ³³ However, this coexistence was believed to be symptomatic (i.e., insomnia symptoms “secondary” to HTN), thus, the association of insomnia with clinical and subclinical measures of elevated blood pressure (BP) or BP dysregulation remained largely unexplored during the following decades. Several reviews ^14,34–36 and meta-analyses ^{16,24–26,31,32} have now been published on the association of insomnia with HTN and BP dysregulation. As shown in Table 1, population-based studies using self-reported data have shown a significant relationship between insomnia symptoms and HTN with estimated odds of prevalent HTN of 1.41-fold²⁵ and estimated risk of incident HTN of 1.05-fold to 1.21-fold,^16,24 depending on the subjective definition used and symptom examined. For example, these longitudinal meta-analyses showed a slightly stronger association of DMS (RR=1.20-1.27) than DIS (RR=1.14-1.17) with incident HTN.^16,24 Most of the available large epidemiologic studies covered by meta-analytic evidence adjusted for important demographic, clinical and behavioral factors such as sex, age, obesity, depression or smoking; however, they did not include PSG and were, thus, unable to control for the presence of OSA, or were able to examine the association for chronic insomnia disorder. For this and other reasons, the degree of association found in previous meta-analysis ¹⁶ and reviews ³⁷ was regarded as preliminary.

Vgontzas and colleagues seminal work in the Penn State Adult Cohort (PSAC) addressed those previous limitations and showed a synergistic effect between chronic insomnia and objective sleep duration < 6 hours on prevalent HTN, while adjusting for the potential effect of multiple demographic, clinical and behavioral factors, including sex, age, race/ethnicity, OSA, obesity, T2D, smoking, alcohol or depression among many others. This study first reported 5.1-fold increased odds of prevalent HTN in the insomnia with short sleep duration (ISSD) phenotype. ³⁸ Fernandez-Mendoza and colleagues further addressed the longitudinal association of the ISSD phenotype with incident HTN in the PSAC, reporting a 3.5-fold increased odds of incident HTN in the ISSD phenotype, while adjusting for the potential effect of multiple demographic, clinical and behavioral factors and baseline BP levels. ³⁹ Both studies reported that the insomnia with normal sleep duration (INSD) phenotype was not associated with significantly increased odds of either prevalent or incident HTN. ^38,39 Thereafter, several studies have replicated these findings, ^40–44 with the exception of a study in the Freiburg Insomnia Cohort (FIC). ⁴⁵ A meta-analysis based on crude descriptive data from 7 studies in 3,034 subjects showed that the ISSD phenotype was associated with a 1.54-fold significantly increased odds of prevalent HTN. ³¹ Another meta-analysis, based on multivariable-adjusted estimates (i.e., odds, hazard or relative risk ratios), from 6 studies in 5,914 subjects and from 2 studies in 1,963 subjects, including those from the PSAC, FIC or Sleep Heart Health Study (SHHS), showed that the ISSD phenotype was associated with 2.67-fold increased odds of prevalent HTN and a 1.95-fold increased risk of incident HTN. ³²

Of note, studies using subjective, instead of objective, measures of sleep duration to identify the ISSD phenotype report imprecise estimates, smaller effect sizes, and/or non-significant results. ^40,44,46,47 In addition, other studies have examined whether the association of insomnia with HTN is stronger when relying on objective measures of daytime alertness, such as the MSLT, given that individuals with chronic insomnia take longer to fall asleep on the MSLT compared to good sleepers, a finding primarily found in those with the ISSD phenotype. ^9,10,48 For example, one study showed that subjects with chronic insomnia who took longer than 14 minutes to fall asleep on the MSLT were associated with 3.3-fold to 4.3-fold increased odds of prevalent HTN, whereas subjects with chronic insomnia who fell asleep within 14 minutes on the MSLT were not significantly associated with prevalent HTN. ⁴⁹

Studies using objective nighttime sleep or daytime alertness measures, however, have not examined subclinical markers of BP regulation such as nighttime BP or non-dipping BP levels, as performed in studies on chronic insomnia without objective measures. ^50–52 Two recent meta-analyses did not support a statistically significant association between insomnia with BP-dipping or heart rate variability (HRV). ^26,30 A systematic review of 26 observational studies comprising 1,484 subjects suggested that 80% of the studies did find significant differences in cardiovascular activity, including BP dysregulation or impaired HRV, ⁵³ but that the ISSD phenotype presented most consistent findings in markers of impaired cardiovascular regulation across five studies. ^53–57 Well-controlled studies are clearly needed to examine the impact of the ISSD phenotype on BP regulation (e.g., circadian pattern, non-dipping), on different BP levels beyond stage 2 HTN (e.g., elevated systolic blood pressure, stage 1 HTN) as well as diurnal and nocturnal HRV, particularly using 24-h ambulatory measures and prospective designs.

In summary, current evidence supports an association of insomnia with HTN. However, early epidemiologic studies pertained to insomnia symptoms, rather than chronic insomnia disorder, were not able to control for OSA, and reported modest effect sizes. The replicated findings on the association of the ISSD phenotype with HTN have strengthened the evidence and helped identify who among those with chronic insomnia complaints have a clinically meaningful elevated HTN risk.

