Clinical Endocrinology (2015) 82, 165–177
doi: 10.1111/cen.12597
REQUESTED REVIEW
Obstructive sleep apnoea, insulin resistance and adipocytokines David C.L. Lam, Karen S.L. Lam and Mary S.M. Ip Department of Medicine, University of Hong Kong, Hong Kong SAR, China
Summary Obstructive sleep apnoea (OSA) is associated with multiple cardiometabolic abnormalities. Obesity is considered a major risk factor for the development of OSA, and it is also an established risk factor for insulin resistance and other cardiometabolic disorders. The enigma remains whether OSA has any causal role in the adverse metabolic profile, independent of or beyond that due to obesity. Sleep apnoeas and hypopnoeas result directly in intermittent hypoxaemia and cerebral arousals, both of which may evoke a cascade of downstream biologic responses in various body tissues and cells. Adipose tissue is a major source of adipocytokines many of which play important roles in the regulation of various metabolic functions. It is hypothesized that OSA may, through its unique pathophysiology, affect metabolic function through modulation of production or action of adipocytokines. This review focuses on insulin resistance, glucose metabolism and relevant adipocytokines in the context of OSA. (Received 30 April 2014; returned for revision 24 May 2014; finally revised 18 August 2014; accepted 18 August 2014)
Introduction Obstructive sleep apnoea (OSA) refers to the sleep-related breathing disorder wherein recurrent events of complete or partial collapse of the upper airway at pharyngeal level occur during sleep, hindering airflow into the lungs, resulting in apnoeas, hypopnoeas or increased respiratory efforts.1,2 Sleep studies [polysomnograms, (PSG)] document a range of sleep parameters, including the number of respiratory events that occur per hour of sleep, widely termed as Apnoea hypopnea index (AHI), which forms the basis of diagnosis and severity grading.1,2 Symptomatic OSA (OSA syndrome), defined with the cardinal feature of excessive daytime sleepiness, has been reported to affect about 2–4% of middle-aged adults in different communities, while as many as 9% of women and 24% of men in a
Correspondence: Professor Mary S.M. Ip, Division of Respiratory and Critical Care Medicine, Department of Medicine, 4/F, Professorial Block, Queen Mary Hospital, Pokfulam, Hong Kong SAR, China. Tel.: (852) 2255 4455; Fax: (852) 2816 2863; E-mail: msmip@hkucc. hku.hk © 2014 John Wiley & Sons Ltd
random general population in USA in the 1990s demonstrated an AHI of ≥5 events/hour which is considered as above adult norm.1–3 Given the global escalation of obesity which is the strongest risk factor for OSA irrespective of ethnicity, current prevalence of OSA is estimated to be much higher.3 Treatment strategies include the use of continuous positive airway pressure (CPAP) devices, oral appliances, surgery and other measures including weight reduction in the overweight or obese subject.1 Majority of patients with OSA are overweight or obese, while OSA of a range of severity is found in 40–90% of obese subjects.4,5 OSA is highly associated with various cardiometabolic diseases, and studies showed a five- to sevenfold risk of increase of metabolic syndrome in OSA.4–6 Insulin resistance is strongly though not exclusively associated with obesity, in particular visceral obesity. It is a core component of the metabolic syndrome and a key mechanistic pathway in the pathogenesis of cardiometabolic derangements, especially of type 2 diabetes mellitus (DM). Apart from being a precursor of glucose intolerance and type 2 DM, insulin resistance is also closely linked to other metabolic dysfunction such as hypertension and dyslipidaemia,7 but there is scanty literature regarding the relationships of insulin resistance and adiopocytokines with these metabolic parameters in the context of OSA. Although OSA and many cardiometabolic aberrations share the common risk factor of obesity, the pathophysiologic effects in OSA can potentially influence metabolic homeostasis, and it is postulated that OSA itself may confer cardiometabolic risks ‘independent’ of obesity. Other then direct influence on metabolic pathways without involving adipose tissues, OSA may interact with obesity to propagate or amplify adverse metabolic effects beyond that contributed by adiposity in the absence of OSA, and which may run in a vicious cycle.4–6 Adipocytokines, derived predominantly from adipose tissues, are important mediators of metabolic functions.8,9 Hence, modulation of the expression/secretion or function of adipocytokines by OSA is one potential mechanistic pathway for metabolic dysfunction in OSA. The purpose of this review is to provide a current overview of the relationships between OSA, insulin resistance, glucose metabolism and relevant adipocytokines.
Pathophysiology of OSA The mechanisms of upper airway collapse are multiple and not completely defined. Factors that have been identified to deter165
166 D. C. L. Lam et al. mine the propensity for upper airway collapse during sleep include craniofacial morphology, amount and distribution of adipose tissues in the body, upper airway dilator muscle function, arousal threshold and ventilatory control.2,5 Obstructive breathing events result in transient hypoxaemia which quickly reverts with resumption of unobstructed breathing – a phenomenon termed intermittent hypoxaemia/hypoxia, as well as cerebral arousals and changes in sleep architecture. A plethora of potential systemic pathophysiologic effects may occur, including hemodynamic shear stress, sympathetic nerve activation, oxidative stress, inflammation and hormonal changes of the hypothalamic–pituitary–adrenal axis such as cortisol.2,4– 6,10–12 These pathways are also known to underpin many common cardiometabolic pathologies.4–6 Furthermore, short sleep duration has been associated with obesity and adverse glucose metabolism, while experimental sleep restriction demonstrated adverse insulin–glucose metabolism, lending relative sleep loss in OSA as an additional mediating pathway for metabolic dysfunction.13 Thus, it has been hypothesized that OSA, per se or interacting with obesity and/or other concurrent risk factors, plays a causal role in cardiometabolic derangements (Fig. 1).4–6,10
Obesity and OSA Obesity is the commonest risk factor for development of OSA, although body mass index (BMI) accounts for only a small part of the variability of AHI, implying the multifactorial nature of pathogenesis of OSA.2 A 10% gain in body weight increased the chance of developing moderate–severe OSA by sixfold, while every 1% increase in body weight was associated with a 3% increase in AHI, and weight loss helped to reduce OSA severity but to a less substantial effect.3
OSA Intermittent hypoxia
Sleep fragmentation/Loss
Sympathetic activation Hormonal changes (HPA axis, satiety hormones)
Adipose tissue Hypoxia Oxidative stress Inflammation
Altered adipocytokine release Metabolic dysfunction Insulin resistance, glucose dysmetabolism dyslipidaemia, hypertension
Fig. 1 Putative link of OSA, adipocytokines and metabolic dysfunction.
