Apnea–hypopnea and desaturations in heart failure with reduced ejection fraction: Are we aiming at the right target?

International Journal of Cardiology 203 (2016) 1022–1028

Contents lists available at ScienceDirect

International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard

Apnea–hypopnea and desaturations in heart failure with reduced ejection fraction: Are we aiming at the right target? Barnabas Gellen a,b,h,⁎, Florence Canouï-Poitrine c,d,1, Laurent Boyer b,e, Xavier Drouot e,f, Aurélie Le Thuaut c,d, Diane Bodez a, Ala Covali-Noroc e, Marie Pia D'ortho g, Soulef Guendouz a, Stéphane Rappeneau a, Mounira Kharoubi a, Jean-Luc Dubois-Rande a,b,d, Luc Hittinger a,b,d, Serge Adnot b,e, Sylvie Bastuji-Garin c,d, Thibaud Damy a,b a

Cardiology Department, APHP, Henri-Mondor Hospital, F-94010 Créteil, France Mondor Institute Biomedical Research (IMRB), INSERM U955, F-94010 Créteil, France c Public Health Department and Clinical Research Unit (URC-Mondor), APHP, Henri-Mondor Hospital, F-94010 Créteil, France d CEpiA (Clinical Epidemiology and Ageing) EA4393, Medical School, UPEC, F-94010 Créteil, France e Physiology Department, APHP, Henri-Mondor Hospital, F-94010 Créteil, France f Physiology Department, Poitiers University Hospital, F-86021 Poitiers, France g Physiology Department, APHP, Bichat Hospital, F-75018 Paris, France h Cardiology Department, Poitiers University Hospital, F-86021 Poitiers, France b

a r t i c l e

i n f o

Article history: Received 23 June 2015 Received in revised form 28 October 2015 Accepted 16 November 2015 Available online 18 November 2015 Keywords: Heart failure Sleep disordered breathing Apnea–hypopnea index Desaturation Prognosis

a b s t r a c t Background: Sleep disordered breathing (SDB) is common in patients with heart failure with reduced ejection fraction (HFrEF). An increased apnea–hypopnea index (AHI) is associated with poor outcomes. We examined whether an analysis of nocturnal desaturations (NDs) can improve the risk stratification. Methods: Three-hundred seventy-six consecutive patients with stable chronic HFrEF and LVEF ≤45% were prospectively screened using polygraphy. Sleep apnea (SA) was defined as an AHI ≥15. The mean age was 59 ± 13 years, the mean LVEF was 30 ± 6%, and the median AHI was 18 [IQR: 9.33). The composite end-point of death, heart transplantation or LV assistance occurred in 98 patients (26%) within 3 years. Minimal oxygen saturation (MOS) during sleep, the number of desaturations b 90%/h and the time spent with oxygen saturation b 90% were significantly associated with adverse events (adjusted HR 1.25 [1.03–1.52], 1.25 [1.03–1.53], and 1.28 [1.04–1.59]), whereas the AHI was not (1.10 [0.86–1.39]). The best MOS cut-off value for poor outcomes was ≤88%. The patients with an MOS ≤ 88% had a significantly higher event rate (31.9%) than those with an MOS N 88% (15.6%; p b 0.01). The risk assessment using an MOS of ≤88% in addition to established prognostic markers yielded a net reclassification index (NRI) of nearly 6% and was particularly useful in the subgroup of patients with events (NRI: 8.4%). Conclusions: In HFrEF patients, ND ≤88% appears to be predictive of adverse events, independent of the presence of SA. This suggests that the risk assessment in HFrEF should also include ND in top of AHI. © 2015 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Sleep disordered breathing (SDB), which is highly prevalent in patients with CHF [1,2], is characterized by repetitive suspension (apnea) or reduction (hypopnea) of breathing, causing intermittent oxygen desaturation followed by reoxygenation. Recurrent oxygen desaturation has deleterious effects including bursts of increased sympathetic nervous system activity [3], peripheral vasoconstriction, increased heart rate [4] and blood pressure [5], and the release of pro⁎ Corresponding author at: Department of Cardiology, Henri Mondor University Hospital, 51 Avenue du Maréchal de Lattre de Tassigny, F-94010 Créteil, France. E-mail address: [email protected] (B. Gellen). 1 Equally contributing first author.

http://dx.doi.org/10.1016/j.ijcard.2015.11.108 0167-5273/© 2015 Elsevier Ireland Ltd. All rights reserved.

inflammatory cytokines. Instead of physiological nocturnal relaxation, the cardiovascular system functions under stress conditions in patients with SDB. Indeed, SDB in HF is associated with myocardial ischemia, LV remodeling [6] and arrhythmias [7,8]. Central SDB appears to constitute an independent risk factor for mortality in patients with systolic HF [9–15] even if this association remains somewhat controversial [16,17]. The severity of SDB in HFrEF patients is commonly assessed by the number of apneas and hypopneas per hour of sleep (apnea–hypopnea index; AHI) and by the presence of Cheyne–Stokes respiration (CSR). Gottlieb et al. found in HF patients that hypoxemia but not the AHI was associated with nocturnal BNP increase, reflecting hemodynamic stress [18]. Of note that in the group with severe sleep apnea, more than half of the apnea events were obstructive.

