Hypoadiponectinemia is Related to Sympathetic Activation and Severity of Obstructive Sleep Apnea

Abstract

Study Objectives:

Hypoadiponectinemia is associated with cardiovascular morbidity and diabetes mellitus. We hypothesize that adiponectin may be downregulated in sleep apnea through various mechanisms, contributing to cardiometabolic risks. This study investigated the relationship between serum adiponectin and sleep disordered breathing and its potential determinants.

Design:

Cross-sectional study.

Subjects and setting:

Adult men without prevailing medical comorbidity from the sleep clinic in a teaching hospital.

Measurements & Results:

One hundred thirty-four men underwent polysomnography, with mean age of 43.9 (9.8) years, and median apnea-hypopnea index (AHI) of 17.1 (5.7, 46.6). Overnight urine samples for catecholamines and blood samples for analyses of insulin, glucose and adiponectin levels from fasting subjects were taken. Insulin resistance was estimated by homeostasis model assessment (HOMA-IR). Magnetic resonance imaging was performed to quantify the amount of abdominal visceral fat. Serum adiponectin level, adjusted for age, body mass index, and visceral fat volume, was significantly lower in subjects with severe obstructive sleep apnea (AHI ≥ 30) compared with those with an AHI of less than 30: 4.0 (3.1, 5.4) versus 5.4 (3.6, 7.9) μg/mL, P = 0.039. After we adjusted for adiposity, adiponectin levels remained negatively correlated with AHI (P = 0.037), arousal index (P = 0.022), HOMA-IR/fasting insulin (P < 0.001), and urinary norepinephrine and normetanephrine (P < 0.008). In a multiple stepwise regression model, the independent determinants of adiponectin after adjustment for adiposity were HOMA-IR (P < 0.001) and urinary norepinephrine and normetanephrine (P = 0.037).

Conclusions:

Adiponectin was suppressed in subjects with severe obstructive sleep apnea, independent of obesity. Adiponectin levels were determined by insulin resistance and sympathetic activation, factors that may be totally or partially attributed to sleep disordered breathing.

Citation:

Lam JCM; Xu A; Tam S; Khong PL; Yao TJ; Lam DCL; Lai AYK; Lam B; Lam KSL; Ip MSM. Hypoadiponectinemia is related to sympathetic activation and severity of obstructive sleep apnea. SLEEP 2008;31(12):1721–1727.

THERE IS GROWING EVIDENCE TO SUGGEST THAT OBSTRUCTIVE SLEEP APNEA (OSA) INDEPENDENTLY CONTRIBUTES TO CARDIOVASCULAR MORBIDITY and mortality; various pathogenetic mechanisms have been described.¹ OSA is highly associated with obesity and other risk factors for cardiovascular disease, such as insulin resistance and the metabolic syndrome.^2–4 It is plausible that intermittent hypoxia and sleep fragmentation in OSA may directly or indirectly modify the expression of various biomediators of vascular and metabolic effects.¹

Adiponectin is secreted exclusively by adipose tissue and found in high concentrations in the circulation of healthy individuals.^5,6 Its levels correlate negatively with percentage of body fat and central fat distribution or visceral obesity,⁷ as well as with serum insulin levels, and positively with glucose disposal during euglycemic clamp.^8,9 Adiponectin has been shown to promote insulin sensitivity in animals ^10,11 and in humans,¹² and the effect is seen independent of obesity.¹⁰ Plasma adiponectin levels have also been reported to be reduced after insulin infusion in healthy men during insulin clamp studies.¹³ These data suggest that insulin signaling is important in the feedback loop regulating serum levels of adiponectin and that adiponectin expression in fat cells may be suppressed by insulin. Adiponectin also possesses anti-inflammatory properties; hypoadiponectinemia has been demonstrated to be closely associated with endothelial dysfunction and cardiovascular morbidity in a few clinical studies.^14–16

With the growing evidence that OSA may independently contribute to atherosclerosis and insulin resistance or diabetes, we hypothesize that adiponectin may play a role in the mediation of the cardiometabolic risks in OSA. This study was designed to evaluate the relationship between adiponectin levels and sleep apnea and to investigate the potential role of hypoxemia, sleep fragmentation, sympathetic activation, and insulin resistance in the regulation of adiponectin expression.

