Chronic obstructive pulmonary disease among residents of an historically industrialised area

ThoraxAN INTERNATIONAL JOURNAL OF RESPIRATORY MEDICINE

67

10

Volum

e 67 Issue 10 Pages 849–934 TH

OR

AX

O

ctober 2012

October 2012 Volume 67 Issue 10

thorax.bmj.com

thoraxjnl_67_10_Cover.indd 1thoraxjnl_67_10_Cover.indd 1 05/09/12 5:01 PM05/09/12 5:01 PM

Contents Volume 67 Issue 10 | THORAX October 2012

Editorials849 What you don’t know can hurt you; early

asymptomatic lung disease in cystic fi brosis

S Cunningham

851 The new microbiology of cystic fi brosis: it takes a community

J LiPuma

Original articlesCystic fi brosis853 Early antibiotic treatment for Pseudomonas

aeruginosa eradication in patients with cystic fi brosis: a randomised multicentre study comparing two different protocols

G Taccetti, E Bianchini, L Cariani, R Buzzetti, D Costantini, F Trevisan, L Zavataro, S Campana, on behalf of the Italian Group for P aeruginosa eradication in cystic fi brosis

860 Understanding the natural progression in %FEV

1 decline in patients with cystic fi brosis:

a longitudinal study

D Taylor-Robinson, M Whitehead, F Diderichsen, H V Olesen, T Pressler, R L Smyth, P Diggle

867 Long-term cultivation-independent microbial diversity analysis demonstrates that bacterial communities infecting the adult cystic fi brosis lung show stability and resilience

F A Stressmann, G B Rogers, C J van der Gast, P Marsh, L S Vermeer, M P Carroll, L Hoffman, T W V Daniels, N Patel, B Forbes, K D Bruce

874 Lung function is abnormal in 3-month-old infants with cystic fi brosis diagnosed by newborn screening

A-F Hoo, L P Thia, The Thanh Diem Nguyen, A Bush, J Chudleigh, S Lum, D Ahmed, I B Lynn, S B Carr, R J Chavasse, K L Costeloe, J Price, A Shankar, C Wallis, H A Wyatt, A Wade, J Stocks, on behalf of the London Cystic Fibrosis Collaboration (LCFC)

882 Progress in cystic fi brosis and the CF Therapeutics Development Network

S M Rowe, D S Borowitz, J L Burns, J P Clancy, S H Donaldson, G Retsch-Bogart, S D Sagel, B W Ramsey

Cough891 Transient receptor potential channels mediate

the tussive response to prostaglandin E2 and

bradykinin

M Grace, M A Birrell, E Dubuis, S A Maher, M G Belvisi

Chronic obstructive pulmonary disease901 Chronic obstructive pulmonary disease among

residents of an historically industrialised area

A C Darby, J C Waterhouse, V Stevens, Clare G Billings, C G Billings, C M Burton, C Young, J Wight, P D Blanc, D Fishwick

Respiratory infection908 Cigarette smoke and platelet-activating factor

receptor dependent adhesion of Streptococcus pneumoniae to lower airway cells

J Grigg, H Walters, S S Sohal, R Wood-Baker, D W Reid, C-B Xu, L Edvinsson, M C Morissette, M R Stämpfl i, M Kirwan, L Koh, R Suri, N Mushtaq

Smoking914 Smoking, acute mountain sickness and altitude

acclimatisation: a cohort study

T-Y Wu, S-Q Ding, J-L Liu, J-H Jia, Z-C Chai, R-C Dai, J-Z Zhao, Q D Tang, B Kayser

Sleep920 The effect of continuous positive airway

pressure usage on sleepiness in obstructive sleep apnoea: real effects or expectation of benefi t?

M R Crawford, D J Bartlett, S R Coughlin, C L Phillips, A M Neill, C A Espie, G C Dungan II, J P H Wilding, P M A Calverley, R R Grunstein, N S Marshall

PostScript925 Correspondence

This article has been chosen by the Editor to be of special interest

or importance and is freely available online.

Articles carrying the Unlocked Logo are freely available online

under the BMJ Journals unlocked scheme.

See http://thorax.bmj.com/info/unlocked.dtl

This journal is a member of and subscribes to the principles of the

Committee on Publication Ethics

www.publicationethics.org.uk

Journal of the British Thoracic Society

Impact Factor: 6.84

EditorsA Bush (UK)

I Pavord (UK)

Deputy EditorsP Cullinan (UK)

C Lloyd (UK)

Associate EditorsR Beasley (New Zealand) E Lim (UK)

J Brown (UK) N Maskell (UK)

R Buhl (Germany) JL Pepin (France)

JC Celedón (USA) T Sethi (UK)

A Custovic (UK) M Steiner (UK)

A Fisher (UK) D Thickett (UK)

P Gibson (Australia) A Torres (Spain)

J Grigg (UK) Z Udwadia (India)

D Halpin (UK) H Zar (South Africa)

A Jones (UK)

Statistical EditorsJ Gibson (UK)

T McKeever (UK)

StatisticianL Szatkowski (UK)

Journal club EditorJ Quint (UK)

President, British Thoracic SocietyA Greening

Editorial Offi ceBMJ Publishing Group Ltd, BMA House,

Tavistock Square, London WC1H 9JR, UK

T: +44 (0)20 7383 6373F: +44 (0)20 7383 6668E: [email protected]

ISSN: 0040-6376 (print)ISSN: 1468-3296 (online)

Disclaimer: Thorax is owned and published by the British Thoracic Society and BMJ Publishing Group Ltd, a wholly owned subsidiary of the British Medical Association. The owners grant editorial freedom to the Editor of Thorax.

Thorax follows guidelines on editorial independence produced by the World Association of Medical Editors and the code on good publication practice of the Committee on Publication Ethics.

Thorax is intended for medical professionals and is provided without warranty, express or implied. Statements in the Journal are there sponsibility of their authors and advertisers and not authors’institutions, the BMJ Publishing Group Ltd, theBritish Thoracic Society or theBMAunless otherwisespecifi ed or determined by law. Acceptance of advertising does not imply endorsement.

To the fullest extent permitted by law, the BMJPublishing Group Ltd shall not be liable for any loss,injury or damage resulting from the use of Thoraxor any information in it whether based on contract,tort or otherwise. Readers are advised to verify any information they choose to rely on.

Copyright: © 2012 BMJ Publishing Group Ltd and the British Thoracic Society. All rights reserved; no part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without the prior permission of Thorax.

Thorax is published by BMJ Publishing Group Ltd, typeset by Techset and printed in the UK on acid-free paper.

Thorax (ISSN No: 0040–6376) is published monthly by BMJ Publishing Group and distributed in the USA by Air Business Ltd. Periodicals postage paid at Jamaica NY 11431. POSTMASTER: send address changes to Thorax, Air Business Ltd, c/o Worldnet Shipping Inc., 156-15, 146th Avenue, 2nd Floor, Jamaica, NY 11434, USA.

MORE CONTENTS �

Cover caption: The current US Cystic Fibrosis Foundation pipeline showing the progression of new medications through to clinical usage.

thoraxjnl_67_10_TOC.indd 1thoraxjnl_67_10_TOC.indd 1 05/09/12 5:12 PM05/09/12 5:12 PM

Chest clinicAudit, research and guideline update928 British Thoracic Society national bronchiectasis

audit 2010 and 2011

A T Hill, S Welham, K Reid, C E Bucknall, On behalf of the British Thoracic Society

Case based discussion931 A therapeutic conundrum: recurrent cystic-

fi brosis-related haemoptysis complicated by acute pulmonary embolism

W G Flight, R J Bright-Thomas, S Butterfi eld, A M Jones, A Kevin Webb

Images in Thorax933 Pulmonary complications of intravesicular BCG

immunotherapy

B Davies, H Ranu, M Jackson

Miscellaneous866 Journal club: Histamine-releasing factor: a

possible future therapeutic target for asthma and allergy

V E Beasley

873 Journal club: Tracheobronchial transplantation with a bioartifi cal nanocomposite

K Sritharan

881 Journal club: Restoration of function of the ∆F508 mutation in cystic fi brosis

S Rowan

907 Journal club: Autoimmune disorders increase the risk of developing pulmonary embolism

H Umpleby

924 Journal club: Salbutamol infusion worsens outcomes in ARDS

O Okafor

930 Journal club: High doses of vitamin D may reduce exacerbations of chronic obstructive pulmonary disease

E Heiden

Contents Volume 67 Issue 10 | THORAX October 2012

Receive regular table of contents by email.Register using this QR code.

thoraxjnl_67_10_TOC.indd 2thoraxjnl_67_10_TOC.indd 2 05/09/12 5:12 PM05/09/12 5:12 PM

Subscription Information

Thorax is published monthly (subscribers receive all supplements)

Institutional Rates 2012Print£568; US$1,108; €767

OnlineSite licences are priced on FTE basis and allow access by the whole institution. Print is available at deeply discounted rates for online subscribers; details available online at http://group.bmj.com/group/subs-sales/subscriptions or contact the Subscription Manager in the UK (see above right).

Personal Rates 2012Print (includes online access at no additional cost)£237; US$463; €320

Online only£128; US$250; €173

ISSN 0040-6376 (print)

ISSN 1468-3296 (online)

Personal print or online only and institutional print subscriptions may be

purchased online at http://group.bmj.com/group/subs-sales/subscriptions

(payment by (MasterCard/Visa only).

Residents of some EC countries must pay VAT; for details call us or visit

http://group.bmj.com/group/subs-sales/subscriptions/subs-vat

Contact DetailsEditorial Offi ceBMJ Publishing Group Ltd, BMA House, Tavistock

Square London, WC1H 9JR, UK


PermissionsSee http://journals.bmj.com/misc/permissions.dtl

Supplement EnquiriesT: +44 (0)20 7383 6057F: +44 (0)20 7554 6795E: [email protected]

For ALL subscription enquiries and ordersT: +44 (0)20 7383 6270F: +44 (0)20 7383 6402E: [email protected]://group.bmj.com/group/customerservice/journalhelp/journalhelp

AdvertisingT: +44 (0)20 7383 6386F: +44 (0)20 7383 6556E: [email protected]://group.bmj.com/group/advertising

Author ReprintsReprints Administrator


Commercial Reprints (except USA & Canada)

Nadia Gurney-Randall

T: +44 (0)20 8445 5825M: +44 (0)7866 262344F: +44 (0)20 8445 5870E: [email protected]

Commercial Reprints (USA & Canada)

Marsha Fogler

T: +1 800 482 1450 (toll free in the USA)

T: +1 856 489 4446 (outside the USA)

F: +1 856 489 4449E: [email protected]

British Thoracic Society17 Doughty Street

London WC1N 2PL, UK

T: +44 (0)20 7831 8778F: +44 (0)20 7831 8766E: [email protected]://www.brit-thoracic.org.uk/index.html

Aims and Scope: Thorax enjoys an enviable and longstanding reputation for publishing clinical and experimental research articles covering many disciplines, including pathology, immunology and surgery

EditorsA Bush (UK)

I Pavord (UK)

Deputy EditorsP Cullinan (UK)

C Lloyd (UK)

Associate EditorsR Beasley (New Zealand)

J Brown (UK)

R Buhl (Germany)

JC Celedón (USA)

A Custovic (UK)

A Fisher (UK)

P Gibson (Australia)

J Grigg (UK)

D Halpin (UK)

A Jones (UK)

E Lim (UK)

N Maskell (UK)

JL Pepin (France)

T Sethi (UK)

M Steiner (UK)

D Thickett (UK)

A Torres (Spain)

Z Udwadia (India)

H Zar (South Africa)

Statistical EditorsJ Gibson (UK)

T McKeever (UK)

StatisticianL Szatkowshi (UK)

Journal club EditorJ Quint (UK)

AN INTERNATIONAL JOURNAL OF

RESPIRATORY MEDICINE

ThoraxJournal of the British Thoracic Society

Editorial BoardE Abraham (USA)

A Agusti (Spain)

H Aranibar (Chile)

E Baraldi (Italy)

E Bateman (South Africa)

J Bradley (UK)

J Britton (UK)

J de Jongste (Netherlands)

O E Fox (Germany)

J Fahy (USA)

J Gauldie (Canada)

M Han (USA)

J Henderson (UK)

F Holguin (USA)

I Janahi (Qatar)

R Kalhan (USA)

A Knox (UK)

C Kuehni (Switzerland)

G Lack (UK)

F Maltais (Canada)

F Martinez (USA)

A Morice (UK)

P O’Byrne (Canada)

M Peters-Golden (USA)

C Pison (France)

E Pizzichini (Canada)

A Quittner (USA)

F Ratjen (Canada)

J Scullion (UK)

J Simpson (UK)

R Stein (Brazil)

R Taylor (New Zealand)

M Thomas (UK)

T Troosters (Belgium)

C Wainwright (Australia)

D Warburton (USA)

A Woodcock (UK)

President, British ThoracicSocietyA Greening

Production EditorAimee Knight

Journal ManagerBryony Lovelock

PublisherAllison Lang

Guidelines for Authors and ReviewersFull instructions are available online

at http://thorax.bmj.com/ifora.

Articles must be submitted

electronically http://submit-thorax.

bmj.com. Authors retain copyright

but are required to grant Thorax an

exclusive licence to publish http://

thorax.bmj.com/ifora/licence.dtl

Impact Factor: 6.84

thoraxjnl_67_10_EB.indd 1thoraxjnl_67_10_EB.indd 1 05/09/12 5:01 PM05/09/12 5:01 PM

Highlights from this issue

doi:10.1136/thoraxjnl-2012-202596 Andrew Bush, Ian Pavord, Editors

Cystic fibrosis (CF): the new TB?Not because it is anything other than avery minor player in the men of death com-pared with that great global scourge, TB,but because CF is following TB into theepicentre of evidence based medicine. ThisCF themed issue of Thorax, coinciding withthe US Cystic Fibrosis Foundation annualmeeting, celebrates the achievements of theCF community in general, and theFoundation in particular, in driving througha series of focussed randomised controlledtrials taking medications forward from the‘wouldn’t it be good if ’ stage to FDA licens-ing. These are summarised in the reviewby Rowe et al (see page 882) and visuallyon the front cover of the journal. Thereview summarises the few failures as wellas the many triumphs, and highlightslessons learned by the Foundation. Weinclude a further trial, comparing two dif-ferent, relatively short course (28 days)Pseudomonas aeruginosa eradication regimeswhich showed equivalence of inhaled colis-tin and tobramycin when combined withoral ciprofloxacin (see page 853). CF is abeacon for other diseases, and an inspir-ation for other patient groups as to whatcan be achieved. These trials worthy suc-cessors to the MRC trials on TB treatmentfromway back when.

Wrong from the startNot another medical cri du coeur abouthealth service reform, but CF lungdisease in babies diagnosed by newbornscreening (NBS). There is conflicting evi-dence about whether NBS CF babieshave normal or abnormal lung functionvery early on, and then subsequentlydeteriorate. The issue of lung functionearly on is definitively nailed by Hoo et al(see page 874) in this issue of thejournal. They showed that, comparedwith contemporaneous controls, morethan a third of NBS CF infants age

3 months had abnormal lung function,namely airway obstruction using theraised volume squeeze technique, hyper-inflation, or distal gas mixing (lung clear-ance index). What is not clear is whetherthese changes are due to early postnatalinfection and inflammation, or, as the CFpig studies might suggest (Am J Respir CritCare Med 2010;182:1251–61), a structural,antenatally-mediated effect of CFTR dys-function. Either way, these data challengethe CF community to address the issues ofearly NBS CF lung disease, and find bio-markers for intervention trials in veryyoung infants, as highlighted in the accom-panying editorial by Cunningham (seepage 849). What you can’t see willundoubtedly come back to bite you.

Two’s company, three’sa crowd, and hundredsare baffling!Life used to be so simple—the normallower airway is sterile, and CF is paucimi-crobial. It is now clear from sophisticatedmolecular microbiology that the normallower airway is teeming with micro-organisms, that this flora is an essentialpart of modulating immune maturation,and that the more diverse the species,the better for the child. Stressmann et alreport on the longitudinal bacterial florain CF (see page 867), and showed thatairway flora were diverse and tended tobe stable over the yearlong study period;interestingly, individuals with very differ-ent clinical courses seemed to have thesame microbiological profiles (so what isdriving their clinical course?); that anti-biotics transiently perturbed the commu-nity, but for whatever reasons, theprevious community usually soonreformed; and that lots of differentorganisms were a good thing. In his edi-torial (see page 851), LiPuma discusses

the need to move from single organismthinking to the concept of a ‘pathogenicunit’, and, reassuringly for those of uswho are struggling with the new micro-biology, frankly sets out the large areas ofignorance we all share. Rightly he cau-tions against therapeutic nihilism or par-alysis in the face of ignorance; currentantibiotic strategies, albeit shooting inthe dark, have been associated withmarked improvements in prognosis.However, these new data will likelyprovoke a radical re-think, and the needfor new intervention trials, as we (hope-fully!) increasingly understand howfavourably to modulate the microbio-logical soups that are bubbling away inevery human airway.

‘Why, sometimes I’ve believedas many as six impossiblethings before breakfast’The White Queen could accomplish thisfeat with practice, and to our politicianssix is mere child’s play, but in our case-based discussion doing a mere twoimpossible (or at least mutually contra-dictory) things was quite a challenge.So what does the CT show, why is themanagement problematic, and what’s itgot to do with CF? Turn to page 932 tofind out.

Thorax October 2012 Vol 67 No 10

Airwaves

What you don’t know can hurtyou; early asymptomatic lungdisease in cystic fibrosisSteve Cunningham

Clinicians caring for children with cysticfibrosis (CF) should now take note; despiteyour best clinical care, lung disease in CFinfants develops surreptitiously and ill-defined by early symptoms. By the timelung disease has symptomatically declareditself, it’s probably structurally establishedand too late to reverse. What you don’tknow can hurt you.

Hoo and colleagues in this issue provideevidence of early lung disease in patientswith CF diagnosed by newborn screening.1

At a mean 3 months of age, 34% of CFpatients had abnormalities of pulmonaryfunction (including lung clearance index(LCI) and/or forced expiratory volume in0.5 second (FEV0.5)) when compared withhealthy controls. The presence of symp-toms, sometimes aggressively treated, didnot reliably identify those with abnormalpulmonary function. This study, bythe London CF Collaborative, usedcontemporaneous healthy controls andadhered to treatment protocols from diag-nosis. The research examines with greatclarity the vital question of how early lungdisease starts in those born with CF. Thechildren studied were provided with everyopportunity for optimal health, by the useof newborn screening, regular chest physio-therapy and prophylactic oral flucloxacillin.Disappointingly, our current best care isnot good enough to prevent lung disease.Of even greater disappointment is thatthose with lung disease do not stand outfrom the crowd.

This is not isolated data. Studies overthe past decade have shown that lungdisease is established very early in lifewith bronchiectasis and ventilationinhomogeneity preceding both symp-toms and a notable decline in FEV1.Children at 12 years of age on theUS CFF Registry (http://www.cff.org),maintain mean FEV1% predicted atc95%, yet at the same age bronchiec-tasis will be established in 59%.2

Bronchiectasis is present, and persists, in33% of children between 2 and 3 yearsof age.3 In 17-month-old CF toddlers,airways have thicker walls and smallerlumens when compared with healthycontrols.4 Even as young as 3 months ofage, the trailblazing Australian AREST CFstudy of newborn-screened infants hasdemonstrated a degree of chest CT abnor-mality in 81%.5 The physiological chron-ology has been demonstrated too, withmore sensitive measures of lung functionable to identify abnormality at youngerages. Measurement of ventilationinhomogeneity by multiple breathwashout (MBW) LCI enables a sensitiveassessment of pulmonary function acrossage ranges. MBW LCI is above thehealthy control normal range in 95% ofschool age CF children,6 73% of preschoolchildren,7 32% of toddlers at 18 monthsof age8 and 21% of infants at 3 monthsof age.1 In infants and children, such pro-gressive anatomical and physiologicalchange is commonly disassociated fromsymptoms,1 5 8 despite the findings thatLCI is more abnormal in those with evi-dence of lower respiratory tractinfection.6 8

Bronchiectasis, destructive and sub-stantially irreversible, is considered ofmost concern in CF, defining the finalcommon pathway of lung destruction torespiratory failure. Limiting the develop-ment of bronchiectasis is considered keyto reducing CF morbidity; prevention ofbronchiectasis is the primary outcome inthe current Australian study ofAzithromycin in CF infants (COMBATCF: clinical trials.gov NCT01270074).Averting bronchiectasis is understood,but what about ventilation inhomogen-eity? Are early abnormalities of ventila-tion inhomogeneity fully reversible, or ifunchecked, how do they evolve intolonger-term structural change with asso-ciated symptoms?

