Assessment of systolic and diastolic function in heart failure using ambulatory monitoring with acoustic cardiography

Abstract

Introduction. The circadian variation of heart function and heart sounds in patients with and without heart failure (HF) is poorly understood. We hypothesized HF patients would exhibit less circadian variation with worsened cardiac function and sleep apnea. Methods. We studied 67 HF patients (age 67.4 ± 8.2 years; 42% acute HF) and 63 asymptomatic control subjects with no history of HF (age 61.6 ± 7.7 years). Subjects wore a heart sound/ECG/respiratory monitor. The data were analyzed for sleep apnea, diastolic heart sounds, and systolic time intervals. Results. The HF group had significantly greater prevalence of the third heart sound and prolongation of electro-mechanical activation time, while the control group had an age-related increase in the prevalence of the fourth heart sound. The control group showed more circadian variation in cardiac function. The HF subjects had more sleep apnea and higher occurrence of heart rate non-dipping. Conclusions. The control subjects demonstrated an increasing incidence of diastolic dysfunction with age, while systolic function was mostly unchanged with aging. Parameters related to systolic function were significantly worse in the HF group with little diurnal variation, indicating a constant stimulation of sympathetic tone in HF and reduction of diurnal regulation.

Key words::

• Heart failure
• heart sounds
• Holter
• sleep apnea
• systolic dysfunction

Abbreviations
BMI	=	body mass index
EMAT	=	electro-mechanical activation time
HF	=	heart failure
LV	=	left ventricle
LVH	=	left ventricular hypertrophy
LVST	=	left ventricular systolic time
NYHA	=	New York Heart Association
S3	=	third heart sound
S4	=	fourth heart sound
%S3	=	percentage of third heart sounds detected
%S4	=	percentage of fourth heart sounds detected

Key messages

• For the first time it has become possible to monitor systolic and diastolic function even in combination with sleep-disordered breathing in a simple way for 24 hours or more under ambulatory conditions.

• Ambulatory monitoring of heart sounds enables the assessment and study of circadian variations in a patient's diastolic and systolic heart function and, as such, can aid in the diagnosis of cardiac diseases and the titration of corresponding treatment.

• Sleep apnea is a prevalent finding in heart failure patients, and it can be detected by the herein presented simple Holter-based method.

Introduction

Little is known about the circadian variation of diastolic heart sounds and systolic time intervals in patients with and without heart failure. Doppler echocardiography is the most established tool to assess left ventricular filling patterns and systolic function but represents only a short time of hemodynamic state during the examination at rest, and it is not available nor is it practical in all ambulatory patients. In the past years, acoustic cardiography has emerged as a new technology based on the simultaneous assessment and automated analysis of ECG and heart sounds. Its utility to reliably quantify cardiac function through systolic time intervals and presence of diastolic heart sounds has been proven and shown in various publications.

Using a 10-second acoustic cardiographic recording, Marcus et al. (1) found the third heart sound (S3) to be associated with elevated left ventricular end-diastolic pressure and reduced ejection fraction, while Collins et al. (2) studied 343 patients in the emergency department with signs or symptoms of acute decompensated heart failure and found the third heart sound to be specific for primary heart failure. Roos et al. (3) defined left ventricular systolic dysfunction (LVSD) as maximum dP/dt <1,600 mmHg/s on a population of 108 patients undergoing cardiac catheterization. They found the systolic time interval, electro-mechanical activation time (EMAT; Q wave onset to the S1), to be prolonged in the group with LVSD (106 ± 21 ms versus 85 ± 11 ms) and to be superior to angiographic ejection fraction and QRS duration to detect left ventricular systolic dysfunction. In a study on 164 out-patients referred for echocardiographic examination, acoustic cardiography has also been shown to improve the performance of B-type natriuretic peptide to detect left ventricular systolic dysfunction (4). A previous acoustic cardiographic study, also using a 10-second recording technique, on 1,329 asymptomatic adults over a wide range reported the prevalence of diastolic heart sounds and systolic time intervals but did not address postural and circadian variations (5).

