The use of sputum cell counts to evaluate asthma medications

Abstract

Total and differential cell counts from hypertonic-induced, dithiothreitol-dispersed sputum provide reproducible measurements of airway inflammatory cell counts, which are responsive to treatment with anti-inflammatory drugs. They have helped to understand the kinetics of inflammatory cell changes in asthma after the reduction of corticosteroids and the subsequent re-introduction of treatment. They have identified that the presence of sputum eosinophilia in asthma, chronic cough and chronic airflow limitation is a predictor of steroid-responsiveness and of lack of ‘asthma control’. They can be used to study the dose–response effect of inhaled corticosteroids and may be useful to establish the relative potency of different corticosteroid formulations and delivery devices. Sputum cell counts are also useful to study the potential anti-inflammatory effects of drugs like theophylline, long-acting β-adrenoceptor agonists, leukotriene antagonists and newer drugs in development. They may be helpful to select add-on therapy to corticosteroids in ‘difficult-to-control’ asthma.

Introduction

In recent years, since asthma was recognized to be associated with a chronic inflammation of the airways, anti-inflammatory agents, in particular corticosteroids, have been the corner-stone of asthma management. International guidelines have emphasized the need to treat even mild uncontrolled asthma or mild persistent asthma with inhaled corticosteroids [1, 2]. However, the inflammatory component of asthma or the anti-inflammatory effects of drugs are not examined in the routine management of asthma. They are indirectly assessed from measurements of symptoms, airflow limitation and airway responsiveness, which can be misleading [3]. A major limitation of the direct assessment of airway inflammation by bronchial mucosal biopsy or bronchoalveolar lavage is the invasive nature of the investigation. In recent years, the examination of induced sputum for cellular and fluid-phase markers of inflammation has provided a relatively noninvasive, valid and repeatable method of assessing airway inflammation [4]. Sputum induction can be performed successfully, and safely and can be readily repeated in the same patient [5]. This review will summarize the information available from sputum examination on the anti-inflammatory effects of ingested and inhaled corticosteroids, short and long-acting β-adrenoceptor agonists, theophylline and a leukotriene antagonist. For the purposes of this review, we have not considered the information from bronchoscopic procedures.

Sputum examination

The methods of sputum induction and examination have been reviewed in detail elsewhere [6–8]. The induction is carried out, usually after premedication with inhaled salbutamol, with an aerosol of hypertonic saline, e.g. for three periods of 7 min each with 3%, 4% and 5% saline or for about 12–15 min with a fixed concentration of 3% or 4.5% saline. Salivary contamination and postnasal drip are minimized by blowing the nose, rinsing the mouth with water and swallowing the water before expectoration. The expectorate is a mixture of sputum and saliva. Salivary contamination can be further reduced by selecting the more opaque or dense portions of sputum from the clear saliva. Minimization of salivary squamous cell contamination increases the interobserver reliability of subsequent measurements [9]. The major advances in recent years in sputum processing have been the dispersion of sputum with 0.1% dithiothreitol and the preparation of cytospins for differential cell counts. The remaining suspension of cells can be centrifuged to remove these and the supernatant can be stored at −70 °C for later measurements of soluble mediators, cytokines, chemokines, proteins, etc. The majority of research on the effects of anti-inflammatory drugs have so far focused on cell counts.

Observations with corticosteroids

Anti-inflammatory effects and kinetics of response

Three models, using induced sputum, have been used to investigate the anti-inflammatory effects of glucocorticoids on airway inflammation.