Both the degree of adiposity and its distribution determine the risk of having OSA. Increased adipose and soft tissues around the upper airway narrow the pharyngeal calibre and increase the likelihood of airway collapse during sleep. OSA and its severity have been reported to be associated with central obesity and abdominal visceral fat, more so than with BMI, particularly in men.5 It has further been proposed that obesity and OSA may be mutually conducive. Repetitive apnoeas and arousals during sleep lead to excessive daytime sleepiness which may reduce motivation to physical activity and predispose to weight gain over time. Presence of sleep apnoea in men with central obesity attenuated the metabolic improvement in response to a lifestyle intervention programme, compared to those without OSA.14 Studies on the impact of CPAP treatment on body weight or abdominal fat have yielded conflicting data,15–18 with two randomized, sham-controlled trials showing no change.17,18 It has been suggested that OSA subjects may be inherently predisposed to weight gain and may experience difficulty in losing weight compared to non-OSA subjects, relating to selective differences in body metabolism regulated by hormones such as insulin, leptin or ghrelin.19 Lifestyle modifications or other interventions resulting in weight loss clearly benefit both metabolic function and OSA in obese subjects.4
OSA, insulin resistance and glucose metabolism The relationship between OSA and a range of glucose dysmetabolism from insulin resistance to overt diabetes mellitus has been investigated, and readers are referred to several recent comprehensive reviews.4,20,21 Epidemiologic studies have consistently shown up an association of OSA and impaired glucose metabolism after adjustments of known confounders.22,23 The Wisconsin Sleep Cohort (mean age 50 years) showed an independent association between severity of OSA and prevalence of DM at baseline, but no increase in incident diabetes at 4-year follow-up,22 while other cohorts found associations between baseline OSA and incident diabetes over 3–10 years.21,23,24 Several cross-sectional studies of community populations consistently found that OSA was associated with increased insulin resistance or impaired glycaemic status, despite adjustment for obesity and other confounders.25–28 Analyses of the European Sleep Apnea Database showed that OSA or its severity was associated with worse glycaemic status (HbA1c levels) in both subjects with and without DM.29,30 Majority of clinical studies in non-diabetic subjects have reported independent associations between OSA and insulin resistance/sensitivity, with dose-dependent effect of OSA on metabolic impairment, although some studies found that the association was abolished after adjusting for BMI and/or other measures of adiposity.20,21 A small proportion of OSA subject are not obese, and in a clinic-based study of Chinese subjects without DM, homeostatic model assessment index on insulin resistance (HOMA-IR) was higher in both obese and non-obese subjects with OSA.31 Clinical studies focusing on non-obese OSA also showed that they had higher insulin resistance or glucose dysmetabolism compared with lean counterparts without OSA.32,33 © 2014 John Wiley & Sons Ltd Clinical Endocrinology (2015), 82, 165–177
Adipocytokines and insulin resistance in OSA 167 In a study of 118 subjects without DM, the frequent sample intravenous glucose tolerance test showed ‘OSA dose-dependent’ impairment of insulin sensitivity and also of pancreatic beta-cell function.34 Despite reduced insulin sensitivity, patients with OSA did not increase pancreatic beta-cell insulin output. To the contrary, in a study of healthy lean young men free of cardiometabolic disease, the presence of mild-to-moderate OSA (mean AHI 112 29 per sleep hour) was associated with insulin resistance and an increase in insulin secretion suggestive of a compensatory mechanism.33 Another study of 45 severely obese adults also found that OSA was associated with increased betacell function, indicated by homeostasis model assessment calculations, in those with normal glucose metabolism.35 It is plausible that homeostatic mechanisms of enhanced insulin secretion may be lost with chronic insult, especially with ageing or prediabetic state. It has been proposed that sleepiness may be a phenotypic marker for insulin resistance in OSA. In non-diabetic subjects with similar BMI and AHI, those with excessive daytime sleepiness had higher HOMA-IR than those without sleepiness, and insulin resistance was improved with CPAP treatment only in the group with baseline excessive sleepiness.36 Associated abdominal or visceral obesity in OSA could contribute to sleepiness in OSA via hypercytokinaemia.5 Despite the rather abundant positive data supporting an independent association of OSA and insulin resistance (or impaired insulin sensitivity), cross-sectional studies cannot be definitive for a causal link. Reported data on the effect of treatment of OSA on insulin–glucose metabolism remain highly controversial. Insulin sensitivity, measured by the hyperinsulinaemic euglycaemic clamp, was acutely improved in 40 non-diabetic subjects with OSA treated with CPAP for 2 days, and the improvement was maintained at 3 months (n = 31). The improvement in insulin sensitivity was driven mainly by the subset of lean subjects with BMI < 30 kg/m2.37 Other observational studies of small sample sizes, in either subjects with or without DM, also suggested improvements in insulin sensitivity/resistance with CPAP treatment, but results are by no means consistent and a number of studies did not find any change.20,21 Most of the randomized controlled trials (RCTs) regarding CPAP treatment of OSA did not provide definitive evidence for improvement in insulin sensitivity/resistance or other measures of glucose metabolism. A RCT involving middle-aged non-diabetic Chinese men showed improvement in insulin sensitivity measured with the short insulin tolerance test, but not in HOMA-IR, after 1 week of CPAP treatment (n = 30) compared to the control group on observation.38 However, contrary to the previous Caucasian study,37 improvement in insulin sensitivity was driven by the subgroup of subjects with BMI >25 kg/m2 (mean BMI: 28 kg/m2), who were obese by Asian criteria. In a cross-over controlled trial of 8-week CPAP duration, no improvement in insulin sensitivity index derived from oral glucose tolerance test was found, but subset analysis of those with severe OSA (AHI ≥ 30) showed significant improvement in insulin sensitivity with CPAP treatment compared to sham CPAP.39 In another RCT comprising of 65 non-diabetic men, © 2014 John Wiley & Sons Ltd Clinical Endocrinology (2015), 82, 165–177