B. Gellen et al. / International Journal of Cardiology 203 (2016) 1022–1028

The prognostic impact of nocturnal desaturations (ND) on the occurrence of sudden cardiac death has been recently demonstrated in patients with obstructive SDB [19]. However, the prognostic value of ND in HFrEF patients remains unknown. 2. Methods 2.1. Study design and setting The study design was an observational prospective cohort survey. The source population consisted of consecutive patients with stable symptomatic HFrEF who were routinely admitted, mostly on an outpatient basis, to the HF clinic of the Cardiology Department of Mondor University Hospital (Créteil/France) from 2005 to 2010 for reevaluation of symptoms, verification of optimal medical treatment and compliance, and therapeutic education if necessary. All the patients were systematically screened for SDB using overnight polygraphy during the study period. SA treatment was then initiated if indicated and if accepted by the patient. Patients who were treated with nocturnal ventilation or oxygen therapy prior to the index polygraphy were not included. The study complied with the Declaration of Helsinki, with the statement of ethical publishing as appears in the International of Cardiology [20], and was approved by the French Society of Cardiology Ethics Committee. Informed consent was obtained from all patients. The study protocol was approved by our local ethics committee (Créteil) and by the French Comité National Informatique et Liberté (CNIL number 1431858).

1023

desaturation index was defined as the number of oxygen desaturations of more than 3% per hour of sleep. The CSR was recorded. The diagnosis of SDB was considered if the AHI was ≥5/h of sleep. The subjects were divided depending on the severity of SDB accordingly to the American Academy of Sleep Medicine recommendations. The diagnosis of SA was considered for AHI ≥15. SDB was considered central if more than 50% of the apnea–hypopnea events were central and obstructive if more than 50% of the events were obstructive.

2.4. Outcomes The patients were followed-up systematically over 3 years after index polygraphy. The primary endpoint was a composite of death, heart transplantation, or a ventricular assist device implant within 3 years of follow-up. Events were defined accordingly to ACC.AHA Key Data Elements and Definitions for Cardiovascular Endpoint Events in Clinical Trials [22]. The first event was recorded as the date after the polygraphy recording. The status and dates of death were obtained from the patients' medical records. If these data were unavailable or if the patient had not reached an endpoint, his or her status was ascertained by a telephone call to their referring cardiologist, heart surgeon, pneumologist, and/or physician. If the information was still unavailable, a death certificate was searched for by the council in the area where the patient was born. If their status was still not available, the patient was considered lost to follow-up at the time of their last visit to the hospital, as recorded in their medical records.

2.2. Patient selection Patients were included in the study if they had an LVEF ≤ 45% as assessed using a transthoracic echocardiography, had a stable clinical status (unchanged NYHA class), and had received unchanged medical therapy for at least 4 weeks before inclusion. The non-inclusion criteria were known ventilatory-treated SA, non-cardiac life-threatening diseases (cancer, infectious disease, stroke, etc.), an acute HF episode within 4 weeks of the polygraphy, or significant valvular heart disease. In particular, patients with significant functional mitral regurgitation (MR; N20 mm2 of effective regurgitation orifice area) were excluded since in these patients the valvulopathy by itself can worsen the congestive state in top of the LV dysfunction, and might thereby influence polygraphy parameters and outcomes. 2.3. Echocardiography, nocturnal polygraphy Two-dimensional and Doppler (pulsed, continuous, and TDI) echocardiography were performed as recommended by the American Society of Echocardiography [21]. LVEF was calculated according to the biplane method of disks (Simpson's rule). An overnight polygraphy was performed using a computerized dataacquisition system (EmblettaTM, ResMed, Saint Priest, France). Oronasal airflow was measured from the nasal pressure and a mouth thermistor, the chest and abdominal effort were assessed using inductance plethysmography, and pulse oximetry, snoring, actimetry, and body position were recorded. The patients completed the Epworth sleepiness scale questionnaire. All the recordings were scored manually by two scorers who were blinded to the clinical and echocardiographic data. We took into account only recordings of a minimum duration of 4 h without any missing traces. The sleep time was assessed using a standardized questionnaire reporting the time of going to bed, the estimated moment of falling asleep, and the time of definite arousal. The number of AHI events was determined after the exclusion of periods with movement, which were considered to be wakeful periods. The oxygen saturation parameters of the polygraphies were collected retrospectively. Daytime oxygen saturation was considered as the mean saturation of the first 5 min of the polygraphy recording. The oxygen

Fig. 1. Study flow chart. HFrEF = heart failure with reduced ejection fraction; MR = mitral regurgitation.

1024

B. Gellen et al. / International Journal of Cardiology 203 (2016) 1022–1028

2.5. Statistical analyses The quantitative data were expressed as the mean (standard deviation, SD) or median (quartile 1–quartile 3, Q1–Q3) as appropriate. The qualitative data were expressed in n (%). The incidence rates of the primary endpoint and each component of the endpoint were separately expressed with their 95% confidence intervals. Kaplan–Meier survival curves were computed for the entire cohort and according to the baseline polygraphic parameters. There are no validated prognostic thresholds for polygraphic parameters except for the AHI. Therefore, thresholds based on the quartile values of each parameter in the event-free group were used for defining the categories. The association of baseline clinical, biological and polygraphic characteristics with the primary endpoint was assessed using a Cox proportional hazards model adjusted for the year of inclusion and expressed by hazard ratios and 95% confidence intervals. Correlations between polygraphic parameters were assessed using Spearman's correlation coefficients. Variables associated with the primary endpoint with p b 0.25 were considered for the multivariate analysis. Bivariate analysis was performed to identify confounding factors. We searched for interactions. A multivariate analysis was performed using a Cox proportional hazards model. We verified the Schönfeld residuals to check the assumption of proportional hazards. The association of each variable with the main endpoint was assessed using an adjusted hazard ratio and 95% confidence interval.