METHODS

Subjects and Study Protocol

Consecutive subjects undergoing polysomnography at the sleep laboratory, University Department of Medicine, Queen Mary Hospital, have been recruited since October 2002 (except March to September 2003 when sleep laboratory services were suspended) into a cohort study to explore various cardiometabolic aspects in relation to OSA, targeting subjects without other prevailing comorbidities. In the current analysis, we addressed serum adiponectin levels and sympathetic activity in relation to sleep apnea in those subjects who were recruited between October 2002 and April 2007. Inclusion criteria were adult men who underwent sleep studies. Exclusion criteria were history of hypertension, taking medications for diabetes mellitus or hyperlipidemia, coronary artery disease, any chronic illness, regularly taking any medications, and fasting serum glucose of 7 mmol/L or higher.

Subjects underwent evaluation of demographic, anthropometric, and clinical profile and had sleep studies. Overnight urine samples for catecholamines and fasting blood samples for insulin, glucose, and adiponectin were collected. Magnetic resonance imaging (MRI) of the abdomen for fat quantification was done within 2 weeks of sleep studies.

The study was approved by the ethics committee of The University of Hong Kong, and all subjects gave written informed consent.

Polysomnography

Subjects underwent overnight polysomnography in the sleep laboratory (Alice 4/5 Diagnostics System, Respironics, Pittsburgh, PA). The polysomnogram consisted of continuous polygraphic recording from surface leads for electroencephalography, electrooculography, electromyography, electrocardiography, nasal pressure transducer (supplemented by thermistor) for nasal and oral airflow, thoracic and abdominal impedance belts for respiratory effort, pulse oximeter for oxyhemoglobin level, tracheal microphone for snoring, and sensors for leg and sleep position. Polysomnographic records were scored manually. Sleep data and arousals were scored according to established criteria.¹⁷ Respiratory events were scored according to American Academy of Sleep Medicine criteria¹⁸: apnea was defined as complete cessation of airflow lasting 10 seconds or more; hypopnea was defined as an abnormal respiratory event lasting at least 10 seconds with at least a 30% reduction in thoracoabdominal movement or airflow, as compared with baseline, and with an oxyhemoglobin desaturation of at least 4%.¹⁹ The average number of episodes of apnea and hypopnea per hour of sleep (the apnea-hypopnea index, AHI) was calculated as the summary measurement of sleep-disordered breathing.

Adiponectin Assay

Sandwich ELISA for Human Adiponectin

Serum adiponectin concentrations were measured with an inhouse monoclonal antibody-based immunoassay established in our laboratory.^20–22 Human serum was diluted 1:5000, and 100 μL of the diluted samples was applied to each well along with the standard, incubated at 37°C for 1 hour, washed 3 times with PBS-T, and then incubated with 100 μL of the biotinylated antihuman adiponectin monoclonal antibody (2 μg/mL) for another 1 hour. Following 3 washings, the wells were incubated with streptavidin-conjugated horseradish peroxidase for 60 minutes and subsequently reacted with tetramethyl-benzidine reagent for 15 minutes. We then added 100 μL of 2 M-H₂SO₄ to each well to stop the reaction, and the absorbance at 492 nm was measured. The intra-assay and inter-assay coefficients of variance were determined by measuring 5 serum samples from healthy subjects in a total of 6 independent assays with duplicate determinations. The intra-assay and inter-assay coefficients of variance were 4.4% to 5.9% and 5.2% to 6.3%, respectively, and the sensitivity and linear range of the assay were 0.5 ng/mL and 0.5 to 50 ng/mL, respectively.

