|Home | About | Journals | Submit | Contact Us | Français|
Unrecognized obstructive sleep apnea (OSA) is highly prevalent in obesity. Both obesity and OSA are associated with vascular endothelial inflammation and increased risk for cardiovascular diseases. We investigated directly whether endothelial alterations that are commonly attributed to obesity are in fact related to OSA.
Seventy-one subjects with body mass index (BMI) ranging from normal to obese underwent attended polysomnography. To assess directly vascular inflammation and oxidative stress, we quantified expression of nuclear factor kappa B (NFκB) and nitrotyrosine by immunofluorescence in freshly harvested venous endothelial cells. To evaluate basal endothelial nitric oxide (NO) production and activity, we quantified expression of endothelial NO synthase (eNOS) and phosphorylated eNOS (P-eNOS). Vascular reactivity was measured by brachial artery flow-mediated dilation (FMD). Expression of eNOS and P-eNOS and FMD were significantly lower whereas expression of nitrotyrosine was significantly greater in OSA patients (n=38) than in OSA-free subjects (n=33) regardless of central adiposity. Expression of NFκB was greater in obese OSA patients than in obese OSA-free subjects (p=0.004). Protein expression and FMD were not significantly affected by increasing BMI or central obesity in OSA patients and in OSA-free subjects. After 4 weeks of continuous positive airway pressure (CPAP) therapy, FMD and expression of eNOS and P-eNOS significantly increased whereas expression of nitrotyrosine and NFκB significantly decreased in OSA patients who adhered with CPAP≥4 hours daily.
Untreated OSA rather than obesity is a major determinant of vascular endothelial dysfunction, inflammation and elevated oxidative stress in obese patients.
The prevalence of obstructive sleep apnea (OSA) increases markedly with increasing adiposity.1 Fifty to seventy percent of overweight and obese subjects have OSA.2 Despite such overwhelming association, OSA remains unrecognized in the vast majority of obese subjects.3,4 Obesity and OSA are considered independent risk factors for cardiovascular diseases.5,6 Both obesity and OSA have been associated with vascular endothelial alterations that underlie the development and progression of atherosclerosis.7-13 Whereas obesity is regarded as a confounding factor when evaluating vascular endothelial function in patients with OSA, the likely presence of OSA is not routinely considered when evaluating endothelial function in obesity.
Considering the high prevalence of unsuspected OSA among obese subjects, the vascular endothelial alterations that are commonly attributed to obesity may, in fact, be related to OSA. Muscle sympathetic nerve activity is increased when obesity and OSA coexist whereas it is unaffected by obesity alone.14 Recurrent hypoxia during transient cessation of breathing in OSA may exacerbate visceral adipose tissue hypoxia and thereby promote macrophage infiltration of adipose tissue.15-17 Whether unrecognized OSA is an important determinant of the vascular endothelial alterations commonly attributed to obesity has not been investigated. Such possibility is therapeutically relevant considering the challenge of weight reduction and the beneficial impact of therapy for OSA on cardiovascular risk.18
Accordingly, the present study was undertaken to assess directly vascular endothelial function in normal-weight, overweight, and obese subjects who were systematically evaluated for OSA. We hypothesized that OSA rather than obesity is largely responsible for vascular endothelial alterations in obese subjects.
We prospectively recruited normal-weight (BMI<25 kg/m2), overweight (BMI 25-29.9 kg/m2) and obese (BMI ≥30 kg/m2) subjects from the community through advertising (n=40) and from the Sleep Disorders Center at the Columbia University (n=31) between March 2006 and April 2009. None of the study participants had been evaluated for OSA before the present study. All study participants underwent polysomnography. An apnea-hypopnea index (AHI) ≥5 obstructive events per hour of sleep established the diagnosis of OSA. Study participants with AHI<5 events/hour were considered OSA-free. Except for possible obesity, physical examination and routine laboratory tests were normal in all study participants. Waist circumference was defined as the minimal abdominal circumference between the lower edge of the rib cage and the iliac crest. Waist to hip ratio was defined as the waist circumference divided by the circumference determined over the femoral heads (hip circumference). The circumferences were obtained with a flexible tape measure, while maintaining close contact with skin and without compressing the underlying tissues. Subjects were in a standing position and breathing normally. Normal waist circumference was defined as <40 in for men and <35 in for women. Normal waist to hip ratio was defined as ≤0.90 for men and ≤0.85 for women.19 Patients with hypertension, coronary artery disease, heart failure, a history of stroke, diabetes mellitus, chronic obstructive or restrictive pulmonary disease, chronic kidney disease, dyslipidemias, or tobacco use within the past 10 years were ineligible for the study. The study participants were not receiving medications or nutritional supplements. The Columbia University Committee on Human Research approved the study. All study participants gave written informed consent.
