Cardiac Sympathetic Dominance and Systemic Inflammation in COPD

Abstract

Cardiac autonomic dysfunction is an independent determinant of adverse outcomes in many diseases. The available literature on the relative changes in sympathetic and parasympathetic components in chronic obstructive pulmonary disease (COPD) is equivocal, the clinical and physiological correlates are poorly defined and association with markers of systemic inflammation has not been explored. As both autonomic dysfunction and systemic inflammation may contribute to cardiovascular morbidity in COPD, we hypothesized that these may be associated. Sixty three stable patients of COPD and 36 controls underwent spirometry, estimation of diffusion capacity, six-minute walk test and measurements of serum interleukin-6 (IL-6) and high-sensitivity C-Reactive protein. Cardiac autonomic activity was evaluated by standard five-minute heart rate variability (HRV) recordings to obtain time- and frequency-domain indices and the averaged heart rate. We observed that HRV indices of overall autonomic modulation, the standard deviation of time intervals between consecutive normal beats (SDNN) and total power, were greater in patients with higher levels of indices of both parasympathetic and sympathetic activity. The heart rate was significantly higher in patients indicating an overall sympathetic dominance and was inversely correlated with diffusion capacity. Serum IL-6 was inversely correlated with pNN50, an index of parasympathetic activity, and positively with LF/HF ratio, a measure of sympathetic: parasympathetic balance. None of the HRV indices was significantly correlated with physiological measures of severity. It was concluded that patients with COPD have increased cardiac autonomic modulation with sympathetic dominance. This is associated with decreased lung diffusion capacity and systemic inflammation.

Keywords:

• autonomic dysfunction
• chronic obstructive pulmonary disease
• inflammation
• heart rate
• heart rate variability
• sympathetic

Introduction

COPD is a major cause of death and disability worldwide and has shown increasing trends in mortality (1). Cardiovascular involvement is a major reason for mortality and is believed to be causally associated with systemic inflammation (2, 3). Cardiac autonomic dysfunction has been demonstrated in COPD in ­several studies, including one by our group, though results on the relative balance of the sympathetic and parasympathetic components have differed (4–16).

Heart rate variability (HRV), the oscillation in the intervals between consecutive heart beats, is an established noninvasive tool to study autonomic activity. The methodology for its measurement has been standardized (17). It yields several indices in time and frequency domains that quantify sympathetic and parasympathetic modulation of heart rate. Normal cardiac heart rate modulation has a parasympathetic dominance (18). Sympathetic dominance in HRV, i.e. a shift in the balance towards the sympathetic, marks a poor prognosis in diabetes, heart failure, renal failure and post-myocardial state by predisposing to arrhythmias and sudden death (19–23).

A neurohumoral association has been proposed in COPD and other disease states with sympathetic overactivity promoting endothelial dysfunction and systemic inflammation (24, 25). Whereas HRV measurements are technically exacting that therefore precludes their routine clinical application, the resting heart rate is an excellent surrogate for bedside evaluation of autonomic balance with an increased rate reflecting sympathetic dominance. Similar to the adverse prognostic implications of a sympathetic shift in HRV, increased heart rate also has been shown to be associated with increased mortality in cardiac diseases and in COPD (26, 27). Thus, it may independently provide clinically useful information as a prognostic marker.

Apart from the inconsistency in literature on the autonomic balance, the clinical and physiological correlates of cardiac autonomic dysfunction in COPD are poorly defined and its association with markers of systemic inflammation has not been explored. As both sympathetic dominance and systemic inflammation may contribute to cardiovascular morbidity in COPD, we hypothesized that these may be associated. Therefore, we investigated cardiac autonomic dysfunction using HRV and heart rate measurements and examined its clinical and physiological correlates, and, association with markers of systemic inflammation.

Materials and Methods

Sample size calculations were based on the magnitude of high frequency component of HRV in COPD and healthy controls reported in literature (10). To obtain similar difference as significant with a power of 80%, a requirement of 52 cases and 26 controls was calculated. Sixty-three stable male patients of COPD, all ex-smokers, diagnosed and under treatment according to the GOLD guidelines (28) were included in the study. Patients with chronic hypoxemia (documented by arterial blood gas analysis or pulse oximetry), a recent exacerbation during the last 4 weeks, diabetes mellitus, and evidence of any co-existent pulmonary, cardiac or other systemic disease, including anaemia, were excluded.