Ventilation inhomogeneity, particularlyin the young, carries a clinical sense thatif only we worked harder with chestphysiotherapy and mucolytics we may beable to reverse such change. Possibly.

In children under 5 years treated withnebulised hypertonic saline as part of theISIS trial, while the primary outcome forthe trial was negative, measures of forcedexpiratory flow demonstrated significantimprovement in the subset of patientstested, suggesting that more sensitivemeasures may be able to demonstrateeffectiveness of therapies when studyingyounger children.9

The association between changes in ven-tilation inhomogeneity and structuralchange on imaging also needs more investi-gation. Inevitably CT Chest, readily avail-able, will be the predominant anatomicalcomparator in such studies, but hyperpo-larised helium MRI (HeMRI) may alsooffer useful insight, as it is able to providesynchronous imaging and quantifiablemeasure of focal ventilation inhomo-geneity. In a recent interventional studyof Ivacaftor, focal areas of ventilationinhomogeneity (assessed by HeMRI) wereimproved by therapeutic intervention, butthe same areas ‘relapsed’ once the interven-tion was terminated.10 If focal areas of ven-tilation inhomogeneity become chronicallysusceptible to injury, then early interven-tion trials to prevent such change may becritical to avert early CF lung disease. As atechnology, HeMRI has constrained appli-cation in the clinical context, so the abilityof MBW LCI to sensitively and specificallyidentify ventilation defects detected byHeMRI warrants further research.

To prevent early CF lung disease, bron-chiectasis and ventilation inhomogeneity,as surrogate biomarkers should be madeprimary outcomes in clinical researchstudies. To become primary outcomes, sur-rogate biomarkers require to be approvedby regulatory authorities and adopted bypharmaceutical companies for phase IIIclinical trials. The sensitivity of CT chestand ventilation inhomogeneity to earlylung disease in CF has been recognised inrecent US NHLBI workshop11 andEuropean Respiratory Society ResearchSeminar (Rotterdam, March 2012). BothCt and MBW LCI were identified as criticalareas for further development and stand-ardisation, to enable their use as primaryoutcomes in multicentre interventionalstudies. The CF research community isgrowing in confidence to challenge andsupport both regulators and pharmaceut-ical companies to make more effectiveassessments of potential therapies usingthese novel surrogate biomarkers. This isparticularly pressing for infants and pre-school children where current regulatoryrequirements may be considered insuffi-ciently demanding of patient benefit. Inresponse, regulators will request that

Correspondence to Dr Steve Cunningham,Department of Respiratory and Sleep Medicine, RoyalHospital for Sick Children, Sciennes Road, EdinburghEH9 1LF, UK; [email protected]

Thorax October 2012 Vol 67 No 10 849

Editorial

http://www.cff.org

http://www.cff.org

proposed novel surrogate biomarkers areexposed to appropriate levels of criticalscrutiny: demonstrated to be safe, quantifi-able and reproducible, sensitive to mean-ingful changes in patient health, andreflect relevant clinical patient benefit.

Several CF interventional clinical studieshave now reported CT chest and LCI astrial endpoints, supporting the case thatthese novel surrogate biomarkers can besensitive and meaningful measures ofpatient health. Treatment of a respiratoryexacerbation in CF children at mean age of3 years with a course of intravenous anti-biotics and intensive chest physiotherapyis associated with an improvement inchest CTscores, particularly bronchial dila-tation/bronchiectasis.12 Further evidencethat novel pulmonary function measures(LCI) may provide a more sensitive signalof therapeutic benefit has recently beenreported in three studies where patientsreceiving the intervention had normalrange spirometry. Statistically significantimprovements in LCI over placebo wasdemonstrated for hypertonic saline,13

dornase α14 and ivacaftor in G551Dpatients.15 Systematic amalgamation ofcurrent evidence, supported by unpub-lished standardisation data, should enablethese biomarkers to enter routine use in CFclinical trials. The entrance will be timely;multiple novel agents are in Phase II andIII trials, and could benefit from more sen-sitive markers of effect for pharmacologicalintervention of cystic fibrosis transmem-brane regulator (CFTR) malfunction (UKCF Gene Therapy Multidose Trial, Vertex

Pharmaceutical interventions for Class IIand IV mutations, and Class III mutationsin young children, PTC Therapeutics forClass I mutations).

Clinicians increasingly understandhow much they must strive to delivertheir patients to the point of futurepharmacogenetic correction with no, orlimited, lung damage. The report byLum and colleagues adds clinical pressureat an earlier age. But, how can we dothis, if we can’t see it, can’t hear it andcan’t routinely measure it? In this age ofenlightenment to the dangers of earlyCF lung disease, early intervention trialswith appropriately approved biomarkershold the key.

Competing interests None.

Provenance and peer review Commissioned;internally peer reviewed.

Thorax 2012;67:849–850.doi:10.1136/thoraxjnl-2012-202465

REFERENCES1. Hoo A-F, Thia LP, Nguyen TTD, et al. Lung function is

abnormal in 3-month-old infants with cystic fibrosisdiagnosed by newborn screening. Thorax2012;67:874–81.

2. Gustafsson PM, De Jong PA, Tiddens HA, et al.Multiple-breath inert gas washout and spirometryversus structural lung disease in cystic fibrosis.Thorax 2008;63:129–34.

3. Mott LS, Park J, Murray CP, et al. Progression ofearly structural lung disease in young children withcystic fibrosis assessed using CT. Thorax2012;67:509–16.

4. Martinez TM, Llapur CJ, Williams TH, et al.High-resolution computed tomography imaging ofairway disease in infants with cystic fibrosis. Am JRespir Crit Care Med 2005;172:1133–8.

5. Sly PD, Brennan S, Gangell C, et al. Lung disease atdiagnosis in infants with cystic fibrosis detected bynewborn screening. Am J Respir Crit Care Med2009;180:146–52.

6. Aurora P, Gustafsson P, Bush A, et al. Multiplebreath inert gas washout as a measure of ventilationdistribution in children with cystic fibrosis. Thorax2004;59:1068–73.

7. Aurora P, Bush A, Gustafsson P, et al.Multiple-breath washout as a marker of lung diseasein preschool children with cystic fibrosis. Am JRespir Crit Care Med 2005;171:249–56.

8. Belessis Y, Dixon B, Hawkins G, et al. Early cysticfibrosis lung disease detected by bronchoalveolarlavage and lung clearance index. Am J Respir CritCare Med 2012;185:862–73.

9. Rosenfeld M, Ratjen F, Brumback L, et al. Inhaledhypertonic saline in infants and children younger than6 years with cystic fibrosis: the ISIS randomizedcontrolled trialInhaled hypertonic saline in childrenWith CF. JAMA 2012:1–9.

10. Altes T, Johnson MA, Miller GW, et al.Hyperpolarized sas MRI of ivacaftor therapy insubjects with cystic fibrosis who have theG551D-CFTR mutation. J Cyst Fibros 2012;11(Supp 1):S67.

11. Ramsey BW, Banks-Schlegel S, Accurso FJ, et al.Future directions in early cystic fibrosis lung diseaseresearch: an NHLBI workshop report. Am J RespirCrit Care Med 2012;185:887–92.

12. Davis SD, Fordham LA, Brody AS, et al. Computedtomography reflects lower airway inflammation andtracks changes in early cystic fibrosis. Am J RespirCrit Care Med 2007;175:943–50.

13. Amin R, Subbarao P, Jabar A, et al. Hypertonicsaline improves the LCI in paediatric patients with CFwith normal lung function. Thorax 2010;65:379–83.

14. Amin R, Subbarao P, Lou W, et al. The effect ofdornase alfa on ventilation inhomogeneity inpatients with cystic fibrosis. Eur Respir J2011;37:806–12.

15. Davies JC, Sheridan H, Lee P-S, et al. Effect ofivacaftor on lung function in subjects with CF who havethe G551D-CFTR mutation and mild lung disease: acomparison of lung clearance index (LCI) vs.spirometry. J Cyst Fibros 2012;11(Supp 1):S15.

850 Thorax October 2012 Vol 67 No 10

Editorial

The new microbiology of cysticfibrosis: it takes a communityJohn LiPuma

For the past 40 years, the approach tostudying infections in the airways ofpersons with cystic fibrosis (CF) haslargely parallelled that taken in the studyof other human infectious diseases. Amicroorganism (the causative agent)recovered in culture from an infected siteis studied in isolation using a variety of invitro and in vivo models intended toapproximate some facet of the humaninfection. Although this strategy hasyielded a wealth of information regardingmicrobial virulence factors and pathogenicmechanisms for many human infections,its limitations, when applied to thechronic, polymicrobial infections thattypify CF, are becoming increasinglyobvious. We now appreciate that respira-tory tract infection in CF most ofteninvolves diverse communities of opportu-nistic bacterial species that are welladapted to the peculiarities of this niche.We have, furthermore, come to under-stand that species within this communityare not merely living unaffected by theirmicrobial neighbours, but rather, areactively engaged with each other. And weare steadily decoding the rules andmechanisms that govern this microbialconcert.

Given this expanding understanding ofinfection in CF, it seems reasonable to shiftour attention towards a conceptual frame-work that considers the airway microbialcommunity as the ‘pathogenic unit.’Adoption of this framework leads to severalrelevant but unanswered questions. Howare these communities structured? Arethey relatively static or dynamic? Arechanges in the community stochastic orcanonical? What drives changes incommunity structure, and how do thesechanges impact community function andhost response? Are there communities thatare more or less associated with progressionof lung disease? More importantly, to whatextent can different communities actuallyeffect the progression of lung disease?

And, in the absence of our ability to ‘ster-ilize’ the lower airways (ie, if indeed, either‘natural’ or healthy), can we envisagemanipulation of communities to favourthose that may be kinder to their humanhost?The work presented by Stressmann

and colleagues1 adds another piece tothis puzzle. This group of investigatorshas been at the forefront of culture-independent bacterial communityprofiling for the past decade, having beenone of the first to describe the complexityof the bacterial community in CFairways,2 and having made severalimportant observations in this area since.This team, led by Professor KennethBruce, now describes the characterisationof bacterial communities in serial sputumsamples obtained from 14 CF patientsduring the course of 1 year. The primaryquestion being addressed pertained to thestability of communities during thisperiod, with particular attention tocommunity response to antibiotictherapy. The authors conclude that whilenot true in all cases, the microbialcommunities that chronically infect theairways of CF patients vary little overa year, despite antibiotic perturbation.A closer look at the data presented in

this report provides further insight. Thepatients included in the study variedwidely with respect to their degree oflung disease; the mean per cent predictedforced expiratory volume in one second(%FEV1) for each patient ranged from16.5% to 84% during the study period.Predictably, patients with more advancedlung disease experienced, on average,more exacerbations, and received morecourses of antibiotics than patients withless lung disease (based on mean %FEV1).Intuitively, one might have expected thatthe clinical instability of the formergroup of patients would have beenreflected in less stable bacterial commu-nities in serial sputum samples. It is easyto imagine that exacerbation of respira-tory symptoms might result from achange in the bacterial community or,conversely, that symptomatic changesin the host’s clinical (and presumably,

inflammatory) state might perturb anotherwise stable resident airwaycommunity.However, the opposite effect was

observed. Although the degree ofcommunity stabilitydmeasured as thedissimilarity in community structuresbetween sequential pairs of samples fromindividual patientsdvaried considerably,more change in community structure wasobserved, on average, in patientswith greater lung function and fewerexacerbations. Similar results wererecently reported by Zhao et al, whoexamined bacterial communities in serialrespiratory samples obtained over thecourse of 8e9 years from six adult CFpatients.3

Further consideration of these unex-pected results begins to draw a clearerpicture of the ecology of airway infectionin the context of advancing lung diseasein CF. Similar to previous studies,4e6

Stressmann et al found an inverse rela-tionship between airway bacterialcommunity diversity and lung diseaseseverity. That is, as %FEV1 decreases(with progressing lung disease), so toodoes community diversity. It is not clearwhether this decrease in diversity isa driver of lung disease or, conversely,results, in part, from the increasinglyintensive antibiotic therapy given topatients as lung disease advances, assuggested by Zhao et al3 Both studiesfound that the decrease in diversity inpatients with advanced lung disease wasa reflection of a decrease in the number ofspecies in the community (referred to ascommunity ‘richness’) and an increase inthe relative abundance of Pseudomonasaeruginosa (or, in ecological parlance,a decrease in the ‘evenness’ of speciesabundance in the community). Thus, wesee that as lung disease progressesdasindicated by increasing frequency ofexacerbations, decreasing %FEV1, and,presumably, by increasing antibioticusedbacterial community richnessdecreases, as does community evenness,which is marked by an increase in therelative abundance of P. aeruginosa.The findings reported in these studies

describe a model in which airway bacterialcommunities at earlier stages of lung diseaseare diverse, having the ‘bandwidth’ toexperience changes in their structure inrelation to the host’s clinical state andtherapy. As lung disease progresses,community richness and evenness decrease,and this more constrained community isrelatively more stable (ie, less able to bemoved by perturbations imposed by the

Correspondence to Dr John LiPuma, University ofMichigan, 1150 W. Medical Center Dr, 8323 MSRB III,SPC 5646, Ann Arbor, MI 48109, USA;[email protected]

Editorial


host or, presumably, by antimicrobialtherapy provided to the host). Additionallongitudinal studies are underway to deter-mine if this model holds.

But what does all this mean in terms ofadvancing our understanding and/orimproving our management of CF airwaysinfection? Are diverse, less stable micro-bial communities ‘better ’ for the host?Does the presence of less diverse, relativelyfixed communities signify a ‘terminal’microbial state that presages end-stagelung disease? Do these findings suggestthat we ought to modify the way wemanage CF infection in the earlier or latestages of lung disease?

At this point in time, we simply do notknow the answers to these questions.What is clear, however, is that we oughtnot to readily retreat from the currentantimicrobial strategies that have beenassociated with significant improvementsin life expectancy in CF during the pasttwo decades.7 But, equally important, we

now have an emerging conceptual frame-work, the methodological tools, anda budding foundation of knowledge thatwill enable us to focus on the structureand function of the airway microbialcommunity in CF, rather than (or morelikely, in addition to) on select members ofthat community. This new microbiologywill be a critical component of a systemsbiology approachdan approach thatconsiders the complex interplay betweencommunity microbes, and between themicrobial community and the hostdthatis needed to take us to the next level in ourunderstanding of airways infection in CF.



doi:10.1136/thoraxjnl-2012-202018

REFERENCES1. Stressmann FA, Rogers GB, van der Gast CJ, et al.

Long-term cultivation-independent microbial diversity

analysis demonstrates that bacterial communitiesinfecting the adult cystic fibrosis lung show stabilityand resilience. Thorax

2. Rogers GB, Hart CA, Mason JR, et al. Bacterialdiversity in cases of lung infection in cystic fibrosispatients: 16S ribosomal DNA (rDNA) lengthheterogeneity PCR and 16S rDNA terminal restrictionfragment length polymorphism profiling. J ClinMicrobiol 2003;41:3548e58.

3. Zhao J, Schloss PD, Kalikin LM, et al. Decade-longbacterial community dynamics in cystic fibrosisairways. Proc Nat Acad Sci U S A 2012;109:5809e14.

4. Cox MJ, Allgaier M, Taylor B, et al. Airway microbiotaand pathogen abundance in age-stratified cysticfibrosis patients. PLoS One 2010;5:e11044.

5. Klepac-Ceraj V, Lemon KP, Martin TR, et al.Relationship between cystic fibrosis respiratory tractbacterial communities and age, genotype, antibioticsand Pseudomonas aeruginosa. Environ Microbiol2010;12:1293e303.

6. van der Gast CJ, Walker AW, Stressman FA, et al.Partitioning core and satellite taxa from within cysticfibrosis lung bacterial communities. ISME J2010;5:780e91.

7. Van Devanter DR, LiPuma JJ. Microbial diversity inthe cystic fibrosis airways: where is thy sting? FutureMicrobiol 2012;7:1e3.

PAGE fraction trail=1.5

Editorial


Thorax 2012;67:851–852.

2012; :874e81.67

ORIGINAL ARTICLE

Early antibiotic treatment for Pseudomonasaeruginosa eradication in patients with cystic fibrosis:a randomised multicentre study comparing twodifferent protocols

Giovanni Taccetti,1 Elisa Bianchini,2 Lisa Cariani,3 Roberto Buzzetti,4 Diana Costantini,5

Francesca Trevisan,1 Lucia Zavataro,1 Silvia Campana,1 on behalf of the Italian Groupfor P aeruginosa eradication in cystic fibrosis*

ABSTRACTBackground Pseudomonas aeruginosa chronicpulmonary infection is an unfavourable event in cysticfibrosis. Bacterial clearance is possible with an earlyantibiotic treatment upon pathogen isolation. Currently,no best practice exists for early treatment. The efficacyof two different regimens against initial P aeruginosainfection was assessed.Methods In a randomised, open-label, parallel-groupstudy involving 13 centres, the superiority of inhaledtobramycin/oral ciprofloxacin compared with inhaledcolistin/oral ciprofloxacin (reference treatment) over28 days was evaluated. Patients were eligible if theywere older than 1 year with first or new P aeruginosaisolation. Treatments were assigned equally bycentralised balanced randomisation, stratified by age andforced expiratory volume in 1 s values. The participantsand those giving the intervention were not masked toarm assignments. The primary endpoint wasP aeruginosa eradication, defined as three successivenegative cultures in 6 months. Analysis was by intentionto treat. This trial was registered with EudraCT, number2008-006502-42.Results 105 patients were assigned to inhaled colistin/oral ciprofloxacin (arm A) and 118 to inhaled tobramycin/oral ciprofloxacin (arm B). All patients were analysed.P aeruginosa was eradicated in 66 (62.8%) patients inarm A and in 77 (65.2%) in arm B (OR 0.90, 95% CI 0.52to 1.55, p¼0.81). Following treatment, an increase inStenotrophomonas maltophilia was noted (OR 3.97, 95%CI 2.27 to 6.94, p¼0.001) with no differences betweenthe two arms (OR 0.89, 95% CI 0.44 to 1.78, p¼0.88).Conclusions No superiority of treatment under studywas demonstrated in comparison to the referencetreatment. Early eradication treatment was associatedwith an increase in S maltophilia.

INTRODUCTIONChronic pulmonary infection by Pseudomonasaeruginosa is an unfavourable event in patients withcystic fibrosis (CF) associated with an increase inmorbidity and mortality. Once chronic infection isestablished, P aeruginosa is virtually impossible toeradicate.1e4 However, various studies have shownthat bacterial clearance from the respiratory tract ispossible with early antibiotic treatment upon

isolation of this pathogen,4e16 so aggressive anti-biotic treatment of initial P aeruginosa infection isnow recommended.2 Several strategies exist to treatearly P aeruginosa infection using inhaled colistinwith oral ciprofloxacin or inhaled tobramycin orintravenous antibiotics.1e17 To date, there is noclearly identified best practice for early treatment ofP aeruginosa infection. Few studies have comparedthe efficacy of different treatments.1 6 7 10 14e16 TheELITE trial compared the efficacy of a 28-dayregimen with a 56-day regimen of inhaled tobra-mycin and showed no differences.10 The EPICstudy investigated the efficacy of four anti-pseu-domonal regimens in children aged 1e12 years withrecently acquired infection. No difference wasfound between cycled and culture-based therapies.Adding oral ciprofloxacin produced no benefits.17

The primary aim of this multicentre, randomisedstudy was to evaluate the efficacy of two differenteradication treatments (oral ciprofloxacin andinhaled tobramycin compared with oral cipro-floxacin and inhaled colistin) over 28 days againstinitial P aeruginosa infection. Eradication is definedas three negative cultures within a 6-monthperiod.3

Key messages

What is the key question?< To clarify the efficacy of two different eradica-

tion treatments, oral ciprofloxacin and inhaledtobramycin (test treatment), compared with oralciprofloxacin and inhaled colistin (referencetreatment) over 28 days against initial Pseudo-monas aeruginosa infection in cystic fibrosis.

