The Relationship Between Cough Reflex Sensitivity and Exacerbation Frequency in Chronic Obstructive Pulmonary Disease

Chronic obstructive pulmonary disease (COPD) exacerbations are acute episodes of sustained deterioration in respiratory symptoms beyond normal daily variation [1]. Such episodes are associated with impaired quality of life [2], accelerated decline in lung function [3] and increased health care utilisation [4]. The identification of individuals with COPD who are more susceptible to exacerbation may lead to targeted preventative therapies. Increased cough and/or sputum volume or purulence are common symptoms of COPD exacerbations [5, 6]. Independent authors have reported that higher levels of both cough symptoms and sputum production in the stable state are associated with exacerbation frequency [2, 7, 8].

The mechanism of cough in general is complex and not fully understood [9, 10]. One potential mechanism is the activation of bronchopulmonary C fibres [9]. C fibres are relatively quiescent in health but can be activated or sensitised by inflammatory mediators [11, 12]. For example, cough reflex sensitivity to capsaicin, a potent activator of C fibres, is heightened during viral upper respiratory tract infection [13, 14]. The changes in cough reflex sensitivity during acute exacerbation of COPD have not previously been investigated.

The aim of this pilot study was, firstly, to determine if cough reflex sensitivity is heightened during exacerbation of COPD and, secondly, to investigate an association between cough reflex sensitivity and exacerbation frequency. Thirdly, we sought to determine the repeatability of cough reflex sensitivity measurements in patients with stable COPD.

This was a prospective observational study of participants hospitalised with exacerbation of COPD (winter seasons of 2017/2018) and participants with stable COPD (April to August 2017), conducted at a single site (King’s College Hospital, Denmark Hill, London, UK) in accordance with the principles of the Declaration of Helsinki. The study was granted research ethics committee approval (Health Research Authority South East Scotland Research Ethics Service, 16/SS/0189). All participants provided written informed consent.

All participants underwent investigations at two timepoints, six weeks apart. At timepoint 1, demographic and anthropometric data were collected, and participants underwent spirometry, inhaled capsaicin tussive challenge and 24-h cough frequency monitoring. Participant-reported subjective assessments of COPD health status, cough severity and cough-specific health status were also performed. At timepoint 2, all previous measures from timepoint 1 were repeated, and the participants also underwent body plethysmography and measurements of carbon monoxide transfer factor coefficient.

Cough reflex sensitivity (CRS) was assessed according to the European Respiratory Society (ERS) guidelines [19]. Capsaicin solution (Sigma-Aldrich, Missouri, USA) was delivered in incremental doubling doses (0.49–1000 µmol L⁻¹) at 1-min intervals, as 10-µL single breath inhalations, via an air-powered dosimeter (KoKo Digidoser, nSpire Health Inc, Longmont, CO, USA). To diminish the effect of anticipation, 0.9% saline solution was interspersed randomly [19]. A single characterised nebuliser (DeVilbiss Healthcare, Port Washington, NY, USA) with an output of 1.232 mL min⁻¹ was employed for all participants throughout the study. In addition, a valve limited inspiratory flow to 0.5 L s⁻¹ [19, 20]. The number of coughs elicited by each dose administration was counted for 15 s following each dose inhalation with the aid of an audio recorder (ICD-PX333, Sony Corporation, Tokyo, Japan) [19, 20]. The challenge test continued until ≥ 5 coughs were elicited by a single-dose administration. The threshold concentrations of capsaicin required to elicit 1 (C1), 2 (C2) and 5 (C5) coughs were calculated by interpolation [21].

Cough frequency was assessed objectively over 24 h with the Leicester Cough Monitor (LCM) [22]. The LCM is a validated ambulatory system, which comprises a MP3 recorder (ICD-PX333, Sony Corporation, Tokyo, Japan), a free-field lapel microphone (LFH9173, Philips, Amsterdam, Netherlands) and a semi-automated cough detection software. Coughs were detected as characteristic sounds, whether occurring in isolation or as part of a bout of prolonged coughing [22]. Total cough count (number of coughs over 24 h), mean cough frequency (coughs h⁻¹ over 24 h), awake cough counts (number of coughs over estimated time spent awake) and awake 24-h cough frequency (coughs h⁻¹) were recorded. Participants recorded and reported their time spent asleep.

