ARTICLE IN PRESS

Respiratory Medicine (2007) 101, 1652–1658

0954-6111/$ - see frdoi:10.1016/j.rmed.

$The study was f�Corresponding au

E-mail address: h

Effects of montelukast compared to double dosebudesonide on airway inflammation andasthma control$

Neil Barnesa, Michel Lavioletteb, David Allenc, Patrick Flood-Paged,Frederick Hargreavee, Paul Corrisf, Brian J. O’Connorg, Helen Tateh,�,Debbie Parkeri, Ian Pavordi

aDepartment of Respiratory Medicine, London Chest Hospital, Bonner Road, London E2 9JX, UKbCentre de pneumolgie, Hopital Laval, 2725 Chemin Sainte-Foy, Quebec (Qc), Canada G1V 4G5cDepartment of Respiratory Medicine, North Manchester General Hospital, Delauneys Road, Crumpsall,Manchester M8 5RB, UKdNewport Chest Clinic, 129 Stow Hill, Newport, South Wales NP20 4GA, UKeFirestone Institute for Respiratory Health, St. Joseph’s Healthcare, 50 Charlton Avenue East, Hamilton, Ont.,Canada L8N 4A6fInstitute of Cellular Medicine, Newcastle University and Freeman Hospital, Newcastle upon Tyne NE7 7DN, UKgDepartment of Asthma, Allergy and Respiratory Science, Kings College London School of Medicine, Bessemer Road,London SE5 9PJ, UKhWCDMO, Merck Sharp and Dohme Ltd., Hertford Road, Hoddesdon, EN11 9BU, UKiDepartment of Respiratory Medicine Clinical Sciences, Glenfield Hospital, Groby Road, Leicester LE3 9QP, UK

Received 16 October 2006; accepted 15 March 2007Available online 3 May 2007

KEYWORDSAsthma;Clinical trial;Eosinophils;Mediators;Sputum

ont matter & 20072007.03.007

unded by Merck, Sthor. Tel.: +44 199

elen_tate@merck

SummaryMany patients with asthma remain symptomatic with impaired airway function on inhaledsteroids. This study investigates the relationship between the clinical effect seen inresponse to additional treatment and the effect on airway inflammatory indices.Seventy-five adult asthmatic patients, incompletely controlled on 800mcg budesonide/day, were randomised following a 4 week run-in period, to a double-blind, multi-centrecontrolled clinical trial of doubling inhaled corticosteroid (budesonide 1600mcg/day) oradding 10mg montelukast for 12 weeks. Induced sputum was collected at baseline and endof treatment and analysed for eosinophil and neutrophil percentages, leukotrienes C4, D4

and E4, IL-8, Eosinophil Cationic Protein (ECP) and histamine.Sputum evidence of inflammation (X2:0% eosinophils) was seen in only 29% of thesepatients and the percentage of eosinophils and other markers of airway inflammation did

Elsevier Ltd. All rights reserved.

harp & Dohme Ltd., Hoddesdon, Herts, UK.2 452161; fax: +44 1992 458724.

.com (H. Tate).

ARTICLE IN PRESS

Asthma: Sptum cells and mediators 1653

not change over the study period in either treatment group. There were significantimprovements in am PEF (montelukast: 31.7 L/min, budesonide: 32.3 L/min) and quality oflife with both treatments.We conclude that while both treatments showed similar improvements in lung function andquality of life, there was no evidence from these sputum markers measured that theeffects were mediated via a reduction in airway inflammation or that the level of pre-treatment markers was associated with outcome.& 2007 Elsevier Ltd. All rights reserved.

Introduction

In mild to moderate asthmatic patients taking bronchodila-tors only, numerous studies have demonstrated the presenceof airway inflammation.1–6 In biopsy studies there is anincrease in eosinophils and CD4 positive T lymphocytes,thickening of the basement membrane and epithelialdisruption.1 In sputum the characteristic finding in sympto-matic patients on bronchodilators is of sputum eosinophi-lia.2–6 When patients are treated with inhaledcorticosteroids many of these abnormalities improve6 butsymptoms can continue in up to 50% of patients.7 The extentto which these residual symptoms are causally linked toeosinophilic inflammation is unclear, because the inflamma-tion was not measured.