3.2. Insulin resistance, glucose dysregulation, and type 2 diabetes

Multiple cross-sectional and longitudinal studies have reported significant associations between the presence of insomnia symptoms and insulin resistance, increased fasting glucose levels, prevalent type 2 diabetes (T2D), or incident T2D. ^{15,20,25,31,47,59–62} As shown in Table 1, two meta-analyses of up to 11 longitudinal studies comprising up to 289,588 subjects, estimated that the risk of incident T2D associated with insomnia symptoms ranged from 1.55-fold to 1.84-fold, ^15,20 while another meta-analysis of 12 cross-sectional studies comprising 151,299 subjects estimated that the odds of prevalent hyperglycemia, including T2D, associated with insomnia symptoms were about 1.29-fold. ²⁵ The longitudinal meta-analyses showed a stronger association of DMS (RR=1.74-1.84) than DIS (RR=1.55-1.57) with incident T2D. The vast majority of these epidemiologic studies relied on broad self-reported definitions of insomnia symptoms and were able to adjust for important demographic, clinical and behavioral factors such as sex, age, obesity, depression or smoking, but not for OSA given the absence of PSG or other sleep testing. In fact, some in-lab studies did not find an association between chronic insomnia disorder and impaired metabolism. ^63,64 As with HTN above, the risk of T2D has also been examined for the ISSD and INSD phenotypes. Vgontzas and colleagues showed in the PSAC that the ISSD phenotype, but not the INSD phenotype, was associated with 2-to-3-fold significantly increased odds of prevalent T2D, even after adjusting for sex, age, race/ethnicity, obesity, OSA, HTN, smoking, alcohol use, and depression. ⁶⁵ Additional studies reported similar findings suggestive of insulin resistance and metabolic dysregulation in the ISSD phenotype, but not in the INSD phenotype. ^{45,58,66–68} A meta-analysis based on crude descriptive data from 6 studies showed that the ISSD phenotype is associated with a 1.63-fold significantly increased odds of prevalent T2D; ³¹ there is no meta-analysis in the literature based on multivariable-adjusted odds, hazard or relative risk ratios on prevalent or incident T2D across these insomnia phenotypes.

The evidence for an association between insomnia and metabolic syndrome (MetS) – the clustering of central obesity, insulin resistance, elevated BP and dyslipidemia – has been limited and inconsistent. ^69–72 Only one meta-analysis of 12 cross-sectional studies comprising up to 151,299 subjects was able to estimate the odds of each component of MetS associated with insomnia symptoms, which ranged from a non-significant 1.12-fold for prevalent hyperlipidemia to a significant 1.41-fold for prevalent elevated BP; ²⁵ however, the meta-analysis did not provide estimates for each insomnia symptom (e.g., DIS, DMS) or for chronic insomnia disorder, neither for prevalent MetS (i.e., the actual clustering of at least 3 cardiometabolic components). Interestingly, individuals with chronic insomnia disorder are typically non-obese, not significantly heavier than healthy controls, and not likely to develop obesity despite sleeping objectively shorter than controls, ^73–75 findings consistent with the presence of increased whole-body metabolic rate and physiologic hyperarousal. ^60,76 In addition, PSG sleep duration does not significantly correlate with body mass index in individuals with chronic insomnia disorder. ⁷⁴ These data indicate that chronic insomnia disorder may be linked to impaired glucose levels and T2D through underlying mechanisms other than central obesity. A preliminary study in the PSAC showed that the ISSD phenotype, but not the INSD phenotype (OR=1.03), was associated with a 2.04-fold significantly increased odds of prevalent MetS, to which elevated BP, hypercholesterolemia, and hyperglycemia, but not central obesity or hypertriglyceridemia, were the main contributors. ⁷⁷

In summary, evidence suggests a significant association between insomnia and metabolic dysfunction, including increased risk of T2D, with the available evidence suggesting that this risk is driven by the ISSD phenotype among those with chronic insomnia disorder.

3.3. Cardiovascular and cerebrovascular diseases

Clinical patients with chronic insomnia complaints have been observed to have a history of heart disease or stroke at a greater frequency than expected for the general population or similar outpatients. ^33,34,37 However, whether insomnia was associated with an increased risk of incident CVD or CBVD remained largely unexplored from the 1980s through the 2000s. As shown in Table 1, several systematic reviews and meta-analyses have been published on the association of insomnia with CVD and CBVD, ^{17–19,22,23,27–29} estimating that individuals with insomnia symptoms have a 1.28-fold to 1.69-fold increased risk of incident CVD or CBVD, with a stronger association of DIS (RR=1.20-1.45) than DMS (RR=1.02-1.11) with incident CVD or CBVD even after adjusting for multiple demographic, clinical and behavioral factors. ^19,22,23 However, most of this meta-analytic evidence pertains to insomnia symptoms, rather than chronic insomnia disorder,²³ and most the epidemiologic studies included did not include PSG or other sleep testing to control for OSA or examine insomnia phenotypes.

Following the findings in the PSAC on the association of the ISSD phenotype with increased risk of HTN and T2D, ^38,39,65 other independent investigators examined whether the ISSD phenotype is also more strongly associated with CVD or CBVD than the INSD phenotype. ^40,78–81 For example, a study in 4,437 subjects from the SHHS showed that the ISSD phenotype was associated with a 29% increased risk of incident CVD or CBVD after 11.4 years of follow-up, while individuals with the INSD phenotype were not at significantly increased risk of incident CVD or CBVD. ⁷⁸ Another study in 1,258 subjects from the PSAC showed that the ISSD phenotype was associated with 2.5-fold increased odds of incident CVD or CBVD after 9.2 years of follow-up, while the INSD phenotype was not. ⁷⁹ As with findings observed for HTN or T2D, subjective sleep duration did not provide adequate risk stratification to identify the ISSD phenotype and its increased risk of CVD or CBVD in either of these 2 large studies. ^78,79 Analyses of the Multi-Ethnic Study of Atherosclerosis, including between 1,429 and 2,188 older adults, have focused on subclinical markers of CVD finding that the ISSD phenotype, but not the INSD phenotype, is associated with increased circulating levels of cardiac troponin T. ^80,81 Together, these epidemiologic studies support an association of insomnia with CVD or CBVD, an association primarily driven by the ISSD phenotype.