Discrimination was quantified by calculating the Net Reclassification Improvement Index (NRI): the NRI indicates the proportion of patients who are correctly classified in the group with the event and without the event when adding the polygraphic marker to the multivariate model adjusted for clinical and echocardiographic risk factors [23,24]. The thresholds for defining the risk categories were defined using the likelihood ratio method as proposed by Takahara and colleagues [25]. The baseline characteristics of the patients and the incidence rate of the main endpoint were compared according to the polygraphic profiles of the patients using Pearson's chi-square or Fischer's exact tests or the Mann–Whitney test as appropriate. Post-hoc two-by-two comparisons were performed after the Bonferroni correction was applied. All the tests were two-tailed, and the significance threshold was set at b0.05. No multiple imputations were performed for missing data. The statistical analyses were conducted using Stata v12.0 (College Station, TX, USA). 3. Results Among the 607 patients with chronic HFrEF screened over the inclusion period, 376 (62%) patients agreed to participate in the study protocol, had a valid polygraphy recording, and fulfilled all the predefined inclusion and exclusion criteria (Fig. 1). Follow-up was complete in all included patients. Among these patients, 311 (82.7%) had SDB (AHI ≥ 5) and 207

Table 1 Clinical characteristics. Overall (n = 376)

No event

Event

HRyear adjusted [IC 95%]

pa

n = 278 (73.9)

n = 98 (26.1)

Age, mean Sex, male BMI, mean Ischemic etiology NYHA III/IV (n = 374) Arterial hypertension Diabetes Atrial fibrillation (n = 374) Pace-maker (n = 373) Cardiac resynchronization therapy LVEF, median [Q1–Q3] LVEF b35% SBP, median [Q1–Q3] (n = 372) DBP, median [Q1–Q3] (n = 370) Heart rate, median [Q1–Q3] (n = 375) BNP (n = 277) Plasma sodium (n = 316) Creatinine (n = 321) Clearance (Cockroft) Beta-blocker (n = 320) ACE inhibitor (n = 313) ARB (n = 300) MR antagonist (n = 290) Diuretics (n = 272)

58.7 (13.2) 308 (81.9) 27.3 (5.3) 181 (48.1) 157 (42.0) 181 (48.1) 124 (33.0) 70 (18.7) 101 (27.1) 74 (19.7) 30 [25–36] 233 (62.0) 120 [103–133] 70 [60–79] 75 [65–87] 318 [119–704] 139 [137–141] 115 [97–140] 66.9 [46.9–83.3] 278 (86.7) 246 (78.6) 56 (18.7) 153 (52.8) 218 (79.0)

58.5 (12.8) 207 (81.5) 27.7 (5.3) 122 (48.0) 79 (31.3) 128 (50.4) 82 (32.3) 41 (16.3) 53 (19.3) 31 (12.2) 30 [26–38] 147 (57.9) 120 [107–135] 70 [60–80] 75 [65–86] 215 [94.5–552.5] 140 [138–141] 108 [96–131] 69.1 [51.5–86.1] 213 (91.8) 184 (80.7) 30 (17.1) 111 (52.1) 148 (74.8)

59.4 (14.3) 101 (82.8) 26.2 (5.0) 59 (48.4) 78 (63.9) 53 (43.4) 42 (34.4) 29 (23.8) 48 (49.0) 43 (35.2) 25 [20–35] 86 (70.5) 110 [100–126] 67 [56–72] 78 [66–88] 658 [344–1213] 138 [133.5–140] 132 [105–151] 49.8 [38.6–76.2] 62 (73.9) 62 (72.9) 18 (23.1) 42 (54.6) 67 (90.5)

1.02 [0.83–1.24] 0.76 [0.47–1.25] 0.95 [0.91–0.99] 1.02 [0.69–1.52] 3.84 [2.50–5.89] 0.69 [0.46–1.04] 1.09 [0.72–1.66] 1.50 [0.93–2.41] 2.79 [1.87–4.14] 2.34 [1.54–3.55] 0.93 [0.91–0.96] 1.60 [1.03–2.47] 0.62 [0.50–0.77] 0.66 [0.54–0.80] 1.20 [0.99–1.44] 1.84 [1.51–2.23] 0.91 [0.86–0.95] 1.85 [1.14–3.00] 0.75 [0.62–0.90] 0.29 (0.18; 0.47) 0.66 (0.41; 1.07) 1.45 (0.85–2.45) 1.18 (0.75–1.85) 3.24 (1.49–7.07)

0.88 0.28 0.02 0.9 b0.001 0.08 0.67 0.09 b0.001 b0.001 b0.001 0.03 b0.001 b0.001 0.06 b0.001 b0.001 0.01 0.002 b0.001 0.09 0.17 0.47 0.003

Polygraphic parameters AHI Sleep apnea (AHI ≥ 15) Number of desaturation of ˃3%/h (n = 375) Mean oxygen saturation during sleep (%) Time spent with oxygen saturation b90% (min, n = 372) Minimal nocturnal oxygen saturation during sleep (%) Number of desaturations b90%/h (n = 362) Cheyne–Stokes respiration (n = 375)

18 (9–33) 207 (55.1) 11 (4–25) 94.2 (92.6–95.5) 5.7 [0.6–35.9] 83.5 (77–88) 2 (0.3–7.9) 96 (25.6)

18 (9–33) 152 (54.7) 10 (3–25) 94.2 (92.6–95.5) 5.3 [0.3–35.5] 84 (78–88) 1.9 (0.1–7.8) 63 (22.7)

17.5 (9–32) 55 (56.1) 11 (5–23) 94.1 (92.9–95.5) 10.7 [1.1–44.2] 83 (76–87) 2.4 (0.5–9.2) 33 (33.7)

