Prognostic value of sleep apnea and nocturnal hypoxemia in patients with decompensated heart failure

Abstract Background Nocturnal hypoxemia is an important factor underlying the impact of sleep apnea on heart failure. It remains unclear whether nocturnal hypoxemia has a greater prognostic value in acute decompensated heart failure (ADHF) compared with the frequency of sleep apnea. Hypothesis Nocturnal hypoxemia might be better than the frequency of sleep apnea in predicting the outcomes in ADHF. Methods Sleep studies were prospectively performed during an ADHF hospitalization from January 2015 to December 2017. Sleep apnea was defined as the apnea‐hypopnea index (AHI) ≥15/h. The severity of nocturnal hypoxemia was determined by the percentage of time with saturation below 90% (T90%). The endpoint was the first event of all‐cause death, heart transplantation, implantation of left ventricular assist device, unplanned hospitalization for worsening heart failure, acute coronary syndrome, significant arrhythmias, or stroke. Results Of 382 patients, 189 (49.5%) had sleep apnea. The endpoint incidence did not differ between AHI categories (≥15/h vs <15/h: 52.4% vs 44.6%, log rank P = .353), but did between T90% categories (≥3.6% vs <3.6%: 54.5% vs 42.4%, log rank P = .023). Multivariate Cox regression analysis showed that T90% was independently associated with the endpoint (hazard ratio [HR] 1.008, 95% confidence interval [CI] 1.001‐1.016, P = .033), whereas AHI was not; the risk of the endpoint increased by 40.8% in patients with T90% ≥3.6% (HR 1.408, 95%CI 1.030‐1.925, P = .032). Conclusion Nocturnal hypoxemia had a greater prognostic value in ADHF than the frequency of sleep apnea.


| INTRODUCTION
Sleep apnea, typically categorized as predominantly obstructive (OSA) or central (CSA), is highly prevalent in both acute decompensated heart failure (ADHF) 1,2 and chronic stable heart failure. 3,4 Sleep apnea is responsible for multiple cardiovascular pathophysiological changes in heart failure, such as myocardial ischemia, 5 increased pulmonary arterial pressure, 6 and abnormal cardiac electrophysiological activities, 7,8 based on complex mechanisms, including nocturnal hypoxemia, increased sympathetic activity, enhanced renin-angiotensinaldosterone system, and chronic inflammation. 9,10 It has been reported that sleep apnea, generally scored by the apnea-hypopnea index (AHI), might be an independent risk factor of adverse outcomes in heart failure. 1,2,11 However, AHI has been questioned as a prognostic predictor of heart failure in some studies. 12,13 AHI is only a metric reflecting the frequency of apneas and hypopneas during sleep and does not take the lengths of apneas and hypopneas into consideration on its own definition. Therefore, more importance should be attached to detailed characteristics of sleep apnea.
Nocturnal hypoxemia, as a composite consequence of apneas and hypopneas, might better represent the adverse effects of nocturnal respiratory events in heart failure. Gottlieb et al reported that increased hemodynamic stress in heart failure was related to the percentage of time with saturation below 90% (T90%), but not to the AHI. 14 Evidence also suggested that nocturnal hypoxemia appeared to be more robust to predict outcomes in stable chronic heart failure compared with AHI. 11,15 However, it is unclear whether nocturnal hypoxemia is better than AHI in predicting the outcomes in ADHF. Therefore, in the present study, we aimed to compare AHI and several parameters of nocturnal hypoxemia in evaluating the prognosis in hospitalized heart failure patients.

| Patients
This single-center, prospective, observational study was performed in Heart Failure Center, Fuwai Hospital. From January 2015 to December 2017, patients with ADHF were consecutively enrolled, including both new-onset heart failure and decompensation of chronic heart failure. ADHF was diagnosed based on symptoms/signs of fluid overload and/or hypoperfusion, and appropriate additional investigations such as chest X-ray, electrocardiogram, N-terminal pro-brain natriuretic peptide (NT-proBNP), and echocardiography according to the European Society of Cardiology Guidelines. 16 The exclusive criteria were as follows: age <18 or >80 years; any coronary event within the previous 3 months or at the time of enrollment, namely, acute coronary syndrome (ACS), percutaneous coronary invention, coronary artery bypass grafts; implantation of pacemaker, implantable cardioverter defibrillation (ICD), or cardiac resynchronization therapy within the previous 3 months; heart valvular surgeries within the previous 3 months; stroke within the previous 6 months; dialysis; chronic obstructive pulmonary disease; acute myocarditis or infective endocarditis; significant uncorrected valvular heart disease; malignancy; pregnancy; diagnosed sleep apnea, or previously receiving any type of positive pressure ventilation or oxygen therapy. Patients were also excluded if they were admitted to hospital for cardiovascular interventions and surgeries. The study protocol conformed to the Declaration of Helsinki and was approved by the institutional review board of Fuwai Hospital. Individual informed consents were signed.