What is the bottom line?< To date, there is no clearly identified best

practice for early treatment of P aeruginosainfection.

Why read on?< To verify the efficacy of test treatment

compared with reference treatment in eradi-cating P aeruginosa, increasing forced expiratoryvolume in 1 s values and modifying airways flora.

1Department of Sciences forWoman and Child’s Health,University of Florence, CysticFibrosis Centre, Anna MeyerChildren’s University Hospital,Florence, Italy2Centro CoordinamentoSperimentazioni Cliniche-ITT,AOU Careggi-Pad. San Damiano,Florence, Italy3Laboratory for Cystic FibrosisMicrobiology, FondazioneIRCCS, Ca’ GrandadOspedaleMaggiore Policlinico, Milano,Italy4Clinical Epidemiologist,Bergamo, Italy5Cystic Fibrosis Centre,Fondazione IRCCS, Ca’GrandadOspedale MaggiorePoliclinico, Milano, Italy

Correspondence toDr Giovanni Taccetti, CysticFibrosis Centre, Anna MeyerChildren’s University Hospital,Viale Pieraccini 24, Firenze 50139,Italy; [email protected]

For author footnote see end ofthe article.

Received 22 July 2011Accepted 26 January 2012

Cystic fibrosis

Thorax 2012;67:853–859. doi:10.1136/thoraxjnl-2011-200832 853

Published Online First29 February 2012

The secondary aims were to verify if the treatment isequally effective in patients with first-ever P aeruginosa infec-tion and previously treated patients; to evaluate FEV1 atbaseline and following treatment; to determine if differencesexist between the two types of treatment referring to the Paeruginosa-free period; and to determine whether the eradica-tion treatment is associated with the emergence of Burkholderiacepacia complex, other non-fermenter Gram negatives andAspergillus spp.

METHODSThis was a multicentre, open-label, parallel-group study toevaluate the superiority of inhaled tobramycin and oral cipro-floxacin compared with inhaled colistin and oral ciprofloxacinconsidered as the reference treatment for eradication.3 4 6 12 18

Patients in regular clinical and microbiological follow-upwere considered eligible if older than 1 year with first or newP aeruginosa infection. New infection was defined as P aeruginosaisolation following bacterial clearance documented by threenegative cultures in the previous 6 months.3

CF diagnosis was based on clinical features of the disease andconcentration of chloride in sweat >60 mmol/litre.1e3

Exclusion criteria were concomitant infection by other non-fermenter Gram negatives, respiratory exacerbations, Paeruginosaresistance to any of the antibiotics used in the study, concomi-tant use of quinolones, chronic macrolide use, abnormal kidneyor liver function tests.

Thirteen CF Centres participated in this trial. The studyreceived ethical approval from the Meyer Hospital EthicsCommittee and was registered with EudraCT, number 2008-006502-42.

RandomisationA balanced randomisation sequence was created using statisticalsoftware to assign treatments within permuted blocks of size10. Randomisation assignment was organised by email. Patientswere allocated 1:1 and distributed into two groups, stratifiedaccording to age and FEV1 values as an expression of illnessseverity:< age 1e5 years< age >5e12 with FEV1 <70%< age >5e12 with FEV1 >70%< age >12 with FEV1 <70%< age >12 with FEV1 >70%.

The people involved in randomisation and in the treatmentassignments were kept separate.

ProceduresMicrobiological analyses were performed following publishedliterature.19 Strain typing in the case of next P aeruginosa isola-tion was performed with repetitive PCR.20 FEV1 values weremeasured according to American Thoracic SocietyeEuropeanRespiratory Society standards.21

Patients received one of the two types of treatments for28 days: twice-daily inhalation of 2 000 000 IU colistin and30 mg/kg/day of oral ciprofloxacin, divided into two doses (armA or active comparator) or twice-daily inhalation of 300 mgtobramycin solution for inhalation and 30 mg/kg/day of oralciprofloxacin divided into two doses (arm B or test treatment).Colistin was reconstituted immediately before use and admin-istered in isotonic solution (two phials of 1 000 000 IU in 2 ml ofwater for injections plus 2 ml of 0.9% sodium chloride).3

Tobramycin solution for inhalation was dispensed as a ready-to-

use solution in either 300 mg/5 ml or 300 mg/4 ml phials. Thistrial was conducted using jet nebulisers.3

Each physician in the respective centre provided the patients’clinical management, according to standards of care.18 Thetreatment was suspended in the case of adverse effects orpulmonary exacerbation.22

Nebulisers and drugs for treatment were covered by theItalian Health Service at no charge to the patients.

OutcomesTreatment efficacy was evaluated according to the UK CF Trustcriteria.3 Results of cultures and clinical records were used toassess secondary aims. P aeruginosa microbiological status wasevaluated using the Leeds criteria23 after a median follow-uptime of 16 months since recruitment.

Statistical analysesThis trial was designed calculating the sample size as a balancebetween statistical considerations and experience regardingepidemiology of first or new P aeruginosa infection.24 We calcu-lated sample size assuming 65% response to early eradicationtreatment lasting 4 weeks in the active control arm. Wehypothesised recruiting 200 patients in a 2-year trial. Thea (type I) error was set at 0.05, the b (type II) error at 70%. A13% greater rate of eradication in arm B in comparison with armA was considered as the level of superiority.In relation to the primary aim of the study (bacterial eradi-

cation), a sequential statistical analysis using O’Brien-Flemingboundaries was used to determine the appropriate rules for trialinterruption.24 Four interim analyses (every 6 months) wereplanned before the start of the trial. The results of the interimanalyses were presented to the Data Safety Monitoring Boardoverseeing the study to consider the option of early termination.The association of patients’ baseline demographical and clinicalcharacteristics on the primary outcome was evaluated usinga test for interaction. For all subgroup analyses, unadjusted ORand relative 95% CI were used as a measure of treatment effect.Data regarding the primary aim was independent, with oneobservation per participant, and evaluated according to inten-tion-to-treat analysis.24

For secondary aims, differences in continuous and categoricalvariables between groups were evaluated using descriptivestatistics, c2 test, and t test. P aeruginosa-free time after treat-ment was calculated according to the KaplaneMeier product-limit method. Comparisons of estimated Paeruginosa-free curveswere performed by means of the log-rank test. HRs andappropriate 95% CIs were also calculated by means of Cox’sproportional hazard model.

RESULTSFigure 1 shows the trial profile. A total of 263 patients wereassessed for eligibility in the period from 1 February 2008 to 31March 2010 and 223 patients were randomisedd105 (47.1%)patients were assigned to arm A (oral ciprofloxacin and inhaledcolistin) as active controls and 118 (52.9%) patients wereassigned to arm B (oral ciprofloxacin and inhaled tobramycin)as test treatment. Two patients in each arm refused the treat-ment after randomisation. Table 1 shows patients’ clinicalcharacteristics.Microbiological cultures were performed on pharyngeal swab

specimens in 132 of 223 patients (59.2%) and sputum in 91 of223 patients (40.8%). At baseline, P aeruginosa phenotype wasnon-mucoid and all strains were susceptible to ciprofloxacin,colistin and tobramycin.

Cystic fibrosis

854 Thorax 2012;67:853–859. doi:10.1136/thoraxjnl-2011-200832

One hundred and twenty-one of 223 patients (54.3%), 57 inarm A and 64 in arm B, had first-ever Paeruginosa infection while102 (45.7%), 48 in arm A and 54 in arm B, had previously had Paeruginosa infection cleared with early eradication treatment. Theage (mean6 SD) of patients with first-ever Paeruginosa infectionwas significantly lower than in patients who had undergoneprevious eradication treatment (7.8967.4 years; 11.3368.5 years;difference: 3.44, 95% CI �5.54 to 1.34, p¼0.001). For the age(mean 6 SD) at first-ever P aeruginosa infection, we observed nostatistically significant difference between females and males(7.1265.7 years; 8.5168.5 years; difference: 1.39, 95% CI �3.31to 0.53, p¼0.15).

One hundred and twenty-eight of 223 patients (57.3%)5 years or older, 60 in arm A and 68 in arm B, were able toperform spirometry.

Thirty-eight of 223 randomised patients (17%) dropped out ofthe study. Causes of dropout are shown in table 2.

Primary endpointP aeruginosa was eradicated in 66 of 105 patients (62.8%) in armA and in 77 of 118 patients (65.2%) in arm B. The proportion ofP aeruginosa eradication between the two groups was notstatistically significant (OR 0.90, 95% CI 0.52 to 1.55, p¼0.81).

The results of interim analyses and the final analysis areshown in figure 2.

Treatment result by age and FEV1 values is shown in table 3.We observed no statistically significant difference in P aerugi-nosa eradication in the strata considered (Pearson c2¼4.45,p¼0.35).

Subgroup analyses are reported in figure 3. In patients withFEV1 > 70%, the proportion of P aeruginosa eradication washigher although not statistically significant in patients in the

test treatment arm in comparison to the active control arm(unadjusted OR 2.14, 95% CI 0.99 to 4.67, p¼0.07).

Secondary endpointsEffectiveness of treatmentAt the end of the follow-up period, having performed culturesand patients’ clinical examination at 2, 4 and 6 months, earlyeradication treatment was found to be effective in 80 of 121patients (66.1%) with first-ever P aeruginosa infection (failure in41 patients) and in 63 of 102 patients (61.7%) who hadsuccessfully undergone early eradication treatment (failure in 39patients). Difference was not statistically significant (OR 1.20,95% CI 0.69 to 2.09, p¼0.59).

FEV1 change following treatmentFollowing treatment, after a mean time of observation of54639 days, the mean FEV1 (6SD) relative change (percentageof predicted) from baseline was 2.15% (68.50) in arm A and4.55% (611.54) in arm B (p¼0.18).

Period in which patients remained P aeruginosa freeAfter 6 months 143 of 223 patients (64.1%) were P aeruginosafree, 66 in arm A and 77 in arm B (HR 0.95, 95% CI 0.58 to 1.56p¼0.85).Table 4 illustrates the percentage of P aeruginosa-free patients

for each of the three cultures following early treatment in botharms of the study in a 6-month range.Molecular analyses were performed on paired P aeruginosa

samples isolated from 47 patients at baseline and within a6-month period following treatment. The same P aeruginosagenotype was found in isolates from 36 patients, a differentgenotype in isolates from 11 patients.

Figure 1 Trial profile.

Cystic fibrosis


Post-trial follow-up: emergence of other non-fermenter Gramnegatives and Aspergillus spp in patients’ airwaysThe post-trial follow-up was completed in 205 of 223 rando-mised patients (91.9%) (missing data: 18 patients, 8 in arm Aand 10 in arm B). After a median observation period of16 months (range 12e28), 135 patients (60.5%), 61 in arm A and74 in arm B, were still P aeruginosa free, 56 (25.1%), 28 in arm Aand 28 in arm B, were intermittently infected and 14 (6.3%), 8 inarm A and 6 in arm B, chronically infected.

We evaluated the isolation frequency of other Gram-negativepathogens and Aspergillus spp during a period of 16 months(median) pre/post treatment. Following treatment, Steno-trophomonas maltophilia was found in 40 patients (17.9%) andAchromobacter xylosoxidans in 14 patients (6.2%). We observed noemergence of B cepacia complex. Table 5 illustrates the increasein isolation frequency of S maltophilia and A xylosoxidans in the

airway cultures of 54 patients in the pre/post observation periodindicated above. While the isolation frequency of A xylosoxidansappeared substantially unvaried, following treatment weobserved an increase in S maltophilia isolations (OR 3.97, 95% CI2.27 to 6.94, p¼0.001). Twenty-five of 80 S maltophilia isolations(31.3%) were observed in arm A and 55 (68.7%) in arm B.Aspergillus spp isolates were found in 20 patients (8%) before

treatment and in 31 (14%) after treatment (OR 1.63, 95% CI0.90 to 2.97, p¼0.13).In the 205 patients completing follow-up there were 52

Aspergillus spp positive cultures of 1025 in the pre-observationperiod and 59 of 1237 in the post-observation period (OR 1.06,95% CI 0.72 to 1.56, p¼0.81).As shown in table 6, no statistically significant differences

were observed in the emergence of non-fermenters and Asper-gillus spp in the two treatment arms (S maltophilia, OR 0.89,95% CI 0.44 to 1.78, p¼0.88; A xylosoxidans, OR 1.52, 95% CI0.51 to 4.57, p¼0.62; Aspergillus spp, OR 0.47, 95% CI 0.21 to1.07, p¼0.10).

DISCUSSIONThe present study is a comparison of two different schedulesof early eradication regimens of inhaled antibiotics associatedwith oral ciprofloxacin in the treatment of initial P aeruginosainfection. No superiority of the treatment under study wasdemonstrated when compared with the reference treatment. Afollow-up conducted using three cultures per patient over6 months showed no differences between the two arms inrelation to the period in which patients remained P aeruginosa

Table 2 Adverse events leading to treatment discontinuation and othercauses of dropout per arm

Arm A, oralciprofloxacin/inhaled colistin

Arm B, oralciprofloxacin/inhaled tobramycin

n (%) n (%)

All adverse events 2 (1.9) 3 (2.5)

Vomiting 1 2

Photosensitivity 1 e

Wheezing e 1

Pulmonary exacerbations during earlyeradication treatment

4 (3.8) 5 (4.2)

Lack of compliance with follow-upprotocol

11 (10.4) 13 (11)

Table 1 Baseline demographic and clinical characteristics of patients

Arm A (active control),oral ciprofloxacin/inhaled colistin(n[105)

Arm B (test treatment),oral ciprofloxacin/inhaled tobramycin(n[118)

Age in years, median (IQR) 7.45 (1e25.5) 7.64 (1e35.2)

Gender, n (%)

Female 53 (50.5) 54 (45.8)

Male 52 (49.5) 64 (54.2)

Cystic fibrosis genotype, n (%)

F508del homozygotes 25 (23.8) 29 (24.5)

F508del compoundheterozygotes

44 (41.9) 53 (44.9)

Other genotype 36 (34.3) 36 (30.6)

Coinfection with other pathogens at baseline, n (%)

Staphylococcus aureus 62 (59) 63 (53.5)

Methicillin-resistant S aureus 7 (6) 2 (1.6)

Haemophilus influenzae 10 (9.5) 12 (10)

Streptococcus pneumoniae 0 2 (1.6)

Pseudomonas aeruginosa infection, n (%)

First-ever infection 57 (54.3) 64 (54.2)

Previously infected 48 (45.7) 54 (45.8)

FEV1 (% of predicted)

Mean (6SD) 93.1 (623.1) 91.8 (619.1)

Patients able to performspirometry, n (%)

60 (57.1) 68 (57.6)

Age and FEV1 group, n (%)

#5 years 36 (34.3) 41 (34.8)

>5e12 years, FEV1<70% 9 (8.6) 11 (9.3)

>5e12 years, FEV1>70% 32 (30.4) 35 (29.7)

>12 years, FEV1<70% 5 (4.8) 5 (4.2)

>12 years, FEV1>70% 23 (21.9) 26 (22)

FEV1, forced expiratory volume in 1 s.

Figure 2 Sequential analysis graph showing results of four interimanalyses and the final analysis.

Table 3 Intention-to-treat analysis by strata

Pseudomonas aeruginosaeradication, n

Treatmentfailure, n

Age 1e5 years 48 29

5e12 years with FEV1 <70% 16 4

5e12 years with FEV1 >70% 45 22

Age >12 years with FEV1 <70% 7 3

Age >12 years with FEV1 >70% 27 22

Total 143 80

Cystic fibrosis


free. Patients previously treated with early eradication treatmenthad the same microbiological outcome compared with patientswith first-ever infection. Early eradication treatment is associ-ated with an increase in S maltophilia in the airways. The highpercentage of patients receiving intervention considering thenumber assessed for eligibility indicates treatment acceptability.The dropout percentage was due more to the lack of compliancewith microbiological follow-up than to the treatment itself.

A large number of P aeruginosa infections occurred in the 1e5age range, as shown by other studies.16

P aeruginosa strains isolated at baseline were susceptible toantibiotics. Only a small percentage of bacterial strains associ-ated with initial infection are resistant to antibiotics.3 25

However, as susceptibility in early isolates cannot be presumed,susceptibility tests should continue to be performed to assist thechoice of treatment.26

We observed no differences in the primary outcome betweenpatients with first-ever P aeruginosa infection and patientspreviously treated. This underlines the importance of payinggreat attention in the initial phases of infection, even in previ-ously treated patients.

The eradication rates in both arms of the study were lowerthan some other published trials. Eradication rates are not easilycomparable as many differences exist in the literature regarding

study design, schedule and length of eradication treatments,follow-up period, methods used to collect airway specimens,patients’ clinical characteristics and definition of eradication.1e17

In our study the analysis was by intention to treat, which mayhave affected the percentage of eradication.24 Moreover, patientsincluded in this trial were treated using a single cycle of anti-biotics. Experiences published in the literature underlinea greater efficacy of oral ciprofloxacin and inhaled colistin whentreatment is extended to 3 months.4 12 27 Recent studiesconducted only with inhaled tobramycin showed no improve-ment in the microbiological outcome when treatment lengthwas extended from 28 to 56 days.10 It is debatable whether suchcontrasting results for the microbiological outcome after treat-ment extension depend on associated use of antibiotics ratherthan using a single drug. The evaluation of treatments beyond28 days is not included in the aims of the present study.Following treatment, we observed an increase in FEV1 values

in both arms. Limited data exist about the efficacy of earlyeradication treatment in preventing lung function decline.4 10 12

A longer follow-up is required to evaluate the effect of thistreatment on lung function. Moreover, considering thatpulmonary function testing cannot usually be performed onchildren under 5 years, FEV1 may not be the ideal marker toevaluate lung function in eradication trials.17

Figure 3 Subgroup analyses ofvariables among patients receiving oralcipofloxacin and inhaled colistin (arm A)compared with those receiving oralciprofloxacin and inhaled tobramycin(arm B). FEV1, forced expiratory volumein 1 s.

Table 4 Treatment failure at first, second and third culture for treatment arm

N

Pseudomonasaeruginosa-free, n

Treatmentfailure(cumulative), n

KaplaneMeiersurvival 95% CI

Arm A (105 patients)

Dropout 4

Missing data 4

Completed follow-up 97

At first culture (60 days) 84 13 0.87 0.78 to 0.92

At second culture (120 days) 69 28 0.71 0.66 to 0.82

At third culture (180 days) 66 31 0.68 0.58 to 0.76

Arm B (118 patients)

Dropout 5

Missing data 5

Completed follow-up 108

At first culture (60 days) 92 16 0.85 0.79 to 091

At second culture (120 days) 80 28 0.74 0.67 to 0.82

At third culture (180 days) 77 31 0.71 0.60 to 0.78

Cystic fibrosis


Our data confirm that next P aeruginosa isolation within6 months from the eradication treatment is mainly due totreatment failure.3 The possibility that after early eradicationtreatment next Paeruginosa infections could be due to different Paeruginosa strains is described in the literature, indicating theefficacy of treatment.8 10 12 28 Considering that molecular typingtechniques are not easily carried out in clinical practice, cultureresults remain the standard for evaluating treatment efficacy.