3.4. Stress and immune mechanisms and health behaviors

Insomnia has been conceptualized as a sleep disorder for which its etiology is explained by a diathesis (predisposing traits) - stress (precipitating events) interaction, ^82,83 and its chronicity explained primarily by cognitive-behavioral (perpetuating) factors (Figure 2). ⁸³ This model has not only been supported by observational and experimental studies, but also by the efficacy of CBT-I. There is, however, neurobiological evidence indicating that physiologic changes in the arousal and stress systems are also involved the pathophysiology of insomnia, particularly the ISSD phenotype, ^9,84,85 and that these mechanisms may be responsible for the association of this insomnia phenotype, ant not the INSD phenotype, with adverse heart and brain health outcomes.

Figure 2. Pathophysiology and adverse cardiometabolic and neurocognitive outcomes of the insomnia with objective short sleep duration phenotype.

This diagram depicts the pathophysiology of insomnia with objective short sleep duration and its associated adverse heart and brain outcomes; it does not depict the etiology or etiopathogenesis (i.e., predisposing/diathesis and precipitating/stress factors) of chronic insomnia or its phenotypes. The diagram depicts how chronic insomnia, when identified solely based on subjective complaints of difficulty initiating and/or maintaining sleep, is primarily the result of cognitive-behavioral perpetuating factors, including cognitive and emotional arousal (e.g. racing mind and underlying high EEG frequency power, emotional inhibition/internalization, fear of sleeplessness) and specific sleep-inhibitory (e.g., cognitively/emotionally-arousing activities prior to bedtime), sleep-incompatible (e.g., activities in bed other than sleep and sex) and sleep-compensatory (e.g., excessive time in bed, catch-up sleep) behaviors. The diagram also depicts how objective short sleep duration in individuals with chronic insomnia complaints is primarily the result of physiologic hyperarousal, which includes hyperactivation of the arousal system including ascending projections from the brainstem and hypothalamus to the diencephalon, limbic system, basal forebrain, and neocortex (B-Hth-BF-Ctx) as well as descending projections regulating the autonomic nervous system. These ascending and descending projections interact with the acute or chronic activation of the HPA and SAM axes of the stress system (i.e., hypercortisolemia and sympathetic hyperactivation). Chronic insomnia and objective short sleep duration therefore intersect in a synergistic manner into a specific phenotype associated with adverse health outcomes, above and beyond those for which cognitive-behavioral factors are responsible for in chronic insomnia (i.e., psychiatric disorders). This elevated risk of heart and brain disorders is the result of purported mechanistic pathways, including endothelial dysfunction, blood pressure (BP) dysregulation, impaired cardiac autonomic modulation (CAM), chronic low-grade inflammation, insulin resistance/metabolic dysfunction or oxidative stress, that lead to CMR, CVD, CBVD and MCI/ADRD via coronary artery calcification, carotid intima media thickness or vascular insults, among many other central and peripheral subclinical changes. The diagram does not ignore the potential relative contribution of known constitutional factors in the pathophysiology of cardiovascular, metabolic and brain morbidity and mortality. It also depicts the potential relative role of inadequate health behaviors in increasing cardiometabolic disease risk. CBVD, cerebrovascular disease; CMR, cardiometabolic risk factors; CVD, cardiovascular disease; HPA, hypothalamic-pituitary-adrenal; SAM, sympatho-adrenal-medullary.

Studies dating back to the late 1960s ⁸⁶ focused on the neuroendocrine stress response such as increased hypothalamic-pituitary-adrenal (HPA) axis activation, as measured by cortisol levels, ^{9,84,85,87–90} or overnight norepinephrine and catecholamine metabolite levels in individuals with chronic insomnia, which were found primarily in the ISSD phenotype or to correlate with the degree of objective sleep disturbance. ^{9,84,87,88,90,91} Investigators have also found that insomnia is associated with increased nocturnal HR, blunted HRV, altered impedance cardiography, increased or altered low-grade inflammation, or increased central nervous system activation during wake and sleep. ^{53,56,57,92,93} Two systematic reviews and a meta-analysis have examined the association of insomnia with HR, HRV or other indices of cardiac autonomic modulation with mixed findings. ^30,53,92 Overall, the multiple studies to date suggest increased sympathetic modulation during both wake and nighttime periods in individuals with chronic insomnia disorder. ⁵³ Importantly, studies in which a positive association was reported, individuals with chronic insomnia were carefully screened and showed objective sleep disturbances, while studies that did not find an association between chronic insomnia and HRV parameters defined the disorder solely based on subjective reports. ^30,53 Spiegelhalder and colleagues ⁹³ exemplified this very clearly by finding that chronic insomnia, defined solely based on subjective criteria, was not associated with either resting HR or nighttime HRV; when investigators identified the ISSD phenotype within their sample, it showed higher HR and lower HRV than good sleepers, while the INSD phenotype showed similar HR and HRV as good sleepers. ⁹³ An additional study by Jarrin and colleagues replicated these findings in the ISSD phenotype. ⁵⁷

The reported association of insomnia with immune system activity has been modest. ²¹ Epidemiologic studies in adults reported no significant association between insomnia symptoms and C-reactive protein (CRP) levels, an acute-phase inflammatory protein of hepatic origin that increases following interleukin-6 (IL-6) secretion. ⁹⁴ However, in-lab controlled studies found increased inflammation in individuals with chronic insomnia compared to good sleepers, as measured by the level ⁹⁵ and diurnal pattern ⁹⁶ of IL-6 secretion. In these latter studies, individuals with chronic insomnia were carefully-screened and showed objective sleep disturbances. Evidence has continued to accumulate showing that biomarkers of stress and immune system hyperarousal are primarily present in individuals with the ISSD phenotype, including across the lifespan. ^{9,66,97–103} For example, a study in 378 subjects from the Penn State Child Cohort (PSCC) showed that adolescents with the ISSD phenotype had significantly higher CRP levels compared to good sleepers, while adolescents with the INSD phenotype had similar CRP levels as good sleepers. ¹⁰⁴ In addition, Tempaku and colleagues found that the ISSD phenotype was associated with 4.2-fold increased odds of short leukocyte telomere length in adults. ¹⁰⁵

