CPAP may promote an endothelial inflammatory milieu in sleep apnoea after coronary revascularization

Summary

Background

Continuous positive airway pressure (CPAP) has failed to reduce cardiovascular risk in obstructive sleep apnoea (OSA) in randomized trials. CPAP increases angiopoietin-2, a lung distension-responsive endothelial proinflammatory marker associated with increased cardiovascular risk. We investigated whether CPAP has unanticipated proinflammatory effects in patients with OSA and cardiovascular disease.

Methods

Patients with OSA (apnoea-hypopnea index [AHI] ≥15 events/h without excessive sleepiness) in the Randomized Intervention with CPAP in Coronary Artery Disease and OSA study were randomized to CPAP or usual care following coronary revascularization. Changes in plasma levels of biomarkers of endothelial (angiopoietin-2, Tie-2, E-selectin, vascular endothelial growth factor [VEGF-A]) and lung epithelial (soluble receptor of advanced glycation end-products [sRAGE]) function from baseline to 12-month follow-up were compared across groups and associations with cardiovascular morbidity and mortality assessed.

Findings

Patients with OSA (n = 189; 84% men; age 66 ± 8 years, BMI 28 ± 3.5 kg/m², AHI 41 ± 23 events/h) and 91 patients without OSA participated. Angiopoietin-2 remained elevated whereas VEGF-A declined significantly over 12 months in the CPAP group (n = 91). In contrast, angiopoietin-2 significantly declined whereas VEGF-A remained elevated in the usual care (n = 98) and OSA-free groups. The changes in angiopoietin-2 and VEGF-A were significantly different between CPAP and usual care, whereas Tie-2, sRAGE and E-selectin were similar. Greater 12-month levels of angiopoietin-2 were associated with greater mortality. Greater CPAP levels were associated with worse cardiovascular outcomes.

Interpretation

Greater CPAP levels increase proinflammatory, lung distension-responsive angiopoietin-2 and reduce cardioprotective angiogenic factor VEGF-A compared to usual care, which may counteract the expected cardiovascular benefits of treating OSA.

Funding

National Institutes of Health/National Heart, Lung, and Blood Institute; Swedish Research Council; Swedish Heart-Lung Foundation; ResMed Foundation.

Research in context.

Evidence before this study

• •Cardiovascular risk persists after Continuous Positive Airway Pressure (CPAP) therapy in patients with obstructive sleep apnoea (OSA) despite the elimination of intermittent hypoxia and fragmented sleep, the purported causes of increased cardiovascular risk in OSA.

Added value of this study

• •CPAP therapy alters unfavourably the natural trajectory of biomarkers of endothelial and lung epithelial inflammation and angiogenesis in patients with OSA and coronary artery disease following coronary revascularization.
• •A similar biomarker milieu is associated with worse outcomes in mechanically ventilated patients with acute lung injury.
• •Greater CPAP level is associated with more unfavourable trajectory of biomarkers of inflammation following coronary revascularization.

Implications of all the available evidence

• •Our findings raise the possibility that by increasing lung volumes, CPAP may contribute to lung endothelial and epithelial inflammation in a manner analogous to ventilator-induced lung injury.
• •More conservative CPAP levels that control daytime OSA symptoms, even if they do not eliminate all apnoeas and hypopneas, may need to be considered to minimize potential side effects from lung hyperinflation.

Introduction

Obstructive sleep apnoea (OSA), a condition that affects approximately a quarter of adults in Western societies, is associated with more than a two-fold increased risk for hypertension, ischemic stroke, coronary artery disease and mortality.¹ Observational studies have suggested that continuous positive airway pressure (CPAP) therapy improves cardiovascular outcomes in OSA; however, three recent randomized clinical trials of CPAP therapy for secondary prevention of cardiovascular disease in patients with OSA failed to confirm those findings in intention-to-treat analyses.2, 3, 4 This indicates that cardiovascular risk persists despite the elimination of intermittent hypoxia and fragmented sleep, the purported causes of increased cardiovascular risk in OSA. While these trials have been criticized for poor adherence to CPAP, even among participants with good CPAP adherence there was no significant reduction in cardiovascular risk in the two largest trials.^2,3 Whether CPAP therapy may have unrecognized, potentially detrimental side effects that counteract its beneficial effects in eliminating hypoxia burden and sleep fragmentation has not been considered.

We recently reported that good adherence to CPAP (≥4 h/night) for 4 weeks to 6 months was associated with an unexpected increase in circulating levels of angiopoietin-2 (Ang-2), a proangiogenic factor that amplifies endothelial inflammation and disturbs endothelial junctional integrity, suggesting a surprising, potentially detrimental effect of CPAP therapy on inflammatory milieu in OSA.^5,6 Elevated levels of Ang-2 are associated with increased risk for cardiovascular disease in the Framingham Heart Study and increased all-cause and cardiovascular mortality in the general population.^7,8 Circulating levels of Ang-2 are already significantly greater in otherwise healthy, untreated patients with OSA than in OSA-free controls and correlate with OSA severity and nocturnal hypoxia burden.⁹ The unexpected further increase in Ang-2 levels after CPAP therapy is concerning, particularly in patients with acute coronary syndrome, in whom greater Ang-2 level is associated with increased five-year mortality.¹⁰ Levels of Ang-2 decline over several months following an acute coronary event, a period that typically coincides with the initiation of cardioprotective pharmacotherapy, including statins.¹¹ Interestingly, addition of statin to CPAP therapy blunts the increase in circulating Ang-2 levels in otherwise healthy patients with OSA with good CPAP adherence by inhibiting its release from endothelial cell storage granules.^6,9