Acoustic cardiography is now available in a Holter device (see Figure 1), which can record continuous ECG, sound, and respiratory data for up to 48 hours on ambulatory subjects (6). Dual function ECG/sound sensors are placed on the chest in addition to three ECG electrodes, while the main unit contains a mechanical sensor for position, activity, and respiration measurements. Acoustic cardiographic parameters produced by this technique include those to assess systolic function including the presence of the S3 and EMAT (7), and those to evaluate diastolic function (the presence of an S4 which has been shown to be associated with increased left ventricular stiffness (8)). In addition, the respiration data can be used to detect periods of reduced chest movement that, when combined with respiration-induced changes in heart rate and snoring sounds, can be used to determine the presence of central or obstructive sleep apnea. In patients with heart failure, in particular, sleep apnea further complicates therapy and reduces quality of life, and causes repetitive arousals at night that impact the sympathetic nervous system and adrenergic receptors.

Figure 1. Ambulatory acoustic cardiography monitor. Placement of the recorder unit and the ECG/sound sensors. Regular ECG monitoring electrodes are placed on the right arm, left arm, and left leg locations for ECG monitoring. Combined ECG/sound sensors are placed in the precordial V3 and V4 positions. The recorder unit is either placed on the patient's chest or abdomen.

The purpose of this study is to analyze circadian variations in systolic and diastolic heart function and sleep disorders at the same time in a group of chronic and acute heart failure patients and age-matched asymptomatic control subjects. We hypothesized that the heart failure population would have less circadian variation, more sleep-disordered breathing, and worsened cardiac function particularly in the acute heart failure (HF) subgroup.

Patients and methods

Patients

The local Medical Ethics Committee approved the heart failure study performed in Lucerne, Switzerland. After obtaining informed consent, we evaluated a convenience sample of patients with a history of heart failure equal to or over the age of 50 years who were interned for a heart failure exacerbation or stable patients in the clinic. The stable, asymptomatic heart failure clinic patients will be referred to as ‘chronic’, whereas those interned with signs and symptoms of acute decompensation will be referred to as ‘acute’. The study on asymptomatic control subjects received IRB approval from Liberty Institutional Review Board, Deland, FL, USA, and informed consent was obtained. Subjects were included if in sinus rhythm and the subject was not having chest pain, syncope, or shortness of breath. The subjects were asked about their medical history, specifically any history of heart failure, and all medications they were currently taking. There were no control subjects with a history of heart failure, and none were taking medications typical for heart failure patients. Additional exclusion criteria for both studies included acute coronary syndrome, presence of significant heart murmur, active infection, and continuous intravenous therapy.

A total of 63 asymptomatic control subjects and 67 heart failure patients were recruited. There were no subjects undergoing chemotherapy. Base-line demographics, medications, and medical history, especially history of hypertension or heart failure and cardiovascular drug therapy, were recorded on all subjects. History of previous or current smoking status was not recorded. Blood pressure measurements and previous cardiac history including Doppler echocardiographic results were obtained on all heart failure patients.

Acoustic cardiographic Holter recordings

Acoustic cardiographic Holter data were collected using a small device attached to the patient's chest (see Figure 1). Standard ECG electrodes were placed in modified limb locations and dual-purpose (ECG and sound) AUDICOR® sensors (Inovise Medical, Inc., Beaverton, OR, USA) were placed in either the standard V4 position or both the V3 and V4 positions while the subject was in a supine position. Adequate data quality was confirmed by visual assessment by study personnel. Cardiac acoustic, ECG, and data from the activity/respiration sensor were recorded simultaneously for a mean duration of 16 hours (awake 8.8 ± 3.6 hours; sleep 7.7 ± 1.7 hours).

The data were analyzed for the presence of diastolic heart sounds (third (S3) and fourth (S4) heart sounds) and for systolic time intervals including electro-mechanical activation time (EMAT; onset of the Q wave to the first heart sound (S1)). The S3 and S4 strength is a combination of intensity and persistence and is expressed on a scale from 0 to 10, and if the strength is > 5.0 an S3 or S4 is considered to be present. The respiration signal was visually over-read by trained personnel (Figure 2) and episodes of apnea documented as well as presence or absence of snoring on the sound signal.

