Multicenter, Prospective Study on Respiratory Stability During Recovery From Deterioration of Chronic Heart Failure

Abstract

Background: The respiratory instability frequently observed in advanced heart failure (HF) is likely to mirror the clinical status of worsening HF. The present multicenter study was conducted to examine whether the noble respiratory stability index (RSI), a quantitative measure of respiratory instability, reflects the recovery process from HF decompensation.

Methods and Results: Thirty-six of 44 patients hospitalized for worsening HF completed all-night measurements of RSI both at deterioration and recovery phases. Based on the signs, symptoms, and laboratory data during hospitalization, the Central Adjudication Committee identified 22 convalescent patients and 14 patients with less extent of recovery in a blinded manner without any information on RSI or other respiratory variables. The all-night RSI in the convalescent patients was increased from 27.8±18.4 to 34.6±15.8 (P<0.05). There was no significant improvement of RSI, however, in the remaining patients with little clinical improvement. Of the clinical and laboratory variables, on stepwise linear regression modeling, body weight, peripheral edema, and lung congestion were closely related to the RSI of recovered patients and accounted for 56% of the changes in RSI (coefficient of determination, R²=0.56).

Conclusions: All-night RSI, a quantitative measure of respiratory instability, could faithfully reflect congestive signs and clinical status of HF during the recovery process from acute decompensation.

Despite medical and device-based therapeutic advances, patients with heart failure (HF) are still at high risk of morbidity and mortality.¹ The incidence of rehospitalization for HF is also high,^2,3 resulting in a major burden on patients, caregivers, and the health-care system in the aging society. HF will become more prevalent, with a strong impact on the overall medical economy.¹ Under these circumstances, new management strategies are warranted on the basis of the pathophysiology of worsening HF. One such strategy is to detect early decompensation when appropriate intervention is possible. Unlike blood pressure in hypertension or hemoglobin A1c in diabetes mellitus, however, there are no appropriate quantitative parameters of clinical severity of HF. This is because HF is a syndrome involving complex pathophysiological aspects of the circulatory system, and hemodynamic, neurohumoral, and biochemical derangement.^4–7 Consequently, the severity of HF has been assessed categorically using patient-reported functional capacity. New York Heart Association (NYHA) functional classification or quality of life score based on questionnaire, although simple and widely adopted in numerous clinical studies, has several disadvantages in reporting functional state.^8–11 Given that the classification is non-parametric and based on patient symptoms, there are intrinsic difficulties in using the derived functional capacity to evaluate the pathophysiology of HF.¹² Older adults do not report symptoms to health-care professionals because they believe vague symptoms to be related to aging rather than illness, resulting in a delay in identification of worsening HF. Furthermore, dyspnea and body weight gain leading to hospitalization usually occur late in the course of decompensation,¹² and thus early deterioration of HF fails to be detected.¹³

Two potential methods of detecting development of pulmonary congestion have been introduced in clinical settings: one is a measurement of intrathoracic impedance reflecting fluid status,^14,15 and the other is direct monitoring of pulmonary artery pressure with an implanted pressure sensor,¹⁶ allowing hemodynamic-guided medical management. Despite the potential benefit of direct hemodynamic monitoring of pulmonary congestion, these invasive approaches are limited to selected patients and are not applicable to all patients with HF. To date, there has been no reliable, sensitive, non-invasive, and widely applicable indicator for the detection of early deterioration of HF.

Recently, we have developed a new quantitative and non-invasive measure of respiratory instability (respiratory stability index: RSI). RSI reflects not only Cheyne-Stokes respiration but also irregular rapid and shallow respiration without periodicity.¹⁷ We have demonstrated that RSI measured during the daytime is a strong and independent predictor of poor prognosis in patients with chronic HF, suggesting that the respiratory instability measured by RSI is an ominous sign of worsening HF.¹⁷ The aim of the present study was therefore to examine whether respiratory instability assessed on all-night RSI faithfully reflects the clinical severity of HF as patients improved from acute decompensation of congestive state immediately after admission, to the recovery phase before discharge.

Methods

Subjects and Study Design

We recruited patients with NYHA functional class III or IV who were admitted to hospital (Appendix S1) for acute decompensated HF (excluding HF exacerbation following infectious diseases) between November 2015 and April 2016. All patients were aged ≥20 years and had had no respiratory equipment such as positive pressure ventilation or oxygen inhalation. Key exclusion criteria were acute coronary syndrome in the 3 months before enrollment, chronic obstructive pulmonary disease, pneumonia or other infection, central nervous disorder, chronic renal failure with hemodialysis, or symptomatic malignancy. This observational study was undertaken at 9 centers. Enrollment began in November 2015 and ended in April 2016, with follow-up completed in June 2016.