We observed no substantial differences in the primaryoutcome regarding sex, patient age and stratification by age andFEV1 values. In the subgroup analyses we observed a trend inarm B towards a greater efficacy in P aeruginosa eradication inpatients with FEV1 values >70% of predicted. A higher effec-tiveness of nebulised tobramycin in comparison with colistinhas been described as improving lung function in patients withchronic P aeruginosa infection.29 30 The results of such studiescannot be generalised to the treatment of early phases of Paeruginosa infection.

The clinical meaning of the emergence of other non-fermenterGram negatives has yet to be established. Data regarding Axylosoxidans are limited19 and as far as S maltophilia is concerned,contrasting opinions exist about pathogenicity.31e33 The causesof the increased isolations of non-fermenter Gram negatives arenot clear. It has been hypothesised that selective pressure exertedon bacterial populations by aerosolised aminoglycosides coulddetermine the emergence of S maltophilia, but results fromprevious studies are contrasting. While the ELITE trial showedno S maltophilia increase during the follow-up, the EPIC trialobserved the emergence of the pathogen in up to 20% ofpatients.10 17 Quinolones have also been associated with anincreased risk of S maltophilia.33 We cannot exclude that theincreasing isolation of such pathogens may be caused by cross-infections among patients in CF centres or by contamination ofaerosol devices.34 However, molecular studies to evaluate thepossibility of cross infections among patients and microbiolog-ical surveillance of aerosol devices were not part of our study.

The duration of the follow-up and the lack of internationallyaccepted definitions for the chronicity of such infections makeclinical outcome evaluation difficult.By delaying chronic P aeruginosa infection, early eradication

treatment constitutes a positive result for patients. Currently,we lack direct evidence of the effect of eradication on patients’survival. According to recently published data, even mucoid Paeruginosa strains can be eradicated, but with a lower percentageof success compared with the treatment of non-mucoidphenotype strains.13 35

Further research is necessary to assess the effect of earlyantibiotic treatment on respiratory exacerbations and antibioticuse. Even though a gold standard has yet to be established, datafrom our study indicate that treatment efficacy does not seem todepend on the drugs used, or having previous eradication treat-ment, or the patients’ stratification according to age and FEV1

values.The extent to which treatment timelines can affect the

microbiological outcome needs to be established.

Author footnote*All members mentioned as Italian Group for P aeruginosa eradication in cystic fibrosiscontributed to the interpretation of the data, commented extensively, revised thereport critically, and approved the final version. GT had full access to all the data in thestudy and takes responsibility for the integrity of the data and the accuracy of the dataanalysis. Microbiological analyses were performed in their respective CF Centres by EManso (Ancona), A D’Aprile (Cerignola), T Borio (Gualdo Tadino), S Campana, NRavenni, D Dolce, P Cocchi, G Mergni (Firenze), L Cariani, G Defilippi (Milano), TPensabene (Palermo), E Fiscarelli (RomadOspedale Bambino Gesu). L Cariani and GDefilippi performed molecular analyses on P aeruginosa strains.

AcknowledgementsWe thank the following CF Centres Directors/Medical doctorsfor having made this study possible: R Gagliardini (Ancona); R Padoan (Brescia);L Ratclif (Cerignola); M Ambroni (Cesena); C Braggion (Firenze); L Minicucci(Genova); A Cosimi (Gualdo Tadino); A Negri, E. Micheletti (Livorno); F Forte(Matera); G Magazzu (Messina); C Colombo (Milano); F Pardo, M.Collura (Palermo);V Lucidi, E Fiscarelli (Roma - Ospedale Bambino Gesu). We give grateful thanks toDr Sergio Zuffo and all the physiotherapists of the Cystic Fibrosis Centres who tookpart in the study.

Contributors The trial was designed by GT, LC, RB, DC, FT, LZ and SC. LZ coordinatedthe randomisation unit and kept the randomisation list. The trial was coordinated intheir respective CF Centres by R Gagliardini (Ancona), R Padoan (Brescia), L Ratclif(Cerignola), M Ambroni (Cesena), G Taccetti (Firenze), R Casciaro, L Minicucci(Genova), A Cosimi, T Borio (Gualdo Tadino), A Negri (Livorno), F Forte (Matera), GVieni, C Zinnarello (Messina), D Costantini, L Claut (Milano), E Fiscarelli, V Lucidi(Roma), M Collura (Palermo). All the coordinators contributed to the interpretation ofdata in their centres. GT, EB, LC, RB, DC, FT, LZ and SC wrote the first draft of thepaper, which all authors reviewed. EB coordinated statistical analyses, performedinterim analyses and the final analysis, and wrote the Results section.

Funding Italian Cystic Fibrosis Research Foundation (Grant FFC#17/2007) with thecontribution of ‘Delegazione FFC di Vicenza’.


Patient consent Obtained.

Ethics approval Ethics Committee of Meyer Hospital (Florence, Italy).

Provenance and peer review Not commissioned; externally peer reviewed.

Members of Italian Group for P. aeruginosa eradication in cystic fibrosisRolando Gagliardini, Esther Manso, Ospedali Riuniti di Ancona, Ancona, Italy; RitaPadoan, Elena Soncini, Clinica Pediatrica, Ospedale dei Bambini, AO Spedali CiviliBrescia, Brescia, Italy; Angelica D’Aprile, Luigi Ratclif, Ospedale G Tatarella,Cerignola, Italy; Maura Ambroni, Divisione di Pediatria e Patologia Neonatale, OspedaleBufalini, Cesena, Italy; Rosaria Casciaro, Laura Minicucci, Ospedale G Gaslini DivisionePediatria I Genova, Italy; Tatiana Borio, Angelo Cosimi, Ospedale di Gubbio-GualdoTadino, Gualdo Tadino, Italy; Vanessa Boni, Gemma Braccini, Priscilla Cocchi, DanielaDolce, Filippo Festini, Valeria Galici, Gianfranco Mergni, Novella Ravenni, Sergio Zuffo,Anna Meyer Children’s University Hospital, Department of Sciences for Woman andChild’s Health - University of Florence Florence, Italy; Amalia Negri, Divisione diPediatria, Ospedale di Livorno, Livorno, Italy; Fabio Forte, Divisione di Pediatria eNeonatologia Dipartimento Donna, Maternita e Infanzia, Ospedale di Matera, Matera,Italy; Giuseppe Vieni, Clara Zinnarello, Ospedale “G Martino”, Messina, Italy; Laura

Table 5 Isolation frequency of Stenotrophomonas maltophilia andAchromobacter xylosoxidans in the cultures of 54 patients in a mediantime of observation pre and post-treatment of 16 months

Pretreatment, n Post-treatment, n

Cultures positive for S maltophilia 16 80

Cultures positive for A xylosoxidans 15 19

Total number of cultures positive for non-fermenter Gram negatives

31 99

Total number of cultures negative for non-fermenter Gram negatives

303 381

Table 6 Post-trial follow-up (completed in 205 patients):microbiological status for Stenotrophomonas maltophilia, Achromobacterxylosoxidans and Aspergillus spp in the two arms

Arm A, n Arm B, n Total, n

Patients completing the post-trial follow-up 97 108 205

Missing data on patients 8 10 18

S maltophilia

Culture-positive 18 22 40

Culture-negative 79 86 165

A xylosoxidans



Aspergillus spp



Cystic fibrosis


Claut, Gloria Defilippi, Divisione Pediatria dell’Universita degli Studi di MilanoFondazione IRCCS, Ca’ Granda - Ospedale Maggiore Policlinico, Milano, Italy; ErsiliaFiscarelli, Ospedale Pediatrico Bambino Gesu, Roma, Italy; Mirella Collura, TizianaPensabene, Divisione Pediatria Ospedale dei Bambini "G. Di Cristina", Palermo, Italy.

REFERENCES1. Langton Hewer SC, Smyth AR. Antibiotic strategies for eradicating Pseudomonas

aeruginosa in people with cystic fibrosis. Cochrane Database Syst Rev 2009;(4):CD004197.

2. Doring G, Høiby N; Consensus Study Group. Early intervention and prevention of lungdisease in cystic fibrosis: a European consensus. J Cyst Fibros 2004;3:67e91.

3. The UK Cystic Fibrosis Trust Infection Control Group. Pseudomonas aeruginosaInfection in People with Cystic Fibrosis. Suggestions for Prevention and InfectionControl. 2nd edn. Bromley, UK: Cystic Fibrosis Trust, 2004.

4. Frederiksen B, Koch C, Hoiby N. Antibiotic treatment of initial colonization withPseudomonas aeruginosa postpones chronic infection and prevents deterioration ofpulmonary function in cystic fibrosis. Pediatr Pulmonol 1997;23:330e5.

5. Steinkamp G, Tummler B, Malottke R, et al. Treatment of Pseudomonas aeruginosacolonisation in cystic fibrosis. Arch Dis Child 1989;64:1022e8.

6. Valerius NH, Koch C, Høiby N. Prevention of chronic Pseudomonas aeruginosacolonisation in cystic fibrosis by early treatment. Lancet 1991;338:725e6.

7. Wiesemann HG, Steinkamp G, Ratjen F, et al. Placebo-controlled, double blind,randomized study of aerosolized tobramycin for early treatment of Pseudomonasaeruginosa colonization in cystic fibrosis. Pediatr Pulmonol 1998;25:88e92.

8. Munck A, Bonacorsi S, Mariani-Kurkdjan P, et al. Genotypic characterization ofPseudomonas aeruginosa strains recovered from patients with cystic fibrosis afterinitial and subsequent colonization. Pediatr Pulmonol 2001;32:288e92.

9. Ratjen F, Doring G, Nikolaizik WH. Effect of inhaled tobramycin on earlyPseudomonas aeruginosa colonisation in patients with cystic fibrosis. Lancet2001;358:983e4.

10. Ratjen F, Munck A, Kho P, et al; ELITE Study Group. Treatment of earlyPseudomonas aeruginosa infection in patients with cystic fibrosis: the ELITE trial.Thorax 2010;65:286e91.

11. Griese M, Muller I, Reinhard D. Eradication of initial Pseudomonas aeruginosacolonization in patients with cystic fibrosis. Eur J Med Res 2002;7:79e80.

12. Taccetti G, Campana S, Festini F, et al. Early eradication therapy againstPseudomonas aeruginosa in cystic fibrosis patients. Eur Respir J 2005;26:458e61.

13. Nixon GM, Armstrong D, Carzino R, et al. Clinical outcome after early Pseudomonasaeruginosa infection in cystic fibrosis. J Pediatr 2001;138:699e704.

14. Proesmans M, Boulanger L, Vermeulen F, et al. Eradication of recent Pseudomonasaeruginosa isolation: TOBI versus colistin/ciprofloxacin. J Cyst Fibros 2008;7(Suppl 2):S64.

15. Gibson RL, Emerson J, McNamara S, et al. Significant microbiological effect ofinhaled tobramycin in young children with cystic fibrosis. Am J Respir Crit Care Med2003;167:841e9.

16. Douglas TA, Brennan S, Gard S, et al. Acquisition and eradication of P. aeruginosa inyoung children with cystic fibrosis. Eur Respir J 2009;33:305e11.

17. Treggiari MM, Retsch-Bogart G, Mayer-Hamblett N, et al. Comparative efficacy andsafety of 4 randomized regimens to treat early Pseudomonas aeruginosa infection inchildren with cystic fibrosis. Arch Pediatr Adolesc Med 2011;165:847e56.

18. Elborn JS, Hodson M, Bertram C. Implementation of European standards of care forcystic fibrosisdcontrol and treatment of infection. J Cyst Fibros 2009;8:211e17.

19. Saiman L, Siegel J; Cystic Fibrosis Foundation Consensus Conference on InfectionControl Participants. Infection control recommendations for patients with cysticfibrosis: microbiology, important pathogens, and infection control practices to preventpatient-to-patient transmission. Am J Infect Control 2003;31(3 Suppl):S1e62.

20. Syrmis MW, O’Carroll MR, Sloots TP, et al. Rapid genotyping of Pseudomonasaeruginosa isolates harboured by adult and paediatric patients with cystic fibrosisusing repetitive-element-based PCR assays. J Med Microbiol 2004;53:1089e96.

21. Miller MR, Hankinson J, Brusasco V, et al. Standardisation of spirometry. Eur RespirJ 2005;26:319e38.

22. Saiman L, Marshall BC, Mayer-Hamblett N, et al; Macrolide Study Group.Azithromycin in patients with cystic fibrosis chronically infected with Pseudomonasaeruginosa: a randomized controlled trial. JAMA 2003;290:1749e56.

23. Lee TW, Brownlee KG, Conway SP, et al. Evaluation of a new definition for chronicPseudomonas aeruginosa infection in cystic fibrosis patients. J Cyst Fibros2003;2:29e34.

24. Schulz KF, Altman DG, Moher D; CONSORT Group. CONSORT 2010 Statement:updated guidelines for reporting parallel group randomised trials. Ann Intern Med2010;152:726e32.

25. Ho SA, Lee TW, Denton M, et al. Regimens for eradicating early Pseudomonasaeruginosa infection in children do not promote antibiotic resistance in this organism.J Cyst Fibros 2009;8:43e6.

26. Macdonald D, Cuthbertson L, Doherty C, et al. Early Pseudomonas aeruginosainfection in individuals with cystic fibrosis: is susceptibility testing justified? JAntimicrob Chemother 2010;65:2373e5.

27. Hansen CR, Pressler T, Høiby N. Early aggressive eradication therapy for intermittentPseudomonas aeruginosa airway colonization in cystic fibrosis patients: 15 yearsexperience. J Cyst Fibros 2008;7:523e30.

28. Schelstraete P, Deschaght P, Van Simaey L, et al. Genotype based evaluation ofPseudomonas aeruginosa eradication treatment success in cystic fibrosis patients.J Cyst Fibros 2010;9:99e103.

29. Hodson ME, Gallagher CG, Govan JR. A randomised clinical trial of nebulisedtobramycin or colistin in cystic fibrosis. Eur Respir J 2002;20:658e64.

30. Adeboyeku D, Scott S, Hodson ME. Open follow-up study of tobramycinnebuliser solution and colistin in patients with cystic fibrosis. J Cyst Fibros2006;5:261e3.

31. Goss CH, Mayer-Hamblett N, Aitken ML, et al. Association betweenStenotrophomonas maltophilia and lung function in cystic fibrosis. Thorax2004;59:955e9.

32. Dalbøge CS, Hansen CR, Pressler T, et al. Chronic pulmonary infection withStenotrophomonas maltophilia and lung function in patients with cystic fibrosis.J Cyst Fibros 2011;10:318e25.

33. Graff GR, Burns JL. Factors affecting the incidence of Stenotrophomonas maltophiliaisolation in cystic fibrosis. Chest 2002;121:1754e60.

34. Denton M, Rajgopal A, Mooney L, et al. Stenotrophomonas maltophiliacontamination of nebulizers used to deliver aerosolized therapy to inpatients withcystic fibrosis. J Hosp Infect 2003;55:180e3.

35. McPherson H, Rosenthal M, Bush A. Can mucoid Pseudomonasaeruginosa be eradicated in children with cystic fibrosis? Pediatr Pulmonol2010;45:566e8.

Cystic fibrosis


ORIGINAL ARTICLE

Understanding the natural progression in %FEV1decline in patients with cystic fibrosis:a longitudinal study

David Taylor-Robinson,1 Margaret Whitehead,1 Finn Diderichsen,2

Hanne Vebert Olesen,3 Tania Pressler,4 Rosalind L Smyth,5 Peter Diggle6

ABSTRACTBackground Forced expiratory volume in 1 s asa percentage of predicted (%FEV1) is a key outcome incystic fibrosis (CF) and other lung diseases. As peoplewith CF survive for longer periods, new methods arerequired to understand the way %FEV1 changes overtime. An up to date approach for longitudinal modellingof %FEV1 is presented and applied to a unique CF datasetto demonstrate its utility at the clinical and populationlevel.Methods and findings The Danish CF register contains70 448 %FEV1 measures on 479 patients seen monthlybetween 1969 and 2010. The variability in the data ispartitioned into three components (between patient,within patient and measurement error) using theempirical variogram. Then a linear mixed effects model isdeveloped to explore factors influencing %FEV1 in thispopulation. Lung function measures are correlated forover 15 years. A baseline %FEV1 value explains 63% ofthe variability in %FEV1 at 1 year, 40% at 3 years, andabout 30% at 5 years. The model output smooths out theshort-term variability in %FEV1 (SD 6.3%), aiding clinicalinterpretation of changes in %FEV1. At the populationlevel significant effects of birth cohort, pancreatic statusand Pseudomonas aeruginosa infection status on %FEV1are shown over time.Conclusions This approach provides a more realisticestimate of the %FEV1 trajectory of people with chroniclung disease by acknowledging the imprecision inindividual measurements and the correlation structure ofrepeated measurements on the same individual overtime. This method has applications for clinicians inassessing prognosis and the need for treatmentintensification, and for use in clinical trials.

INTRODUCTIONUnderstanding the long-term natural history ofchanges in lung function in people with lungdiseases is a research priority.1 In order to do this,objective measures of disease progression arenecessary. The per cent predicted forced expiratoryvolume in 1 s (%FEV1) is commonly used tomonitor lung function, and to describe diseaseseverity in cystic fibrosis (CF)2 and chronicobstructive pulmonary disease (COPD).3 %FEV1 isused to inform clinical decisions about changing orintensifying treatment, and as an outcome measurein clinical studies.4e6 Furthermore %FEV1 has beenshown to be related to survival in CF. Kerem et al’s

study in 1992 demonstrated that patients witha %FEV1 <30 had a 2-year mortality over 50%,7

though a more recent study shows that survivalrates at low levels of lung function have improvedin subsequent cohorts.8

Interpreting the significance of changes in %FEV1

in CF to inform patient management and tocounsel patients regarding prognosis requires anunderstanding of the inherent variability of %FEV1

measures within individuals, to determine whatconstitutes a clinically significant deterioration in%FEV1, rather than a change due to measurementerror, or recoverable day-to-day fluctuation in lungfunction.9 10 Furthermore, this variability needs tobe understood to make valid inferences about theassociation between covariates and %FEV1 inobservational studies.As survival in CF improves with successive

cohorts, there are many more people surviving intolate adulthood. An implication of this, coupledwith the availability of long-term follow-up data inCF registers, is that up to date methods should beadopted to interpret the long-term dynamics oflung function in CF. Statistical techniques for

Key messages

What is the key question?< Now that people with cystic fibrosis are living

much longer, how can we optimally describe thechanges in forced expiratory volume in 1 s asa percentage of predicted (%FEV1) over time ina way that is useful for clinicians at theindividual and the population level?

What is the bottom line?< We describe a novel modelling approach for

analysing changes in %FEV1 over time that canbe applied at the individual level to interpret theclinical significance of sudden changes in %FEV1, and at the population level to quantify theeffect of factors such as Pseudomonas aerugi-nosa acquisition.

Why read on?< Lung function measures are correlated for over

15 years, and a baseline %FEV1 value explains63% of the variability in %FEV1 at 1 year, 40% at3 years and about 30% at 5 years.

< Additional materials arepublished online only. To viewthese files please visit thejournal online (http://dx.doi.org/10.1136/thoraxjnl-2011-200953).1Department of Public Healthand Policy, University ofLiverpool, Liverpool, UK2Department of Social Medicine,University of Copenhagen,Copenhagen, Denmark3Cystic Fibrosis Center, AarhusUniversity Hospital, Aarhus,Denmark4Cystic Fibrosis Center,Rigshospitalet, Copenhagen,Denmark5Division of Child Health,University of Liverpool,Liverpool, UK6School of Health and Medicine,Lancaster University, Lancaster,UK

Correspondence toDr David Taylor-Robinson, MRCPopulation Health Scientist,Department of Public Health andPolicy, Whelan Building,University of Liverpool, LiverpoolL69 3GB, UK; [email protected]

RLS and PJD are joint seniorauthors.

