Sleep Medicine 15 (2014) 485–495
Contents lists available at ScienceDirect
Sleep Medicine journal homepage: www.elsevier.com/locate/sleep
Review Article
Insights into obstructive sleep apnea research Mohammad Badran a, Najib Ayas b, Ismail Laher a,⇑ a
Department of Pharmacology and Therapeutics, Faculty of Medicine, University of British Columbia, Vancouver, BC V6T 1Z3, Canada Divisions of Critical Care and Respiratory Medicine, Department of Medicine, University of British Columbia, Sleep Disorders Program, UBC Hospital, Division of Critical Care Medicine, Providence Health Care, Vancouver, BC, Canada b
a r t i c l e
i n f o
Article history: Received 25 October 2013 Received in revised form 3 January 2014 Accepted 8 January 2014 Available online 10 February 2014 Keywords: Intermittent hypoxia Obesity Oxidative stress Inflammation Endothelium Animal models
a b s t r a c t Moderate to severe obstructive sleep apnea (OSA) occurs in 10–17% of middle aged men and 3–9% of middle-aged women with a higher prevalence among obese subjects. This condition is an independent risk factor for many cardiovascular diseases. Intermittent hypoxia is a major pathophysiologic character of OSA; it can lead to oxidative stress and inflammation, which in their turn cause endothelial dysfunction, a hallmark of atherosclerosis. Many animal models have been designed to mimic OSA in human patients to allow more in-depth investigation of biological and cellular mechanisms of this condition. This review discusses the cardiovascular outcomes of OSA and some of the animal models that are being used to investigate it. Ó 2014 Elsevier B.V. All rights reserved.
1. Obstructive sleep apnea Obstructive sleep apnea (OSA) is characterized by momentary cessations in breathing (apnea) or significant reductions in breathing amplitude (hypopnea) caused by an obstructed or collapsed upper airway; both conditions can cause significant arterial hypoxemia and hypercapnia. The apnea/hypopnea index (AHI) describes the total number of apnea/hypopnea episodes per hour of sleep, which is usually 30 categorize patients with sleep apnea as mild, moderate, and severe, respectively [1]. An obstructed airway increases resistance to airway flow that results in a greater breathing effort and swings in intrathoracic pressure, resulting in disruption of sleep, arousal, and reopening of the airway [2]. It is estimated that there are currently approximately 14% of men and 5% of women in the USA who have an AHI P 5, plus symptoms of daytime sleepiness, with 13% of men and 6% of women also having moderate to severe OSA (AHI P 15) [3]. Based on the average of prevalence estimates from many clinical studies, it is estimated that nearly one of every five adults has at least mild OSA and that one of every 15 has at least moderate OSA; moderate OSA occurs predominantly at body mass index (BMI) values of 25– 28 [4–6]. Despite numerous advancements in medicine, the majority of those affected with OSA remain undiagnosed [7]. OSA is sus⇑ Corresponding author. Tel.: +1 (604) 822 5882; fax: +1 (604) 224 5142. E-mail address: [email protected] (I. Laher). http://dx.doi.org/10.1016/j.sleep.2014.01.009 1389-9457/Ó 2014 Elsevier B.V. All rights reserved.
pected in people who are obese, hypertensive, hypersomnolent and habitual snorers [8]. Polysomnography is the main method for assessing patients with suspected sleep apnea [9]. Sleep stages are recorded along with oxyhemoglobin saturation, breathing, and airflow. In addition, limb and eye movements and the electrocardiogram are also monitored [10]. OSA creates a huge economic burden when compared to other chronic diseases. In 2000, OSA-related automobile collisions alone attributed to 1400 fatalities and a total cost of 15.9 billion dollars in the USA. Treatment with continuous positive airway pressure (CPAP) resulted in saving 7.9 billion dollars and 1000 lives [11]. It is well established that the outcomes of OSA can lead to serious vascular diseases. Data from different studies implicate OSA in the development of hypertension and, to some extent, cardiac ischemia, congestive heart failure, arrhythmias, cerebrovascular disease, and stroke [12]. Many intermediary mechanisms, such as sympathetic activation, endothelial dysfunction, vascular oxidative stress, inflammation, increased coagulation, and metabolic dysregulation, link OSA to vascular disease [13]. 2. OSA and cardiovascular disease Evidence that relates OSA directly to vascular disease comes from small longitudinal studies of incident cardiovascular disease and studies assessing the outcomes of CPAP intervention. Nevertheless, many studies can only indirectly implicate OSA in the
486
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
etiology of cardiovascular disease, mainly because of the cost of establishing the diagnosis of OSA in large population samples, which means that most large-scale epidemiologic studies do not monitor OSA. Another important reason is that patients with OSA also have coexisting morbidities such as hypertension or obesity, making the independent risk of OSA on vascular disease more difficult to assess. 2.1. Hypertension About 50% of OSA patients have hypertension and, importantly, another 30% of hypertensive patients have undiagnosed OSA [14,15]. Cross-sectional studies show that there is a great association between OSA and hypertension. In a study of 2677 adults who were referred to a sleep clinic, the odds of hypertension increased by 1% for every increase in AHI unit with the prevalence levels for hypertension being 22.8% in control, 36.5% in mild, 46% in moderate, and 53.6% is severe OSA patients (after adjusting for age, BMI, and gender) [16]. In a more recent study, OSA (AHI: >15 events per hour) was the most common condition associated with resistant hypertension (64.0%), followed by primary aldosteronism (5.6%), renal artery stenosis (2.4%), renal parenchymal disease (1.6%), oral contraceptives (1.6%), and thyroid disorders (0.8%) in 125 patients with resistant hypertension [17]. The Wisconsin Sleep Cohort (WSCS) reports that patients with AHI of 15 or higher have a threefold increased risk of developing hypertension during this four-year study [18]. However, the Sleep Heart Health Study (SHHS) showed that associations were weak and not statistically significant (SHHS subjects with an AHI P30 events/h had an adjusted 1.5-fold increased risk of developing hypertension compared to subjects without OSA at baseline) [19]. Data from a series of cross-sectional studies strongly support an association between OSA and hypertension; however, the adjusted odds ratios for hypertension seem to vary considerably between studies and do not clearly show causality. Furthermore, not all longitudinal studies in adults support a causal relationship. Additional cohort studies in younger patients who are at risk of hypertension and OSA will provide some useful insights [20]. Studies in rats and mice also show that chronic intermittent hypoxia (IH) increases blood pressure [21,22]. OSA is now included as one of the main causes of hypertension in the sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure [23]. 2.2. Type 2 diabetes The overall prevalence of OSA in type 2 diabetic patients is nearly 71%, based on the average of data from five separate studies [24–28]. These data also suggest that approximately 19 million diabetic patients may have untreated or undiagnosed OSA. However, the prevalence of type 2 diabetes in OSA patients ranges between 15% and 30%, depending on the methods used to diagnose type 2 diabetes and OSA severity [29]. Untreated OSA can lead to poor glycemic control as Aronsohn et al. reported in a study of 60 diabetic patients who underwent full laboratory polysomnography. OSA severity was directly proportional to poorer glycemic control after controlling for age, sex, BMI, level of exercise, and total sleep time. Compared to non-OSA patients (adjusted mean hemoglobin A1c: 5.7%), the adjusted mean hemoglobin A1c was 7.3%, 7.7%, and 9.7% in mild, moderate, and severe OSA patients, respectively [24]. Botros et al. conducted an observational cohort study that examined 1233 patients for sleep-disordered breathing; 544 participants completed a full polysomnogram with no preexisting diabetes. The population was divided based on AHI and the main outcome was incident diabetes, as defined by fasting glucose levels of >126 mg/dL. They found that an increasing severity of OSA
was associated with an increased risk of diabetes. After adjusting for sex, BMI, age, and weight change, they reported an independent association between OSA and incident diabetes with a hazard ratio of 1.43 [30]. Overall, there is evidence from many clinical trials that untreated OSA can worsen glycemic control in type 2 diabetic patients [29]. There is still a controversy whether CPAP treatment in OSA improves glucose control. Current data suggest that obesity and the amount of CPAP can influence the metabolic response to CPAP. Large-scale randomized control trials assessing insulin sensitivity and glucose tolerance are required to determine the metabolic effects of CPAP. 2.3. Coronary artery disease OSA is also related to coronary artery disease (CAD) and stroke, because the prevalence of OSA among hospitalized men with acute myocardial infarction is nearly 70% [31]. IH, sympathetic vasoconstriction, and changes in intrathoracic pressure can all contribute to cardiac ischemia and atherosclerosis. In a prospective cohort study, 408 patients aged 70 years or younger with diagnosed coronary disease were followed up for a median period of 5.1 years. An AHI of P10 and an oxygen desaturation index (ODI) of P5 were used as the diagnostic criteria for sleep-disordered breathing. The primary end point was a composite of death, cerebrovascular events, and myocardial infarction. There was a 70% relative increase and a 10.7% absolute increase in the primary composite end point in patients with disordered breathing as defined by an ODI of P5. Similarly, patients with an AHI of P10 had a 62% relative increase and a 10.1% absolute increase in the composite end point [32] These data were confirmed by Shah et al. who reported that after adjusting for cardiovascular risk factors such as BMI and hypertension, OSA retained a significant association (hazard ratio 2.06) with the composite outcome of myocardial infarction, coronary artery revascularization procedures, and death after 2.9 years of follow-up [33]. A study of 200 patients without a history of CAD shows that the median coronary artery calcification score (Agatston units) was nine in OSA patients and 0 in non-OSA patients. This was measured by electron beam computed tomography on these patients within 3 years of polysomnography [34]. The median calcification score increased with the severity of OSA. A recent study of more than 500 subjects showed that OSA patients are more likely to have a family history of premature death from CAD than non-OSA patients. The results were independent of BMI, gender, and personal history of CAD [35]. A five-year follow-up of 62 patients with CAD reported a higher mortality rate in OSA patients (38%) compared to non-OSA patients (9%) [36]. Drager et al. evaluated the effects of 4 months of CPAP therapy on early markers of atherosclerosis, arterial stiffness, 24-h blood pressure (BP) monitoring, plasma C-reactive protein (CRP), and catecholamines in patients with severe OSA. Their study recruited only relatively young patients without significant comorbidities and who were unmedicated. A four-month CPAP therapy significantly improved validated markers of atherosclerosis, for example, reductions in inflammation markers of inflammation and sympathetic activity, in these patients [37]. In addition, OSA can affect outcomes after percutaneous coronary intervention (PCI) in patients with acute coronary syndrome (ACS). Yumino et al. followed ACS patients with OSA for 227 days and reported that the incidence of major cardiac events (cardiac death, reinfarction, and vessel revascularization) was significantly higher in ACS patients with OSA (23.5%) when compared to ACS patients without OSA (5.3%). In addition, binary restenosis was higher in patients with OSA when compared with those without (36.5% vs. 15.4%) [38]. Although the prevalence of OSA in patients with CAD is high, it is suggested that patients with OSA have less severe cardiac injury
