Fifteen sedentary males with smoking-related COPD1 were enrolled in this observational, clinical, cross-sectional study. All patients were evaluated in an outpatient specialized clinic by the same physician and optimally treated before study initiation for at least 3 months (Table 1). All patients presented with a clinical and spirometric diagnosis of COPD (FEV1/forced vital capacity <0.7 and post-bronchodilator FEV1<80% predicted) at stages II, III, or IV.1 All subjects reported suffering from chronic dyspnea (modified Medical Research Council scale score). No decompensation episodes occurred in any enrolled subject for at least one month prior to study initiation and no subjects had participated in a regular physical exercise program during the previous six months. The exclusion criteria included long-term O2 therapy (for at least 6 months), type I or non-controlled type II diabetes mellitus and peripheral vascular disease, orthopedic/rheumatological/neurological conditions that would preclude participation in the protocol, uncontrolled hypertension, other concomitant respiratory diseases, current alcoholism, use of theophylline, history of cardiac arrhythmias or potential electrocardiogram alterations, exercise-induced asthma, unstable angina, acute myocardial infarction during the previous six months, syncope, any contraindication to exercise testing according to American Heart Association guidelines,17 and inability to understand and cooperate with the procedures. Approval was granted by the Medical Ethics Committee of São Paulo Hospital, São Paulo, Brazil (protocol 473.529/2013). All subjects were informed about study objectives, experimental procedures and potential risks and gave written informed consent before study initiation.

All patients were submitted to a comprehensive evaluation, divided into three days: (1) clinical evaluation by a physician and a physical therapist, followed by lung function tests (resting blood gases, spirometry, static lung volumes – SLV, and diffusion capacity of the lung for carbon monoxide) and Doppler echocardiography; (2) HRV data collection (stand position) and 6MWT; (3) incremental CPET.

All subjects were evaluated during the same time of the day (from 1pm to 5pm) in order to avoid differences due to circadian rhythm. They were instructed to abstain from caffeinated and alcoholic beverages and to avoid exercise on the day before data collection.

Lung function: Spirometry, gas transfer, and SLV (1085 ELITE DTM, Medical Graphics Corporation, St. Paul, MN, USA) were measured according to American Thoracic Society/European Respiratory Society guidelines.18–20 Resting blood gases were obtained by samples from the radial artery.21

Doppler echocardiography: All individuals were submitted to a 2D echocardiogram using an iE33 system (Philips, Andover, MA, USA) with a 2–5MHz matrix transducer and tissue Doppler imaging software. Patients were studied in the left lateral decubitus position by the same physician. Quantification of the cardiac chambers was performed according to the American Society of Echocardiography.22

Acquisition of R-R interval (RRi): RRi was registered using the Polar® system (Polar® S810i, Kempele, Finland), at rest in the standing position (10min), and during 6MWT. An elastic belt (Polar T31 transmitter, Polar Electro, Kempele, Finland) was attached to the subject's chest at the level of the lower third of the sternum. The belt contains a stable case with HR electrodes, an electronic processing unit, and an electromagnetic field transmitter. The HR signals are continuously transmitted to the Polar Advantage receiver unit via an electromagnetic field.23 All data were transferred to a computer using Polar Pro-Trainer 5TM® software.

HRV analysis: HRV was analyzed using Kubius HRV® software (MATLAB, version 2.1, Kuopio, Finland). The total period of RRi collection was scrutinized and, within a 10-minute period in the standing position and in the final minute of the 6MWT, the most stable noise-independent segment (i.e., without missing data and/or noise events), with at least 256 points,16 was selected for analysis. Time domain, frequency domain, and non-linear analysis were performed at rest and during the 6MWT. The mean RRi, the standard deviation of the normal RR intervals (SDNN), and the square root of the mean squared differences of successive RR intervals (RMSSD) were obtained for time domain linear analysis. LF and HF (normalized units – nu) included frequency domain HRV indices. The LF/HF ratio was calculated to verify the sympathovagal balance.16 Non-linear HRV analysis was performed from SD1 (standard deviation measuring the dispersion of points in the plot perpendicular to the line-of-identity), SD2 (standard deviation measuring the dispersion of points along the line-of-identity), alpha1 and alpha2 (respectively, short and long-term fluctuations of detrended fluctuation analysis), and approximate entropy (ApEn) indices. SD1 is related to parasympathetic activity, while SD2 reflects total variability. Alpha1 and alpha2 were used to quantify the fractal property of the RRi temporal series. In healthy conditions, alpha1 should be close to 1 and higher than alpha2.24 ApEn detects changes in a time series and provides a non-negative number to the series. Higher values indicate higher complexity.25 In order to verify responses due to sympathetic stimuli (exercise), HRV indices were also expressed in delta (Δ) (Δ rest-6MWT=HRV index at rest minus HRV index during 6MWT).