Despite the evidence above that the ISSD phenotype is associated with stress and immune changes that may play a mechanistic role in the increased cardiometabolic risk observed in this insomnia phenotype, it is likely that inadequate health-related behaviors may be another potential pathway by which insomnia is linked to adverse cardiometabolic outcomes (Figure 2). Individuals with insomnia are more likely to report smoking, excessive alcohol or caffeine use, poor diet, lack of physical activity or show lower cardiorespiratory fitness. ¹⁰⁶ However, little evidence is available on how these health behaviors distribute across insomnia phenotypes and whether they play a relevant role. ¹⁰⁷ In fact, most epidemiological and physiological studies have controlled for many of these health behaviors in their designs or analyses. Nevertheless, more work is needed to establish the relative contribution and potential causal role of these inadequate health behaviors to the increased cardiometabolic risk associated with insomnia, above and beyond the other putative stress- and immune-related mechanisms. For example, recent studies have shown that the ISSD phenotype is associated with higher concentrations of acetate, butyrate, propionate, and total short-chain fatty acids as well as with higher levels of eight microbiota metabolites, benzophenone, pyrogallol, 5-aminopental, butyl acrylate, kojic acid, deoxycholic acid (DCA), trans-anethole, and 5-carboxyvanillic acid, as assayed in fecal samples, which speaks to potential pathways in the gut-brain axis contributing to adverse heart and brain health outcomes in this insomnia phenotype. ^108,109

In summary, biomarkers of neuroendocrine dysregulation, cardiac autonomic imbalance and chronic low-grade inflammation are found in ISSD, a phenotype in which 24-h physiologic hyperarousal is the primary pathophysiologic mechanism (Figure 2).

4. Insomnia and Brain Health

Experimental evidence has established that acute sleep deprivation and chronic sleep restriction degrade neurocognitive functions associated with vigilance, attention and memory, ¹¹⁰ via induction of long-term neuro-modulatory changes in brain physiology and metabolism, and activation of distributed neural networks and connectivity. ¹¹¹ Sleep has also been shown to play a role in the process of transferring newly acquired information into long-term memory, which is referred as sleep-dependent memory consolidation. ¹¹¹ This section briefly reviews the adverse consequences of chronic insomnia, and its phenotypes, on neurocognitive functioning, structural and functional brain changes, and NCD, including mild cognitive impairment (MCI), vascular cognitive impairment (VCI), and dementia. As shown in Table 2, a total of nine systematic reviews and meta-analyses have been conducted on these questions. ^29,112–119

Table 2.

Meta-analyses on the association between insomnia and adverse brain health outcomes

First author, year (studies’ design)	N (# of studies)	Insomnia Definition	Outcome	Findings
Fortier-Brochu, 2012 112 (Cross-sectional)	1,197 (24)	Insomnia disorder	Episodic memory Problem solving WM manipulation WM retention	ES = −0.51* ES = −0.42* ES = −0.42* ES = −0.22*
Tahmasian, 2018 113 (Cross-sectional)	799 (19)	Insomnia disorder	sMRI fMRI PET	P = 0.914
de Almondes, 2016 114 (Longitudinal)	37,237 (5)	Insomnia symptoms	ACD	RR = 1.53*
Shi, 2018 115 (Longitudinal)	246,786 (9 / 18)	Insomnia symptoms	ACD AD VaD	RR = 1.17 RR = 1.51* RR = 1.13
Wardle-Pinkston, 2019 116 (Cross-sectional)	4,539 (48)	Insomnia symptoms or disorder	Overall Complex attention Episodic memory Perception Problem solving WM manipulation WM retention	g = −0.24* g = −0.36* g = −0.29* g = −0.24* g = −0.39* g = −0.52* g = −0.30*
Wu, 2023 29 (Longitudinal)	>1 Million (25)	Insomnia symptoms	AD	RR = 1.51*
Weihs, 2023 117 (Cross-sectional)	1,085 (3)	Insomnia disorder	sMRI	P = NS
Ballesio, 2023 118 (Cross-sectional)	1,172 (8)	Insomnia disorder	BDNF	g = −0.86*
Ren, 2023 119 (Cross-sectional)	1,339 (5)	Insomnia disorder < 6h of sleep	Overall Attention Memory Executive function	g = −0.56* g = −0.86* g = −0.47* g = −0.39*
Ren, 2023 119 (Cross-sectional)	1,339 (5)	Insomnia disorder > 6h of sleep	Overall Attention Memory Executive function	g = −0.003 g = 0.19 g = −0.22 g = 0.14

ACD = all-cause dementia. AD = Alzheimer’s disease. BDNF = brain-derived neurotrophic factor. ES = effect size. fMRI = functional magnetic resonance imaging. g = Hedges’ g effect size. sMRI = structural magnetic resonance imaging, including voxel-based morphometry. NS = non-statistically significant. PET = positron emission tomography. RR = relative risk. VaD = vascular dementia. WM = working memory.

^* = Statistically significant.

