INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Chronic cough, defined as a cough that persists for 3 wk or longer (1), is a common troublesome symptom, being one of the most frequent complaints to the primary-care physician and to the pulmonologist (1, 2). More than 90% of the cases result from one of five common causes, smoking, postnasal drip, asthma, gastroesophageal reflux, and chronic bronchitis (1, 3), but an important subgroup will have more than one factor causing cough (4). The so-called "cough variant asthma" is a well known cause of cough (1, 5) and a less common non-asthmatic eosinophilic bronchitis has also been described (6).

In some patients the cause of cough will be readily apparent and will require only minimal investigations. The standard diagnostic approach of chronic cough includes pulmonary function testing, chest and sinus radiology, methacholine challenge, and, in some cases, bronchoscopy and 24-h esophageal pH monitoring (4, 7, 8).

Bucca and colleagues (9, 10) and others (6, 11) have shown that inflammation is a common feature in many of these patients with chronic cough, whether the final diagnosis is asthma, rhinosinusitis, gastroesophageal reflux, or ACE inhibitor-induced cough. Exhaled nitric oxide (ENO) levels are increased in patients with asthma (14), viral respiratory infections (17), or bronchiectasis (18). ENO decreases with corticosteroid treatment of asthma, likely reflecting reduced inflammation (19). However, not all forms of airway inflammation are accompanied by increased levels of ENO; cigarette smoking reduces ENO levels while causing airway inflammation.

ENO measurement has been suggested as a simple and noninvasive way of monitoring inflammation of the respiratory tract with a potential role in the diagnosis of disease (20, 21). As the diagnosis of cough may require invasive procedures, a noninvasive diagnostic tool could be of value.

We hypothesized that ENO levels would be elevated in patients with chronic cough caused by asthma but would be normal in patients with nonasthmatic causes of cough. Thus, ENO would have a role in the diagnostic approach to chronic cough.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Chronic cough was defined as cough for more than 3 wk, with a normal chest radiograph and FEV₁ > 80% of predicted. The exclusion criteria were the use of codeine or any other medication for chronic cough, upper respiratory infection 4 wk prior the study, use of corticosteroids within 6 wk of the study, current smoking, any significant medical conditions, and the standard contraindications to methacholine challenge.

Thirty-eight consecutive adult patients referred to the outpatient clinics of The Asthma Centre of the Toronto Hospital, a tertiary referral center, or to an affiliated community respiratory clinic met these criteria and were enrolled in the study. The study was approved by the Ethics Committee of the Toronto Hospital. All patients signed the Consent Form.

Study Design

The study was a cross-section observational design, characterizing patients with chronic cough of unknown cause by means of ENO measurement and conventional diagnostic tools, including methacholine challenge. ENO levels were also measured in positive and negative control groups (patients with known asthma and healthy volunteers).

Each patient underwent a standard clinical assessment, which included history and physical examination, medical questionnaire, laboratory tests and chest roentgenogram, spirometry before and after bronchodilator (22), allergy skin testing to 12 common allergens, and methacholine challenge using the method of Cockroft and colleagues (23). In addition, each patient performed exhaled NO (ENO) measurements by a technique described previously (24), done prior to spirometry and methacholine challenge. Pulmonary function testing and ENO measurements were done within 24 h; skin prick testing was done within 10 d of the first interview.

Clinical diagnoses were made by an experienced respiratory physician (KRC) using a diagnostic algorithm and blinded to the result of ENO assay. The patients were divided in two categories: asthmatics (A) and nonasthmatics (NA), on the basis of their responses to inhaled albuterol and to inhaled methacholine. Patients whose FEV₁ increased 12% or more above baseline values 15 min after the inhalation of four MDI puffs of albuterol (360 µg; Glaxo Wellcome Canada Inc., Mississauga, ON, Canada) were considered to have asthma. These patients were not subjected to methacholine challenge. Patients without bronchodilator response underwent methacholine challenge and were considered to have asthma if their methacholine PC₂₀ was 8 mg/ml or less. Subjects were considered atopic if they had at least one positive skin prick test relative to positive and negative histamine and saline controls, respectively.

Control Group

For the additional purpose of comparison between the asthmatics of the chronic cough group and asthmatics in whom wheezing was the main complaint, we also studied a group of 44 subjects with mild or moderate atopic asthma. This group was defined according to ATS standards (25), in which wheezing or dyspnea but not cough were the main complaint (WA). Skin tests were positive in all subjects of this group. Twenty-three healthy nonsmoking volunteers were recruited as controls (C) from the staff of the hospital. They had no history of pulmonary disease or atopy and had normal spirometry. These two sets of patients were also not receiving steroids and met all the same inclusion and exclusion criteria of the previous groups.