To investigate whether CPAP therapy prevents the natural decline in levels of Ang-2 after coronary revascularization in patients with OSA, we assessed circulating levels of Ang-2 before and after one year of randomly allocated CPAP therapy vs. usual care. In order to contextualize findings related to Ang-2, we simultaneously measured circulating levels of its receptor Tie-2, the endothelial inflammatory marker E-selectin, the angiogenic marker vascular endothelial growth factor (VEGF-A), and the soluble receptor for advanced glycation end products (sRAGE), a validated biomarker for pneumocyte type I injury.^12,13 Additionally, we assessed whether baseline and post-CPAP levels of these biomarkers of endothelial and lung epithelial function as well as CPAP levels are associated with cardiovascular morbidity and mortality in patients with OSA after coronary revascularization.

Methods

Study participants and design

This study is a secondary analysis using existing plasma samples collected in the Randomized Intervention with CPAP in CAD and OSA (RICCADSA) randomized clinical trial of CPAP therapy for OSA (NCT00519597), which was conducted in the Skaraborg County of West Götaland, Sweden between 2005 and 2010. Follow-up for the primary outcomes was completed in May 2013 and reported in 2016.² Consecutive patients with coronary artery disease (CAD) who underwent coronary artery bypass grafting or percutaneous coronary intervention within the previous six months were screened for OSA (Fig. 1). Patients with moderate to severe OSA (apnoea-hypopnea index [AHI] ≥ 15/h) without excessive daytime sleepiness (Epworth Sleepiness Scale scores <10) were randomly allocated to auto-titrating CPAP or usual care. OSA was diagnosed by home sleep apnoea testing (Embletta® Portable Digital System, Embla, Broomfield, CO, USA) and confirmed by fully attended nocturnal polysomnography as reported previously.² Criteria for the cardiovascular diagnosis were defined by the independent clinical event committee as described previously.² The reasons for participant refusal in the RICCADSA trial was included in the primary RICCADSA report.² Baseline anthropometrics, medical history of study population including hypertension and diabetes, the severity of CAD, angiographic findings, and type of revascularization procedure (percutaneous coronary intervention or coronary bypass grafting), medications and smoking status were obtained from combination of the patients’ self-report and physician diagnosis reported in patient medical records and national registers.

Fig. 1.

Flow diagram and study protocol.

Inclusion criteria

Newly diagnosed moderate to severe OSA (AHI ≥15 events/h of sleep) after percutaneous coronary intervention or coronary artery bypass grafting for angiography-verified coronary artery disease.

Exclusion criteria

Mild OSA (AHI 5–14.9/h of sleep), treated OSA, presence of central apnoea on HSAT, excessive daytime sleepiness. Patients with OSA with excessive daytime sleepiness (Epworth Sleepiness Scale score ≥10) could not be denied CPAP therapy owing to the societal risk of motor vehicle and industrial accidents; thus, these patients were excluded.

Adherence with CPAP therapy was ascertained as reported previously.² Clinical outcomes were determined by an Independent Clinical Event Committee, which reviewed data obtained from hospital records and death certificates, as previously described.² For patients who experienced more than one cardiovascular event during the follow-up period, only the first event was included in the composite outcome. Information was obtained from patients’ medical records, the Swedish Hospital Discharge Register and the Swedish National Cause of Death Registry.

Out of 189 patients, the proportion of observed events (composite cardiovascular outcome) was 28.6% (n = 54), which leaves the censoring rate of 71.4%. The proportion of observed all-cause mortality events was 4.2% (n = 8), which leaves the censoring rate of 95.8%. Those events were censored in May 2013, which marked the end of the study. Hence, censoring is independent.

This study was a secondary analysis using existing plasma samples; thus, sex and gender were not ascertained directly from participants for this secondary analysis. Sex was reported as male or female based on data obtained in the original study by the patients’ self-report.²

The primary outcome of this study was the change in circulating levels of Ang-2, Tie-2, sRAGE, E-selectin and VEGF-A from baseline to a 12-month follow up in CAD patients who were newly diagnosed with OSA after coronary revascularization and were randomized to CPAP vs. usual care. To assess natural trends in levels of these biomarkers after coronary revascularization, we also measured these markers in patients with CAD after coronary revascularization who were OSA-free (AHI <5/h). Secondary outcomes were the association of CPAP levels with biomarkers and a composite cardiovascular outcome (a composite of repeat revascularization, myocardial infarction, stroke, hospitalization for cardiovascular causes, and cardiovascular mortality), and the association of biomarkers with a composite cardiovascular outcome and all-cause mortality.

The study protocol was approved by the Ethics Committee of the Medical Faculty of the University of Gothenburg (approval no. 207-05; 09/13/2005; amendment T744-10; 11/26/2010; amendment T512-11; 06/16/2011), and written informed consent was obtained from all participants. The trial was registered (FoU i Sveriged Research and Development in Sweden; researchweb.org; VGSKAS-4731; 04/29/2005; and www.clinicaltrials.gov; NCT00519597).