no difference in insulin sensitivity index was observed between the therapeutic and sham CPAP group at 12 weeks. However, at the end of 24 weeks, when the whole study group was receiving CPAP treatment, a significant improvement in insulin sensitivity compared with baseline was seen.40 In a 24-week RCT in obese adults with moderate-to-severe OSA, CPAP alone did not alter insulin resistance, while CPAP combined with weight loss provided incremental reduction in insulin resistance and serum triglyceride compared to that achieved by weight loss alone.41 In diabetic subjects, data from observational studies tended to be favourable with improvement in insulin sensitivity/resistance, while this has not been supported by data from the only randomized sham-CPAP-controlled study reported to date, in which neither HbA1c nor insulin sensitivity showed improvement with therapeutic CPAP treatment.42 Rapid-eye-movementrelated (REM-related) apnoeas and hypopnoeas were found to be particularly associated with worse glycaemic control in type 2 DM, with potential treatment implications regarding adequate inclusion of REM sleep periods in the overnight CPAP usage.43 Recent studies have shown conflicting data that pancreatic insulin secretion may be enhanced or impaired in clinical subjects or experimental human models of OSA, which may reflect differences in age, duration of OSA, glycaemic status or other factors that influence pancreatic function and hence islet cell response to insult from OSA.33–35,44,45 The controversial clinical data on the link between OSA and insulin resistance can be attributed to several factors, including the use of a range of techniques in the evaluation of insulin–glucose metabolism, each with their merits and limitations (Table 1), the limited sample size of most studies with heterogenous patient characteristics including their degree of obesity, and the observational nature and short duration of intervention studies. Experiments in healthy human subjects have shown supporting evidence for a causal role of OSA on insulin–glucose dysmetabolism. Healthy subjects exposed to 5 hours of intermittent hypoxia/normoxia to mimic OSA during wakefulness found impairment in both insulin sensitivity and insulin secretion.44 Cerebral arousals related to recurrent obstructed breathing in OSA may also be pathogenetic for metabolic dysregulation, probably via sympathetic activation. Studies using acoustic stimuli to induce sleep fragmentation, with increased daytime sympathetic activity as assessed by heart rate variability, showed that sleep disruption without hypoxia could lead to decrease in insulin sensitivity, as well as impaired non-insulin-dependent glucose disposal45 or inadequate compensatory increase in insulin secretion.53 Furthermore, impairment of insulin signalling has been demonstrated in human subcutaneous adipocytes collected from seven healthy subjects after experimental sleep restriction compared to after normal sleep.54 Intracellular hypoxia in hypertrophied fat cells in obesity has been proposed as one mechanism for dysmetabolism.55 In mice model of intermittent hypoxia (IH), insulin resistance increased in lean and genetically or dietary-induced obese mice56,57 and also in acute IH-exposed mice that had pharmacologic denervation of sympathetic and parasympathetic nervous systems.58 In
168 D. C. L. Lam et al. Table 1. Assessment tools for glucose metabolism Test
Brief methodology
Parameters measured
Remarks
Blood glucose46
Fasting venous blood sample for plasma glucose level
• Fasting glucose level
• Conventional test for diagnosis of
Spot venous blood sample for glycated Hb level
• Glycaemic status over past
Haemoglobin A1c47 (HbA1c)
DM/impaired fasting glucose
• Used in clinical practice to assess
2–3 months
glycaemic control in past 2–3 months in DM
• HbA1c ≥65% is used for diagnosis of DM (ADA/WHO)
• HbA1c 57–64% is used for diagnosis Oral glucose tolerance test48 (OGTT)
Oral glucose loading (75 g) followed by evaluation of 2-h postloading blood glucose
• Impaired glucose tolerance (IGT)
diagnosis of DM
Oral glucose loading followed by evaluation • Insulin sensitivity of glucose every 30 min. Simultaneous insulin levels measured
Hyperinsulinaemic euglycaemic clamp49
of prediabetes (ADA)
• 2 h glucose ≥111 mmol/l for
A dose–response curve to exogenous insulin • Insulin sensitivity is generated by measuring the variable infusion rate of glucose required to maintain euglycaemia
• 78–11 mmol/l for diagnosis of IGT • May be reflecting insulin secretion in response to glucose loading rather than insulin sensitivity. Poor test reproducibility due to variability of gastrointestinal absorption and other factors
• Gold standard for assessing insulin sensitivity.
• The steady state rate of peripheral
Fasting venous blood sample with glucose and insulin measurements HOMA-IR = insulin (lU/ml) 9 glucose (mmol/l)/225 HOMA-b = [20 9 insulin (lU/ml)]/ [glucose (mmol/l) 35]
• Insulin resistance HOMA-IR
• •
• Insulin secretion HOMA-beta
•
Frequently sampled intravenous glucose tolerance test (FSIGT, FSIVGTT)51
Fasting baseline blood glucose (and insulin), followed by frequent sampling after glucose injection (for insulin sensitivity, insulin is injected 20 min later) for 3 h. A computer model describing plasma dynamics (minimal model) is applied for deriving metabolic parameters
• Assesses both pancreatic beta-
Short insulin tolerance test (SITT)52
Administration of exogenous insulin followed by monitor of fall in blood glucose over the next 30 min to derive the glucose disappearance rate
Homeostasis model assessment (HOMA)50
•
cell secretory capacity and peripheral glucose uptake in response to the bolus IV glucose.
glucose utilization (M value) is measured as milligrams of glucose used per kilogram of body weight per minute. Labour-intensive investigation First derived from epidemiologic studies Measures basal insulin resistance and insulin secretion Reflects mainly hepatic insulin resistance
• Validated for insulin sensitivity against hyperglycaemic euglycaemic clamp. No need for online measurements or external control of infusion.
• Additional information on
• Reflects whole body insulin sensitivity insulin sensitivity is gained by administration of insulin 20 min after the glucose load • Insulin sensitivity • Validated for insulin sensitivity against hyperglycaemic euglycaemic clamp. No need for online measurements or external control of infusion
the rodent model where lean or obese were subjected to normoxia, IH cycles of low (12/hour) or high (60/hour) frequency or sustained hypoxia for 12 h, oxygen tension in fat tissue showed little fluctuation with IH treatment despite swinging arterial oxygen saturation. The overall fat tissue oxygen tensions in IH
groups were lower than the normoxia group and were similar in lean and obese mice. Increased insulin resistance was found in all hypoxic exposure regimens.57 Intermittent hypoxia, combined with glucose infusion, amplified the reversal of diurnal glucose rhythm and also led to pancreatic beta-cell replication in mice.59
© 2014 John Wiley & Sons Ltd Clinical Endocrinology (2015), 82, 165–177
Adipocytokines and insulin resistance in OSA 171 Table 2. (continued) Authors
Sample characteristics
Results in brief
Drummond et al.74
98 patients with moderate–severe OSA
Cuhadaroglu et al.77
44 OSA treated with CPAP for 8 weeks
Li et al.78
141 children with median age 108 years (85–128). Obese vs non-obese; OSA 43 vs no OSA 49 OSA and 30 non-OSA control subjects with both groups stratified by BMI
Baseline leptin levels positively correlated with BMI, fat distribution and OSA severity. BMI was the only predictor of basal leptin levels. No significant change in leptin levels after auto CPAP for 6 months Leptin levels correlated with HOMA-S and HOMA-IR at baseline and FU. No correlation between HOMA-IR and AHI. In CPAP ≥4 h/night (n = 31), reduced total cholesterol, low-density lipoprotein and leptin Diastolic blood pressure, triglyceride, height and BMI z-score were independently associated with leptin concentration. No significant change in leptin levels after intervention for OSA No significant differences in leptin and adiponectin levels between case and controls. Both leptin and adiponectin levels were reduced after 3 months of CPAP treatment Men: OSA had high leptin, IL-6 hsCRP, IR Women: borderline obese, OSA had higher hsCRP. RCT: no change in leptin, IL-6, AD, hsCRP, IR
Sanchez de la Torre et al.80 Kritikou et al.32