Spirometry, body plethysmography and diffusing capacity of carbon monoxide (Jaeger MS-PFT Analyser Unit with Sentry Suite software version 2.19.96, Vyaire Medical, Mettawa, IL, USA) were measured as per guidelines by the ERS and the American Thoracic Society [26‐28].

The distribution of data was assessed using the D’Agostino-Pearson test. Parametric data were summarised with means (standard deviation, SD), whilst non-parametric data with medians (interquartile range, IQR). Capsaicin challenge and cough frequency data were presented as geometric mean (geometric standard deviation, SD). Parametrically and non-parametrically distributed paired data were analysed with paired t test and the Wilcoxon matched-pairs signed rank test, respectively. Fisher’s exact test and Chi-squared test were used for categorical data. Correlations between variables were assessed with Pearson product-moment correlation coefficient (r) for parametric data, and Spearman’s rank-order correlation coefficient (ρ) for non-parametric data. Repeatability was analysed with the Bland–Altman method, and with intraclass correlation coefficients (ICC) based on a single-rater, absolute agreement, two-way mixed-effects model. p values < 0.05 were considered statistically significant with the exception for the analysis of factors associated with future exacerbation frequency when a Bonferroni correction was applied, whereby p values < 0.007 were considered statistically significant. The threshold concentrations of capsaicin required to elicit 1, 2 and 5 coughs (C1, C2 and C5, respectively) were calculated by interpolation of the log dose–response curve [21]. A value of 1000 µmol L⁻¹ was assigned to any interpolated values which were > 1000 µmol L⁻¹ [21].

A previous study has reported a significant reduction in capsaicin-evoked cough reflex sensitivity in 14 participants following recovery from an upper respiratory tract infection [13]. A study by Dicpinigaitis reported the reproducibility of the capsaicin cough challenge in a sample size of 10 subjects [20]. We therefore aimed to recruit 10 participants to investigate the repeatability of CRS in COPD, and 15 participants to investigate CRS during and following an exacerbation of COPD. All analyses were performed using Prism® version 8.1.1 (GraphPad Software, San Diego, California, USA), except the Bland–Altman and intraclass correlation correlations analyses which were performed with R version 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria) for macOS version 10.14.5.

Subsequent exacerbations of COPD occurred in 15 of the 17 individuals who were followed up for 12 months following the initial hospitalisation. The mean (SD) number of further COPD exacerbations per participant during follow-up was 2.5 (2.0) episodes. There were a total of 42 exacerbation episodes, of which 30 led to utilisation of primary care services, 12 to presentation to the emergency department, and 9 to hospital admission. The median (IQR) time to the next exacerbation following recruitment at T_E was 135 (78–238) days.

There was a significant association between the change in C5 from T_E to T_R and the rate of acute exacerbation of COPD in the following 12 months (ρ = − 0.687, p = 0.003) (Table 4 and Fig. 2). In addition, there was no significant association between the rate of exacerbation of COPD in the following 12 months and the changes from T_E to T_R in either objective cough frequency, CAT scores (p = 0.116), cough severity VAS (p = 0.596), LCQ or FEV₁% predicted (Table 4 and Fig. 3). Neither at time T_E or T_R was there an association between the rate of exacerbation of COPD in the following 12 months and CRS, objective cough frequency, CAT scores, cough severity VAS, LCQ or FEV₁% predicted (all p > 0.06).

No participant was excluded from the study of CRS repeatability. Intraclass correlation coefficients to determine repeatability of the cough reflex sensitivity measures log(C1), log(C2) and log(C5) were 0.37, 0.54 and 0.85, respectively, in participants with stable COPD (Table 5). The mean differences (95% CI) in C1, C2 and C5 over 6 weeks were 0.34 (− 0.13–2.40), 0.31 (− 0.11–2.16) and 0.21 (− 0.40–1.79) doubling doses, respectively (Fig. 4). C5 was the most repeatable capsaicin challenge test threshold and was therefore used for the main analysis.