Sleep apnea testing

The home sleep apnoea testing system consisted of a nasal pressure detector using a nasal cannula/pressure transducer system and two respiratory inductance plethysmography belts to record thoraco-abdominal movement and body position, and a finger pulse oximeter detecting oxyhemoglobin saturation (SpO₂) and heart rate. Apnea was defined as at least 90% cessation of airflow, and hypopnea as at least a 50% reduction in thoraco-abdominal movement and/or in nasal pressure amplitude for at least 10 s.² Events with a 30% reduction in either thoracoabdominal movement or nasal pressure amplitude for a minimum of 10 s were also scored as hypopneas if there was ≥4% oxygen desaturation. The oxygen desaturation index (ODI) was determined as the number of desaturations in SpO₂ of at least 4% from the immediately preceding baseline value per hour of sleep. Obstructive sleep apnoea (OSA) was defined as an apnoea-hypopnea index (AHI) of at least 15 events/h of sleep. OSA-free participants were defined as AHI <5 events/h of sleep.

All patients with CAD and a diagnosis of OSA based on the home sleep apnoea testing underwent fully attended overnight polysomnography (PSG) in hospital using a computerized recording system (Embla A10®, Embla, Broomfield, CO, USA). The PSG system included sleep monitoring through three-channel electroencephalography (EEG [C4/A1, C3/A2, CZ/A1]), two-channel electrooculography (EOG), one-channel submental electromyography (EMG), bilateral tibial EMG and two-lead electrocardiogram (ECG) in addition to the cardiorespiratory channels as described for the Embletta system above. PSG recordings were scored based on 30-s epochs according to the Rechtschaffen and Kales criteria by an observer blinded to clinical data and home sleep apnoea testing results. Obstructive events on the PSG were scored according to the same AASM criteria applied for the home sleep apnoea testing. CAD patients without OSA on home sleep apnoea testing did not undergo overnight PSG as reported previously.²

Blood collection and assessment of circulating biomarkers of endothelial and epithelial function

Venous blood samples were collected in a fasting state between 07:00 and 08:00 AM following the baseline sleep study and at a 12-month follow-up visit. Plasma samples were aliquoted and stored at −70 °C.

Levels of Ang-2, sRAGE, Tie-2, VEGF-A, and E-Selectin were measured in 30 μL of plasma diluted 1:2 using Human Magnetic Luminex® Assays (R&D Systems, catalog# LXSAHM) according to the manufacturer's instructions. The data were collected on a BD FACSAria III flow cytometer and analyzed by Flowjo. The mean fluorescence intensity of each pro-inflammatory biomarkers of the standards was used for calculating the standard curve for each biomarker using a log–log curve fit. Average Coefficient of Variance for Protein Standard across all five analytes (intra-plate measurements) was 6.0% and for Replicated Sample across all five analytes (inter-plate measurements) 11.3%.

There are no missing data for the primary analyses of the levels of biomarkers as only participants with available blood sample were included in this secondary analysis.

Statistical analysis

Two-sample t-tests (for continuous variables) and Chi-square tests (for categorical variables) were used to determine whether the participants’ baseline characteristics are comparable between the CPAP and usual care groups. The assumption of normality was tested using both Kolmogorov–Smirnov and Shapiro–Wilk and visually assessed by Q–Q plots. For the primary analysis comparing changes in levels of biomarkers from baseline to 12-month follow-up between groups, the Mann–Whitney U Test was used. The Wilcoxon signed-rank test was used to test within group changes in levels of biomarkers from baseline to 12-month follow-up. The association between CPAP pressure and circulating levels of biomarkers was assessed by linear regression.

For secondary analyses of the association between levels of biomarkers and mortality, time dependent Cox regression was used, where the outcome is the time at death or censoring at the latest available follow-up time for each participant, and the main exposures include the baseline levels of biomarkers and the change from baseline to 12-month follow-up. We tested for the proportional hazard assumption.¹⁴ For continuous predictors in the Cox models, we also assessed the adequacy of the linearity assumption, using penalized smoothing splines with the optimal degrees of freedom based on AIC criterion to model their potential non-linear associations and likelihood ratio test to determine the adequacy of the linear Cox models. The start times for the survival analysis are the enrollment times of individual participants. The first participant was enrolled on November 24, 2005. Age, sex, baseline BMI, hypertension, diabetes, and severity of OSA (baseline AHI) were included as controlling variables. Composite cardiovascular outcome was a recurrent event from baseline throughout the entire follow up period. Considering that non-fatal cardiovascular events occurring between baseline and 12 months may affect biomarker levels at the latter time point, a two-stage landmark analysis was used to assess the impact of levels of biomarkers on cardiovascular morbidity.¹⁵ In the first stage, a mixed effect Gamma frailty model for recurrent events was used to evaluate the effect of baseline biomarker level on cardiovascular morbidity prior to 12-month follow-up.¹⁶ We used Cox mixed effect model incorporating subject-specific random intercept. The analysis was conducted using R with coxme package (https://cran.r-project.org/web/packages/coxme/coxme.pdf). The random intercepts were included to summarize the individual cardiovascular risks prior to the landmark time. In the second stage, the same model was used to evaluate associations of biomarker levels at 12-month follow-up (the landmark time) with subsequent cardiovascular events while adjusting for the individual cardiovascular burden prior to 12-month follow-up (i.e., the random intercept from the stage 1 model) and other covariates. For analyses examining the effect of biomarkers on mortality and cardiovascular outcomes, we adjusted for covariates selected consistent with a modified disjunctive cause criterion,¹⁷ including baseline measures of age, sex, BMI, OSA severity, hypertension and diabetes mellitus, which are known causes of the exposure or outcome of interest and are not known to be instruments for the exposure or intermediates on the causal pathway.

