Airway diseases account for a significant proportion of the respiratory physician's workload. Traditionally, conditions are classified according to clinical symptoms and abnormalities of function, although the major categories of airway disease (asthma, chronic cough and chronic obstructive pulmonary disease (COPD)) show considerable overlap in terms of both clinical picture and response to antiinflammatory treatment with corticosteroids. All of the above conditions are associated with airway inflammation, although to date there has been little interest in its routine measurement in clinical practice. The recent development of simple, safe and valid noninvasive techniques for the assessment of airway inflammation has increased interest in such an approach 1, 2. A number of techniques are available, ranging from measurement of exhaled nitric oxide levels to performing differential cell counts and assessment of mediator concentrations in induced sputum. Relatively little is known about the relationship between these different markers, although it is likely that they measure different aspects of the inflammatory response 3. A clearer understanding is desirable since some techniques are more suited to routine clinical use than others. Induced sputum cell and mediator measurements are particularly well validated 1, and normal ranges from a relatively large adult population have been published 4.
The present article summarises clinical studies pointing to the usefulness of induced sputum analysis in the following conditions: 1) asthma; 2) cough; 3) COPD; and 4) other respiratory conditions.
Asthma
Asthma is commonly associated with sputum eosinophilia. Up to 80% of corticosteroidnaive subjects 1, 2 and >50% of corticosteroidtreated subjects 5 with currently symptomatic asthma have a sputum eosinophil count that is outside the normal range. Subjects with severe acute asthma usually exhibit marked sputum eosinophilia, although predominant neutrophilia has been noted in some studies in which subjects taking oral corticosteroids at the time of the exacerbation are not excluded 6. The validity of a high sputum eosinophil count for the identification of asthma (defined as consistent symptoms with objective evidence of abnormal variable airflow obstruction) is better than peak expiratory flow variability, defined as the mean percentage amplitude, and the acute bronchodilator response; it approaches the sensitivity and specificity of measurement of airway responsiveness 7. There is at best a weak relationship between the severity of asthma as defined by lung function, airway responsiveness or symptoms and the sputum eosinophil count 5, 8–10.
The shortterm response to inhaled corticosteroids differs markedly according to the sputum eosinophil count, with little evidence of improvement in symptoms and airway responsiveness in subjects with a baseline sputum eosinophil count of <3% 11. These findings suggest that measuring the underlying airway inflammation might provide a better guide as to the need for corticosteroid treatment than assessment of functional abnormality. More work is needed to define the relationship between airway inflammation, symptoms, the underlying functional defect and the response to corticosteroids in subjects with asthma. The underlying pathophysiology of noneosinophilic asthma needs to be better understood, and studies are required to determine whether inhaled corticosteroids have any longerterm benefits in this group. Finally, there is a pressing need for a study examining whether asthma management aimed at normalising the sputum eosinophil count results in better patient outcomes than a standard approach where the aim of treatment is to suppress symptoms.
Categorising subjects according to the presence of sputum eosinophilia might be particularly useful in asthmatics who are symptomatic despite treatment with inhaled corticosteroids since the additional treatment options (long acting β2‐agonists, higherdose inhaled corticosteroids, leukotriene antagonists and theophylline) differ in their effects on eosinophilic airway inflammation 12–15. This is an important group (50% of asthmatics in the UK are receiving low or moderate doses of inhaled corticosteroids and >50% have sufficient symptoms to warrant increased treatment); it is therefore necessary to understand better why some patients exhibit persistent eosinophilic airway inflammation despite corticosteroid treatment and whether this group do particularly well with additional antiinflammatory treatment.
A number of investigators have examined the practical use of assessment of induced sputum markers of airway inflammation in occupational asthma 16–18. In general, occupational asthma is associated with similar sputum characteristics to nonoccupational asthma, and occupational challenges are associated with an increase in sputum eosinophilia 16, 17 in much the same way as allergen challenge is. There is some evidence that sputum eosinophil counts increase during workplace exposures in subjects with occupational asthma 17. Further work is required to determine whether incorporation of assessment of airway inflammation into a standard protocol consisting of assessment of peak expiratory flow ands airway responsiveness at work and away from work provides additional diagnostically useful information.
