Disclosure: The authors declared no conflict of interest.
Sleep apnea modifies the long-term impact of surgically induced weight loss on cardiac function and inflammation†
Article first published online: 25 MAY 2013
Copyright © 2012 The Obesity Society
Volume 21, Issue 4, pages 698–704, April 2013
How to Cite
Kardassis, D., Grote, L., Sjöström, L., Hedner, J. and Karason, K. (2013), Sleep apnea modifies the long-term impact of surgically induced weight loss on cardiac function and inflammation. Obesity, 21: 698–704. doi: 10.1002/oby.20115
- Issue published online: 25 MAY 2013
- Article first published online: 25 MAY 2013
- Accepted manuscript online: 5 NOV 2012 05:52PM EST
- Manuscript Accepted: 18 SEP 2012
- Manuscript Received: 26 MAR 2012
Obesity is frequently associated with obstructive sleep apnea (OSA). Both conditions are proinflammatory and proposed to deteriorate cardiac function. We used a nested cohort study design to evaluate the long-term impact of bariatric surgery on OSA and how weight loss and OSA relate to inflammation and cardiac performance.
Design and Methods:
At 10-year follow-up in the Swedish Obese Subjects (SOS) study, we identified 19 obese subjects (BMI 31.2 ± 5.3 kg m−2), who following bariatric surgery at SOS-baseline had displayed sustained weight losses (surgery group), and 20 obese controls (BMI 42.0 ± 6.2 kg m−2), who during the same time-period had maintained stable weight (control group). All study participants underwent overnight polysomnography examination, echocardiography and analysis of inflammatory markers.
The surgery group displayed a lower apnea hypopnea index (AHI) (19.9 ± 21.5 vs. 37.8 ± 27.7 n/h, P = 0.013), lower inflammatory activity (hsCRP 2.3 ± 3.0 vs. 7.2 ± 5.0 mg L−1, P < 0.001), reduced left ventricular mass (165 ± 22 vs. 207 ± 22 g, P < 0.001) and superior left ventricular diastolic function (E/A ratio 1.24 ± 1.10 vs. 1.05 ± 0.20, P = 0.006) as compared with weight stable obese controls. In multiple regression analyses including all subjects (n = 39) and controlling for BMI, the AHI remained independently associated with hsCRP (β = 0.09, P < 0.001), TNF-α (β = 0.03, P = 0.031), IL-6 (β = 0.01, P = 0.007), IL 10 (β = −0.06; P = 0.018), left ventricular mass (β = 0.64, P < 0.001), left atrial area (β = 0.08, P = 0.002), pulmonary artery pressure (β = 0.08, P = 0.011) and E/Ea ratio (β = 0.04, P = 0.021).
Patients with sustained weight loss after bariatric surgery display less severe sleep apnea, reduced inflammatory activity, and enhanced cardiac function. Persisting sleep apnea appears to limit the beneficial effect of weight loss on inflammation and cardiac performance.
Obesity is an important risk factor for obstructive sleep apnea (OSA), a syndrome characterized by the partial or complete recurrent collapse of the pharyngeal airway during sleep (1). The prevalence of OSA increases along with the degree of obesity (2, 3) and the two conditions share a common influence on inflammatory activity and cardiac function.
Accumulation of visceral adipose tissue generates a rich source of humoral mediators, including proinflammatory cytokines such as TNF-α and IL-6. Obesity is also associated with increased levels of C-reactive protein (CRP), reflecting a state of low-grade inflammation (4). OSA, per se, has been linked to higher concentrations of TNF-α and IL- 6 (5) and circulating cytokine levels are reduced following treatment of the sleep disorder (6). Recently, Punjabi and Beamer (7) reported on an association between OSA and elevated levels of CRP independent of obesity markers.
Obesity is known to influence the structure and function of the heart, partly through the increased hemodynamic load that accompanies body fat accumulation. Sleep apnea may also contribute to left ventricular hypertrophy and diastolic dysfunction, most likely via hypoxia-induced sympathetic activation and hypertension (8). Interestingly, pro-inflammatory cytokines have been implicated in the pathogenesis of cardiac dysfunction, presumably by direct induction of myocardial hypertrophy, fibrosis and apoptosis (9).