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
during an acute nonfatal myocardial infarction when compared to patients without OSA. Data from Shah et al. suggest that a higher AHI was associated with lower peak troponin-T levels. This may suggest a cardioprotective role of sleep apnea during acute myocardial infarction via ischemic preconditioning [39]. It is likely that OSA worsens the prognosis of CAD patients, but more research is needed to determine whether diagnosing and treating OSA can influence the pathophysiological outcomes in patients with CAD. 2.4. Stroke In a cross-sectional study of over 6000 subjects from the SHHS, the odds ratio of prevalent stroke was higher (1.58) in patients with moderate sleep apnea (AHI P 11) [40]. Yaggi et al. studied 1022 patients (where 697 had OSA with AHI of five or higher) to determine the independent effect of OSA on the composite outcome of stroke or death from any cause. They found that OSA was associated with stroke or death with a hazard ratio of 2.24. After adjusting for age, sex, BMI, diabetes, hypertension, and dyslipidemia, OSA retained a significant association with stroke and death (hazard ratio 1.97). In a trend analysis, increased severity of OSA was associated with increased risk of the development of composite end point of death [41]. Redline et al. followed up a total of 5422 participants without a history of stroke at the baseline examination and untreated for OSA for a median of 8.7 years. Around 193 strokes were detected. A significant positive association between ischemic stroke and AHI was observed in men (p = 0.016), such that those in the highest AHI quartile (>19) had an adjusted hazard ratio of 2.86. In the mild to moderate range (AHI, 5–25), each one-unit increase in AHI in men was estimated to increase stroke risk by 6% [42]. Furthermore, Arzt et al. performed cross-sectional and longitudinal analyses on 1475 and 1189 subjects, respectively, from the general population. A protocol that included polysomnography, history, and risk factors of stroke was repeated at four-year intervals. In the cross-sectional analysis, subjects with AHI P20 had an increased odds ratio for stroke (odds ratio 4.33) after adjusting to confounding factors when compared to subjects with AHI < 5. In the prospective analysis, subjects with AHI P 20 were at an increased risk of suffering a first-ever stroke over the next 4 years (unadjusted odds ratio 4.31) when compared to those with AHI < 5. However, after adjustment for confounding factors, the odds ratio was still elevated, but no longer significant [43]. There are many mechanisms that link OSA to stroke. The cerebral responsiveness to CO2 (hypercapnic vasodilation) was studied by Morgan et al. in OSA patients. They measured cerebral flow velocity and end-tidal-CO2 in 373 participants of the Wisconsin Sleep Cohort and quantified cerebrovascular responsiveness to CO2 as the slope of the linear relationship between flow velocity and end-tidal-CO2 during rebreathing. The main independent variables were AHI and SaO2. There was a positive correlation between mean SaO2 levels and cerebrovascular CO2 responsiveness, where each 5% decrease in SaO2 during sleep predicted a decrease in cerebrovascular reactivity of 0.4 cm s1 mmHg1 PETCO2. These data suggest that hypoxemia in OSA can blunt hypercapnic vasodilation in the cerebral circulation [44]. However, the relationship between OSA and stroke remains circumstantial. The population of patients who are at risk of OSA is demographically similar to the patients who are at risk of stroke. The fact that only survivors of strokes are tested complicates the causal association between stroke and OSA [12]. 2.5. Heart failure It is estimated that 37% of 450 patients with systolic heart failure were diagnosed with sleep apnea after polysomnography. The prevalence was greater in men (38%) with obesity being the main
487
risk factor, while the prevalence in women was 31% with the main risk factor being old age [45]. Gottlieb et al. reported that OSA predicted incident heart failure in men but not in women (adjusted hazard ratio 1.13 per 10-unit increase in AHI). Men with AHI P 30 were 58% more likely to develop heart failure than those with AHI < 5 [46]. Although one study documented OSA in more than 50% of heart failure patients with preserved systolic function [47], the majority of recent studies show increased prevalence (30–62%) of OSA in patients with ejection fraction 645 [48–50]. Patients with OSA can develop heart failure through many mechanisms, including increases in blood pressure, left ventricular afterload, and greater risk of myocardial infarction [51]. It is clear that OSA is related to many cardiovascular diseases and its treatment is a necessity. 3. Obesity and OSA Overweight and obesity are considered the biggest health burden throughout the world, nearly affecting every aspect of life, and remain a great challenge to medical practice. It is estimated that by 2030, there will be 1.35 billion overweight adults (BMI > 25 kg/m2) and nearly 570 million obese (BMI > 30 kg/m2) in the world [52]. One of every three adults in the USA is obese [53]. Almost 20% of individuals in Western countries are obese and 1–2% are morbidly obese (BMI > 40 kg/m2). It appears that the prevalence of morbid obesity is increasing rapidly [53]. Furthermore, it has been predicted that life expectancy for obese adults may decrease [54]. In the USA, health care costs are already over 30% higher for obese subjects as compared to those with normal BMI. In Europe, obesity is estimated to be responsible for 2–8% of health costs and 10–13% of deaths [54]. Obesity is strongly associated with OSA, with nearly 60–90% of OSA patients being obese [55]. Thus, gaining weight worsens the severity of OSA, while losing weight improves it [56]. However, the association between obesity and OSA seems to be bi-directional, meaning that OSA can also lead to weight gain [57,58]. Localization of excess adipose tissue, for example, around the neck, affects pharyngeal neural and mechanical mechanisms that mediate airway collapsibility, thereby promoting OSA [59]. Obesity can also narrow the upper airway and reduce chest wall compliance [60] Obesity itself is considered a cardiovascular risk factor and similar to OSA, it is associated with male gender, cardiovascular morbidity, insulin resistance, hypertension, type 2 diabetes, and stroke [61]. Oxidative stress is thought to be a unifying mechanism leading to the development of comorbidities in obesity. Evidence suggests that the sources of oxidative stress in obesity come from a variety of sources including hyperleptinemia, hyperglycemia, decreased antioxidant capacity, increased reactive oxygen species (ROS) formation, increased tissue lipids levels, enzymatic sources within the endothelium, and chronic inflammation [62]. Furukawa et al. showed that fat accumulation in nondiabetic subjects closely correlates with markers of oxidative stress [63]. The Framingham Heart Study also demonstrates that BMI, diabetes, and smoking are independently and significantly associated with oxidative stress markers [64]. Unfortunately, OSA is not commonly considered in studies investigating oxidative stress in obese subjects and the contribution of OSA to oxidative stress cannot thus be excluded. In animal models of obesity, oxidative stress was mediated through the development of metabolic syndrome via dysregulated adipokine production. The mechanisms involved in increased oxidative stress include decreased expression of antioxidant enzymes in adipocytes and the upregulation of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase [63]. Adipocytes exposed to hypoxia show some dysregulation of some adipokines such as tumor necrosis factor (TNF) alpha and
488
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
leptin [65]. It seems that activation of nuclear factor kappa B (NF-jB) by hypoxia and/or by increased adipokines and free fatty acids released by excess adipose tissue is the final common inflammatory pathway linking obesity, OSA, and the metabolic syndrome both individually and, in many cases, synergistically [66]. Table 1 explores the biological effects of some adipokines and their levels in OSA patients. 4. Oxidative stress in OSA Oxidative stress results from an imbalance in oxidant production and antioxidant defense mechanisms; this means that either the overproduction of reactive oxygen and nitrogen species (ROS/ RNS) and/or a decreased antioxidant capacity leads to oxidative stress [74]. Free radicals play an important role in regulating cellular function and signal transduction. However, overproduction of free radicals can affect many cellular and physiological mechanisms by damaging lipids, proteins, and DNA. Recent studies show important links between the formation of hypoxia-related, free radical-related oxidative stress and cardiovascular disease in OSA patients. Pathways linking OSA/IH and oxidative stress and endothelial dysfunction are summarized in Fig. 1. 4.1. ROS and their sources in OSA ROS are atoms or molecules that possess one or more unpaired electrons in their outer orbit and thus are chemically unstable and highly reactive. When two radicals react, the product is a non-radical. But when radicals react with non-radicals, the product is a new radical and, therefore, the radical chain reaction propagates. ROS are by-products of oxygen metabolism, which are generated during normal cellular respiration and their elimination occurs through enzymatic and nonenzymatic antioxidant systems. When ROS generation exceeds the capacity of the antioxidant systems, oxidative stress and damage to tissues and cells ensue. This mechanism can eventually contribute to the pathological conditions of cardiovascular disease. One of the most abundant ROS is the superoxide anion; although it is a weak radical, it reacts to give rise to other potent oxidants such as hydroxyl radical and peroxynitrite. The latter is an RNS, which has a major role in causing endothelial dysfunction because it is the product of superoxide anion reacting with nitric oxide. As a result, nitric oxide bioavailability decreases, and the relaxing ability of the blood vessels is compromised. Free radical leakage from mitochondria during oxidative phosphorylation is the major source of superoxide anions. It is estimated that at least 3–5% of the oxygen that is consumed is
Fig. 1. Oxidative stress and inflammation in OSA. ROS formation induced by intermittent hypoxia activates an inflammatory cascade via activation of nuclear factor kappa-b, which influences the transcription of inflammatory cytokines and adhesion molecules. These in turn activate different blood cells and transcription factors. Activated platelets and leukocytes generate greater amounts of ROS, proinflammatory cytokines, and adhesion molecules, thus exacerbating the inflammatory and oxidative cycle leading eventually to endothelial dysfunction and promoting cardiovascular disease.