4.1. Neurocognitive deficits

An insomnia diagnosis requires the presence of both nighttime sleep difficulties and daytime functioning impairments. Chronic insomnia patients often complain of moderate-to-severe cognitive problems in alertness, attention, concentration, and memory; however, meta-analytic neurocognitive data reveal only small-to-moderate deficits. ^112,116,120 Variable testing protocols and heterogeneous samples may contribute to the subtle cognitive deficits observed. ^{112, 116,120} To address the heterogeneity issue, several studies examined the relationship between the INSD and ISSD phenotypes with neurocognitive performance in adults who have not yet developed MCI or dementia. Recent meta-analytic data (Table 2) from five of these observational studies across 1,339 adults ^121–125 has shown that the ISSD phenotype was associated with overall neurocognitive performance deficits, while the INSD phenotype was not and showed similar performance in neurocognitive tests as good sleepers. ¹¹⁹ Neurocognitive deficits in the ISSD phenotype were particularly in the domains of set-switching attention, memory, and executive functions (Table 2). ¹¹⁹ Using a different approach, Edinger and colleagues (2013) showed that individuals with chronic insomnia and increased daytime sleep latency (MSLT > 8 minutes) had lower nighttime sleep efficiency, suggesting 24-hour physiologic hyperarousal, and greater error rates on set-switching attention tasks. ¹²⁶ Together, these data further indicate that objective measures like short PSG sleep or long MSLT latencies can serve as insomnia-specific markers of physiologic hyperarousal and predict its impact on higher-order executive functions, ^127,128 which cooperate in the dorsolateral (DLPFC) and medial (MPFC) areas of the prefrontal cortex and anterior cingulate cortex (ACC). These findings are important given the recognized role of vascular, metabolic and cardiac contributors to NCD (i.e., VCI), including Alzheimer’s disease (AD). ^4,129–131

4.2. Neuroimaging changes

Over the past two decades, numerous neuroimaging studies have been conducted on individuals with chronic insomnia, but the results have been inconsistent and diverse. A meta-analysis of 19 studies using task-based functional magnetic resonance imaging (fMRI), resting-state fMRI, voxel-based structural morphometry (sMRI), and positron emission tomography found no statistically significant convergent evidence for a combination of structural atrophy or functional disturbances in individuals with chronic insomnia (Table 2). ¹¹³ Another meta-analysis of sMRI data could not identify any association between an insomnia brain score and the presence of insomnia symptoms in independent cohort studies or an association with subjective insomnia severity, except for subjective sleep quality (Table 2). ¹¹⁷ However, a systematic review of resting-state fMRI in chronic insomnia and OSA concluded that the salience network is crucial in hyperarousal and affective symptoms in insomnia, while dysfunctional connectivity of the posterior default-mode network underlies the cognitive and depressive symptoms of OSA. The central executive network, which is involved in higher-order cognitive functions, was not found to be associated with chronic insomnia. ¹³² A recent meta-analysis concluded that inconsistencies across sMRI and fMRI studies in chronic insomnia might be related to clinical heterogeneity, small sample sizes, explorative nature of studies, different experimental designs, and the variety of preprocessing and statistical approaches used, recommending that future neuroimaging studies should include large, well-characterized samples and standard imaging and analysis protocols. ¹¹³ Only one study to date has examined the association of the ISSD phenotype with brain changes as measured by fMRI, ¹³³ and none with sMRI. Magnetic resonance spectroscopy (MRS) has been used to study neurochemical brain changes in individuals with chronic insomnia. Seven studies have examined neurotransmitters and amino-acids in the brain, with inconsistent findings. While five studies showed decreased γ-aminobutyric acid (GABA) and/or increased glutamate/glutamine concentrations in chronic insomnia, ^134–138 Spiegelhalder and colleagues ¹³⁹ could not replicate such findings and rather found that glutamate/glutamine levels in the DLPFC increased across the day and shorter habitual sleep duration was associated with lower GABA levels in the ACC in those with chronic insomnia. The authors suggested that reduced GABA levels may be a trait marker of objective sleep disturbances, while increasing glutamate/glutamine levels may reflect hyperarousal at bedtime in insomnia patients. ¹³⁹ Miller and colleagues found that the ISSD phenotype was associated with reduced aspartate and glutamine concentrations in the left occipital cortex. ¹⁴⁰ Additionally, a recent meta-analysis of 8 studies comprising 1,172 subjects showed that insomnia is associated with significantly reduced brain-derived neurotrophic factor (BDNF) levels. ¹¹⁸ Interestingly, two studies have shown that BDNF levels are reduced in the ISSD phenotype, but not in the INSD phenotype. ^141,142 Together, these data suggest that changes in brain biochemistry may be primarily found in individuals with the ISSD phenotype.

4.3. Neurocognitive disorders

A systematic review and meta-analysis of five studies found that insomnia symptoms were associated with a 1.53-fold risk of incident all-cause dementia, with high heterogeneity across studies. ¹¹⁴ Two meta-analyses of up to 25 longitudinal studies found that individuals who self-reported insomnia symptoms had a higher risk of incident all-cause dementia, AD, and vascular dementia, with specific analyses indicating a 1.51-fold increased risk of AD, even after adjusting for multiple demographic, clinical and behavioral factors. ^{29, 115} None of these meta-analyses were able to provide estimates for each insomnia symptom (e.g., DIS, DMS) on their association with prevalent or incident NCD. All three meta-analyses, however, pointed out methodological issues in the original studies in terms of the definition of insomnia used (symptoms vs. disorder) and absence of objective sleep measures to examine objectively-defined insomnia phenotypes. A recent study of 1,524 adults in the PSAC showed that the ISSD phenotype was associated with 2.2-fold increased odds of MCI and 2.3-fold increased odds of possible VCI, defined by the presence of both MCI and CMR, CVD or CBVD; in contrast, the INSD phenotype was not associated with increased odds of either MCI (OR=0.75) or possible VCI (OR=0.76). ¹²⁵ Another study in 208 subjects older than 60 years old, of whom 124 had MCI and 84 were not cognitively impaired, showed that MCI subjects with the ISSD phenotype, but not with the INSD phenotype, had increased cortisol levels. ¹⁴³ In conclusion, the ISSD phenotype is associated with deficits in higher-order neurocognitive functioning and increased risk of NCD. However, the underlying neural mechanisms and source of these neurocognitive deficits remain largely unknown, given the lack of neuroimaging studies in objectively-measured insomnia phenotypes. Future studies should examine whether the ISSD phenotype in middle-age is indeed associated with neurochemical, structural and/or functional brain changes that lead to cognitive decline and eventual dementia in older adulthood.