In the uncontrolled asthma model, subjects are selected with uncontrolled asthma and an increase in the proportion of eosinophils in sputum. An increased proportion of eosinophils, indicating an eosinophilic bronchitis, is needed so that there is an adequate signal to investigate the effect of the drug on this marker of inflammation. Not all subjects with symptomatic asthma or exacerbations of asthma are associated with sputum inflammatory cell infiltration or the same inflammatory process [10, 11]. Evidence so far suggests that glucocorticoid treatment helps those patients with an eosinophilic bronchitis but not those without sputum eosinophilia in asthma [12, 13], chronic cough [14] or chronic airflow limitation due to smoker's bronchitis [15]. Prednisolone treatment for 2 weeks decreased sputum eosinophilia (from a mean of 6.7% to 1.0%) and sputum levels of eosinophil cationic protein and eosinophil peroxidase [16]. In another study, when the anti-inflammatory effects of prednisolone (30 mg daily for 2 weeks) were compared with high dose inhaled fluticasone (2000 µg daily for 2 weeks), the median decrease in sputum eosinophil count tended to be greater with the inhaled steroid (Δ 4.2%) compared with prednisolone (Δ 3.5%) [17]. The kinetics of resolution of inflammation following treatment of severe asthma exacerbation with prednisone has been examined [13]. Sputum eosinophils did not become significantly reduced until 48 h after prednisone treatment and they continued to fall to normal between 3 and 7 days (Figure 1). In contrast, sputum fluid-phase fibrinogen decreased more slowly and did not become significantly reduced until the 7th day, suggesting that airway oedema resolves more slowly. Prednisone-dependent asthma is also associated with an eosinophilic bronchitis, which can be reversed with higher doses of prednisone [18]. In eosinophilic bronchitis without asthma, sputum eosinophils fell to normal values within 1 week after budesonide 800 µg day⁻¹ [19].

Figure 1.

Kinetics of sputum markers after treatment of a severe asthma exacerbation with prednisone. Sputum eosinophil count and fluid-phase eosinophil cationic protein (ECP) level are significantly reduced in 48 h and continue to fall to normal between 3 and 7 days. In contrast, fluid-phase fibrinogen decreased more slowly and did not become significantly reduced until the 7th day (from [13]).

In the allergen-challenge model, the effect of pretreatment with corticosteroid on allergen-induced inflammatory changes can be studied. Pre-treatment with a single dose of beclomethasone 500 µg partly inhibited the late asthmatic response, but did not inhibit the early asthmatic response or prevent the allergen-induced sputum eosinophilia either at 7 or 24 h after allergen inhalation [20, 21]. Pre-treatment with a single dose of fluticasone 250 µg partly inhibited the late asthmatic response and the sputum eosinophilia at 7 h, but not at 24 h [22]. The latter was inhibited by inhaled budesonide (400 µg day⁻¹) by Turbuhaler for 1 week before the allergen inhalation [23, 24]. However, if the regular treatment with inhaled corticosteroid was discontinued for as little as 24 h before the allergen inhalation, this protective effect seemed to be lost [22]. This, together with the observation that the attenuating effect of inhaled corticosteroids on some of the sputum markers of inflammation are short-lived [25], raises questions about the duration of the anti-inflammatory effects of inhaled corticosteroids.

In the steroid-reduction model, the steroid-sparing effects of different drugs can be investigated by the addition of the test drug during controlled steroid reduction [26]. This will be discussed later with long-acting β-adrenoceptor agonists and theophylline. The steroid-reduction model allows overtreatment with inhaled corticosteroids to be identified; many patients are able to discontinue glucocorticoid without an exacerbation.

Dose–response

Few well-conducted studies have been able to show a dose–response effect of inhaled corticosteroids on the usual measurements of asthma control, i.e. symptoms, spirometry or airway responsiveness [27]. Recent studies have examined the effects of varying doses of inhaled corticosteroids on sputum inflammatory markers. Jatakanon and colleagues [28] demonstrated a dose-dependent reduction in sputum eosinophils in a randomized, placebo-controlled, parallel-group study of 100, 400 and 1600 µg of budesonide dry powder inhalation each for 4 weeks in steroid-naive asthmatics. However, Taylor and colleagues [29], using similar doses of ciclesonide dry powder inhalation for 2 weeks demonstrated a significant reduction in the proportion of eosinophils in induced sputum with 400 and 1600 µg, but no dose–response. Further studies are required. However, if a dose-dependent effect on eosinophils can be demonstrated, this would be a valid method to compare the relative potencies of different formulations and delivery devices of inhaled corticosteroids.