Received 15 August 2011Accepted 26 February 2012

This paper is freely availableonline under the BMJ Journalsunlocked scheme, see http://thorax.bmj.com/site/about/unlocked.xhtml

Cystic fibrosis


Published Online First3 May 2012

longitudinal data analysis have been the subject of muchmethodological development over the past 20 years, and therandom intercept and slope model has become a popular analysisframework.4 5 11e14 While this is often appropriate for relativelyshort follow-up periods, there are theoretical reasons to suggestthat this approach makes assumptions that will lead to incorrectinferences if applied over longer follow-up periods. One centralassumption is that the variability in %FEV1 increases asa quadratic function over time (in proportion to time squared),which leads to estimates that diverge unrealistically over longertime periods. Methods for undertaking these analyses overlonger time periods have been described,15 but have not beencommonly applied.

In this study we analyse a unique population-level dataset ofpeople with CF that includes longitudinal %FEV1 measurestaken monthly for up to 30 years. We apply these methods todevelop a general model for %FEV1 decline that goes beyond thepopular random-intercept and slope approach, and explicitlydescribes the variability in %FEV1 within individuals over time.We show how this could be applied clinically to help interpretthe significance of changes in lung function, and at a populationlevel to explore the association of covariates (eg, Pseudomonasaeruginosa acquisition) with %FEV1 decline.

METHODSSubjectsAll patients aged over 5 years whose %FEV1 data were enteredon the Danish CF database between 1969 and 2010 were eligible.Post-transplant data from patients who had received a lungtransplant were excluded. Patients attending the two Danish CFcentres (Copenhagen and Aarhus) are seen routinely everymonth in the outpatient clinic for evaluation of clinical status,pulmonary function and microbiology of lower respiratory tractsecretions. It is estimated that coverage of people with CFresident in Denmark is almost complete from 1990 when CFcare was centralised. This coverage and the unparallelledfrequency of measurement make this a unique dataset forepidemiological research. The study was approved by the DanishData Inspectorate (Datatilsynet).

Lung function testingThe primary outcome for this analysis was %FEV1. Pulmonaryfunction tests were performed according to internationalrecommendations,16 measuring FEV1, expressed as a percentageof predicted values for sex and height using reference equationsfrom Wang or Hankinson.17 18

CovariatesCovariates in the analysis were age, sex, genotype coded as thenumber of Delta F508 alleles (0, 1 or 2), onset of chronic Pseu-domonas infection (coded 0 or 1 as a time-varying covariate),pancreatic insufficiency determined on the basis of pancreaticenzyme usage (coded 0 or 1 as a baseline covariate), birth cohort(six 10-year cohorts starting at 1948), and CF-related diabetes(CFRD) diagnosed using the WHO criteria (coded 0 or 1 asa time-varying covariate).

Statistical analysisA detailed explanation is given in the online appendix. Repeated%FEV1 measures on individuals are correlated, and this must beaccommodated to obtain valid inferences. We used a linearmixed effects model with longitudinally structured correla-tion,15 19 and contrasted our approach with the widely usedrandom intercept and slope model.20 We modelled randomvariation in %FEV1 over time for an individual subject so thatthe strength of the correlation of the random variation betweentwo values depends on the corresponding time separation. Themodel decomposed the overall random variation in the data intothree components: between subjects, between times withinsubjects, and measurement error.First, we fit a provisional model for the mean response by

ordinary least squares and used the empirical variogram of theresiduals (see figure E1 in the online appendix) to provide initialestimates for the three components of variation, and for theshape of the correlation function of the between-times-within-subjects component. We then re-estimated all of the modelparameters by maximum likelihood estimation, and usedgeneralised likelihood ratio statistics to compare nested models,and Wald statistics to test hypotheses about model parameters.We assessed associations between single or multiple covariatesand the population mean %FEV1 over time, and explored alter-natives to a linear function for the population-averaged timetrend.

RESULTSPopulation characteristicsThe dataset contained 70 448 lung function measures on 479patients seen between 1969 and 2010 in Denmark (table 1). Themedian number of %FEV1 measures per person was 101 (range2e597). The median follow-up period was 10.5 years (range0.1e31.5), with a total of 6500 person-years of follow-up. Forty-two patients were followed up for more than 30 years (see alsofigures E2 and E3 in the online appendix).

Table 1 Baseline characteristics of the Danish cystic fibrosis (CF) population

Birth cohort

‡1948 ‡1958 ‡1968 ‡1978 ‡1988 ‡1998 Total

N (%) 7 (1.5) 42 (8.8) 110 (23) 105 (21.9) 141 (29.4) 74 (15.4) 479 (100)

Women 1 (14.3) 19 (45.2) 48 (43.6) 52 (49.5) 74 (52.5) 42 (56.8) 236 (49.3)

No. Delta F508 ¼ 0 0 (0) 0 (0) 1 (0.9) 4 (3.8) 5 (3.5) 5 (6.8) 15 (3.1)

No. Delta F508 ¼ 1 2 (28.6) 14 (33.3) 26 (23.6) 24 (22.9) 42 (29.8) 19 (25.7) 127 (26.5)

No. Delta F508 ¼ 2 5 (71.4) 28 (66.7) 83 (75.5) 77 (73.3) 94 (66.7) 50 (67.6) 337 (70.4)

Developed chronic Pseudomonas 6 (85.7) 31 (73.8) 84 (76.4) 55 (52.4) 20 (14.2) 5 (6.8) 201 (42)

Missing infection information 0 (0) 5 (11.9) 2 (1.8) 2 (1.9) 1 (0.7) 0 (0) 10 (2.1)

Pancreatic insufficient 7 (100) 42 (100) 105 (95.5) 99 (94.3) 133 (94.3) 73 (98.6) 459 (95.8)

Copenhagen 7 (100) 38 (90.5) 83 (75.5) 72 (68.6) 79 (56) 50 (67.6) 329 (68.7)

Alive 4 (57.1) 27 (64.3) 79 (71.8) 77 (73.3) 132 (93.6) 74 (100) 393 (82)

Developed CFRD 3 (42.9) 21 (50) 41 (37.3) 31 (29.5) 22 (15.6) 1 (1.4) 119 (24.8)

CFRD, cystic fibrosis related diabetes.

Cystic fibrosis


Limitations of random intercept and slope modelThe high degree of short-term and long-term variation inpredicted %FEV1 is illustrated in figure 1. The standard randomintercept and slope model approach is illustrated over long andshort follow-up periods in figure 1A,C. This approach assumesthat any deviation of an individual’s trajectory from the popu-lation mean is linear in time over the whole of the follow-upperiod apart from independent random errors. One can see thatthis assumption is reasonable over short time periods, as illus-trated by the fit of the shorter dotted-line segments (figure 1A,C), but over longer time periods the individual data tracesdiverge unrealistically from their fitted linear mean trajectories(long solid line). Our proposed model produces a much closerfit to the data (figure 1B,D), and one that better reflectsthe relative magnitude of the three estimated components ofvariation in %FEV1 over time.

Quantifying the variability in %FEV1 over timeThe empirical variogram quantifies the variability in the dataset(figure 2A). The intercept at time zero represents measurementerror because there can be no true within-person variation at

a time lag of zero. Of the total variance in the Danish dataset,about half is due to systematic differences between patients (eg,genotype, sex or pancreatic status), two-fifths is within patients,representing change over time (disease progression), and one-tenth is ‘measurement error ’. In practice, this last componentrepresents the combined effects of technical errors, and physi-ological variability occurring at time intervals less than themonthly interval of measurement, for example, day-to-dayvariability. This error variance equates to an average SD of 6.3%for repeated measures on the same individual at short timeintervals. Figure 2B shows the proportion of the within-personvariability in %FEV1 at follow-up time (t), which can beexplained by their %FEV1 value at baseline. For example, about50% of the within-patient variability at t¼2.5 years is explainedby the baseline measurement, and about 30% at t¼5 years.Overall, the dependence on baseline measures gradually decaysand is negligible at 15 years.

Clinical utility of our proposed modelThe model can be used to guide interpretation of sudden changesin lung function. Consider seeing the person in figure 1B at

Figure 1 Comparison of conventionalrandom intercept and slope model overshort and long follow-up periods,versus our proposed Gaussian processmodel. (A) Data for a single individual,illustrating that a linear trend fitsreasonably well over short timeperiods, but gives a very poor fit to thisindividual’s complete data; linear trendsare fitted by ordinary least squares. (B)The same data with the fitted trajectoryof the stationary Gaussian processmodel. The smoothed fitted trace isa better representation of the ‘true’underlying lung function, and could beused in real time to guide theinterpretation of sudden changes in lungfunction. For instance, the sudden dropto under 30% indicated by the arrow isnot mirrored in the model trace,suggesting that this may be recoverablerandom fluctuation. (C, D)Corresponding plots for a secondindividual. %FEV1, forced expiratoryvolume in 1 s as a percentage ofpredicted.

Cystic fibrosis

862 Thorax 2012;67: doi:10.1136/thoraxjnl-2011-200953860–866.

around age 9 (as indicated by the arrow in the figure), when herlung function has dropped to below 30%. On the basis of thisone-off measurement, one might be quite guarded in terms ofprognosis. However, our modelled trace (thick black line in figure1B) suggests that her underlying lung function is changing lessdramatically, with a modelled %FEV1 of around 50%. We suggestthat this estimate provides a more realistic assessment ofunderlying lung function by smoothing out the short-termvariability. This could be a useful adjunct to clinical decision-making. As well as providing information about the significanceof a sudden change in lung function, figure 2B also quantifies thepredictive value of a contemporary %FEV1 measure. In terms ofcounselling patients, this means that a higher %FEV1 today isassociated with a higher %FEV1 at subsequent time points, butthe predictive value deteriorates over time as illustrated in thefigure.

Effect of covariates on lung function in the Danish populationWe explored the effect of covariates that have been associatedwith %FEV1 in previous studies to demonstrate how this modelcan be used to answer questions at the population level (seetable E1 online appendix for univariate associations).4 There was

no evidence to suggest that covariate effects were nonlinear (seefigure E4 in online appendix). The final model included age,Pseudomonas status, pancreatic status, cohort and CFRD(table 2). Note that the estimated covariate effects in table 2 arepopulation-averaged effects, that is, they describe average valuesof %FEV1 for sub-populations of individuals sharing the sameexplanatory characteristics, rather than for any one individual.The most prominent effects are associated with birth cohort,pancreatic function and the onset of Pseudomonas infection(figure 3). There is clear separation between the three mostrecent birth cohorts, with a successive increase in the interceptterm at age 5 (83% in the 1978e88 cohort vs 96% in the post-1998 cohort) (figure 3A and figures E9eE10 in online appendix).There is a large change in the point estimate for the rate ofchange of lung function in the post-1998 (0.24%) compared withthe 1988e98 cohort (�1% per year), such that the post-1998cohort appears to be improving over the period of measurement.The three cohorts spanning the years 1948e1978 have a similaroverall rate of decline around �0.3% per year, with an interceptat age 5 of 66%. Pancreatic insufficiency is associated witha significantly steeper rate of decline of lung function (�0.92%per year, 95% CI �1.7 to �0.3), as is acquisition of Pseudomonas

Figure 2 Quantifying the variability inforced expiratory volume in 1 s asa percentage of predicted (%FEV1) withthe variogram approach. (A) Scaledempirical variogram for the Danish data.The solid line (variogram function)represents the variance of thedifference between residual errorswithin individuals at time lags from 0 to30 years. The variogram functionincreases up to about 15 years,corresponding to a decreasingcorrelation between paired lungfunction measures withincreasing time separation. Thevariogram partitions the variability inthe data into three components: withinperson, between person, and error. (B)Proportion of variability in an individual’s %FEV1 at follow-up time t that is explained by their %FEV1 at baseline. This shows that the variogram canpredict 63% of the variability from the population average at 1 year, which decreases to around 60%, 40%, 30% and 10% at 2, 3, 5 and 10 yearsrespectively.

Table 2 Estimates from final multivariate model

Point estimate Lower 95% CI Upper 95% CI p Value

Intercept at age 5 years 66.02 61.13 70.92 <0.001

CFRD �2.47 �3.58 �1.37 <0.001

Age �0.26 �0.49 �0.03 0.025

Cohort$1948 (reference 1968) 1.20 �25.50 27.90 0.930

Cohort$1958 �0.75 �10.01 8.51 0.874

Cohort$1978 16.60 10.15 23.05 <0.001

Cohort$1988 25.19 19.11 31.27 <0.001

Cohort$1998 29.81 22.85 36.78 <0.001

Pancreatic sufficiency 2.78 �10.43 15.99 0.679

Pseudomonas aeruginosa infection �0.51 �0.72 �0.29 <0.001

Age3cohort$1948 �0.03 �0.67 0.61 0.920

Age3cohort$1958 0.06 �0.23 0.34 0.699

Age3cohort$1978 �0.72 �1.00 �0.44 <0.001

Age3cohort$1988 �0.72 �1.09 �0.35 <0.001

Age3cohort$1998 0.50 �0.41 1.42 0.280

Age3pancreatic sufficiency 0.98 0.29 1.67 0.005

CFRD, cystic fibrosis related diabetes.

Cystic fibrosis


infection (�0.5% per year, 95% CI �0.72 to �0.3) (figure 3B andfigure E8 in online appendix). CFRD is associated with a drop inintercept of �2.5% (95% CI �3.6% to �1.37%), but has noeffect on the rate of decline of lung function.

DISCUSSIONWe describe a novel longitudinal modelling technique specificallyaimed at analysing long sequences of repeated measurements,and apply this to %FEV1 from a CF population. We show howthis approach could be used to inform patient management, byaiding the interpretation of sudden changes in lung function,and by quantifying the predictive value of a baseline %FEV1

measure up to 15 years later. At the population level, we showhow our model can be used to quantify the effect of covariateson populations or sub-populations. Translation of these methodsinto clinical practice is important because people with CF areliving longer, and we have shown how commonly appliedapproaches are unhelpful over long follow-up periods.

This study quantifies the short-term variability in %FEV1 inthis population (SD 6.3%), and demonstrates that %FEV1

measures within individuals are correlated over time lags of15 years or more. We have also explored the effect of previouslystudied risk factors for lung function decline in the Danish CFpopulation, and have demonstrated significant effects of birthcohort, pancreatic status and Pseudomonas infection status.

The findings from this study have a number of clinicalapplications. Quantifying the variability in lung functionmeasures is essential to make correct clinical interpretation.10

Exploiting the unusually high frequency of data collection inDenmark, this study implies that on average a change in %FEV1

of >13% (ie, twice the error SD, to give a 95% confidence range)is likely to represent true within-patient variation over time(disease progression), whereas anything less than this could bedue to short-term fluctuation, which may recover. Stanbrooket al21 found a pooled within-subject %FEV1 SD of 4.5% whenmeasured over a 9-day period in 21 stable adults with CF. Thispopulation is different to the population in our study, who weremeasured regardless of clinical status, and one would thereforeexpect greater variability. Other studies have shown that peoplewith CF, asthma and COPD have more short-term variability inlung function tests22e24 and that more impaired lung function isassociated with greater variability.25

Our model can be used to generate an underlying represen-tation of an individual’s ‘true’ lung function trajectory (figure1B,D) that smoothes out the noise inherent in %FEV1 measures.These smoothed traces could be used to inform clinical decision-

makingdthe model fit curves in figure 1 provide more realisticestimates of underlying lung function, and more valid criteria forclinical decisions. We propose that this model could be used todevelop a real-time smoothing tool embedded in electronicpatient records to aid clinical interpretation of spirometry data.We suggest that access to this information would provide somere-assurance to patents experiencing lower than expected lungfunction values, since lung function can recover quite dramati-cally, and these data suggest that a linear or stepwise decline inlung function over time is not the norm.We have generated, for the first time to our knowledge, the

variogram function for %FEV1 in people with CF over longfollow-up periods. This precisely quantifies how %FEV1

measures are correlated over time. Furthermore we have donethis for the whole CF population of Denmark. This quantifiesthe degree to which a baseline %FEV1 measure can be usedto predict subsequent %FEV1 measures over long follow-upperiods, and is likely to be of interest to clinicians andpatients. We demonstrate a long-term correlation between levelsof %FEV1 within an individual. This suggests that there is long-term predictive value in a high %FEV1 measuredpeople withCF with a high %FEV1 at baseline are more likely to have a high%FEV1 up to 15 years later than individuals with a lowerbaseline %FEV1 (figure 2B). However, the predictive value of a%FEV1 measure drops away rapidly over this period. We can saythat on average a %FEV1 reading today explains about 63% ofthe variability in %FEV1 at 1 year, 40% at 3 years, and about30% at 5 years.This corroborates Rosenthal’s study,26 which found that

baseline %FEV1 explains 66% of the variability in %FEV1 at1 year, and Mastella et al’s study of European registry data inwhich differences in lung function at enrolment at age 5, cate-gorised as mild, moderate or severe, tracked through the study toage 40.27 Konstan et al also describe how a lower %FEV1 fora given age can be used to characterise the aggressiveness of lungdisease.28 Other studies have shown a high %FEV1 to be anindependent risk factor for a greater rate of decline of %FEV1

over the next few years.4 29 This is not at odds with our findingshere; a high %FEV1 can be a risk factor for greater decline inthe short term, while still being associated with a relativelyhigher %FEV1 over the longer term.28

At the population level we show how our approach can beapplied to quantify the effect of covariates on changes inlung function. Furthermore, the partitioning of the variability in%FEV1 and the precise description of the correlation structurecaptured in the model provide important information for sample

Figure 3 Effect of covariates onforced expiratory volume in 1 s asa percentage of predicted (%FEV1). (A)Birth cohort effect in the final model.There is clear separation between thethree most recent birth cohorts, witha successive increase in the interceptterm at age 5 years. (B) Effect ofpancreatic insufficiency andPseudomonas infection on thepredicted population trajectory fora person born in the 1988e1998cohort. CF, cystic fibrosis; PA,Pseudomonas aeruginosa.

Cystic fibrosis


size calculations in longitudinal clinical studies with %FEV1 asan outcome. Increasingly longitudinal outcomes are being usedin randomised control trials, and to undertake an a priori samplesize calculation it is essential to have information on thecorrelation structure. Furthermore, our modelled %FEV1 tracecould be used as an outcome in its own right.

As with other studies of patients with CF,30 there is a strikingcohort effect evident in this population. The treatment of CFlung disease has been transformed over the period captured inthis analysis, from 1969 to the present day. Particularlyimpressive is the improvement in lung function in the post-1998cohort by comparison with preceding birth cohorts. Althoughpatients in this group are early in their disease progression, theoverall picture suggests that new therapeutic strategies arecontinuing to provide improvements in respiratory function inCF.

Our approach to modelling changes in %FEV1 can be appliedover long follow-up periods. This is in contrast to the widelyused random intercept and slope approach that has been appliedin studies of CF and COPD over short-term4 27 31 32 and longer-term follow-up periods.10 11 14 33 The development and testingof the new approach is facilitated by the nature of the DanishCF registerdto our knowledge there are no other datasets thatcontain such frequent (monthly) measures of lung function onindividuals measured over very long periods (up to 31.5 years).However, the fact that the data are from Denmark does notinfluence the validity of the methods we have described, sincethese are essentially context free. Furthermore, this method doesnot exploit any features of our data that are unique to CF, and isequally applicable to other clinical areas that generate longsequences of repeated measurements. As a next step werecommend that this method be applied to longitudinal datacollected in other CF registries, such as the UK, to clarify howrobust this approach is in terms of predicting changes in %FEV1

over time, and to better understand how this might informclinical decision making. Future research could explore theutility of our proposed model in other diseases such as COPD.