6MWT: Subjects were asked to walk at their own maximal pace, without running, along a 30m perimeter. Patients were allowed to stop, but they could start again, if possible, within the allocated 6min.10 Oxyhemoglobin saturation by pulse oximetry (SpO2 – %, Nonin Medical, Plymouth, MN) and HR were continuously recorded. Breathlessness and leg effort were rated according to the 10-point Borg category-ratio scale at the 6th min.26 The test was performed twice, with at least 30min between tests, and the best 6MWD was recorded.

CPET: Symptom-limited, incremental (5 or 10W) exercise testing was performed on an electronically braked cycle ergometer (Corival 400, Lode BV, Netherlands) using the Vmax229d Cardiopulmonary Exercise Testing System (SensorMedics).27V˙O2, L/min, V˙CO2, L/min, respiratory exchange ratio RER=V˙CO2/V˙O2, V˙E, L/min, tidal volume (VT, ml), end tidal carbon dioxide tension (PETCO2, mmHg), and respiratory frequency (breaths/min) were obtained breath-by-breath and averaged as 20s bins. SpO2 was continuously recorded. Breathlessness and leg effort were rated according to the 10-point Borg category-ratio scale at the peak of exercise.26 The V˙E/V˙CO2 relationship from the beginning of exercise to peak exercise was derived via least squares linear regression (i.e., y=mx+b, m=slope) (Microsoft Excel, Microsoft Corp., Bellevue, WA, USA).28 The OUES was calculated using the log-transformation (base 10) of V˙E (L/min) on the x-axis and V˙O2 (L/min) in the y-axis.29 CP (mmHgmlO2/kg/min) was defined as the product of peak V˙O2 and peak SBP.5 EVP (mmHg) was defined as peak SBP divided by the V˙E/V˙CO2 slope.6

The patients’ baseline characteristics and functional variables at rest and at peak exercise are reported as means±standard deviation. According to variable distribution (Shapiro–Wilk test), Pearson's or Spearman's moment correlation coefficients were used to test the association between HRV indices (dependent variables) and 6MWT and CPET variables (independent variables). The magnitude of correlations was determined considering the following classification scheme for r-values: (0.26–0.49: low or weak; 0.50–0.69: moderate; 0.70–0.89: strong or high; 0.90–1.0: very high).30 Considering a moderate association of 0.6, the calculated effect size for the current sample was 0.7.31 Post hoc analysis revealed a power of 0.93 for the current sample at an α level of 0.05 (GPower software, version 3.1.3, Germany). All statistical analysis was conducted at a 95% level of significance. Statistical analysis was performed using the Sigma Plot for Windows, version 11.0 (Sigma Plot, San Jose, CA, USA).

Twenty-six COPD patients from a specialized outpatient clinic (sample of convenience) were recruited. Eleven did not fulfill the inclusion criteria [long-term O2 therapy (1), recent decompensation episode (4), current alcoholism (1), participation in pulmonary rehabilitation (1), refusal to participate (3), unable to perform movement of cycling (1)]. The patients’ baseline characteristics are shown in Table 1. Table 2 shows functional variables obtained at rest, during 6MWT, and during CPET.

