Author + information
- Received April 17, 2017
- Revision received July 7, 2017
- Accepted July 9, 2017
- Published online September 4, 2017.
- Michele Emdin, MD, PhDa,b,∗ ( )(, )
- Gianluca Mirizzi, MDa,b,
- Alberto Giannoni, MD, PhDb,
- Roberta Poletti, MDb,
- Giovanni Iudice, BScb,
- Francesca Bramanti, BScb and
- Claudio Passino, MDa,b
- aInstitute of Life Sciences, Scuola Superiore Sant’Anna, Pisa, Italy
- bFondazione Toscana Gabriele Monasterio, Pisa, Italy
- ↵∗Address for correspondence:
Dr. Michele Emdin, Cardiology and Cardiovascular Medicine Division, Fondazione G. Monasterio CNR, Regione Toscana, Via Giuseppe Moruzzi 1, 56124 Pisa, Italy.
Background Large trials using noninvasive mechanical ventilation to treat central apnea (CA) occurring at night (“sleep apnea”) in patients with systolic heart failure (HF) have failed to improve prognosis. The prevalence and prognostic value of CA during daytime and over an entire 24-h period are not well described.
Objectives This study evaluated the occurrence and prognostic significance of nighttime, daytime, and 24-h CA episodes in a large cohort of patients with systolic HF.
Methods Consecutive patients receiving guideline-recommended treatment for HF (n = 525; left ventricular ejection fraction [LVEF] of 33 ± 9%; 66 ± 12 years of age; 77% males) underwent prospective evaluation, including 24-h respiratory recording, and were followed-up using cardiac mortality as an endpoint.
Results The 24-h prevalence of predominant CAs (apnea/hypopnea index [AHI] ≥5 events/h, with CA of >50%) was 64.8% (nighttime: 69.1%; daytime: 57.0%), whereas the prevalence of predominant obstructive apneas (OA) was 12.8% (AHI ≥5 events/h with OAs >50%; nighttime: 14.7%; daytime: 5.9%). Episodes of CA were associated with neurohormonal activation, ventricular arrhythmic burden, and systolic/diastolic dysfunction (all p < 0.05). During a median 34-month follow-up (interquartile range [IQR]: 17 to 36 months), 50 cardiac deaths occurred. Nighttime, daytime, and 24-h moderate-to-severe CAs were associated with increased cardiac mortality (AHI of </≥15 events/h; log-rank: 6.6, 8.7, and 5.3, respectively; all p < 0.05; central apnea index [CAI] of </≥10 events/h; log-rank 8.9, 11.2, and 10.9, respectively; all p < 0.001). Age, B-type natriuretic peptide level, renal dysfunction, 24-h AHI, CAI, and time with oxygen saturation of <90% were independent predictors of outcome.
Conclusions In systolic HF patients, CAs occurred throughout a 24-h period and were associated with a neurohormonal activation, ventricular arrhythmic burden, and worse prognosis.
Two centuries ago, a periodic breathing pattern was described in patients with heart failure (HF) by the Irish physicians Cheyne (1) and Stokes (2). Cheyne-Stokes respiration (CSR) (3) is a rhythmic waxing and waning of respiration, with alternating periods of central apnea (CA) episodes and deep, rapid breathing (3). Increased central/peripheral chemosensitivity and circulatory time delays are pathophysiological triggers (3). Notably, both Cheyne and Stokes described patients who were awake. Episodes of CA were thought to occur only during sleep at nighttime and were referred to as “central sleep apnea” (CSA) (4), analogous to episodes of obstructive sleep apnea (OSA). Polysomnographic studies show that CSA actually predominates over OSA during the night in patients with systolic HF (5–7). Central sleep apnea is associated with adrenergic activation and life-threatening arrhythmia and has independent prognostic value in most studies (5–7).
Treatments for CA in HF have thus been applied primarily during the night (8–12), with disappointing results. Indeed, the CANPAP (Canadian Continuous Positive Airway Pressure [C-PAP] for Patients with Central Sleep Apnea and Heart Failure) and SERVE-HF (Treatment of Sleep-Disordered Breathing with Predominant Central Sleep Apnea by Adaptive Servo Ventilation in Patients with Heart Failure) trials failed to show survival benefit and showed poor compliance (13,14). These negative results have been interpreted as evidence supporting a compensatory, beneficial role of CSR in HF (15,16), suggesting that this phenomenon should not be treated (17). Alternatively, the failure of noninvasive mechanical ventilation may be because it targets CA only during sleep, with no effect on or even a potential rebound in daytime apnea (18).