There is a range of treatment choices in symptomaticpatients on inhaled steroids including increasing the doseof inhaled corticosteroids, the addition of long-actinginhaled bronchodilator, the addition of leukotriene receptorantagonist or the addition of slow-release theophyllines.7

There is continuing debate about which of these is thepreferred option. Part of this debate is fuelled by differingviews regarding the relative importance of the differentmodes of action of these agents. Inhaled corticosteroids’major mode of action is to decrease airway inflammation.Long-acting inhaled b2-agonists primarily relax airwaysmooth muscle, and there is little compelling evidence ofanti-inflammatory effect of these agents in chronic asthma.8

The leukotriene receptor antagonists relax airway smoothmuscle by antagonising the effect of broncho-constrictorcysteinyl leukotrienes but they also have an anti-inflamma-tory effect, distinct from that of inhaled corticosteroids.9

There are only a limited number of studies that haveinvestigated the additional anti-inflammatory effects ofdifferent strategies in patients already taking inhaledcorticosteroids.

There is evidence from a number of placebo controlledstudies that the addition of a leukotriene antagonist, inpatients inadequately controlled on inhaled corticosteroids,produces improvements in pulmonary function, symptoms,b2-agonist use and asthma exacerbations,10,11 althoughwhether these effects are associated with anti-inflammatoryeffects is unknown. We investigated in symptomatic patientswith impaired airway function on moderate doses of inhaledsteroid (budesonide 800mcg/day) the effect of the additionof a leukotriene receptor antagonist (montelukast 10mg/day) versus doubling the dose of inhaled corticosteroids andmeasured airway inflammation. We used identical inclusionand exclusion criteria and an identical protocol to a largerstudy12 but added measurement of sputum markers of

inflammation at baseline and end of treatment to provide aninsight into the effect.

Methods

Subjects

The study design was identical to that of the larger clinicalstudy.12 Subjects were non-smokers aged 15–70 years with41 year history of asthma symptoms. Their currenttreatment was inhaled steroids (600–1200mcg/day budeso-nide or equivalent) and b2-agonist only. Their bronchodilatorreversibility was X12% ðFEV1Þ or X15% (peak expiratoryflow, PEF) and they had 450% predicted FEV1=PEF. Despitetreatment, subjects remained symptomatic during the last 2weeks of the 4 week run-in period, as assessed by a symptomdiary and daily b2-agonist use (on average at least one puffper day). Exclusions included any other pulmonary disorder,emergency treatment for asthma within 1 month, hospita-lisation within 2 months or respiratory tract infection within3 weeks.

Study design

The study was a multi-centre (four centres in the UK and twocentres in Canada), double-blind, parallel group randomisedcontrolled trial. Inhaled budesonide (800mcg/day) wasgiven during the 4 week run-in period, following which 75patients were randomised to receive either 800mcgbudesonide via Turbuhaler plus 10mg montelukast daily or1600mcg budesonide plus placebo for the 12 week treat-ment period. Induced sputum samples were taken atbaseline and after the 12 week treatment period. Dailydiary cards were used to record symptoms,13 PEF and b2-agonist use throughout the 16 weeks of the study period.Asthma-specific quality of life questionnaire14 was com-pleted at baseline and end of treatment. Compliance wasassessed from both the diary card and the pill count.

Induction and processing of sputum was conducted ateach centre using methods described elsewhere.5 Differ-ential counts were performed in each country (GlenfieldHospital, UK; St. Joseph’s Hospital, Canada). Histamine,cysteinyl leukotrienes and IL-8 were measured by ELISA, ECPby fluoroenzymeimmunassay at Glenfield Hospital usingmethods described previously.15

The primary endpoint was change from baseline in sputumeosinophil percentage. Secondary endpoints included othersputum inflammatory markers, percentage change frombaseline (last 10 days of the run-in period) in morning PEF

ARTICLE IN PRESS

N. Barnes et al.1654

(best of 3 readings) over the last 10 weeks of treatment,asthma-specific quality of life and percentage of days withasthma exacerbations (defined as: decrease in am PEF of420%; am PEFo180 L=min; increase in b2-agonist use of470%; increase in symptom score of 450% or an asthmaattack resulting in an unscheduled medical visit or corticos-teroid tablets).