During the period from deterioration to recovery, all-night respiratory measurements were performed twice: first “at deterioration” (immediately after admission or after weaning from assisted ventilation when used); and again “at recovery” (just before discharge). When hospitalization lasted >2 weeks, an additional all-night respiratory measurement was repeated when possible at the midpoint of hospitalization in order to assess the trend of RSI more accurately during the recovery phase. Given that changes in respiratory indices may differ depending on the HF conditions, the following background information was collected: gender, age, body weight, underlying heart disease, history of cardiovascular therapy, HF treatment prior to admission (concomitant drugs and therapies), the reason for hospitalization, hospitalization status, and the baseline cardiac function on echocardiography. During hospitalization, severity of HF was followed up by checking symptoms, body weight, physical findings, brain natriuretic peptide (BNP) level, chest X-ray, echocardiographic variables, and additional therapeutic regimen on or near the day of all-night respiratory measurements. Patient-reported functional capacity was also assessed using a specific activity scale.¹¹

All-night respiratory measurements were carried out using a portable respiratory monitoring device (SAS-3200, Nihon Kohden, Tokyo, Japan). Air flow was measured using a nasal cannula with pressure sensor. For the assessment of sleep apnea, arterial oxyhemoglobin saturation (SpO₂) was continuously measured with a pulse oximeter (SAS-2100, Nihon Kohden) attached to the subject’s finger using a flexible probe. Signals of nasal pressure were digitized and sampled at 32 Hz, and SpO₂ at 2 Hz. These data were analyzed off-line using a personal computer.

The program to calculate RSI, created by the author, underwent validation by the Academic Research Organization (ARO) of Kyushu University Hospital. Analysis of the source code and movement inspection using raw data were performed for the RSI calculation system including the RSI calculation program, along with that of the peripheral equipment, by the ARO, who confirmed that they worked according to the specifications.

The study protocol was approved by the Ethics Committee at each center. The study was conducted in accordance with the Declaration of Helsinki (revised in October 2013) and “Ethical Guidelines for Medical and Health Research Involving Human Subjects” amended on 22 December 2014. Patients or their proxies were given written informed consent forms approved by the Ethics Committees, were fully informed about the study in writing, and then freely signed the consent forms.

Analysis of All-Night RSI

The data to be included in the analysis were restricted to those for the fixed hours from 23:00 through 05:00 hours. All data were collected in the ARO of Kyushu University Hospital and automatically accumulated and analyzed, so that any third person could not arbitrarily select the period of time to be included in the analysis. Respiratory flow signals measured with nasal pressure sensor were integrated to obtain instantaneous ventilatory signals and resampled at 4 Hz. All-night respiratory signals were divided into serial 5-min segments every 50 s and then subjected to spectral analysis with the maximum entropy method (MEM; Figure 1). More than 420 segments of 5-min data (≥350 min) were analyzed. For each variable, the power spectra were serially calculated using MEM. We focused on 2 frequency ranges to estimate RSI as previously described.¹⁷ One range consisted of the respiratory frequency components retrieved from the instantaneous ventilatory signals by removing high- or low-frequency noise through a 5th-order band-pass Butterworth filter with cut-off frequencies of 0.11 and 0.5 Hz. The other components comprised the very low frequency range of respiration, as an index reflecting the magnitude of periodic breathing. This periodic breathing was obtained by tracing peaks of the instantaneous ventilatory signals with the baseline adjusted to zero, and then applying the band-pass filter with cut-off frequencies of 0.008, and 0.04 Hz. MEM was applied to these respiratory and periodic breathing curves to extract the spectral components of respiratory variations from each wave. All spectral power was normalized using the ratio (%) of the maximum power of the respiratory components. In the evaluation of respiratory instability, we focused mainly on respiratory interval variations and equally adopted all respiratory frequency points that had spectral power >10% of the maximum respiratory power. The very low frequency points of the periodic breathing curve were also taken into account in evaluating RSI only when the maximum power of the very low frequency components was >50% of the maximum power of the respiratory components. The distribution of these respiratory frequency points was evaluated using standard deviation, and RSI was defined as the reciprocal of the standard deviation. Serial changes in all-night RSI were averaged to serve as a representative of all-night respiratory instability.

Figure 1.

All-night respiratory signals and respiratory stability index (RSI). All-night respiratory signals were divided into serial 5-min segments every 50 s, and then spectral analysis was carried out using the maximum entropy method. All spectral power was normalized by the ratio (%) of the maximum power of the respiratory components. The distribution of these respiratory frequency points was evaluated using SD, and RSI was defined as the reciprocal of the SD. Normal all-night RSI demonstrates oscillatory patterns with a similar cycle interval to rapid-eye-movement (REM)/non-REM sleep rhythm. PSD, power spectral density; RespHzSD, standard deviation of respiratory frequencies.

The maximum fall of SpO₂ from baseline and the periods in which SpO₂ decreased by ≥3% from baseline were determined every 5 min from serial segments of SpO₂. We calculated the hourly number of episodes of ≥3% desaturation as the oxygen desaturation index.

Endpoints and Evaluation of Recovery

The primary outcome was all-night RSI along with the recovery from acute decompensation of HF. Secondary outcomes were change in all-night respiratory measurements including mean respiratory rate, apnea-hypopnea index (AHI), and 3% oxygen desaturation index. The Central Adjudication Committee (Appendix S2) determined, in a blinded manner without any information about RSI or other respiratory variables, whether clinical status of HF had improved or was unchanged during the observation periods of hospitalization. Unchanged was defined as no change in congestive symptoms and signs, cardiac size or lung congestion on chest X-ray, and/or BNP on data collected on day 7 or later after hospitalization, compared with those at admission. In contrast, these abnormalities related to worsening HF disappeared in the improved group. None of the patients had worsening during hospitalization.