Men: 20 OSA vs 18 non-OSA, non-obese; women: 18 OSA vs 21 non-OSA; RCT: 35 CPAP vs 35 sham CPAP, 8 weeks
lated that OSA may modulate the production of adiponectin resulting in hypoadiponectinaemia and thus lead to adverse cardiometabolic effects. Majority of studies in OSA found that hypoadiponectinaemia had strong correlation with obesity and insulin resistance as in general populations, and the relationship between adiponectin levels and OSA was heavily confounded by obesity (Table 3). In a study of middle-aged Chinese men, adiponectin levels were not determined by any sleep apnoea parameter independent of obesity, but by insulin resistance (HOMA-IR) and also urinary noradrenaline levels taken to be reflective of sympathetic activation in OSA.89 When Caucasian children were studied, OSA was associated with worse insulin resistance, and worsening OSA was associated with lower levels of adiponectin and with increasing levels of urinary catecholamines although no direct relationship between the two biomarkers could be identified.92 A few observational studies reported increase in adiponectin levels after CPAP treatment for OSA, but several studies including RCTs have not found any change,17,32,42 and a case-controlled study found, contrary to expectation, a further decrease in adiponectin, after 3 months of CPAP treatment.80 Intermittent hypoxia exposure of mouse adipose tissue culture ex vivo could reduce the expression of high molecular weight adiponectin, which is the active moiety of adiponectin.93 Sustained hypoxic exposure and IH reduced adiponectin levels in mice90 and 3T3-L1 adipocytes, respectively,83 while acute IH exposure in mice model modulated insulin resistance and leptin production but not adiponectin.44 In vitro data have demonstrated suppression of adiponectin gene expression by synthetic sympathomimetic in pre-adipocyte cell lines.94
Adipocyte fatty acid binding protein Fatty acid binding proteins (FABP), a family of small intracellular lipid-binding proteins that reversibly bind hydrophobic ligands which influence energy metabolism and inflammation, is © 2014 John Wiley & Sons Ltd Clinical Endocrinology (2015), 82, 165–177
present in abundant amounts in adipocytes and in macrophages. Adipocyte-FABP, also known as FABP4, plays an important role in regulation of insulin resistance, glucose metabolism and lipid synthesis, and higher circulating A-FABP level was found to be an independent predictor of the subsequent development of the metabolic syndrome and diabetes.95 To date, there are only several published studies on FABP in OSA.96–100 The first study involved 124 Chinese men and those with severe OSA (mean AHI: 58 events/hour of sleep) showed higher serum A-FABP levels compared to obesity-matched subjects with no or less severe OSA.96 Serum A-FABP levels showed significant correlations, independent of obesity and age, with various sleep hypoxaemia parameters and also with insulin resistance (HOMA-IR). In a study of 45 obese OSA subjects, CPAP treatment was found to decrease serum A-FABP levels.98 In a study of 309 children, childhood obesity and OSA were associated with higher plasma FABP4 levels, and selective variants of the FABP4 gene appeared to contribute to the increased circulating levels.99 Recently, another study from Spain, comprising of 125 subjects, reported that levels of FAPB-4 (i.e. A-FABP) but not epidermal fatty acid binding protein (FABP-5) were elevated in OSA,100 but FABP-4 levels did not decrease with CPAP treatment except in those with severe OSA. These studies suggest a potential role of FABP as a mediator of metabolic regulation in OSA, but more data are needed for confirmation.
C-reactive protein, interleukin 6, tumour necrosis factor-alpha, plasminogen activator inhibitor-1 and others These adipocytokines are produced by a variety of cells including adipocytes and macrophages in adipose tissues, with prominent pro-inflammatory properties. Systemic and tissue inflammation play an important part in the pathogenesis of metabolic and vascular dysfunction.9 IL-6 and tumour necrosis factor-alpha (TNFa) have been implicated for somnogenic
170 D. C. L. Lam et al. Table 2. Selected studies examining the relationships between OSA and leptin and related adipokines Authors
Sample characteristics
Results in brief
Cross-sectional, OSA associated with higher leptin levels independent of BMI or fat Phillips et al.19 32 male healthy OSA vs 32 matched non-OSA Higher leptin levels in OSA. Higher weight gain in the year before subjects diagnosis of OSA compared to non-OSA groups Patel et al.66 138 normal or OSA subjects Correlation of both morning and evening levels of leptin with AHI, but correlation removed after adjustment of BMI. Independent association between evening : morning leptin ratio and AHI Tatsumi et al.69 96 male patients with OSAS vs 52 control Leptin levels correlated with BMI, VFA, SFA, AHI, mean arterial SaO2 matched for BMI and lowest SaO2. Mean arterial SaO2 and lowest SaO2 are explantory factors for serum leptin levels but not others McArdle et al.70 21 OSA compated with 21 BMI-matched OSA had higher leptin levels. No difference in adiponectin and TNF-a non-OSA levels Tauman et al.72 130 children (82 28 years); 51 obese vs Leptin was increased in obesity, and correlated with AHI and 79 non-obese; nocturnal hypoxaemia after controlling for BMI Z-score. Leptin levels 43 SDB children, 42 mild SDB, 45 controls correlated with insulin–glucose ratio Kapsimalis et al.75 15 non-OSAS, 26 mild–moderate OSAS, 26 HOMA-IR was not associated with OSA severity independent of obesity. severe OSAS Severity of nocturnal hypoxaemia was associated with leptin and CRP levels independent of obesity Al Lawati et al.76 176 patients with suspected OSA, 68% male OSA severity independently associated with serum levels of leptin and sVCAM-1 Cross-sectional, no independent association between OSA and leptin levels Sch€afer et al.64 85 consecutive male patients with suspected Abdominal fat predicted presence and degree of OSA. Leptin not OSA related to degree of OSA when controlled for body fat Sharma et al.71 40 apneic obese subjects vs 40 non-apnoeic OSA has no independent association with lipid abnormalities, insulin obese controls vs 40 normal weight control resistance, serum leptin and adiponectin levels. Obesity was the major determinant of metabolic abnormalities in this cohort Ursavas et al.79 55 OSAS and 15 age-matched controls No difference in levels of leptin, adiponectin and resistin. Serum ghrelin levels were higher in OSA group Arnardottir et al.81 452 OSA subjects in the Icelandic Sleep Obesity and gender, not OSA, were the predominant determinants of Apnoea Cohort leptin levels. OSA was an independent determinant (32% variance) of leptin levels only in non-obese non-hypertensive subjects Higher abdominal fat in OSA cf non-OSA despite similar BMI. No Hargens et al.82 Young healthy subjects, 12 overweight difference in HOMA-IR, leptin, AD, TNF, IL-6 on adjustment for OSA (moderate) abdominal fat 18 overweight non-OSA 15 normal weight controls Intervention studies of OSA and leptin with or without cross-sectional baseline data Chin et al.15 22 NCPAP treated patients vs 9 controls Leptin decreased after 3 days of CPAP, no change in fasting insulin. Visceral fat area decreased in both groups with/out reduction in body weight after 6 months CPAP, subcutaneous fat area decreased only in the weight reduction group Ip et al.63 30 OSA vs 30 non-OSA BMI-matched Higher serum leptin levels were found in OSA subjects. CPAP for 6 subjects months reduced leptin Baseline, OSA higher leptin and lower ghrelin levels. Leptin levels Harsch et al.65 30 male OSA vs BMI-matched non-OSA. showed no correlation compared to AHI after adjustment for BMI. Evaluated at baseline and CPAP for 2 days Reduced leptin levels after 2 months CPAP (n = 13), especially in and 2 months those with BMIor= 4 h/night (n = 16), decreased blood Dorkova et al.73 syndrome pressure, total cholesterol, ApoB, HOMA-IR, MDA, TNF-a but no significant changes in leptin, IL-6, CRP. No significant changes in groups with CPAP