Observations with short-acting β-adrenoceptor agonists

The regular use of high doses of short-acting β-adrenoceptor agonists like fenoterol and salbutamol may worsen asthma control and airway responsiveness in patients with moderate to severe asthma [30], and increase airway responses to inhaled allergen [31]. Aldridge et al. [32], in a four period, randomized, double-blind, placebo-controlled study, compared sputum cell counts after 6 weeks of treatment with regular terbutaline (1 mg four times a day), regular budesonide (400 µg twice a day), combined treatment or placebo in patients with mild-to-moderate asthma. The percentage of eosinophils in induced sputum at the end of terbutaline treatment alone (median 8.3%) was higher compared with the end of treatment with placebo (median 4.4%), budesonide alone (median 1.7%) and combined treatment (median 2.1%). Gauvreau et al. [31] examined the effects of regular treatment with 800 µg a day of salbutamol for 7 days on allergen-induced increases in sputum cell counts in 10 mild asthmatics in a randomized, placebo-controlled, cross-over study. Allergen inhalations were performed 12 h after the final dose of salbutamol. Salbutamol treatment significantly increased the late asthmatic response compared with placebo treatment and significantly enhanced the number of sputum eosinophils and sputum eosinophilic cationic protein (ECP) at 7 h, but not at 24 h after allergen inhalation. These observations suggest that the regular use of short-acting β-adrenoceptor agonist increases the allergen-induced late asthmatic response by increasing events associated with eosinophil influx into the airways.

Observations with long-acting β-adrenoceptor agonists

Salmeterol and formoterol are potent bronchodilators with questionable anti-inflammatory properties. Sputum examination has proved useful in understanding these effects. Both drugs attenuate the early and late asthmatic responses after allergen inhalation, without inhibiting the allergen-induced increases in sputum eosinophils [20, 21], suggesting that allergen-induced airway responses are modified through functional antagonism rather than inhibition of inflammatory cell infiltration. In contrast to these conclusions, Dente and coworkers [33] reported an inhibitory effect of pretreatment with a single dose of salmeterol on allergen-induced sputum eosinophilia at 24 h post allergen inhalation. This observation is questionable because of the nature of the cross-over design in which baseline measurements were not made before the second allergen inhalation test and may not have been comparable with those before the first test [34]. Further evidence of the lack of anti-inflammatory effect of salmeterol was provided by Turner et al. [35] in a randomized, double-blind, parallel group study of salmeterol, beclomethasone and placebo in mildly uncontrolled asthmatics with sputum eosinophilia. Salmeterol improved airway function and symptoms, but had no effect on sputum eosinophilia. Taken together, these observations suggest that long-acting β-adrenoceptor agonists do not control eosinophilic inflammation. Their effects on other types of inflammation have not been investigated by sputum examination.

International guidelines currently recommend the addition of a long-acting β-adrenoceptor agonist if asthma is not adequately controlled with a moderate dose of inhaled corticosteroid (i.e. beclomethasone 1000 µg daily or equivalent) [1, 2]. These recommendations were based on observations of improvement in peak expiratory flow (PEF) and spirometry when the two drugs were used in combination compared to doubling the dose of inhaled corticosteroid [36, 37]. Recent observations of the effects of the combination of long-acting β-adrenoceptor agonists and inhaled corticosteroids on sputum cell counts are useful in making future recommendations.

As part of a large randomized, double-blind, parallel-group, multicentre study, Kips et al. [38] evaluated the effect of adding formoterol to a low dose of budesonide, compared with a higher dose of budesonide, on sputum cell counts and asthma control. After a 4 week run-in period of treatment with 800 μg twice daily of budesonide, 60 patients were randomly assigned to 1 years treatment with 400 µg twice daily of budesonide and placebo twice daily or 100 µg twice daily of budesonide and 12 mg twice daily of formoterol. At the end of the run-in period the median sputum eosinophil count was significantly reduced from 4.5% to 0.7%. At randomization, sputum eosinophils were similar in both groups (0.9% and 0.6%, respectively). During the randomized treatment period, a slight, nonsignificant increase in the proportion of eosinophils was observed in the group treated with the lower dose of budesonide (3.4% vs 1.7% in the group treated with the higher dose of budesonide). The change over the entire study period in either of the two treatment groups was not significant. Clinical asthma control was also not significantly different between each group. This suggests that a long-acting β-adrenoceptor agonist can be added on to a moderate maintenance dose of inhaled corticosteroid to achieve asthma control without worsening airway eosinophilic inflammation.