Target Sample Size

This study was not designed as a verification study, and the nature of this study was proof of concept. Thus, a power calculation was not done for the sample size. To determine a feasible sample size, we performed a survey prior to the start of the study and found that an average of 2 patients were hospitalized for exacerbation of HF per month during the winter period (December–April) at each center. Assuming an entry rate of 50%, we set the target sample size at 40 so that patient enrollment would be completed in 6 months.

Statistical Analysis

All numerical data are summarized as mean±SD and categorical variables as prevalence (%). We used the paired t-test or McNemar test to evaluate changes in variables between soon after admission and before discharge. Two-sided P<0.05 was considered significant. We also investigated important predictors of variation in RSI during the period of recovery from admission to discharge. For this purpose, we performed univariate and multivariate analyses in all 36 patients, consisting of 22 recovered patients and 14 patients who failed to demonstrate improvement. To select a set of predictors that have a close relationship with the changes in RSI, we applied a stepwise variable selection process based on P-value. Candidate predictors consisted of functional capacity, heart rhythm, parameters related to volume overload and congestion, cardiac function, and medication added after treatment. P<0.15 derived from the stepwise regression analysis was used to identify important predictors. Given that the objective of analysis was to identify variables that change in accordance with RSI, we excluded age and sex from the explanatory variables.

Results

The patient selection process is shown in Figure 2. Of the 44 patients enrolled, 2 withdrew before the start of the study and 1 withdrew before RSI at the time of recovery could be measured, resulting in 41 patients with complete RSI measurements at both admission and discharge. Reviewing the records of the HF clinical index data, the Central Adjudication Committee judged 25 patients to be qualified for primary analysis. The condition of these patients was definitely deteriorating at the time of RSI testing on admission, and improving at the time of RSI testing at discharge. The clinical status of the remaining 16 patients was judged to be unchanged. Three of the 25 convalescent patients and 2 of 16 patients with less extent of improvement did not complete RSI test before discharge because of patient refusal. Consequently, 22 patients successfully completed RSI test at both time points and were included in the analysis of the change in RSI during recovery. The results from the remaining 14 patients who failed to demonstrate improvement were referred to for better understanding of the results obtained from the primary analysis.

Figure 2.

Patient selection process. RSI, respiratory stability index.

Table 1 lists the baseline characteristics of the 41 patients whose RSI data were obtained at admission. The mean age was 77.2 years, and half of the patients had either dilated cardiomyopathy or ischemic heart disease accompanied by atrial fibrillation or hypertension. Most of the patients were severely ill at admission, as shown by NYHA class III or IV, the average specific activity scale of 2.3 MET, low left ventricular ejection fraction of 44%, and high BNP of 974 pg/mL. Two patients had improved from NYHA class III to II, with intensive treatment in the acute phase of recovery.

Table 1. Baseline Subject Characteristics: RSI Obtained at Admission and Discharge

Variable	n=41
Age (years)	72.2±12.08
Sex (male)	25 (61.0)
Underlying heart disease
Dilated cardiomyopathy	6 (14.6)
Ischemic heart disease	15 (36.6)
Valvular heart disease	5 (12.2)
Hypertensive heart disease	11 (26.8)
Others	4 (9.8)
Complications
Atrial fibrillation	23 (56.1)
Diabetes mellitus	13 (31.7)
Hypertension	24 (58.5)
Chronic kidney disease	14 (34.1)
Functional capacity
NYHA
I	0 (0.0)
II	2 (4.9)
III	31 (75.6)
IV	8 (19.5)
Specific activity scale (MET)	2.3±1.07
BNP (pg/mL)	975±1,602
Medications
Diuretic	26 (63.4)
ACEI/ARB	22 (53.7)
β-blocker	19 (46.3)
Inotropic drug	5 (12.2)
Aldosterone antagonist	21 (51.2)

Data given as mean±SD or n (%). ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor antagonist; BNP, brain natriuretic peptide; NYHA, New York Heart Association; RSI, respiratory stability index.

The HF indices at the time of admission and discharge are listed in Table 2. In the convalescent patients, symptomatic improvement was shown before discharge by decreased prevalence of NYHA class III and IV from 96 to 36% (P<0.01) and improved specific activity scale from 2.2 to 4.7 MET (P<0.01). Concomitantly, heart rate and body weight fell with disappearance of the third heart sound; congestive signs such as rales, cervical venous distension, hepatomegaly, leg edema, lung congestion, and pleural effusion almost disappeared. Laboratory data before discharge also documented reduction of cardiac size evaluated using cardiothoracic ratio and left ventricular dimension, a decrease in BNP from 701 to 326 pg/mL (P<0.01), lowered prevalence of atrial fibrillation from 52 to 32% (P<0.05), lowered AHI from 18.6 to 12.7/h (P<0.05), and an increased all-night RSI from 27.8 to 34.6 (P<0.05). In patients who were judged as unchanged by the Central Adjudication Committee, there was no significant improvement in respiratory variables, including RSI.