The study was limited to male patients as COPD in Indian women is mostly due to exposure to biomass fuels and therefore sufficient numbers of female smoker patients were not available within the study period. Current treatment included inhaled long- and short-acting bronchodilators, anticholinergics and corticosteroids according to GOLD guidelines-defined severity (28). Thirty-six age-matched healthy volunteers, 18 smokers and 18 nonsmokers, served as controls. A written informed consent was obtained from the study participants. The study was approved by the Institutional Ethics Committee.

After a detailed history and physical examination, blood counts, biochemistry panel, plain chest radiograph and a 12-lead standard electrocardiogram were obtained and a two-dimensional and color doppler echocardiography was carried out to ensure compliance with the inclusion ­criteria. The patients underwent spirometry using standardized procedure recommended by the American Thoracic Society-European Respiratory Society task force guidelines (29) on a dry rolling-seal spirometer (Benchmark lung function equipment, P.K. Morgan, Kent, UK), measurement of single-breath diffusion capacity for carbon monoxide (DLCO_SB), and a six-minute walk test (30).

DLCO_SB was divided by alveolar volume (V_A) to obtain the Krogh's constant, K_CO. The severity of COPD was categorized according to the GOLD criteria (28). The BODE index, a validated marker of severity and predictor of mortality and morbidity in COPD was calculated (31). It is a composite score based on body-mass index, degree of airflow obstruction, British Medical Research Council grade of dyspnea, and the six-minute walk distance index. Ten ml of venous blood was withdrawn and serum was separated and stored at −80ºC for later measurements of high-sensitivity C-reactive protein (hs-CRP) and interleukin-6 (IL-6). These were measured on an ELISA reader using commercial kits (Assaypro, Saint Charles, Missouri, USA). Smoking was not allowed on the day of the investigations for smoker controls.

Following withdrawal of inhaled bronchodilators for 24 hours, a 5-minute HRV was recorded using a standardized procedure (17). Recordings were obtained around the same time of the day in all subjects in a quiet, dimly-lit and comfortable room maintained around 22–25°C. ­Caffeinated/alcoholic beverages were not permitted on the day of the test. A standard lead II ECG was recorded for twenty minutes. The analog signal was fed into a bioamplifier, digitized using an analog-to-digital converter and displayed on a computer with the data acquisition and analyzing software (Power Lab 4/30 data acquisition system with Lab Chart Pro 7.0, AD Instruments-Australia). The recordings were manually examined and regions with more than 1% ectopic beats or 5% artifacts were excluded from the analysis. The average heart rate over 5 minutes was calculated. Time domain and frequency domain analyses using non-parametric Fast Fourier Transform (FFT) were performed following task force standards (17).

In the time-domain analysis, the following indices were measured: (i) SDNN: standard deviation of the time intervals between consecutive normal beats, i.e., the NN intervals, a global index of HRV; (ii) RMSSD: root mean square of successive differences between adjacent NN intervals, a marker of parasympathetic activity; and (iii) pNN50: percentage of number of NN interval with difference ≥50 ms, a marker of parasympathetic activity. In frequency-domain analysis, the following were measured: (i) Total power (TP): the area under the spectral curve from 0.01 to 0.4 Hz, a global index of HRV; (ii) LF, power in the low frequency band, i.e., 0.04–0.15 Hz: reflects both sympathetic and parasympathetic activity; (iii) LFnu, normalized low frequency power: LF divided by the balance of TP minus very low frequency; (iv) HF, power in the high frequency band, i.e., 0.15–0.4 Hz: reflects the parasympathetic activity; (v) HFnu, normalized high frequency power: HF divided by the balance of TP minus very low frequency; and (vi) LF/HF ratio, low to high frequency power ratio averaged: reflects sympathetic:parasympathetic balance. A minimum of three artefact-free segments with average heat rate and SDNN within 5% of each other were selected. HRV indices described above were obtained from these and mean was taken for analysis.