Adverse events were assessed for severity, duration andstudy drug causality. The study received ethical reviewcommittee approval. Patients gave written informed con-sent.

Statistical considerations

The analysis was primarily descriptive. Protocol violators,identified before the study was unblinded, were excludedfrom the main efficacy analyses. The primary efficacymeasurement was analysed non-parametrically, because ofthe large proportion of zero counts; a 95% confidenceinterval for the median difference between treatmentgroups was estimated.16 Additional Poisson and negative-binomial modelling was undertaken. Neutrophil percentageswere analysed parametrically using analysis of covariancewith model factors of country and treatment group and thebaseline value as a covariate. The country by treatmentinteraction was assessed and omitted if not significant. A95% confidence interval for the mean difference in changefrom baseline between treatments was derived. All othermarkers followed highly positively skewed distributions andwere transformed to logarithms prior to parametric analysis,leading to a 95% confidence interval for the ratio of changebetween groups. Secondary efficacy measurements: changein morning PEF and asthma-specific quality of life wereassessed by analysis of covariance. The percentage of dayswith asthma exacerbations followed a highly skewed patternand therefore was analysed non-parametrically. Graphicalmethods and Spearmen’s correlation coefficients were usedto assess possible relationships between baseline eosinophilcounts and PEF and changes in eosinophil levels withchanges in PEF, within each treatment group.

Planned analysis included only patients with baselineeosinophils X2%. During recruitment, few patients hadeosinophils X2% and the decision was made to include allpatients and perform a subgroup analysis on those X2%.Little data were available on eosinophil counts when thestudy was designed and the sample size was based onprecision considerations. Thirty patients per group wouldensure the between group comparison of change from

Table 1 Demography and baseline characteristics.

Montelukast + budeso800 mg ðN ¼ 37Þ

Age (years) mean (SD) 41.5 (11.7)Gender M:F 14:23Height (cm) mean (SD) 167.6 (10.4)AM average PEF1 (L/min) mean (SD) 436.6 (149.3)% predicted FEV1 mean (SD) 74.1 (13.0)Quality of life mean (SD) 4.9 (1.1)

baseline was reasonably precise (the 95% confidence intervalwould span a distance approximately equal to the standarddeviation unit).

Results

Seventy-five patients were randomised (Table 1). Compli-ance was over 98% for all drugs from the diary card and 77%for montelukast and 84% for placebo from drug accounting.One patient in the montelukast group was designated aprotocol violator because he took prednisolone. In thebudesonide group there were four protocol violators (twoviolated the inclusion criteria, and two took prednisolone)and one patient withdrew consent. Paired sputum eosinophilvalues were available for 24 subjects in the montelukastgroup and 28 in the double dose budesonide group.

The eosinophil percentages indicating the change for eachpatient (paired data) are shown in Figs. 1A and B, for themontelukast and double dose budesonide groups, respec-tively. At baseline 15 out of the 52 patients for whom pairedeosinophil counts were available (six in the montelukastgroup and nine in the double dose budesonide group) had nodetectable eosinophils and altogether 37 (17 in themontelukast group and 20 in the double dose budesonidegroup) had values lower than 2.0%. The distribution ofpercentage eosinophils was highly skewed, the largest valuebeing 27% and the median 0.8%. The median sputumeosinophil counts before and after montelukast were 1.4%and 0.9%, respectively. The median sputum eosinophilcounts before and after double dose budesonide were 0.5%and 0.7%. There were no significant changes over thetreatment period and the confidence interval for the mediandifference between groups was �0:60 to 0:95%, indicatingthat the groups were similar. The Poisson distribution andnegative binomial distributions did not fit the data well; theoverall estimates for the treatment group difference usingthese models were in line with the conclusion of notreatment difference seen from the non-parametric ap-proach. A subgroup analysis of those patients with a baselinevalue of 2% or over (seven in the montelukast group andeight in the double dose budesonide group, see Fig. 1)showed that the median sputum eosinophil count decreasedfrom 6.3% to 2.0% (95% CI of median change �7:0% to 21:3%)with montelukast and from 8.7% to 1.6% (95% CI of medianchange �7:3% to 5:3%) with double dose budesonide.