doi: 10.1111/cen.12597
REQUESTED REVIEW
Obstructive sleep apnoea, insulin resistance and adipocytokines David C.L. Lam, Karen S.L. Lam and Mary S.M. Ip Department of Medicine, University of Hong Kong, Hong Kong SAR, China
Summary Obstructive sleep apnoea (OSA) is associated with multiple cardiometabolic abnormalities. Obesity is considered a major risk factor for the development of OSA, and it is also an established risk factor for insulin resistance and other cardiometabolic disorders. The enigma remains whether OSA has any causal role in the adverse metabolic profile, independent of or beyond that due to obesity. Sleep apnoeas and hypopnoeas result directly in intermittent hypoxaemia and cerebral arousals, both of which may evoke a cascade of downstream biologic responses in various body tissues and cells. Adipose tissue is a major source of adipocytokines many of which play important roles in the regulation of various metabolic functions. It is hypothesized that OSA may, through its unique pathophysiology, affect metabolic function through modulation of production or action of adipocytokines. This review focuses on insulin resistance, glucose metabolism and relevant adipocytokines in the context of OSA. (Received 30 April 2014; returned for revision 24 May 2014; finally revised 18 August 2014; accepted 18 August 2014)
Introduction Obstructive sleep apnoea (OSA) refers to the sleep-related breathing disorder wherein recurrent events of complete or partial collapse of the upper airway at pharyngeal level occur during sleep, hindering airflow into the lungs, resulting in apnoeas, hypopnoeas or increased respiratory efforts.1,2 Sleep studies [polysomnograms, (PSG)] document a range of sleep parameters, including the number of respiratory events that occur per hour of sleep, widely termed as Apnoea hypopnea index (AHI), which forms the basis of diagnosis and severity grading.1,2 Symptomatic OSA (OSA syndrome), defined with the cardinal feature of excessive daytime sleepiness, has been reported to affect about 2–4% of middle-aged adults in different communities, while as many as 9% of women and 24% of men in a
Correspondence: Professor Mary S.M. Ip, Division of Respiratory and Critical Care Medicine, Department of Medicine, 4/F, Professorial Block, Queen Mary Hospital, Pokfulam, Hong Kong SAR, China. Tel.: (852) 2255 4455; Fax: (852) 2816 2863; E-mail: msmip@hkucc. hku.hk © 2014 John Wiley & Sons Ltd
random general population in USA in the 1990s demonstrated an AHI of ≥5 events/hour which is considered as above adult norm.1–3 Given the global escalation of obesity which is the strongest risk factor for OSA irrespective of ethnicity, current prevalence of OSA is estimated to be much higher.3 Treatment strategies include the use of continuous positive airway pressure (CPAP) devices, oral appliances, surgery and other measures including weight reduction in the overweight or obese subject.1 Majority of patients with OSA are overweight or obese, while OSA of a range of severity is found in 40–90% of obese subjects.4,5 OSA is highly associated with various cardiometabolic diseases, and studies showed a five- to sevenfold risk of increase of metabolic syndrome in OSA.4–6 Insulin resistance is strongly though not exclusively associated with obesity, in particular visceral obesity. It is a core component of the metabolic syndrome and a key mechanistic pathway in the pathogenesis of cardiometabolic derangements, especially of type 2 diabetes mellitus (DM). Apart from being a precursor of glucose intolerance and type 2 DM, insulin resistance is also closely linked to other metabolic dysfunction such as hypertension and dyslipidaemia,7 but there is scanty literature regarding the relationships of insulin resistance and adiopocytokines with these metabolic parameters in the context of OSA. Although OSA and many cardiometabolic aberrations share the common risk factor of obesity, the pathophysiologic effects in OSA can potentially influence metabolic homeostasis, and it is postulated that OSA itself may confer cardiometabolic risks ‘independent’ of obesity. Other then direct influence on metabolic pathways without involving adipose tissues, OSA may interact with obesity to propagate or amplify adverse metabolic effects beyond that contributed by adiposity in the absence of OSA, and which may run in a vicious cycle.4–6 Adipocytokines, derived predominantly from adipose tissues, are important mediators of metabolic functions.8,9 Hence, modulation of the expression/secretion or function of adipocytokines by OSA is one potential mechanistic pathway for metabolic dysfunction in OSA. The purpose of this review is to provide a current overview of the relationships between OSA, insulin resistance, glucose metabolism and relevant adipocytokines.
Pathophysiology of OSA The mechanisms of upper airway collapse are multiple and not completely defined. Factors that have been identified to deter165
166 D. C. L. Lam et al. mine the propensity for upper airway collapse during sleep include craniofacial morphology, amount and distribution of adipose tissues in the body, upper airway dilator muscle function, arousal threshold and ventilatory control.2,5 Obstructive breathing events result in transient hypoxaemia which quickly reverts with resumption of unobstructed breathing – a phenomenon termed intermittent hypoxaemia/hypoxia, as well as cerebral arousals and changes in sleep architecture. A plethora of potential systemic pathophysiologic effects may occur, including hemodynamic shear stress, sympathetic nerve activation, oxidative stress, inflammation and hormonal changes of the hypothalamic–pituitary–adrenal axis such as cortisol.2,4– 6,10–12 These pathways are also known to underpin many common cardiometabolic pathologies.4–6 Furthermore, short sleep duration has been associated with obesity and adverse glucose metabolism, while experimental sleep restriction demonstrated adverse insulin–glucose metabolism, lending relative sleep loss in OSA as an additional mediating pathway for metabolic dysfunction.13 Thus, it has been hypothesized that OSA, per se or interacting with obesity and/or other concurrent risk factors, plays a causal role in cardiometabolic derangements (Fig. 1).4–6,10
Obesity and OSA Obesity is the commonest risk factor for development of OSA, although body mass index (BMI) accounts for only a small part of the variability of AHI, implying the multifactorial nature of pathogenesis of OSA.2 A 10% gain in body weight increased the chance of developing moderate–severe OSA by sixfold, while every 1% increase in body weight was associated with a 3% increase in AHI, and weight loss helped to reduce OSA severity but to a less substantial effect.3
OSA Intermittent hypoxia
Sleep fragmentation/Loss
Sympathetic activation Hormonal changes (HPA axis, satiety hormones)
Adipose tissue Hypoxia Oxidative stress Inflammation
Altered adipocytokine release Metabolic dysfunction Insulin resistance, glucose dysmetabolism dyslipidaemia, hypertension
Fig. 1 Putative link of OSA, adipocytokines and metabolic dysfunction.