5. Clinical and Public Health Implications

Considering that about 75% of those with insomnia disorder persist over time and that about 15-20% who report insomnia symptoms will eventually develop the disorder, ^144–148 it is clear that it should be a target of public health policies. A clinical concern is how to disentangle insomnia as a “secondary” symptom of cardiometabolic and neurocognitive disorders vs. a comorbid disorder. As early posited, ^83,149 it is unequivocal that insomnia symptoms can arise from such conditions and/or their medical treatments, however, the nature of chronic insomnia disorder being “secondary” to medical conditions is on a much more restrictive scale than typically believed. ¹⁴⁹ During the acute onset of cardiovascular, metabolic or neurocognitive disorders, there can be in some individuals an insomnia response that does parallel the course of the medical disorder, thus, the importance of criterion F in current nosology. ⁷ However, insomnia symptoms “secondary” to a medical precipitant during an acute and transient phase will acquire self-sustaining perpetuation after a period of time by virtue of the patient’s cognitive-behavioral response to the nighttime sleep disturbance, and what was “secondary” will become “comorbid”. ^83,149 The importance of cognitive-behavioral factors in perpetuating chronic insomnia complaints is depicted in Figure 1. Moreover, the prospective evidence included in Tables 1 and 2 support that insomnia, in a significant proportion of cases, can precede cardiometabolic and neurocognitive morbidity and serve as a risk or vulnerability factor for adverse heart and brain outcomes. An issue to be addressed in longitudinal studies is the lack of repeated measures of insomnia, objective sleep and clinical factors (including changes in medication use) that can properly model the lifespan trajectories of each of these factors as a function of aging and age-related cardiometabolic risk and cognitive decline.

An improved detection of insomnia via brief, reliable, valid and easy to use self-reported tools, such as the insomnia severity index (ISI) and consensus sleep diary, as well as objective measures of sleep duration such as home-based PSG, ACT or consumer wearables should lead to early identification and phenotyping of those at greatest risk of adverse cardiometabolic and brain outcomes. This is important because, once insomnia is diagnosed, objective short sleep serves as a marker of its biological severity (Figures 1 and 2), while the resulting ISSD and INSD phenotypes do not reliably differ in terms of their reported perceived insomnia severity (e.g., ISI scores). ^40,45,57,101 In addition, this approach will allow properly controlling for the potential effect of conditions such as restless legs syndrome, periodic limb movements or circadian misalignment, a limitation of the existing epidemiological evidence. This improved phenotyping should also lead to early targeted treatment, helping prevent downstream consequences on health, including cardiometabolic and neurocognitive risk.

Although the currently recommended first-line treatment of insomnia, CBT-I, has been shown to improve subjective sleep continuity and insomnia severity in individuals with comorbid insomnia and heart or brain health disorders, ^150–152 RCTs that have prospectively assessed whether CBT-I, or the most commonly-used pharmacological therapies (e.g., benzodiazepine receptor agonists, sedative anti-depressants, dual-orexin receptor antagonists), improve cardiovascular, metabolic or neurocognitive functions as primary outcomes are scarce. The available systematic reviews and meta-analyses of existing RCTs indicate inconclusive findings, pointing to high risk of bias, heterogeneity, and design issues of otherwise preliminary RCTs examining cardiometabolic or neurocognitive effects. ^153–158 For example, a systematic review of 15 RCTs involving 2,067 subjects concluded that CBT-I appeared to be associated with improved hemoglobin A1c and CRP levels (k = 6 studies), however, nine of the 15 RCTs were pilot studies, five were in specific patient populations (e.g., cancer, hemodialysis, bipolar disorder), and there was a high risk of bias, which limited the interpretation of the findings. ¹⁵³ A meta-analysis of 12 RCTs involving 2,044 subjects showed that CBT-I and/or sleep hygiene was associated with significantly improved hemoglobin A1c levels, particularly among those with T2D (k = 9 studies); however, the same limitations applied to this meta-analytic evidence, including a high heterogeneity of treatment effects. ¹⁵⁶ In addition, a systematic review of 18 RCTs involving 923 subjects concluded that CBT-I was associated with small-to-moderate improvement in subjective cognitive functioning; however, this evidence was deemed as preliminary, while that for objective cognitive performance as too limited (k = 4 studies). ¹⁵⁷ Commensurate, a recent meta-analysis of 24 RCTs involving non-pharmacological (i.e., , yoga, CBT-I, relaxation, sleep restriction) and pharmacological (i.e., ayurveda, zolpidem, vestipitant, trazodone, melatonin, ramelteon, eszopiclone, temazepam, and flurazepam) therapies showed a stronger effect on cognitive functioning when measured subjectively than objectively, with similar effect sizes for non-pharmacological and pharmacological therapies; however, the merging of diverse sleep interventions and high heterogeneity of effects observed, importantly limits the interpretation of these findings. ¹⁵⁸ Overall, the majority of RCTs that have tested the effect of CBT-I on cardiometabolic or neurocognitive measures were pilot studies and failed to show meaningful effects not confounded by extraneous clinical or therapeutic factors.