It has been proposed that the cardiovascular risk associated with obesity is amplified by OSA (10, 11) and that low-grade systemic inflammation may provide a common intermediary pathway between these conditions and the development of cardiac disorders. However, to what degree interactions between obesity, OSA and inflammation are involved in the pathogenic process of cardiovascular disease in these patients has not yet been resolved.
Short-term weight loss provides a benefit with respect to obesity-related OSA (12, 13), inflammation (14) and cardiac dysfunction (15). However, to what extent OSA is improved long-term following surgical obesity intervention is still largely unresolved (16, 17). This is of importance since persistent sleep apnea may have a negative impact on the outcomes of bariatric surgery. The aim of the current study was to investigate the long-term effects of surgically induced weight loss on sleep disordered breathing and to evaluate how the presence of sleep apnea relates to circulating inflammatory markers and cardiac performance.
The SOS study is a prospective, controlled, surgical interventional trial, which enrolled 4,047 obese subjects at 25 surgical departments and 480 primary health care centers in Sweden. The study protocol is described in detail elsewhere (18, 19). Briefly, the surgery group comprised 2,010 eligible subjects desiring surgery and, at the same time, a matched control group of 2,037 obese subjects. Inclusion criteria were age ranging from 37 to 60 years and BMI of ≥ 34 for men and of ≥38 or more for women. The exclusion criteria, aiming to obtain operable patients, were the same for both study groups. At SOS-baseline information about sleep apnea was collected through self-administered questionnaires, using questions that have been validated against polysomnography and applied in previous sleep-surveys in Sweden (20, 21). Patients were asked if a family member had observed frequent pauses in breathing during sleep (yes/no). Patients were also asked to rank on a five-point scale the presence of load and disrupting snoring and daytime sleepiness (never, rarely, sometimes, often, and very often). Subjects reporting “often” or “very often” were considered to be frequent snorers or to have frequent daytime sleepiness.
Present study population
Patients enrolled in the present smaller substudy were recruited amongst participants of the SOS study who were subjected to 10-year follow-up at our local study center at Sahlgrenska hospital during 2007. To be included in the study, surgery patients were to display a weight loss >15% compared with SOS-baseline and obese control patients a weight change of <5%. In all, we recruited 39 subjects, including 19 surgery patients and 20 obese controls. The study group comprised 26 women and 13 men in the age range 52-71 years. The study protocol was approved by the ethics committee at the University of Gothenburg (Approval number: S 341 01), and all study subjects gave their informed consent to participate.
Clinical measurements and inflammatory markers
Weight and height were determined to the nearest 0.5 kg and 0.5 cm, respectively. Blood pressure was measured in the supine position after 10-min rest and the mean of two recordings was registered. Blood samples were obtained in the morning after a 12-h fast and inflammatory cytokines, including high sensitivity C-reactive protein (hs-CRP), tumor necrosis factor-alpha (TNF-α), interleukin 6 (IL-6) and interleukin 10 (IL-10) were quantified using an ultra-sensitive bead-based assay (Human TH1/TH2 10-Plex Ultra-Sensitive kit; MSD, Gaithersburg, MD).
Polysomnography at 10-year follow-up included the following signals: two leads of EEG (C4A1; C3A2), two leads of EOG, and a submental EMG continuously registered by surface electrodes. A one-channel ECG was continuously registered. Leg movements were detected by an anterior tibialis electromyogram. Airflow was monitored by combined oronasal thermistors and a nasal pressure canula, abdominal and chest wall movements by inductive plethysmography, and arterial oxygen saturation was measured by finger pulse oximetry. All recordings were stored in a computerized polysomnography system (Embla A10, Flaga, Reykjavik). Sleep stages were manually scored by a trained sleep technician using the Rechtschaffen and Kales criteria (22) and the ASDA criteria (23) for event and arousal scoring. An apnea/hypopnea event was scored if there was a clear decrease (≥50%) in the amplitude of a valid measure of airflow (either by thermistors or nasal canula pressure transducer) during sleep (for hypopnea a ≥3% oxygen desaturation or an associated arousal was required) or the combination of a ≥30% reduction in airflow (compared to pre-event baseline) with at least a 4% reduction of oxygen saturation. A minimum event duration of 10 s is required. The total number of apnea and hypopnea episodes per hour was divided by total sleep time to calculate the apnea–hypopnea index (AHI), which was used as the primary study parameter. The number of dips SaO2 ≥4% per hour of sleep (oxygen saturation index—ODI) was also determined.