Both the degree of adiposity and its distribution determine the risk of having OSA. Increased adipose and soft tissues around the upper airway narrow the pharyngeal calibre and increase the likelihood of airway collapse during sleep. OSA and its severity have been reported to be associated with central obesity and abdominal visceral fat, more so than with BMI, particularly in men.5 It has further been proposed that obesity and OSA may be mutually conducive. Repetitive apnoeas and arousals during sleep lead to excessive daytime sleepiness which may reduce motivation to physical activity and predispose to weight gain over time. Presence of sleep apnoea in men with central obesity attenuated the metabolic improvement in response to a lifestyle intervention programme, compared to those without OSA.14 Studies on the impact of CPAP treatment on body weight or abdominal fat have yielded conflicting data,15–18 with two randomized, sham-controlled trials showing no change.17,18 It has been suggested that OSA subjects may be inherently predisposed to weight gain and may experience difficulty in losing weight compared to non-OSA subjects, relating to selective differences in body metabolism regulated by hormones such as insulin, leptin or ghrelin.19 Lifestyle modifications or other interventions resulting in weight loss clearly benefit both metabolic function and OSA in obese subjects.4
OSA, insulin resistance and glucose metabolism The relationship between OSA and a range of glucose dysmetabolism from insulin resistance to overt diabetes mellitus has been investigated, and readers are referred to several recent comprehensive reviews.4,20,21 Epidemiologic studies have consistently shown up an association of OSA and impaired glucose metabolism after adjustments of known confounders.22,23 The Wisconsin Sleep Cohort (mean age 50 years) showed an independent association between severity of OSA and prevalence of DM at baseline, but no increase in incident diabetes at 4-year follow-up,22 while other cohorts found associations between baseline OSA and incident diabetes over 3–10 years.21,23,24 Several cross-sectional studies of community populations consistently found that OSA was associated with increased insulin resistance or impaired glycaemic status, despite adjustment for obesity and other confounders.25–28 Analyses of the European Sleep Apnea Database showed that OSA or its severity was associated with worse glycaemic status (HbA1c levels) in both subjects with and without DM.29,30 Majority of clinical studies in non-diabetic subjects have reported independent associations between OSA and insulin resistance/sensitivity, with dose-dependent effect of OSA on metabolic impairment, although some studies found that the association was abolished after adjusting for BMI and/or other measures of adiposity.20,21 A small proportion of OSA subject are not obese, and in a clinic-based study of Chinese subjects without DM, homeostatic model assessment index on insulin resistance (HOMA-IR) was higher in both obese and non-obese subjects with OSA.31 Clinical studies focusing on non-obese OSA also showed that they had higher insulin resistance or glucose dysmetabolism compared with lean counterparts without OSA.32,33 © 2014 John Wiley & Sons Ltd Clinical Endocrinology (2015), 82, 165–177
Adipocytokines and insulin resistance in OSA 167 In a study of 118 subjects without DM, the frequent sample intravenous glucose tolerance test showed ‘OSA dose-dependent’ impairment of insulin sensitivity and also of pancreatic beta-cell function.34 Despite reduced insulin sensitivity, patients with OSA did not increase pancreatic beta-cell insulin output. To the contrary, in a study of healthy lean young men free of cardiometabolic disease, the presence of mild-to-moderate OSA (mean AHI 112 29 per sleep hour) was associated with insulin resistance and an increase in insulin secretion suggestive of a compensatory mechanism.33 Another study of 45 severely obese adults also found that OSA was associated with increased betacell function, indicated by homeostasis model assessment calculations, in those with normal glucose metabolism.35 It is plausible that homeostatic mechanisms of enhanced insulin secretion may be lost with chronic insult, especially with ageing or prediabetic state. It has been proposed that sleepiness may be a phenotypic marker for insulin resistance in OSA. In non-diabetic subjects with similar BMI and AHI, those with excessive daytime sleepiness had higher HOMA-IR than those without sleepiness, and insulin resistance was improved with CPAP treatment only in the group with baseline excessive sleepiness.36 Associated abdominal or visceral obesity in OSA could contribute to sleepiness in OSA via hypercytokinaemia.5 Despite the rather abundant positive data supporting an independent association of OSA and insulin resistance (or impaired insulin sensitivity), cross-sectional studies cannot be definitive for a causal link. Reported data on the effect of treatment of OSA on insulin–glucose metabolism remain highly controversial. Insulin sensitivity, measured by the hyperinsulinaemic euglycaemic clamp, was acutely improved in 40 non-diabetic subjects with OSA treated with CPAP for 2 days, and the improvement was maintained at 3 months (n = 31). The improvement in insulin sensitivity was driven mainly by the subset of lean subjects with BMI < 30 kg/m2.37 Other observational studies of small sample sizes, in either subjects with or without DM, also suggested improvements in insulin sensitivity/resistance with CPAP treatment, but results are by no means consistent and a number of studies did not find any change.20,21 Most of the randomized controlled trials (RCTs) regarding CPAP treatment of OSA did not provide definitive evidence for improvement in insulin sensitivity/resistance or other measures of glucose metabolism. A RCT involving middle-aged non-diabetic Chinese men showed improvement in insulin sensitivity measured with the short insulin tolerance test, but not in HOMA-IR, after 1 week of CPAP treatment (n = 30) compared to the control group on observation.38 However, contrary to the previous Caucasian study,37 improvement in insulin sensitivity was driven by the subgroup of subjects with BMI >25 kg/m2 (mean BMI: 28 kg/m2), who were obese by Asian criteria. In a cross-over controlled trial of 8-week CPAP duration, no improvement in insulin sensitivity index derived from oral glucose tolerance test was found, but subset analysis of those with severe OSA (AHI ≥ 30) showed significant improvement in insulin sensitivity with CPAP treatment compared to sham CPAP.39 In another RCT comprising of 65 non-diabetic men, © 2014 John Wiley & Sons Ltd Clinical Endocrinology (2015), 82, 165–177