Cough
Chronic cough is associated with predominant sputum neutrophilia 19, but up to 40% of subjects with cough have a sputum eosinophil count of >3% 20, 21. Half of them show no functional evidence of asthma but suffer from eosinophilic bronchitis 20, 21. Assessment of airway inflammation is the only way of identifying these subjects and is, therefore, an important addition to the chronic cough investigation algorithm 20. Further work is required to determine the sputum cell characteristics of cough associated with gastrooesophageal reflux, rhinitis and viral infections, although a recent bronchoscopic study suggests that no particular patterns will emerge 22. Lipidladen macrophages are noted more commonly in induced sputum from patients with oropharyngeal reflux than controls, suggesting that this might be a useful noninvasive marker of oropharyngeal reflux with gastric aspiration 23. Patients with cough and sputum eosinophilia exhibit an objective response to corticosteroid treatment that occurs in parallel with a treatmentassociated fall in the sputum eosinophil count 20, 24. In contrast, patients without sputum eosinophilia do not respond 25.
Chronic obstructive pulmonary disease
In COPD, the sputum neutrophil count is usually raised, and the absolute or differential neutrophil count is related to reduced forced expiratory volume in one second (FEV1) and the increased rate of decline in FEV1, suggesting that neutrophilic airway inflammation is functionally important 26, 27. Up to 40% of subjects with COPD have a sputum eosinophil count of >3% 28, 29. These subjects are in other respects identical to subjects without sputum eosinophilia (i.e. no more likely to have features suggesting asthma). There is increasing evidence that the presence of sputum eosinophilia predicts an objective response to corticosteroid treatment in COPD 28, 29. In one study, the response to a 2‐week course of oral prednisolone increased as the baseline sputum eosinophil count increased and was associated with a marked treatmentinduced fall in the sputum eosinophil count but no change in sputum markers of neutrophilic inflammation 29. This suggests that eosinophilic airway inflammation is functionally important in some subjects with COPD, and that the effects of corticosteroids are due to modification of this aspect of the complex airway inflammation. Further work is required to determine whether reduction in exacerbation frequency and decreased rate of decline in FEV1 with longterm corticosteroid treatment is confined to subjects with sputum eosinophilia. Potentially, sputum assessment could be used as a screening test before deciding on longterm corticosteroid treatment in COPD.
Other conditions
Little is known about induced sputum cell features of interstitial diseases, although sarcoidosis and extrinsic allergic alveolitis are associated with airway involvement and so might be usefully investigated using induced sputum 30. One problem is that induced sputum might not be the best way of investigating lymphocytic airway inflammation since lymphocytes are scarce in induced sputum. There have been intriguing findings of mineralogical particles in induced sputum in occupational lung diseases associated with dust exposure 31 and haemosiderinladen macrophages in subjects with left ventricular failure 32 which deserve further study. Common colds are associated with predominant neutrophilic airway inflammation in normal subjects and subjects with asthma 33. Further work is also required in this area. Potentially, induced sputum could be a useful means of identifying different patterns of inflammation associated with different pathogens or differences in host response to the same pathogen.
Key points
1) There is considerable overlap of the sputum inflammatory cell characteristics of the main clinical categories of airway disease. 2) The relationship between eosinophilic airway inflammation and disease expression is uncertain. 3) Antiasthma therapies differ in their effects on eosinophilic airway inflammation. 4) Eosinophilic bronchitis is a common and treatable cause of chronic cough. It can only be identified by measuring airway inflammation. 5) Sputum eosinophilia predicts a response to corticosteroid therapy in asthma, cough and COPD. 6) There is no evidence that corticosteroids are helpful in the absence of a sputum eosinophilia.
Outstanding questions
1) Does measuring airway inflammation improve diagnosis and treatment outcome? What is the best technique to use? 2) Can patterns of inflammation that are associated with differences in prognosis or treatment response be identified?
- Received April 4, 2002.
- Accepted April 16, 2002.
- © ERS Journals Ltd