1.06 [0.85–1.31] 1.07 [0.71–1.59] 1.10 [0.90–1.35] 0.93 [0.74–1.15] 1.18 [0.96–1.43] 0.89 [0.75–1.06] 1.15 [0.96–1.38] 1.61 [1.03–2.51]

0.61 0.74 0.34 0.5 0.11 0.24 0.12 0.04

Sleep apnea treatment in patients with AHI ≥15 (N = 207) No treatment CPAP BiPAP ASV

119 (57.5) 22 (10.6) 3 (1.5) 63 (30.4)

82 (54.0) 21 (13.8) 2 (1.3) 47 (30.9)

37 (67.3) 1 (1.8) 1 (1.8) 16 (29.1)

1 (reference) 0.09 [0.01–0.65] – 0.58 [0.32–1.07]

0.08

NYHA = New York Heart Association; LVEF = left ventricular ejection fraction; SBP = systolic blood pressure; DBP = diastolic blood pressure; BNP = brain natriuretic peptide; ACE = angiotensin converting enzyme; ARB = angiotensin receptor blocker; MR = mineralocorticoid receptor; AHI = apnea–hypopnea index; CPAP = continuous positive airway pressure; BiPAP = bi level positive pressure ventilation; ASV = adaptive servoventilation. a Year-adjusted Cox proportional hazards model.

B. Gellen et al. / International Journal of Cardiology 203 (2016) 1022–1028

(55%) had SA (AHI ≥ 15). The demographic, clinical, biological, and polygraphic variables are summarized in Table 1. There was a total of 98 (26.1%) events, corresponding to an incidence rate of 29% (95% CI: 24– 34). The mortality rate was 21.5% (95% CI: 17.3–26.5; n = 68); the heart transplantation rate was 9.5% (95% CI: 6.6–13.5; n = 28); and the LV assistance rate was 3.6% (95% CI: 2–6.3; n = 12). Compared to the group without a primary endpoint, the group with adverse events had, as expected, higher NYHA classes, a lower LVEF, a higher BNP, lower sodium levels and creatinine clearance, a lower baseline BMI, more pacemakers and more CRT. The patients who experienced events had a comparable daytime oxygen saturation (95.5%; 94.1–96.7) to those patients who did not experience events (96.0%; 94.6–96.9). Regarding polygraphic parameters, CSA and Cheyne–Stokes breathing were more frequent in patients with events. There was a tendency toward more time spent with oxygen saturation b 90% (hypoxia burden) and a higher number of desaturations b90%/h in the group with adverse events than in those without. Because of the high correlation between polygraphic parameters, each tested parameter was analyzed separately. In the multivariate analysis after adjustment for confounders (NYHA, LVEF, BMI, atrial fibrillation, SBP, CRT, year of index polygraphy, and SA treatment status), the number of desaturations N3%/h, a hypoxia burden b90%, the minimal oxygen saturation (MOS), and the number of desaturations b 90%/h were independently associated with the primary endpoint (Fig. 2). Systolic blood pressure was identified as a negative confounder explaining the absence of significant association between desaturation parameters and outcomes in the crude analysis. In other terms, in the overall population, patients with more altered desaturation parameters had more likely higher blood pressure (data not shown). We examined whether there was a clinically relevant threshold of MOS allowing for the better identification of patients with a higher risk for adverse events. Minimal oxygen saturation was divided into quartiles, and adjusted hazard ratios were calculated for each quartile. We found a significantly higher risk of adverse events for patients with MOS ≤88% (first three quartiles) than for the fourth quartile of patients with MOS N 88%. Fig. 3 shows the event-free survival of patients according to this threshold, with a significantly more favorable outcome for patients with MOS N88%. In the patients with MOS ≤88%, daytime oxygen saturation was comparable between those with an AHI ≥ 15 (95.2%; 94.0–96.7) and those with an AHI b 15 (95.4%; 93.8–96.5).

1025

By using MOS ≤ 88% in addition to the established prognostic markers of HFrEF, the risk assessment could be significantly improved in the group of patients with events by up-grading their risk (NRI 8.4%, with 90% of correct reclassification), whereas it was useless in the group of patients without events (Table 2). The year of polygraphy was included in the model in order to adjust for the evolution of the treatment of HFrEF between 2005 and 2010. For example, the recommendation level and therefore the implementation of CRT have significantly evolved over this period. Using the presence of SA and the presence of MOS ≤88%, we identified 4 distinct phenotypes of patients (Table 3). In patients with SA, the AHI ranges were 16–68 in those without MOS ≤ 88% and 16–124 in those with MOS ≤88%, respectively. The patients with MOS ≤ 88% had high event rates independent of the presence of SA. The presence or absence of SA appeared to be related to prognosis in only the patients without nocturnal desaturations ≤ 88%. Daytime oxygen saturation was comparable between the four groups. 4. Discussion This is the first large-scale long-term study evaluating the prognostic effect of nocturnal desaturation (ND) in stable HFrEF patients. Our results showed that the severity of ND is strongly associated with poor outcomes and that it improves risk stratification. The prognosis of patients with severe ND is comparably poor regardless of the presence of clinically significant sleep apnea (SA). Most studies focus on characterizing SA by the number (AHI) and type (obstructive, central, and mixed) of the reduction of breathing activity, and less or no importance is attached to a detailed characterization of desaturation (amplitude and duration). Nocturnal desaturation was suggested to have direct detrimental hemodynamic effects in HF patients independently of SA in a physiological cross-sectional study [26]. Therefore, it appears important to analyze the prognostic value of AHI separately from that of desaturation because patients with a high AHI might not desaturate significantly, whereas patients with few apnea–hypopnea events might show important decreases in oxygen saturation during sleep. In SA patients without HF, nocturnal desaturation is a consequence of reduced breathing activity; therefore, AHI adequately reflects the severity of SA. In patients with congestive heart failure, nocturnal