Time dependent Cox regression analysis was used to assess the association between the use of CPAP pressure above vs. below the median and the composite cardiovascular outcome.

This study was a secondary analysis using existing plasma samples collected in the RICCADSA trial (NCT00519597); thus, sample size was determined by the available cohort rather than a prospective sample size calculation. The available sample provided an anticipated power of 80% to detect an effect size of 0.41 (Cohen's D) using a two-sample t-test at a 0.05 significance level for the primary analysis. Statistical analysis was performed using IBM SPSS Statistics for Windows, version 28.0 (IBM Corp., 2022) and R Studio Programming. All biomarkers were selected based on a priori hypothesis. Therefore, a 2-sided p < 0.05 was used to assess statistical significance. Testing of the major assumptions underlying each statistical test confirmed that these assumptions were not violated.

Role of the funding sources

None of the funding sources had any direct influence on the design of the study, the analysis of the data, the data collection, drafting of the manuscript, or the decision to publish.

Results

Patients with OSA and coexistent CAD were enrolled within 6 months of coronary revascularization and randomized to CPAP therapy (n = 91) or usual care (n = 98). OSA-free patients with CAD who underwent coronary revascularization were also analyzed to determine trends in biomarkers levels after revascularization (n = 91) (Fig. 1). Home sleep apnoea test was performed 62 ± 30 (mean ± SD) days and blood was collected 98 ± 35 days after coronary revascularization in the total sample (n = 270). The demographic and clinical characteristics of the study participants are shown in Table 1 and Supplemental Table E1. The CPAP group had more hypertension and tended to have more diabetes than the usual care group (Table 1). Time elapsed from coronary revascularization to blood collection was similar in CPAP and usual care groups (96 ± 34 vs. 99 ± 39 days, p = 0.54) as well as in the entire OSA cohort and OSA-free participants (98 ± 37 vs. 97 ± 31 days, p = 0.95). Median (1st-3rd quartiles) follow up was 57 (44–74) months. Cardioprotective pharmacotherapy utilization was similar in CPAP and usual care groups as well as in OSA-free participants, including statins (Table 1 and Supplemental Table E1). Median (1st-3rd quartiles) CPAP adherence was 3.5 (0.0–5.7) h/night in the CPAP group. Median (1st-3rd quartiles) residual AHI was 4.4 (3.3–7.1) in the CPAP group.

Table 1.

Demographic and clinical characteristics of Patients with OSA after coronary revascularization randomized to CPAP therapy vs. usual care.

	CPAP (n = 91)	Usual care (n = 98)	p
Demographic Characteristics
Age, y	65 (8.0)	67 (8.0)	0.24
Male sex, N (%)	75 (82)	84 (86)	0.53
Body Mass Index kg/m2	28.4 (3.6)	28.4 (3.3)	0.97
Epworth Sleepiness Scale	6.0 (2.2)	5.0 (2.0)	0.38
Current Smoking at baseline, N (%)	15 (17)	14 (14)	0.67
Comorbidities
Hypertension, N (%)	67 (74)	54 (55)	0.0080
Acute myocardial infarction, N (%)	54 (59)	47 (48)	0.12
Coronary bypass grafting, N (%)	24	28	0.59
Diabetes Mellitus, N (%)	28 (31)	18 (18)	0.047
Stroke, N (%)	8 (9)	9 (9)	0.91
Pulmonary Disease, N (%)	3 (3)	7 (7)	0.33
Medications
Lipid-lowering agent, N (%)	86 (98)	90 (93)	0.17
Diuretic, N (%)	23 (26)	25 (26)	0.96
β Blocker, N (%)	77 (88)	84 (87)	0.86
Clopidogrel, N (%)	53 (59)	49 (51)	0.25
Warfarin, N (%)	3 (3)	9 (9)	0.14
Calcium Channel Blocker, N (%)	18 (21)	16 (17)	0.49
Angiotensin Converting Enzyme, N (%)	37 (42)	49 (51)	0.25
Angiotensin II Receptor Blocker, N (%)	14 (16)	15 (16)	0.93
Selective Serotonin Reuptake Inhibitors, N (%)	4 (5)	2 (2)	0.42
Sleep Study
AHI, events/h	33.1 (21.3–61.9)	39.3 (24.9–56.9)	0.49
ODI4%, events/h	14.3 (7.7–26.5)	18.3 (7.9–30.8)	0.49
Time spent <90% SpO2,%	4.5 (0.8–23.9)	4.3 (0.8–16.7)	0.93
Nadir SpO2, %	85.0 (79.0–87.0)	84.0 (80.0–87.0)	0.89

Continuous data are presented as mean (SD) or median (1st-3rd quartiles). The Student's t-test or Mann–Whitney U test for the continuous variables, and χ² test or Fisher's exact test for categorical variables. CPAP = continuous positive airway pressure; AHI = apnoea hypopnea index; ODI4% = oxygen desaturation index, number of 4% desaturation events per hour; SpO₂ = arterial oxyhemoglobin saturation.