Other sputum cell and mediator results, before and aftertreatment, are shown in Table 2. There were no within groupdifferences over the treatment period for either treatment

nide Budesonide 1600mg ðN ¼ 38Þ

45.0 (14.2)20:18169.9 (9.7)449.7 (145.0)73.6 (14.2)4.9 (0.9)

ARTICLE IN PRESS

Sp

utu

m e

osin

op

hil

co

un

t (%

)

Sp

utu

m e

osin

op

hil

co

un

t (%

)

N = 2

N = 50

1

10

100

Baseline End of Treatment

0

1

10

100

Baseline End of Treatment

Figure 1 Graphs of individual sputum eosinophil counts (%) at baseline and end of treatment. Panel A: budesonide 800mgþmontelukast 10mg group ðn ¼ 24Þ, Panel B: budesonide 1600mg group ðn ¼ 28Þ. Solid lines join observations in the same subject. Theheavy solid line joins the median values at the two time points.

Asthma: Sptum cells and mediators 1655

group for any of the inflammatory markers. No differencesbetween treatment groups in change from baseline wereapparent—all confidence intervals include the null value (0for mean/median for eosinophils, neutrophils, days withasthma exacerbations and QoL and 1 for the ratios for othermarkers).

The treatment effect on PEF over the study period isshown in Fig. 2 (paired data). Morning PEF increased by31.7 L/min (95% CI 10.8 to 52.6) in the montelukast groupand 32.3 L/min (95% CI 10.7 to 53.9) in the double dosebudesonide group (95% confidence interval for the differ-ence between groups was �29:4 to 30:6 L=min). The medianpercentage of days with asthma exacerbations (as definedabove) was similar in the two groups (see Table 2).

There was a clinically significant improvement in theadjusted total asthma-specific quality of life score of 0.73points (95% CI 0.46 to 1.00) with montelukast and 0.48points (95% CI 0.20 to 0.77) with double dose budesonide.There was no significant difference between treatmentgroups ðp ¼ 0:21Þ.

Correlation plots showed no associations between base-line sputum eosinophil count or change in sputum eosinophilcount and change in morning PEF in either treatment group(Spearmen’s correlation coefficients were all less than 0.25,associated p values were all greater than 0.2).

There were no serious adverse events. Twenty-eightpatients in the montelukast group (76%) and 33 (87%) inthe double budesonide group experienced minor adverseevents.

Discussion

This study is one of the first to investigate the effect of anti-asthmatic drugs on airway inflammation in patients alreadytaking inhaled corticosteroids. It was a large multi-centrestudy with a total of 75 patients taking part. Our aim was toinvestigate the effect of doubling the dose of inhaledsteroids or adding montelukast on sputum measures of

airway inflammation in patients taking inhaled corticoster-oids and to see if there was any relationship betweenchanges in airway inflammation and clinical outcomes. Thestudy was designed to give a mechanistic insight into thelarger COMPACT study12 and used identical inclusion andexclusion criteria. The patient groups recruited had similardemography and severity of asthma to those in the COMPACTstudy. The results of the current study show very similarclinical effects to the COMPACT study, with an improvementin PEF and QoL in both the montelukast and budesonidetreatment groups. As in the larger COMPACT study the timecourse of improvement in PEF was sustained throughout thetreatment period, strongly suggesting genuine drug effectsrather than regression to the mean. The important newfinding is that these clinical benefits were not associatedwith changes in sputum markers of airway inflammation.

Our findings are consistent with the findings of earlierstudies showing that montelukast had no effect on eosino-philic airway inflammation in patients treated with inhaledcorticosteroids.17,18 In contrast, two studies have shownthat treatment with montelukast is associated with asignificant reduction in sputum eosinophil counts in patientswith asthma treated with bronchodilators only.19,20 Thesefindings suggest a difference between the effects ofmontelukast on eosinophilic airway inflammation in patientson and off inhaled corticosteroids.