Fig. 2. Multivariate analysis of polygraphy parameters for death, heart transplantation or LV assistance within 3 years. *Multivariate Cox-proportional hazard model adjusted for NYHA, LVEF, BMI, atrial fibrillation, SBP, CRT, and year of polygraphy. The time spent with oxygen saturation b90% and the number of desaturations b90%/h are expressed for each 1 SD increase. The minimal oxygen saturation during sleep is expressed for one SD decrease. The time spent with oxygen saturation b90% and the number of desaturations b90%/h are log-transformed. NYHA = New York Heart Association; LVEF = left ventricular ejection fraction; BMI = body mass index; SBP = systolic blood pressure; CRT = cardiac resynchronization therapy.

1026

B. Gellen et al. / International Journal of Cardiology 203 (2016) 1022–1028

Fig. 3. Kaplan–Meier curve of event-free survival according to minimal nocturnal saturation.

desaturation might be provoked, or at least accentuated, by worsened hemodynamic conditions in the supine position. Rostral fluid shift from the legs toward the chest might aggravate pulmonary congestion and compromise blood oxygenation [27]. In these patients, nocturnal desaturation is therefore not necessarily a consequence of reduced breathing activity: HF patients with a normal (b 5) or near-to-normal [5–15] AHI might desaturate significantly during sleep, and the magnitude and the duration of these desaturations might affect the prognosis. We demonstrated for the first time an independent association between desaturation parameters and prognosis in HFrEF patients. Our findings are in line with recent data from a large retrospective cohort of more than 10,000 SA patients showing a significant association between nocturnal saturation and cardiovascular outcomes and all-cause mortality, without a significant association between AHI and outcome [28]. The objective of this study was to assess the prognostic value of saturation parameters using index polygraphy in the real life, i.e., by applying the recommendations for nocturnal ventilation if indicated. The majority of our patients with an AHI ≥15 were indeed treated using nocturnal ventilation, which might impact on outcomes in HFrEF [29,30].

Therefore, the design of the cohort was not appropriate to compare the prognostic effect of AHI with that of desaturation. The absence of a statistical correlation between the AHI and the outcome does not call into question the negative prognostic value of untreated SA in chronic HF. However, there is some mechanistic evidence in favor of the importance of desaturation over AHI in HFrEF. Gottlieb et al. had previously found that in HFrEF patients, nocturnal hemodynamic stress is a consequence of a lack of oxygen rather than of the frequency of apnea episodes and arousals [18]. They characterized nocturnal lack of oxygen by the time spent with oxygen saturation below 90%, which was called the “burden of hypoxemia”. In line with these findings, a higher burden of desaturation was associated with increased adverse events (HR: 1.28) in our cohort. A minimal nocturnal saturation of ≤88% appeared more adequate for identifying patients with a particularly poor prognosis in our cohort than the traditionally used 90% threshold. Nearly one-third of the patients in our cohort presented significant nocturnal desaturations (MOS ≤88%) without significant SA (AHI b 15). The clinical outcomes in this group were as poor as those of patients with SA (AHI ≥ 15) and significant nocturnal desaturations. These data suggest that adverse outcomes might be at least as strongly related to desaturations as to reduced breathing activity in HFrEF patients. Three hypotheses can be put forward to partly explain the fact that desaturations were associated with poor outcomes and should be considered in HFrEF. First, important desaturations in the supine position might be strongly related to hemodynamic compromise in HFrEF patients and might therefore represent a more powerful marker of disease severity. Second, ventilation therapy was not prescribed to the 169 patients (45%) with an AHI b15, and 70% of these patients (N = 118) showed desaturations ≤ 88%. These 118 (39%) out of the 306 patients with important desaturations did not receive ventilation or oxygen therapy that could have positively influenced the nocturnal lack of oxygen and the outcome. Third, even in patients with an AHI ≥15 that were prescribed nocturnal ventilation, desaturations might have still occurred even in the absence of reduced breathing activity. With regard to risk stratification, taking in account nocturnal desaturation ≤ 88% in addition to established prognostic HF markers allowed us to redefine individual risk in a non-negligible proportion of

Table 2 Reclassification by addition of minimal oxygen saturation during sleep to risk assessment (n = 351). Model with minimal oxygen saturation during sleep (> or <=88%)* Low risk (<15%) Event

Intermediate risk (15– 41%)

High risk (>41%)

Reclassified to higher risk

Reclassified to

Net correct

lower risk

reclassification, %

94

Model : NYHA +LVEF+ BMI+ CRT + SBP +AF + year of polygraphy* Low risk (<15%)

13

2

0

0

22

7

0

1

50

Low risk (<15%)

83

22

0

Intermediate risk (15–41%)

18

84

8

0

5

36

Intermediate risk (15–41%) High risk (>41%) No event

High risk (>41%) Net reclassification improvement, %

9

1

8.4

30

23

–2.7

256

5.7

Net Reclassification Index (NRI) represents the proportion of patients who are reclassified into a more adequate risk class using a multivariate model with minimal oxygen saturation during sleep compared to a multivariate predictive model without minimal oxygen sleep duration. NRI = (proportion of patients with an events reclassified into higher risk category with the model included minimal oxygen saturation − proportion of patients with an events reclassified into lower risk category with the model included minimal oxygen saturation) / (patients without events reclassified into a lower risk category with the model included minimal oxygen saturation − patients without events reclassified into a higher risk category with the model included minimal oxygen saturation). NYHA = New York Heart Association; LVEF = left ventricular ejection fraction; BMI = body-mass index; CRT = cardiac resynchronization therapy; SBP = systolic blood pressure; AF = atrial fibrillation. Multivariate Cox proportional hazards model.