Over the last 2 decades, short-term (20-min) polygraphic recordings in up to 500 HF patients have demonstrated that CAs are present even under waking conditions (13,19) and are associated with detrimental outcomes in HF (13,19). One study performed with a 24-h polygraphic recordings (14) in 62 patients demonstrated that CSR occurred in 62% of patients at night and 16% of patients during the day, using an apnea/hypopnea index (AHI) cutpoint of >15 events/h.
We hypothesized that sleep studies alone may not adequately characterize CA, and we prospectively tested the presence, time course, and severity of CAs throughout a 24-h period in a large cohort of consecutive patients with systolic HF to define the clinical relevance and prognostic value of daytime, nighttime, and 24-h CA burden and compare it to those of OSA.
From January 2006 to December 2013, 525 consecutive patients with systolic HF (stage C [American College of Cardiology/American Heart Association criteria]) and echocardiographic evidence of impaired left ventricular systolic function (left ventricular ejection fraction [LVEF] of <50%) receiving stable (≥3 months) guideline-recommended treatment were prospectively enrolled. Exclusion criteria were severe pulmonary or neurological disease; thyroid dysfunction; or concurrent therapy with morphine or derivatives, theophylline, oxygen, benzodiazepines, acetazolamide, continuous positive airway pressure, or servoventilation.
All patients underwent 2-dimensional transthoracic echocardiogram examinations (model IE33 ultrasound machine with X5-1 transducer; Philips Medical Systems, Palo Alto, California) (20,21), 24-h electrocardiography Holter recording (Elamedical, Paris, France), symptom-limited cardiopulmonary exercise testing (VMAX, Sensormedics, Conshohocken, Pennsylvania), 24-h cardiorespiratory monitoring (see below), and biohumoral characterization, including plasma catecholamines, aldosterone, renin activity, and N-terminal pro–B-type natriuretic peptide (NT-proBNP) levels (22). All patients also underwent 24-h cardiorespiratory polygraphic recording for screening of CA occurrence during daytime, nighttime, and the entire 24-h period. All examinations were performed within a 3-day period.
All patients gave informed consent for the study, which was approved by the Institutional Ethics Committee and conducted in accordance with Declaration of Helsinki of the World Medical Association.
24-h cardiorespiratory polygraphic recording
All patients underwent 24-h continuous polygraphic recording, including electrocardiography, respiration by chest and abdominal inductance plethysmography belts, nasal airflow detection, and oxygen saturation (SaO2) (Somté model PSG2, Compumedics, Abbotsford, Australia). Cardiorespiratory polygraphy was performed by experienced sleep technicians (G.I. and F.B.), who reviewed the raw data. Each analysis was then checked by a physician with specific relevant clinical and research experience (A.G., M.E., or C.P.).
According to guidelines on portable respiratory systems (23), 3 signals were used to score respiratory events and to distinguish between central and obstructive apnea episodes: 1) nasal airflow; 2) respiratory activity recorded from thoracic/abdominal belts; and 3) oxygen saturation. In case of technical issues that made the recording indecipherable, the cardiorespiratory monitoring was immediately repeated.
There were only 29 recording failures (5%) that could not be repeated. These failures were due to technical issues such as loss of airflow signal (n = 7), loss of both thoracic and abdominal bands (n = 10), or loss of oxygen saturation signal (n = 12).
An apnea was defined as a cessation of airflow lasting at least 10 s. A hypopnea was defined as an abnormal respiratory event lasting at least 10 s, with at least a 50% reduction in airflow compared to baseline, without complete cessation, usually in association with a reduction in SaO2. An obstructive apnea (OA) was defined as a pause in respiration of >10 s associated with ongoing ventilatory effort recorded by thoracic and abdominal bands. A central apnea was defined as a >10-s pause in respiration with no associated respiratory effort recorded by thoracic and abdominal bands. Without the availability of an esophageal pressure transducer or diaphragmatic electromyogram to correctly score hypopneas as either central or obstructive and, given the poor reliability of indirect algorithms (24), hypopneas were considered to follow the distribution of most of the apneic events (25). Therefore, the severity of either CA or OA was primarily quantified by the frequency of apnea and hypopnea episodes per hour, or the AHI (26). Patients were randomized to the CA group if their AHI was ≥5 events/h, with >50% of apneic events being central, or to the obstructive apnea group if they had an AHI of ≥5 events/h, with >50% of apneic events being obstructive, as previously reported (26). The average value of AHI during the whole 24-h recording at night (10:00 pm to 6:59 am) and during the daytime (7:00 am to 9:59 pm) were computed and graded in apnea severity for both OA and CA, according to the following grading system: negligible (AHI of 0 to 4.9 events/h); mild (AHI of 5 to 14.9 events/h); moderate (AHI of 15 to 29.9 events/h); and severe (AHI of ≥30 events/h).
