Correlation between cardiorespiratory coupling and six-minute walking tests in patients with chronic obstructive pulmonary disease

Background : Cardiovascular disease is a common comorbidity and cause of mortality among patients with chronic obstructive pulmonary disease (COPD). However, the interaction between the heart and lungs in COPD patients has yet to be fully elucidated. Aim : Our objective in this study was to characterize cardiorespiratory interactions in terms of cardiorespiratory coupling (CRC) using the synchrogram index of the heart rate and respiration flow signals. Methods : This prospective study examined 10 normal subjects and 55 COPD patients. Linear regression and forward stepwise regression were used to determine the correlation between the synchrogram index and the six-minute walking test. Results : K-means clustering analysis was used to separate the 55 COPD patients into a synchronized group (median 0.89 (0.64-0.97), n=43) and a desynchronized group (median 0.23 (0.02-0.51), n=12) based on the synchrogram index. In this study, the synchrogram index was significantly correlated with the six-minute walking distance (r 2 =0.3, sigma T=0.02) and the distance saturation product (r 2 =0.3, sigma T =0.03). Note that age was a significant confounding factor. Conclusion : The synchrogram index shows clinical potential for the stratification of COPD patients for treatment.


Introduction
Chronic obstructive pulmonary disease (COPD) is characterized by irreversible airflow limitations resulting from the destruction of the alveolar attachment. [1] Possible causes include exposure or inhalation of noxious gases or particles and systemic inflammation. COPD has been shown to affect nutritional status, exercise tolerance, and the cardiovascular system. [2] The prevalence of cardiovascular disease as a comorbidity and cause of mortality underlines the need to consider the interaction between the heart and lungs in patients with COPD. [3][4][5] Heart-lung interactions can be classified according to the underlying mechanisms: (i) respiratory sinus arrhythmia, (ii) cardioventilatory coupling, and (iii) respiratory stroke volume synchronization. [6] During inspiration, central inspiratory drive [7] and negative intrathoracic pressure [8] both contribute to an increase in heart rate. [9] Negative intrathoracic pressure promotes filling of the right ventricle and impedes filling of the left ventricle. [10] A decrease in arterial blood pressure tends to increase respiratory rate and tidal volume through the baroreflex. [11] Several studies have described heart rate variability (HRV) and its relationship with the respiratory system among subjects with COPD. Researchers have found that cardiac autonomic dysfunction is more prevalent among COPD patients than among normal subjects, due presumably to inspiratory muscle weakness [12,13] and poor health-related quality of life. [14] Note however that heart-lung interactions in COPD subjects have not been fully elucidated. Synchrogram is a widely applied signal processing tool that facilitates the detection of coupling between two oscillatory signals. [15,16] It has been used to study the heart-lung interactions in normal subjects [17], infants [18], and schizophrenia patients. [19] In addition, it is applied to identify sleep stages [20] in patients with obstructive sleep apnea. Essentially, this analysis can be divided into two steps: (1) detecting the instantaneous phase; and (2) quantifying the degree of phase coupling. The term instantaneous phase describes the oscillatory status of a waveform signal at a given point in time. The degree of coupling is calculated by obtaining the synchrogram of the instantaneous phases of two oscillatory signals. In this prospective study, we explored the phase coupling of the instantaneous heart rate and respiratory flow signal with the aim of elucidating an important mechanism underlying heart-lung interactions (i.e., respiratory sinus arrhythmia) in patients with COPD. [20] Methods

Study Design and Patients
This prospective study included normal subjects and patients with a clinical diagnosis COPD, based on the Global Initiative for Obstructive Lung Disease Criteria (GOLD). The study was conducted at Chang Gung Memorial Hospital during January and December in 2019. Patients with heart failure (ejection fraction <40%), known malignancy, or atrial fibrillation as well as those using anti-arrhythmic agents for arrhythmia or oxygen were excluded. All COPD patients underwent cardiac echo analysis, biochemical analysis (eosinophils, high sensitivity C-reactive protein (HS-CRP), and IgE), pulmonary function tests, a six-minute walking test, and a coupling test during the first visit. Clinical profiles, a list of inhalation medicines, and acute exacerbation history [21] were also recorded ( Figure 1). A total of 55 patients with COPD (69 (51-84) years old, 54 male) and ten normal subjects (64.5 (50-71) years old, 5 male) were enrolled. All participants signed informed consent prior to enrollment. The study was approved by the Ethics Committee of Chang Gung Memorial Hospital, Linkou, Taiwan (201702150B0).