All study participants underwent attended polysomnography as previously described.20 Endothelial cells harvesting and flow-mediated dilation (FMD) were performed between 9:00 and 11:00 AM after polysomnography while study participants were in a fasting state. All newly diagnosed patients with OSA underwent continuous positive airway pressure (CPAP) titration. All experimental procedures were repeated after a 4-week treatment period in all OSA patients.
Polysomnography and CPAP titration were performed according to the recommendations of the American Academy of Sleep Medicine.21-23 Patients with OSA who had AHI≥20 events/hour (n=19) underwent split-night polysomnography. The titration protocol for split-night titration studies was identical to that of full-night titration studies. All patients had respiratory disturbance index (RDI) < 5 and a minimum SaO2 above 90% at the prescribed CPAP pressure.23 (Details regarding polysomnography are available in the online-only Data Supplement.) Experimental procedures were performed 48 hours after the split-night or the full-night titration polysomnography in order to avoid any CPAP interference. Adherence with CPAP was defined as CPAP use ≥4 hours daily.18 Adherence was assessed by using CPAP device with compliance software.
A 20-gauge angiocatheter was inserted into a superficial forearm vein. Under sterile conditions, 3 J-shaped vascular guide wires (Arrow, Reading, PA) were sequentially advanced into the vein up to 10 cm. Endothelial cells were retrieved from wire tips by washing with endothelial cell dissociation buffer. Endothelial cells were recovered by centrifugation and fixed with 3.7% formaldehyde in PBS for 10 min, washed twice with PBS, transferred to poly-L-lysine coated slides (Sigma, St. Louis, MO), and air dried at 37°C. The slides were stored at -80°C until analyzed. Harvesting yielded 1365 endothelial cells with a range from 921 to 1767.
To assess vascular inflammation and oxidative stress, we measured venous endothelial expression of nuclear factor kappa B (NFκB) and nitrotyrosine by quantitative immunofluorescence.13 To evaluate basal endothelial nitric oxide (NO) production and activity, we measured venous endothelial expression of total NO synthase (eNOS) and activated eNOS (phosphorylated eNOS at serine1177 [P-eNOS]).
Endothelial cells were permeabilized in PBS/0.5% Triton X-100. Non-specific sites were blocked with PBS-5% donkey serum. Endothelial cells were incubated with monoclonal antibodies against endothelial nitric oxide synthase (eNOS) (Becton Dickinson Transduction Laboratories, San Jose, CA), phosphorylated eNOS at serine1177 (P-eNOS), nitrotyrosine (Upstate Biotechnology, Chicago, IL) and nuclear factor kappa B (NFκB) (Santa Cruz Biotechnologies, Santa Cruz, CA), and followed by Cy3-conjugated donkey anti-mouse antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA). Appropriate negative control slides were generated using preimmune IgG. Polyclonal anti-von Willebrand factor antibodies (DAKO, Glostrup, Denmark) were then used, followed by FITC-conjugated secondary antibodies. Nuclei were stained with diaminophenylindole (DAPI) (Molecular Probes, Carlsbad, CA). Between experiments variability was standardized using reference slides of human umbilical venous endothelial cells (HUVEC) obtained from the same culture dish. Slides from study participants were stained concurrently with one slide of HUVEC. Endothelial cells were analyzed with a fluorescent microscope under identical conditions (Nikon Eclipse E600, Melville, NY), and were captured by digital camera (Q Imaging Retiga EXi, Surrey, BC, Canada). The reader was blinded to subjects' identity. Nuclear and von Willebrand factor staining identified endothelial cells. Slides were systematically read left to right and top to bottom. Twenty-five consecutive endothelial cells were analyzed from each slide. The number of positive (bright) intracellular pixels was quantified using commercially available software (Adobe Photoshop 7.0), and normalized to reference HUVEC slides to calculate pixel ratios (arbitrary units). Quantification by immunofluorescence has been repeatedly validated against immunoblotting in venous endothelial cells with coefficient correlations of 0.99 and 0.97.9,24 Reproducibility of quantitative immunofluorescence of protein expression in venous endothelial cells harvested on the same day was previously assessed.24 The overall coefficient of variation and the mean measurement error for protein expression in venous endothelial cells (17 duplicate measurements) were 11% and 286 pixels, respectively. Reproducibility over time was assessed in 10 healthy OSA-free subjects. The overall coefficient of variation in protein expression was 8% and the mean measurement error 267 pixels in cells harvested 4 weeks apart from the same site.