For objective cough frequency, the ICC coefficients for repeatability of log(total 24-h cough count), log(total cough frequency), log(awake cough count) and log(awake cough frequency) over 6 weeks were 0.86, 0.85, 0.70 and 0.84, respectively (Table 5). The mean differences (95% CI) in total cough frequency, awake cough count and awake cough frequency were 0.76 (0.47–1.22), 1.32 (0.76–2.29) and 0.11 (0.79–2.12) fold differences, respectively (Fig. 5). The ICC for repeatability of cough severity VAS scores over 6 weeks was 0.56 (Table 5 and Fig. 5).

We investigated the sensitivity of the cough reflex during exacerbations of COPD and its relationship with the frequency of future exacerbations. Cough reflex sensitivity (CRS) was heightened during exacerbations of COPD. The magnitude of change in cough reflex sensitivity 6 weeks after hospitalisation for exacerbation was associated with the frequency of future exacerbations. There was no association between subjective cough severity, objective cough frequency or FEV₁ and the frequency of exacerbations. In stable COPD patients, C5 was the most repeatable measure of CRS.

To the best of our knowledge, our study is the first to investigate CRS during exacerbations of COPD. CRS to capsaicin was heightened during exacerbation. Viral upper respiratory tract infections are commonly associated with exacerbations of COPD [29]. There are studies which have investigated CRS in healthy subjects with acute cough associated with upper respiratory tract infection [13, 14]. O’Connell et al. demonstrated that CRS was heightened during upper respiratory tract infection, and reduced following 4 weeks of recovery in healthy participants with non-productive cough [13]. Dicpinigaitis et al. also reported that CRS to capsaicin was heightened during upper respiratory tract infection, and reduced following 4 weeks of recovery [14]. Capsaicin is a selective agonist of transient receptor potential vanilloid 1 ion channel receptors (TRPV1), which are expressed on C-fibre sensory nerves [30]. Bronchopulmonary C fibres are relatively quiescent in health and can be sensitised or activated by inflammatory mediators [11]. Exposure to ozone [12], lactic acid [12], cationic proteins [31] and prostaglandin-E₂ [11] are known to enhance pulmonary C-fibre responses to mechanical and chemical stimuli in a reversible fashion. Airway inflammation in exacerbation of COPD could therefore lead to sensitisation or activation of C fibres.

A key finding from our study was that the magnitude of change in CRS following an exacerbation was associated with prospective frequency of exacerbations. The frequency of exacerbation in our study was comparable to those in previous studies [2, 17]. Further exacerbations were more frequent in individuals demonstrating a lower reduction in cough reflex sensitivity at six weeks after the index episode. Previous studies have investigated the predictive factors for exacerbations of COPD, including exacerbation frequency in the preceding 12 months, disease severity, gastro-oesophageal reflux and health status [17, 32]. A failure of CRS to change during recovery from exacerbation has not been previously studied as a potential marker of future exacerbation risk. Terada et al. conducted the only study of CRS and its association with exacerbations, and investigated CRS in patients with stable COPD [33]. In contrast to our findings, Terada et al. did not find an association between CRS and exacerbation of COPD defined by a change of medication [33]. There are several key differences between the studies that may explain the contrasting findings. Terada et al. assessed a single measure of CRS during a stable phase and did not assess it longitudinally for the persistence of CRS following exacerbation of COPD [33]. Terada et al. did however report a weak correlation with frequency of exacerbations of COPD when defined on the basis of a change of symptoms rather than a change in therapy for exacerbation of COPD [33].

The mechanism by which a persistently heightened cough reflex sensitivity may predispose to future exacerbations is not clear. An associated persistence of airway inflammation following an exacerbation may be important but this has seldom been studied. Airway inflammation in patients with stable COPD however has been reported to predispose to exacerbations. Bhowmik et al. demonstrated that raised levels of cytokines, IL-6 and 8, in sputum were associated with frequent exacerbations [34]. Gompertz et al. reported reduced levels of sputum protease inhibitors in those with frequent COPD exacerbations [35]. Another possible mechanism for persistent cough reflex sensitivity, and predisposition for future exacerbation and symptoms, is changes in central neural pathways and cognitive influences, such as those that have been reported in inflammatory airway diseases such as asthma using functional neuroimaging [36].