In a statistical subanalysis, patients were categorized according to whether their AHI was above or below 20, which was close to the median AHI value for the total study population, and allowed for statistical comparison between subsets. Thus, for each study group, two subsets were generated reflecting a high (AHI > 20) versus low (AHI < 20) intensity of sleep apnea. For convenience, subjects with AHI > 20 are designated in the text as having “high levels of OSA activity” and those with AHI < 20 as having “low levels OSA activity.” In a recent study by Gooneratne et al. (24), an AHI above 20 was associated with a higher mortality hazard ratio. Further, this cut-off level has been applied in studies examining how sleep disordered breathing relates to congestive heart failure (25) and outcomes after cardiac resynchronization therapy (25, 26).
Echocardiography at 10-year follow-up was performed using commercially available standard equipment. Data was acquired with the subject in the left lateral decubitus position at end-expiration. All measurements were averages deriving from three consecutive cardiac cycles. All recordings were made by experienced physicians and analyzed by a single observer blinded to study subject classification using customized dedicated research software (Echopac, GE Vingmed Sound, Horten, Norway). The left ventricular mass was determined using the truncated ellipsoid model according to Byrd et al. (27). Planimetry of the left atrium was performed from a late systolic stop frame with the maximum atrial area. LV ejection fraction was calculated according to the Simpson's rule (28). Pulmonary artery systolic pressure was obtained by calculating the pressure gradient between the right ventricle and atrium from the peak velocity of tricuspid regurgitant flow and adding the estimated value of right atrial pressure. Measurements of early flow velocity (E) and peak velocity during atrial systole (A) were obtained and the ratio E/A was calculated. Myocardial tissue Doppler imaging was performed from the apical four-chamber view. Early (Ea) diastolic tissue velocities were recorded at the septal corner of the mitral annulus and the ratio of early diastolic mitral flow velocity to early diastolic tissue velocity (E/Ea) was subsequently calculated.
Statistical analyses were performed using the SPSS for Windows version 15 (SPSS, Chicago, IL). Descriptive data are summarized as the mean (standard deviation) or proportions (%). Comparisons between groups and between subsets within groups were made by unpaired t tests for continuous variables and Fisher's exact test for categorical data. After pooling data from the two study groups, multiple regression analyses were used to examine how apnea/hypopnea index and BMI were related to inflammatory markers and echocardiography measures of cardiac structure and function. A P value of <0.05 was considered significant.
Clinical characteristics and sleep-questionnaire data for the surgery and control groups at SOS-baseline are shown in Table 1. At this time-point, before intervention, the two study groups were comparable with respect to gender, age, BMI, blood pressure, and comorbidities. Furthermore, the two groups reported similar rates of witnessed sleep apnea, loud and disruptive snoring and daytime sleepiness indicating a comparable prevalence of OSA at SOS-baseline.