To assess the 12-month trends in levels of biomarkers of endothelial and epithelial function after coronary revascularization, we first measured these markers in patients with CAD after coronary revascularization who do not have OSA. In accordance with previous reports of short-term trends,¹¹ circulating levels of Ang-2 declined whereas VEGF-A remained elevated over 12 months follow-up (Table 2). Levels of E-selectin and Tie-2 also declined whereas levels of sRAGE were unchanged at 12-month follow up (Table 2).

Table 2.

Circulating levels of biomarkers of inflammation at baseline and after 12 months follow up in patients with coronary artery disease (CAD) after coronary revascularization with obstructive sleep apnoea (OSA: randomized to CPAP therapy vs. usual care) or without OSA (controls).

	CAD with OSA	CAD with OSA	pa	CAD without OSA	pb	pc
	CPAP (n = 91)	Usual care (n = 98)		Controls (n = 91)
Angiopoietin 2
Baseline	3.40 (2.51, 4.11)	3.52 (2.77, 4.21)	0.27	3.17 (2.51, 4.05)	0.66	0.10
Change: baseline to 12 months	0.05 (−0.65, 0.78)	−0.28 (−0.94, 0.19)	0.018	−0.33 (−1.12, 0.38)	0.0089	0.89
pd	0.52	0.0039		0.0028
RAGE
Baseline	2.94 (2.47, 3.74)	3.18 (2.61, 3.93)	0.17	3.04 (2.64, 3.70)	0.42	0.47
Change: baseline to 12 months	−0.01 (−0.53, 0.58)	−0.26 (−0.74, 0.47)	0.25	−0.16 (−0.69, 0.54)	0.42	0.72
pd	0.96	0.15		0.33
Tie 2
Baseline	17.54 (14.63, 21.26)	18.36 (15.22, 21.26)	0.26	18.30 (15.73, 22.98)	0.21	0.90
Change: baseline to 12 months	−1.10 (−4.85, 1.10)	−2.12 (−5.76, 0.54)	0.17	−1.89 (−5.04, 0.53)	0.28	0.70
pd	0.00076	<0.0001		<0.0001
VEGF
Baseline	26.90 (19.45, 35.51)	25.11 (18.92, 37.18)	0.69	24.31 (17.90, 34.98)	0.23	0.40
Change: baseline to 12 months	−3.40 (−11.68, 6.29)	0.25 (−6.59, 8.73)	0.047	2.55 (−6.79, 13.54)	0.0073	0.31
pd	0.019	0.67		0.10
E-Selectin
Baseline	35.40 (28.27, 43.09)	35.85 (24.72, 46.88)	0.97	35.03 (24.96, 43.10)	0.53	0.64
Change: baseline to 12 months	−2.07 (−6.61, 1.91)	−2.83 (−8.01, 0.33)	0.20	−1.41 (−6.94, 1.89)	0.76	0.29
pd	0.0063	<0.0001		0.0022

Continuous data are presented as median (1st-3rd quartiles); Mann–Whitney U test was used to test differences between CPAP and usual care groups, and between CPAP or usual care groups and controls; Wilcoxon Signed Rank Test was used to test differences between baseline and 12-month follow-up within groups.

^ap values for CPAP vs. usual care.

^bp values for CPAP vs. controls.

^cp values for usual care vs. controls.

^dp values for baseline vs. 12-month follow-up.

Effects of CPAP on circulating levels of biomarkers of endothelial and lung epithelial function in patients with OSA after coronary revascularization

We next investigated whether randomly allocated CPAP therapy affects levels of these biomarkers in patients with OSA with CAD after coronary revascularization. Baseline levels of Ang-2, Tie-2, VEGF-A, E-selectin and sRAGE were similar in patients with OSA randomized to CPAP or usual care (Table 2). After 12 months of CPAP therapy, circulating levels of Ang-2 failed to decline compared with baseline in the CPAP group (Table 2). In contrast, levels of Ang-2 declined in the usual care group (Table 2). The change in Ang-2 levels from baseline to 12 months was significantly different between CPAP and usual care group (median difference = −0.32, 95% CI −0.57, −0.06, p = 0.018; Table 2). After adjustment for age, sex, BMI, hypertension, diabetes, and the severity of OSA (baseline AHI), the change in Ang-2 levels from baseline to 12 months remained significantly different between CPAP and usual care group (p = 0.039). Levels of sRAGE, Tie-2 and E-selectin also declined more with usual care than with CPAP (Table 2); however, these changes did not significantly differ between groups. Similar to OSA-free participants, levels of VEGF-A remained elevated in the usual care group (Table 2). In contrast, CPAP therapy reduced levels of VEGF-A in patients with OSA (Table 2). The change in VEGF-A levels from baseline to 12 months was significantly different between CPAP and usual care group (median difference = 3.65, 95% CI 0.62, 5.68, p = 0.047; Table 2).