Table 2. Change in HF Indices

Variable	Improved group				Unchanged group
	n	Admission	Discharge	Difference	n	Admission	Discharge	Difference
No. days between admission and HF indices assessment	25	2.4±1.8			16	9.6±5.3
Age (years)	25	72.6±12.1			16	71.5±12.4
Sex (male)		16 (64.0)				9 (56.3)
Dyspnea
At rest		18 (72.0)	0 (0.0)	−0.7±0.5**		9 (56.3)	0 (0.0)	−0.6±0.5**
Walking		25 (100.0)	6 (24.0)	−0.8±0.4**		16 (100.0)	8 (50.0)	−0.5±0.5**
Functional capacity
NYHA	25			−1.0±0.8**	16			−0.6±0.6
I		0 (0.0)	3 (12.0)			0 (0.0)	2 (12.5)
II		1 (4.0)	13 (52.0)			1 (6.3)	5 (31.3)
III		18 (72.0)	9 (36.0)			13 (81.3)	9 (56.3)
IV		6 (24.0)	0 (0.0)			2 (12.5)	0 (0.0)
Specific activity scale (MET)	25	2.2±1.0	4.7±1.3	2.5±1.5**	16	2.5±1.1	3.9±1.4	1.4±1.0**
Physical findings
Heart rate (beats/min)	25	96.4±36.4	70.7±12.7	−25.8±33.9**	16	78.4±17.8	70.2±12.6	−8.2±12.3*
SBP (mmHg)	25	126.0±24.1	109.4±11.3	−16.6±25.3**	16	116.3±18.7	115.1±18.9	−1.2±12.9
DBP (mmHg)	25	78.0±17.3	66.7±10.5	−11.3±17.0**	16	67.9±14.4	63.0±9.9	−4.9±11.8
Body weight (kg)	25	63.5±14.4	58.3±14.0	−5.3±4.3**	16	58.4±18.7	56.4±19.0	−1.9±1.8**
Third heart sound		11 (44.0)	1 (4.0)	−0.4±0.5**		7 (43.8)	4 (25.0)	−0.2±0.4
Rales		14 (56.0)	1 (4.0)	−0.5±0.5**		6 (37.5)	1 (6.3)	−0.3±0.5*
Jugular venous distention		19 (76.0)	1 (4.0)	−0.7±0.5**		9 (56.3)	4 (25.0)	−0.3±0.5*
Hepatomegaly		11 (44.0)	2 (8.0)	−0.4±0.5**		2 (12.5)	2 (12.5)	–
Leg edema		22 (88.0)	3 (12.0)	−0.8±0.4**		10 (62.5)	2 (12.5)	−0.5±0.5**
Laboratory examination
Lung congestion		20 (80.0)	0 (0.0)	−0.8±0.4**		13 (81.3)	1 (6.3)	−0.8±0.4**
Pleural effusion		20 (80.0)	4 (16.0)	−0.6±0.5**		13 (81.3)	9 (56.3)	−0.3±0.4
CTR (%)	24	64.7±8.1	59.1±7.8	−5.5±4.1**	16	60.7±7.2	56.9±6.5	−3.7±4.7**
Atrial fibrillation		13 (52.0)	8 (32.0)	−0.2±0.4*		6 (37.5)	4 (25.0)	−0.1±0.3
BNP (pg/mL)	25	701±391	326±281	−376±368**	16	1,401±2,507	770±707	−631±2,319
Sodium (mEq/L)	25	140.0±3.9	138.3±4.5	−1.7±3.4*	16	137.6±4.3	137.1±3.6	−0.4±3.1
LVEF (%)	19	45.2±18.0	49.5±17.4	4.3±8.8*	14	45.3±13.5	46.9±16.2	1.6±8.2
LVDd (mm)	19	56.1±9.8	54.5±9.7	−1.6±3.1*	14	52.6±6.2	53.0±6.8	0.4±2.5
LVDs (mm)	19	43.5±12.1	40.7±11.8	−2.7±4.6*	14	40.7±7.9	39.5±8.4	−1.2±4.3
Deceleration time (ms)	19	132.0±38.5	191.2±53.1	59.1±49.5**	14	156.7±90.4	181.4±56.0	24.7±103.7
E/e’	17	19.2±10.8	15.7±6.3	−3.5±9.0	13	21.4±8.8	17.9±3.9	−3.5±7.7
All-night monitoring
Respiratory rate (/min)	22	16.6±2.6	15.4±2.1	−1.2±2.3*	14	19.7±2.7	18.9±2.8	−0.8±2.1
AHI	23	18.6±16.3	12.7±10.5	−5.9±12.9*	14	15.7±22.7	14.6±22.8	−1.1±8.6
3%ODI	23	19.3±16.2	13.3±11.0	−6.0±14.4	14	16.0±21.3	13.3±19.0	−2.7±8.0
Mean RSI	22	28.7±18.4	34.6±15.8	5.9±11.2*	14	28.3±15.7	30.9±17.9	2.6±15.0
Additional medication after admission
Oral
Diuretics		9 (36.0)				6 (37.5)
ACEI/ARB		2 (8.0)				4 (25.0)
β-blocker		2 (8.0)				2 (12.5)
Vasodilators		2 (8.0)				0 (0.0)
Inotropic agents		0 (0.0)				1 (6.3)
I.v.
Diuretics		10 (40.0)				12 (75.0)
Vasodilators		6 (24.0)				5 (31.3)
Inotropic agents		4 (16.0)				3 (18.8)
β-blocker		1 (4.0)				2 (12.5)

Data given as mean±SD or n (%). *P<0.05, **P<0.01. Categorical variables were tested using McNemar test. Continuous variables were tested using paired t-test. 3%ODI, 3% oxygen desaturation index; AHI, apnea-hypopnea index; CTR, cardiothoracic ratio; DBP, diastolic blood pressure; E/e’, ratio of transmitral early filling velocity to early diastolic tissue velocity; HF, heart failure; LVDd, left ventricular end-diastolic diameter; LVDs, left ventricular end-systolic diameter; LVEF, left ventricular ejection fraction; SBP, systolic blood pressure. Other abbreviations as in Table 1.