converted to superoxide anion during aerobic respiration. IH resembles ischemia/reperfusion injury, where there is an overproduction of free radicals during the period of hyperoxia. ROS production is elevated because of excessive mitochondrial reduction [74,75]. Phagocytes also generate large amounts of ROS when activated. Phagocytes are the immune system’s first line of defense against pathogens, and ROS production is one of their mechanisms for destroying invading pathogens. They generate superoxide anion through NADPH oxidase and other ROS (H2O2, HClO and NO) through other enzymes [76]. Although this mechanism of producing ROS can kill pathogens, it can also cause damage to surrounding tissues. NADPH oxidase is also found in nonphagocytic cells where it generates less superoxide anions for different purposes such as signaling [77]. For example, vascular cells express NADPH oxidase where superoxides play an essential role in vascular cell growth, migration, and alterations of extracellular matrix [78].
Table 1 Biology of some adipokines and their levels in OSA patients. Adipokine
Function Acts as a satiety signal with direct effects on the hypothalamus Stimulates lipolysis Inhibits lipogenesis Improves insulin sensitivity Increases glucose metabolism Stimulates fatty acid oxidation
Levels in OSA Increased [67,68] Unchanged [69,70]
Leptin
(1) (2) (3) (4) (5) (6)
Resistin
(1) Induces severe hepatic insulin resistance-increased rate of glucose production in rat (2) Shows functions controversial in humans
Unchanged [70,71]
Adiponectin
(1) (2) (3) (4) (5) (6) (7)
Decreased [72,73] Unchanged [68,70]
Increases fatty acid oxidation with reduction in plasma fatty acid levels Decreases plasma glucose levels Increases insulin sensitivity Anti-inflammatory Antioxidant Anti-atherogenic Anticancer properties through the inhibition TNF-a-mediated NF-jB pathway
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
489
4.2. Evidence of Oxidative stress in OSA
5. Inflammation in OSA
Many studies confirm that OSA is associated with oxidative stress that is caused by ROS generation and confirmed by measurements of oxidative stress markers. A study by Schulz et al. reports increased production of ROS in stimulated neutrophils and monocytes from OSA patients [79], while others report that ROS production was also significantly higher in non-stimulated monocytes of OSA patients [80,81]. Rats subjected to IH for 2 weeks have increased vascular production of ROS [82]. IH-induced pulmonary hypertension in mice is associated with increased lung levels of the NADPH oxidase subunits NOX4 and p22phox, indicating that NADPH oxidase-derived ROS contributes to the development of pulmonary hypertension caused by chronic IH [83]. NADPH oxidase is activated in many animal models of IH in tissues such as the myocardium, brain, carotid body, and liver [84,85]. Oxidative stress markers of lipid peroxidation, protein carbonylation, and DNA oxidation are increased in OSA patients. Lipid peroxidation is an important marker of oxidative stress because lipids are easily oxidized. Many studies show that lipid peroxidation increases in OSA patients. In an overnight study of OSA patients with and without cardiovascular disease, thiobarbituric acid (TBARS) had significantly increased [86]. In another study, 14 males with severe OSA fasted all night and TBARS levels were measured the next morning. TBARS levels in those patients were significantly higher (28.1 nmol MDA mg/LDL protein), when compared to 13 healthy age-matched controls (20.0 nmol MDA mg/LDL protein) [87]. Oxidized low-density lipoprotein (LDL) also increases in OSA, as shown in a prospective study of 37 subjects who have OSA and 38 who do not, where plasma levels of oxidized LDL were higher in OSA patients (43.6 U/L) compared to control (32.3 U/L) [88]. Protein carbonylation increases similarly in OSA patients, because as shown by a study of 17 moderate to severe OSA patients, protein carbonyl levels were significantly higher (1.11 lmol/g protein) when compared to matched control (0.99 lmol/g protein). On the other hand, the increase was not significant in mild OSA patients (1.03 lmol/g protein) [72]. 8-Hydroxyl-2deoxyguanosine (8-OHdG), a marker of DNA oxidation, is also elevated in OSA patients. Urinary excretion of 8-OHdG significantly correlates with the severity of OSA [89]. In animals, many chronic IH studies show significant increases in oxidative stress markers. A month of IH significantly increased malondialdehyde (MDA) levels in mice [90]. This is in accordance with another study where Polotsky et al. found that serum MDA levels increased fourfold in mice subjected to chronic IH for 6 months when compared to control [91]. Oxidative stress markers are also elevated in tissues such as the liver and brain [92,93]. Antioxidant capacity is impaired in OSA patients. Although the antioxidant capacity in OSA subjects and controls did not differ in their study, Christou et al. showed a linear negative relationship between antioxidant capacity and AHI (R = 0.551, p = 0.041) [94]. When Barcelo et al. tested total antioxidant status in OSA patients, they found that it had significantly decreased (1.4 mmol/L) when compared to that in healthy subjects (1.5 mmol/L, p = 0.0001). They also noticed lower levels of vitamin A (64 lg/dL) and vitamin E (1525 lg/dL) when compared to control (74 and 1774 lg/dL, respectively) [95], while Katsoulis et al. reported some unexpected results where they found that total antioxidant status before and after sleep was significantly lower in OSA patients with AHI < 30 (1.73 vs. 1.65 mmol/L, p = 0.01) but not in severe OSA patients with AHI > 30 (1.64 vs. 1.58 mmol/L, p = 0.07). A possible explanation could be the difference between acute effect of hypoxia immediately resulting from apneic sleep and any chronic state of oxidative stress that may be sustained in severe OSA patients even during daytime [96]. Combined animal and clinical studies indicate that OSA is a condition of increased oxidative stress.