Association of circulating levels of biomarkers of endothelial and epithelial function with CPAP level and adherence

Greater post-CPAP levels of Ang-2 were associated with greater median CPAP pressure (p = 0.039 after adjustment for baseline Ang-2 levels; Fig. 2) but not with hours per night of CPAP adherence (p = 0.34). Levels of Tie-2, E-selectin, VEGF-A and sRAGE after 12 months of randomly allocated CPAP therapy were not significantly associated with median CPAP level (p = 0.14, p = 0.91, p = 0.24, p = 0.13, respectively) or adherence (p = 0.32, p = 0.57, p = 0.76, p = 0.31, respectively). Median CPAP level did not correlate with severity of OSA as assessed by AHI (r = 0.03, p = 0.80).

Fig. 2.

Circulating levels of angiopoietin-2 are associated with median CPAP pressure in patients with obstructive sleep apnoea after coronary revascularization (linear regression slope 0.30, 95% CI 0.03, 0.57, p = 0.039). The shaded area represents 95% confidence band. CPAP, continuous positive airway pressure.

Association of biomarkers of endothelial and epithelial function with cardiovascular morbidity and mortality in patients with OSA after coronary revascularization

All tests for non-proportional hazard ratios and non-linearity assumptions in the Cox models are insignificant. Based on the linear Cox regression, increase in level of Ang-2 from baseline to 12 months was associated with greater mortality in patients with OSA after coronary revascularization (HR 1.45 per ng/ml increase in Ang-2; 95% CI 1.02, 2.07, p = 0.039; adjusted for age, sex, BMI, hypertension, diabetes, severity of OSA [baseline AHI], group assignment and baseline Ang-2 levels).

Change in levels of sRAGE, Tie-2, VEGF-A and E-selectin from baseline to 12-months were not significantly associated with mortality in patients with OSA (p = 0.10, 0.66, 0.47 and 0.96, respectively).

Landmark analysis showed that the 12-month levels of endothelial markers Ang-2, VEGF-A, Tie-2 and E-selectin were not significantly associated with the composite cardiovascular outcome after a 12-month follow up in patients with OSA who underwent coronary revascularization (p = 0.92, 0.11, 0.85 and 0.97, respectively). However, the 12-month levels of sRAGE, a marker of lung epithelial injury, were associated with greater incidence of the composite cardiovascular outcome (HR 1.27 per ng/ml increase in sRAGE; 95% CI 1.04, 1.55, p = 0.019; adjusted for age, sex, BMI, hypertension, diabetes, severity of OSA [baseline AHI], group assignment and baseline sRAGE levels).

To assess whether levels of CPAP are associated with cardiovascular morbidity in patients with OSA after coronary revascularization, we performed Cox regression analysis for the composite cardiovascular outcome stratified by the CPAP levels below vs. above the median pressure. The median CPAP level was 7 cm H2O in patients with OSA randomized to CPAP. The use of CPAP levels >7 cm H2O was associated with greater incidence of the composite cardiovascular outcome compared with the use of CPAP levels ≤7 cm H2O (HR 2.81; 95% CI 1.22, 6.44, p = 0.021; adjusted for age, sex, BMI, and severity of OSA [baseline AHI]; Fig. 3).

Fig. 3.

The association of the composite cardiovascular outcome with CPAP pressure. Cox regression analysis for the composite cardiovascular outcome stratified by the CPAP pressure below vs. above the median pressure, adjusted for age, sex, BMI and baseline AHI (HR 2.81; 95% CI 1.22, 6.44, p = 0.021). The shaded areas represent 95% confidence bands. CPAP, continuous positive airway pressure.

Discussion

The major findings of this study are that randomly allocated CPAP therapy alters the pattern of plasma biomarker expression after coronary revascularization in patients with OSA in a direction that favors inflammatory over angiogenic processes (Fig. 4). Compared with usual care, patients with OSA randomized to CPAP therapy for 12 months had persistently elevated circulating levels of Ang-2, a pro-inflammatory factor, and a decline in levels of VEGF-A, a proangiogenic factor that is elevated after coronary revascularization and whose levels correlate inversely with severity of CAD.^11,18 Post-CPAP Ang-2 levels correlated directly with median CPAP levels. These findings are in accordance with two recent reports that patients with OSA adherent to CPAP therapy for 4 weeks to 6 months in 3 independent cohorts had an unexpected CPAP-induced increase in circulating levels of Ang-2,^5,6 findings which prompted the current study. Although Tie2, sRAGE, and E-selectin also fell more in the usual care than the CPAP group, these differences were not statistically significant. Greater 12-month levels of Ang-2 and sRAGE, a marker of lung epithelial injury, were associated with significantly greater mortality and an increased incidence of cardiovascular events, respectively. The use of CPAP levels >7 cm H2O was associated with worse cardiovascular outcomes compared with CPAP pressures ≤7 cm H₂O.

Fig. 4.