Figure 3 shows all-night RSI and respiratory rate in patients with severe and moderate HF. RSI in severe HF (Figure 3A) was low and flat during acute decompensation soon after admission, while it doubled after the recovery period, before discharge. In patients with moderate HF (Figure 3B), RSI remained suppressed with a transient rise during acute decompensation but showed repetitive increases in RSI after recovery before discharge. The oscillatory patterns in RSI found during the recovery phase have distinctive 90-min oscillations corresponding to the rapid-eye-movement (REM)/non-REM sleep cycle. There was no appreciable change in respiratory rate before or after recovery.

Figure 3.

Recovery of all-night respiratory stability index (RSI) in patients with (A) severe heart failure and (B) moderate heart failure. (A) The increase in RSI before discharge was accompanied by distinctive 90-min oscillations corresponding to rapid-eye-movement (REM)/non-REM sleep cycle. (B) After recovery before discharge, RSI showed repetitive increases with 90-min cycle length (REM/non-REM sleep cycle).

Number of days between admission and assessment of RSI and HF indices was an average of 2.4±1.8 days for the patients with improved outcome and 9.6±5.3 days for those with unchanged outcome. The delayed assessment after admission was considered to have brought about some extent of improvement in the HF signs and symptoms in the unchanged-outcome group, resulting in less improvement from admission to discharge in this group compared with the improved-outcome group. Nine of the 22 patients with improved outcome underwent RSI assessment halfway between deterioration and recovery. The change in RSI from deterioration to the midpoint, and from deterioration to the time of discharge was 1.33 and 5.11, respectively, suggesting that RSI faithfully reflects the status of patient improvement over the recovery period.

On univariate analysis, changes in heart rate and lung congestion during recovery tended to be related to those in RSI. Using a stepwise method with a multiple linear regression model based on P-value (<0.15), body weight (P=0.001), edema (P=0.002), lung congestion (P=0.003), serum sodium concentration (P=0.115), and an additional treatment with β-blocker (P=0.136) were identified as explanatory variables, with coefficient of determination R²=0.43 (Table 3). Given that congestive signs, body weight, edema, and lung congestion had a close relationship with changes in RSI, we created another multiple linear regression model using these congestive parameters as predictor variables, and changes in RSI as an explanatory variable, in 22 convalescent patients. The coefficient of determinant R² was 0.56, suggesting that the model explained 56% of the variability in RSI (Figure 4).

Table 3. Indicators of Change in RSI

Variable	Univariate analysis	Multivariate analysis
	P-value	Regression coefficient	P-value
Intercept		−2.6	0.584
Disappearance of dyspnea at rest	0.78
ΔNYHA	0.93
ΔSpecific activity scale	0.59
ΔHeart rate	0.08
ΔBody weight	0.22	−2.0	0.001
Disappearance of third heart sound	0.31
Improvement of leg edema	0.18	10.9	0.002
Improvement of lung congestion	0.06	−12.6	0.003
Improvement of pleural effusion	0.33
ΔCTR	0.87
Disappearance of atrial fibrillation	0.31
ΔBNP	0.75
ΔSodium	0.40	0.9	0.115
ΔLVEF	0.97
ΔLVDd	0.30
ΔDeceleration time	0.56
Addition of oral diuretics	0.32
Addition of oral ACEI/ARB	0.90
Addition of oral β-blocker	0.58	−9.8	0.136
Addition of oral vasodilators	0.92
Addition of oral inotropic agents	0.92
Addition of i.v. diuretics	0.79
Addition of i.v. vasodilators	0.70
Addition of i.v. inotropic agents	0.31
Addition of i.v. of β-blocker	0.97

Aabbreviations as in Tables 1,2.

Figure 4.

Prediction of change in respiratory stability index (RSI) by congestive signs of heart failure. On multiple linear regression modeling involving 22 convalescent patients, a 56% improvement in RSI (coefficient of determinant R²=0.56) was determined according to changes in congestive signs such as lung congestion, edema, and body weight.

Discussion

The main finding in this study is that respiratory instability quantified on all-night RSI became stabilized during the recovery process in hospitalized patients with worsening HF. A consistent improvement of RSI was also found at the interim point of the recovery process, whereas RSI did not increase significantly in patients who were judged to have no significant improvement of clinical status by the Central Adjudication Committee. Another new finding in the present study was that the major determinant for all-night RSI was systemic and pulmonary congestion. We have previously reported that RSI measured during the daytime has independent prognostic importance in patients with chronic HF.¹⁷ These unique pieces of respiratory information suggest that RSI serves as a non-invasive and quantitative measure of severity of HF instead of NYHA class or quality of life score based on questionnaires.