Of note, RCTs have shown that CBT-I does not statistically and/or clinically improve PSG- or ACT-measured sleep parameters, including lengthening TST. ^150,159 Recent meta-analytic data of nine RCTs has shown that the ISSD phenotype, as identified by PSG or ACT, has a 29% lower response rate (48% vs. 77%) and a 26% lower remission rate (32% vs. 58%) than the INSD phenotype after CBT-I. ¹⁶⁰ In addition, omnibus analyses showed that the ISSD phenotype responded significantly worse than the INSD phenotype after CBT-I on secondary outcomes; particularly, PSG-measured sleep (including TST), insomnia severity index, and dysfunctional beliefs and attitudes about sleep. ¹⁶⁰ No physiological or cardiometabolic data were available across studies that could be meta-analyzed. However, one of the pilot studies included showed that trazodone, but not CBT-I, significantly lengthened ACT-measured TST and decreased salivary-assayed cortisol levels the ISSD phenotype. ¹⁶¹ These data were consistent with prior observations with doxepin in chronic insomnia ¹⁶² and suggest that other therapeutic approaches are needed for those with the ISSD phenotype ^163,164 in order to achieve response and remission rates comparable to those obtained by individuals with the INSD phenotype after CBT-I (i.e., 60% or higher). These initial therapeutic findings provide further support for the validity of insomnia phenotyping efforts based on objectively-measured sleep duration. Nevertheless, evidence from well-designed, statistically powered prospective RCTs is needed to test the differential effectiveness of current insomnia therapies (behavioral, pharmacological or their combination), in improving subclinical and prognostic markers of cardiovascular, metabolic and neurocognitive risk in chronic insomnia disorder and, in particular, its ISSD and INSD phenotypes, ¹⁶⁵ given the high-risk of bias, heterogeneity and design issues of the current RCT evidence. ¹⁶⁰

6. Needed Avenues for Basic, Translational and Clinical Science

Despite the literature reviewed above, there are still major issues related to the etiopathogenesis and pathophysiology of insomnia and its phenotypes that laboratory and clinical studies need to address in order to answer critical questions pertaining to cardiometabolic risk.

There are many studies on the epidemiology of insomnia; however, few have methodological strengths that allow studying individuals with chronic insomnia disorder, over time with longitudinal designs, in diverse populations, using a combination of self-reports and objective sleep measures, and including preferably physiologic tests to account for OSA’s comorbidity and more accurately ascertain physiologic sleep duration or other electrophysiological metrics for an improved phenotyping. For example, given the high comorbidity of insomnia and depression, observational studies should further disentangle their independent and synergistic contributing risk on developing cardiometabolic and neurocognitive disorders; although there is no solid causal evidence that depression mediates the impact of insomnia on heart and brain health or vice versa, ¹⁶⁶ many of the pathways through which depression influences heart and brain health overlap with putative pathways through which insomnia (Figure 2) may affect the pathophysiology and clinical course of these disorders. ¹⁶⁶ As it pertains to improved insomnia phenotyping, the identification of the ISSD phenotype has been supported by five unsupervised data-driven studies of up to 100,000 adults that did not use a priori cut-offs for objective sleep measures and provide validity its phenotyping. ^{101,140,167–169} As it pertains to its stability, researchers have estimated a short-term stability of 75% for the ISSD phenotype in a sample of 150 adults with chronic insomnia complaints assessed with three consecutive PSG nights ¹⁷⁰ and 58% in a sample of 61 adults with chronic insomnia complaints assessed with two consecutive PSG nights. ¹⁷¹ The long-term stability of the ISSD and INSD phenotypes has been estimated at 70-75% across two PSG nights 2.5 years apart. ¹⁷⁰ Methodological issues addressing a more precise identification of insomnia phenotypes need to be addressed in future studies addressing cardiovascular, metabolic and brain health risk, including those leveraging existing “big data” resources. ¹⁷²

Genetic and epigenetic studies of insomnia remain an area in need of innovative research in order to address its biological predisposition and interaction with environmental precipitating events. ¹⁷³ This is important from heart and brain health standpoint because two genome-wide association studies conducted in 1,331,010 and 2,365,010 subjects have identified between 200 to 500 loci associated with insomnia symptoms, which suggests extreme polygenicity, ^174,175 and showed that genes associated with metabolic and psychiatric neural pathways were involved in insomnia symptoms via genome-wide meta-analysis ¹⁷⁵ and identified causal effects of insomnia symptoms on diabetes, cardiovascular disease, and depression via mendelian randomization. ¹⁷⁴ However, no genetic or epigenetic study to date has relied on deeper measures of insomnia and identified objectively-measured phenotypes at greatest risk of cardiometabolic and neurocognitive morbidity and mortality. Only one study in 173 adults has used serum metabolomics and found 52 serum metabolites related to the ISSD phenotype compared to good sleepers, of which indoxyl sulfate was associated with higher BP levels. ¹⁷⁶ In a subset of 34 adults with ISSD, fecal 16S rDNA amplicon sequencing was also obtained to explore gut microbiota distribution, to find that indoxyl sulfate was positively associated with bacteroidetes abundance and negatively associated with firmicutes abundance. ¹⁷⁶ Large, population-based studies are needed to replicate these findings and uncover the molecular underpinnings of the increased cardiovascular risk, or lack thereof, across insomnia phenotypes.