Statistical analysis was carried out using the Statistical Package for Social Sciences (SPSS, Chicago; IL, USA), version 16.0 and Graph Pad Prism 5.01 softwares. Data are presented as mean ± sd. The data was examined for normality of distribution using the Shapiro-Wilk test. Frequency domain indices of HRV had skewed distribution. Therefore, natural logarithmic transformation was done to normalize the data. Student's unpaired t-test was used for two-group (patient-control) comparisons. Multiple groups were compared by analysis of variance (ANOVA) with post hoc Bonferroni's test for between-group comparisons. Levene's test for homogeneity of variances was performed to check the appropriateness of ANOVA. Correlation between HRV indices and parameters of disease severity was evaluated using the Pearson test. A p value of < 0.05 was considered statistically significant.

Results

The demographic characteristics and lung function data of patients and controls are shown in Table 1. The groups were matched for age. The smoker and nonsmoker controls had similar lung function. Three (4.8%) patients had mild, 33 (52.4%) had moderate, 25 (39.7%) had severe and 2 (3.2%) had very-severe COPD. The patients were also categorized into four quartiles according to the BODE index score. The patient distribution was: 1st quartile (score < 2), 14 (22.2%); 2nd quartile (score = 2), 13 (20.6%); 3rd quartile (score = 3), 14 (22.2%); and 4th quartile (score = ≥ 4), 22 (34.9%). All the patients had normal blood counts, serum chemistries and ECG. None of the patients had any valvular disease or wall motion abnormality, and had normal chamber dimensions and left ventricular ejection fraction on echocardiography.

Table 1.  Demographic and lung function tests data

	COPD	Nonsmoker Controls	Smoker Controls
	(n = 63)	(n = 18)	(n = 18)
Age	55.5 ± 5.8^ns	52.4 ± 8.9	53.1 ± 5.8
BMI	20.5 ± 3.9^***	26.0 ± 3.8	22.184 ± 2.4
Pack years of smoking	39.4 ± 27.5^***	—	19.2 ± 15.2
FEV1/FVC ratio	44.6 ± 11.0^***	77.4 ± 5.4	78.1 ± 6.3
FEV1% predicted	54.1 ± 16.7^***	106.5 ± 24.0	105.0 ± 13.9
DLCOSB% predicted	66.2 ± 23.4^***	101.9 ± 20.7	105.2 ± 18.5
KCO% predicted	79.3 ± 28.3^***	112.1 ± 18.3	110.0 ± 17.3

Data expressed as mean ± SD; Comparisons of three groups using ANOVA, ***p < 0.0001, ns: not significant; FEV₁/FVC ratio and FEV₁% predicted are post-bronchodilator values.

The 5-minute averaged heart rate (beats per minute, bpm) was significantly higher in patients (83.7 ± 11.9) compared to controls (nonsmoker controls: 76.3 ± 7.5, and smoker controls: 72.8 ± 9.4) (ANOVA, p < 0.0001, Figure 1). Post hoc comparisons showed higher heart rate in patients than in nonsmoker (p < 0.01) and smoker controls (p < 0.01) but similar heart rates in the two control groups.

Figure 1. The 5-minute averaged heart rate (beat per minute, bpm) in patients and controls (ANOVA p < 0.0001).

The internal consistency of the measurement of HRV parameters was established by the strong correlations between SDNN and total power (r = 0.93, p < 0.0001), both reflecting the overall modulation of the NN intervals, and between the time domain and frequency domain indices reflecting parasympathetic activity, i.e., RMSSD and HF (r = 0.88, p < 0.0001), and between pNN50 and HF (r = 0.66, p < 0.0001).

Comparison of HRV parameters in the three groups (Table 2) showed significant differences in SDNN (p < 0.05) and RMSDD (p < 0.05) in time domain parameters, and total power (p < 0.05), low frequency (LF) modulation (p < 0.0001), and high frequency (HF) modulation in frequency domain parameters (p < 0.05). Post-hoc comparisons revealed significant greater values of SDNN (p < 0.05), RMSDD (p < 0.05), total power (p < 0.05), low frequency (LF) modulation (p < 0.0001), and high frequency (HF) modulation (p < 0.05) in patients compared to smoker controls. The two control groups did not differ in these parameters. When the two control groups were combined, the differences with patients in these indices remained significant (Figure 2).