A limitation of this study is the likely influence of survivorbias on lung function estimates in the earlier birth cohorts. Inthe 1948e1978 period, the intercept at age 5 appears signifi-cantly lower than in the other cohorts, but there is also a shal-lower rate of decline of lung function. This is likely to be due tothe incomplete capture of patients in earlier cohorts, withcensoring due to death leaving only the more stable survivors.This is a common problem in datasets of this type.34 Fitting themodel by maximum likelihood automatically corrects for selec-tion bias that depends on a patient’s observed lung functionmeasurements prior to death, although not for any additionaldependence on unmeasured features of their lung functiontrajectory.15 19

Pancreatic sufficiency had an important effect on the overallrate of decline of lung function (+0.9% per year). In Konstan’sstudy4 pancreatic sufficiency was the most important protectivefactor in the age group 6e8 years (+1.33% per year). The smallnumber of pancreatic-sufficient individuals in the Danishdataset (n¼20, 5%) have a notably different lung functionphenotype, maintaining near-normal lung function over theperiod of follow-up (see plot in online appendix). The onset ofPseudomonas infection was associated with a significant increasein the rate of decline of lung function, by around �0.5% per year,similar to that reported in the study by Konstan, in whichPseudomonas colonisation was associated with an increased rateof decline of FEV1 of �0.31% per year in the 6e8-year-old agegroup, and �0.22 in the 9e12-year-old age group.4

In conclusion, our modelling approach provides a more real-istic estimate of the %FEV1 trajectory in CF, which could beapplied in real time to help clinicians interpret the significance ofchanges in %FEV1. Furthermore, our approach quantifies thepredictive value of a baseline %FEV1 measure, over threedecades. This method is equally applicable to the longitudinalassessment of %FEV1 in other lung diseases, and can enable morerobust comparisons of populations, including groups studied inclinical trials. As people are now living for many decades withthese diseases, the development of tools to better understand thenatural history of this important outcome will be essential forimproved clinical care, as well as being a key research priority.1

Acknowledgements We thank Professor Peter Oluf Schiøtz for his support inaccessing the data for this analysis.

Contributors DTR, MMW, FD, TP, RLS and PD conceived and designed the study. TPand HVO collected the data. DTR undertook the analysis and PD supervised analysis.DTR, MMW, RLS and PD interpreted the results and drafted the paper. All authorscontributed to and approved the final draft for publication.

Funding This work was supported by an MRC Population Health Scientist Fellowshipto DTR (G0802448). The funders had no role in study design, data collection andanalysis, decision to publish, or preparation of the manuscript.


Ethics approval The study was approved by the Danish Data inspectorate(Datatilsynet). Danish CF registry data were used, analysed anonymously.


REFERENCES1. Holgate ST. Priorities for respiratory research in the UK. Thorax 2007;62:5e7.2. Davies JC, Alton EW. Monitoring respiratory disease severity in cystic fibrosis.

Respir Care 2009;54:606e17.3. Rabe KF, Hurd S, Anzueto A, et al; Global Initiative for Chronic Obstructive Lung

Disease. Global strategy for the diagnosis, management, and prevention of chronicobstructive pulmonary disease: GOLD executive summary. Am J Respir Crit Care Med2007;176:532e55.

4. Konstan MW, Morgan WJ, Butler SM, et al. Risk factors for rate of decline in forcedexpiratory volume in one second in children and adolescents with cystic fibrosis.J Pediatr 2007;151:134e9, 139.e1.

5. Konstan MW, Schluchter MD, Xue W, et al. Clinical use of ibuprofen is associatedwith slower FEV1 decline in children with cystic fibrosis. Am J Respir Crit Care Med2007;176:1084e9.

6. Ramsey BW, Dorkin HL, Eisenberg JD, et al. Efficacy of aerosolized tobramycin inpatients with cystic fibrosis. N Engl J Med 1993;328:1740e6.

7. Kerem E, Reisman J, Corey M, et al. Prediction of mortality in patients with cysticfibrosis. N Engl J Med 1992;326:1187e91.

8. George PM, Banya W, Pareek N, et al. Improved survival at low lung function incystic fibrosis: cohort study from 1990 to 2007. BMJ 2011;342:d1008.

9. Corey M. Power considerations for studies of lung function in cystic fibrosis. ProcAm Thorac Soc 2007;4:334e7.

10. Hnizdo E, Yu L, Freyder L, et al. The precision of longitudinal lung functionmeasurements: monitoring and interpretation. Occup Environ Med2005;62:695e701.

11. Stern DA, Morgan WJ, Wright AL, et al. Poor airway function in early infancy andlung function by age 22 years: a non-selective longitudinal cohort study. Lancet2007;370:758e64.

12. van Diemen C, Postma D, Siedlinski M, et al. Genetic variation in TIMP1 but notMMPs predict excess FEV1 decline in two general population-based cohorts. RespirRes 2011;12:57.

13. Harber P, Tashkin DP, Simmons M, et al. Effect of occupational exposures ondecline of lung function in early chronic obstructive pulmonary disease. Am J RespirCrit Care Med 2007;176:994e1000.

14. Corey M, Edwards L, Levison H, et al. Longitudinal analysis of pulmonary functiondecline in patients with cystic fibrosis. J Pediatr 1997;131:809e14.

15. Diggle P, Heagerty P, Liang KY, et al. Analysis of Longitudinal Data. 2nd edn. Oxford:Oxford University Press, 2002.

16. Miller MR, Hankinson J, Brusasco V, et al. Standardisation of spirometry. Eur RespirJ 2005;26:319e38.

17. Wang X, Dockery DW, Wypij D, et al. Pulmonary function between 6 and 18 years ofage. Pediatr Pulmonol 1993;15:75e88.

18. Hankinson JL, Odencrantz JR, Fedan KB. Spirometric reference values froma sample of the general U.S. population. Am J Respir Crit Care Med1999;159:179e87.

19. Fitzmaurice GM, Laird NM, Ware JH. Applied Longitudinal Analysis. Hoboken, NewJersey: Wiley, 2004.

Cystic fibrosis


20. Laird NM, Ware JH. Random-effects models for longitudinal data. Biometrics1982;38:963e74.

21. Stanbrook MB, Corey M, Tullis DE. The repeatability of forcedexpiratory volume measurements in adults with cystic fibrosis. Chest2004;125:150e5.

22. Hruby J, Butler J. Variability of routine pulmonary function tests. Thorax1975;30:548e53.

23. Pennock BE, Rogers RM, McCaffree DR. Changes in measured spirometric indices.What is significant? Chest 1981;80:97e9.

24. Cooper PJ, Robertson CF, Hudson IL, et al. Variability of pulmonary function tests incystic fibrosis. Pediatr Pulmonol 1990;8:16e22.

25. Enright PL, Beck KC, Sherrill DL. Repeatability of spirometry in 18,000 adult patients.Am J Respir Crit Care Med 2004;169:235e8.

26. Rosenthal M. Annual assessment spirometry, plethysmography, and gas transfer incystic fibrosis: do they predict death or transplantation. Pediatr Pulmonol2008;43:945e52.

27. Mastella G, Rainisio M, Harms HK, et al. Allergic bronchopulmonary aspergillosis incystic fibrosis. A European epidemiological study. Epidemiologic Registry of CysticFibrosis. Eur Respir J 2000;16:464e71.

28. Konstan MW, Wagener JS, VanDevanter DR. Characterizing aggressiveness andpredicting future progression of CF lung disease. J Cyst Fibros 2009;8(Suppl 1):S15e19.

29. Vandevanter DR, Wagener JS, Pasta DJ, et al. Pulmonary outcome prediction(POP) tools for cystic fibrosis patients. Pediatr Pulmonol 2010;45:1156e66.

30. Elborn JS, Shale DJ, Britton JR. Cystic fibrosis: current survival and populationestimates to the year 2000. Thorax 1991;46:881e5.

31. Vestbo J, Sorensen T, Lange P, et al. Long-term effect of inhaled budesonide in mildand moderate chronic obstructive pulmonary disease: a randomised controlled trial.Lancet 1999;353:1819e23.

32. MacLean JE, Atenafu E, Kirby-Allen M, et al. Longitudinal decline in lung volume ina population of children with sickle cell disease. Am J Respir Crit Care Med2008;178:1055e9.

33. Kohansal R, Martinez-Camblor P, Agusti A, et al. The natural history of chronicairflow obstruction revisited: an analysis of the Framingham offspring cohort. Am JRespir Crit Care Med 2009;180:3e10.

34. Frederiksen B, Lanng S, Koch C, et al. Improved survival in the Danish center-treated cystic fibrosis patients: results of aggressive treatment. Pediatr Pulmonol1996;21:153e8.


Cystic fibrosis

866

Journal club

Histamine-releasing factor: a possible future therapeutictarget for asthma and allergyMast cells and basophils are key players in the IgE-dependent allergic response. These cellstrigger an inflammatory cascade by secreting numerous preformed pro-inflammatory chemicalmediators such as histamine, proteases and cytokines into the blood. Histamine-releasingfactor (HRF), found in nasal, skin blister and bronchoalveolar lavage fluids, is a proteinsecreted by macrophages that can stimulate histamine, interleukin 4 and IL-3 production fromIgE-sensitised basophils and mast cells. Despite considerable efforts, research has failed toidentify a HRF receptor. However, HRF is known to have intracellular and extracellularfunctions, with the latter implicated in late-phase allergic reactions and chronic inflammation.This study identified a subset of IgE and IgG antibodies as HRF-interacting molecules in

vitro. Through complex molecular research techniques it was confirmed that HRF togetherwith HRF-reactive IgE triggered mast cell activation in vitro, confirming its pro-inflammatoryrole. Specific HRF inhibitor peptides were also characterised. Through several differentintricate experiments these peptides were found to suppress passive cutaneous anaphylaxisand mast cell-dependent airway inflammation and substantially reduce allergic airwayinflammation. These findings provide further clarification that HRF promotes allergicinflammation in the skin and lung via an immediate hypersensitivity reaction.The authors conclude that HRF could be a future novel therapeutic target for asthma and

allergy. However, this development may be reliant on the discovery of a specific HRF receptor.Furthermore, the additional intracellular activities of HRF are extensive and more specificresearch related to these functions may also be essential.

< Kashiwakura JC, Ando T, Matsumoto K, et al. Histamine-releasing factor has a proinflammatory role in mouse models ofasthma and allergy. J Clin Invest 2012;122:218e28.

Victoria Elizabeth Beasley

Correspondence to Dr Victoria Elizabeth Beasley, ST2. Imperial NHS Trust, Department of Renal Medicine, HammersmithHospital, Du Cane Road, London W12 0HS, UK; [email protected]


Provenance and peer review Commissioned; internally peer reviewed.

Thorax 2012;67:866. doi:10.1136/thoraxjnl-2012-201726

Published Online First 1 March 2012


ORIGINAL ARTICLE

Long-term cultivation-independent microbial diversityanalysis demonstrates that bacterial communitiesinfecting the adult cystic fibrosis lung show stabilityand resilience

Franziska Anne Stressmann,1 Geraint B Rogers,1 Christopher J van der Gast,2

Peter Marsh,3 Louic S Vermeer,4 Mary P Carroll,5 Lucas Hoffman,6

Thomas W V Daniels,5 Nilesh Patel,7 Benjamin Forbes,1 Kenneth Deans Bruce1

ABSTRACTBackground Culture-independent analysis of therespiratory secretions of people with cystic fibrosis (CF)has identified many bacterial species not previouslydetected using culture in this context. However, little isknown about their clinical significance or persistence inCF airways.Methods The authors characterised the viablebacterial communities in the sputum collected from 14patients at monthly intervals over 1 year usinga molecular community profiling techniquedterminalrestriction fragment length polymorphism. Clinicalcharacteristics were also collected, including lungfunction and medications. Ecological communitymeasures were determined for each sample. Microbialcommunity change over time within subjects wasdefined using ecological analytical tools, and thesemeasures were compared between subjects and toclinical features.Results Bacterial communities were stable withinsubjects over time but varied between subjects,despite similarities in clinical course. Antibiotic therapytemporarily perturbed these communities whichgenerally returned to pretreatment configurationswithin 1 month. Species usually considered CFpathogens and those not previously regarded as suchexhibited similar patterns of persistence. Less diversesputum bacterial communities were correlated to lungdisease severity and relative abundance of Pseudomonasaeruginosa.Conclusion Whilst not true in all cases, the microbialcommunities that chronically infect the airways ofpatients with CF can vary little over a year despiteantibiotic perturbation. The species present tended tovary more between than within subjects, suggesting thateach CF airway infection is unique, with relatively stableand resilient bacterial communities. The inverserelationship between community richness and diseaseseverity is similar to findings reported in other mucosalinfections.

INTRODUCTIONWhilst advances in longevity continue, respiratoryfailure remains the leading cause of death in cysticfibrosis (CF). The lung damage that leads to respi-ratory failure results from the chronic bacterial

infection and host immune response that is typicalby adulthood, if not before. Infection by certainbacterial species has been associated with poorerclinical outcomes.1 2 These species have beendetected previously using selective culture-basedapproaches that are inherently biased. Morerecent studies are increasingly characterising thebacteria present in the CF airways using culture-independent approaches that avoid these biases.3e6

In this way, much greater bacterial diversity hasbeen found typically in these airway secretionsthan previously appreciated. It is also important tonote that bacterial species, many of them anaer-obes, identified using culture-independent methodsare frequently of abundances comparable to speciesroutinely cultivated and treated as pathogens in theCF airway.6e8

From the earliest of these culture-independentstudies, it has been postulated that species presentin the lower airways represent an interactingcommunity,9 implying roles for ‘non-traditional’species in CF lung disease pathogenesis and treat-ment response. However, support for sucha community model requires determination of the

Key messages

What is the key question?< To what degree do the bacterial species present

in cystic fibrosis (CF) sputum represent chroniccolonisation of the lower airways?

What is the bottom line?< The CF bacterial community, including species

only recently reported, as well as knownpathogens such as Pseudomonas aeruginosa,is in many instances stable and resilient overextended periods.

Why read on?< These findings provide a basis for understanding

the clinical importance of the bacterial commu-nity in disease progression, and a context forstudies that assess the impact of antimicrobialtherapy.

< Additional materials arepublished online only. To viewthese files please visit thejournal online (http://dx.doi.org/10.1136/thoraxjnl-2011-200932).1Molecular MicrobiologyResearch Laboratory, Instituteof Pharmaceutical Science,King’s College London, London,UK2NERC Centre for Ecology andHydrology, Wallingford, UK3Health Protection AgencySouth East, SouthamptonGeneral Hospital, Southampton,UK4Institute of PharmaceuticalScience, King’s College London,London, UK5Cystic Fibrosis Unit,Southampton UniversityHospitals NHS Trust,Southampton, UK6Department of Pediatrics,University of Washington,Seattle, Washington, USA7Department of Pharmacy,Kingston University, Kingston,UK

Correspondence toDr Geraint Rogers, MolecularMicrobiology ResearchLaboratory, Institute ofPharmaceutical Science, King’sCollege London, 150 StamfordStreet, Franklin-Wilkins Building,London SE1 9NH, UK;[email protected]

Received 10 August 2011Accepted 9 May 2012

Cystic fibrosis


Published Online First15 June 2012

persistence of the various community members, the communityresponse to perturbations, for example antibiotics, and its rela-tionship with disease severity. Some species may remain in thelower airways for relatively short periods, while others mayestablish chronic infections. Infection over longer periods mayalso be more likely to adversely impact upon respiratory health.In support of these concepts, CF airways communities havebeen suggested to contain both core and transient species,10

with community structure and membership related to age andlung disease severity.11 However, such studies are generally crosssectional or short term,12 and a longitudinal analysis of multiplesubjects has not yet been performed.

Longitudinal analysis has followed how specific CF pathogenschange with time during chronic airway infection.13 14 Thisapproach must now be applied to the entire community. Thefirst step is to define the variability of CF airway microbiota overtime, during clinical changes, and with antibiotic treatment.Community constituency could change in a number of ways. Atone extreme, the composition of the community may changefrequently, with the community found in one specimen notresembling those in subsequent samples from the same subject.At the other extreme, the community may be so stable that thesame species are recovered at any sampling point. CF airwaydisease is typified by periods of relative respiratory healthpunctuated by exacerbation; with augmented treatment, usuallyrequirement for antibiotics. As such, we hypothesised that thevarying treatment that a patient with CF receives over thecourse of a year, and the dynamic nature of respiratory andgeneral health, would be associated with detectable changes inthe bacterial community and a ‘highly variable’ composition. Totest this hypothesis required the collection of samples frommultiple subjects spanning an extended period.

Here, we report the culture-independent analysis of viablebacteria present in monthly samples from airways of 14 adultswith CF over a year. These data were collected alongside detailedclinical metadata. A series of analytical tools tested thecompositional stability of these communities over time.10 Wealso analysed the relationship between community structureand lung disease severity.

MATERIALS AND METHODSClinical samplesThis observational study of 14 adult patients with CF over12 months was undertaken with ethical approval from South-ampton and South West Hampshire Research Ethics Committee(06/Q1704/26). Eligible subjects were aged 18 or over and hadexperienced at least three pulmonary exacerbations (CFPEs)requiring antibiotic treatment in the prior 12 months. Samplesfrom these patients were previously used in an analysis of coreand satellite bacterial taxa.10

Patient clinical details are summarised in supplementary table 1.The start of a CFPE was defined by the clinician’s decision toinitiate antibiotic therapy for deteriorating clinical status,broadly based on factors described previously.15 The end of CFPEwas defined by the decision to cease antibiotic therapy due tostabilisation or improvement in signs and symptoms. A total of39 episodes of CFPE were experienced by these patients over theyear, ranging from 0 (patients 3 and 8) to 6 (patient 9) witha mean of 2.8, SD 1.4. Overall, 17 different antibiotic combi-nations were used with five ‘elective’ antibiotic courses given forreasons other than worsening respiratory symptoms.

One sputum sample was obtained from each patient atapproximately monthly intervals. The majority of samples(65%) were collected at least 21 days prior to, or 21 days after

cessation of, antibiotics for CFPE (supplementary table 2). Allsputa were stored at 48C immediately after expectoration,shipped at 48C (in accordance with handling guidelines16) andstored at �808C prior to processing, or cultured by the HealthProtection Agency South East, UK following standard operatingprocedures.

Clinical dataClinical measures (including forced expiratory volume in 1 s(FEV1) and forced expiratory volume in 6 s (FEV6) in litres,temperature, and patient reported outcome (PRO) scores) wererecorded at sample collection. PRO scores were recorded usingvisual analogue scores for individual symptoms (‘breathlessness’,‘cough severity ’, ‘sputum production’ and ‘general well-being’)from 0 (no symptoms) to 100 (worst symptoms).

Culture-independent analysisExclusion of DNA from non-viable cells in sputa via cross-linking using propidium monoazide17 18 and subsequent nucleicacid extraction were performed as described previously.19

PCR and terminal restriction fragment length polymorphism(T-RFLP) profiling were carried out as previously described3

(see supplementary materials and methods).

Statistical analysisBacterial species richness (the number of bacterial species ina sample) was inferred from T-RF band number as previouslydescribed.20 Cumulative bacterial taxon richness (the totalnumber of different bacterial taxa identified after each successivetime point) was assessed using taxaetime relationships (TTRs);these describe how richness increases with the time over whichthe community is monitored20 21 (see supplementary materialsand methods).Distanceedecay relationships (DDRs) of bacterial community

similarity with time (temporal differences in days) were deter-mined. DDRs describe how similarity in taxa compositionbetween two communities varies with a measure of distance(here, the time between pairs of sample collections)22 (seesupplementary materials and methods).Principal component analysis (PCA), used to identify differ-

ences in community structures, was carried out with softwaredeveloped in house, using the Python programming languagewith the Numpy package. The principal components werecalculated using eigenvalue decomposition of mean-centred andauto-scaled data.

RESULTSMeasures of community species constituency, and clinical data,were compared for sputa collected monthly over a year from 14patients. Bacterial community measures assessed the changesin the relative abundance of species detected both withinindividuals over time, and between subjects.

Bacterial diversityBacterial species dynamics were determined using selectiveculture-based analysis and DNA-based profiling. Culture-basedmicrobiological analysis (supplementary table 3) showed thatPseudomonas aeruginosa and Pseudomonas spp. were the mostfrequently reported (70.8% of samples). Routine culture alsocommonly identified ‘oral flora’ (56.0%). ‘Unidentified isolates’(1.2%), ‘coliform’ (2.4%), ‘Stenotrophomonas maltophilia’ (0.6%),‘methicillin-sensitive Staphylococcus aureus (MSSA)’ (10.7%),‘Staphylococcus sp.’ (1.8%) and ‘Streptococcus Group F’ (0.6%) werealso reported. No significant correlations were found between

Cystic fibrosis


species detected by this method and clinical status, or inthe coculture of separate species (Pearson correlations withBonferroni corrections).