Insulin and Glucose Measurements

Serum insulin was assayed by microparticle enzyme immunoassay using a monoclonal mouse anti-human insulin antibody (Abbott Laboratories, Tokyo, Japan). The intra-assay and interassay coefficients of variation were 4.8% and 5.1%, respectively. Plasma glucose was measured by the glucose oxidase method on an autoanalyzer (Beckman Instruments, Bream, CA). The average glucose values of 3 fasting blood samples taken over 10 minutes were used for calculation of the homeostasis model assessment index of insulin resistance (HOMA-IR) with the formula: fasting plasma glucose (mmol/L) x fasting serum insulin (mIU/L)/22.5.³

Urine Catecholamines

Urine was collected from 22:00 to 08:00, including during all sleep hours on the night of the sleep study, in dark glass bottles, each containing 20 mL of 2 M sulfuric acid. The levels of norepinephrine and epinephrine, and their corresponding O-methylated metabolites, normetanephrine and metanephrine, were measured by isocratic ion-pairing high-performance liquid chromatography after column extraction.^23,24 All results were corrected for urine volume and creatinine clearance and expressed as nmol per mmol of creatinine.

Magnetic Resonance Imaging (MRI) for Abdominal Fat

Patients underwent abdominal MRI using a 1.5 T magnet (Signa Horizon LX, General Electric Medical Systems, Milwaukee, WI) in the morning, having fasted for at least 4 hours. Four-slice parasagittal localizer scans were performed to locate the lumbar vertebrae. Axial spin echo scans (TR 350 ms, TE 12 ms, slice thickness 10 mm) were obtained from the inferior endplate of L1 to the inferior endplate of L5.

A total of 15 to 19 consecutive slices were obtained from the subjects, depending on their length, for interrogation of the visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT), and total abdominal adipose tissue. Given the high grayscale variation within the image regions representing fat tissue and other tissues like liver and bowels, the determination of an adequate threshold of the signal intensity was performed in each patient individually to distinguish between adipose and lean tissues on magnetic resonance images. The images were then segmented, and regions of interests of the SAT and VAT were created and their respective areas measured using Analyze 7.0 (Biomedical Imaging Resource, Mayo Clinic, Rochester, MN) software.

Abdominal SAT area was defined as the area of adipose tissue between the skin and the outermost aspect of the abdominal muscle wall. All adipose tissue pixels within the intra-abdominal cavity at the innermost aspect of the abdominal and oblique wall musculature and the anterior aspect of the vertebral body were considered VAT. High-intensity non-fat pixels arising from fatty intestinal contents were avoided or manually removed when possible. These areas were multiplied by slice thickness, and the products were summated to obtain the volume of adipose tissue.

Statistical Analysis

All parameters were summarized by descriptive statistics. Logit and log transformations were performed for percentile and continuous data, respectively, to obtain normality when necessary before analysis. Continuous and categorical parameters were compared using the Student t-test and Chi-square test, respectively. Comparison of a continuous parameter between groups after adjustment for age, body mass index (BMI), waist circumference, or visceral fat was made by performing a linear regression of the parameter on the grouping factor and these confounders and testing the significance of the coefficient of the grouping factor. Linear association of 2 continuous parameters was assessed by Pearson's correlation coefficient without adjustment and by partial correlation coefficients with adjustment of confounding factors. Comparisons of groups with OSA of different severity, as defined by AHI, were done by post hoc analysis with adjustment for age, BMI, waist or visceral fat volume. Stepwise multiple linear regression was used to determine factors that independently predicted adiponectin levels. Age and obesity or visceral obesity parameters were forced into the model to ensure adjustments. All tests were 2-sided with a P value of less than 0.05 as the level of significance. Statistical analyses were performed with SPSS for Windows software, version 14.0.