Instrumentation
Experiments were performed in a quiet room with the temperature maintained at 22-24 °C.
Participants were instructed to avoid bronchodilators, such as beta-2 agonists, xanthene derivatives, alcohol, and coffee prior to the test. The chest skin was abraded using gel and then cleaned using alcohol to reduce electrode impedance prior to the attachment of electrocardiogram (ECG) electrodes.
Prior to the examination, recordings were obtained of blood pressure, heart rate, and oxygen saturation. The subjects wore a pulse oximeter on the index finger and ECG electrodes on the chest wall. To avoid signal quality issues, the patient practiced breathing through the flow tube for 1 min before recording. ECG signals and flow signals were then recorded continuously for 5 min using three Actiwave devices (CamNtech Ltd, Cambridge, UK). The recorded signals were transferred in European Data Format to LabChart 8 software (ADInstruments, Dunedin, New Zealand), and then exported to text files for analysis.

Signal Processing and Synchrogram index
R peaks were detected using a standard R peak detection algorithm, and the time differences between consecutive R peaks were converted into an instantaneous heart rate (IHR) time series using a standard interpolation algorithm. [22] Instantaneous phases of the respiratory signals and IHR time series were extracted using the synchrosqueezing transform (SST). [23] The time-varying broad pass issue associated with the Hilbert transform was avoided using the generalized SST form. After obtaining the instantaneous phases of the IHR (denoted as Φ H ) and the respiratory signal (denoted as Φ R ), the synchrogram was used to quantify phase synchronization. [17,24] Note that a synchrogram is a graphical tool used to depict phase coupling between two oscillatory signals. In the current study, we first obtained the timestamps t k at the point where the IHR phase attained 0 (i.e., Φ H (t k ) mod 2π = 0). We then measured the respiratory phase at t k as follows: Plotting Ψ(t k ) against t k made it possible to obtain a horizontal stripe in cases where the IHR and respiratory signals were synchronized; otherwise, we obtained scattered points. Cardiorespiratory synchronization was quantified by measuring the synchrogram index λ [24] as follows: where M is the number of detected cycles in the IHR. The synchronization index was closer to 1 in synchronized systems and closer to 0 in unsynchronized systems.

Statistical Analysis
All results are presented as median (range) or mean ± standard deviation. The nonparametric exact two-tailed Mann-Whitney U test was used to determine the statistical significance between two groups of continuous variables, and Fisher's exact tests were used for categorical variables. Variance differences were evaluated using the F-test. Pearson's correlation was used to examine the association between distance, distance saturation product, and synchrogram index. Forward stepwise multiple regression analysis of the distance and distance saturation product was used to evaluate the contribution of age, body mass index, and synchrogram index. All reported P values were two-sided, with P < 0.05 considered statically significant. All data were analyzed using R version 3.5.2 (R foundation for statistical computing).