Exhaled NO Measurement

A rapid-response chemiluminescent NO analyzer (Sievers 280; Sievers, Boulder, CO) was employed. A daily two-point calibration was performed, first with 100% nitrogen for zero, then with an analyzed standard gas for the span (1.6 ppm NO, balance nitrogen). The lower detection limit for NO was 1 to 2 ppb. Ambient air NO was recorded before and after each subject was studied. Zero and span gases were checked periodically. The NO analyzer signal output was fed to a computer data acquisition program (DasyLab for Windows; DasyTec Corp., Amherst, NH) with a real-time chart-recorder-like display of NO versus time written directly to the computer's hard disk as a data file. NO concentrations were measured using a data analysis program, written in-house (Microsoft Visual Basic).

The ENO measurement was performed with a previously described technique (24). The seated subject inserted a mouthpiece and inhaled to TLC from a reservoir of medical grade compressed air that contained < 1 ppb NO. The subject then exhaled via a high resistance and maintained a mouth pressure of 20 mm Hg, which was displayed on a pressure gauge. The resultant expiratory flow was 45 ml/s. The steady-state NO plateau was taken as the ENO value. Repeated exhalations were performed to achieve three ENO values that agreed at the 5% level.

Statistical Analysis

A SAS system version 6.12 for Windows was used for the analyses. Tests for normality (Shapiro-Wilk) were done for the distributions of NO values. Two-by-two tables of NO (low/high) versus asthma diagnosis (yes/no) were prepared using each of the 10th, 25th, 50th, 75th, and 90th percentiles of the NO distribution as cut points. Pretest probability of disease, sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios were calculated for each of these tables (26). A receiver operating characteristic curve (ROC) was plotted in order to choose the best cutoff point. Analysis of variance on the ranks was done as a general test (Kruskal-Wallis test). If the result of this test was statistically significant a nonparametric test (Wilcoxon's two-sample) was used for pairwise post-hoc comparison of groups (26). No correction for multiple comparisons was used. Statistical significance was considered at p < 0.05. All tests were two-tailed.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient characteristics of the groups studied are shown in Table 1. Cough had been present for an average of 53.8 wk (range, 3 to 152 wk). Bronchial hyperresponsiveness on methacholine challenge was present in eight of 38 patients with chronic cough (asthma prevalence of 21%).

ENO for the whole group was not normally distributed. The median ENO concentration was 16.7 ppb (25 to 75%: 11.0 to 21.7 ppb) for the NA group, whereas in the A group it was 75.0 ppb (25 to 75%: 34.1 to 104.0 ppb). The median ENO in the WA group was 69.0 ppb (25 to 75%: 51.9 to 101.0 ppb), and in the C group it was 28.3 (25 to 75%: 23 to 30 ppb) (Figure 1).

View larger version (11K): [in this window] [in a new window]   Figure 1.   Exhaled nitric oxide by diagnostic category.

Analysis of variance on the ranks of the ENO values by groups was statistically significant (p = 0.0001). Post-hoc pairwise comparisons, using the Wilcoxon's nonparametric two-sample test, showed that groups A and WA were statistically similar (p = 0.73), and both were different from groups NA and C (p = 0.0014, p = 0.007, respectively). The patients with nonasthmatic chronic cough (NA) had a lower ENO than did all other groups, including the control group (C) (p = 0.0008).

The ROC curve for ENO in this group of patients, where the best sensitivity and specificity was achieved using the cut point of 30 ppb, is shown in Figure 2. The distribution of the patients with chronic cough according to this ENO cutoff point and according to groups A or NA is shown in Table 2.

View larger version (11K): [in this window] [in a new window]   Figure 2.   ROC curve for exhaled nitric oxide; data labels show the selected cutoff points of ENO levels.

In this group of patients, in which the prevalence of asthma was 21%, the sensitivity was 75%, the specificity was 87%, the positive predictive value was 60%, and the negative predictive value was 93%; the positive likelihood ratio was 5.8 and the negative likelihood ratio was 0.3.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have shown for the first time that ENO has a role in the evaluation of chronic cough. ENO below 30 ppb in untreated nonsmoking adults with chronic cough had high specificity (87%), a high negative predictive value (93%), and a helpful negative likelihood ratio (0.3) for the absence of asthma. However, an ENO value above 30 ppb had reasonable sensitivity (75%) and a weak positive predictive value (60%) for the presence of asthma to explain chronic cough. Thus, ENO measurement may have greater value as a negative predictive index in distinguishing cough variant asthma from nonasthmatic cough.