Complete sputum data were only available for 69% ofpatients, partially because this was a multi-centre trialinvolving some centres with no prior experience in inducedsputum methodology. The incomplete data are unlikely tocause a systematic bias in the study as they were evenlydistributed between the montelukast and doubled dosesteroid groups. This does lead to a reduction in power, butthe primary approach to the analysis was descriptive ratherthan hypothesis testing.

A relevant finding in the current study is that approxi-mately two-thirds of patients with symptomatic asthma oninhaled corticosteroids do not have a sputum eosinophilcount outside the normal range ðo2%Þ. This is consistent

ARTICLE IN PRESS

Table

2Su

mmarystatistics

andch

ange

sfrom

baselineforou

tcom

emea

sures(paireddata).

Mon

teluka

stþ

bud

eson

ide80

0mg

Bud

eson

ide16

00mg

Adjusted

differenc

eor

ratiobetwee

ngrou

pswith95

%CI

Baseline

Endof

trea

tmen

tBaseline

Endof

trea

tmen

t

Eosino

phils

%med

ian(ran

ge)

1.4(0.0,9.0)

0.9(0.0,27.5)

0.5(0.0,27

.0)

0.7(0.0,16

.0)

0.00

(�0.60

,0.95)

Neu

trop

hils

%mea

n(SD)

39.3

(22.5)

36.3

(25.3)

49.8

(29.8)

44.6

(31.5)

2.4(�

17.4,12.6)

ECPng

/mlge

ometricmea

n(ran

ge)

302.9(22.8,

9540

)37

0.0(23.3,

3456

)46

9.9(18.5,

2583

0)31

7.1(32.1,

1161

0)0.74

(0.34,

1.62

)C4,D4,E 4

ng=m

lge

ometricmea

n(ran

ge)

1.6(0.4,29

.9)

2.3(0.3,18

.2)

1.9(0.5,17

.9)

1.7(0.3,36

.6)

0.68

(0.37,

1.28

)IL-8

ng/m

lge

ometricmea

n(ran

ge)

9.5(1.1,74

.0)

13.3

(0.2,84

.8)

12.6

(0.1,83

.8)

13.9

(2.5,84

.0)

1.02

(0.51,

2.04

)Histamineng

/mlge

ometricmea

n(ran

ge)

2.3(0.2,14

6.6)

3.1(0.2,88.9)

2.6(0.2,40.0)

2.5(0.2,72.8)

0.72

(0.32,

1.64

)Morning

PEF

L/min

mea

n(SD)

441.5(148

.3)

473.4(167

.4)

448.4(153

.8)

480.7(154

.6)

0.6(�

29.4,30

.6)

Day

swithasthmaex

acerbations

%med

ian(ran

ge)

3.4(0.0,94

.3)

6.9(0.0,34

.1)

0.9(�

1.2,

6.1)

Totalasthmaspec

ificqua

lity

oflife

mea

n(SD)

4.96

(1.10)

5.60

(1.03)

4.84

(0.93)

5.36

(0.92)

�0.25

(�0.64

,0.15

)

N. Barnes et al.1656

with the view that there is not a simple causal relationshipbetween eosinophilic inflammation, asthma symptoms anddisordered lung function.21

Overall there was no evidence for a decrease ineosinophilic inflammation with either a doubling dose ofinhaled corticosteroids or the addition of montelukast.There are several possible explanations for these findings.It could be that sputum eosinophilia does not capture theanti-inflammatory effects of either high dose inhaledsteroids or the addition of a leukotriene receptor antagonistand that alternative techniques such as bronchial biopsy ormeasurement of exhaled markers of inflammation arenecessary to detect an effect. However, experience withthe anti-IL-5 monoclonal antibodies suggests that theinduced sputum eosinophil count is more responsive tochange than tissue eosinophil counts,22 suggesting that thisis not the case. It is possible that montelukast and increasedinhaled steroid strategies are affecting other functionallyimportant aspects of airway inflammation that are notdemonstrated by induced sputum analysis such as airwayoedema23 or the number of mast cells in airway smoothmuscle.24