The second tool used to analyse bacterial species dynamicswas the culture-independent technique, T-RFLP. T-RFLPprovides measures of species constituency when based on 16SrRNA gene analysis. In the 168 samples, a total of 1158 T-RFbands were detected, representing 37 distinct T-RF lengths (eachrepresenting one or more species). The mean number of T-RFbands in profiles from individual patients over the study rangedfrom 2.6 (SD 2.4, n¼12) (patient 12) to 12.7 (SD 1.6, n¼12)(patient 10), with an overall mean of 6.9 (SD 3.9, n¼168).

T-RFLP data were used to characterise which specieswere present over the study period in each subject. In thefollowing sections, data for three patients are selected to repre-sent the types of community dynamics observed; all other dataare presented as supplementary information. These patientsrepresented the highest (patient 2) and lowest (patient 4) rela-tive bacterial community stability (Bray-Curtis similarityindex), and the relative stability closest to the group mean(patient 13).

Community change over timeTTR analysis identified the cumulative number of speciesdetected in each sample (figure 1 for patients 2, 4 and 13;

remainder in supplementary figure 1). For the overall dataset,scaling component values (w) were observed to range from 0.05(patient 6) to 0.38 (patient 12) (supplementary figure 1). Thisindicated that some patients acquired new species morefrequently than others. No significant relationship was identifiedhowever between the number of species at the beginning of thestudy and the subsequent rate of new species acquisition (pair-wise Pearson correlation across all patients, p¼0.699, r2¼0.5).The overall change in community composition over the 12-

month sampling period was assessed using the inverse of theBray-Curtis similarity index. This was applied to all possiblepairs of samples from individual subjects and the mean of eachsubject’s resulting values calculated for the year (n¼11) (resultsexpressed as % change; supplementary table 4). While the meanchange in bacterial composition for the group overall was 35.5%(SD 18.2%, n¼154 samples), low levels of change were seen incertain patients (eg, patients 2, 3 and 5 had 4.3%, 18.1% and18.8% change respectively; supplementary table 4), with higherrelative levels of change seen in other patients (eg, patients 4 and14 had 70.1% and 64.3% change, respectively; supplementarytable 4). Together, these results indicated that bacterialcommunity variation is inherently different among patientswith CF. No relationship was identified between mean richnessof a patient’s community and its variability over time (Pearsoncorrelation, p¼0.252, R¼0.389).

Figure 1 Species richness dynamics and taxaetime relationship. Species richness (black circles, black line) and the cumulative species richness(white circles, no line) are shown. Power regression lines were fitted to the cumulative species richness. Also given are the taxaetime relationshippower law equation (S ¼ cTw), and coefficients of determination (r2). All regression coefficients were significant (p<0.001).

Figure 2 Distanceedecay relationships. The distanceedecay of bacterial community similarity with time. Power regression lines were fitted, andpower law regression equations (SBC ¼ cTd) and coefficients of determination (r2) are given in each instance. Only two relationships were significant(p¼0.040 for patients 4 and 7).

Cystic fibrosis


To test whether similarity between samples was due solely tothe time interval between sample collection, DDR were used(figure 2 and supplementary figure 2). A statistically significantdistance-decay relationship was identified for two subjects (4and 7, p¼0.040 for both). For all other subjects, no relationshipbetween community composition at two given sample pointsand time interval between them was found (supplementarytable 5). This suggests that the community stability observedwas not simply a function of sampling frequency.

PCA was performed on community data from each subject(figure 3 and supplementary figure 3). While the communities ineach patient varied over time, no trends within patients wereobserved consistently across the patient group as a whole.Further, samples collected during periods of exacerbation werenot found consistently to be significantly different from non-exacerbation communities (Pearson correlations for communityrichness, p¼0.216, R¼0.501, structure (slope), p¼0.056,R¼0.765, and P aeruginosa relative abundance, p¼0.023,R¼0.648). PCA analysis on data from all subjects are shown infigure 4. When the two components that explain the highestlevels of variance are examined, the sample points from indi-vidual patients cluster more tightly with each other than withcommunities from other subjects in a number of cases. Therelatively small proportion of total variation represented by anyindividual component (14% for component 1) indicates thatbacterial community differences between patients cannot beascribed to differences in a small number of correlated factors,but rather are likely to represent many unrelated factors, eachcontributing a small amount of variance.

Clinical dataFigure 5 shows the variation in FEV1, PRO measures and anti-biotics for patients 2, 4 and 13 respectively; data for all patientsare in supplementary figure 4, with FEV6 data shown insupplementary figure 5. The FEV1 (% predicted) values rangedfrom 5.6% (patient 9) to 94.2% (patient 10), mean 49.1% (SD17%, n¼163). FEV6 (litres) values ranged from 0.56 (patient 2) to3.85 (patient 10), mean 2.27 (SD 0.79, n¼162). To reduce theeffect of short-term lung function fluctuations, a weekly meanfor FEV1 and FEV6 was also calculated (here, the mean wascalculated for all measurements 3 days preceding and followingeach sampling point). PRO scores are also displayed as a sumof the respective values for the component symptom scores(breathlessness, sputum production, cough and generalwellbeing).Apart from patients 1, 2 and 11, for whom a substantial

decline was observed, lung function for the other patients wasrelatively stable over the year. PRO scores varied greatly betweenpatients, however no clear relationship was observed in relationto lung function changes.

Correlating clinical and microbial datasetsAirway community characteristics were correlated with clinicalcharacteristics. Subjects were separated into higher and lowerhalves based on FEV1, with these resulting groups examined forcommunity characteristic differences. Five of the six patientswith FEV1 above the median (patients 1, 6, 10, 11 and 14) sharedthree or more of the following characteristics: species number>7, a mean relative P aeruginosa abundance of <35%, a TTRvalue of <0.1, and a DDR value of <0.1 with no statisticalsignificance. All of these patients had experienced at least threeCFPEs, except patient 6 (two CFPEs), and showed a change incommunity composition over the year above the median(31.7%), except patient 6 (24.8%). In contrast, only one subjectwith a low mean FEV1 (patient 9) had four of these criteria.

The statistical significance of these relationships was testedthrough Pearson’s coefficient correlation analyses (table 1).Three significant relationships were identified after the applica-tion of Bonferroni correction; higher PRO scores reported bywomen than men (p¼0.002, R¼0.758); a steeper rank abun-dance slope (the slope of the regression line fitted to speciesabundance ordered from highest to lowest abundance) withdetection of low species number (p¼0.002, R¼�0.763) anda steeper rank abundance slope correlated with high Paeruginosarelative abundance (p¼0.002, R¼�0.763).In addition, pairwise comparisons between clinical and

microbiological data were performed across all 168 samplesindependent of patient identity. Significant relationshipsafter Bonferroni correction between FEV1 and FEV6 and higherspecies number per sample, steeper rank abundance slope,and lower P aeruginosa relative abundance were identified(table 1).

Figure 3 Principal componentsanalysis of individual patients. Plots forcommunity profiles by component 1(PC1) and component 2 (PC2) areshown. The percentage of variance inthe data explained by each componentis indicated on the relevant axis.Sampling points are indicated withnumbers 1e12, with samples collectedduring pulmonary exacerbationsindicated by boxed numbers.

Figure 4 Principal components analysis (PCA) of all subjects. Thevariance explained by each component is shown in brackets. Solidshapes indicate samples collected during pulmonary exacerbations.

Cystic fibrosis


DISCUSSIONThis study defined the stability of communities over time inrelation to host and clinical characteristics. In general, whilesome degree of overlap was observed, the airways of individualswith CF contained communities that differed from those inother subjects, often despite similar clinical courses. In contrast,subjects’ bacterial communities changed little over the year,despite intervening respiratory exacerbation periods; the changesobserved during antibiotic treatment did not persist, returningto approximate pretreatment structures within a month.

Of the previous studies that applied culture-independentmethods to examine the microbiota of CF airways,3 5 6 23 24 twofocused on the relationship with lung disease severity. Inagreement with these studies,10 11 subjects with higher lung

function in our study tended to have more diverse airwaycommunities and vice versa. Furthermore, as also shown previ-ously,10 11 23 Paeruginosa was associated with lower communityrichness and lower lung function.The feature that distinguished this study from previous work

was its longitudinal analysis of viable cells. Most earlier studiesidentified the bacteria in individual CF samples in cross-sectionalanalyses or, less often, in two or more samples taken over a shorttime period (eg, in treatment of exacerbation).8 12 Tunney et al12

applied a combination of T-RFLP and culture to demonstratethat sputum communities changed little during and soon afterantibiotic treatment. While this and other short-term studieswere revealing, a longer-term study of multiple subjects wasrequired to define long-term airway bacterial dynamics. We

Figure 5 Forced expiratory volume in 1 s (FEV1), patient reported outcomes (PROs) and antibiotic treatment at sampling. Lung function FEV1 (litres)(black diamonds) is shown at each sampling point, and a calculated weekly mean FEV1 (black solid line). The individual patient reported outcomescores (breathlessness, sputum production, cough, general wellbeing) were summed for each sampling point and are shown as grey areas (the higherthe score, the worse the symptoms). Pulmonary exacerbations with antibiotic treatments are indicated as black dashes at the bottom of the graph withthe details for each pulmonary exacerbation (marked with an asterisk and the respective number) shown in supplementary table 1.

Table 1 Summary of significant correlations*

Correlate 1 Correlate 2Pearson correlationcoefficient (r) Significance (p) n

Significant patient-specific pairwise correlations

Clinical: clinical correlations

FEV1 Temperature 0.539 0.047 14

FEV6 No. CFPEs �0.573 0.032 14

No. CFPEs Age �0.533 0.05 14

Genotype Age 0.652 0.012 14

PRO sum Gender 0.758 0.002* 14

BMI Nebulised colomycin 0.621 0.018 14

Clinical: microbiological correlations

FEV6 Rank abundance slope �0.602 0.023 14

FEV6 Species number 0.549 0.042 14

Microbiological: microbiological correlations

Pseudomonas aeruginosa T-RF % Rank abundance slope �0.763 0.002* 14

P aeruginosa T-RF % Species number �0.763 0.002* 14

P aeruginosa T-RF % Bray-curtis similarity �0.715 0.004 14

Species number Rank abundance slope �0.711 0.004 14

Significant patient independent pairwise correlations across all samples





FEV1 P aeruginosa T-RF % �0.706 0.0001 168

FEV6 P aeruginosa T-RF % �0.691 0.0001 168

*Correlations for which p<0.05 after Bonferroni correction.BMI, body mass index; CFPE, cystic fibrosis pulmonary exacerbation; FEV1, forced expiratory volume in 1 s; FEV6, forced expiratoryvolume in 6 s; PRO, patient-reported outcome; T-RF, terminal restriction fragment.

Cystic fibrosis


sought to distinguish transient from chronic infectious charac-teristics, as the latter are more likely to be associated with lungdisease progression. While Sibley et al6 previously analysedbacterial dynamics over a period comparable to that examinedhere, their study focused on a single individual, precludinggeneralisation and inter-subject comparison.

Further, we focused on only viable cells18 19 to ensure anaccurate characterisation of community stability. This mini-mised the contribution of dead bacteria, for example, fromantibiotic treatment or immune response. Thus, bacterial speciesconsistently observed can be described with some confidence asbeing chronically present in the CF lower airways.

Given the many perturbations experienced by the airwaymicrobiota in an adult patient (including antibiotic therapy,immune response, and cough clearance), we hypothesised thateach subject’s bacterial community would vary greatly inmembership over time. Cross-sectional analyses of bacterialcommunities in CF airways have reported a wide range ofbacteria.10 11 Such variation could be explained by high speciesturnover. Surprisingly though, we found CF lung bacterialcommunities to be, to a large extent, stable over a year.

In contrast, the composition of airway communities differedamong patients. Interestingly, this was reminiscent of findingsfrom other mucosal microbial communities, for example, thegastrointestinal tract.25 The reasons for this stability withinpatients, and divergence between patients, are unknown;however, disease severity and its correlate, treatment history,may play selective roles. We also observed that CF airwaybacterial communities were resilient. While antibiotic therapycan have a significant short-term impact on CF lung commu-nities,8 12 our data indicate that these communities approachpretreatment configurations rapidly.

Our results also extend the previous finding that CF lungcommunity richness is inversely correlated with lung diseaseseverity.10 Neither community richness nor lung functionchanged significantly within subjects during the study. As such,we could not define the contribution of individual species, orspecific communities, to lung disease pathogenesis. However, wedid find an inverse relationship between community richnessand lung function, similar to previous cross-sectional studies.11

This is reminiscent of relationships found for other mucosalmicrobial communities.26 27 In this regard, while no causalitycan or should be inferred, the current results may indicate thatCF lung health is highest in the presence of a diverse microbiota.Alternatively, repeated antibiotic exposure, or the toxic productsof pathogens such as P aeruginosa, could be responsible forcommunity simplification. Antibiotic treatment could not becontrolled for in this study. As such, further investigation of therelationships observed between community richness and diseaseseverity is now warranted.

In summary, we found that the species that constitute CFairway communities are present for prolonged periods, and thatcommunities can change little over a year, despite antibiotictreatment. This stability indicates that much of the widerbacterial diversity seen in airway samples does not representshort-term colonisation.

Contributors FS, GR, CvdG, PM, LV, MC, LH, TD, NP, BF and KB were all involved inthe conception, design, interpretation of data, and revising and final approval of thearticle. In addition, CvdG and LV were responsible for statistical analysis, and TD forsample and metadata collection. GR is the guarantor. In addition, collaborator KD wasinvolved in the processing of samples, PC in data processing.

Funding This work was supported by the Anna Trust (grant number KB2008) and theSeattle Children’s Hospital Center for Clinical and Translational Research.


Patient consent Detail has been removed from this case description/these casedescriptions to ensure anonymity. The editors and reviewers have seen the detailedinformation available and are satisfied that the information backs up the case theauthors are making.

Ethics approval Ethics approval was provided by Southampton and South WestHampshire Research Ethics Committee.


REFERENCES1. Courtney JM, Dunbar KE, McDowell A, et al. Clinical outcome of Burkholderia

cepacia complex infection in cystic fibrosis adults. J Cyst Fibros 2004;3:93e8.2. Ballmann M, Rabsch P, von der Hardt H. Long-term follow up of changes in FEV1

and treatment intensity during Pseudomonas aeruginosa colonisation in patients withcystic fibrosis. Thorax 1998;53:732e7.

3. Rogers GB, Carroll MP, Serisier DJ, et al. Characterization of bacterial communitydiversity in cystic fibrosis lung infections by use of 16s ribosomal DNA terminalrestriction fragment length polymorphism profiling. J Clin Microbiol 2004;42(Suppl 11):5176e83.

4. Stressmann FA, Rogers GB, Klem ER, et al. Analysis of the bacterial communitiespresent in lungs of patients with cystic fibrosis from American and British centers. JClin Microbiol 2011;49(Suppl 1):281e91.

5. Bittar F, Richet H, Dubus JC, et al. Molecular detection of multiple emergingpathogens in sputa from cystic fibrosis patients. PLoS One 2008;3(Suppl 8):e2908.

6. Sibley CD, Parkins MD, Rabin HR, et al. A polymicrobial perspective of pulmonaryinfections exposes an enigmatic pathogen in cystic fibrosis patients. Proc Natl AcadSci U S A 2008;105:15070e5.

7. Tunney MM, Field TR, Moriarty TF, et al. Detection of anaerobic bacteria in highnumbers in sputum from patients with cystic fibrosis. Am J Respir Crit Care Med2008;177:995e1001.

8. Worlitzsch D, Rintelen C, Bohm K, et al. Antibiotic-resistant obligate anaerobesduring exacerbations of cystic fibrosis patients. Clin Microbiol Infect 2009;15(Suppl 5):454e60.

9. Rogers GB, Hart CA, Mason JR, et al. Bacterial diversity in cases of lung infection incystic fibrosis patients: 16S ribosomal DNA (rDNA) length heterogeneity PCR and16S rDNA terminal restriction fragment length polymorphism profiling. J ClinMicrobiol 2003;41(Suppl 8):3548e58.

10. van der Gast CJ, Walker AW, Stressmann FA, et al. Partitioning core andsatellite taxa from within cystic fibrosis lung bacterial communities. ISME J2010;5:780e91.

11. Cox MJ, Allgaier M, Taylor B, et al. Airway microbiota and pathogen abundance inage-stratified cystic fibrosis patients. PLoS One 2010;5(Suppl 6):e11044.

12. Tunney MM, Klem ER, Fodor AA, et al. Use of culture and molecular analysis todetermine the effect of antibiotic treatment on microbial community diversity andabundance during exacerbation in patients with cystic fibrosis. Thorax2011;66:579e84.

13. Jelsbak L, Johansen HK, Frost AL, et al. Molecular epidemiology and dynamics ofPseudomonas aeruginosa populations in lungs of cystic fibrosis patients. InfectImmun 2007;75(Suppl 5):2214e24.

14. Romling U, Fiedler B, Bosshammer J, et al. Epidemiology of chronic Pseudomonasaeruginosa infections in cystic fibrosis. J Infect Dis 1994;170(Suppl 6):1616e21.

15. Fuchs HJ, Borowitz DS, Christiansen DH, et al. Effect of aerosolized recombinanthuman DNase on exacerbations of respiratory symptoms and on pulmonary functionin patients with cystic fibrosis. The Pulmozyme Study Group. N Engl J Med1994;331:637e42.

16. HPA. Standards Unit, Department for Evaluations, Standards and Training, Centre forInfections. 2008. Investigation of bronchoalveolar lavage, sputum and associatedspecimens, BSOP 57. http://www.evaluations-standards.org.uk (accessed 23 May2012).

17. Nocker A, Sossa-Fernandez P, Burr MD, et al. Use of propidium monoazide for live/dead distinction in microbial ecology. Appl Environ Microbiol 2007;73(Suppl 16):5111e17.

18. Rogers GB, Stressmann FA, Koller G, et al. Assessing the diagnostic importance ofnonviable bacterial cells in respiratory infections. Diagn Microbiol Infect Dis 2008;62(Suppl 2):133e41.

19. Rogers GB, Marsh P, Stressmann AF, et al. The exclusion of dead bacterial cells isessential for accurate molecular analysis of clinical samples. Clin Microbiol Infect2010;16:1656e8.

20. van der Gast CJ, Ager D, Lilley AK. Temporal scaling of bacterial taxa is influencedby both stochastic and deterministic ecological factors. Environ Microbiol2008;10:1411e18.

21. Rogers GB, Skelton S, Serisier DJ, et al. Determining cystic fibrosis-affected lungmicrobiology: comparison of spontaneous and serially induced sputum samples byuse of terminal restriction fragment length polymorphism profiling. J Clin Microbiol2010;48:78e86.

22. van der Gast CJ, Gosling P, Tiwari B, et al. Spatial scaling of arbuscularmycorrhizal fungal diversity is affected by farming practice. Environ Microbiol2011;13:241e9.

23. Klepac-Ceraj V, Lemon KP, Martin TR, et al. Relationship between cystic fibrosisrespiratory tract bacterial communities and age, genotype, antibiotics andPseudomonas aeruginosa. Environ Microbiol 2010;12:1293e303.

Cystic fibrosis


24. Harris JK, De Groote MA, Sagel SD, et al. Molecular identification of bacteriain bronchoalveolar lavage fluid from children with cystic fibrosis. Proc Natl Acad SciU S A 2007;104:20529e33.

25. Wu GD, Chen J, Hoffmann C, et al. Linking long-term dietary patterns with gutmicrobial enterotypes. Science 2011;334:105e8.