Inflammation is a process of dynamic complex cytologic changes, mediator release, and cellular infiltration that occurs to the blood vessel and adjacent tissue in response to an injury or abnormal stimulation. Inflammation plays an important role in the pathogenesis of atherosclerosis, where levels of inflammatory circulatory markers are associated with cardiovascular risk. Inflammatory cascades involve nuclear transcription factors where the primary target is NF-jB, while the markers include adhesion molecules, CRP, TNF alpha, and others. Expression of adhesion molecules and inflammatory cytokines facilitates the recruitment of macrophages loaded with oxidized lipids known as foam cells. The accumulation of foam cells eventually leads to formation of atherosclerotic plaque, which is destined to rupture at one point, and leads to cardiovascular complications [97]. Lines of evidence from clinical studies show that inflammatory markers are elevated in OSA patients. IH in animals and cell cultures also demonstrates activation of inflammatory pathways [98]. Inflammatory pathway behind OSA/IH is illustrated in Fig. 1. ROS, hypoxia, cytokines, bacterial and viral products, and many other mediators can activate NF-jB. Once activated, NF-jB translocates to the nucleus and binds to promoters of specific genes initiating transcription of many products such as interleukin (IL-1), TNF-alpha, COX-2, adhesion molecules, and acute-phase proteins [99]. Nine patients with OSA showed significantly higher levels of p65 (a marker of NF-jB activation) when compared to seven healthy matched subjects (median, 0.037 ng/lL vs. 0.013 ng/lL) [100]. Htoo et al. showed that neutrophils from patients with mild to moderate and severe OSA had 4.8-fold and 7.9-fold increases in NF-jB-binding activity compared to control group [101]. Cytokines are molecules that participate in the process of inflammation, but are also essential to tissue repair, growth, and differentiation. Cytokines are synthesized and released by many cells and participate in the regulation of the immune system through interactions with other cytokines and transcription factors. They play a major role in progression of atherosclerosis because they control macrophage activation, smooth muscle cell proliferation, NO production, and activation of endothelial cells [102]. Many cytokines are regulated by free radicals through the activation of NF-jB, such as TNF alpha, IL-1, and IL-8 [103]. In OSA patients, it has been shown that the levels on TNF alpha in serum, monocytes, and T lymphocytes are higher when compared to control [104–107]. Moreover, IL-2 and IL-8 had increased in OSA patients [104,108–111]. These cytokines are all involved in the process of endothelial dysfunction and cardiovascular disease [110]. CRP is an acute-phase reactant secreted by the liver, adipose tissue, and other cell types, and it is considered to be one of the main markers of inflammation with a strong predictor value of CAD and cardiovascular events [112,113]. CRP potentially links OSA to inflammation, oxidative stress, and atherosclerosis. CRP has prooxidant effects when added to cultured coronary artery smooth muscle cells. Moreover, vascular smooth muscle cells and macrophages obtained from vulnerable plaque of coronary artery patients undergoing atherectomy co-express CRP protein and its mRNA with NADPH oxidase [114]. In addition to that, CRP induces adhesion molecule expression in cultured endothelial cells [115]. It is clear that CRP contributes to cardiovascular complications via inflammatory and oxidative stress processes. In OSA, plasma CRP levels in 22 patients were significantly higher (0.33 mg/dL) when compared to 20 control subjects (0.02 mg/dL), but importantly, CRP levels were independently associated with OSA severity (p = 0.032) [116]. A recent study reports that CRP levels had
490
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
significantly increased in OSA patients (4.03 mg/L) compared to the control group (2.41 mg/L) [117]. These data suggest that OSA is associated with elevated levels of CRP, a marker of cardiovascular risk. However, obesity is prevalent among OSA patients with some studies correlating increased CRP levels with BMI in OSA patients [118,119]. Guilleminault et al. monitored 239 patients and found that CRP remained normal in most patients. With CRP as the dependent variable, regression analysis with other factors such as BMI, sex, age, and lowest SaO2 suggested that a significant relationship existed only with BMI [118]. It is difficult to separate the effects of obesity and OSA on CRP levels, because their interactions can significantly increase in CRP levels. Circulating leukocytes and platelets are the main source of inflammatory cytokines and adhesion molecules. These molecules facilitate interactions between endothelial cells and activated blood cells and increase adhesion to the vascular wall. Selectins are one group thought to be responsible for weak interactions between leukocytes, platelets, and endothelial wall. Selectins are divided into L-selectin, E-selectin, and P-selectin. Another group of adhesion molecules is the integrins, which allows for a firmer binding to endothelial cells [120]. Under physiological conditions, these cells express low levels of intracellular inflammatory cytokines and adhesion molecules. But when activated by different stimuli including hypoxia/reoxygenation or OSA, expression levels increase dramatically, which leads to a pronounced inflammatory response, increases endothelial cell injury, and promotes atherosclerosis [120]. The expression of adhesion molecules in monocytes, polymorphonuclear cells, and T-lymphocytes increases in OSA patients [80,105,121]. Importantly, treatment of OSA patients with CPAP reduced the expression of P-selectin [122,123]. This inflammatory pathway also occurs in endothelial cells because they are also capable of expressing inflammatory cytokines and adhesion molecules [124]. Endothelial cells provide a vascular permeability barrier, regulate vascular tone, and maintain an antithrombotic and anti-inflammatory phenotype in their nonactivated state. Endothelial cells resist adhesion to platelets, red blood cells (RBCs), and adhesion molecules. However, in response to injury such as hypoxia/reoxygenation, increased expression of adhesion molecules is triggered and the interaction between the endothelium and circulating cells is mediated [125,126]. Many studies show that the expression of adhesion molecules is increased in OSA patients. Ursavas et al. measured the levels of intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1) in 39 OSA patients and found that both ICAM-1 (480.1 vs. 98.6 ng/mL) and VCAM-1 (1156.6 vs. 878.8 ng/ mL) had significantly increased in the OSA group compared to their matched controls [127]. There is also increased in vivo endothelial cell activation in OSA patients, which is a procoagulant and proinflammatory state where there is an increase in white blood cell interaction with endothelial cells, leading eventually to atherosclerosis [128].
6. Endothelial dysfunction in OSA Diminished endothelial function is an important consequence of OSA and is frequently measured as impaired endotheliumdependent vasodilatation [129]. Different studies show lower levels of circulating NO in OSA, for example, by the reduced levels of serum nitrite/nitrate (by-products of normal NO metabolism) in OSA subjects (38.9 lM vs. 63.1 lM in controls) [130]. This was confirmed in other studies where nitrate/nitrite levels were significantly lower in OSA patients (35.6 lM) when compared to control (72.6 lM) [131]. Many mechanisms have been suggested for endothelial dysfunction because of OSA or IH, including (1) interaction of NO and ROS forming peroxynitrite, (2) uncoupling
of endothelial nitric oxide synthase (eNOS), and (3) decreased endothelial expression of eNOS and increased levels of endogenous eNOS inhibitors [132]. Because of its short half-life and large volume of distribution, peroxynitrite is hard to measure and these factors explain the lack of difference in nitrotyrosine levels between OSA and healthy subjects [133,134]. However, Jelic et al. found an increased expression of nitrotyrosine in endothelial cells derived from OSA patients [128]. In all the forms of NOS, including the endothelial one, enzymatic activity requires five cofactor groups to incorporate oxygen into the amino acid L-arginine to produce NO. Those cofactors include flavin adenine dinucleotide (FAD), flavin mononucleotide (FMN), heme, tetrahydrobiopterin (BH4), and Ca2+-calmodulin. If NOS lacks L-arginine or any of these cofactors, it will produce superoxide anion instead of NO through an uncoupled state of NOS [135]. Antoniades et al. showed that increased ROS production during hypoxia could lead to BH4 oxidation and increased levels of arginase II that degrades L-arginine, leading to further eNOS uncoupling [136]. Patients with OSA have increased levels of asymmetrical dimethylarginine (ADMA), a competitive inhibitor of NOS [137]. Studies by Tanaka et al. suggest that eNOS activation is sensitive to regulation by redox status and that oxidative stress leads to decreased eNOS phosphorylation, thereby reducing its activity [138], while Jelic et al. supported the latter findings when they reported decreased ratios of total to phosphorylated eNOS in endothelial cells from patients with OSA [128]. 7. Animal models of sleep apnea OSA patients usually have comorbidities such as obesity, diabetes, or hypertension, which likely will affect cause–effect relationships. Creating animal models of OSA would minimize the influence of comorbidities and behavioral variables common in humans. Using animal models also permits the use of pharmacological agents to study the pathological mechanisms under a well-controlled environment. Ideally, animal models should mimic OSA in humans in at least three ways: (a) share aspects of the underlying pathophysiology, (b) have similar symptoms and the spectrum of disease severity that occur in humans, and (c) respond to treatment modalities that are useful in humans. Furthermore, a short life span (to allow for the unveiling of a wide range of disease-related complications within a reasonable time period), routine availability, cost effectiveness, and availability of diseasefree littermates add to the usefulness of animal models. There are additional considerations when using animals that need to be considered for sleep-related research. For instance, rodents are nocturnal animals that sleep in the prone position; there are no sufficient data to evaluate how these differences relate to sleep apnea studies undertaken in humans [1]. Animal models for studying sleep-disordered breathing should address at least one (or a combination) of the three main injurious consequences of sleep apnea: IH/hypercapnia, strained breathing because of mechanical obstruction, and sleep fragmentation. In this regard, rodents are amenable to genetic manipulation suitable for the production of phenotypes that may characterize OSA in humans. One advantage of using rodent models to examine neurophysiological aspects of sleep apnea in humans is the high degree of similarity between the structures of the nervous systems of rodents, such as rats and mice, and humans. 7.1. The natural model of sleep-disordered breathing A natural animal model of OSA is the English bulldog. There is a strong resemblance in sleep apnea between humans and bulldogs, making this animal model a suitable candidate for experimental
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
use. It was noticed that these dogs snore and have hypopneas and frequent arousals from sleep, mainly because of an abnormal upper airway anatomy, with an enlarged soft palate and a narrowing of the oropharynx. These animals have episodes of both central and obstructive apnea with hemoglobin desaturation (
Contents lists available at ScienceDirect
Sleep Medicine journal homepage: www.elsevier.com/locate/sleep
Review Article
Insights into obstructive sleep apnea research Mohammad Badran a, Najib Ayas b, Ismail Laher a,⇑ a
Department of Pharmacology and Therapeutics, Faculty of Medicine, University of British Columbia, Vancouver, BC V6T 1Z3, Canada Divisions of Critical Care and Respiratory Medicine, Department of Medicine, University of British Columbia, Sleep Disorders Program, UBC Hospital, Division of Critical Care Medicine, Providence Health Care, Vancouver, BC, Canada b
a r t i c l e
i n f o
Article history: Received 25 October 2013 Received in revised form 3 January 2014 Accepted 8 January 2014 Available online 10 February 2014 Keywords: Intermittent hypoxia Obesity Oxidative stress Inflammation Endothelium Animal models
a b s t r a c t Moderate to severe obstructive sleep apnea (OSA) occurs in 10–17% of middle aged men and 3–9% of middle-aged women with a higher prevalence among obese subjects. This condition is an independent risk factor for many cardiovascular diseases. Intermittent hypoxia is a major pathophysiologic character of OSA; it can lead to oxidative stress and inflammation, which in their turn cause endothelial dysfunction, a hallmark of atherosclerosis. Many animal models have been designed to mimic OSA in human patients to allow more in-depth investigation of biological and cellular mechanisms of this condition. This review discusses the cardiovascular outcomes of OSA and some of the animal models that are being used to investigate it. Ó 2014 Elsevier B.V. All rights reserved.