Considering the potential to misclassify hypopneas (24,27), an analysis was also performed based only on apnea episodes by using the central apnea index (CAI) and OA index (OAI). The burden of desaturation was evaluated as the minimum SaO2 value reached and the time spent with SaO2 below 90% [T90].
All patients were followed at the hospital outpatient clinic until December 31, 2013, and outcome status was determined from the medical records or telephone interviews with patients, relatives, or general practitioners. No patient was lost at follow-up. The endpoint was death attributable to cardiac cause (sudden death, progressive HF-related death, or acute myocardial infarction). Patients who died of noncardiac causes and those who underwent heart transplantation or left ventricular assist device implantation were censored at the time of the event.
Data were reported as mean ± SD for normally distributed variables; otherwise, they were expressed as median (interquartile range [IQR]). Mean differences between groups were evaluated using the analysis of variance or Kruskal-Wallis test for variables with skewed distribution, with Bonferroni post hoc analysis. Discrete variables were compared by using the chi-square test with Yates correction or the Fisher exact test when appropriate. Survival analysis was estimated by Kaplan-Meier method and log-rank statistics; patients were stratified according to an AHI of ≥15 events/h and a CAI of ≥15 events/h over the 24-h, the nighttime, or the daytime period.
For univariate Cox regression analysis, the candidate independent variables were selected on the basis of the strength of association with outcome shown by previous studies in similar populations, that is, age, estimated glomerular filtration rate (eGFR), assessed using the Modification of Diet in Renal Disease (MDRD) formula; plasma NT-proBNP concentrations; and LVEF. Other assessed variables were AHI and CAI, computed during the night, daytime, and during the whole 24-h period; and OAI and T90 over the whole 24-h period, all considered continuous variables to reduce the potential loss of power associated with dichotomization. All univariate predictors were entered in the backward stepwise multivariate analysis where AHI during the night, day, and the whole 24-h period, alongside CAI during the night, daytime, and 24-h; and T90 values were entered in separate models to avoid multicollinearity. The number of variables included in the multivariate analysis was weighted against the number of cardiac events to minimize the risk of overfitting.
The predictive power of a variable was quantified as the area under the receiver operating characteristic (ROC) curve. Analyses were performed using R version 3.1.1 software (Vienna, Austria). A p value of ≤0.05 was considered statistically significant.
Distribution of apnea over 24-h
Using guideline definitions, prevalence rates of patients with normal breathing (NB) (AHI of <5 events/h) and those with obstructive and central apnea episodes (≥5 events/h, respectively) at night, during the day, and over the whole 24-h period are shown in the Central Illustration. The prevalence of NB increased in the daytime, the OA prevalence decreased, and the CA prevalence remained predominant throughout the 24-h period.
Using an AHI cut off of ≥5 events/h, the prevalence rates of patients with CA at night, during the day, and throughout the 24-h period were 69.1%, 57.0%, and 64.8%, respectively, whereas the prevalence rates of patients with OA were 14.7%, 5.9%, and 12.7%, respectively (Central Illustration). In patients with predominantly CA, 94.5% of apnea episodes were of central origin, whereas in patients with predominantly OA, 85.8% of apnea episodes were of obstructive origin.