The assessment of CRS may be a tool to identify patients at high risk of future exacerbations. The existing cough challenge methodology is not practical for routine clinical use and simpler protocols would need to be developed. The capsaicin challenge test was however well tolerated by severe COPD patients suffering from acute exacerbation study. This is the first study to assess cough reflex sensitivity during an exacerbation and may facilitate future research investigating the mechanisms involved in exacerbation of COPD. It is not possible to elucidate the mechanism between the magnitude of change in CRS following exacerbation and the frequency of future exacerbations from our study. A better understanding of this mechanism may lead to targeted therapy for the prevention of exacerbations.

Previous studies have investigated whether other measures of cough predict COPD exacerbation frequency, with conflicting findings. Burgel et al. and Seemungal et al. reported that the presence of self-reported cough is associated with exacerbation frequency in stable COPD [2, 7]. In contrast, Hurst et al.[17], in common with this current study, did not find an association between self-reported cough and numbers of exacerbation of COPD. Neither did we observe an association between objective cough counts and exacerbation frequency. Objective CRS, subjective symptom scales and objective cough counts assess very different dimensions of cough, and a poor association between them is well reported [37]. It is likely that CRS reflects the interaction between airway inflammation and sensory nerves.

There are limitations to our study. We studied a relatively small sample size and were therefore unable to assess the predictors of exacerbations of COPD with multivariate regression analysis. Our patients had severe COPD and therefore our findings should be interpreted with caution for those with mild disease. We were also unable to assess the effects of factors such as smoking status that may influence CRS due to the small sample size. To standardise the timing of the cough challenge test during the hospital admission, patients were assessed on the first day that they were deemed fit for a challenge test by the clinical and research teams. It is possible that the variability of the timing of the cough challenge test between the patients may have impacted the cough reflex sensitivity assessment. The assessment of CRS in the recovery phase may have been too soon after the exacerbation. We did, however, demonstrate that CRS is highly repeatable over 6 weeks in stable COPD with no order effect. Furthermore, the CRS values following recovery from exacerbation in our study were similar to those reported previously in stable COPD by Sumner et al.[38]. FEV₁% predicted during exacerbation, following recovery and the change between 2 occasions were not associated with the frequency of future exacerbations despite previous reports of an association between severity of COPD according to FEV₁% predicted and exacerbation frequency [17, 39]. This may be due to the small sample size and the relatively limited range of disease severity in our study. Patients with more severe disease were more likely to have exacerbations, and therefore be included in our study. The subjective cough measures were not associated with CRS, which may be due to the small sample size in our study. This is not unexpected as health status and CRS assess very different dimensions of cough, and is consistent with previous reports [37, 40]. We did not find an association between cough reflex sensitivity and objective cough frequency, and this is consistent with a previous study reporting a poor association [38].

The technology for measuring cough frequency in the current study has not yet been widely used in COPD, but is well established in chronic cough in other contexts [40]. Validity for our approach though is suggested by the similar magnitude and range of cough frequencies we report in stable COPD compared to those of a study by Sumner et al. using an alternative method to determine cough frequency [38]. Whilst there was a slight excess of female participants in our cohort, there was no significant difference in C5 between gender. A slight excess of female participants is consistent with other studies of patients hospitalised for exacerbation of COPD [16, 41]. Whilst C5 generally appeared a repeatable measure of cough reflex sensitivity, this observation was not universal; some participants with stable COPD demonstrated marked differences in C5 between the two timepoints. This may therefore impact the utility of the C5 test in future studies. We did not investigate the cause of COPD exacerbation or markers of inflammation in sputum and serum, and these should be assessed in future studies. We only assessed CRS with inhaled capsaicin. It is possible that COPD patients may respond differently to other tussive agents, and produce different results [42, 43].

In conclusion, cough reflex sensitivity is heightened during acute exacerbation of COPD and is associated with the frequency of future exacerbations. Further studies are needed to replicate our findings in a larger study, and to investigate the mechanisms for persistent cough reflex hypersensitivity as this may lead to targeted therapy for the prevention of COPD exacerbations.