However, caution has to be exercised when the dose of inhaled corticosteroid is reduced in asthmatics on a long-acting β-adrenoceptor agonist. McIvor et al. [39] studied the steroid-sparing and ‘masking’ effects of salmeterol vs placebo in 13 asthmatic patients requiring a daily high dose of inhaled beclomethasone, in a randomized, double-blind, cross-over trial. Corticosteroid doses were reduced weekly until criteria were met for an exacerbation or the corticosteroid was fully withdrawn. Subjects were restabilized on their original dose of inhaled corticosteroid for 4 weeks before cross-over to the alternate treatment. Mean corticosteroid dose was reduced by 87% during salmeterol treatment vs 69% with placebo. Sputum eosinophils increased before exacerbations despite stable symptoms, FEV₁ and PEF. In the week before the clinical exacerbation, mean sputum eosinophil counts were higher in the salmeterol-treatment arm (20%) compared with the placebo-treatment arm (9%). Thus, in this model of steroid-reduction, salmeterol controlled symptoms and lung function until inflammation became significantly more advanced, thereby suggesting that the bronchodilating and symptom-relieving effects of salmeterol can mask increasing inflammation and delay awareness of worsening asthma.

Observations with theophylline

With the recognition that low dosage of theophylline may have anti-inflammatory properties, it may have an adjunct role with inhaled corticosteroids in the management of chronic asthma. Very few studies have examined the anti-inflammatory effects of theophylline in sputum. An uncontrolled, open-label study by Tohda et al. [40] in 18 mild-to-moderate asthmatics did not show any effect of 4 weeks of treatment with 400 mg of a long-acting theophylline on sputum eosinophil number. However, there was a significant decrease in sputum ECP level compared with pretreatment measurement. In another uncontrolled, open-label study [41] in 24 asthmatic patients there was a significant decrease in sputum eosinophilia at the end of 8 weeks of treatment with a long-acting theophylline 400 mg daily. In a further randomized, controlled study of theophylline withdrawal in asthmatic patients controlled on low dose theophylline and inhaled budesonide, Minoguchi and coworkers [42] demonstrated a significant increase in the percentage of total and activated eosinophils in induced sputum at the end of 6 weeks, compared with the group in which the dose of theophylline was maintained. Recently, Louis et al. examined the effects of 4 weeks treatment with sustained-release oral theophylline on sputum eosinophilia in steroid-naive asthmatics in a randomized, placebo-controlled study [43]. While the 10 subjects who received placebo did not have any decrease in their sputum eosinophil count, the mean sputum eosinophil count of the 11 subjects who received theophylline decreased from 20% to 14.5%. This 35% decrease in eosinophil count was also associated with a significant decrease in the eosinophil chemotactic activity. However, it was not associated with any significant improvement in symptoms, FEV₁ or airway responsiveness. Taken together, these observations suggest that low dose theophylline may have an adjunct role to corticosteroids in controlling airway eosinophilic inflammation.

