Adaptive servoventilation in diastolic heart failure and Cheyne–Stokes respiration

Patients

A total of 60 patients were included with HFNEF, defined according to current European Society of Cardiology (ESC) guidelines 13, moderate-to-severe CSR, an apnoea/hypopnoea index (AHI) of >15 events·h⁻¹ and a proportion of CSR of >50%. All patients had to be in a stable clinical condition and in New York Heart Association (NYHA) class II–III. Prior to inclusion, evidence of diastolic dysfunction was obtained invasively (P_LV,ED of >20 mmHg) and patients had to be on stable medication for ≥4 weeks. In order to exclude patients with a significant obstructive and/or restrictive pulmonary disease, spirometric testing was performed. Other exclusion criteria were reduced systolic left ventricular function, significant valvular heart disease, a history of sleep-disordered breathing (SDB) or ongoing treatment of SDB, a proportion of CSR of <50%, evidence of pulmonary disease, hypercapnia (P_c,CO2 of >45 mmHg) in capillary blood gas samples, pregnancy and acute coronary syndrome or acute cardiac decompensation.

The present study represents a prospective trial that was initiated in July 2006, lasting until December 2008. All patients included in the study were invited to reattend the sleep laboratory of the Dept of Cardiology (Heart and Diabetes Center North Rhine-Westphalia, Ruhr University Bochum, Bad Oeynhausen, Germany) every 3 months for either measurement of cardiac function (echocardiography, cardiopulmonary exercise testing, measurement of N-terminal-pro-brain natriuretic peptide (NT-proBNP) and blood gas testing) and sleep studies (therapy group) or measurement of cardiac function alone (control group). Individual end-point data from every patient (mean 11.6±3 months; range 3–24 months) formed the basis of the present study. During follow-up, patients undergoing invasive procedures (e.g. percutaneous coronary intervention) were excluded from the study. Although the ESC does not recommend any specific pharmacological treatment for HFNEF patients 14, patients with changes in heart failure medication (except for dosage changes) were excluded. The study was approved by the local institutional review board of the Ruhr University Bochum.

Sleep studies

Sleep studies were performed using full in-hospital polysomnography (N7000/S7000; Embla, Munich, Germany). Electroencephalographic data were collected from central and auricular positions according to Rechtschaffen and Kales 15. Eye movements, as well as mandibular and tibial muscle efforts, nasal airflow (pressure), chest and abdominal effort, pulse oximetry and body position, were recorded continuously. In addition, patients were observed by video. Analyses were performed using Somnologica Studio 5^TM software (Medcare, Embla) and reviewed by two independent SDB specialists not involved in the patients’ treatment.

Hypopnoea was defined as a ≥30% reduction in airflow in combination with an oxygen desaturation of >3% by pulse oximetry. Apnoea was defined as a cessation of airflow for ≥10 s. There had to be at least three cycles of crescendo and decrescendo change in breathing amplitude and ≥5 events·h⁻¹ of central sleep apnoeas or hypopnoeas during sleep for it to be scored as CSR 16. Automatic data from treatment devices were visualised using ResScan^TM software (ResMed, Martinsried, Germany).

Adaptive servoventilation

ASV (AutoSet CS^TM2, ResMed) was introduced to 51 patients. ASV represents a bilevel ventilation system with automatic anticyclic adaptation of pressure support. Individual pressure support was introduced during wakefulness under continuous blood pressure monitoring and adjusted according to nocturnal measurements. At therapy introduction, 31 patients were not willing to receive therapy or rejected therapy after an introductory night. In addition, eight patients withdrew from therapy during follow-up due to nonspecific personal reasons (n = 2; all within 3 months) or due to lacking compliance (mean daily use of <4 h; n = 6; five within 3 months and one within 6 months). For balancing group sizes and due to the fact that lack of compliance with ventilation therapy is associated with poorer outcome in heart failure patients 17, these patients were regarded as untreated, and thus added to the control group. After a follow-up period, 39 patients undergoing ASV therapy and 21 patients without ASV therapy (12 due to nonspecific reasons (e.g. personal lifestyle and sense of shame), five due to mask intolerance and four due to subjective intolerance to positive airway pressure ventilation) were capable of finishing the study. The study flow chart is given in figure 1. In the control group, sleep studies were performed at baseline only.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1–

Study flow chart. PCI: percutaneous coronary intervention; PTSMA: percutaneous transluminal septal myocardial ablation.