Figure 2. ECG and respiration in sleep apnea. Example of ECG (top) and respiration signal (bottom) for episodes of apnea.

Statistical analysis

Results are given as mean values ± standard deviations for continuous variables with Gaussian distribution. Categorical data are presented as exact numbers and proportions. After confirming that parameters had Gaussian distribution, we used two-tailed unpaired t tests and a priori selected P values < 0.05 to be considered statistically significant. Results were determined by age decade but not for separate genders because we did not find significant differences in diastolic heart sounds or systolic time intervals between men and women, which is consistent with the findings of others (5).

Results

Basic demographics and clinical history are summarized for each entire group and by age decade in Table I, whereas Table II contains cardiac characteristics by age decade for the heart failure population. The asymptomatic control group consisted of 63 subjects (52% male; age 61.6 ± 7.7 years) with 33% having a history of hypertension, 5% with prior myocardial infarction, none with pacemakers, and the remaining 63% with no significant medical history. Within the asymptomatic control population 39% were not taking any medications. The remainder of these subjects were taking antihypertensives (30%), statins (24%), and beta-blockers (1.3%). The heart failure group was composed of 67 patients (85% male; age 67.4 ± 8.2 years; 21 with pacemakers). All heart failure subjects had a left ventricular ejection fraction <50% (mean 30.7 ± 9.5%), 64% also had diastolic dysfunction, and 42% were in hospital for acute decompensation. The majority of heart failure subjects had a diagnosis of dilated cardiomyopathy (60%).

Table I. Demographics and clinical history.

	Asym n = 63	Chronic HF n = 37	Acute HF n = 28
Age, years	61.6 ± 7.7 (50–81)	65.6 ± 8.5 (52–82)	69.8 ± 7.6 (50–85)
% Male	52%	87%^a	75%^a
BMI, kg/m^2	27.4 ± 5.9 (19–50)	28.7 ± 4.1 (21–42)	27.5 ± 4.9 (20–39)
Medical history
None	63%	0%	0%
HTN	33%	44%	86%^a
OMI	5%	31%^a	54%^a
Pacer	0%	39%^a	25%^a
Afib	0%	25%^a	25%^a
Renal	0%	28%^a	32%^a
CP	0%	6%	4%
SOB	0%	3%	22%^a
DM	0%	36%^a	29%^a
Other	0%	33%^a	39%^a

Mean ± standard deviation, range (min–max).

^aP value < 0.05 across asymptomatic control to heart failure.

Afib = atrial fibrillation; Asym = asymptomatic subjects; HF = heart failure patients; BMI = body mass index; HTN = hypertension; OMI = old myocardial infarction; Pacer = pacemaker; CP = angina; SOB = shortness of breath; DM = diabetes mellitus.

Table II. Clinical characteristics of heart failure subjects, by age decade.

	All n = 67	50 ≤ Age < 60 n = 11	60 ≤ Age < 70 n = 28	Age ≥ 70 n = 28
Ejection fraction, %	30.7 ± 9.5	34.8 ± 8.3	29.3 ± 8.5	30.6 ± 10.6
NYHA class:
Class 1	4%	9%	0%	7%
Class 2	43%	45%	46%	39%
Class 3	42%	36%	39%	47%
Class 4	6%	0%	7%	7%
Unknown	4%	10%	8%	–
Acute HF, %	42%	18%	32%	61%
LVH	66%	64%	71%	61%
Cardiac diagnosis:
CAD	64%	45%	54%	82%
VHD	48%	45%	39%	57%
DCM	60%	64%	68%	50%
HHD	16%	0%	14%	25%
Sleep apnea	43%	27%	50%	43%
Nocturnal HR < 10% change	74%	54%	81%	74%

Mean ± standard deviation.

LVH = left ventricular hypertrophy; CAD = coronary artery disease; VHD = valvular heart disease; DCM = dilated cardiomyopathy; HHD = hypertensive heart disease.