no difference in insulin sensitivity index was observed between the therapeutic and sham CPAP group at 12 weeks. However, at the end of 24 weeks, when the whole study group was receiving CPAP treatment, a significant improvement in insulin sensitivity compared with baseline was seen.40 In a 24-week RCT in obese adults with moderate-to-severe OSA, CPAP alone did not alter insulin resistance, while CPAP combined with weight loss provided incremental reduction in insulin resistance and serum triglyceride compared to that achieved by weight loss alone.41 In diabetic subjects, data from observational studies tended to be favourable with improvement in insulin sensitivity/resistance, while this has not been supported by data from the only randomized sham-CPAP-controlled study reported to date, in which neither HbA1c nor insulin sensitivity showed improvement with therapeutic CPAP treatment.42 Rapid-eye-movementrelated (REM-related) apnoeas and hypopnoeas were found to be particularly associated with worse glycaemic control in type 2 DM, with potential treatment implications regarding adequate inclusion of REM sleep periods in the overnight CPAP usage.43 Recent studies have shown conflicting data that pancreatic insulin secretion may be enhanced or impaired in clinical subjects or experimental human models of OSA, which may reflect differences in age, duration of OSA, glycaemic status or other factors that influence pancreatic function and hence islet cell response to insult from OSA.33–35,44,45 The controversial clinical data on the link between OSA and insulin resistance can be attributed to several factors, including the use of a range of techniques in the evaluation of insulin–glucose metabolism, each with their merits and limitations (Table 1), the limited sample size of most studies with heterogenous patient characteristics including their degree of obesity, and the observational nature and short duration of intervention studies. Experiments in healthy human subjects have shown supporting evidence for a causal role of OSA on insulin–glucose dysmetabolism. Healthy subjects exposed to 5 hours of intermittent hypoxia/normoxia to mimic OSA during wakefulness found impairment in both insulin sensitivity and insulin secretion.44 Cerebral arousals related to recurrent obstructed breathing in OSA may also be pathogenetic for metabolic dysregulation, probably via sympathetic activation. Studies using acoustic stimuli to induce sleep fragmentation, with increased daytime sympathetic activity as assessed by heart rate variability, showed that sleep disruption without hypoxia could lead to decrease in insulin sensitivity, as well as impaired non-insulin-dependent glucose disposal45 or inadequate compensatory increase in insulin secretion.53 Furthermore, impairment of insulin signalling has been demonstrated in human subcutaneous adipocytes collected from seven healthy subjects after experimental sleep restriction compared to after normal sleep.54 Intracellular hypoxia in hypertrophied fat cells in obesity has been proposed as one mechanism for dysmetabolism.55 In mice model of intermittent hypoxia (IH), insulin resistance increased in lean and genetically or dietary-induced obese mice56,57 and also in acute IH-exposed mice that had pharmacologic denervation of sympathetic and parasympathetic nervous systems.58 In
168 D. C. L. Lam et al. Table 1. Assessment tools for glucose metabolism Test
Brief methodology
Parameters measured
Remarks
Blood glucose46
Fasting venous blood sample for plasma glucose level
• Fasting glucose level
• Conventional test for diagnosis of
Spot venous blood sample for glycated Hb level
• Glycaemic status over past
Haemoglobin A1c47 (HbA1c)
DM/impaired fasting glucose
• Used in clinical practice to assess
2–3 months
glycaemic control in past 2–3 months in DM
• HbA1c ≥65% is used for diagnosis of DM (ADA/WHO)
• HbA1c 57–64% is used for diagnosis Oral glucose tolerance test48 (OGTT)
Oral glucose loading (75 g) followed by evaluation of 2-h postloading blood glucose
• Impaired glucose tolerance (IGT)
diagnosis of DM
Oral glucose loading followed by evaluation • Insulin sensitivity of glucose every 30 min. Simultaneous insulin levels measured
Hyperinsulinaemic euglycaemic clamp49
of prediabetes (ADA)
• 2 h glucose ≥111 mmol/l for
A dose–response curve to exogenous insulin • Insulin sensitivity is generated by measuring the variable infusion rate of glucose required to maintain euglycaemia
• 78–11 mmol/l for diagnosis of IGT • May be reflecting insulin secretion in response to glucose loading rather than insulin sensitivity. Poor test reproducibility due to variability of gastrointestinal absorption and other factors
• Gold standard for assessing insulin sensitivity.
• The steady state rate of peripheral
Fasting venous blood sample with glucose and insulin measurements HOMA-IR = insulin (lU/ml) 9 glucose (mmol/l)/225 HOMA-b = [20 9 insulin (lU/ml)]/ [glucose (mmol/l) 35]
• Insulin resistance HOMA-IR
• •
• Insulin secretion HOMA-beta
•
Frequently sampled intravenous glucose tolerance test (FSIGT, FSIVGTT)51
Fasting baseline blood glucose (and insulin), followed by frequent sampling after glucose injection (for insulin sensitivity, insulin is injected 20 min later) for 3 h. A computer model describing plasma dynamics (minimal model) is applied for deriving metabolic parameters
• Assesses both pancreatic beta-
Short insulin tolerance test (SITT)52
Administration of exogenous insulin followed by monitor of fall in blood glucose over the next 30 min to derive the glucose disappearance rate
Homeostasis model assessment (HOMA)50
•
cell secretory capacity and peripheral glucose uptake in response to the bolus IV glucose.
glucose utilization (M value) is measured as milligrams of glucose used per kilogram of body weight per minute. Labour-intensive investigation First derived from epidemiologic studies Measures basal insulin resistance and insulin secretion Reflects mainly hepatic insulin resistance
• Validated for insulin sensitivity against hyperglycaemic euglycaemic clamp. No need for online measurements or external control of infusion.
• Additional information on
• Reflects whole body insulin sensitivity insulin sensitivity is gained by administration of insulin 20 min after the glucose load • Insulin sensitivity • Validated for insulin sensitivity against hyperglycaemic euglycaemic clamp. No need for online measurements or external control of infusion
the rodent model where lean or obese were subjected to normoxia, IH cycles of low (12/hour) or high (60/hour) frequency or sustained hypoxia for 12 h, oxygen tension in fat tissue showed little fluctuation with IH treatment despite swinging arterial oxygen saturation. The overall fat tissue oxygen tensions in IH
groups were lower than the normoxia group and were similar in lean and obese mice. Increased insulin resistance was found in all hypoxic exposure regimens.57 Intermittent hypoxia, combined with glucose infusion, amplified the reversal of diurnal glucose rhythm and also led to pancreatic beta-cell replication in mice.59
© 2014 John Wiley & Sons Ltd Clinical Endocrinology (2015), 82, 165–177
Adipocytokines and insulin resistance in OSA 171 Table 2. (continued) Authors
Sample characteristics
Results in brief
Drummond et al.74
98 patients with moderate–severe OSA
Cuhadaroglu et al.77
44 OSA treated with CPAP for 8 weeks
Li et al.78
141 children with median age 108 years (85–128). Obese vs non-obese; OSA 43 vs no OSA 49 OSA and 30 non-OSA control subjects with both groups stratified by BMI
Baseline leptin levels positively correlated with BMI, fat distribution and OSA severity. BMI was the only predictor of basal leptin levels. No significant change in leptin levels after auto CPAP for 6 months Leptin levels correlated with HOMA-S and HOMA-IR at baseline and FU. No correlation between HOMA-IR and AHI. In CPAP ≥4 h/night (n = 31), reduced total cholesterol, low-density lipoprotein and leptin Diastolic blood pressure, triglyceride, height and BMI z-score were independently associated with leptin concentration. No significant change in leptin levels after intervention for OSA No significant differences in leptin and adiponectin levels between case and controls. Both leptin and adiponectin levels were reduced after 3 months of CPAP treatment Men: OSA had high leptin, IL-6 hsCRP, IR Women: borderline obese, OSA had higher hsCRP. RCT: no change in leptin, IL-6, AD, hsCRP, IR
Sanchez de la Torre et al.80 Kritikou et al.32
Men: 20 OSA vs 18 non-OSA, non-obese; women: 18 OSA vs 21 non-OSA; RCT: 35 CPAP vs 35 sham CPAP, 8 weeks