RESULTS

A total of 933 men referred for diagnostic polysomnography were screened; 203 subjects were eligible according to criteria, but 30 subjects were not included in this study because of nonavailability of MRI studies due to hospital logistic issues. Of the remaining 173 subjects, 11 of them refused to participate. Thus, 162 subjects were enrolled, and those with the following parameters were excluded from further analyses: exceedingly high triglyceride levels (n = 1), urine catecholamine assay interference (n = 4), adiponectin assay interference (n = 5), total sleep time less than 240 minutes during polysomnographic recording (n = 3), and defaulted MRI quantification of abdominal fat (n=15). One hundred thirty-four subjects were finally evaluated, with a mean (SD) age of 43.9 (9.8) years, BMI of 26.6 (4.1) kg/m², waist circumference of 93.0 (10.1) cm, and abdominal visceral fat of 1053332 (490582) mm³ and with a median (interquartile) AHI of 17.2 (5.7, 46.6) and adiponectin level of 4.9 (3.3, 6.8) μg/mL.

When subjects were divided into 3 groups with AHI less than 5, 5 to less than 30, and 30 or higher, adiponectin levels adjusted for adiposity showed a significant trend (adjusted for waist: P = 0.059, adjusted for visceral fat volume, P = 0.065), and post hoc analysis indicated that there was a significant difference between the group with an AHI less than 5 and an AHI of at least 30 (P = 0.02). Further comparisons with the cohort divided into AHI less than 30 and 30 or higher are shown in Table 1. The severe OSA group was significantly more obese and more insulin resistant. They had higher urine norepinephrine and normetanephrine levels (P < 0.001), and their serum adiponectin levels were lower (P = 0.003); these 2 parameters remained significantly different after adjustment for age, BMI, and visceral fat volume (P < 0.001 and P = 0.039, respectively).

Table 1.

The Clinical Profiles of 134 Subjects with AHI < 30 and AHI ≥ 30

	AHI < 30 n = 87	AHI ≥ 30 n = 47	P value
Age, y	43.5 ± 10.1	44.8 ± 9.5	0.459
BMI, kg/m2	25.6 ± 3.7	28.4 ± 4.1	< 0.001
Waist circumference, cm	90.4 ± 9.8	97.9 ± 8.7	< 0.001
MRI visceral fat, mm3	945694 ± 467468	1252578 ± 474159	0.001
Smoker, no. (%)	40 (46)	23 (49)	0.743
AHI, events/ha	9.2 (2.0, 17.0)	57.8 (44.4, 70.9)	< 0.001
Arousal index, events/ha	16.4 (11.3, 25.1)	51.8 (35.5, 64.0)	< 0.001
Duration with saturation < 90%, mina	3.6 (0.5, 11.0)	105 (47.5, 178.5)	< 0.001
Minimum O2 saturation, %b	82 (79, 88)	62 (50, 73)	< 0.001
Sleep effi ciency, %b	86 (78, 91)	83 (78, 90)	0.347
Serum insulin, mIU/La	9.3 (6.4,13.2)	12.0 (9.6, 19.5)	0.001
Serum glucose, mmol/L	5.1 ± 0.6	5.3 ± 0.6	0.039
HOMAa	2.0 (1.4, 3.0)	2.9 (2.2, 4.6)	0.001
Urine (NE+NME), nmol/mmol creatinine	24.5 ± 96	33.7 ± 11.3	< 0.001
Serum adiponectin, μg/mLa	5.4 (3.6, 7.9)	4.1 (3.1, 5.4)	0.003

Abbreviations: BMI refers to body mass index; AHI, apnea-hypopnea index; HOMA, homeostasis model assessment for estimating insulin resistance; (NE+NME), urine norepinephrine and normetanephrine

Data are presented as mean ± SD or median (interquartile range) unless otherwise stated. Compared by 2-tailed t-test or Chi-Square test.

^alog-transformed before analysis

^blogit-transformed before analysis

As expected, serum adiponectin levels correlated negatively with all obesity indices: BMI, waist circumference and visceral fat volume. After adjustment for age and obesity, adiponectin levels remained significantly correlated with severity of OSA, arousal index, measures of insulin resistance, and urine norepinephrine and normetanephrine levels (Table 2).

Table 2.