Heart rate significantly decreased at night in all age-groups in the asymptomatic control population but in none of the heart failure age-groups, and the heart rate at night in the asymptomatic control population was significantly lower as compared to the heart failure patients in all age-groups. Within the asymptomatic control population there were 19% with untreated sleep apnea and 9.5% with nocturnal heart rate that decreased less than 10% from their day-time heart rate (nocturnal non-dipping). Within the heart failure group there were 43% with untreated sleep apnea and 74% with nocturnal heart rate that decreased less than 10% from their day-time heart rate, neither significantly different from the chronic to the acute heart failure groups.

QRS duration did not increase with age in the asymptomatic control group nor change from day to night. QRS duration was significantly longer in the heart failure populations and also did not change from day to night. QRS duration showed no power in discriminating the following conditions within the heart failure population: acutely decompensated heart failure, sleep apnea, heart rate non-dipping, left ventricular hypertrophy (LVH), diastolic dysfunction, history of hypertension, or New York Heart Association (NYHA) class 2 or 3/4.

Parameters reflecting systolic function

Electro-mechanical activation time (EMAT) in the asymptomatic control subjects had little circadian variation, with a non-significant change from awake to sleep across age-groups and little change with increasing age (mean 89 ± 9 ms day/night and across groups). In the heart failure group, EMAT increased steadily with increasing age and was significantly higher as compared to the asymptomatic control population in all but the youngest age decade (Figure 3). Used previously as an index of systolic function, EMAT divided by left ventricular systolic time (LVST; time from S1 to S2) showed little variation with age in the asymptomatic control population and was consistently lower at night. This index was significantly higher in the heart failure population, demonstrated an increasing trend with age, and only showed a slight decrease at night in the youngest group.

Figure 3. Electro-mechanical activation time. Electro-mechanical delay (EMAT; time from Q onset to the S1, ms) for each age decade, day and night for acute heart failure (red), chronic heart failure (blue), and asymptomatic (black) subjects. *P value < 0.05 across control to heart failure.

The third heart sound was significantly more prevalent in the heart failure population as compared to the asymptomatic control subjects both day and night in all age decades (Figure 4).

Figure 4. Percentage of S3 detected (S3 strength ≥ 5.0, %) in each age decade, day and night for acute heart failure (red), chronic heart failure (blue), and asymptomatic (black) subjects. *P value < 0.05 across control to heart failure; ⁺P value < 0.05 across acute to chronic heart failure.

Parameters reflecting diastolic function

The prevalence of the fourth heart sound (S4) was significantly greater at night in the asymptomatic control subjects and demonstrated an increasing trend with age (Figure 5).

Figure 5. Percentage of S4 detected (S4 strength ≥ 5.0, %) in each age decade, day and night for acute heart failure (red), chronic heart failure (blue), and asymptomatic (black) subjects. *P value < 0.05 across control to heart failure; ⁺P value < 0.05 within control group across day and night.

Comparison of asymptomatic subjects to chronic or acute heart failure patients

The heart failure patients consisted of 28 that were hospitalized for an exacerbation of their symptoms and 37 stable, chronic heart failure patients. The remaining 2 patients were not classified as acute or chronic. Table III contains a comparison of measurements comparing the asymptomatic control subjects to these two subgroups. Heart rate was significantly higher in the acute HF patients. EMAT and %S3 were significantly higher in both heart failure populations as compared to the asymptomatic control group with %S3 at 25% in the acute heart failure group compared to 3% in the asymptomatic control group and 13% in the chronic HF group.

Table III. Comparison of asymptomatic subjects to acute or chronic heart failure patients.

	Asymptomatic n = 63	Chronic HF n = 37	Acute HF n = 28
Heart rate, bpm	71.0 ± 9.1	74.4 ± 14.0	81.7 ± 13.8^a
EMAT, ms	89.2 ± 9.4	122.0 ± 29.4^a	118.0 ± 24.3^a
S3, %	3.0 ± 5.0	13.0 ± 20.0^a	25.0 ± 22.0^a,b
S4, %	16.8 ± 18.0	10.0 ± 14.0	13.0 ± 22.0
Sleep apnea	19%	39%	46%
Nocturnal non-dipping	9.5%	74%	75%

^aP value < 0.05 heart failure groups compared to asymptomatic population.