lated that OSA may modulate the production of adiponectin resulting in hypoadiponectinaemia and thus lead to adverse cardiometabolic effects. Majority of studies in OSA found that hypoadiponectinaemia had strong correlation with obesity and insulin resistance as in general populations, and the relationship between adiponectin levels and OSA was heavily confounded by obesity (Table 3). In a study of middle-aged Chinese men, adiponectin levels were not determined by any sleep apnoea parameter independent of obesity, but by insulin resistance (HOMA-IR) and also urinary noradrenaline levels taken to be reflective of sympathetic activation in OSA.89 When Caucasian children were studied, OSA was associated with worse insulin resistance, and worsening OSA was associated with lower levels of adiponectin and with increasing levels of urinary catecholamines although no direct relationship between the two biomarkers could be identified.92 A few observational studies reported increase in adiponectin levels after CPAP treatment for OSA, but several studies including RCTs have not found any change,17,32,42 and a case-controlled study found, contrary to expectation, a further decrease in adiponectin, after 3 months of CPAP treatment.80 Intermittent hypoxia exposure of mouse adipose tissue culture ex vivo could reduce the expression of high molecular weight adiponectin, which is the active moiety of adiponectin.93 Sustained hypoxic exposure and IH reduced adiponectin levels in mice90 and 3T3-L1 adipocytes, respectively,83 while acute IH exposure in mice model modulated insulin resistance and leptin production but not adiponectin.44 In vitro data have demonstrated suppression of adiponectin gene expression by synthetic sympathomimetic in pre-adipocyte cell lines.94
Adipocyte fatty acid binding protein Fatty acid binding proteins (FABP), a family of small intracellular lipid-binding proteins that reversibly bind hydrophobic ligands which influence energy metabolism and inflammation, is © 2014 John Wiley & Sons Ltd Clinical Endocrinology (2015), 82, 165–177
present in abundant amounts in adipocytes and in macrophages. Adipocyte-FABP, also known as FABP4, plays an important role in regulation of insulin resistance, glucose metabolism and lipid synthesis, and higher circulating A-FABP level was found to be an independent predictor of the subsequent development of the metabolic syndrome and diabetes.95 To date, there are only several published studies on FABP in OSA.96–100 The first study involved 124 Chinese men and those with severe OSA (mean AHI: 58 events/hour of sleep) showed higher serum A-FABP levels compared to obesity-matched subjects with no or less severe OSA.96 Serum A-FABP levels showed significant correlations, independent of obesity and age, with various sleep hypoxaemia parameters and also with insulin resistance (HOMA-IR). In a study of 45 obese OSA subjects, CPAP treatment was found to decrease serum A-FABP levels.98 In a study of 309 children, childhood obesity and OSA were associated with higher plasma FABP4 levels, and selective variants of the FABP4 gene appeared to contribute to the increased circulating levels.99 Recently, another study from Spain, comprising of 125 subjects, reported that levels of FAPB-4 (i.e. A-FABP) but not epidermal fatty acid binding protein (FABP-5) were elevated in OSA,100 but FABP-4 levels did not decrease with CPAP treatment except in those with severe OSA. These studies suggest a potential role of FABP as a mediator of metabolic regulation in OSA, but more data are needed for confirmation.
C-reactive protein, interleukin 6, tumour necrosis factor-alpha, plasminogen activator inhibitor-1 and others These adipocytokines are produced by a variety of cells including adipocytes and macrophages in adipose tissues, with prominent pro-inflammatory properties. Systemic and tissue inflammation play an important part in the pathogenesis of metabolic and vascular dysfunction.9 IL-6 and tumour necrosis factor-alpha (TNFa) have been implicated for somnogenic
170 D. C. L. Lam et al. Table 2. Selected studies examining the relationships between OSA and leptin and related adipokines Authors
Sample characteristics
Results in brief
Cross-sectional, OSA associated with higher leptin levels independent of BMI or fat Phillips et al.19 32 male healthy OSA vs 32 matched non-OSA Higher leptin levels in OSA. Higher weight gain in the year before subjects diagnosis of OSA compared to non-OSA groups Patel et al.66 138 normal or OSA subjects Correlation of both morning and evening levels of leptin with AHI, but correlation removed after adjustment of BMI. Independent association between evening : morning leptin ratio and AHI Tatsumi et al.69 96 male patients with OSAS vs 52 control Leptin levels correlated with BMI, VFA, SFA, AHI, mean arterial SaO2 matched for BMI and lowest SaO2. Mean arterial SaO2 and lowest SaO2 are explantory factors for serum leptin levels but not others McArdle et al.70 21 OSA compated with 21 BMI-matched OSA had higher leptin levels. No difference in adiponectin and TNF-a non-OSA levels Tauman et al.72 130 children (82 28 years); 51 obese vs Leptin was increased in obesity, and correlated with AHI and 79 non-obese; nocturnal hypoxaemia after controlling for BMI Z-score. Leptin levels 43 SDB children, 42 mild SDB, 45 controls correlated with insulin–glucose ratio Kapsimalis et al.75 15 non-OSAS, 26 mild–moderate OSAS, 26 HOMA-IR was not associated with OSA severity independent of obesity. severe OSAS Severity of nocturnal hypoxaemia was associated with leptin and CRP levels independent of obesity Al Lawati et al.76 176 patients with suspected OSA, 68% male OSA severity independently associated with serum levels of leptin and sVCAM-1 Cross-sectional, no independent association between OSA and leptin levels Sch€afer et al.64 85 consecutive male patients with suspected Abdominal fat predicted presence and degree of OSA. Leptin not OSA related to degree of OSA when controlled for body fat Sharma et al.71 40 apneic obese subjects vs 40 non-apnoeic OSA has no independent association with lipid abnormalities, insulin obese controls vs 40 normal weight control resistance, serum leptin and adiponectin levels. Obesity was the major determinant of metabolic abnormalities in this cohort Ursavas et al.79 55 OSAS and 15 age-matched controls No difference in levels of leptin, adiponectin and resistin. Serum ghrelin levels were higher in OSA group Arnardottir et al.81 452 OSA subjects in the Icelandic Sleep Obesity and gender, not OSA, were the predominant determinants of Apnoea Cohort leptin levels. OSA was an independent determinant (32% variance) of leptin levels only in non-obese non-hypertensive subjects Higher abdominal fat in OSA cf non-OSA despite similar BMI. No Hargens et al.82 Young healthy subjects, 12 overweight difference in HOMA-IR, leptin, AD, TNF, IL-6 on adjustment for OSA (moderate) abdominal fat 18 overweight non-OSA 15 normal weight controls Intervention studies of OSA and leptin with or without cross-sectional baseline data Chin et al.15 22 NCPAP treated patients vs 9 controls Leptin decreased after 3 days of CPAP, no change in fasting insulin. Visceral fat area decreased in both groups with/out reduction in body weight after 6 months CPAP, subcutaneous fat area decreased only in the weight reduction group Ip et al.63 30 OSA vs 30 non-OSA BMI-matched Higher serum leptin levels were found in OSA subjects. CPAP for 6 subjects months reduced leptin Baseline, OSA higher leptin and lower ghrelin levels. Leptin levels Harsch et al.65 30 male OSA vs BMI-matched non-OSA. showed no correlation compared to AHI after adjustment for BMI. Evaluated at baseline and CPAP for 2 days Reduced leptin levels after 2 months CPAP (n = 13), especially in and 2 months those with BMIor= 4 h/night (n = 16), decreased blood Dorkova et al.73 syndrome pressure, total cholesterol, ApoB, HOMA-IR, MDA, TNF-a but no significant changes in leptin, IL-6, CRP. No significant changes in groups with CPAP