Observations with leukotriene antagonists

Inhibitors of the 5-lipooxygenase (LO) pathway including 5-LO inhibitors, 5-LO activating protein inhibitors, leukotriene B₄ antagonists and the more recent leukotriene D₄ receptor antagonists have proven anti-inflammatory effects and are useful as adjunct therapy in chronic asthma. Two studies have examined the effects of montelukast, a leukotriene D₄ anatgonist, on sputum cell counts. Pizzichini et al. [44] studied the effects of 4 weeks of treatment with montelukast on sputum cell counts in 40 adult mildly uncontrolled asthmatics with sputum eosinophilia, in a randomized, double-blind, placebo-controlled, parallel-group study. Montelukast treatment caused a significant reduction in mean sputum eosinophils from 7.5% to 3.9%. This was accompanied by an improvement in asthma symptoms, β-adrenoceptor agonist use and morning PEF. Placebo treatment, in contrast, was associated with an increase in sputum eosinophils from 14.5% to 17.9%. This study demonstrated that montelukast decreases airway eosinophilic inflammation in addition to improving clinical outcomes. Diamant and coworkers [45] examined the anti-inflammatory effect of pretreatment with montelukast on allergen inhalation in 12 mild asthmatic patients. While three oral doses of montelukast 10 mg administered 36 h and 12 h before and 12 h postallergen significantly inhibited the early and late asthmatic responses compared to placebo, they did not decrease allergen-induced sputum eosinophilia or sputum ECP level. This may be related to inadequate dosing or insufficient number of study subjects. Yoshida and colleagues examined the anti-inflammatory effects of 4 weeks treatment with pranlukast in 32 patients with mild or moderate asthma, in a randomized, double-blind, placebo-controlled, cross-over study [46]. Pranlukast decreased the mean total sputum eosinophil count from 93×10⁴/ml to 12×10⁴/ml. This was accompanied by an improvement in symptoms, but not by a clinically significant change in airway responsiveness. Further studies are needed to compare the anti-inflammatory effects of leukotriene antagonists with that of inhaled corticosteroids in patients with mildly uncontrolled asthma as well as those with prednisone-dependent asthma.

Other drugs and newer drugs

Sputum cell counts have been used to investigate the anti-inflammatory properties of nebulized lignocaine [47] and nebulized heparin [48]. Both have been reported to lower sputum eosinophilia, however, this has not been demonstrated in randomized, placebo-controlled clinical trials. The effect of cromones on sputum cell counts has not been investigated. However, a randomized, placebo-controlled study [49] of 40 mg four times daily of cromoglycate for 4 weeks has been shown to reduce eosinophil count from 10% to 4% in bronchial mucus aspirated from proximal airways. A recent study [50] suggests that cell counts from bronchial mucus aspirate are comparable to those from induced sputum. The effects of an intravenous anti-IgE monoclonal antibody on allergen induced airway responses and sputum cell counts have been examined [51]. While 9 weeks of treatment significantly reduced the median baseline (3.7% to 0.2%) and allergen induced sputum eosinophilia (7.9% to 0.7%) compared with the pretreatment values, these differences were not significantly different from that of the placebo treatment arm. Recently, the effects of therapy directed at cytokines that decrease eosinophil number have been reported [52, 53]. A single intravenous infusion of humanized monoclonal antibody to IL-5 significantly attenuated baseline and allergen induced sputum eosinophilia compared with placebo in a randomized, parallel study [52]. The effect persisted for up to 30 days after the infusion. Similarily, weekly subcutaneous injection of recombinant human IL-12 in increasing doses for 4 weeks significantly reduced baseline and allergen-provoked sputum eosinophilia in a randomized, placebo-controlled, parallel study [53]. Surprisingly, in both these studies, the attenuation of allergen-induced sputum eosinophilia was not accompanied by either a reduction in the allergen-induced late asthma response or by an improvement in the baseline or induced airway hyperresponsiveness. However, the studies were under-powered to detect differences in these outcome measurements. Furthermore, even on the placebo-treatment arm, allergen-inhalation did not cause a significant increase in airway hyperresponsiveness. These results may imply either that the eosinophil is not a central effector cell in asthma or that targeting therapy against a single cytokine and a single inflammatory cell is not adequate for asthma control. The latter seems to be the more likely explanation, although this cannot be inferred from the above studies. Indeed eosinophilic airway inflammation is only one component of the disease process in asthma [54] and is not the only inflammatory cell that determines the clinical features like variable airflow limitation and airway hyperresponsiveness. These observations emphasize the fact that sputum cell counts measure airway inflammation and not necessarily asthma control or severity, although they may be interrelated.