Comparing coupling tests with six-minute walking test
Patients in the synchronized group presented longer walking distances (468 (328-624) vs. 408 (182-517), unit=m, p =0.009) and a higher distance saturation product (421.2 (255.6-536.6), vs 373.2 (149.2-464.6), unit=m%, p=0.02) ( Table 3). The synchrogram index correlated significantly with distance (r=0.42, p =0.0012) (Figure 3c) and distance saturation product (r=0.4, p= 0.002) ( Figure   3d). Also note that even after discounting the effects of age and BMI, the synchrogram index remained a significant factor correlating with distance and distance saturation product. (Supplement 1, 2) Discussion CRC has been widely studied in normal subjects [17] and infants [28], and as an indication of sleep stages; [20,29] however, it has not been examined in the context of COPD. In this study, we compared COPD patients and normal subjects using the results of synchronization analysis. We analyzed the interaction between IHR and respiratory signals. We also used the synchrogram index to quantify the interaction between the heart and lungs. The synchrogram index distribution was wide among COPD patients and narrow among normal subjects. Using k-means clustering to classify COPD patients as synchronized or desynchronized, we found that the synchrogram index distribution in the synchronized group was similar to that among normal subjects.
This study led to two major findings. The distance and distance saturation product from six-minute walking tests were significantly lower in the desynchronized group than in the synchronized group.
Even after discounting the effects of age and BMI in a forward stepwise regression model, the synchrogram index remained a significant factor correlating with distance and distance saturation product.
Researchers have previously demonstrated that six-minute walking distance is an important predictor of survival in COPD patients [30,31] and heart failure patients. [32] Our study revealed that walking distance was lower among desynchronized patients; however, these results could not be confirmed from the perspective of outcomes due to inadequate follow-up duration. The poor walking distance demonstrated by COPD patients can be attributed to desaturation, the severity of emphysema, dyspnea scores, and inspiratory capacity as determined by ECLIPSE (Evaluation of COPD longitudinally to identify predictive surrogate endpoints). [33] Note that poor six-minute walking distance was also observed in patients presenting with heart failure with preserved ejection function.
[34] Del Buono et al. reported that several comorbid conditions, such as skeletal muscle dysfunction, impaired autonomic regulation, and nutritional factors, may coexist and contribute to exercise intolerance in patients suffering from heart failure. [35] It has also been reported that impaired aerobic function contributes to low exercise intolerance in patients with COPD and in those suffering heart failure, due to negative cardiopulmonary-muscular interaction. [36] Desaturation levels, Borg score, inspiratory capacity, flow limitations (FEV 1 , % predicted), and dynamic hyperinflation (Δ inspiratory capacity) were similar in these two groups, and these showed no connection to walking distance in this study. This is an indication that walking distance is independent of desaturation and impaired pulmonary or heart function. Instead, it appears that nutritional status, peripheral muscle condition, [37] oxygen utilization by peripheral muscle, and negative cardiopulmonary-muscle interactions [36] should be taken into account.
Distance saturation product (i.e., the product of nadir saturation during exercise and walking) is a known predictor of mortality among patients with bronchiectasis [38] and interstitial lung disease.
[39] Nonetheless, it has not yet been discussed in the context of COPD. In this study, patients in the desynchronized group presented a lower distance saturation product, which suggests an elevated likelihood of poor outcomes. Our results indicate that nadir saturation was not correlated with the synchrogram index, suggesting that distance contributed to distance saturation product but not desaturation. A strong heart-lung interaction may improve ventilation and perfusion matching, resulting in better oxygen transport; [40] however, we did not observe any discrepancy between the synchronized and desynchronized groups in terms of saturation. This may be explained by the fact that we excluded patients who were using oxygen on a daily basis. Note that there may be a link between desaturation and coupling in those patients. To further study this relationship, we should explore COPD patients who used oxygen daily. There is also the possibility that the patients stopped walking prior to the onset of desaturation. To examine this possibility, we need more data and carry out the subgroup analysis in patients who have poor performance in walking distance to evaluate the cause of stopping walking.
This study faced a number of limitations. First, despite measuring and quantifying the coupling between respiration flow signals and ECG signals, we found no indications of causality. Second, the strict inclusion criteria prevented us from analyzing patients using oxygen on a daily basis. Third, most of the patients in the study were male and all were of East Asian decent; i.e., this sample is not representative of COPD patients overall. Finally, the follow-up time of this preliminary study was of insufficient duration to evaluate mortality outcomes and cardiac vascular events. We will continue monitoring the subjects in this study in order to observe the clinical impact of synchronization in heart-lung interactions.

Conclusions
This study conducted CRC analysis to describe respiratory sinus arrhythmia, a mechanism underlying heart-lung interactions in COPD patients. CRC is significantly correlated with six-minute walking distance and distance saturation product. Note that age was a significant confounding factor. The