Vascular reactivity of the brachial artery was measured in the contralateral arm to the endothelial harvesting site by FMD according to the guidelines of the International Brachial Artery Reactivity Task Force.25 (Details regarding FMD are available in the online-only Data Supplement.)
Continuous data are presented as mean±SD and median [Q1-to-Q3], and compared using Wilcoxon rank sum tests. Categorical data are presented as frequency and percentage, and compared using chi-squared tests.
Two-way analysis of variance (ANOVAs) that included BMI category, the diagnosis of OSA, and multiplicative interaction terms for BMI category and the diagnosis of OSA as fixed effects was used to examine the independent associations between these factors and expression of eNOS, P-eNOS, nitrotyrosine, NFκB, and FMD. Waist circumference and waist to hip ratio categories were examined in a similar fashion to BMI category. Tukey's multiple comparison procedure was used to control the type I error rate. Spearman correlation coefficient was used to examine unadjusted associations between BMI, waist circumference and waist to hip ratio and protein expression and FMD in OSA patients and OSA-free subjects. Generalized additive models with loess smoothing functions for continuous variables were used to examine linear relationship between severity of OSA (as assessed by AHI, nadir arterial oxyhemoglobin saturation [SaO2], and time spent below SaO2 of 90% during sleep [t<SaO2 90%]) and protein expression and FMD with adjustment for age, gender and BMI. Linear mixed-effects models with CPAP adherence, age, gender, BMI, time (baseline or follow-up) and an interaction term for CPAP adherence and time as fixed effects and subjects as random effects were used to examine the change in protein expression and FMD in OSA patients from baseline to follow-up. Statistical significance was defined as two-tailed p values <0.05. Statistical analysis was performed with SAS 9.1 (SAS Institute, Cary, NC) and the gam function in R 2.8.1 (R Foundation, Vienna, Austria).
Seventy-one subjects were studied. Thirty-eight subjects were found to have OSA and 33 subjects were free of OSA. The clinical and laboratory characteristics of the study subjects are presented in Table 1. Age, gender, BMI, blood pressure, fasting blood glucose and total cholesterol level were similar in OSA patients and in OSA-free subjects. Waist circumference and waist to hip ratio tended to be greater in OSA patients than in OSA-free subjects. Patients with OSA had significantly lower SaO2 nadir during sleep and had more daytime sleepiness as measured by Epworth Sleepiness Scale than OSA-free subjects.
Expression of eNOS and P-eNOS (markers of endothelial NO production and activity) was significantly lower whereas expression of nitrotyrosine (a marker of oxidative stress) was significantly greater in OSA patients than in OSA-free subjects regardless of central adiposity (Figure 1). Expression of eNOS and P-eNOS was significantly lower whereas expression of NFκB (a marker of inflammation) was significantly greater in obese OSA patients than in obese OSA-free subjects. Expression of NFκB was also significantly greater in OSA patients with increased waist circumference and waist to hip ratio than in their OSA-free counterparts (Figure 1). Neither BMI nor central obesity (increased waist circumference or waist to hip ratio) significantly altered protein expression in OSA patients and OSA-free subjects (Figure 1). The effects of obesity on endothelial protein expression were also analyzed with BMI, waist circumference and waist to hip ratio expressed as continuous variables. Endothelial protein expression did not significantly correlate with BMI, waist circumference and waist to hip ratio in OSA patients and OSA-free subjects (Supplemental Figure 1). Expression of eNOS, phosphorylated eNOS, nitrotyrosine and NFκB was similar in obese (BMI≥30 kg/m2) and non-obese study participants (BMI<30 kg/m2) when not stratified by the presence of OSA (p > 0.20 for all).
Brachial artery FMD, an indirect marker of endothelial NO-mediated reactivity, was significantly lower in normal-weight OSA patients than in OSA-free subjects of similar weight. Overweight OSA patients tended to have lower FMD than overweight OSA-free subjects (Figure 2). Patients with OSA and normal or increased waist circumference and normal waist to hip ratio had lower FMD than their OSA-free counterparts (Figure 2). Resting brachial diameter and percent reactive hyperemia after arm cuff deflation were similar in OSA patients and OSA-free subjects. Body mass index, waist circumference and waist to hip ratio did not significantly alter FMD in OSA patients and OSA-free subjects (Figure 2). Brachial artery FMD did not significantly correlate with BMI, waist circumference and waist to hip ratio in OSA patients and OSA-free subjects (Supplemental Figure 2).