B. Gellen et al. / International Journal of Cardiology 203 (2016) 1022–1028

1027

Table 3 Patient phenotypes depending on the presence of sleep apnea (SA; AHI ≥ 15) and of desaturations ≤88%. Overall (n = 376)

No SA/no desaturation n = 51 (13.6)

SA/no desaturation n = 19 (5.0)

No SA/desaturation n = 118 (31.4)

SA/desaturation n = 188 (50.0)

pa

Age, median (e–t) Male sex BMI (kg/m2), mean (e-t) Ischemic etiology NYHA III/IV Arterial hypertension Diabetes Atrial fibrillation Pace-maker (n = 373) Cardiac resynchronization therapy LVEF b 35% SBP (mm Hg) DPB (mm Hg) Heart rate (bpm) BNP (ng/l; n = 277) Plasma sodium (mmol/l; n = 316) Creatinine (μmol/l; n = 321)

58.7 (13.2) 308 (81.9) 27.3 (5.3) 181 (48.1) 157 (42.0) 181 (48.1) 124 (33.0) 70 (18.7) 101 (27.1) 74 (19.7) 233 (62.0) 120 [104–133] 70 [60–79] 75 [65–87] 318 [119–704] 139 [137–141] 115 [97–140]

53.7 (15.3) 36 (70.6) 25.1 (4.1) 22 (43.1) 16 (31.4) 22 (43.1) 12 (23.5) 6 (11.8) 13 (25.5) 8 (15.7) 23 (45.1) 116 [100–130] 64 [60–73] 75 [65–88] 118 [58–320] 139 [136–140] 106 [90–125]

59.0 (12.1) 18 (94.7) 26.2 (3.3) 12 (63.2) 4 (21.0) 10 (52.6) 3 (15.8) 0 (0) 4 (22.2) 3 (15.8) 15 (78.9) 110 [100–135] 70 [60–80] 75 [63–88] 311 [123–497] 139 [135–140] 130 [97–151]

57.5 (12.2) 87 (73.7) 27.7 (5.4) 50 (42.4) 52 (44.4) 55 (46.6) 41 (34.7) 18 (15.4) 37 (31.4) 27 (22.9) 69 (58.5) 119 [105–136] 70 [60–80] 76 [65–86] 317 [105–704] 140 [137–141] 110.5 [99–136]

60.8 (13.0) 167 (88.8) 27.8 (5.5) 97 (51.6) 85 (45.4) 94 (50.0) 68 (36.2) 46 (24.6) 47 (25.3) 36 (19.1) 126 (67.0) 120 [105–131] 70 [60–80] 75 [65–87] 399 [160–896] 139 [137–141] 121 [98–141]

0.01 b0.001 0.01 0.19 0.07 0.8 0.13 0.01 0.64 0.68 0.01 0.55 0.05 0.97 0.001 0.87 0.1

AHI Oxygen desaturation of ˃3%/h (n = 375) Oxygen desaturations b90%/h (n = 362) Mean oxygen saturation during sleep (%) Time spent with oxygen saturation b90% (min, n = 372) Minimal nocturnal oxygen saturation during sleep (%)

18 [9–33] 11 [4–25] 2 [0.3–7.9] 94.2 [92.6–95.5] 5.7 [0.6–35.9]

5 [3–9] 2 [1–4] 0 [0–0] 95.9 [94.6–96.9] 0 [0–0]

27 [18–43] 13 [7–20] 0.1 [0–0.3] 95.3 [94.2–96.5] 0 [0–0.2]

9 [5–12] 6 [3–10] 1.1 [0.3–3] 94.0 [92.6–95.4] 4.85 [1.35–20.3]

30.5 [22–44] 23 [13–36] 6.8 [2–18.8] 93.5 [91.9–94.8] 20.9 [4.1–64.5]

b0.001 b0.001 b0.001 b0.001 b0.001

83.5 [77–88]

91 [90–92]

89 [89–91]

84 [78–86]

80 [74–85]

b0.001

Event rates within 3 years, n (%)

98 (29)

8 (18.9)

1 (6.7)

35 (32.0)

54 (31.7)

0.058

NYHA = New York Heart Association; LVEF = left ventricular ejection fraction; SBP = systolic blood pressure; DBP = diastolic blood pressure; BNP = brain natriuretic peptide; SA = sleep apnea; AHI = apnea–hypopnea index; bpm = beats per minute. Significant SA = AHI N15. a Pearson's chi-square test for qualitative variables, Mann–Whitney test for quantitative variables, log-rank test for comparison of events within 3 years.

HFrEF patients. In the subgroup of patients with adverse events, the risk of 11% of the patients was reclassified using the ≤88% saturation threshold, of which 90% were reclassified correctly into a higher category of risk. Thus, the presence of nocturnal desaturations ≤88% in HFrEF patients appears to be helpful in identifying those with adverse outcomes. The results of the SERVE-HF study suggest that the treatment of predominant CSA with Adaptive Servo-Ventilation in HFrEF patients is deleterious [31]. However, definitive conclusions should not be drawn before a careful analysis of all data, including desaturation parameters. There are other questions that remain still unanswered and require further research. First, does nocturnal desaturation merely reflect the severity of supine hemodynamic impairment in HFrEF, or is it causally implicated in adverse events by triggering arrhythmias and the worsening of congestion? Second, in HFrEF patients with desaturations that do not have indications for nocturnal ventilation therapy (AHI b 15), would nocturnal oxygen therapy, beyond reducing daytime fatigue and sleepiness [32], improve outcomes? The randomized Oxygen-HF and the North-East Oxygen Network (NEON) trials examining the effects of home oxygen therapy in HFrEF will answer these questions [33].