1. Obstructive sleep apnea Obstructive sleep apnea (OSA) is characterized by momentary cessations in breathing (apnea) or significant reductions in breathing amplitude (hypopnea) caused by an obstructed or collapsed upper airway; both conditions can cause significant arterial hypoxemia and hypercapnia. The apnea/hypopnea index (AHI) describes the total number of apnea/hypopnea episodes per hour of sleep, which is usually 30 categorize patients with sleep apnea as mild, moderate, and severe, respectively [1]. An obstructed airway increases resistance to airway flow that results in a greater breathing effort and swings in intrathoracic pressure, resulting in disruption of sleep, arousal, and reopening of the airway [2]. It is estimated that there are currently approximately 14% of men and 5% of women in the USA who have an AHI P 5, plus symptoms of daytime sleepiness, with 13% of men and 6% of women also having moderate to severe OSA (AHI P 15) [3]. Based on the average of prevalence estimates from many clinical studies, it is estimated that nearly one of every five adults has at least mild OSA and that one of every 15 has at least moderate OSA; moderate OSA occurs predominantly at body mass index (BMI) values of 25– 28 [4–6]. Despite numerous advancements in medicine, the majority of those affected with OSA remain undiagnosed [7]. OSA is sus⇑ Corresponding author. Tel.: +1 (604) 822 5882; fax: +1 (604) 224 5142. E-mail address: [email protected] (I. Laher). http://dx.doi.org/10.1016/j.sleep.2014.01.009 1389-9457/Ó 2014 Elsevier B.V. All rights reserved.
pected in people who are obese, hypertensive, hypersomnolent and habitual snorers [8]. Polysomnography is the main method for assessing patients with suspected sleep apnea [9]. Sleep stages are recorded along with oxyhemoglobin saturation, breathing, and airflow. In addition, limb and eye movements and the electrocardiogram are also monitored [10]. OSA creates a huge economic burden when compared to other chronic diseases. In 2000, OSA-related automobile collisions alone attributed to 1400 fatalities and a total cost of 15.9 billion dollars in the USA. Treatment with continuous positive airway pressure (CPAP) resulted in saving 7.9 billion dollars and 1000 lives [11]. It is well established that the outcomes of OSA can lead to serious vascular diseases. Data from different studies implicate OSA in the development of hypertension and, to some extent, cardiac ischemia, congestive heart failure, arrhythmias, cerebrovascular disease, and stroke [12]. Many intermediary mechanisms, such as sympathetic activation, endothelial dysfunction, vascular oxidative stress, inflammation, increased coagulation, and metabolic dysregulation, link OSA to vascular disease [13]. 2. OSA and cardiovascular disease Evidence that relates OSA directly to vascular disease comes from small longitudinal studies of incident cardiovascular disease and studies assessing the outcomes of CPAP intervention. Nevertheless, many studies can only indirectly implicate OSA in the
486
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
etiology of cardiovascular disease, mainly because of the cost of establishing the diagnosis of OSA in large population samples, which means that most large-scale epidemiologic studies do not monitor OSA. Another important reason is that patients with OSA also have coexisting morbidities such as hypertension or obesity, making the independent risk of OSA on vascular disease more difficult to assess. 2.1. Hypertension About 50% of OSA patients have hypertension and, importantly, another 30% of hypertensive patients have undiagnosed OSA [14,15]. Cross-sectional studies show that there is a great association between OSA and hypertension. In a study of 2677 adults who were referred to a sleep clinic, the odds of hypertension increased by 1% for every increase in AHI unit with the prevalence levels for hypertension being 22.8% in control, 36.5% in mild, 46% in moderate, and 53.6% is severe OSA patients (after adjusting for age, BMI, and gender) [16]. In a more recent study, OSA (AHI: >15 events per hour) was the most common condition associated with resistant hypertension (64.0%), followed by primary aldosteronism (5.6%), renal artery stenosis (2.4%), renal parenchymal disease (1.6%), oral contraceptives (1.6%), and thyroid disorders (0.8%) in 125 patients with resistant hypertension [17]. The Wisconsin Sleep Cohort (WSCS) reports that patients with AHI of 15 or higher have a threefold increased risk of developing hypertension during this four-year study [18]. However, the Sleep Heart Health Study (SHHS) showed that associations were weak and not statistically significant (SHHS subjects with an AHI P30 events/h had an adjusted 1.5-fold increased risk of developing hypertension compared to subjects without OSA at baseline) [19]. Data from a series of cross-sectional studies strongly support an association between OSA and hypertension; however, the adjusted odds ratios for hypertension seem to vary considerably between studies and do not clearly show causality. Furthermore, not all longitudinal studies in adults support a causal relationship. Additional cohort studies in younger patients who are at risk of hypertension and OSA will provide some useful insights [20]. Studies in rats and mice also show that chronic intermittent hypoxia (IH) increases blood pressure [21,22]. OSA is now included as one of the main causes of hypertension in the sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure [23]. 2.2. Type 2 diabetes The overall prevalence of OSA in type 2 diabetic patients is nearly 71%, based on the average of data from five separate studies [24–28]. These data also suggest that approximately 19 million diabetic patients may have untreated or undiagnosed OSA. However, the prevalence of type 2 diabetes in OSA patients ranges between 15% and 30%, depending on the methods used to diagnose type 2 diabetes and OSA severity [29]. Untreated OSA can lead to poor glycemic control as Aronsohn et al. reported in a study of 60 diabetic patients who underwent full laboratory polysomnography. OSA severity was directly proportional to poorer glycemic control after controlling for age, sex, BMI, level of exercise, and total sleep time. Compared to non-OSA patients (adjusted mean hemoglobin A1c: 5.7%), the adjusted mean hemoglobin A1c was 7.3%, 7.7%, and 9.7% in mild, moderate, and severe OSA patients, respectively [24]. Botros et al. conducted an observational cohort study that examined 1233 patients for sleep-disordered breathing; 544 participants completed a full polysomnogram with no preexisting diabetes. The population was divided based on AHI and the main outcome was incident diabetes, as defined by fasting glucose levels of >126 mg/dL. They found that an increasing severity of OSA
was associated with an increased risk of diabetes. After adjusting for sex, BMI, age, and weight change, they reported an independent association between OSA and incident diabetes with a hazard ratio of 1.43 [30]. Overall, there is evidence from many clinical trials that untreated OSA can worsen glycemic control in type 2 diabetic patients [29]. There is still a controversy whether CPAP treatment in OSA improves glucose control. Current data suggest that obesity and the amount of CPAP can influence the metabolic response to CPAP. Large-scale randomized control trials assessing insulin sensitivity and glucose tolerance are required to determine the metabolic effects of CPAP. 2.3. Coronary artery disease OSA is also related to coronary artery disease (CAD) and stroke, because the prevalence of OSA among hospitalized men with acute myocardial infarction is nearly 70% [31]. IH, sympathetic vasoconstriction, and changes in intrathoracic pressure can all contribute to cardiac ischemia and atherosclerosis. In a prospective cohort study, 408 patients aged 70 years or younger with diagnosed coronary disease were followed up for a median period of 5.1 years. An AHI of P10 and an oxygen desaturation index (ODI) of P5 were used as the diagnostic criteria for sleep-disordered breathing. The primary end point was a composite of death, cerebrovascular events, and myocardial infarction. There was a 70% relative increase and a 10.7% absolute increase in the primary composite end point in patients with disordered breathing as defined by an ODI of P5. Similarly, patients with an AHI of P10 had a 62% relative increase and a 10.1% absolute increase in the composite end point [32] These data were confirmed by Shah et al. who reported that after adjusting for cardiovascular risk factors such as BMI and hypertension, OSA retained a significant association (hazard ratio 2.06) with the composite outcome of myocardial infarction, coronary artery revascularization procedures, and death after 2.9 years of follow-up [33]. A study of 200 patients without a history of CAD shows that the median coronary artery calcification score (Agatston units) was nine in OSA patients and 0 in non-OSA patients. This was measured by electron beam computed tomography on these patients within 3 years of polysomnography [34]. The median calcification score increased with the severity of OSA. A recent study of more than 500 subjects showed that OSA patients are more likely to have a family history of premature death from CAD than non-OSA patients. The results were independent of BMI, gender, and personal history of CAD [35]. A five-year follow-up of 62 patients with CAD reported a higher mortality rate in OSA patients (38%) compared to non-OSA patients (9%) [36]. Drager et al. evaluated the effects of 4 months of CPAP therapy on early markers of atherosclerosis, arterial stiffness, 24-h blood pressure (BP) monitoring, plasma C-reactive protein (CRP), and catecholamines in patients with severe OSA. Their study recruited only relatively young patients without significant comorbidities and who were unmedicated. A four-month CPAP therapy significantly improved validated markers of atherosclerosis, for example, reductions in inflammation markers of inflammation and sympathetic activity, in these patients [37]. In addition, OSA can affect outcomes after percutaneous coronary intervention (PCI) in patients with acute coronary syndrome (ACS). Yumino et al. followed ACS patients with OSA for 227 days and reported that the incidence of major cardiac events (cardiac death, reinfarction, and vessel revascularization) was significantly higher in ACS patients with OSA (23.5%) when compared to ACS patients without OSA (5.3%). In addition, binary restenosis was higher in patients with OSA when compared with those without (36.5% vs. 15.4%) [38]. Although the prevalence of OSA in patients with CAD is high, it is suggested that patients with OSA have less severe cardiac injury
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