Associations of Serum Adiponectin Level With Other Parameters in 134 Subjects

Serum adiponectin, μ g/mLa
Pearson's correlation	r	P value
Age, y	0.053	0.543
BMI, kg/m2	−0.249	0.004
Waist circumference, cm	−0.251	0.003
Visceral fat volume, mm3	−0.255	0.003
Partial correlation coefficients with adjustment for age, BMI	+ waist circumference		+ visceral fat volume
	r	P value	r	P value
AHI, events/ha	−0.172	0.050	−0.183	0.037
Duration with O2 saturation < 90%, mina	−0.120	0.171	−0.127	0.147
Arousal index, events/ha	−0.202	0.021	−0.200	0.022
Minimum O2 saturation, %b	0.143	0.105	0.141	0.109
Sleep efficiency, %b	0.084	0.341	0.079	0.368
Serum insulin, mIU/La	−0.410	< 0.001	−0.394	< 0.001
Serum glucose, mmol/L	−0.253	0.004	−0.262	0.003
HOMAa	−0.431	< 0.001	−0.417	< 0.001
Urinary (NE+NME), nmol/mmol creatinine	−0.234	0.007	−0.231	0.008

Abbreviations: BMI refers to body mass index; AHI, apnea-hypopnea index; HOMA, homeostasis model assessment for estimating insulin resistance; NE+NME, norepinephrine and normetanephrine level

By Pearson's correlation coeffi cients and partial correlation coeffi cients with adjustment for age, BMI, and waist circumference or visceral obesity

^alog-transformed before analysis

^blogit-transformed before analysis

The levels of urine norepinephrine and normetanephrine, individually or summatively, were highly correlated with various sleep indices (P < 0.001), independent of age and visceral obesity (Table 3), whereas the other catecholamines, epinephrine and metanephrine, did not show any significant correlation (Data not shown).

Table 3.

Associations Between Urine Catecholamine Levels and Other Parameters in 134 Patients

Urine (NE+NME) level (nmol/mmol creatinine)
Pearson's correlation	r	P value
Age, y	0.438	< 0.001
BMI, kg/m2	−0.021	0.807
Waist circumference, cm	0.038	0.666
Visceral fat volume, mm3	0.124	0.154
Partial correlation coefficients with adjustment for age, BMI	+ waist circumference		+ visceral fat volume
	r	P value	r	P value
AHI, events/h	0.400	< 0.001	0.403	< 0.001
Duration with O2 saturation < 90%, minutesa	0.392	< 0.001	0.394	< 0.001
Arousal index, events/ha	0.395	< 0.001	0.394	< 0.001
Minimum O2 saturation, %b	−0.382	< 0.001	−0.382	< 0.001
Sleep effi ciency, %b	0.028	0.751	0.029	0.743

Abbreviations: BMI refers to body mass index; AHI, apnea-hypopnea index.

By Pearson's correlation coeffi cients and partial correlation coeffi cients with adjustment for age, BMI, and waist circumference or visceral obesity

^alog-transformed before analysis

^blogit-transformed before analysis

In the stepwise regression model, adjusted for age, BMI, waist circumference or visceral fat volume, HOMA-IR, and urine ( norepinephrine + normetanephrine) emerged as significant determinants of adiponectin levels (Table 4).

Table 4.

Stepwise Multiple Regression Models of Adiponectin Level in 134 Patients^a

Model 1—Adjusted for Age, BMI, and Waist Circumference
Adjusted R2 = 23.9%
	Adjusted R2	Estimate (SE)	P value
Age, BMI and waist	6.2 %	—	—
HOMAa	15.6 %	−0.407 (0.083)	< 0.001
Urinary (NE+NME), nmol/mmol creatinine	2.1 %	−0.008 (0.004)	0.037
Model 2—Adjusted for Age, BMI, and Visceral Fat Volume
Adjusted R2 = 22.6%
	Adjusted R2	Estimate (SE)	P value
Age, BMI, and visceral fat volume	4.4 %	—	—
HOMAa	17 %	−0.430 (0.084)	< 0.001
Urinary (NE+NME), nmol/mmol creatinine	2.1 %	−0.008 (0.004)	0.035