By using a higher AHI cutoff of ≥15 events/h, which identifies patients with moderate-to-severe apnea, the prevalence rates of CAs at nighttime, daytime, and throughout the 24-h were 49.9% (24.3% AHI of ≥15 but <30 events/h; 25.5% AHI of ≥30 events/h), 28.4% (20.7% AHI of ≥15 but <30 events/h; 7.7% AHI of ≥30 events/h), and 38.2% (26.8% AHI of ≥15 but <30 events/h; 11.4% AHI of ≥30 events/h), respectively, whereas the prevalence rates of OSA were 9.9% (6.7% AHI of ≥15 but <30 events/h; 3.2% AHI of ≥30 events/h), 1.5% (1.1% AHI of ≥15 but <30 events/h; 0.4% AHI of ≥30 events/h), and 4.5%, respectively (3.7% AHI of ≥15 but <30 events/h; 0.8% AHI of ≥30 events/h) (Central Illustration).
For patients with CAI of ≥5 events/h, the prevalence rates of CA were 44.0%, 26.5%, and 36.2%, respectively, at night, during the day, and during the whole 24-h period; for CAI ≥10 events/h, rates were 46.3%, 27.4%, and 37.9%, respectively, at night, during the day, and during the whole 24-h period, respectively. The prevalence rates of OA in patients with OAI of ≥5 events/h were 10.5%, 1.7%, and 5.5% at night, during the day, and over the whole 24-h period, respectively; and for OAI of ≥10 events/h, rates were 3.0%, 0.2%, and 0.6% for OAI of ≥15 events/h at night, during the day, and over the whole 24-h period, respectively.
The circadian distribution of apnea episodes is shown in Figure 1. In patients with predominantly CA (n = 363 patients), the mean AHI decreased from 20 at night to 10 events/h during the day. Central apnea was the main contributor to the AHI in this subset (highest nocturnal CAI was 15 events/h; highest daytime CAI was 7 events/h).
Notably, 3% of patients experienced CA during the whole 24-h period, whereas 10% of patients showed CA for ≥16 h/day, respectively, well beyond the physiological length of sleep time.
In patients with predominantly OA (n = 77), the AHI decreased from 23 at night to 5 events/h during the daytime, and hypopneas were the main contributors to the AHI values during the daytime with respect to OA.
One-half of the patients presented with moderate-to-severe CA both at night and during the daytime (AHI of ≥15 events/h; n = 138 out of 262 [53%]) (Figure 2A), whereas only a few patients presented with isolated moderate-to-severe daytime apnea (daytime AHI of ≥15 events/h; nighttime AHI of <15 events/h; n = 11 out of 101 [11%]). Patients with moderate-to-severe apnea during both the day and the night also experienced a higher night AHI than patients with moderate-to-severe CA events only at night (39 ± 13 events/h vs. 26 ± 9 events/h; p < 0.001). Daytime and nighttime AHI and CAI showed a significant correlation (R = 0.73 and 0.82, respectively; both, p < 0.01) (Figure 2B).
Clinical characteristics and apnea burden
The clinical characteristics of the whole population and of the subsets with NB, OSA, and CSA are shown in Table 1. Patients with either CSA or OSA at night were older, more frequently males, presented more frequently with ischemic causes, and had lower functional capacity (peak maximum rate of oxygen consumption [Vo2/kg]), compared with patients with NB.
Patients with OSA had the highest body mass index, whereas patients with CSA showed the worst left ventricular systolic and diastolic function and the highest severity of mitral regurgitation, neurohormonal activation, and history of atrial fibrillation. Nonsustained ventricular tachyarrhythmia episodes were more frequent in CSA patients than in patients with NB and OSA. No differences in the desaturation burden were found between patients with CSA and those with OSA.
Clinical, humoral, functional, and echocardiographic parameters according to severity of the OSA and CA (or CSA) occurrence at night (mild vs. moderate vs. severe) are summarized in Tables 2 and 3, respectively. Conversely, Tables 4 and 5 show the impact of graded severity of CAs during the daytime and throughout the 24-h period, respectively.
Patients with severe OSA (i.e., AHI of ≥30 events/h) at night showed lower eGFR and higher plasma NT-proBNP levels than patients with mild OSA. They also showed higher nighttime AHI and OAI and higher daytime and 24-h AHI and OAI episodes. They also experienced a worse burden of desaturation but not increased plasma norepinephrine levels or worsened arrhythmic profiles.
Patients with severe CA (i.e., AHI of ≥30 events/h) during nighttime, daytime, and throughout the 24 h were more frequently males, more symptomatic, had a more severe diastolic dysfunction, and showed higher neurohormonal activation and ventricular arrhythmic burden. Finally, patients with severe CA also experienced a higher burden of desaturation.