4.1. Limitations of the study Brain natriuretic peptide levels could not be included in the risk stratification because our laboratory switched from BNP to NT-proBNP during the inclusion period, and there is no reliable conversion method between both peptide measurements that allows for statistical analysis. The study was not sufficiently powered to examine the prognostic effects of desaturations according to the different types of sleep apnea (OSA and CSA). Patient adherence to ventilation therapy was not systematically assessed during the complete follow-up, and its potential effect on the outcomes could therefore be not determined. The 88% cut-off value for ND was identified retrospectively and needs therefore confirmation by a prospective study.

4.2. Conclusions and perspectives Nocturnal desaturation appears to be a powerful tool to predict poor outcomes in HFrEF patients with SA, independent of nocturnal ventilation therapy and in addition to routinely used markers of HF severity. HFrEF patients without clinically significant SA (AHI b 15) presenting significant nocturnal desaturations have similar event rates to those with SA, which highlights the prognostic importance of desaturation events. A saturation threshold of ≤ 88% adequately identifies patients with a poor prognosis. A prospective study based on oxygen saturation parameters is required to evaluate the clinical effect of a more intense follow-up of HFrEF patients presenting nocturnal desaturations below 88%. In HFrEF, the prognostic effect of nocturnal oxygen therapy in patients presenting important ND should be investigated in a large randomized-controlled trial. Conflict of interest None. References [1] O. Oldenburg, B. Lamp, L. Faber, H. Teschler, D. Horstkotte, V. Topfer, Sleepdisordered breathing in patients with symptomatic heart failure: a contemporary study of prevalence in and characteristics of 700 patients, Eur. J. Heart Fail. 9 (2007) 251–257. [2] R. Schulz, A. Blau, J. Borgel, H.W. Duchna, I. Fietze, I. Koper, et al., Sleep apnoea in heart failure, Eur. Respir. J. 29 (2007) 1201–1205. [3] P. Solin, D.M. Kaye, P.J. Little, P. Bergin, M. Richardson, M.T. Naughton, Impact of sleep apnea on sympathetic nervous system activity in heart failure, Chest 123 (2003) 1119–1126. [4] C.P. O'Donnell, C.D. Schaub, A.S. Haines, D.E. Berkowitz, C.G. Tankersley, A.R. Schwartz, et al., Leptin prevents respiratory depression in obesity, Am. J. Respir. Crit. Care Med. 159 (1999) 1477–1484. [5] J.P. Hobkirk, T. Damy, M. Walters, A. Bennett, S.J. Smith, L. Ingle, et al., Effects of reducing inspired oxygen concentration for one hour in patients with chronic heart failure: implications for air travel, Eur. J. Heart Fail. 15 (2013) 505–510.

1028

B. Gellen et al. / International Journal of Cardiology 203 (2016) 1022–1028

[6] T. Damy, A. Paulino, L. Margarit, X. Drouot, M. Stoica, E. Vermes, et al., Left ventricle remodelling is associated with sleep-disordered breathing in non-ischaemic cardiopathy with systolic dysfunction, J. Sleep Res. 20 (2011) 101–109. [7] S. Javaheri, W.S. Corbett, Association of low PaCO2 with central sleep apnea and ventricular arrhythmias in ambulatory patients with stable heart failure, Ann. Intern. Med. 128 (1998) 204–207. [8] T. Bitter, N. Westerheide, C. Prinz, M.S. Hossain, J. Vogt, C. Langer, et al., Cheyne– Stokes respiration and obstructive sleep apnoea are independent risk factors for malignant ventricular arrhythmias requiring appropriate cardioverter–defibrillator therapies in patients with congestive heart failure, Eur. Heart J. 32 (2011) 61–74. [9] H. Wang, J.D. Parker, G.E. Newton, J.S. Floras, S. Mak, K.L. Chiu, et al., Influence of obstructive sleep apnea on mortality in patients with heart failure, J. Am. Coll. Cardiol. 49 (2007) 1625–1631. [10] P.J. Hanly, N.S. Zuberi-Khokhar, Increased mortality associated with Cheyne–Stokes respiration in patients with congestive heart failure, Am. J. Respir. Crit. Care Med. 153 (1996) 272–276. [11] S. Javaheri, R. Shukla, H. Zeigler, L. Wexler, Central sleep apnea, right ventricular dysfunction, and low diastolic blood pressure are predictors of mortality in systolic heart failure, J. Am. Coll. Cardiol. 49 (2007) 2028–2034. [12] D.D. Sin, A.G. Logan, F.S. Fitzgerald, P.P. Liu, T.D. Bradley, Effects of continuous positive airway pressure on cardiovascular outcomes in heart failure patients with and without Cheyne–Stokes respiration, Circulation 102 (2000) 61–66. [13] U. Corra, M. Pistono, A. Mezzani, A. Braghiroli, A. Giordano, P. Lanfranchi, et al., Sleep and exertional periodic breathing in chronic heart failure: prognostic importance and interdependence, Circulation 113 (2006) 44–50. [14] D. Yumino, H. Wang, J.S. Floras, G.E. Newton, S. Mak, P. Ruttanaumpawan, et al., Relationship between sleep apnoea and mortality in patients with ischaemic heart failure, Heart 95 (2009) 819–824. [15] P.A. Lanfranchi, A. Braghiroli, E. Bosimini, G. Mazzuero, R. Colombo, C.F. Donner, et al., Prognostic value of nocturnal Cheyne–Stokes respiration in chronic heart failure, Circulation 99 (1999) 1435–1440. [16] S. Andreas, G. Hagenah, C. Moller, G.S. Werner, H. Kreuzer, Cheyne–Stokes respiration and prognosis in congestive heart failure, Am. J. Cardiol. 78 (1996) 1260–1264. [17] T. Roebuck, P. Solin, D.M. Kaye, P. Bergin, M. Bailey, M.T. Naughton, Increased longterm mortality in heart failure due to sleep apnoea is not yet proven, Eur. Respir. J. 23 (2004) 735–740. [18] J.D. Gottlieb, A.R. Schwartz, J. Marshall, P. Ouyang, L. Kern, V. Shetty, et al., Hypoxia, not the frequency of sleep apnea, induces acute hemodynamic stress in patients with chronic heart failure, J. Am. Coll. Cardiol. 54 (2009) 1706–1712. [19] A.S. Gami, E.J. Olson, W.K. Shen, R.S. Wright, K.V. Ballman, D.O. Hodge, et al., Obstructive sleep apnea and the risk of sudden cardiac death: a longitudinal study of 10,701 adults, J. Am. Coll. Cardiol. 62 (2013) 610–616. [20] L.G. Shewan, A.J. Coats, Adherence to ethical standards in publishing scientific articles: a statement from the International Journal of Cardiology, Int. J. Cardiol. 161 (2012) 124–125. [21] American College of Cardiology Foundation Appropriate Use Criteria Task F, American Society of E, American Heart A, American Society of Nuclear C, Heart