during an acute nonfatal myocardial infarction when compared to patients without OSA. Data from Shah et al. suggest that a higher AHI was associated with lower peak troponin-T levels. This may suggest a cardioprotective role of sleep apnea during acute myocardial infarction via ischemic preconditioning [39]. It is likely that OSA worsens the prognosis of CAD patients, but more research is needed to determine whether diagnosing and treating OSA can influence the pathophysiological outcomes in patients with CAD. 2.4. Stroke In a cross-sectional study of over 6000 subjects from the SHHS, the odds ratio of prevalent stroke was higher (1.58) in patients with moderate sleep apnea (AHI P 11) [40]. Yaggi et al. studied 1022 patients (where 697 had OSA with AHI of five or higher) to determine the independent effect of OSA on the composite outcome of stroke or death from any cause. They found that OSA was associated with stroke or death with a hazard ratio of 2.24. After adjusting for age, sex, BMI, diabetes, hypertension, and dyslipidemia, OSA retained a significant association with stroke and death (hazard ratio 1.97). In a trend analysis, increased severity of OSA was associated with increased risk of the development of composite end point of death [41]. Redline et al. followed up a total of 5422 participants without a history of stroke at the baseline examination and untreated for OSA for a median of 8.7 years. Around 193 strokes were detected. A significant positive association between ischemic stroke and AHI was observed in men (p = 0.016), such that those in the highest AHI quartile (>19) had an adjusted hazard ratio of 2.86. In the mild to moderate range (AHI, 5–25), each one-unit increase in AHI in men was estimated to increase stroke risk by 6% [42]. Furthermore, Arzt et al. performed cross-sectional and longitudinal analyses on 1475 and 1189 subjects, respectively, from the general population. A protocol that included polysomnography, history, and risk factors of stroke was repeated at four-year intervals. In the cross-sectional analysis, subjects with AHI P20 had an increased odds ratio for stroke (odds ratio 4.33) after adjusting to confounding factors when compared to subjects with AHI < 5. In the prospective analysis, subjects with AHI P 20 were at an increased risk of suffering a first-ever stroke over the next 4 years (unadjusted odds ratio 4.31) when compared to those with AHI < 5. However, after adjustment for confounding factors, the odds ratio was still elevated, but no longer significant [43]. There are many mechanisms that link OSA to stroke. The cerebral responsiveness to CO2 (hypercapnic vasodilation) was studied by Morgan et al. in OSA patients. They measured cerebral flow velocity and end-tidal-CO2 in 373 participants of the Wisconsin Sleep Cohort and quantified cerebrovascular responsiveness to CO2 as the slope of the linear relationship between flow velocity and end-tidal-CO2 during rebreathing. The main independent variables were AHI and SaO2. There was a positive correlation between mean SaO2 levels and cerebrovascular CO2 responsiveness, where each 5% decrease in SaO2 during sleep predicted a decrease in cerebrovascular reactivity of 0.4 cm s1 mmHg1 PETCO2. These data suggest that hypoxemia in OSA can blunt hypercapnic vasodilation in the cerebral circulation [44]. However, the relationship between OSA and stroke remains circumstantial. The population of patients who are at risk of OSA is demographically similar to the patients who are at risk of stroke. The fact that only survivors of strokes are tested complicates the causal association between stroke and OSA [12]. 2.5. Heart failure It is estimated that 37% of 450 patients with systolic heart failure were diagnosed with sleep apnea after polysomnography. The prevalence was greater in men (38%) with obesity being the main
487
risk factor, while the prevalence in women was 31% with the main risk factor being old age [45]. Gottlieb et al. reported that OSA predicted incident heart failure in men but not in women (adjusted hazard ratio 1.13 per 10-unit increase in AHI). Men with AHI P 30 were 58% more likely to develop heart failure than those with AHI < 5 [46]. Although one study documented OSA in more than 50% of heart failure patients with preserved systolic function [47], the majority of recent studies show increased prevalence (30–62%) of OSA in patients with ejection fraction 645 [48–50]. Patients with OSA can develop heart failure through many mechanisms, including increases in blood pressure, left ventricular afterload, and greater risk of myocardial infarction [51]. It is clear that OSA is related to many cardiovascular diseases and its treatment is a necessity. 3. Obesity and OSA Overweight and obesity are considered the biggest health burden throughout the world, nearly affecting every aspect of life, and remain a great challenge to medical practice. It is estimated that by 2030, there will be 1.35 billion overweight adults (BMI > 25 kg/m2) and nearly 570 million obese (BMI > 30 kg/m2) in the world [52]. One of every three adults in the USA is obese [53]. Almost 20% of individuals in Western countries are obese and 1–2% are morbidly obese (BMI > 40 kg/m2). It appears that the prevalence of morbid obesity is increasing rapidly [53]. Furthermore, it has been predicted that life expectancy for obese adults may decrease [54]. In the USA, health care costs are already over 30% higher for obese subjects as compared to those with normal BMI. In Europe, obesity is estimated to be responsible for 2–8% of health costs and 10–13% of deaths [54]. Obesity is strongly associated with OSA, with nearly 60–90% of OSA patients being obese [55]. Thus, gaining weight worsens the severity of OSA, while losing weight improves it [56]. However, the association between obesity and OSA seems to be bi-directional, meaning that OSA can also lead to weight gain [57,58]. Localization of excess adipose tissue, for example, around the neck, affects pharyngeal neural and mechanical mechanisms that mediate airway collapsibility, thereby promoting OSA [59]. Obesity can also narrow the upper airway and reduce chest wall compliance [60] Obesity itself is considered a cardiovascular risk factor and similar to OSA, it is associated with male gender, cardiovascular morbidity, insulin resistance, hypertension, type 2 diabetes, and stroke [61]. Oxidative stress is thought to be a unifying mechanism leading to the development of comorbidities in obesity. Evidence suggests that the sources of oxidative stress in obesity come from a variety of sources including hyperleptinemia, hyperglycemia, decreased antioxidant capacity, increased reactive oxygen species (ROS) formation, increased tissue lipids levels, enzymatic sources within the endothelium, and chronic inflammation [62]. Furukawa et al. showed that fat accumulation in nondiabetic subjects closely correlates with markers of oxidative stress [63]. The Framingham Heart Study also demonstrates that BMI, diabetes, and smoking are independently and significantly associated with oxidative stress markers [64]. Unfortunately, OSA is not commonly considered in studies investigating oxidative stress in obese subjects and the contribution of OSA to oxidative stress cannot thus be excluded. In animal models of obesity, oxidative stress was mediated through the development of metabolic syndrome via dysregulated adipokine production. The mechanisms involved in increased oxidative stress include decreased expression of antioxidant enzymes in adipocytes and the upregulation of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase [63]. Adipocytes exposed to hypoxia show some dysregulation of some adipokines such as tumor necrosis factor (TNF) alpha and
488
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
leptin [65]. It seems that activation of nuclear factor kappa B (NF-jB) by hypoxia and/or by increased adipokines and free fatty acids released by excess adipose tissue is the final common inflammatory pathway linking obesity, OSA, and the metabolic syndrome both individually and, in many cases, synergistically [66]. Table 1 explores the biological effects of some adipokines and their levels in OSA patients. 4. Oxidative stress in OSA Oxidative stress results from an imbalance in oxidant production and antioxidant defense mechanisms; this means that either the overproduction of reactive oxygen and nitrogen species (ROS/ RNS) and/or a decreased antioxidant capacity leads to oxidative stress [74]. Free radicals play an important role in regulating cellular function and signal transduction. However, overproduction of free radicals can affect many cellular and physiological mechanisms by damaging lipids, proteins, and DNA. Recent studies show important links between the formation of hypoxia-related, free radical-related oxidative stress and cardiovascular disease in OSA patients. Pathways linking OSA/IH and oxidative stress and endothelial dysfunction are summarized in Fig. 1. 4.1. ROS and their sources in OSA ROS are atoms or molecules that possess one or more unpaired electrons in their outer orbit and thus are chemically unstable and highly reactive. When two radicals react, the product is a non-radical. But when radicals react with non-radicals, the product is a new radical and, therefore, the radical chain reaction propagates. ROS are by-products of oxygen metabolism, which are generated during normal cellular respiration and their elimination occurs through enzymatic and nonenzymatic antioxidant systems. When ROS generation exceeds the capacity of the antioxidant systems, oxidative stress and damage to tissues and cells ensue. This mechanism can eventually contribute to the pathological conditions of cardiovascular disease. One of the most abundant ROS is the superoxide anion; although it is a weak radical, it reacts to give rise to other potent oxidants such as hydroxyl radical and peroxynitrite. The latter is an RNS, which has a major role in causing endothelial dysfunction because it is the product of superoxide anion reacting with nitric oxide. As a result, nitric oxide bioavailability decreases, and the relaxing ability of the blood vessels is compromised. Free radical leakage from mitochondria during oxidative phosphorylation is the major source of superoxide anions. It is estimated that at least 3–5% of the oxygen that is consumed is
Fig. 1. Oxidative stress and inflammation in OSA. ROS formation induced by intermittent hypoxia activates an inflammatory cascade via activation of nuclear factor kappa-b, which influences the transcription of inflammatory cytokines and adhesion molecules. These in turn activate different blood cells and transcription factors. Activated platelets and leukocytes generate greater amounts of ROS, proinflammatory cytokines, and adhesion molecules, thus exacerbating the inflammatory and oxidative cycle leading eventually to endothelial dysfunction and promoting cardiovascular disease.