During a median 34-month follow-up (IQR: 17 to 36 months), 50 deaths occurred (41 due to HF progression, 5 to sudden cardiac deaths, 4 to fatal myocardial infarctions). Nonsurvivors were older (73 ± 9 years of age vs. 65 ± 12 years of age, respectively; p = 0.001), more symptomatic (NYHA functional class III to IV: 56.0% vs. 32.2%, respectively; p = 0.001), and showed lower LVEF (29 ± 9% vs. 32 ± 9%, respectively; p = 0.01), reduced eGFR (44 ± 15 ml/min vs. 68 ± 26 ml/min, respectively; p = 0.001), and higher plasma NT-proBNP levels (5475 ng/l [IQR: 2747 to 9596 ng/l] vs. 1197 ng/l [IQR: 461 to 2719 ng/l], respectively; p = 0.001).
Nonsurvivors had higher T90 levels (18.0 min [IQR: 7.5 to 27.5 min] vs. 6.5 min [IQR: 2.0 to 12.0 min], respectively; p = 0.001), higher incidence of apnea episodes during the daytime (AHI of 12 events/h [IQR: 5 to 22 events/h] vs. 8 events/h [IQR: 2 to 16 events/h], respectively; p = 0.021) and the whole 24-h period (AHI of 15 events/h [IQR: 9 to 26 events/h] vs. 12 events/h [IQR: 5 to 21 events/h], respectively; p = 0.042) but no significant differences in nighttime CAs (AHI of 24 events/h [IQR: 11 to 36 events/h] vs. 18 events/h [IQR: 8 to 32 events/h], respectively; p = 0.056).
The Kaplan-Meier survival analysis comparing NB, OSA, and CSA patients is shown in Figure 3. Central apnea patients demonstrated the worst prognostic profile (log-rank: 7.2; p = 0.028). Either an AHI of ≥15 events/h or a CAI of ≥10 events/h in patients with CAs was able to stratify mortality (Figure 4) during nighttime, daytime, and for the whole 24-h period. Patients with CAs during both the day and the night showed a worse prognostic profile than patients with CAs occurring during nighttime only (Figure 5).
Univariate analysis was performed to assess the relative contributions of candidate variables to occurrence of major cardiac events. Univariate predictors of increased risk of events were age, LVEF, NT-proBNP, eGFR; night, day, and 24-h AHI; night, day, and 24-h CAI; and T90 (Online Table 1).
With multivariate analysis, independent predictors of events were age, NT-proBNP, eGFR, 24-h AHI; night, day, and 24-h CAI; and T90 (Table 6). ROC analyses performed for 24-h AHI, nighttime CAI, daytime CAI, 24-h CAI, and T90 AUC were 0.59, 0.63, 0.64, 0.64, and 0.7, respectively; ROC curves for 24-h AHI, CAI, and T90 are shown in Online Figure 1.
This prospective study is the largest performed so far that includes the 24-h period in systolic HF patients receiving guideline-recommended treatment (96% on beta-blockers, 93% on angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers, 76% on mineralocorticoid antagonists, 26% with cardiac resynchronization therapy [CRT]).
Our prospective study confirms a significant prevalence of CA during the night (69% AHI of ≥5 events/h; 50% AHI of ≥15 events/h; 24% CAI ≥5 events/h). A significant portion of patients with significant night CAs presented with the same phenomenon during the daytime (with a daytime AHI of ≥5 events/h: daytime AHI of ≥15 events/h was 58%: daytime CAI ≥5 events/h was 11%).
Use of a portable monitor gave a reliable and clinically informative picture of the 24-h apnea phenomenon, with greater patient compliance and wider applicability to outpatient clinics than standard polysomnography. The only study using a similar 24-h analytical approach (14) recruited only 60 HF patients (85% on beta-blockers, no CRT, mean LVEF of 26%) and found a similar nocturnal incidence (62%) but a lower diurnal incidence (16%) of CA (using an AHI of ≥15 events/h).
Studies by Poletti et al. (13) and La Rovere et al. (19) found a prevalence of diurnal CAs of 59% and 38%, respectively, by attended 20-min short-term polygraphic recordings with no grading of CA severity.