| Sleep study
Patients received sleep studies by means of Apnealink Plus (Resmed, Ltd, Martinsried, Germany) from 22:00 to 6:00 after an initial improvement of heart failure during the hospitalization period by intensive therapy. Patients undergoing sleep study were requested to relieve from edema and lie in a supine position without dyspnea under room air.
Sleep studies were not done on patients who were hemodynamically unstable, had nocturnal dyspnea, needed oxygen supplement or ventilation. Nasal airflow amplitude and oxygen saturation were measured by a nasal flow pressure cannula and a finger pulse oximeter, respectively.
The recorded data were analyzed by two-step method. First, the data were analyzed automatically by software, Apnealink Version 10.20. Then the recordings were manually reanalyzed by a sleep specialist who was blinded to the clinical status of patients. In recordings, only time periods with both sufficient airflow and saturation signals were considered valid recording time. We only took account those sleep studies with a minimum 4-hour valid recording time. Apnea was defined as breathing amplitude decreased by ≥90% for ≥10 seconds. Hypopnea was defined as breathing amplitude decreased by ≥30% lasting for ≥10 seconds, accompanied by a ≥3% drop in oxygen saturation. 17 AHI was defined as the total number of apneas and hypopneas per hour. Sleep apnea was defined as AHI ≥15/h. Oxygen desaturation index (ODI) was defined as the total number of desaturation events where oxygen saturation decreased by ≥3% per hour. The mean saturation (meanSO 2 ), the minimal saturation (minSO 2 ), and T90% during sleep were also recorded.

| Blood samples and echocardiography
Blood samples were routinely collected for every patient. We examined a series of blood parameters, including NT-proBNP, hemoglobin, serum creatinine (SCr), blood urea nitrogen, potassium, sodium, glycated hemoglobin, total cholesterol, and low-density lipoprotein cholesterol. The renal function was evaluated by eGFR (mL/min/1.73 m 2 ) based on SCr using modification of diet in renal disease (MDRD) equation. Renal dysfunction was defined as eGFR <60 mL/min/1.73 m 2 . Echocardiography was performed using ultrasound system (Vivid E9; GE, Norway) on admission.

| Follow-up and endpoint
The enrolled patients were systematically followed up every 3 months by outpatient reviews or telephone calls after discharge until December 31, 2018. Follow-up was terminated when death, heart transplantation, or implantation of left ventricular assist device (LVAD) occurred. The endpoint was defined as the first event of death from any cause, heart transplantation, LVAD implantation, unplanned hospitalization for worsening heart failure, ACS, significant arrhythmias, and stroke. Significant arrhythmia event was defined as sustained ventricular tachycardia, ventricular fibrillation of asystole. Information of the adverse events was obtained from the medical records for those patients who were followed up at our hospital. For those patients who were not followed up at our hospital, detailed information was obtained by telephone calls with patients' families and local medical institutions they were admitted to if necessary. Data regarding the adverse events were collected and determined by two blinded cardiologists.