^bP value < 0.05 chronic heart failure compared to acute heart failure.

EMAT = electro-mechanical activation time.

Variability analysis

Coefficients of variation were calculated for each population over the second hour of sleep. In the heart failure group the coefficient of variation of heart rate was 19.5%, while for the asymptomatic control group it was 14.5%. The coefficient of variation of EMAT was 24.2% in the HF population and 20.5% in the asymptomatic control group. The third and fourth heart sounds had the greatest coefficients of variation (S3: 40.1% heart failure, 42.8% asymptomatic; S4: 48.7% heart failure, 36.2% asymptomatic). The high coefficients of variation in diastolic heart sounds likely reflect changes in preload/afterload and diastolic filling patterns, particularly with sleep apnea.

Discussion

This study highlights changes in parameters reflecting systolic function, diastolic filling patterns, and their change with heart failure. In the asymptomatic control population we found little age-related change in parameters related to systolic function. Similar to the findings of Roos et al. (3) in subjects without left ventricle (LV) systolic dysfunction, electro-mechanical activation time was constant at an average of 89 ms for age 50 years and older, and with little awake-to-sleep variation. Voutilainen et al. (9) studied circadian variation of left ventricular function in healthy people. They found that systolic function was less dependent on the time of day, but that diastolic function did include a nocturnal decrease and a day-time increase in the rate of left ventricular relaxation that they attributed to sympathoadrenal activity. The findings in this study on electro-mechanical activation time support the conclusion of less variation from day to night in systolic function in asymptomatic control subjects.

In the heart failure group, EMAT increased steadily with increasing age and was significantly higher as compared to the asymptomatic control group. The increased EMAT in the heart failure patients was related to NYHA class (2 versus class 3 or 4), both awake and asleep (all P values < 0.05). De Oliveira (10) studied 55 patients with heart failure due to systolic dysfunction and 60 normal subjects and found an increase in EMAT in patients with heart failure that was correlated with the symptomatic status of the patient and degree of ventricular dysfunction. Another study that quantified EMAT on heart failure patients by Rahko et al. found a delay in the mitral and tricuspid closure related to the degree of systolic dysfunction and elevated filling pressure (11). Thus, prolongation of EMAT reflects worsening systolic dysfunction and is reflected in symptoms experienced by the patient.

The third heart sound is caused by an abrupt limitation of left ventricular inflow during early diastole that causes vibration of the cardiohemic system. The low prevalence of S3 in asymptomatic control subjects over the age of 40 years (less than 5% mean awake value) strengthens the findings of others that a detectable S3 in older subjects is specific for cardiac pathology (12) and is associated with elevated left ventricular filling pressures (7). In our data there was a significantly higher prevalence of the awake S3 in those heart failure patients with NYHA class 3 or 4, as compared to those with NYHA class 2 (P = 0.02). Also, in those patients with acute decompensated heart failure there was a significantly higher proportion of the S3 both awake (P = 0.01) and asleep (P = 0.004).

The fourth heart sound is produced by the abrupt deceleration of the A wave and a stiff left ventricle. The prevalence of the fourth heart sound was found to be as high as 73% in healthy persons in earlier phonocardiographic studies (13–15), and its significance was debated. More recent studies using acoustic cardiography have found lower prevalence of the fourth heart sound (2) and its presence to be associated with increased left ventricular end-diastolic stiffness (8), impaired relaxation (16), and stress-induced ischemia (17). Our findings in the asymptomatic control group suggests an age-related increase in the prevalence of the S4 similar to age-related changes in left ventricular filling patterns determined using echocardiography. In a randomly recruited population of 539 people, Kuznetsova et al. (18) measured the prevalence of left ventricular diastolic dysfunction using echocardiography and found a significant decline in the E/A ratio with age due to a significant decrease in E velocity as well as an increase in A velocity. The finding of altered E and A velocities with aging is consistent with an increase in the prevalence of the S4 with age, particularly in those over the age of 60 years.