Pathophysiology of Respiratory Instability in HF

Respiratory instability is known to be linked to hemodynamic impairment and augmented sympathetic discharge through direct coupling between the respiratory and autonomic centers, and through reflexes from lung mechanoreceptors and muscle metaboreceptors in worsening HF.^4,7,18,19 The mechanisms for respiratory instability in HF, however, have not been fully described because of the multiple factors involved and the lack of a simple clinical measure to quantify not only periodic breathing but also irregular rapid and shallow respiration without periodicity. All-night RSI solved the challenge of quantifying these unstable respiratory patterns and raised the possibility of respiratory instability serving as an integrated measure for the expression of complex hemodynamic, neurohumoral, and chemical interaction in worsening HF. Periodic breathing frequently occurs in patients with HF when CO₂-dependent negative feedback for respiratory control becomes unstable by prolonged circulation time due to low cardiac output and enhanced central and peripheral CO₂ chemosensitivity.^7,20,21 We have previously reported that increased central sympathetic nervous activity observed in patients with HF could play an important role under these conditions, because central sympathoinhibition with guanfacine has a potential benefit in stabilizing the enhanced central CO₂ chemosensitivity and excessive exercise ventilation.⁷ A close correlation between RSI and HF-related fluid accumulation observed in the present study indicates that respiratory instability is also provoked by factors other than the unstable chemoreflex feedback control.

With regard to randomly irregular, rapid, and shallow breathing or non-periodic temporal apnea, lung congestion is likely to be facilitating respiratory instability. When lung compliance is decreased by congestion or increased central blood volume in HF, the respiratory rate rises with a reduction in tidal volume (rapid and shallow respiration) so that breathing should produce the required alveolar ventilation with the minimal amount of work required by the respiratory muscles. This concept is based on efficiency from the viewpoint of the work performed/energy consumed ratio.²² In experimental studies, an increase in pulmonary artery pressure and the development of congestive pulmonary edema could be detected by 4 vagal nerve sensors of the lungs.²³ These vagal sensors work in different ways. Pulmonary C-fibers (type J-receptors), which comprise 70–80% of the afferent nerve from small airways, cause temporal respiratory inhibition, rapid shallow breathing, and a dyspneic sensation, thus are considered to play a critical role as interstitial pressure receptors. The augmented C-fiber discharge is prolonged both when pulmonary artery pressure is elevated leading to lung congestion, and when interstitial edema remains, even after pulmonary artery pressure has been normalized.²³ Bronchial C-fibers are also stimulated by more severe lung congestion and initiate these reflex responses. While stimulated by a high pulmonary artery pressure, rapidly adapting stretch receptors can stimulate deep breathing or sighing through excitatory input to the respiratory center. As pulmonary congestion develops, slowly adapting stretch receptors seem to cause rapid breathing by shortening the inspiratory duration.²³ Thus, these 4 vagal reflexes evoked by vagal sensor stimulation stemming from pulmonary edema have simultaneous counteracting influences on breathing, and could therefore result in irregularly rapid or slow respiratory profiles with non-periodic apnea. These non-periodic unstable respiratory patterns are also taken into account by RSI for the assessment of respiratory instability.¹⁷ Therefore, RSI may serve as a surrogate measure of pulmonary fluid overload in HF with the aid of a built-in sensor system for pulmonary artery pressure and congestion.

Clinical Implications

Given that RSI is defined as 1/(SD of the power spectrum: Hz), it has the dimension of time and its units are seconds. Therefore, RSI is a characteristic time. Theoretically, the width of a power spectrum is the reciprocal value of the decay time (T), characterized by the loss of correlation between the signal at time t and time t+T. For regular respiratory signals, T is very large. In contrast, loss of regularity of respiration over time (respiratory instability) is characterized by the magnitude of T, which corresponds to RSI. In the present study (Figure 3A), RSI after admission was 13 (s) and respiratory rate was 15/min (respiratory interval=4 s). These conditions suggest that breath is temporally correlated with up to the 3rd breath (13/4=nearly 3), gradually losing the correlation with the next 4th breath. Thus, RSI has a physiological meaning of the duration (seconds) of respiratory stability. Therefore, it would be more comprehensive to rename RSI “respiratory stability time” (RST).

We have already documented the usefulness of RSI obtained from 5-min daytime recordings of respiratory signals in the supine position, in which daytime RSI had prognostic importance independent of sympathetic nerve activation in patients with clinically stable HF.¹⁷ The reason why we performed all-night RSI measurement in this study is that respiratory stability is influenced by lung congestion or an increased central blood volume, as well as by the negative feedback of respiratory control of carbon dioxide concentration. In HF, a long period in the lying position at night gradually augments venous return by the fluid shift from interstitial tissues to the venous circuit, while influences by other factors such as skeletal muscle mechano- and metabo-reflexes and higher brain function are diminished. Consequently, respiration during sleep in patients with HF is exclusively controlled through the negative feedback for maintaining carbon dioxide concentration at the predetermined level, and external input from lung stretch receptors, which are stimulated by lung congestion or an increased pulmonary arterial pressure. As shown in the present study, RSI was closely related to volume overload associated with worsening HF. Therefore, all-night RSI could provide information about irritant vagal afferents from lung stretch receptors related to lung congestion, as well as unstable respiratory feedback related to Cheyne-Stokes respiration.