|Surgery group (n = 19)||Control group (n = 20)||P value|
|Age (yrs)||49.6 (5.1)||50.8 (4.7)||Ns|
|Sex, male/female (no)||8/11||5/15||Ns|
|BMI, kg m−2||40.6 (4.3)||39.6 (4.0)||Ns|
|Systolic BP, mmHg||142.6 (14.0)||140.1 (15.8)||Ns|
|Diastolic BP, mmHg||91.5 (8.6)||88.9 (8.0)||Ns|
|No (%) on antihypertensives||5 (26.3)||6 (30.0)||Ns|
|No (%) with diabetes||1 (5.3)||1 (5.0)||Ns|
|No (%) current smoker||5 (26.3)||6 (30.0)||Ns|
|No (%) with witnessed sleep apnea||4 (21)||4 (20)||Ns|
|No (%) with frequent snoring||6 (32.6)||8 (40)||Ns|
|No (%) with daytime sleepiness||7 (36.8)||6 (30)||Ns|
Clinical characteristics, polysomnography findings, inflammatory markers, and echocardiography results from the 10-year cross-sectional examination are shown in Table 2. The surgery group displayed a weight loss of 27.2 kg (P < 0.001), while the obese group showed no significant weight change, which was in accordance with the inclusion criteria. Also, the surgery group had lower blood pressure, but there were no significant differences between groups with respect to antihypertensive treatment, diabetes and smoking. Patients treated with bariatric surgery showed substantially lower AHI and ODI as compared with the control group. Also, nearly two thirds of surgery patients had sleep disordered breathing that was only minimal (AHI < 5) or mild (AHI 5-15), whereas 60% of control subjects were found to have OSA that was severe (AHI > 30) (Figure 1). The surgery group had lower levels of pro-inflammatory cytokines hsCRP, TNF-α, and IL-6, and higher concentrations of the anti-inflammatory cytokine IL-10. Surgically induced long-term weight loss was also associated with lower left ventricular mass, atrial area, pulmonary artery pressure and E/Ea ratio, and higher E/A ratio. However, left ventricular ejection fraction did not differ between study groups.
|Surgery group (n = 19)||Control group (n = 20)||P value|
|BMI, kg m−2||31.2 (5.3)||42.0 (6.2)||<0.001|
|Systolic BP, mmHg||130.4 (13.9)||140.1 (14.7)||0.042|
|Diastolic BP, mmHg||77.7 (6.0)||83.9 (11.6)||0.044|
|No (%) on antihypertensives||7 (36.8)||13 (65)||0.113|
|No (%) with diabetes||2 (10.5)||6 (30.0)||0.235|
|No (%) current smoker||6 (31.6)||3 (15.0)||0.273|
|Sleep apnea activity|
|AHI||19.9 (21.5)||37.8 (27.7)||0.013|
|ODI4||8.6 (10.8)||21.3 (25.1)||0.018|
|hsCRP mg L−1||2.3 (3.0)||7.2 (5.0)||<0.001|
|TNF- α pg mL−1||8.7 (1.0)||14.7 (1.6)||<0.001|
|IL- 6 pg mL−1||1.0 (2.4)||2.4 (1.0)||<0.001|
|IL- 10 pg mL−1||11.0 (4.9)||5.3 (2.6)||<0.001|
|Cardiac structure and function|
|Left ventricular mass (g)||165 (22)||207 (22)||<0.001|
|Left atrial area (cm2)||21.8 (3.9)||24.6 (4.7)||0.051|
|PASP (mmHg)||25.3 (2.8)||31.8 (4.3)||<0.001|
|E/A ratio||1.24 (0.10)||1.05 (0.20)||0.006|
|E/Ea ratio||8.6 (2.2)||10.7 (1.7)||0.011|
|Left ventricular ejection fraction (%)||66.3 (5.2)||64.0 (5.3)||0.103|
Findings following stratification of patients into subsets with and without clinically significant sleep apnea (AHI >20 and <20) are shown in Table 3 and Figures 2 and 3. Clinical variables were similar between subsets within study groups, apart from BMI, which was somewhat higher in strata with AHI > 20 as compared to those with AHI < 20, but this was not statistically significant (Table 3). On the other hand, proinflammatory markers hsCRP, TNF-alpha and IL-6 were, or tended to be, significantly lower and anti-inflammatory cytokine Il-10 higher in patient subsets with low levels of sleep apnea (Figure 2). Also, left ventricular mass, left atrial size, pulmonary artery pressure and E/Ea ratio were, or tended to be, significantly lower in patient strata with low OSA activity (Figure 3). In particular, sleep apnea influenced hsCRP, IL-6, left ventricular mass and left atrial size.