Echocardiography

The diameters and dimensions of the left atrium and left ventricle were measured using M-mode following American Society of Echocardiography guidelines 18. Since multiple echocardiographic parameters are available for assessing diastolic function, a comprehensive approach was used in the present study. Transmitral Doppler flow, supplying peak early (E) and peak atrial Doppler mitral inflow velocity (A), the ratio E/A and deceleration time of E were used as traditional markers, with limited validity due to load dependence. New methods, such as speckle tracking imaging, provide accurate well-validated data regarding diastolic properties and filling pressures of the left ventricle. Using speckle tracking technology offline evaluation, mean early diastolic lengthening velocity (e’) was obtained from different segments of the mitral annulus in four-, three- and two-chamber view. Left ventricular filling pressure was determined from the ratio E/e’.

Cardiopulmonary exercise testing

Symptom-limited bicycle cardiopulmonary exercise testing (ZAN Ferraris, Oberthulba, Germany), starting at 10 W and increasing at 10 W·min⁻¹, was performed at baseline and during follow-up. Oxygen consumption (V’_O2), carbon dioxide production (V’_CO2), instantaneous expiratory gas concentrations throughout the respiratory cycle and minute ventilation (V’_E) were measured continuously on a breath-by-breath basis. Cardiac frequency and blood pressure (by sphygmomanometer) were measured at rest, during each stage of exercise and at peak exercise. Peak V’_O2 (V’_O2,peak), V’_O2 at the individual aerobic–anaerobic threshold (V’_O2,AT) and the ratio V’_E/V’_CO2, as well as maximum workload and total exercise time, were recorded. Predicted V’_O2,peak was calculated automatically taking the patients’ sex and age into account.

V’_O2,AT was defined as: 1) the point at which the ventilatory equivalent for oxygen (V’_E/V’_O2) was minimal and followed by a progressive increase; 2) the point after which the respiratory gas exchange ratio consistently exceeded the resting respiratory gas exchange ratio; and 3) the V’_O2 after which a nonlinear increase in V’_E occurred relative to V’_O2. V’_O2,peak was defined as the highest 30-s mean V’_O2 in the last minute of exercise.

Left heart catheterisation

Left heart angiography was performed using a 5F pigtail catheter (Cordis, Langenfeld, Germany) for acquisition of the left ventricular ejection fraction and P_LV,ED.

Capillary blood gas analysis

P_c,CO2, capillary oxygen tension (P_c,O2) and capillary oxygen saturation were measured using an ABL 330 (Radiometer, Copenhagen, Denmark).

N-terminal-pro-brain natriuretic peptide

NT-proBNP was used as an additional marker of HFNEF severity. Analyses were performed using the Elecsys 2010 analyser (Roche, Basle, Switzerland).

Pulmonary function test

In order to exclude patients with significant pulmonary disease, a pulmonary function test was performed. Spirometry and body plethysmography were performed with the use of a constant-volume body plethysmograph (ZAN Ferraris). Vital capacity, forced vital capacity, the ratio of forced expiratory volume in 1 s to forced vital capacity, peak expiratory flow, mid-expiratory flow when 25, 50 or 75% of forced vital capacity remains in the lung, airway resistance, total lung capacity and the ratio of residual volume to total lung capacity were used for the final analysis. The single-breath technique was used for measurement of diffusing capacity for carbon monoxide. For final analysis, the haemoglobin-corrected lung transfer factor for carbon monoxide (T_L,CO) and the carbon monoxide transfer coefficient (or T_L,CO/alveolar volume in millimols per minute per kilopascal (1 kPa = 7.502 mmHg)) were selected.

Statistics

ANCOVA, t-tests and Wilcoxon tests were applied in the statistical analyses. Using ANCOVA, the follow-up value of each variable reviewed was used as response variable. The associated baseline value of the corresponding variable and the two-level factor, ASV or no ASV, were used as explanatory variables. The first step of the analysis was to fit the full model, consisting of different slopes and intercepts for each level of the factor. Subsequently, the model was simplified in order to find the minimal adequate model by removing all nonsignificant terms until a model with only significant terms was reached. Model simplifications were justified by the explanatory power of the model. The simpler model was preferred as long as it did not explain significantly less than the more complex model (α = 0.05). Hence the full model with interaction was checked for significance and compared to the simplified model without interaction. In the case that the full model were preferred, which indicates a significant interaction, further analyses were not continued.