Figure 2. HRV parameters (SDNN, RMSDD, total power, LF modulation and HF modulation) in patients and controls. SDNN: standard deviation of all the NN intervals; RMSSD: root mean square of successive differences between adjacent NN intervals; LF modulation: power in the low frequency band, i.e., 0.04–0.15 Hz; HF modulation: power in the high frequency band, i.e., 0.15–0.4 Hz; Units: SDNN and RMSDD in ms and total power, LF modulation and HF modulation in ms²; values are natural log- transformed; *p < 0.05, ***p < 0.001, + 0.05 > p < 0.10.

Table 2.  Heart rate variability indices in the three groups

HRV Indices	COPD	Nonsmoker Controls	Smoker Controls
	(n = 63)	(n = 18)	(n = 18)
Time domain indices
Ln SDNN (ms)	3.8 ± 0.5*	3.73 ± 0.5	3.48 ± 0.5
Ln RMSSD (ms)	3.66 ± 0.6*	3.46 ± 0.8	3.15 ± 0.8
Ln pNN50	1.58 ± 0.9^ns	1.48 ± 1.1	1.44 ± 1.2
Frequency domain indices
Ln Total Power (ms^2)	7.68 ± 1.0*	7.34 ± 1.2	6.94 ± 0.9
Ln LF (ms^2)	6.54 ± 1.3^***	6.04 ± 1.2	5.19 ± 0.9
Ln LFnu (ms^2)	3.86 ± 0.7^ns	3.64 ± 0.5	3.63 ± 0.6
Ln HF (ms^2)	6.14 ± 1.4*	6.06 ± 1.1	5.14 ± 1.4
Ln HFnu (ms^2)	3.52 ± 0.7^ns	3.59 ± 0.3	3.59 ± 0.5
Ln LF/HF Ratio	0.19 ± 1.0^ns	0.05 ± 0.6	0.06 ± 1.1

Data shown in natural log transformed values expressed as mean ± SD, Comparisons of three groups using ANOVA, ***p < 0.0001, *p < 0.05, ns: not significant; Between-group comparisons presented in text under the results section; SDNN (standard deviation of all the NN intervals), RMSSD (root mean square of successive differences between adjacent NN intervals), pNN50 (percentage of number of NN interval with difference ≥ 50 ms), Total power (area under the spectral curve from 0.01 to 0.4 Hz), LF (power in the low frequency band 0.04–0.15 Hz), LFnu (low frequency power divided by the balance of TP-very low frequency), HF (power in the high frequency band i.e. 0.15–0.4 Hz), HFnu (high frequency power divided by the balance of TP-very low frequency), LF/HF ratio (low to high frequency power ratio).

Table 3 shows the HRV indices by GOLD severity and by quartiles of BODE index scores. Due to insufficient numbers in mild and severe groups, the patients were re-classified into two broad GOLD severity groups: mild-to-moderate and severe-to-very severe. RMSSD was significantly higher in the latter (p < 0.05) while other indices of HRV and heart rate were not different between these two groups. None of the indices of HRV or the heart rate differed significantly across the BODE index quartiles.

Table 3.  Heart rate and heart rate variability indices by GOLD severity and by quartiles of BODE index scores