Endothelial expression of eNOS and P-eNOS correlated inversely with AHI whereas expression of NFκB and nitrotyrosine correlated directly with AHI after adjustment for age, gender, and adiposity (Figure 3). Similar correlations were noted between baseline protein expression and t<SaO2 90% and SaO2 nadir (p<0.001 for all).
Twenty-six patients with OSA tolerated CPAP for 4 weeks while 12 did not due to discomfort. Adherence with CPAP ranged from 1 to 8 hours. Nineteen patients adhered with CPAP≥4 hours daily for an average of 6.0±1.4 hours. The 7 remaining patients adhered with CPAP<4 hours daily (average of 2.1±0.4 hours). Adherence with CPAP therapy was similar in patients who underwent split-night and full-night diagnostic polysomnography. Age, severity of obesity and OSA, daytime sleepiness, blood pressure, fasting glucose and total cholesterol levels as well as baseline endothelial protein expression and FMD were similar in patients who tolerated and did not tolerate CPAP. Body weight and blood pressure remained unchanged during CPAP therapy.
Endothelial expression of eNOS and P-eNOS increased whereas expression of NFκB and nitrotyrosine decreased in patients who adhered with CPAP≥4 hours daily after adjustment for age, gender, and BMI (Figure 4). Similarly, FMD increased in patients who adhered with CPAP≥4 hours daily (p=0.02). When patients adhered with CPAP≥4 hours daily, FMD and expression of eNOS and NFκB were no longer different from that of OSA-free subjects (p range 0.08-0.63). Expression of P-eNOS remained lower and expression of nitrotyrosine remained greater than in OSA-free subjects despite adherence with CPAP≥4 hours daily (p=0.005 and 0.001 respectively). Endothelial protein expression and FMD did not change in patients who used CPAP<4 hours daily or declined CPAP (p range 0.10-0.72).
The present findings indicate that OSA rather than obesity is a major determinant of endothelial dysfunction, inflammation and elevated oxidative stress in obese patients. Increased body weight and central obesity were not associated with vascular endothelial dysfunction, inflammation and elevated oxidative stress in the absence of OSA. Furthermore, increasing adiposity does not appear to exacerbate the effects of OSA on the vascular endothelium.
A quarter of the American adult population suffers from OSA that remains overwhelmingly unrecognized.1,3,4 Moreover, only half of the patients diagnosed with OSA receive treatment.4 The prevalence of unrecognized OSA increases markedly with increasing adiposity. When adiposity increases by 1 SD, the risk of unrecognized sleep-disordered breathing increases by a 3-fold.1 Pervasive under-diagnosis of OSA may be due to the lack of daytime symptoms such as excessive sleepiness in the vast majority of OSA patients.1 A thorough history and physical examination cannot reliably exclude the presence of OSA even in severely obese patients.26 Regardless of the presence of daytime symptoms, OSA is an independent risk factor for cardiovascular diseases and increased mortality.5,27
Patients with OSA experience repetitive episodes of hypoxia/reoxygenation during transient cessation of breathing that reduce NO availability and promote inflammation and oxidative stress.11-13,28,29 Endothelial alterations similar to that of OSA have been reported in apparently healthy obese subjects.9 Cross-sectional studies suggest that endothelium-dependent arterial dilation, as assessed by FMD or direct intra-arterial administration of acetylcholine, is reduced in obesity.7,30 Moreover, BMI and visceral obesity correlate inversely with endothelium-dependent vasodilation in apparently healthy subjects.30-32 Elevated levels of soluble intra-cellular and vascular adhesion molecule-1 and E-selectin in obesity are consistent with endothelial inflammation and activation.33-35 Increased systemic oxidative stress is thought to contribute to endothelial dysfunction in obesity.36-39 Central distribution of adiposity, a major determinant of endothelial dysfunction, inflammation, and increased oxidative stress, is strongly associated with OSA.2,33,36,39 Since the majority of obese subjects suffer from unrecognized OSA, it is likely that OSA was present in a substantial subset of apparently healthy obese subjects who participated in the studies that linked obesity to endothelial dysfunction and cardiovascular disease.
Direct evidence that obesity is associated with endothelial inflammation and increased vascular oxidative stress was recently reported.9,40 Venous endothelial expression of NFκB and nitrotyrosine was found to be greater in apparently healthy overweight and obese subjects than in normal-weight counterparts.9 However, as in above studies, the presence of unsuspected OSA was not systematically excluded. In contrast to previous studies that directly and indirectly supported an association between obesity and endothelial dysfunction and inflammation, we could not link central obesity to vascular endothelial dysfunction, inflammation and increased oxidative stress in the absence of OSA. Although various factors may have contributed to these discrepant findings, the systematic exclusion of coexistent OSA clearly differentiates the present study from others.