CPAP therapy alters the pattern of plasma biomarker expression after coronary revascularization in patients with obstructive sleep apnoea (OSA) in a direction that favors inflammatory over angiogenic processes. While CPAP therapy reduces OSA-associated hypoxemia, with expected beneficial effects on endothelial function and inflammation, increased lung volumes during CPAP therapy may promote persistent increase in circulating levels of a pro-inflammatory factor angiopoietin-2 while elimination of intermittent hypoxia reduces levels of a proangiogenic factor VEGF-A. The reduction in VEGF-A in the setting of persistent elevation in angiopoietin-2 observed after CPAP therapy is hypothesized to increase the risk of recurrent events after coronary revascularization. Greater levels of angiopoietin-2 and sRAGE, a marker of lung epithelial injury, after 12 months of CPAP therapy were associated with significantly greater mortality and an increased incidence of cardiovascular events in patients with OSA. CPAP, continuous positive airway pressure; VEGF-A, vascular endothelial growth factor; sRAGE, soluble receptor of advanced glycation end-products.

These findings raise the possibility that, despite reducing nocturnal hypoxia burden and sleep fragmentation, CPAP, in particular greater CPAP levels, may perpetuate rather than ameliorate inflammatory processes in OSA. CPAP is known to substantially increase lung volume in a dose-dependent fashion, with 10 cm H₂O pressure increasing functional residual capacity in healthy adults by more than one liter.¹⁹ By increasing lung volume and alveolar distension, CPAP may contribute to lung endothelial and epithelial inflammation in a manner analogous to that seen with volutrauma in ventilator-induced lung injury.20, 21, 22 Although no prior data are available on the impact of lung inflation on Ang-2 levels in humans or animal models outside the context of acute lung injury, when human umbilical vein endothelial cells cultured on a flexible membrane are stretched to 20% elongation, secretion of Ang-2 rapidly increases 2.7-fold over a period of 6 h.²³ Whether lung endothelial cells are the predominant source of circulating Ang-2 after CPAP therapy cannot be determined; however, lung endothelium comprises a third of the body endothelial cell surface and, therefore, could be a significant source of circulating Ang-2 after alveolar distension. Upon release from pulmonary endothelial cells, Ang-2 activates Rho kinase causing disaggregation of cell–cell junctions and potentiation of the inflammatory nuclear factor κ-B pathway while silencing protective phosphoinositide-3 kinase signaling vital to cell survival.23, 24, 25 The net result is capillary leak, neutrophil transmigration and angiogenesis, which may worsen inflammatory processes in OSA.24, 25, 26

The reason why Ang-2 levels failed to decline in our CPAP-treated OSA group rather than increasing, as observed in previous reports, is likely owing to the natural decline in Ang-2 from the peak at the time of coronary revascularization.¹¹ Indeed, the observed 0.33 ng/ml difference in change over 12 months between CPAP and usual care groups in this study is similar to the increase in Ang-2 observed with CPAP therapy in our previously studied cohorts,^5,6 and is of a magnitude that in a community-based cohort was associated with an approximately 15% increase in risk of cardiovascular and all-cause mortality, an effect that could plausibly neutralize the expected cardiovascular benefit of controlling OSA.⁸ However, the potential effect of CPAP on Ang-2 may be underestimated owing to a high utilization of statins in this cohort after coronary revascularization (98 and 93% in CPAP and usual care group, respectively), which contrasts with a low statin utilization in the general population with OSA.²⁷ We have shown that randomly allocated atorvastatin 10 mg daily vs. placebo for 4 weeks reverses the increase in Ang-2 levels seen in otherwise healthy patients with OSA with good CPAP adherence.⁶ Intermittent hypoxia from OSA increases abundance of free cholesterol in the endothelial cell plasma membrane, which promotes redistribution of CD59, a major complement inhibitor, from the endothelial cell surface to the cell interior.⁹ Internalized, intracellular CD59 interacts with the exocytosis machinery of endothelial cell Weibel-Palade bodies that store Ang-2 and promotes complement-dependent release of Ang-2 in patients with OSA.⁹ By lowering cholesterol, statins stabilize CD59 on the endothelial cell surface, which restores endothelial complement protection and reduces Ang-2 release.^6,9

Plasma levels of sRAGE, a marker of lung epithelial injury that amplifies inflammatory responses despite acting as a “decoy receptor” for its ligands, are increased in patients with ventilator-induced lung injury and are associated with increased mortality.¹² In mechanically ventilated patient with acute lung injury, sRAGE is more abundant in the airspaces and correlates only modestly with plasma level.¹³ This may explain why in the present study, although circulating levels of sRAGE tended to decline in the usual care group whereas they were unchanged in the CPAP group, these differences were not statistically significant. As increased levels of sRAGE were associated with worse cardiovascular outcomes in patients with OSA after coronary revascularization, the possibility that alveolar distension in the setting of CPAP therapy may promote a low-level lung epithelial injury manifesting in higher levels of RAGE in the lung milieu merits further investigation.