Sputum cell counts to select anti-inflammatory drugs and ‘add-on’ therapy

The place of sputum cell counts to monitor anti-inflammatory treatment in asthma requires prospective investigation. Although sputum eosinophilia is a predictor of loss of asthma control in patients with moderate to severe asthma when their dose of inhaled corticosteroid is reduced [55], the relevance of persistent airway eosinophilia in clinically controlled asthma [56] is not known. Whether this is a predictor of asthma exacerbations, loss of asthma control or accelerated decline in lung function needs to be investigated. We have found sputum cell counts to be useful in managing difficult asthma (Table 1) [57]. It may also be useful to select ‘add-on’ therapy [58]. Sputum eosinophilia in the situation of difficult to control asthma despite a moderate dose of inhaled corticosteroid should draw attention to possible noncompliance [59], a common allergen or occupational cause [60], or inadequate steroid treatment. Alternatively, there may be no eosinophilia but a neutrophilia, raising the possibility of misdiagnosis, associated disease or steroid-resistant asthma. If the patient is already on steroid treatment, these findings would raise the possibility that the steroid dose can be reduced or an additional agent can be added, for example, a long-acting β-adrenoceptor agonist, a leukotriene antagonist or theophylline. There is no current evidence that one of these is better than the other.

Table 1.

Assessment and management of ‘difficult asthma’ using sputum cell counts.

Sputum cell count	Possible causes	Possible therapy
Normal	eosinophilia controlled with glucocorticoids hyperventilation vocal-cord dysfunction airway hyperresponsiveness malingering	reduce glucocorticoids psychotherapy voice therapy long-acting β-adrenoceptor agonists or leukotriene antagonists psychotherapy
Eosinophilia	noncompliance allergen or occupational inadequate steroid dose	improve compliance avoidance measures increase steroid dose or add leukotriene antagonist
Neutrophilia	infection smoker's bronchitis glucocorticoid resistance occupational cause, e.g. grain dust	antibiotic long-acting β-adrenoceptor agonist reduce corticosteroid avoidance of cause

Statistical issues in using sputum cell count as outcome measure

A meaningful interpretation of changes in sputum cell count as an outcome measure requires a knowledge about their normal values and their repeatability. Sputum cell counts in normal healthy adults have been examined [61] (Table 2). They are repeatable when measured 48 h apart in patients with stable asthma [4] and following an allergen inhalation [62]. For the eosinophil differential count, Gauvreau and collegaues [62] have demonstrated an intraclass correlation coefficient of 0.60 at 7 h and 0.53 at 24 h after an allergen inhalation. This implies that in a cross-over study design, only five subjects are necessary to observe a 50% attenuation of allergen-induced percent of eosinophils, with 95% power. A minimal clinically important change in eosinophil cell count, however, has not yet been identified. This needs to be determined to calculate adequate sample size for clinical studies.

Table 2.

Normal values for sputum cell counts.

Cell count	Mean (s.d.)	Median (IQR)	90th centile
TCC ( × 106/g)	4.1 (4.8)	2.4 (3.2)	9.7
N (%)	37.5 (20.1)	36.7 (29.5)	64.4
Eo (%)	0.4 (0.9)	0.0 (0.3)	1.1
M (%)	58.8 (21.0)	60.8 (28.9)	86.1
L (%)	1.0 (1.1)	0.5 (1.8)	2.6
MC (%)	0.0 (0.04)	0.0 (0.0)	0.04
BE (%)	1.6 (3.9)	0.3 (1.3)	4.4

TCC: total cell count, N: neutrophil, Eo: eosinophil, M: macrophage, L: lymphocyte, MC: metachromatic cell, BE: bronchial epithelial cell, s.d.: standard deviation, IQR: interquartile range.

Conclusion

Induced sputum examination for inflammatory markers holds much promise in the evaluation of the anti-inflammatory properties of asthma medications and identifying the appropriate medication to control symptoms. They may be useful to compare the relative potencies of different anti-inflammatory drugs. Further research is required to investigate the anti-inflammatory effect of newer asthma drugs, to identify the relevance of suppression of airway eosinophilia in the long term management of asthma and to establish a minimally important clinically significant change in airway inflammatory markers including sputum eosinophil count.