Increasing elderly HF is characterized by a variety of clinical features and frequent and costly hospital readmissions.³ Recent advances in telecommunication technology have created new opportunities to provide a tele-monitoring care system as an adjunct to medical management of HF patients.^13–16,24 For remote management of patients with HF, the home tele-monitoring system has 2 major clinical benefits: facilitation of patient self-care; and early detection of HF decompensation. The latter benefit crucially depends on what is monitored. Although it is recommended to monitor weight daily, it is a relatively poor surrogate for intracardiac filling pressure in HF patients.^12,13,24 In contrast, the CHAMPION trial reported that wireless implantable hemodynamic monitoring of pulmonary artery pressure significantly reduced the risk, compared with the control group, of hospital admission due to worsening HF.¹⁶ Evidence of beneficial outcome in the use of implantable device for monitoring intrathoracic impedance to evaluate central fluid balance in patients with congestive HF, is accumulating.^14,15 Despite these benefits, device implantation is invasive and costly, and therefore applicable to limited patients with HF. Given that RSI, obtained easily and non-invasively from respiratory signals, reflects the congestive state of HF, this parameter may be used in the tele-monitoring management of HF patients equipped with non-attached sensor technology.

Study Limitations

Although RSI is thought to vary according to the congestive and clinical status of HF or of medical treatment, a limitation of the present analysis is that direct comparison between changes in RSI and pulmonary capillary wedge pressure or cardiac output, was not performed, and another was that the present study consisted of a relatively small number of patients. Further investigations that compare RSI and invasive hemodynamic variables, and that enroll a large number of HF patients are warranted to generalize the present findings. Another issue that needs careful consideration is the interpretation of RSI during the acute phase of HF decompensation. After recovery from decompensation, the profile of all-night RSI (Figure 3) featured a prominent REM/non-REM sleep cycle, which had been completely abolished with a very low average RSI during the acute stage of decompensation. This indicates a close relationship between RSI and sleep architecture.²¹ Some of the patients with sleep depletion due to nocturnal dyspnea for a few days before admission usually fell into a deep sleep immediately after the relief of pulmonary congestion with i.v. diuretics and vasodilator in the intensive care unit. The deep sleep to compensate for the sleep debt could restore RSI toward near normal levels even when patients are in the acute treatment phase of HF. Finally, factor analysis was performed using a small number of patients. To explore the pathophysiological aspects of RSI in HF, further studies on a larger scale are warranted as a next step in the monitoring of all-night RSI throughout the hospitalization period.

Conclusions

All-night RSI, a quantitative simple measure of respiratory instability, could faithfully reflect congestive signs and clinical status of HF and of the recovery from acute decompensation. RSI has clinical and economic advantages over other invasive monitoring parameters in that RSI can be easily obtained from any respiratory signal using non-attached sensor technology, imposing no burden on the patient. Further investigations are warranted to examine the applicability and feasibility of RSI as a quantitative variable to detect early deterioration of HF.

Acknowledgments

We would like to thank all of the patients who participated and their families. We are indebted to the physicians, all other co-medical staff and the Independent Data Monitoring Committee (Satsuki Fukushima, Kenichi Yoshimura and Shuji Nakano), who contributed to this study. We also thank the staff of the Clinical Research Support Center Kyushu (CReS Kyushu) for their excellent collection and management of data, assistance with the manuscript, and all other support. The present study was funded by Teijin Pharma Ltd.

Disclosures

H.A. received research funding from Teijin Pharma Limited; S.A. received research funding from Teijin Pharma Limited; S.M. received lecture fees from Teijin Pharma Limited; K.D. and N.K. received a scholarship donation from Merck & Co., Inc., Otsuka Pharmaceutical Co., Ltd., Daiichi Sankyo Company, Limited and Takeda Pharmaceutical Company Limited. The other authors declare no conflicts of interest.

Supplementary Files

Supplementary File 1

Appendix S1. List of Investigators at the Participating Medical Institutions of PRospective Study On Respiratory Stability Through Recovery Process from Deterioration of Heart Failure in Patients with Chronic Heart Failure (PROST)

Appendix S2. Lists of Central Adjudication Committee

Please find supplementary file(s);