converted to superoxide anion during aerobic respiration. IH resembles ischemia/reperfusion injury, where there is an overproduction of free radicals during the period of hyperoxia. ROS production is elevated because of excessive mitochondrial reduction [74,75]. Phagocytes also generate large amounts of ROS when activated. Phagocytes are the immune system’s first line of defense against pathogens, and ROS production is one of their mechanisms for destroying invading pathogens. They generate superoxide anion through NADPH oxidase and other ROS (H2O2, HClO and NO) through other enzymes [76]. Although this mechanism of producing ROS can kill pathogens, it can also cause damage to surrounding tissues. NADPH oxidase is also found in nonphagocytic cells where it generates less superoxide anions for different purposes such as signaling [77]. For example, vascular cells express NADPH oxidase where superoxides play an essential role in vascular cell growth, migration, and alterations of extracellular matrix [78].
Table 1 Biology of some adipokines and their levels in OSA patients. Adipokine
Function Acts as a satiety signal with direct effects on the hypothalamus Stimulates lipolysis Inhibits lipogenesis Improves insulin sensitivity Increases glucose metabolism Stimulates fatty acid oxidation
Levels in OSA Increased [67,68] Unchanged [69,70]
Leptin
(1) (2) (3) (4) (5) (6)
Resistin
(1) Induces severe hepatic insulin resistance-increased rate of glucose production in rat (2) Shows functions controversial in humans
Unchanged [70,71]
Adiponectin
(1) (2) (3) (4) (5) (6) (7)
Decreased [72,73] Unchanged [68,70]
Increases fatty acid oxidation with reduction in plasma fatty acid levels Decreases plasma glucose levels Increases insulin sensitivity Anti-inflammatory Antioxidant Anti-atherogenic Anticancer properties through the inhibition TNF-a-mediated NF-jB pathway
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
489
4.2. Evidence of Oxidative stress in OSA
5. Inflammation in OSA
Many studies confirm that OSA is associated with oxidative stress that is caused by ROS generation and confirmed by measurements of oxidative stress markers. A study by Schulz et al. reports increased production of ROS in stimulated neutrophils and monocytes from OSA patients [79], while others report that ROS production was also significantly higher in non-stimulated monocytes of OSA patients [80,81]. Rats subjected to IH for 2 weeks have increased vascular production of ROS [82]. IH-induced pulmonary hypertension in mice is associated with increased lung levels of the NADPH oxidase subunits NOX4 and p22phox, indicating that NADPH oxidase-derived ROS contributes to the development of pulmonary hypertension caused by chronic IH [83]. NADPH oxidase is activated in many animal models of IH in tissues such as the myocardium, brain, carotid body, and liver [84,85]. Oxidative stress markers of lipid peroxidation, protein carbonylation, and DNA oxidation are increased in OSA patients. Lipid peroxidation is an important marker of oxidative stress because lipids are easily oxidized. Many studies show that lipid peroxidation increases in OSA patients. In an overnight study of OSA patients with and without cardiovascular disease, thiobarbituric acid (TBARS) had significantly increased [86]. In another study, 14 males with severe OSA fasted all night and TBARS levels were measured the next morning. TBARS levels in those patients were significantly higher (28.1 nmol MDA mg/LDL protein), when compared to 13 healthy age-matched controls (20.0 nmol MDA mg/LDL protein) [87]. Oxidized low-density lipoprotein (LDL) also increases in OSA, as shown in a prospective study of 37 subjects who have OSA and 38 who do not, where plasma levels of oxidized LDL were higher in OSA patients (43.6 U/L) compared to control (32.3 U/L) [88]. Protein carbonylation increases similarly in OSA patients, because as shown by a study of 17 moderate to severe OSA patients, protein carbonyl levels were significantly higher (1.11 lmol/g protein) when compared to matched control (0.99 lmol/g protein). On the other hand, the increase was not significant in mild OSA patients (1.03 lmol/g protein) [72]. 8-Hydroxyl-2deoxyguanosine (8-OHdG), a marker of DNA oxidation, is also elevated in OSA patients. Urinary excretion of 8-OHdG significantly correlates with the severity of OSA [89]. In animals, many chronic IH studies show significant increases in oxidative stress markers. A month of IH significantly increased malondialdehyde (MDA) levels in mice [90]. This is in accordance with another study where Polotsky et al. found that serum MDA levels increased fourfold in mice subjected to chronic IH for 6 months when compared to control [91]. Oxidative stress markers are also elevated in tissues such as the liver and brain [92,93]. Antioxidant capacity is impaired in OSA patients. Although the antioxidant capacity in OSA subjects and controls did not differ in their study, Christou et al. showed a linear negative relationship between antioxidant capacity and AHI (R = 0.551, p = 0.041) [94]. When Barcelo et al. tested total antioxidant status in OSA patients, they found that it had significantly decreased (1.4 mmol/L) when compared to that in healthy subjects (1.5 mmol/L, p = 0.0001). They also noticed lower levels of vitamin A (64 lg/dL) and vitamin E (1525 lg/dL) when compared to control (74 and 1774 lg/dL, respectively) [95], while Katsoulis et al. reported some unexpected results where they found that total antioxidant status before and after sleep was significantly lower in OSA patients with AHI < 30 (1.73 vs. 1.65 mmol/L, p = 0.01) but not in severe OSA patients with AHI > 30 (1.64 vs. 1.58 mmol/L, p = 0.07). A possible explanation could be the difference between acute effect of hypoxia immediately resulting from apneic sleep and any chronic state of oxidative stress that may be sustained in severe OSA patients even during daytime [96]. Combined animal and clinical studies indicate that OSA is a condition of increased oxidative stress.