Another reason for a failure overall to detect an anti-inflammatory effect may be that the general level ofeosinophilic inflammation was so low at baseline that therewas no room for either drug to improve this. Some supportfor this suggestion comes from looking at the approximatelyone-third of the patients who had an eosinophil level abovethe normal range ðX2:0%Þ. In this group of patients both theaddition of montelukast and an increase in the dose ofbudesonide produced a small decrease in the sputumeosinophilia which was not statistically significant, althoughpower to detect a difference was very low. This group ofpatients were a pre-specified group described in the analysisplan. In the absence of a placebo group it is possible that thechanges seen were due to regression to the mean. Clearlyfurther investigation of the subgroup with a raised eosino-phil count would be worthwhile, although an earlier smallstudy suggests that montelukast treatment does not reducethe sputum eosinophil count in patients with persistenteosinophilic airway inflammation despite treatment withinhaled corticosteroids.17

A secondary aim of the study was to investigate whetherthere was any correlation between the baseline level ofairway inflammation measured in sputum and improvementin clinical asthma, or changes in airway inflammation andmeasures of improvement in clinical asthma. We found nocorrelation between any of the baseline sputum measure-ments or any of the changes in airway inflammation, withshort-term measures of asthma control such as PEF. Thisadds to data indicating that there is little evidence for arelationship between measures of inflammation or lungfunction and symptoms in patients treated with inhaledcorticosteroids.21,24 There is, however, evidence thatpersisting eosinophilia is related to risk of asthma exacer-bation.24 The patients in the current study did notexperience a sufficient level of exacerbations to investigatewhether there was any relation between risk of exacerba-tion and airway eosinophilia.

In conclusion the current study has confirmed thebeneficial effects of both increasing the dose of inhaledsteroid or addition of montelukast in symptomatic patients

ARTICLE IN PRESS

AM

PE

F (

l/m

in)

420

440

460

480

500

520

Days after Randomisation

-14 -7 0 7 14 21 28 35 42 49 56 63 70 77 84

Figure 2 Morning peak expiratory flow (am PEF) over the 12-week treatment period, paired data (solid line ¼ budesonide 800mgþmontelukast 10mg daily ðn ¼ 24Þ, dashed line ¼ budesonide 1600mg daily ðn ¼ 28Þ). Data represent the mean am PEF measuredbefore administration of study medication.

Asthma: Sptum cells and mediators 1657

treated with moderate doses of inhaled steroid. It demon-strated that despite symptomatic asthma nearly two-thirdsof patients have no measurable eosinophilic airway inflam-mation, which adds to the growing body of evidence that therelationship between asthmatic symptoms, disordered lungfunction and airway eosinophilia is complex.

Acknowledgements

The authors thank David Thompson for data analysis, andSonia Fairbairn, Clare Lindsay and Richard Tomiak for thestudy management and assistance with the manuscript.Thanks also go to the research staff at all of theparticipating sites for their commitment to this study.

References

1. Jeffrey PK, Laitinen A, Venge P. Biopsy markers of airwayinflammation and remodelling. Respir Med 2000;94(Suppl.F):S9–S15.

2. Fahy JV, Boushey HA, Lazarus SC, et al. Safety and reproduci-bility of sputum induction in asthmatic subjects in a multicentrestudy. Am J Respir Crit Care Med 2001;163:1470–5.

3. In’t Veen JCCM, De Gouw HWFM, Smits HH, et al. Repeatabilityof cellular and soluble markers of inflammation in inducedsputum from patients with asthma. Eur Respir J 1996;9:2441–7.

4. Pizzichini E, Pizzichini MMM, Efthimidias A, et al. Indices ofairway inflammation in induced sputum: reproducibility andvalidity of cell and fluid-phase measurements. Am J Respir CritCare Med 1996;154:308–17.

5. Pavord ID, Pizzichini MM, Pizzichini E, Hargreave FE. The use ofinduced sputum to investigate airway inflammation. Thorax1997;52:498–501.