Significant interaction was found by models using the variables V’_E/V’_CO2, oxygen pulse, A, E/A ratio, E/e’ and E. Unless the simpler model were preferred, the interaction could be neglected and it was checked whether the two-level factor was significant. For this purpose, the model without interaction was checked for significance and compared to a model without interaction and the two-level factor (α = 0.05). Provided that the simpler model of these two were preferred, the two-level factor (ASV or no ASV) is nonsignificant and the resulting model does not include the categorical variable. Whenever the model without interaction, consisting of a common slope and different intercepts, was preferred, the two-level factor is significant and this model is the minimal adequate model. Differences between the groups were compared using an unpaired t-test. Weight changes, NYHA class (noncategorical parameter, which could not be included in the ANCOVA analysis) and changes in sleep study results from baseline to follow-up were analysed using a paired t-test if the normality test (Kolmogorov–Smirnov) was passed, and with the help of the Wilcoxon test if the normality test was failed. A p-value of <0.05 was considered significant for all comparisons. Data are presented as mean±sd unless stated otherwise. Owing to missing observations, <60 individuals were used in the various analyses. Statistical analyses were performed using the open source software R-2.6.2 19.

Sleep study results

Sleep study results at baseline are presented in table 1. ASV therapy led to a significant reduction in AHI (43.5±14.7 to 3.5±1.7 events·h⁻¹; p<0.001), longest apnoea (32.3±24.7 to 19.3±16.8 s; p<0.001) and hypopnoea (37.9±10.6 to 23.7±13.1 s; p<0.01), maximum desaturation (82.8±4.8 to 88.1±3.7%; p<0.01), percentage of study time at <90% oxygen saturation (7.6±12.8 to 1.1±3.2%; p<0.01), arousal index (30.7±6.4 to 17.5±8.8 events·h⁻¹; p<0.01) and mean oxygen desaturation (5.6±2.0 to 3.4±1.6%; p<0.01). Total sleep time (359±23 to 344±26 min; p = 0.17) rapid eye movement sleep/total sleep time (13.1±8.1 to 15.4±8.4%; p = 0.10) and mean oxygen saturation (92.5±2.3 to 92.8±2.5%; p = 0.32) remained unchanged. Device-based data analysis illustrated a mean usage of >4 h·day⁻¹ in 64.0±30.4% of all possible treatment nights, with a mean use of 5.3±2.1 h·day⁻¹, an AHI for the entire treatment period of 4.2±4.5 events·h⁻¹ and an apnoea index of 0.2±0.4 events·h⁻¹. Noncompliant patients presented with a mean usage of >4 h·day⁻¹ in 10.1±14.4% of all possible treatment nights, with a mean use of 0.8±1.4 h·day⁻¹. In these patients, the AHI was 3.8±6.9 events·h⁻¹ and the apnoea index 0.2±0.8 events·h⁻¹.

NYHA class

There was a significant reduction in NYHA class in both the ASV treatment group (2.4±0.6 to 2.0±0.8; p<0.001) and the control group (2.5±0.5 to 2.2±0.4; p<0.01).

NT-proBNP

Reductions in NT-proBNP concentration were documented in ASV-treated patients and controls; however, these changes did not differ between the groups (table 2).

Cardiopulmonary exercise testing

The cardiopulmonary exercise testing results are presented in table 2. ASV therapy led to a significant increase in V’_O2,AT, V’_O2,peak, predicted V’_O2,peak and oxygen pulse. V’_E/V’_CO2 and exercise duration tended to improve without reaching a level of significance.

Echocardiography

ASV treatment led to a significant decrease in left atrial diameter and E/A, as well as a significant increase in A, e’ and E/e’. Additional findings are presented in table 2.

Potential limitations

There are some potential limitations to the present study. Treatment allocation was not randomised and blinded to patients and physicians. Even though the groups were similar with respect to measured baseline variables, the groups may differ in factors that cannot be measured due to the lack of randomisation. In addition, the lacking randomisation may have caused a selection bias: patients who comply with ASV therapy may also present with superior compliance to the rest of their heart failure treatment. In a recent study, for instance, it was shown that disease management programmes, as one method of improving patients’ therapy compliance, lead to superior survival in systolic heart failure 35. Another possible limitation is the echocardiographic evaluation of e’ by speckle tracking. Even though the e’ derived from speckle tracking shows significantly lower velocities than that from tissue Doppler, the speckle-tracking-derived E/e’ is at least equal to the tissue Doppler E/e’ for the prediction of elevated left ventricular filling pressures 36. In addition, intraindividual changes that were independent of the evaluation method were focused on. Finally, the study group consisted of patients with HFNEF due to different aetiologies. Owing to this small number of patients, it was not possible to perform a subgroup analysis to clarify the impact of different aetiologies of HFNEF on outcome.

Summary

In patients with HFNEF and CSR, ASV not only improves CSR, with a reduction in AHI and other sleep study parameters, but also has a positive effect on cardiopulmonary exercise capacity and echocardiographic parameters of diastolic function in comparison to an untreated group. This effect may be mediated by eliminating the adverse CSR-associated systemic consequences by controlling nocturnal CSR with ASV. Nevertheless end-point-triggered, randomised and blinded studies are necessary for further evaluation of this therapeutic approach.