Furthermore, a systemic psychobiological perspective is needed in novel experimental studies in order to fine-tune the pathophysiology of chronic insomnia and the contribution of physiological and (cognitive-)behavioral perpetuating mechanisms to its heterogenous phenotypes. For example, experimental studies should directly test whether activation of the renin-angiotensin-aldosterone system, sympathetic nervous system or HPA axis act as pathophysiologic mechanisms (mediators) linking insomnia, and its phenotypes, with cardiovascular, metabolic and neurocognitive functioning. ¹⁶⁶ This is a critical avenue for research as better understanding arousal, homeostatic and circadian pathways may also lead to a more robust mechanistic understanding of the causal impact of insomnia on heart and brain heath, via vascular, immune and metabolic pathways, that can inform phenotype-specific interventions or novel therapeutic approaches. Finally, insomnia remains a challenge for non-human experimental studies aiming to disentangle its etiology and pathophysiology, its associated health risks or treatment response at a cellular and molecular level. ¹⁷⁷ To date, non-human models have shown that 1) psychosocial stress in male rats induces sleep continuity disturbances with short sleep that are the result of activation of cerebral cortex, limbic system, and parts of the arousal and autonomic nervous system with simultaneous activation of sleep-promoting areas of the anterior hypothalamus (e.g., ventrolateral preoptic area and median preoptic nucleus); ¹⁷⁸ 2) selectively bred short sleep in Drosophila is associated with sleep continuity disturbances and a core molecular clock that remains intact, a phenotype that is heritable and is associated with hyperactivity and hyperresponsiveness to environmental perturbations, daytime cognitive impairment, elevated dopamine, triglycerides, cholesterol, and free fatty acids levels, and short life expectancy; ¹⁷⁹ and 3) experimentally mismatching sleep opportunity and ability in short sleeping Drosophila produces sleep continuity disturbances, mutant short sleeping Drosophila exhibit a mismatch between sleep opportunity and sleep ability, and this mismatch can be rectified by changing the light-dark period. ¹⁸⁰ As can be gathered, these non-human organisms appear to model the human ISSD phenotype (i.e., short sleep with sleep continuity disturbances) and may provide support for its pathophysiology and adverse health outcomes; the human INSD phenotype is likely a significant challenge, yet not impossible, for non-human model organisms. It is critical to extend these non-human models to understand the cellular and molecular mechanisms linking insomnia phenotypes to heart and brain health outcomes. ¹⁸¹

Form a clinical and phenotypic standpoint, the comorbidity of insomnia and sleep apnea (COMISA) has been increasingly studied in the past decade given the high coexistence rates (40-50%) between the two disorders and accumulating data supporting increased morbidity and mortality with COMISA than with either condition alone. ¹⁸² Although the physiological bases for these differential effects of COMISA on health are not well understood, Redline and colleagues have proposed that they could reflect the additive or synergistic effects of short sleep (insomnia) and disordered breathing (OSA) on the underlying pathophysiological pathways shown in Figure 2 for cardiometabolic disease. ¹⁸² For example, insomnia may exacerbate the existing sympathetic hyperactivity, increased oxidative stress or low-grade inflammation of OSA. ¹⁸² Indeed, a recent study supports such synergism and indicates that COMISA with the ISSD phenotype is associated with higher plasma IL-6 concentrations, greater glucose/insulin ratio, and 3.9-fold increased odds of HTN compared to COMISA with the INSD phenotype. ¹⁸³ Future studies should further endophenotype COMISA informed by this improved OSA and ISSD phenotyping approach using non-invasive measures of respiratory function (e.g., compromised airway anatomy/collapsibility, low arousal threshold, high loop gain, poor upper airway muscle compensation, cardiopulmonary coupling, odds ratio product or hypoxic burden) and objective (e.g., PSG or wearables), ¹⁸³ rather than subjective, ¹⁸⁴ measures of sleep duration.

Finally, there is a clear need for in-lab and at-home studies using state-of-the-art methods to assess cardiovascular, metabolic, cerebrovascular, and neurocognitive function in individuals with chronic insomnia and its phenotypes in order to identify the most reliable biomarkers for the health outcomes of interest and incorporate them into RCTs.

7. Conclusions

In summary, insomnia is not just a symptom of physical and mental health problems, rather it is most frequently a chronic disorder that puts individuals at risk of CMR, CVD, CBVD, VCI, and dementia. Evidence over the last two decades indicates that the association of insomnia with cardiometabolic and neurocognitive morbidity is primarily present in the ISSD phenotype. Based on the current evidence and effect sizes reviewed herein, ISSD should be included as a modifiable risk factor for cardiovascular, metabolic and neurodegenerative disease risk to a similar extent as moderate-to-severe OSA is. There is a need for in-lab and at-home studies using state-of-the-art methods to assess cardiac physiology, metabolic regulation, central vascularity, white-matter aberrations, and glymphatic activity in individuals with chronic insomnia using objective sleep measures for their phenotyping. Similarly, the field requires large, long-term longitudinal studies capable of establishing the relationship between insomnia phenotypes with the development of subclinical and clinical outcomes of brain health. Finally, well-designed prospective RCTs are needed to test whether jointly improving chronic insomnia complaints and lengthening objective sleep duration leads to favorable cardiovascular, metabolic and neurocognitive outcomes.

Non-Standard Abbreviations and Acronyms

ACC

anterior cingulate cortex

AD

Alzheimer’s disease

BMI

body mass index

BP

blood pressure

CBVD

cerebrovascular disease

CHD

coronary heart disease

CMR

cardiometabolic risk factors

CRP

C-reactive protein

CVD

cardiovascular disease

DBP

diastolic blood pressure

DLPFC

dorsolateral prefrontal cortex

fMRI

functional magnetic resonance imaging

HPA

hypothalamic-pituitary-adrenal

HR

heart rate

HRV

heart rate variability

HTN

hypertension

IL-6

interleukin 6

MCI

mild cognitive impairment

MetS

metabolic syndrome

MI

myocardial infarction

MPFC

medial prefrontal cortex

MRS

magnetic resonance spectroscopy

MSLT

multiple sleep latency test

NCD

neurocognitive disorders

OSA

obstructive sleep apnea

PET

positron emission tomography

PSG

polysomnography

sMRI

structural magnetic resonance imaging

SAM

sympatho-adrenal-medullary

SBP

systolic blood pressure

TNF-α

tumor necrosis factor alpha

T2D

type 2 diabetes