HRV Indices	GOLD Severity		BODE Index Quartiles
	Mild-Moderate	Severe-Very Severe	Q1	Q2	Q3	Q4
Heart Rate (beat/minute)	82.1 ± 12.1	85.8 ± 11.4^ns	82.9 ± 10.8	84.4 ± 7.3	82.7 ± 14.5	84.4 ± 13.5^ns
Ln SDNN (ms)	3.73 ± 0.5	3.9 ± 0.4^ns	3.76 ± 0.4	3.68 ± 0.7	3.92 ± 0.5	3.83 ± 0.3^ns
Ln RMSSD (ms)	3.52 ± 0.7	3.85 ± 0.5*	3.56 ± 0.4	3.49 ± 0.8	3.71 ± 0.8	3.79 ± 0.4^ns
Ln pNN50	1.42 ± 0.9	1.8 ± 0.8^ns	1.36 ± 0.9	1.34 ± 0.7	1.86 ± 1.0	1.68 ± 0.8^ns
Ln Total Power (ms^2)	7.5 ± 1.0	7.92 ± 0.9^ns	7.6 ± 0.9	7.31 ± 1.3	7.96 ± 1.1	7.78 ± 0.8^ns
Ln LF (ms^2)	6.33 ± 1.3	6.83 ± 1.3^ns	6.53 ± 1.1	5.97 ± 1.6	7 ± 1.4	6.59 ± 1.24^ns
Ln LFnu (ms^2)	3.82 ± 0.5	3.92 ± 0.9^ns	3.83 ± 0.4	3.8 ± 0.6	3.92 ± 0.4	3.88 ± 1.1^ns
Ln HF (ms^2)	5.89 ± 1.6	6.46 ± 1.2^ns	6.06 ± 1.1	5.65 ± 1.8	6.41 ± 1.7	6.31 ± 1.2^ns
Ln HFnu (ms^2)	3.4 ± 0.8	3.67 ± 0.4^ns	3.55 ± 0.4	3.12 ± 1.2	3.53 ± 0.6	3.73 ± 0.5^ns
Ln LF/HF Ratio	0.31 ± 1.0	0.04 ± 0.8^ns	0.58 ± 0.7	0.11 ± 1.3	0.39 ± 0.9	−0.14 ± 0.9^ns

Data shown in natural log transformed values expressed as mean ± SD, *p < 0.05, ns: not significant; GOLD severity categories compared by Students's t-test, BODE index score quartiles compared by ANOVA. SDNN (standard deviation of all the NN intervals), RMSSD (root mean square of successive differences between adjacent NN intervals), pNN50 (percentage of number of NN interval with difference ≥ 50 ms), Total power (area under the spectral curve from 0.01 to 0.4 Hz), LF (power in the low frequency band 0.04–0.15 Hz), LFnu (low frequency power divided by the balance of TP-very low frequency), HF (power in the high frequency band, i.e., 0.15–0.4 Hz), HFnu (high frequency power divided by the balance of TP-very low frequency), LF/HF ratio (low to high frequency power ratio).

The heart rate was found to have significant negative correlations with DLCO_SB% predicted (r = −0.34, p < 0.01; Figure 3) and K_CO% predicted (r = −0.32, p < 0.05) but not with post-bronchodilator FEV₁, 6-minute walk distance, and the BODE index score. None of the HRV indices was found to have a significant correlation any of these variables.

Figure 3. Relationship between heart rate and single-breath diffusion capacity for carbon monoxide (DLCO_SB% predicted).

Figure 4 shows the levels of serum hs-CRP (mg/L) and IL-6 (pg/ml). The hs-CRP levels were significantly different among the groups (p < 0.01): COPD 5.63 ± 2.1, nonsmoker controls 3.93 ± 2.3 and smoker controls 3.68 ± 1.9 with significantly higher levels in patients compared to the either controls (p < 0.05). Similarly, the IL-6 levels were significantly different among the groups (p < 0.05): COPD 3.74 ± 1.6, nonsmoker controls 3.14 ± 2.2 and smoker controls 2.16 ± 1.5. Patients had significantly higher levels than either controls (p < 0.05). The two control groups did not differ in the levels of either hs-CRP or IL-6. Combining the two control groups and comparing with patients yielded the same results for both parameters.

Figure 4. Levels of serum C-reactive protein (CRP) expressed in mg/L, and interleukin 6 (IL-6) expressed in pg/ml, in patient and control groups. (ANOVA p < 0.05 for both).

While the CRP levels were not found to be significantly related to any of the indices of HRV, IL6 levels were inversely correlated with pNN50 (r = −0.28, p < 0.05) and positively with LF/HF ratio in patients (r = 0.26, p < 0.05).

Discussion

Our results show that patients with COPD have greater overall modulation of heart rate (increased SDNN and total power) with higher levels of activity of both parasympathetic (RMSDD and high frequency modulation) and sympathetic (low frequency modulation) components of the HRV spectrum compared to controls. Overall, the autonomic balance is shifted towards the sympathetic, as shown by the higher heart rate in patients. The heart rate was correlated inversely, though modestly, with the diffusion capacity. Except for the higher RMSSD level in severe and very severely affected patients, neither the heart rate nor any of the other HRV indices were associated with the several clinical and physiological measures of disease severity. The serum markers of systemic inflammation, hs-CRP and IL-6, were increased in patients. A measure of parasympathetic activity, pNN50, had a significant negative correlation with serum IL-6 levels while LF/HF ratio, a measure of sympathetic/parasympathetic balance, had a direct correlation with it. These observations suggest that sympathetic activity in COPD is associated with systemic inflammation.