6. Pavord ID, Brightling CE, Woltmann G, Wardlaw AJ. Non-eosinophilic corticosteroid unresponsive asthma. Lancet1999;353:2213–4.

7. British Thoracic Society; Scottish Intercollegiate GuidelinesNetwork. British guideline on the management of asthma.Thorax 2003; 58 (Suppl. 1): 1–94.

8. Pizzichini MMM, Kidney JC, Wong BJO, et al. Effect ofsalmeterol compared with beclomethasone on allergen-inducedasthmatic and inflammatory responses. Eur Respir J1996:449–55.

9. Sampson AP, Pizzichini E, Bisgaard H. Effects of cysteinylleukotrienes and leukotriene receptor antagonists on markersof inflammation. J Allergy Clin Immunol 2003;111:S49–59.

10. Laviolette M, Malmstrom K, Lu S, et al. Montelukast added toinhaled becolmethasone in treatment of asthma. Am J RespirCrit Care Med 1999;160:1862–8.

11. Virchow JC, Prasse A, Naya I, Summerton L, Harris A. Zafirlukastimproves asthma control in patients receiving high-dose inhaledcorticosteroids. Am J Respir Crit Care Med 2000;162(2 Pt1):578–85.

12. Price DB, Hernandez Magyar DP, Fiterman J, et al. Randomisedcontrolled trial of montelukast plus inhaled budesonide versusdouble dose inhaled budesonide in adult patients with asthma.Thorax 2003;58:211–6.

13. Santanello NC, Barber BL, Reiss TF, Friedman BS, Juniper EF,Zhang J. Measurement characteristics of two asthma symptomdiary scales for use in clinical trials. Eur Respir J1997;10:646–51.

14. Juniper EF, Guyatt GH, Epstein RS, Ferrie PJ, Jaeschke R, HillerTK. Evaluation of impairment of health related quality of life inasthma: Development of a questionnaire for use in clinicaltrials. Thorax 1992;47(2):76–83.

15. Brightling CE, Ward R, Woltmann G, et al. Induced sputuminflammatory mediator concentrations in eosinophilic bronchitisand asthma. Am J Respir Crit Care Med 2000;162:878–82.

16. Hodges Jr JL, Lehmann EL. Estimates of location based on ranktests. Ann Math Stat 1963;27:598–611.

ARTICLE IN PRESS

N. Barnes et al.1658

17. Jayaram L, Duong M, Pizzichini MMM, et al. Failure ofmontelukast to reduce sputum eosinophilia in high-dosecorticosteroid-dependent asthma. Eur Respir J 2005;25:41–6.

18. Green RH, Brightling CE, McKenna S, et al. Comparison ofasthma treatment given in addition to inhaled corticosteroidson airway inflammation and airway responsiveness. Eur Respir J2006;27:1144–51.

19. Pizzichini E, Leff JA, Reiss TF, et al. Montelukast reduces airwayeosinophilic inflammation in asthma: a randomized, controlledtrial. Eur Respir J 1999;14:12–8.

20. Jayaram L, Pizzichini E, Lemiere C, et al. Steroid naiveeosinophilic asthma: anti-inflammatory effects of fluticasoneand montelukast. Thorax 2005;60:100–5.

21. Green RH, Brightling CE, McKenna S, et al. Asthma exacerba-tions and sputum eosinophil counts: a randomised controlledtrial. Lancet 2002;360:1715–21.

22. Flood-Page PT, Menzies-Gow AN, Kay AB, Robinson DS.Eosinophil’s role remains uncertain as anti-interleukin-5 onlypartially depletes numbers in asthmatic airway. Am J Respir CritCare Med 2003;167(2):199–204.

23. Brown RH, Zerhouni EA, Mitzner W. Airway edema potentiatesairway reactivity. J Appl Physiol 1995;79:1242–8.

24. Brightling CE, Bradding P, Symon FA, Holgate ST, Wardlaw AJ,Pavord ID. Mast cell infiltration of airway smooth muscle inasthma. N Engl J Med 2002;346:1699–705.