Our findings of OSA-related vascular endothelial dysfunction in otherwise healthy obese patients concur with previous reports11,12 When compared to healthy subjects matched for adiposity, endothelium-dependent vasodilation is impaired in patients with OSA and improves after treatment. Our findings are also concordant with the observation that muscle sympathetic nerve activity is increased in obese patients with OSA but not in obesity alone.14 Increased endothelial expression of eNOS and P-eNOS and decreased expression of nitrotyrosine and NFκB after effective CPAP therapy provide further evidence that OSA, rather than obesity, was predominantly responsible for endothelial alterations in our study sample. Patients with OSA exhibited endothelial alterations that were reversible with CPAP regardless of the severity of adiposity. Although CPAP was not allocated randomly to our patients, reversal of endothelial dysfunction and inflammation in the absence of change in body weight strongly suggests that OSA is largely responsible for these endothelial alterations in obesity. The observed correlation between severity of OSA and the extent of endothelial alterations supports further OSA as a major contributor to endothelial dysfunction, inflammation and oxidative stress in obese patients.
While adipose tissue is a well known source of inflammation, the mechanisms that trigger and maintain the inflammatory response in obesity are incompletely understood. Adipose tissue hypoxia has been recently proposed as one of the major triggers of macrophage infiltration in adipose tissue.15-17 However, the cascade of events that leads to adipose tissue hypoxia remains unclear.15 The cyclic hypoxia associated with cessation of breathing in OSA may promote adipose tissue hypoxia and inflammation in obesity.
The precise mechanisms underlying cardiovascular risk in OSA and obesity cannot be ascertained from the study of the venous endothelium. Local biomechanical forces that affect arterial endothelial cells at specific sites play an essential role in determining regional susceptibility to atherosclerosis.41-44 Thus, endothelial biopsy at specific sites of the arterial vasculature will likely be required to determine the precise mechanisms underlying atherosclerosis in OSA and obesity. Endothelial cells undergo significant phenotypic drift when removed from their native environment and cultured in vitro.41 Direct characterization of harvested venous endothelial cells without the artifact of culture conditions provides novel insight into the mechanisms that mediate the vascular response to systemic inflammation in OSA and obesity.
In conclusion, OSA rather than obesity appears to be predominantly responsible for vascular endothelial dysfunction, inflammation and increased oxidative stress in obese patients. Accurate assessment of obesity-related vascular risk requires systematic exclusion of unsuspected OSA. Reversal of vascular inflammation with effective therapy for OSA emphasizes the importance of vigilant search and prompt treatment of OSA in overweight and obese subjects.
Unrecognized obstructive sleep apnea (OSA) is highly prevalent in obese subjects. Both obesity and OSA have been associated with vascular endothelial alterations that underlie the development and progression of atherosclerosis and increased risk for cardiovascular diseases. Whereas obesity is commonly regarded as a confounding factor when evaluating vascular endothelial function in patients with OSA, the likely presence of OSA is not routinely considered when evaluating endothelial function in obesity. We investigated directly whether endothelial alterations that are frequently attributed to obesity are in fact related to OSA. Proteins that regulate basal nitric oxide (NO) production and inflammation, and markers of oxidative stress were quantified in venous endothelial cells harvested from normal-weight, overweight, and obese subjects who were systematically evaluated for OSA. The present data provide direct evidence that OSA rather than obesity is a major determinant of endothelial dysfunction, inflammation and elevated oxidative stress in obese patients. Increased body weight and central obesity were not associated with vascular endothelial dysfunction, inflammation and elevated oxidative stress in the absence of OSA. Furthermore, increasing adiposity did not exacerbate effects of OSA on the vascular endothelium. Systematic exclusion of unsuspected OSA is mandatory for accurate assessment of vascular risk in obesity. Reversal of vascular inflammation with effective therapy for OSA emphasizes the importance of a vigilant search and prompt treatment of OSA in overweight and obese subjects.
The authors thank Rui Liu and Zhi Qiang Zhang for assistance with flow-mediated dilation.
Funding Sources: American Sleep Medicine Foundation 21YI03, Irving Center for Clinical Research RR-0645, and American Lung Association CU-52259701 (S.J.); NIH K23 HL086714 (D.J.L.); NIH HL-66211, HL-71042, and HL-79094 (P.F.). No funding source had role in study design, data collection, analysis and interpretation or in the writing of the report and decision to submit the paper for publication.
Conflict of Interest Disclosures: none.