The higher prevalence of CAs during the night may be partially explained by the removal of cortical influences on respiratory centers during sleep and by the rostral fluid shift due to supine position (28). The persistence of CA in patients who are awake in the upright position may be related to a baseline worse hemodynamic profile, with increased pulmonary venous pressure and activated chemoreflex sensitivity overcoming the diurnal cortical outflow to respiratory centers. A similar pattern of periodic breathing is sometimes observed during exercise and is named exercise-induced ventilatory oscillations (29,30), although the pathophysiology of this condition is unknown.
The prevalence of OSA was similar to that reported by Grimm et al. (31) and lower than that reported by disparate studies (32–34). Significant variability in definition criteria, prevalence, and population characteristics may explain these controversial findings. Most studies realized after the release of guidelines for home monitors (23) performed in the HF setting have used portable systems (approximately 60% after 2007) for apnea screening (31–35). Variations in prevalence are related to the apnea definition used, the specific morphometric/demographic characteristics of the population, and the severity of HF. Compared to previous studies, no significant differences were found in our cohort in terms of age and body mass index and LVEF. However, the high prevalence of diastolic dysfunction and mitral regurgitation may have increased the rate of CAs over that of OSAs (36–38). Indeed, patients with CA showed worse hemodynamic and neurohormonal profile than patients with OSA.
Severe CA phenomena during the night, daytime, or the whole circadian period were associated with older age and male sex, more severe symptoms, greater LV systolic dysfunction and dilation, and worse renal function. Central apnea patients presented with neurohormonal activation (increased plasma B-type natriuretic peptide level, a higher adrenergic activation), and higher incidence of nonsustained ventricular tachyarrhythmias at 24-h electrocardiography recording than patients with OSA.
Our patients with CA had worse outcomes than patients with OSA, in line with previous findings in patients with HF and reduced ejection fraction (34,39). Patients with CA are exposed to a higher number of apnea episodes, as CAs are present throughout the 24-h period and not at night only as in patients with OSA (Figure 2). Patients with CA during both the day and night showed worse outcomes than patients with CAs present at night only. Although an AHI of ≥15 events/h was accompanied by a negative prognostic value either at night or daytime, only the AHI computed throughout the 24-h period was independently associated with detrimental outcomes. The Kaplan-Meier, Cox, and the ROC analyses showed that CAI measurements are likely more informative than AHI for survival prediction.
In our study, T90 had an independent prognostic value, confirming recent findings by Oldenburg et al. (34). Origin of T90 is likely determined by prolonged circulatory time/apnea and/or increased plant gain, due to reduced lung volume or lung diffusivity, and may mirror hypoxia-mediated organ damage.
Conversely, the 24-h AHI and CAI are likely an expression of the chemoreflex gain (40,41) and may mirror the adrenergic system-mediated organ damage. Hypercapnia exerts negative effects on hemodynamics and arrhythmogenesis too, through adrenergic overactivation, through chemoreflex stimulation. Either 24-h AHI or T90 is likely to provide different and additive information.
Hypopnea episodes were considered to follow the main trend of apnea episodes in our recordings, as previously suggested (25), because, first, the use of either an esophageal pressure transducer or diaphragmatic electromyography is impractical in a population as large as ours; and, second, previous studies showed a low reliability and feasibility of indirect scoring algorithms for attribution of hypopnea episodes (24). Our approach could have led to underestimation of obstructive events and imprecision of the AHI due to misclassification of hypopneas. However, when assigning a patient to the CA or OSA subgroup, the percentage of prevailing apnea episodes approached 95% of events, suggesting that hypopneas follow the general apnea trend.
Respiratory recordings did not include electroencephalographic tracings, and the sleep/awake state cannot be clearly identified. However, CA occurs during the entire 24-h period, and it is unlikely that patients slept the entire time. A potential contribution to phases of sleep (naps) is likely and may explain the peak of AHI in both patients with OSA and patients with CAs after lunch time. However, although this peak occurs on the background of a daytime AHI of <5 events/h in OSA patients, AHI is constantly ≥5 events/h in CA patients. Furthermore, 3% and 10% of patients experienced CAs during the whole 24-h period or at least for ≥16 h/day, respectively, indicating a respiratory phenomenon that extended well over the physiological sleep time. The presence of CAs during wakefulness is confirmed by previously attended short-time recordings (13,19).
We also report prevalence rates using OAI and CAI; the prognostic discriminative power of CAI in our population strengthens the validity of the present analysis and suggests a potential novel prognostic index based on apnea episodes only for future studies.