|Surgery group (n = 19)||Control group (n = 20)|
|AHI < 20 (n = 15)||AHI > 20 (n = 4)||P value||AHI < 20 (n = 6)||AHI > 20 (n = 14)||P value|
|Age (yrs)||59.5 (4.7)||60.0 (3.9)||0.837||60.0 (7.5)||61.1 (3.9)||0.656|
|BMI (kg m−2)||30.3 (4.7)||34.9 (7.0)||0.132||39.4 (7.5)||43.1 (5.4)||0.225|
|Systolic BP, mmHg||130.3 (13.9)||130.8 (16.4)||0.950||136.7 (9.9)||141.6 (16.5)||0.511|
|Diastolic BP, mmHg||77.1 (5.9)||79.8 (6.8)||0.456||77.8 (6.3)||86.5 (12.5)||0.128|
|No (%) antihypertensives||4 (26.7)||3 (75)||0.540||4 (66.6)||9 (64.3)||0.962|
|No (%) with diabetes||2 (13.3)||0 (0)||0.470||2 (33.3)||4 (28.6)||0.870|
|No (%) current smoker||5 (33.3)||1 (25)||0.800||2 (33.3)||1 (16.7)||0.550|
Multiple regression analysis displaying the importance of sleep apnea activity and BMI for inflammatory markers and echocardiography measures are shown in Table 4. After including all patients and adjusting for the influence of BMI, AHI was independently and positively related to indices of inflammation, left ventricular mass, diastolic dysfunction and pulmonary artery pressure.
|hsCRP||TNF alpha||IL 6||IL 10|
|Adj R2 (%)||64||53||48||34|
|LVM||LA area||PASP||E/Ea ratio|
|Adj R2 (%)||51||25||31||29|
The present study has demonstrated that long-term weight loss following bariatric surgery is associated with a considerably lower degree of sleep apnea. Further, lower weight and less OSA activity were linked to lower inflammatory activity and better left ventricular function. Notably, residual OSA appeared to limit the potential effects of obesity surgery on inflammation and cardiac function.
Our findings are at large in accordance with those of a recent large meta- analysis (17) suggesting a substantial reduction of AHI in obese patients following bariatric surgery. However, in the present study, the prevalence of residual sleep apnea (mean AHI of more than 15 events per hour) following surgery was found to be lower than that reported in the meta-analysis, 37% as compared to 63%. A sleep apnea event intensity of 15 per hour or more may contribute to adverse medical sequelae, such as hypertension, heart disease, stroke and difficulty with weight control (29).
Albeit our study group was small, the findings still support a long-term beneficiary effect of bariatric surgery with respect to sleep disordered breathing in obese patients. In addition, our results add to previous data (30, 31) that weight loss-related alleviation of inflammation and cardiac dysfunction may, to some extent, relate to a concomitant reduction of sleep apnea.
Obesity, OSA, and inflammation
Adipose tissue and visceral fat depots in particular, produce a variety of proinflammatory cytokines including TNF-α and IL- 6. IL-6 is also known to stimulate the hepatic production of CRP (4, 32), contributing further to the state of low-grade inflammation that characterizes obesity. Obstructive sleep apnea has also been proposed to be linked to systemic inflammation, but the nature of this relationship is confounded by the frequent coexistence of visceral adiposity (33).
In the present study AHI was positively related to various inflammatory markers independently of BMI, thus supporting a separate contribution of sleep apnea to systemic inflammation in obesity. This is in line with earlier studies (34, 35), which have demonstrated elevation of inflammatory cytokines in OSA, independent of the degree of obesity. A more recent study (7) has also shown a direct dose-response relationship between OSA and circulating levels of CRP, which was independent of age, BMI, and per cent of body fat. However, studies addressing CRP in OSA have provided inconsistent results which may relate to differences in study population, comorbid conditions and degree of OSA related hypoxia (36, 37).
The lower degree of systemic inflammation in patients with long-term weight loss observed in our study may, at least in part, have resulted from a concomitant reduction in OSA. This would further support a pathogenic role of OSA with respect to systemic inflammation in obesity.