Autonomic dysfunction has been documented in COPD using different tools (4–16), including in a study from our laboratory (8). However, the observations on the relative shift in the sympathetic-parasympathetic balance have been inconsistent and even conflicting. Gross et al (4) studied atropine-induced bronchodilatation in COPD and concluded that cholinergic tone was increased in proportion to the severity of airways disease. Indirect evidence as well as clinical drug trials suggests that bronchomotor tone is under parasympathetic dominance in COPD (5). However, outside the lungs, parasympathetic autonomic neuropathy was shown in chronically hypoxemic COPD patients by Stewart et al. using bedside tests of cardiovascular autonomic responses and acetylcholine sweat-spot test (6, 7). When the same tests were used in normoxemic patients, we found abnormalities in both parasympathetic and sympathetic responses (8). With reference to the tonic central autonomic drive, muscle sympathetic nerve activity has been reported to be increased in patients with chronic respiratory failure, including those with COPD (9).

Studies on cardiac autonomic modulation in COPD using HRV analysis have also yielded similarly varying results. Volterrani et al. (10) observed a decreased global modulation at rest and on passive head-up tilt, and an increased HF power suggesting increased vagal activity that was not related to the severity of airway disease. Pagani et al. (11) demonstrated a reduced resting HRV with a maintained spectral power distribution suggesting isolated decreased sympathetic modulation while Stein et al observed decreased vagal modulation (12). In another study in patients with chronic hypercapnic respiratory insufficiency, depressed HRV with decreased sympathetic and parasympathetic components was found that was partially corrected by oxygen (13).

However, Tug et al. (14) using sympathetic skin responses and HRV observed a wide spectrum of autonomic dysfunction including isolated parasympathetic dysfunction, sympathetic dysfunction or a mixed disorder in a majority of patients with COPD. Autonomic dysfunction was independent of airflow limitation and hypoxemia. Further, Chen et al. (15) observed enhanced cardiac vagal activity and depressed sympathetic activity in COPD patients with chronic hypoxemia that were not associated with the degree of airway narrowing. Most recently, Raupach et al. (16) reported significantly elevated sympathetic nerve activity in normoxemic COPD patients.The varying and conflicting results in these studies may be due to small sample sizes, differences in severity of disease (presence or absence of hypoxemia), and experimental conditions of measurement (at rest or under autonomic stimulation), end-organ studied and the techniques used to evaluate the autonomic activity.

Although we found increased heart rate modulation with increases in both sympathetic and parasympathetic indices on HRV, the increased heart rate suggested that on balance, there is a sympathetic dominance in COPD. This is in agreement with other reports of sympathetic overactivity in hypoxemic (9) and normoxemic patients (16). Interestingly, a higher heart rate was noted in patients with COPD (32) almost three decades ago though not in the context of autonomic control. More recently, similar observations were made in patients with COPD enrolled in the Copenhagen City Heart Study using heart rate measurements on ECG (26). There is also some evidence supporting increased sympathetic activity in COPD from studies that have used other tools to evaluate autonomic balance (33). Although it was not associated with the severity of airways obstruction, the inverse correlation of heart rate with diffusion capacity raises a possibility that autonomic dysfunction may be a characteristic of the emphysema phenotype of COPD. This requires investigation.

Several mechanisms may modulate sympathetic activity in COPD including oxidative stress, systemic inflammation, hypoxia, physical inactivity, and large intrathoracic pressure changes (25). Increased oxidative stress has been well-documented in COPD, including in a study from our laboratory (34). There is evidence that oxidative stress plays a role in sympathetic excitation in hypertension and heart failure (35, 36). An association between markers of inflammation and sympathetic overactivity has been shown in healthy subjects as well as in patients with heart failure (36, 37). Our results too suggest an association between serum IL-6 and sympathetic activity. The design of the present study however does not allow us to establish a causal relationship between the two.