In patients with systolic HF, CA occurs throughout the 24-h period and is associated with neurohormonal activation, increased ventricular arrhythmic burden, and worse prognosis; these are best predicted by 24-h AHI, CAI, and T90 as measurements of the global apnea burden. This novel observation may at least partially explain why previous therapeutic attempts, such as continuous positive airway pressure (11) or adaptive servoventilation (12), both targeting “sleep” apnea episodes, have failed: targeting only “sleep” apnea may be insufficient in patients who manifest CAs all day. On the other hand, this could explain why only adjustment or upgrade of HF therapy treatment (by guideline-recommended drug therapy and cardiac resynchronization) (42,43) have been associated with a prognostic benefit and with decreasing CA incidence. These treatments likely act on the pathophysiological triggers of CA (in this case, reduced LVEF and hence increased circulatory time) and over the whole circadian period, thus including the subset at major risk.
Comprehensive evaluation of the apnea burden, addressing the presence of CA throughout the 24-h period, could represent a meaningful measure for assessing specific therapies for CA in the future.
COMPETENCY IN MEDICAL KNOWLEDGE: Patients with systolic HF often experience moderate-to-severe CA (AHI of >15 events/h) not only while asleep (60%) but also while awake (30%). An AHI of >15 events/h over 24 h (43% of patients) is associated with increased mortality.
TRANSLATIONAL OUTLOOK: Clinical trials of interventions targeting pathophysiological triggers of CA over 24 h are needed to assess the impact of treatment on clinical outcomes.
For a supplemental figure and table, please see the online version of this article.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Emdin, Mirizzi, and Giannoni contributed equally to this work.
- Abbreviations and Acronyms
- apnea/hypopnea index
- central apnea
- central apnea index
- cardiac resynchronization therapy
- central sleep apnea
- Cheyne-Stokes respiration
- estimated glomerular filtration rate
- heart failure
- Interquartile range
- left ventricular ejection fraction
- normal breathing
- N-terminal pro–B-type natriuretic peptide
- obstructive apnea
- obstructive apnea index
- obstructive sleep apnea
- time with oxygen saturation <90%
- Received April 17, 2017.
- Revision received July 7, 2017.
- Accepted July 9, 2017.
- 2017 American College of Cardiology Foundation
- Cheyne J.
- Stokes W.
- Emdin M.G.,
- Giannoni A.,
- Passino C.
- Yumino D.,
- Wang H.,
- Floras J.S.,
- et al.
- Lanfranchi P.A.,
- Braghiroli A.,
- Bosimini E.,
- et al.
- Kasai T.,
- Kasagi S.,
- Maeno K.,
- et al.
- Abraham W.T.,
- Jagielski D.,
- Oldenburg O.,
- et al.
- Costanzo M.R.,
- Khayat R.,
- Ponikowski P.,
- et al.
- Naughton M.T.
- Linz D.,
- Fox H.,
- Bitter T.,
- et al.
- Emdin M.,
- Passino C.,
- Giannoni A.
- Oldenburg O.,
- Horstkotte D.
- Lang R.M.,
- Bierig M.,
- Devereux R.B.,
- et al.
- Randerath W.J.,
- Treml M.,
- Priegnitz C.,
- Stieglitz S.,
- Hagmeyer L.,
- Morgenstern C.
- Ryan C.M.,
- Floras J.S.,
- Logan A.G.,
- et al.
- Somers V.K.,
- White D.P.,
- Amin R.,
- et al.
- Yumino D.,
- Redolfi S.,
- Ruttanaumpawan P.,
- et al.
- Leite J.J.,
- Mansur A.J.,
- de Freitas H.F.,
- et al.
- Oldenburg O.,
- Lamp B.,
- Freudenberg G.,
- Horstkotte D.
- Chenuel B.J.,
- Smith C.A.,
- Skatrud J.B.,
- Henderson K.S.,
- Dempsey J.A.
- Solin P.,
- Bergin P.,
- Richardson M.,
- Kaye D.M.,
- Walters E.H.,
- Naughton M.T.
- Bitter T.,
- Westerheide N.,
- Prinz C.,
- et al.
- Giannoni A.,
- Emdin M.,
- Bramanti F.,
- et al.
- Sinha A.M.,
- Skobel E.C.,
- Breithardt O.A.,
- et al.