http://dx.doi.org/10.1253/circj.CJ-18-0519

References

• 1.   Shah KS, Xu H, Matsouaka RA, Bhatt DL, Heidenreich PA, Hernandez AF, et al. Heart failure with preserved, borderline, and reduced ejection fraction. J Am Coll Cardiol 2017; 70: 2476–2486.
• 2.   Strom JB, Kramer DB, Wang Y, Shen C, Wasfy JH, Landon BE, et al. Short-term rehospitalization across the spectrum of age and insurance types in the United States. PLoS One 2017; 12: e0180767.
• 3.   Madigan EA, Gordon NH, Fortinsky RH, Koroukian SM, Piña I, Riggs JS. Rehospitalization in a national population of home health care patients with heart failure. Health Serv Res 2012; 47: 2316–2338.
• 4.   Cohn JN, Levine TB, Olivari MT, Garberg V, Lura D, Francis GS, et al. Plasma norepinephrine as a guide to prognosis in patients with chronic congestive heart failure. N Engl J Med 1984; 311: 819–823.
• 5.   Goso Y, Asanoi H, Ishise H, Kameyama T, Hirai T, Nozawa T, et al. Respiratory modulation of muscle sympathetic nerve activity in patients with chronic heart failure. Circulation 2001; 104: 418–423.
• 6.   Harada D, Joho S, Hirai T, Asanoi H, Inoue H. Short term effect of adaptive servo-ventilation on muscle sympathetic nerve activity in patients with heart failure. Auton Neurosci 2011; 161: 95–102.
• 7.   Yamada K, Asanoi H, Ueno H, Joho S, Takagawa J, Kameyama T, et al. Role of sympathoexcitation in enhanced hypercapnic chemosensitivity in patients with heart failure. Am Heart J 2004; 148: 964–970.
• 8.   The criteria committee of the New York Heart Association. Nomenclature and criteria for diagnosis of diseases of the heart and great vessels, 9th edn. Boston, MA: Little, Brown & Co 1994.
• 9.   Oldridge N, Höfer S, McGee H, Conroy R, Doyle F, Saner H (for the HeartQoL Project Investigators). The HeartQoL: Part II. Validation of a new core health-related quality of life questionnaire for patients with ischemic heart disease. Eur J Prev Cardiol 2014; 21: 98–106.
• 10.   Selzer A, Cohn K. Functional classification of cardiac disease: A critique. Am J Cardiol 1972; 30: 306–308.
• 11.   Sasayama S, Asanoi H, Ishizaka S, Miyagi K. Evaluation of functional capacity of patients with congestive heart failure. In: Yasuda H, Kawaguchi H, editors. New aspects in the treatment of failing heart syndrome. Tokyo: Springer-Verlag, 1992; 113–117.
• 12.   Friedman MM. Older adults’ symptoms and their duration before hospitalization for heart failure. Heart Lung 1997; 26: 169–176.
• 13.   Chaudhry SI, Mattera JA, Curtis JP, Spertus JA, Herrin J, Lin Z, et al. Telemonitoring in patients with heart failure. N Engl J Med 2010; 363: 2301–2309.
• 14.   Nishii N, Kubo M, Okamoto Y, Fujii S, Watanabe A, Toyama Y, et al. Decreased intrathoracic impedance associated with OptiVol alert can diagnose increased B-type natriuretic peptide: MOMOTARO (Monitoring and Management of OptiVol Alert to Reduce Heart Failure Hospitalization) study. Circ J 2015; 79: 1315–1322.
• 15.   van Veldhuisen DJ, Braunschweig F, Conraads V, Ford I, Cowie MR, Jondeau G, et al. Intrathoracic impedance monitoring, audible patient alerts, and outcome in patients with heart failure. Circulation 2011; 124: 1719–1726.
• 16.   Abraham WT, Stevenson LW, Bourge RC, Lindenfeld JA, Bauman JG, Adamson PB; CHAMPION Trial Study Group. Sustained efficacy of pulmonary artery pressure to guide adjustment of chronic heart failure therapy: Complete follow-up results from the CHAMPION randomized trial. Lancet 2016; 387: 453–461.
• 17.   Asanoi H, Harada D, Oda Y, Ueno H, Takagawa J, Ishise H, et al. Independent prognostic importance of respiratory instability and sympathetic nerve activity in patients with chronic heart failure. J Cardiol 2017; 70: 476–483.
• 18.   Brack T, Thüer I, Clarenbach CF, Senn O, Noll G, Russi EW, et al. Daytime Cheyne-Stokes respiration in ambulatory patients with severe congestive heart failure is associated with increased mortality. Chest 2007; 132: 1463–1471.
• 19.   Harada D, Asanoi H, Takagawa J, Ishise H, Ueno H, Oda Y, et al. Slow and deep respiration suppresses steady-state sympathetic nerve activity in patients with chronic heart failure: From modeling to clinical application. Am J Physiol Heart Circ Physiol 2014; 307: H1159–H1168.
• 20.   Khoo MCK, Kronauer RE, Strohl KP, Slutsky AS. Factors inducing periodic breathing in humans: A general model. J Appl Physiol Respir Environ Exerc Physiol 1982; 53: 644–659.
• 21.   Bruce EN, Daubenspeck JA. Mechanisms and analysis of ventilator stability. In: Dempsey JA, Pack AI, editors. Regulation of breathing. New York: Marcel Dekker, 1995; 285–313.
• 22.   Casey LC, Weber KT. Ventilatory system mechanics and gas exchange. In: Weber KT, Janicki JS, editors. Cardiopulmonary exercise testing. Philadelphia: Saunders, 1996; 77–91.
• 23.   Roberts AM, Bhattacharya J, Schultz HD, Coleridge HM, Coleridge JC. Stimulation of pulmonary vagal afferent C-fibers by lung edema in dogs. Circ Res 1986; 58: 512–522.
• 24.   Anker SD, Koehler F, Abraham WT. Telemedicine and remote management of patients with heart failure. Lancet 2011; 378: 731–739.