Interactions between Ang-2 and its receptor Tie-2 have dual effects on angiogenesis, depending on the availability of VEGF-A.²⁸ In the presence of VEGF-A, Ang-2 acts as an agonist and stimulates angiogenesis. In the absence of VEGF-A, Ang-2 competitively antagonizes angiopoietin-1-induced Tie-2 phosphorylation resulting in vessel regression. Thus, Ang-2 and VEGF-A act synergistically to produce a stable and functional microvasculature. In this study, CPAP therapy prevented the natural decline in levels of Ang-2 while significantly reducing levels of VEGF-A in patients with OSA after coronary revascularization. The reduction in VEGF-A levels after CPAP therapy in OSA has been reported previously and is likely owing to the CPAP-mediated elimination of hypoxia burden, which is a major stimulus for VEGF-A synthesis.^29,30 While this has generally been interpreted as a beneficial effect of CPAP, there is now considerable evidence that higher levels of VEGF-A may be cardioprotective. A VEGF polymorphism associated with greater VEGF expression is associated with a lower risk of CAD.³¹ Similar to our findings, plasma levels of VEGF-A remained elevated in patients with CAD 20 months after myocardial infarction compared to controls,³² and may exert protective effects on the arterial endothelium.³³ At low concentrations, VEGF-A exerts a weak angiogenic and atheroprotective effect,³³ which may be of particular importance after coronary revascularization.³⁴ Studies in animal models suggesting that VEGF-A destabilizes coronary plaques by promoting plaque angiogenesis have been challenged owing to temporal inconsistencies of VEGF-A effects and development of features of plaque instability.³³ In fact, the VEGF-A inhibitor bevacizumab increases cardiovascular risk compared with placebo in cancer patients.³⁵ Thus, the reduction in VEGF-A in the setting of persistent elevation in Ang-2 observed in the CPAP-treated group could potentially increase the risk of recurrent events after coronary revascularization.²⁸

E-selectin is an endothelial-leukocyte adhesion molecule that derives exclusively from activated endothelium and is therefore a sensitive and specific marker of endothelial activation.³⁶ Hypoxemia acts synergistically with inflammatory mediators including tumor necrosis factor-α to induce endothelial expression of E-selectin,³⁷ which likely explains the known association of circulating E-selectin with OSA. Indeed, in a scan of 5000 plasma proteins analyzed in almost 1400 sleep clinic patients from the Stanford Technology Analytics and Genomics in Sleep (STAGES) study, E-selectin was the protein most strongly correlated with the AHI.³⁸ Moreover, E-selectin is associated with increased cardiovascular risk in patients with moderate to severe OSA.³⁹ While a reduction in E-selectin with CPAP therapy has been reported, this was in uncontrolled observational studies. In the present study, E-selectin fell less in the CPAP-treated group than in the usual care group, although this difference was not statistically significant.

While CPAP is very effective for treatment of daytime symptoms of OSA such as excessive sleepiness and fatigue, and the associated increased risk of motor vehicle and industrial accidents and inability to concentrate, we found that greater CPAP levels were associated with worse cardiovascular outcomes. Importantly, median CPAP level did not correlate with severity of OSA and adjustment for OSA severity, major comorbidities and adiposity did not affect this association. In addition, adherence with CPAP therapy (hours per night) was not significantly associated with cardiovascular outcomes, suggesting that pressure levels and not the hours of usage may drive this association. Use of more conservative CPAP levels that control daytime symptoms, even if they do not eliminate all apnoeas and hypopneas, is a therapeutic approach that should be considered to minimize potential side effects from lung hyperinflation.

Limitations of this study include a possibility of unmeasured confounding (although the randomized design minimizes the chance of unmeasured confounding for the primary analysis), a built-in selection bias in hazard ratios,⁴⁰ and insufficient sample size for precise analyses of the effects of CPAP levels on separate components of the composite cardiovascular outcome.

In conclusion, CPAP changes the trajectory of biomarkers of endothelial and epithelial injury after coronary revascularization in a pattern that is associated with worse cardiovascular outcomes. The possibility that CPAP therapy may exacerbate inflammatory processes in OSA warrants further investigation in both human studies and animal models, particularly as CPAP therapy has unexpectedly failed to reduce cardiovascular risk in patients with OSA in intention-to-treat analysis of recent randomized clinical trials.4, 5, 6

Contributors

Y.P. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Y.C. also had full access to all the data in the study and verified the underlying data. Y.P. recruited study participants and analyzed and discussed data, Y.C., analyzed and discussed data, A.B. performed measurements and discussed data, S.R. analyzed and discussed data, Y.C., J.L. and Y.W. designed statistical plan and performed statistical analysis, D.J.G. and S.J. conceived, designed and managed the study, and wrote the manuscript, which was revised and approved by all authors.

Data sharing statement

Data collected for the study, including deidentified individual participant data will be made available to others within 6 months after the publication of this article, as will additional related documents (study protocol, statistical analysis plan, and informed consent form), for academic purposes (e.g., meta-analyses), upon request to the corresponding authors, and with a signed data access agreement.

Declaration of interests

Y.P. received institutional grants from ResMed Foundation and ResMed Ltd. S.R. received consulting fees from Apnimed Inc., and consulting fees and support for travel from Eli Lilly Inc. S.R. is an unpaid Board member for the National Sleep Foundation and Alliance for Sleep Apnea Partners. D.J.G. received consulting fees from Powell–Mansfield, Inc., and fees for participation on Scientific Advisory Board from Signifier Medical Technologies and Wesper, Inc., and fees for participation on Data Monitoring Committee from Apnimed, Inc. All other authors declare no competing interests.