Since there was a proportion of asymptomatic control subjects with a history of hypertension currently being treated, we analyzed the data by dividing the population into those with and without a history of hypertension. There were no significant differences in diastolic heart sounds or systolic time intervals between those groups, which we attribute to their presently controlled blood pressures. When the heart failure population was divided into those with and without LVH, those with LVH had significantly more S4s at night (P = 0.01) and almost reaching statistical significance during the day (P = 0.06). This increase in the prevalence of the S4 with LVH is most likely due to increased stiffness of the ventricle as well as altered diastolic filling patterns.

Sleep apnea accelerates the progression of heart failure through intermittent apnea-induced surges in sympathetic tone and LV afterload, day-time hypertension, and loss of vagal heart rate regulation (19). The prevalence of sleep-disordered breathing was much lower in the asymptomatic control population (19%) than in the heart failure group (43%). In the asymptomatic group the sleep-disordered breathing was associated with snoring and therefore obstructive in nature, while in the heart failure group the sleep-disordered breathing was not always correlated to the presence of snoring and believed to be central in nature. Sleep apnea in the heart failure patients was not associated with age, ejection fraction, presence of diastolic dysfunction, LVH, or BMI, but was significantly associated with increased NYHA class. Javaheri found a similar prevalence (45%) of central sleep apnea in a heart failure population (20). Sleep apnea in the asymptomatic control population was significantly associated with increased age and obesity.

Several recent studies have shown that heart rate, and non-dipping in sleep heart rate in particular, might be an important predictor of cardiovascular and non-cardiovascular risk (21,22). One study on 457 patients being treated or evaluated for hypertension found the future risk of cardiovascular disease to be 2.4 times higher in those whose heart rate does not dip at least 10% below their day-time values (23). We found a much higher prevalence of nocturnal non-dipping in the heart failure population (74%) as compared to the asymptomatic control group (9.5%). Within the heart failure group there were significantly more patients with a history of coronary artery disease in the non-dipping group (P = 0.005), but we found no association of heart rate non-dipping to history of hypertension, left ventricular hypertrophy, or ejection fraction. The lack of diurnal variation in heart rate and other systolic and diastolic parameters indicates a constant stimulation of the sympathetic tone in heart failure and a reduction of typical diurnal regulation.

We experienced very few problems with the ambulatory device for home monitoring. For the first time it has become possible to monitor systolic and diastolic function even in combination with sleep-disordered breathing in a simple way for 24 hours or more under ambulatory conditions. This allows not only the analysis of systolic and diastolic dysfunction during longer periods than only during office time but in addition the monitoring of heart function and sleep disorders concurrently. In addition, this methodology has the potential to analyze efficacy of medical treatment over a long period or to study alterations such as administration of continuous positive airway pressure (CPAP) on cardiac function. The use of a simple device such as that used in the study to diagnose and monitor treatment of sleep apnea, administration and titration of medications, or modification of devices such as cardiac resynchronization therapy (CRT) timing in heart failure patients is certainly warranted.

Limitations of the study

The asymptomatic, presumably healthy control population had limited information acquired on their cardiac function, compliance to medications, use of herbal and life-style drugs, and other factors that may have impacted the findings of this study. The asymptomatic control and heart failure populations were predominantly Caucasian and geographically confined to the west coast of the United States and Lucerne, Switzerland. The small sample size of the heart failure population does not allow us to demonstrate the ability of the device to discriminate chronic from acute heart failure, nor was this study designed to specifically track heart failure patients as they progress into acute decompensation. Since diastolic dysfunction is more common with increased age there were likely subjects in the control group with some degree of diastolic dysfunction.

Conclusion

This study found a higher incidence of the fourth heart sound in subjects without heart failure that increased steadily with age, likely due to increased prevalence of altered diastolic filling patterns associated with impaired relaxation. In contrast, a higher incidence of sleep apnea, nocturnal non-dipping, and a third heart sound both during the day and night was found in the heart failure subjects, as well as higher EMAT reflecting a reduction in contractility. Patients with acute decompensation if compared to chronic heart failure elicit a higher heart rate and higher incidence of the S3. This study demonstrates continuous monitoring of cardiac function and respiration is possible and worthwhile using a convenient ambulatory device.

Declaration of interest: Dr Arand is an employee of Inovise Medical Inc.