Inflammation is a process of dynamic complex cytologic changes, mediator release, and cellular infiltration that occurs to the blood vessel and adjacent tissue in response to an injury or abnormal stimulation. Inflammation plays an important role in the pathogenesis of atherosclerosis, where levels of inflammatory circulatory markers are associated with cardiovascular risk. Inflammatory cascades involve nuclear transcription factors where the primary target is NF-jB, while the markers include adhesion molecules, CRP, TNF alpha, and others. Expression of adhesion molecules and inflammatory cytokines facilitates the recruitment of macrophages loaded with oxidized lipids known as foam cells. The accumulation of foam cells eventually leads to formation of atherosclerotic plaque, which is destined to rupture at one point, and leads to cardiovascular complications [97]. Lines of evidence from clinical studies show that inflammatory markers are elevated in OSA patients. IH in animals and cell cultures also demonstrates activation of inflammatory pathways [98]. Inflammatory pathway behind OSA/IH is illustrated in Fig. 1. ROS, hypoxia, cytokines, bacterial and viral products, and many other mediators can activate NF-jB. Once activated, NF-jB translocates to the nucleus and binds to promoters of specific genes initiating transcription of many products such as interleukin (IL-1), TNF-alpha, COX-2, adhesion molecules, and acute-phase proteins [99]. Nine patients with OSA showed significantly higher levels of p65 (a marker of NF-jB activation) when compared to seven healthy matched subjects (median, 0.037 ng/lL vs. 0.013 ng/lL) [100]. Htoo et al. showed that neutrophils from patients with mild to moderate and severe OSA had 4.8-fold and 7.9-fold increases in NF-jB-binding activity compared to control group [101]. Cytokines are molecules that participate in the process of inflammation, but are also essential to tissue repair, growth, and differentiation. Cytokines are synthesized and released by many cells and participate in the regulation of the immune system through interactions with other cytokines and transcription factors. They play a major role in progression of atherosclerosis because they control macrophage activation, smooth muscle cell proliferation, NO production, and activation of endothelial cells [102]. Many cytokines are regulated by free radicals through the activation of NF-jB, such as TNF alpha, IL-1, and IL-8 [103]. In OSA patients, it has been shown that the levels on TNF alpha in serum, monocytes, and T lymphocytes are higher when compared to control [104–107]. Moreover, IL-2 and IL-8 had increased in OSA patients [104,108–111]. These cytokines are all involved in the process of endothelial dysfunction and cardiovascular disease [110]. CRP is an acute-phase reactant secreted by the liver, adipose tissue, and other cell types, and it is considered to be one of the main markers of inflammation with a strong predictor value of CAD and cardiovascular events [112,113]. CRP potentially links OSA to inflammation, oxidative stress, and atherosclerosis. CRP has prooxidant effects when added to cultured coronary artery smooth muscle cells. Moreover, vascular smooth muscle cells and macrophages obtained from vulnerable plaque of coronary artery patients undergoing atherectomy co-express CRP protein and its mRNA with NADPH oxidase [114]. In addition to that, CRP induces adhesion molecule expression in cultured endothelial cells [115]. It is clear that CRP contributes to cardiovascular complications via inflammatory and oxidative stress processes. In OSA, plasma CRP levels in 22 patients were significantly higher (0.33 mg/dL) when compared to 20 control subjects (0.02 mg/dL), but importantly, CRP levels were independently associated with OSA severity (p = 0.032) [116]. A recent study reports that CRP levels had
490
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
significantly increased in OSA patients (4.03 mg/L) compared to the control group (2.41 mg/L) [117]. These data suggest that OSA is associated with elevated levels of CRP, a marker of cardiovascular risk. However, obesity is prevalent among OSA patients with some studies correlating increased CRP levels with BMI in OSA patients [118,119]. Guilleminault et al. monitored 239 patients and found that CRP remained normal in most patients. With CRP as the dependent variable, regression analysis with other factors such as BMI, sex, age, and lowest SaO2 suggested that a significant relationship existed only with BMI [118]. It is difficult to separate the effects of obesity and OSA on CRP levels, because their interactions can significantly increase in CRP levels. Circulating leukocytes and platelets are the main source of inflammatory cytokines and adhesion molecules. These molecules facilitate interactions between endothelial cells and activated blood cells and increase adhesion to the vascular wall. Selectins are one group thought to be responsible for weak interactions between leukocytes, platelets, and endothelial wall. Selectins are divided into L-selectin, E-selectin, and P-selectin. Another group of adhesion molecules is the integrins, which allows for a firmer binding to endothelial cells [120]. Under physiological conditions, these cells express low levels of intracellular inflammatory cytokines and adhesion molecules. But when activated by different stimuli including hypoxia/reoxygenation or OSA, expression levels increase dramatically, which leads to a pronounced inflammatory response, increases endothelial cell injury, and promotes atherosclerosis [120]. The expression of adhesion molecules in monocytes, polymorphonuclear cells, and T-lymphocytes increases in OSA patients [80,105,121]. Importantly, treatment of OSA patients with CPAP reduced the expression of P-selectin [122,123]. This inflammatory pathway also occurs in endothelial cells because they are also capable of expressing inflammatory cytokines and adhesion molecules [124]. Endothelial cells provide a vascular permeability barrier, regulate vascular tone, and maintain an antithrombotic and anti-inflammatory phenotype in their nonactivated state. Endothelial cells resist adhesion to platelets, red blood cells (RBCs), and adhesion molecules. However, in response to injury such as hypoxia/reoxygenation, increased expression of adhesion molecules is triggered and the interaction between the endothelium and circulating cells is mediated [125,126]. Many studies show that the expression of adhesion molecules is increased in OSA patients. Ursavas et al. measured the levels of intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1) in 39 OSA patients and found that both ICAM-1 (480.1 vs. 98.6 ng/mL) and VCAM-1 (1156.6 vs. 878.8 ng/ mL) had significantly increased in the OSA group compared to their matched controls [127]. There is also increased in vivo endothelial cell activation in OSA patients, which is a procoagulant and proinflammatory state where there is an increase in white blood cell interaction with endothelial cells, leading eventually to atherosclerosis [128].
6. Endothelial dysfunction in OSA Diminished endothelial function is an important consequence of OSA and is frequently measured as impaired endotheliumdependent vasodilatation [129]. Different studies show lower levels of circulating NO in OSA, for example, by the reduced levels of serum nitrite/nitrate (by-products of normal NO metabolism) in OSA subjects (38.9 lM vs. 63.1 lM in controls) [130]. This was confirmed in other studies where nitrate/nitrite levels were significantly lower in OSA patients (35.6 lM) when compared to control (72.6 lM) [131]. Many mechanisms have been suggested for endothelial dysfunction because of OSA or IH, including (1) interaction of NO and ROS forming peroxynitrite, (2) uncoupling
of endothelial nitric oxide synthase (eNOS), and (3) decreased endothelial expression of eNOS and increased levels of endogenous eNOS inhibitors [132]. Because of its short half-life and large volume of distribution, peroxynitrite is hard to measure and these factors explain the lack of difference in nitrotyrosine levels between OSA and healthy subjects [133,134]. However, Jelic et al. found an increased expression of nitrotyrosine in endothelial cells derived from OSA patients [128]. In all the forms of NOS, including the endothelial one, enzymatic activity requires five cofactor groups to incorporate oxygen into the amino acid L-arginine to produce NO. Those cofactors include flavin adenine dinucleotide (FAD), flavin mononucleotide (FMN), heme, tetrahydrobiopterin (BH4), and Ca2+-calmodulin. If NOS lacks L-arginine or any of these cofactors, it will produce superoxide anion instead of NO through an uncoupled state of NOS [135]. Antoniades et al. showed that increased ROS production during hypoxia could lead to BH4 oxidation and increased levels of arginase II that degrades L-arginine, leading to further eNOS uncoupling [136]. Patients with OSA have increased levels of asymmetrical dimethylarginine (ADMA), a competitive inhibitor of NOS [137]. Studies by Tanaka et al. suggest that eNOS activation is sensitive to regulation by redox status and that oxidative stress leads to decreased eNOS phosphorylation, thereby reducing its activity [138], while Jelic et al. supported the latter findings when they reported decreased ratios of total to phosphorylated eNOS in endothelial cells from patients with OSA [128]. 7. Animal models of sleep apnea OSA patients usually have comorbidities such as obesity, diabetes, or hypertension, which likely will affect cause–effect relationships. Creating animal models of OSA would minimize the influence of comorbidities and behavioral variables common in humans. Using animal models also permits the use of pharmacological agents to study the pathological mechanisms under a well-controlled environment. Ideally, animal models should mimic OSA in humans in at least three ways: (a) share aspects of the underlying pathophysiology, (b) have similar symptoms and the spectrum of disease severity that occur in humans, and (c) respond to treatment modalities that are useful in humans. Furthermore, a short life span (to allow for the unveiling of a wide range of disease-related complications within a reasonable time period), routine availability, cost effectiveness, and availability of diseasefree littermates add to the usefulness of animal models. There are additional considerations when using animals that need to be considered for sleep-related research. For instance, rodents are nocturnal animals that sleep in the prone position; there are no sufficient data to evaluate how these differences relate to sleep apnea studies undertaken in humans [1]. Animal models for studying sleep-disordered breathing should address at least one (or a combination) of the three main injurious consequences of sleep apnea: IH/hypercapnia, strained breathing because of mechanical obstruction, and sleep fragmentation. In this regard, rodents are amenable to genetic manipulation suitable for the production of phenotypes that may characterize OSA in humans. One advantage of using rodent models to examine neurophysiological aspects of sleep apnea in humans is the high degree of similarity between the structures of the nervous systems of rodents, such as rats and mice, and humans. 7.1. The natural model of sleep-disordered breathing A natural animal model of OSA is the English bulldog. There is a strong resemblance in sleep apnea between humans and bulldogs, making this animal model a suitable candidate for experimental
M. Badran et al. / Sleep Medicine 15 (2014) 485–495
use. It was noticed that these dogs snore and have hypopneas and frequent arousals from sleep, mainly because of an abnormal upper airway anatomy, with an enlarged soft palate and a narrowing of the oropharynx. These animals have episodes of both central and obstructive apnea with hemoglobin desaturation (