Abbreviations: BMI refers to body mass index; AHI, apnea-hypopnea index; HOMA, homeostasis model assessment for estimating insulin resistance; NE+NME, norepinephrine and normetanephrine level

Independent variables considered: apnea-hypopnea index (AHI)^a, arousal index^a, duration with O₂ saturation <90% ^a, AHI < 30/AHI ≥30, minimum O₂ saturation^b, homeostasis model assessment for estimating insulin resistance (HOMA)^a, fasting serum insulin^a, and urine norepinephrine and normetanephrine (NE+NME).

^alog-transformed before analysis

^blogit-transformed before analysis

DISCUSSION

In this cohort of men without prevalent cardiovascular disease or diabetes mellitus, we demonstrated that circulating adiponectin levels were reduced and correlated with severity of OSA, independent of age and obesity. We also identified that, apart from adiposity, other possible cardiovascular risk factors, including insulin resistance and nocturnal urine catecholamine levels, were independent determinants of hypoadiponectinemia. It is speculated that there may be different mechanisms with complex interplay in the regulation of adiponectin in OSA.

Several recent studies have investigated the relationship between adiponectin levels and sleep disordered breathing, and the results were conflicting. In a large study comprising more than 200 Japanese men with other comorbidities, Makino et al found that hypoadiponectinemia was strongly related to visceral adiposity and insulin resistance but not to any sleep parameter, including AHI, minimum oxygen saturation, slow wave sleep, or duration of oxygen saturation less than 90%.²⁵ Similarly, Sharma et al found that adiponectin levels were correlated with obesity rather than AHI or sleep hypoxemia in Indian men with other known cardiovascular risk factors,²⁶ and Tauman et al also identified a correlation between adiponectin and BMI rather than AHI in children.²⁷ Wolk et al randomly selected male subjects with or without OSA and found that subjects with OSA had even higher plasma adiponectin, as compared with obesity-matched subjects without OSA, but their blood samples were taken in the evening.²⁸ In another BMI-matched case-controlled study, McArdle et al reported no difference in adiponectin levels in 28 otherwise healthy subjects with moderate OSA, compared with those without OSA, despite significant differences in other parameters, including urine catecholamine levels.²⁹ Furthermore, a randomized controlled study showed no change in adiponectin levels after 3 months of treatment with continuous positive airway pressure (CPAP), compared with sham-CPAP, in men with severe OSA.³⁰ However, the subjects had type 2 diabetes, and many were receiving various medications and could not be compared with our subjects who had no overt diabetes and were free from medication use.

In contrast with the above studies, Masserini and coworkers found a trend of decreasing adiponectin levels in 3 groups of subjects with increasing severity of OSA who were similarly obese, where the significant difference was seen between the group with severe OSA compared with those with an AHI less than 10.³¹ Besides, 2 recent studies also reported reduced adiponectin levels in subjects with severe OSA, compared with subjects without OSA, and the adiponectin level was significantly improved in subjects with severe OSA after a single night of treatment with nasal CPAP treatment. However, the authors did not exclude other comorbidities, and these subjects were on different medications that might have affected the serum adiponectin levels.^32,33

In our study comprising men without any prevailing hypertension, diabetes, or cardiovascular disease, adiponectin levels correlated with both obesity or visceral obesity and severity of sleep disordered breathing. The relationship between adiponectin levels and OSA, independent of adiposity, was significantly related to severity of OSA, as indicated by AHI.