| Statistical analysis
Continuous variables were presented as mean ± SD or median with interquartile range (IQR) as appropriate, while categorical variables were expressed as frequency and percentage. Baseline characteristics were compared with Student's t test or Mann-Whitney U test for continuous variables, and chi-square test or Fisher's exact test for categorical variables. The impact of each sleep study parameters on the time to the endpoint was assessed by Kaplan-Meier analysis using logrank test. The thresholds of sleep study parameters were determined by the median values except for AHI. Factors associated with the endpoint were determined using univariate Cox regression analysis, including age, gender, BMI, coronary artery disease, hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, renal dysfunction, NYHA class, mean arterial blood pressure (MAP) at discharge, NT-proBNP, LVEF, medications prescribed at discharge (ie, angiotensin converting enzyme inhibitor [ACEI] /angiotensin receptor blocker [ARB], β-blocker, spironolactone, calcium channel blocker, and statin) and sleep study parameters. Variables with P < .10 in univariate analysis were included in a multivariate Cox regression analysis to identify the independent risk factors of the endpoint based on stepwise backward selection using a likelihood ratio (P > .5 for exclusion). Sleep study parameters were included in multivariate analysis irrespective of their significance in univariate analysis. Because of potential correlation between sleep study parameters, each tested parameter was analyzed separately in multivariate analysis. Hazard ratios (HR) and 95% confidence intervals (CI) were calculated. A two-tailed P < .05 was considered statistically significant. All data were analyzed using SPSS version 23.0 (IBM corporation, Armonk, New York).

| RESULTS
A total of 420 patients who met the predefined inclusion/exclusion criteria were followed up systematically after discharge. Follow-up Abbreviations: BMI, body mass index; CI, confidence interval; HR, hazard ratio; MAP, mean atrial blood pressure; NT-proBNP, N-terminal pro-brain natriuretic peptide; NYHA, New York Heart Association; T90%, the percentage of time with oxygen saturation below 90%. more diuretics. There were no differences of AHI and AHI categories between patients with and without the endpoint. T90% was significantly higher in patients with the endpoint ( Table 1).
The Kaplan-Meier analysis showed the incidence of the endpoint did not differ between AHI categories (≥15/h vs <15/h: 52.4% vs 44.6%, χ 2 = 0.862, log rank P = .353; Figure 2A)  AHI might not be the best metric to determine the severity of sleep apnea in heart failure. As mainly reflecting the frequency of apneas and hypopneas during sleep on its own definition, AHI does not consider the durations of apneas and hypopneas. As a consequence, AHI cannot differentiate between apneas and hypopneas with the same number but different durations. Moreover, the lengths of apneas and hypopneas are dependent on cardiac function. 22 The greater the extent of cardiac dysfunction, the longer apneas and hypopneas will be. As a result, the total number of apneas and hypopneas is potentially limited in heart failure and the severity of sleep apnea, determined by AHI, is consequently underestimated. Abbreviations: ACEI, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker; BMI, body mass index; CI, confidence interval; HR, hazard raito; MAP, mean atrial blood pressure; MinSO 2 , the minimal oxygen saturation; NT-proBNP, N-terminal pro-brain natriuretic peptide; NYHA, New York Heart Association.
The result of our study demonstrated that some parameters representative of nocturnal hypoxemia (ie, T90% and minSO 2 ) were better than AHI in predicting adverse outcomes in ADHF, which was consistent with some previous studies in stable chronic heart failure.
Oldenburg et al 11  showed significant association with the prognosis. 15 In addition, minSO 2 was also demonstrated a more robust association with fatal or resuscitated sudden cardiac death than AHI. 23 These findings suggested that nocturnal hypoxemia might better represent the detrimental effects of sleep apnea than the frequency of apneas and hypopneas in heart failure. It was reported that T90%, rather than AHI, predicted the elevations in brain natriuretic peptide, 14 indicating nocturnal hypoxemia might be an important factor underlying the impact of sleep apnea on acute hemodynamic stress in heart failure. Another study found that cardiac norepinephrine spillover was correlated with a reduced oxygen saturation, but not with the AHI, 24 indicating increased sympathetic activity was more associated with nocturnal hypoxemia. Overall, nocturnal hypoxemia might be a better measure representative of adverse effects of sleep apnea than AHI, explaining why it was better than the frequency of apneas and hypopneas in predicting the prognosis in ADHF.  34 A randomized controlled trial has been registered to assess the efficacy of an oral appliance for sleep-disordered breathing and cardiac function in patients with heart failure. 35

| CONCLUSION
This study demonstrated that nocturnal hypoxemia was more predictive of adverse outcomes in decompensated heart failure than the frequency of sleep apnea. Prospective studies should be conducted to determine the effect of oxygen therapy on the prognosis of heart failure and sleep apnea.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.
How to cite this article: Huang Y, Wang Y, Huang Y, et al.