Several moderate, statistically significant correlations ranging from −0.53 to −0.65 (not shown) were observed between CPET variables (peak relative V˙O2, peak SpO2, peak SBP, peak DBP, and peak VT) and ΔHRVrest-6MWT (ΔHF, ΔLF, ΔSD2, and ΔLF/HF). Other positive, moderate, statistically significant correlations ranging from 0.58 to 0.65 (not shown) were also found between CPET variables (peak relative V˙O2, peak SpO2, and peak DBP) and ΔHRVrest-6MWT (ΔHF, ΔLF, Δα1, and ΔLF/HF). Linear indices (LF and HF) correlated with CPET variables obtained directly (ΔSpO2, respectively, r=−0.64 and r=0.65, p<0.05) and indirectly (V˙E/V˙CO2 slope, respectively, r=−0.52 and r=0.53, p<0.05; EVP, respectively, r=0.52 and r=−0.53, p<0.05). Moderate correlations were observed between alpha1 (Δrest-6MWT) and ΔSpO2, OUES, and EVP (respectively, r=−0.56, r=0.51 and r=0.57, p<0.05). Fig. 1 illustrates the positive, moderate correlation between the CP and ΔLF/HF rest-6MWT (r=0.52, p<0.05).

Figure 1.

Relationship between sympathovagal balance variation (Δ low- to high-frequency ratio – LF/HF) from rest to submaximal exercise (during the six-minute walk test – 6MWT) with hemodynamic responses during maximal exercise testing represented by circulatory power (product of peak oxygen uptake and peak systolic blood pressure). This result suggests that the modulation of heart rate responses on submaximal intensities is linked to cardiovascular responses during maximal exercise testing. *p<0.05.

(0.07MB).

Table 3 shows the correlation coefficients between 6MWT variables and HRV indices obtained at rest, during exercise (6MWT), and in the transition from rest to submaximal exercise (Δrest-6MWT). A strong, statistically significant correlation (r=0.82, p<0.05) was observed between 6MWD and ΔLF/HF rest-6MWT (Fig. 2).

Figure 2.

Strong relationship between sympathovagal balance variation (Δ low to high frequency ratio – LF/HF) from rest to submaximal exercise (during the six-minute walk test – 6MWT) with walking distance covered. This data suggests that the modulation of heart rate response is linked to the performance of COPD patients. *p<0.05.

(0.06MB).

Leite et al.15 demonstrated resting HRV assessment can become a tool to predict exercise capacity in COPD patients. The authors showed that autonomic cardiac function assessment at rest (sitting position) had a weak, although statistically significant, correlation with variables obtained during CPET performed on a treadmill. Since posture interferes with HRV,16 the patients in the present study were assessed at rest, in the same posture in which 6MWT was carried out, i.e., standing position. In addition, we believe that HRV indices expressed in delta (Δrest-6MWT) characterize HRV response to a given stimulus (sympathetic), which is coherent with the expected response during CPET. Therefore, many correlations were found between CPET variables and HRV indices Δrest-6MWT.

The positive SpO2 variations (i.e., Δ>0) in this sample correlated with more adequate cardiac autonomic responses (ΔLF<0 and ΔHF>0) due to sympathetic stimulation (exercise). Since exercise-related oxygen desaturation is common in COPD patients as the disease advances4 and given that hypoxia has been implicated as a direct cause of peripheral nerve damage in COPD,32 this might explain why those who presented desaturation during exercise also showed poorer cardiac autonomic responses than those who did not. Statistically significant correlations between EVP and ΔLF and ΔHF showed adequate response of both frequency bands, from rest to submaximal exercise, corresponded to lower EVP values, and therefore poorer response to aerobic capacity. It seems to be a contradictory relationship; however, most of the subjects included in this study (n=13) presented with signs of air trapping at rest (augmented residual volume – RV – and RV/total lung capacity ratio) and we speculate they also presented with dynamic hyperinflation during exercise. Therefore, we might infer the elevated starting volume and also high work of breathing at rest and during exercise could have caused parasympathetic overactivation and reset the autonomic cardiac control to a new level, causing a “false positive” response from rest to exercise. In addition, it is known that the V˙E/V˙CO2 slope reflects disease severity and its prognosis, with values below 30 being considered normal.4 EVP combines ventilatory efficiency with systemic hemodynamics during exercise (peak SBP/V˙E/V˙CO2 slope ratio), two important physiologic measures related to the ability to respond to aerobic exertion. Impaired EVP reflects deteriorating right heart function and pulmonary hemodynamics33 and, in CHF patients, an EVP below 3.5mmHg is a strong independent predictor of cardiac events.6 Although an EVP threshold has not been established for COPD patients, the majority of patients in our sample presented with low EVP values, possibly due to a high V˙E/V˙CO2 slope (above 30) suggesting poor response to aerobic exercise.