This relatively low reliability of the test for screening asthmatics among patients with chronic cough can be explained by the relatively low pretest probability of the disease (27). In our study population, with a 21% pretest probability of asthma (eight asthmatics in the group of 38 patients with chronic cough), ENO values above 30 ppb yielded a post-test probability of disease of 60% (six asthmatics in a group of 10 patients with high ENO). Although this greatly increased the diagnostic likelihood of asthma in patients with chronic cough, this result alone was still not great enough to establish a diagnosis of asthma confidently. On the other hand, a negative test (< 30 ppb) changes the pretest probability of disease to a post-test probability of disease of 7% (two asthmatics in a group of 28 patients with low ENO). This latter feature would make it a useful test in the clinical setting to exclude asthma in patients with chronic cough, perhaps avoiding the need of more invasive tests such as bronchial challenge (Table 2).

Methacholine challenge has been shown to have positive predictive values ranging from 60 to 82%, whereas the negative predictive value assessing whether chronic cough is likely to be due to asthma is around 100% (25). In our study, ENO had positive and negative predictive values of 60 and 93%, respectively, showing a potentially valuable role in the screening of asthma among patients with chronic cough.

Our study confirmed previous reports that steroid-naive asthmatics have higher ENO values when compared with normal control subjects (14, 28). We could not based only on ENO measurements distinguish between the two asthmatic groups of patients (those with chronic cough with or without asthma). In both, ENO values were elevated and not statistically different.

Another finding of this study was that patients with chronic cough in the absence of asthma had ENO values significantly lower compared with asthmatics and with healthy non-smoking adult volunteers. This is in agreement with the recently reported findings of Dupont and colleagues (28) that also showed a low ENO level in a similar group of patients. Our data do not provide an explanation for these findings, but several hypotheses not mutually exclusive may be considered. First, processes not associated with airway inflammation may have triggered the chronic cough. This would seem plausible for instances of cough triggered by neural reflexes or associated with psychogenic disorders. However, this would not, by itself, explain the apparent reduction in nitric oxide level. Second, the increased nitric oxide production might be relatively specific for certain types of inflammation such as the eosinophilic inflammation of asthma. Third, patients with long standing unexplained cough might have airway epithelial damage that could reduce the ability of cells to produce or transport nitric oxide, although such an abnormality of the iNANC system has not yet been characterized. Our finding of diminished ENO in nonasthmatic chronic cough warrants further studies, but it also may become in the future a valuable marker of airway injury.

Although as a group our patients with nonasthmatic cough had low ENO levels, four patients had elevated values without bronchial hyperresponsiveness to methacholine. These patients were not considered to have asthma by the study criteria, but it is possible that they suffered from a form of asthma without methacholine hyperreactivity but to other stimuli such as cold air. It is also possible that these patients had eosinophilic bronchitis without airway hyperreactivity, a disorder characterized by Gibson and colleagues (6). In support of these speculations, when three of these four patients were treated with inhaled budesonide 800 µg/d for 4 wk, three responded with clinically significant improvement in or disappearance of cough.

Two patients had bronchial hyperreactivity with low ENO levels. One patient was a surreptitious cigarette smoker, a fact that became apparent only at the conclusion of the study. As cigarette smoking is associated with low ENO values (29), only one of our study subjects had an inexplicably low ENO value in the presence of bronchial hyperresponsiveness. Although this patient was categorized as having asthma by the criteria of this study, this may not have been the correct diagnosis. Bronchial hyperresponsiveness to inhaled methacholine is more prevalent in the population than physician-diagnosed clinical asthma (30, 31). Our patient may have had cough without actual asthma and eosinophilic airway inflammation but a false positive methacholine challenge. The possible false positive diagnosis of asthma (on the basis of methacholine challenge in the setting of undiagnosed cough) could lead to inappropriate or futile inhaled corticosteroid therapy. Thus, ENO measurement might be used also to reduce asthma false positive diagnosis in future. This hypothesis requires prospective study.

The role of ENO measurement in the diagnosis of chronic cough could be confounded by several factors, both technical and clinical. The former include standardization of sampling flow, upper airway contamination, method used and timing relative to spirometric testing (24, 32); the latter include viral infection (17), smoking (29), and steroid treatment (19). These confounders may complicate the interpretation of ENO in this setting.

In summary, our study suggests a potentially useful role for ENO measurement in the investigation of chronic cough among nonsmoking adults. When ENO values are below 30 ppb, the likelihood of asthma is low, and this simple noninvasive measurement may obviate the need for a challenge study. By contrast, the finding of a high ENO value increases the likelihood that asthma is present, but it is insufficiently specific to allow diagnosis without a challenge study. The use of ENO measurement in the diagnosis of chronic cough shows potential value and should be the subject of further research.