Subjects with OSA have been reported to show increased sympathetic activity, as demonstrated by microneurography or catecholamine levels.³⁴ OSA may cause sympathetic activation, probably due to cerebral arousals and hypoxemia following apneas and hypopneas. Obesity per se has been reported to lead to heightened sympathetic activity. In this study, nocturnal urine catecholamines levels were strongly associated with the severity of OSA as well as with adiponectin levels, independent of BMI, and waist circumference or visceral fat volume. On univariate analysis, adiponectin levels, adjusted for visceral obesity, correlated with both AHI and arousal index, whereas, in the multivariate analysis, in which other potential mechanisms of adiponectin regulation—including sympathetic activation and insulin resistance—were also considered, none of the sleep variables emerged as an independent factor. Previous studies have shown that sleep fragmentation is associated with increased levels of lipids and cortisol, increased blood pressure,³⁵ and adverse glucose homeostasis.³⁶ Therefore, with our findings, it can be postulated that sympathetic activation, likely as a result of recurrent arousals in OSA, is a pathway through which sleep disordered breathing may contribute to the determination of adiponectin levels.

The regulation of synthesis and secretion of adiponectin are not yet well delineated. The role of sympathetic stimulation has been investigated in in-vitro studies. In these studies, adiponectin gene expression in preadipocyte cell lines was severely suppressed by the synthetic β-adrenergic agonist, isoproterenol; the adenylate cyclase activator, forskolin; and the synthetic cAMP analogue, dibutyryl-cAMP.³⁷ Although norepinephrine is predominantly an α-adrenergic catecholamine with vasoconstricting properties in vivo and in vitro, it has also been demonstrated to manifest β-adrenergic activities of enhancing glucose entry at the adipocyte level, involving the phosphoinositide-3 kinase pathway,³⁸ which has been recently shown to be involved with the regulation of adiponectin gene expression.³⁹ On the contrary, an in-vivo study of a model showed that administration of adiponectin induced dose-dependent suppression of renal sympathetic nerve activity.⁴⁰ Thus, the relationship of adiponectin and sympathetic activation may be bidrectional.

Several in-vitro studies have demonstrated that hypoxia downregulates the expression of adiponectin.^41,42 Exposure to hypoxia decreased adiponectin concentrations by inhibiting adiponectin regulatory mechanisms at both secretion and transcriptional levels in in-vivo and in-vitro studies, regardless of whether the hypoxia was intermittent or sustained.³³ We were not able to identify nocturnal hypoxemic indices as determinants of serum adiponectin level in subjects with OSA, although we cannot definitively rule out a possible contribution from hypoxemia. Our findings suggest that those patients with severe OSA (AHI > 30) are more at risk for having suppressed adiponectin levels; thus, it is possible that we need to evaluate more subjects with severe OSA and a greater severity of hypoxemia to delineate any possible contribution from hypoxemia. Our results are consistent with sympathetic activation being a common central intermediary pathway in OSA for regulation of adiponectin.

Serum adiponectin levels are well established to correlate inversely with insulin resistance in the general male population.^7,25 It was therefore not unexpected to find a similar correlation in this study, although this is a biased cohort with the majority having some degree of sleep disordered breathing. Animal and cell studies have demonstrated that adiponectin has insulinsensitizing effects,^11,12 whereas insulin may downregulate adiponectin expression.¹³ Given the cross-sectional nature of the study, we cannot delineate a causal relationship between adiponectin levels and insulin resistance, though it is tenable that bidirectional effects are present. We have not studied the effects of CPAP treatment, which would give a better picture of the causal and potentially interactive links between OSA and these metabolic parameters. Furthermore, it has been demonstrated that the high molecular weight oligomer is the major active form of adiponectin, mediating its insulin sensitizing and atheroprotective effects,⁴³ and it would be of interest to investigate if the high molecular weight oligomer is specifically affected by sleep disordered breathing.

Our regression model can account for only about 24% of the variance in circulating adiponectin levels, underscoring that many concurrent factors, unknown at present, contribute to the regulation of adiponectin expression. The clinical relevance of this small statistically significant suppression of serum adiponectin levels in severe OSA is unclear, but, given the intricacy of the metabolic network in the human biologic system, the regulation of any hormones or adipocytokine is likely to be multifactorial and complex. Notwithstanding this, our findings suggest that sleep disordered breathing may contribute toward the down-regulation of adiponectin in men, possibly through sympathetic activation, and that hypoadiponectinemia may be one of the links between OSA and cardiometabolic diseases.