The current results showed alpha1 changes from rest to submaximal exercise correlated with ΔSpO2, OUES, and EVP. The causes of these relationships cannot be elucidated in this study, thus deserve further investigation; however, in the elderly population, changes in alpha1 characterize a powerful predictor of sudden cardiac death, whose causes appear to be explained by sympathetic overactivation.34 OUES, an objective, effort-independent measure to estimate cardiopulmonary functional reserve, may be influenced by changes in partial pressure of CO2, V˙CO2, and physiological dead space,29 abnormalities present in COPD patients.3 It seems the analysis of nonlinear HR dynamics from rest to submaximal exercise may provide relevant clinical information regarding greater hypoxemia and both lower ventilatory efficiency and lower ventilatory power during exercise in moderate-to-severe COPD patients.

In patients with heart disease, functional status can be assessed using CP, a parameter that summarizes the responses of HR, left ventricular systolic volume, arterial blood pressure, and arteriovenous oxygen difference to exercise.5 In CHF patients, the peak CP seems to be a better prognostic marker than peak V˙O2 alone.5 In the present study, patients with higher CP values presented positive ΔLF/HF rest-6MWT (Fig. 1), which may represent sympathovagal balance changes from rest to exercise, situation in which sympathetic predominance and reduced vagal stimulation are expected responses.

A widespread test because of its simplicity, convenience, ease of performance, and low cost, the 6MWT provides important clinical information as the severity of COPD increases.3 The study of Pinto-Plata et al.35 showed 6MWD as a better mortality predictor than other COPD severity markers, such as FEV1 and body mass index. Reduced 6MWD represents a fundamental daily inactivity marker in COPD.36 Our study revealed that HRV measurement at rest can help to infer exercise capacity in COPD patients. Both linear (LF/HF) and nonlinear (SD2) HRV indices correlated with subjective data obtained during 6MWT (peak breathlessness) (Table 3). Although van Gestel et al.37 did not observe a correlation between resting HRV and the variables obtained during the 6MWT, the authors suggested that resting HR can be an independent predictor of exercise capacity in COPD patients.

Possibly the most clinically relevant data of our study involved the HRV complexity analysis using the ApEn index. The loss of HRV complexity characterizes greater regularity in RRi dynamics, which reflects poorer adaptability to physiological stress.38 Our results showed moderate statistically significant negative correlation between ApEn and 6MWD, both at rest and during exercise (Table 3), which may indicate that a system with loss of complexity, i.e., with greater regularity and predictability, interferes with exercise performance. We also suggest checking changes in linear HRV indices (LF and HF) Δrest-6MWT (Table 3) for additional information to infer exercise capacity in COPD patients. The strong correlation observed between sympathovagal balance (ΔLF/HF rest-6MWT) and 6MWD (Fig. 2) suggests the faulty adjustment in cardiac autonomic control might be one of the contributors to poor exercise performance in COPD patients.

Our results add clinically relevant information to exercise capacity appraisal and cardiac autonomic function assessment in COPD patients. We believe that not only HRV analysis at rest but also its investigation after stimulus (e.g., exercise sympathetic stimulation from 6MWT) may reflect exercise ventilatory and hemodynamic abnormalities as well as contribute to relevant information concerning cardiac autonomic nervous system modulation. We suggest that HRV complexity analysis using the ApEn index can compose HRV evaluation at rest, in which a less complex system (i.e., smaller values of ApEn) correlates with poor 6MWT performance. It is worth highlighting this trend can be modified after six weeks of aerobic exercise training in COPD patients.39