Obesity, OSA, and cardiac function
Obesity is frequently associated with left ventricular hypertrophy, pulmonary hypertension, and diastolic dysfunction (8, 38). In the present study, AHI was correlated with a higher left ventricular mass, higher pulmonary artery pressure and signs of impaired diastolic function, including larger atrial size and higher E/Ea ratio, and these relationships were independent of the degree of obesity. Hence, our results are in accordance with previous studies suggesting that OSA contributes to obesity-related left ventricular hypertrophy (8, 39) and the findings of Otto et al (38) and Arias et al. (40), who reported a greater degree of diastolic dysfunction in obese patients with OSA as compared to those without.
Treatment of obesity with ensuing long-term weight loss was associated with lower left ventricular mass, lower pulmonary artery pressure and enhanced diastolic function. Also, for both study groups, the subsets with low levels of sleep apnea displayed a more favorable echocardiographic profile. Thus, OSA that prevails despite previous obesity intervention might preclude the beneficial effects of weight loss on cardiac structure and function. One contributing mechanism may be a sustained activation of the sympathetic nervous system activity due to apnea-related hypoxemia that results in both systemic and pulmonary vasoconstriction and thereby augmentation of blood pressure and ventricular afterload. Also, OSA-related inflammatory activity that persists despite weight loss might act to maintain adverse myocardial remodeling. Proinflammatory cytokines, including TNF-α and IL-6, have been proposed to impair cardiac function through mechanisms of myocardial hypertrophy, fibrosis, and apoptosis (9), although the pathophysiological significance of low-grade inflammation in this context remains unclear.
Strengths and limitations
The major merit and novelty of this study is the long-term follow-up data provided at 10 years after surgical intervention for obesity. Detailed classification of sleep, cardiac function and metabolic/inflammatory status using gold standard methods is provided. Both sleep and echocardiography data were analyzed by single interpreters to avoid inter-rater variability.
The study also has limitations, which may restrict the interpretability of the results. First, polysomnography, echocardiography, and inflammatory data had not been obtained at baseline and we could therefore not perform a direct comparison of treatment effects between the surgical and control groups. Second, the study was not randomized and subjects differed with respect to surgical motivation, which could indicate other and unforeseen differences between groups. Still, given a similar anthropometric profile and a comparable response to a validated sleep-questionnaire, it is highly likely that the two groups had similar levels of sleep apnea, diastolic dysfunction and inflammatory activity at baseline. Third, modifiable risk factors such as smoking, sedentary life style, caffeine and alcohol consumption, which all may have affected the inflammatory markers or cardiac function, could not be controlled for during the 10-year follow-up period and were not systematically addressed. Finally, the small number of subjects limits the statistical power in our study. However, the conclusion of our study is based on a recurrent pattern suggesting that OSA influences the degree of inflammatory activation and cardiac dysfunction in patients with different degrees of obesity. Our data therefore clearly suggests that the influence of sleep apnea should be considered in prospective outcome studies of bariatric surgery in obese patients.
We report that long-term weight loss by bariatric surgery is associated with lower levels of OSA, reduced systemic inflammation and enhanced cardiac function. Our data suggests that persistent sleep apnea, in spite of weight loss, may limits the improvement of cardiac function and reduction of inflammatory activity. Assessment of persistent sleep apnea is likely to be warranted also after successful treatment of obesity.
Dr Dimitris Kardassis contributed to study design, data acquisition, data analysis, and manuscript preparation. Dr Jan Hedner contributed to data analysis, and manuscript preparation. Dr Lars Sjöström contributed to study design. Dr Ludger Grote contributed to data acquisition, data analysis, and manuscript preparation. Dr Kristjan Karason contributed to study design, data acquisition, data analysis, and manuscript preparation.
- 22A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Bethesda, MD.: U.S. Dept. of Health, Education, and Welfare; 1968., .
- 23American Academy of Sleep Medicine. The aasm manual for the scoring of sleep and associated events: Rules, terminology and technical specifications. Westchester, IL: American Academy of Sleep Medicine; 2007.,
- 24Sleep disordered breathing with excessive daytime sleepiness is a risk factor for mortality in older adults. Sleep 2011; 34: 435-442., , , et al.