The mechanisms of interaction between inflammation and autonomic function are not well established. Further, no such association was found with hs-CRP. However, ours is the first study to suggest an association between a marker of systemic inflammation and sympathetic activity in patients with COPD. Whether and how autonomic dominance and systemic inflammation, individually or synergistically, promote cardiovascular morbidity remains to be established. Interestingly, Jensen et al. observed that resting heart rate was associated with markers of chronic inflammation but remained associated with both cardiovascular and all-cause mortality even after adjusting for these thus suggesting that it was an independent risk factor for mortality, and not merely a marker of chronic inflammation (27).

Other possible mechanisms for increased sympathetic activity include impaired left ventricular function that is known to occur in patients with COPD (38), hypoxaemia and medication. These were ruled out as our patients did not have clinical or echocardiographic evidence of cor pulmonale or left ventricular systolic dysfunction, and were clinically stable and normoxaemic. Drugs that may modulate heart rate such as long acting beta-agonists and anti-cholinergics were withdrawn 24 hours before the study. Smoking was unlikely to be causally related to sympathetic dominance as the differences between smoker and nonsmoker controls were not significant for any of the study parameters.

Although indices of parasympathetic activity were also found increased, this likely represents a counter-regulatory attempt, though apparently inadequate, to balance the increased sympathetic activity. This reasoning has support from our recent report of a delayed first-minute heart rate recovery after exercise in patients with COPD. A delayed heart rate recovery after exercise is a sign of inadequate parasympathetic response following sympathetic activation (39). Nevertheless, while it fails to fully counter the increased sympathetic stimulus to the heart, the increased parasympathetic outflow to the lungs would increase the bronchomotor tone and mucosal gland secretions that are both under vagal control. This hypothesis is also consistent with the evidence of increased cholinergic tone in the airways in COPD (4, 5). It also provides an explanation for the efficacy of anticholinergic drugs in the management of COPD.

The study has important clinical implications. Adverse cardiovascular effects of sympathetic excitation that are relevant in COPD include endothelial dysfunction, arterial stiffness, left ventricular hypertrophy and supraventricular and ventricular arrhythmias (25). This may explain the epidemiological observations on the association of elevated heart rate with increased cardiovascular mortality in subjects with heart disease (27) and also across all stages of COPD (30). Sympathetic dominance may also potentiate cardiac toxicity of long-acting beta agonists, an area of concern with the use of these drugs. Heart rate is a more intuitive and much simpler clinical tool than a formal and technically difficult HRV measurement. Nevertheless, it appears to be an acceptable surrogate of the autonomic balance and provides similar prognostic information.

Sympathetic overactivity in COPD also offers a potential avenue for therapeutic intervention. Beta-blockers have been found to reduce all-cause mortality in patients with COPD (40) as well as in those undergoing vascular surgery (41). Finally, slow breathing has been shown to reduce sympathoexcitation in patients with COPD (16) and may therefore be of value as a component of pulmonary rehabilitation for such patients.

A limitation of the study was the inclusion of only males, for reasons discussed earlier. The observations may therefore not be extrapolated to female subjects with COPD. However, there is no evidence in literature of any gender differences in autonomic dysfunction in COPD. There were very few patients with mild COPD. This may have prevented an adequate exploration of the relationship between severity of COPD and cardiac autonomic dysfunction. Whereas an inverse correlation between the heart rate and diffusion capacity suggested that patients with emphysema may have a greater ­sympathetic dominance, confirmatory evidence by high-resolution computed tomography was not available. A question may arise whether the study was underpowered as the calculated sample size of the controls was 26 and we included 18 subjects each in smoker and nonsmoker control groups. This was ruled out as the two control groups did not differ in the study parameters and when the two control groups were combined to get the right sample size, the results of analysis were not different.

Conclusions

In conclusion, there is increased modulation of heart rate in stable patients with COPD with an overall sympathetic dominance. This is associated with systemic inflammation and decreased diffusion capacity. Whether cardiac sympathetic dominance and associated systemic inflammation promote cardiovascular morbidity and portend an adverse prognosis requires confirmation.

Declaration of Interest Statement

The authors state that they have no financial, consulting, and personal relationships with other people or organizations that could influence (bias) the authors’ work. There are no conflicts of interest.

Acknowledgments

We thank the Department of Science and Technology, Government of India for technical approval and financial support for this study.