[22]

[23]

[24] [25] [26] [27]

[28]

[29]

[30]

[31]

[32]

[33]

Failure Society of A, Heart Rhythm S, Society for Cardiovascular A, Interventions, Society of Critical Care M, Society of Cardiovascular Computed T, Society for Cardiovascular Magnetic R, P.S. Douglas, M.J. Garcia, D.E. Haines, W.W. Lai, W.J. Manning, et al., ACCF/ASE/AHA/ASNC/HFSA/HRS/SCAI/SCCM/SCCT/SCMR 2011 appropriate use criteria for echocardiography. A report of the American College of Cardiology Foundation Appropriate Use Criteria Task Force, American Society of Echocardiography, American Heart Association, American Society of Nuclear Cardiology, Heart Failure Society of America, Heart Rhythm Society, Society for Cardiovascular Angiography and Interventions, Society of Critical Care Medicine, Society of Cardiovascular Computed Tomography, and Society for Cardiovascular Magnetic Resonance Endorsed by the American College of Chest Physicians, J. Am. Coll. Cardiol. 57 (2011) 1126–1166. K.A. Hicks, J.E. Tcheng, B. Bozkurt, B.R. Chaitman, D.E. Cutlip, A. Farb, et al., 2014 ACC/ AHA key data elements and definitions for cardiovascular endpoint events in clinical trials: a report of the American College Of Cardiology/American Heart Association Task Force on Clinical Data Standards (writing committee to develop cardiovascular endpoints data standards), J. Am. Coll. Cardiol. 66 (2015) 403–469. M.J. Pencina, R.B. D'Agostino Sr., R.B. D'Agostino Jr., R.S. Vasan, Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond, Stat. Med. 27 (2008) 157–172 (discussion 207-12). M.S. Pepe, H. Janes, C.I. Li, Net risk reclassification p values: valid or misleading? J. Natl. Cancer Inst. (2014) 106 (dju041). M. Takahara, N. Katakami, H. Kaneto, I. Shimomura, Risk categorization for calculating net reclassification improvement, Eur. J. Epidemiol. 28 (2013) 607–609. F.J. Giordano, Oxygen, oxidative stress, hypoxia, and heart failure, J. Clin. Invest. 115 (2005) 500–508. D. Yumino, S. Redolfi, P. Ruttanaumpawan, M.C. Su, S. Smith, G.E. Newton, et al., Nocturnal rostral fluid shift: a unifying concept for the pathogenesis of obstructive and central sleep apnea in men with heart failure, Circulation 121 (2010) 1598–1605. T. Kendzerska, A.S. Gershon, G. Hawker, R.S. Leung, G. Tomlinson, Obstructive sleep apnea and risk of cardiovascular events and all-cause mortality: a decade-long historical cohort study, PLoS Med. 11 (2014), e1001599. T. Damy, L. Margarit, A. Noroc, D. Bodez, S. Guendouz, L. Boyer, et al., Prognostic impact of sleep-disordered breathing and its treatment with nocturnal ventilation for chronic heart failure, Eur. J. Heart Fail. 14 (2012) 1009–1019. C. Jilek, M. Krenn, D. Sebah, R. Obermeier, A. Braune, V. Kehl, et al., Prognostic impact of sleep disordered breathing and its treatment in heart failure: an observational study, Eur. J. Heart Fail. 13 (2011) 68–75. M.R. Cowie, H. Woehrle, K. Wegscheider, C. Angermann, M.P. d'Ortho, E. Erdmann, et al., Adaptive servo-ventilation for central sleep apnea in systolic heart failure, N. Engl. J. Med. 373 (2015) 1095–1105. S. Sasayama, T. Izumi, M. Matsuzaki, A. Matsumori, H. Asanoi, S. Momomura, et al., Improvement of quality of life with nocturnal oxygen therapy in heart failure patients with central sleep apnea, Circ. J. 73 (2009) 1255–1262. A.L. Clark, M.J. Johnson, I. Squire, Does home oxygen benefit people with chronic heart failure? BMJ 342 (2011) d234.