26. Li Y, Ge Y, Saxena D, et al. Genetic profiling of the oral microbiota associated withsevere early-childhood caries. J Clin Microbiol 2007;45:81e7.

27. Wang Y, Hoenig JD, Malin KJ, et al. 16S rRNA gene-based analysis of fecalmicrobiota from preterm infants with and without necrotizing enterocolitis. ISME J2009;3:944e54.


Cystic fibrosis

873Thorax October 2012 Vol 67 No 10

Journal club

Tracheobronchial transplantation with a bioartificalnanocompositeThis proof-of-concept study describes the transplantation of a tracheobronchial airway witha stem-cell-seeded bioartifical nanocomposite in a 36-year-old man presenting with symptomsconsistent of a recurrent primary mucoepidermoid cancer. He had previously been treatedwith debulking surgery and postoperative radiation. The tumour extension was deemedunresectable, thus, transplant with artificial biomaterial was offered.The distal trachea and proximal bronchi were resected for complete tumour removal. The

airway was replaced with a synthetic bioartificial nanocomposite, tailor-made in the shape ofthe patient’s airway. The graft was seeded with autologous bone-marrow cells and cultured ina bioreactor for 36 h prior to transplantation. An extracellular like matrix and proliferatingcells were noted within the graft following incubation. To augment the regeneration process,the patient received granulocyte colony-stimulating factor and epoetin b subcutaneousinjections for 2 weeks postoperatively. No major complications were encountered and thepatient remained asymptomatic and tumour-free 5 months later. Postoperative investigationssuggested integration and remodelling of the transplanted material. Biopsies demonstratedvessel formation and epithelialisation. In addition, up-regulation of miR-34/449 micro-RNA,epoetin-receptor expression and anti-apoptotic genes and an increase in the number of plasmacomponents contributing to regeneration were seen.This study demonstrates the feasibility of tailor-made bioengineered synthetic transplants

in patients. However, further long-term studies are needed to ensure transplant effectiveness,efficacy and safety.

< Jungebluth P, Alici E, Baiguera S, et al. Tracheobronchial transplantation with a stem-cell-seeded bioartificial nanocomposite:a proof-of-concept study. Lancet 2011;378:1997e2004.

K Sritharan

Correspondence to Dr K Sritharan, CT1, North Middlesex University Hospital NHS Trust, Sterling Way, London N181QX, UK;[email protected]

Provenance and peer review Not commissioned; internally peer reviewed.



ORIGINAL ARTICLE

Lung function is abnormal in 3-month-old infants withcystic fibrosis diagnosed by newborn screening

Ah-Fong Hoo,1 Lena P Thia,2 The Thanh Diem Nguyen,2 Andrew Bush,3 Jane Chudleigh,1

Sooky Lum,2 Deeba Ahmed,2 Ian Balfour Lynn,3 Siobhan B Carr,4

Richard J Chavasse,5 Kate L Costeloe,6 John Price,7 Anu Shankar,8 Colin Wallis,1

Hilary A Wyatt,7 Angela Wade,9 Janet Stocks,2 on behalf of the London CysticFibrosis Collaboration (LCFC)

ABSTRACTBackground Long-term benefits of newborn screening(NBS) for cystic fibrosis (CF) have been established withrespect to nutritional status, but effects on pulmonaryhealth remain unclear.Hypothesis With early diagnosis and commencement ofstandardised treatment, lung function at w3 months ofage is normal in NBS infants with CF.Methods Lung clearance index (LCI) and functionalresidual capacity (FRC) using multiple breath washout(MBW), plethysmographic (pleth) FRC and forcedexpirations from raised lung volumes were measured in71 infants with CF (participants in the London CFCollaboration) and 54 contemporaneous healthy controlsage w3 months.Results Compared with controls, and after adjustmentfor body size and age, LCI, FRCMBW and FRCpleth weresignificantly higher in infants with CF (mean difference(95% CI): 0.5 (0.1 to 0.9), p¼0.02; 0.4 (0.1 to 0.7),p¼0.02 and 0.9 (0.4 to 1.3), p<0.001, z-scores,respectively), while forced expiratory volume (FEV0.5) andflows (FEF25e75) were significantly lower (�0.9(�1.3 to �0.6), p<0.001 and �0.7 (�1.1 to �0.2),p¼0.004, z-scores, respectively). 21% (15/70) of infantswith CF had an elevated LCI (>1.96 z-scores) and 25%(17/68) an abnormally low FEV0.5 (below �1.96z-scores). While only eight infants with CF hadabnormalities of LCI and FEV0.5, using both techniquesidentified abnormalities in 35% (24/68). Hyperinflation(FRCpleth >1.96 z-scores) was identified in 18% (10/56)of infants with CF and was significantly correlated withdiminished FEF25e75 (r¼�0.43, p<0.001) but not withLCI or FEV0.5.Conclusion Despite early diagnosis of CF by NBS andprotocol-driven treatment in specialist centres, abnormallung function, with increased ventilation inhomogeneityand hyperinflation and diminished airway function, isevident in many infants with CF diagnosed through NBSby 3 months of age.

INTRODUCTIONThe prognosis of cystic fibrosis (CF) has improveddramatically over the years due to implementationof aggressive treatment to optimise nutrition andpulmonary health following diagnosis andincreasing global uptake of newborn screening(NBS) for CF.

Despite convincing evidence of the long-termbenefits of NBS for CF with respect to improvednutritional status,1 2 the extent to which pulmonaryoutcomes have improved remains controversial.Although some studies have failed to demonstrateany benefits with screening,3 others have reportedsignificantly less pulmonary disease on chest radi-ography4 and stable lung function (LF) with lessmarked decline over time in the screened group.5e8

The Australian Respiratory Early Surveillance Teamfor Cystic Fibrosis (AREST-CF) group initiallyreported normal values of forced expired volume in0.5 s (FEV0.5) in NBS infants with CF in the first6 months of life,9 whereas a more recent report usingslightly different methodology found significantreductions in forced expired flows and volumesduring the same period.10

Even in the absence of any clinical respiratorysymptoms, infants with CF diagnosed clinicallyhave impaired LF shortly after diagnosis11 12 andthis impairment persists into school age despitetreatment in specialist CF centres.13e15 We havepreviously shown that the lung clearance index(LCI) measured by multiple breath washout(MBW) is a more sensitive measure of early lungdisease in preschool and young school-age childrenwith CF than spirometry,16e18 and that

Key messages

What is the key question?< Is lung function normal at 3 months of age in

infants with cystic fibrosis (CF) diagnosed vianewborn screening?

What is the bottom line?< Despite early diagnosis of CF by newborn

screening and protocol-driven treatment inspecialist centres, abnormal lung function isevident in many screened infants with CF by3 months of age.

Why read on?< This study, the largest of its kind and the only

one with contemporaneous healthy controls,describes early lung development in newbornscreened infants with CF.

< Additional materials arepublished online only. To viewthese files please visit thejournal online (http://dx.doi.org/10.1136/thoraxjnl-2012-201747).1Respiratory Unit, Great OrmondStreet Hospital for Children NHSFoundation Trust, London, UK2Portex Respiratory Unit, UCLInstitute of Child Health, London,UK3Department of PaediatricRespiratory Medicine, ImperialCollege & Royal Brompton &Harefield Hospital NHSFoundation Trust, London, UK4Department of PaediatricRespiratory Medicine, Barts &The London Children’s Hospital,London, UK5Queen Mary’s Hospital forChildren, Epsom & St HelierUniversity Hospitals NHS Trust,Surrey, UK6Neonatal Unit, HomertonUniversity Hospital NHSFoundation Trust, Barts & TheLondon School of Medicine andDentistry, Queen MaryUniversity of London, London,UK7Department of PaediatricRespiratory Medicine, KingsCollege Hospital, London, UK8Department of Child Health,Lewisham Healthcare NHSTrust, London, UK9Centre for PaediatricEpidemiology & Biostatistics,UCL Institute of Child Health,London, UK

Correspondence toDr Ah-Fong Hoo, RespiratoryUnit, Great Ormond StreetHospital for Children NHSFoundation Trust, London WC1N3JH, UK;[email protected]

Received 10 February 2012Accepted 25 May 2012

Cystic fibrosis



measurements during the preschool years are highly predictiveof LF at 6e10 years of age.13 During infancy, complementaryinformation is provided if the forced expiratory manoeuvres19

and MBW are undertaken.20 Nevertheless, the extent to whichinfant LF tests should be used to help guide CF managementduring infancy remains controversial.21

Since universal CF NBS was implemented in the UK inOctober 2007, the median age of CF diagnosis has fallen tow1 month.22 However, there have yet to be any large rando-mised controlled trials of treatment in this age group. For theseto be initiated, it is vital to have a greater understanding of thenatural history of lung disease in such infants. This study aimedto determine baseline LF within the first 3 months of age in NBSinfants with CF and compare findings to prospectively recruitedhealthy infants of similar age. We hypothesised that, with earlydiagnosis and commencement of standardised specialist treat-ment, there would be no difference in LF by 3 months of age inNBS infants with CF compared with healthy babies.

METHODSAs part of a longitudinal collaborative research program ofinfants with CF diagnosed by NBS (see online supplement), thisstudy received ethical approval (#09/HO71/314) from theNorth Thames Multi-Centre Research Ethics Committee (REC)and the local RECs of the participating specialist centres.

NBS infants with CF born between January 2009 and July2011 were eligible for recruitment to the London CF Collabo-ration (LCFC). Healthy full-term infants of similar age wererecruited prospectively from a London community (for details,including eligibility criteria, see online supplement, section 2).Lung function tests (LFTs) at 3 months were completed.

Study protocolClinical dataWith parental consent, participating centres prospectivelycompleted case record forms (CRFs) at diagnosis and eachsubsequent clinic visit, documenting mode and date of diagnosis,presentation, genotype, history of respiratory symptoms and/orinfection, microbiology, treatment, somatic growth and addi-tional investigations undertaken for clinical purposes. Theseforms enabled auditing and tracking of participating CF centres’adherence to a standardised study treatment protocol (onlinesupplement, section 6) in accordance to the UK CF Trustguidelines.23

Following diagnosis, all infants were commenced on multivi-tamins, pancreatic supplements if appropriate, and prophylacticflucloxacillin according to the standardised study treatmentprotocol (online supplement, section 6). Cough swabs weretaken every 2e3 months at clinics and whenever symptomatic,using a standardised protocol for collection, storage and analysisof samples.24

Lung function testing and anthropometryAll infants were tested at Great Ormond Street Hospital/UCLInstitute of Child Health at w3 months postnatal age, whenclinically stable and at least 3 weeks after any respiratory tractillness. Infants were weighed and examined, oxygen saturation(SpO2) levels were measured (Masimo Radical-7 pulse oximeter,Irvine, California, USA) and vital signs assessed prior toadministering chloral hydrate orally (60e100 mg/kg). Weightand crown-heel length were expressed as z-scores to adjust forage and sex.25 LFTs were performed during epochs of quiet sleepwith the child lying supine. Heart rate and SpO2 were moni-

tored continuously throughout testing. Infant urine or maternalsaliva samples were collected for cotinine assay to validateparental reported smoking habits (online supplement, section 5,table E4). Parents provided informed written consent and werepresent throughout measurements.LCI, a measure of ventilation inhomogeneity, and functional

residual capacity (FRCMBW) were measured with MBW, usinga respiratory mass spectrometer and customised software.20 TheJaeger BabyBody device (V.4.65; Care Fusion, San Diego, Cali-fornia, USA) was used to measure plethysmographic FRC(FRCpleth),26 27 total respiratory compliance (Crs) and resistance(Rrs),28 forced expiratory volume (FEV0.5), forced vital capacity(FVC) and forced expiratory flows (FEF75, FEF25e75) froman inflation pressure of 30 cmH2O using the raised volumetechnique.19 Measurements were always performed in thatorder. Details of data collection and analysis as describedpreviously20 27e30 are summarised in the online supplement.Results were expressed as z-scores to adjust for body size, sexand age if appropriate using reference equations derived from upto 140 healthy white infants studied in our department over thepast decade using identical equipment and protocols19 31 32

(online supplement, section 3). Abnormal lung function wasdefined as results outside the 95% limits of normal (ie, below �1.96 z-scores for FEV0.5 or >1.96 z-scores for LCI (<2.5th or>97.5th centile respectively).

Statistical analysis and power of studyComparisons of group differences were performed using student ttests for continuous variables (age, body size and LFT z-scores) orc2 analyses for categorical variables (sex, ethnicity, maternalsmoking, parental occupation and asthma) (PASW StatisticsV.18). Within the CF group, the relationship between LFToutcomes was quantified using Pearson correlations. Multiplelinear regression analyses were used to quantify impact ofmaternal smoking (data not shown) and the extent to whichpotential clinical determinants (CF genotype, respiratory symp-toms, cough swab culture and antibiotic use) are associated withLFT outcomes within the CF group after adjustment for sex andcurrent body size. Model estimates and differences betweengroups are presented with 95% CIs. Taking into account twoprimary outcomes (LCI and FEV0.5) a sample size of 70 infantswith CF and 50 controls (equivalent to 58 in each group) willallow detection of differences equivalent to 0.58 or 0.66 z-scores atthe 5% significance level with 80% or 90% power, respectively.33

Statistical significance was taken as p<0.05.

RESULTSDuring the study period, 110 infants screened positive for CF, ofwhom successful LFT measurements were obtained in 79 (81%of those eligible). Inspection of the prospective CRFs, andregular discussion with consultants, suggested that the stand-ardised treatment protocol had been adhered to in all infants atthe time of testing. Details are summarised in figure 1, includingsuccess rates for each LFT. For clarity, the results presented hereare limited to those without meconium ileus (n¼71), althoughincluding such infants did not affect the results (data notshown). Of the 274 families with potentially eligible healthyinfants whom we contacted, 39 (14%) were ineligible. Of theremaining 235, 54 (23% of those eligible) attended for LFTs(figure 1, details in figure E1, online supplement).Median (IQR) age at diagnosis for the CF infants was 3.6

(3.1e4.4) weeks; the majority were either homozygous (59%) orheterozygous (32%) DF508 and 8 (11%) were pancreatic

Cystic fibrosis


sufficient. Sixty five (92%) NBS infants with CF had had norespiratory or only mild coryzal symptoms prior to LFTs(figure 2). The remaining 6 (8%) had had significant cough and/or wheeze, 16 (23%) had had at least one positive cough swab byw3 months of age (two with Pseudomonas aeruginosa; 10 withStaphylococcus aureus and four with Haemophilus influenzae).These children were prescribed antibiotics according to a stand-ardised treatment protocol (online supplement, section 6) inaddition to the prophylactic flucloxacillin prescribed for allinfants on diagnosis. No infant had positive cough swab culturefor Stenotrophomonos maltophilia; Burkholderia cepacia complex orAspergillus fumigatus. Additional antibiotics were also prescribedfor 36 infants with negative cough swabs, 32 (89%) of whomhad had prior cough and/or wheeze. Thus in total, 52 (73%) NBSinfants with CF received additional antibiotics prior to their firstLFTs at w3 months (figure 2), but all were asymptomatic at thetime of testing. Of the 46 infants with CF who had cough swabstaken within 10 days of LFTs, only three yielded positive growth(two for S aureus and one for H influenzae).

Table 1 summarises background characteristics of studyparticipants. With the exception of a slightly lower, but statis-tically significant, gestational age, both groups were very similar.

Infant details at the time of LFTs are shown in table 2. NBSinfants with CF were on average 1 week younger than healthycontrols, with a significantly greater proportion having birthweights <10th percentile. After taking age and sex into account,infants with CF were significantly lighter and shorter and hada lower body mass index than the controls. The change in

weight z-score between birth and LFTwas significantly lower ininfants with CF than in controls (table 2).

Lung function resultsTechnically satisfactory measurements of LCI and FRCMBW

were obtained in 121 infants (97% of those in whom they wereattempted), forced expired volumes and flows in 120 (96%),plethysmographic FRC in 103 (82%), Crs and Rrs in 88 (70%)(figure 1, table 3). Failures were largely due to infants wakingprior to completion of the test protocol or technically unac-ceptable data. With the exception of tidal volume, which wasslightly higher (0.4 z-scores) in those with CF, there were nosignificant differences for any of the tidal breathing outcomes orpassive respiratory mechanics (Crs and Rrs). LCI was signifi-cantly higher in those with CF, whether expressed as absolutevalues (online supplement, table E2) or as z-scores (table 3,figure 3). Similarly after adjustment for age, length and sex,FRCMBW was significantly higher (0.38 z-score) in infants withCF. Additional evidence of hyperinflation and gas trapping wasseen in those with CF, with FRCpleth and DFRC (ie, differencebetween FRCpleth and FRCMBW z-scores) being significantlyelevated (table 3, figure 3). After correction for body size, forcedexpired flows and volumes were significantly decreased ininfants with CF compared with healthy controls, by between0.5 and 0.9 z-scores, indicating the presence of airwaysobstruction (table 3, figure 3). The extent to which interpreta-tion of these results would have differed had they simply been

Figure 1 Flow diagram showing success rates in relation torecruitment and achievement of technically acceptable infant lungfunction outcomes in NBS infants with CF (A) and healthy controls (B).CF, cystic fibrosis; Crs, total respiratory compliance; FRC, functionalresidual capacity; LCI, lung clearance index; MBW, multiple breathwashout; NBS, newborn screening; PFT, pulmonary function test; pleth,plethysmographic; Rrs, total respiratory resistance; RTC, Rapid Thoraco-abdominal Compression.

Figure 2 Clinical symptoms and additional antibiotic treatment* ofnewborn screened infants with cystic fibrosis (CF) prior to lung functiontests. Although mothers of infants with CF occasionally reported mildsymptoms (slight cough or mild snuffles) in the weeks prior to LFTs, onthe day of the test all infants had clear chests on auscultation with nosign of blocked nostrils or cough. Reports of significant symptomsincluded previous wheeze, crackles, tachypnoea with or without coughand cold. *Antibiotic prescribed in addition to prophylactic flucloxacillindue to respiratory symptoms and/or positive growth on cough swab.See online supplement for standardised treatment protocol. AB,antibiotic; HI, Haemophilus influenzae; IV, intravenous; PsA, Pseudo-monas aeruginosa; SA, Staphylococcus aureus.

Cystic fibrosis


expressed as absolute values or weight-corrected ratios ispresented in table E2 of the online supplement.

Additional determinants of lung functionOn multivariable analyses, after adjustment for CF, otherpotential determinants (sex, gestational age, birth weightz-score, pre- or postnatal maternal smoking and maternalasthma) were not significantly associated with any LF z-scores.Among infants with CF, with the exception of a significantlylower FEV0.5 (mean (95% CI): �0.70 (�1.29 to �0.10) z-scores)in those who had received any additional antibiotics for symp-toms or positive cough swab, there was no significant associa-tion between LF outcomes and the infants’ genotype, clinicalstatus or treatment prior to LFTs at w3 months of age (see tableE3, OLS).

Relationship between different lung function outcomes in infantswith CFThe relationship between selected LF outcomes in infants withCF is shown in figure E2 (online supplement). There was nosignificant relationship between the two primary outcomesFEV0.5 and LCI (r¼�0.10, p¼0.432). Twenty-one percent ofinfants had an LCI above 1.96 z-scores, whereas 25% had anFEV0.5 below �1.96 z-scores, while only 12% (8/68) hadabnormalities detected by both these tests (figure E2a); if basedon either test, 35% (24/68) would be identified with abnormalresults. FEF25e75 and FEV0.5 (r¼0.73, p<0.001) detected a similarproportion of infants outside the normal range (24% and 25%,respectively, figure E2b), whereas FEF75 was less discriminative(only detecting abnormalities in 15% of infants; data notshown). FRCpleth z-score and DFRC were highly correlated

(r¼0.66, p<0.001, figure E2d), both detecting a similar propor-tion of infants with abnormally elevated results (18% and 20%,respectively). Forty-four percent (31/71) of NBS infants with CFhad at least one abnormal result if based on LCI, FEV0.5 orFRCpleth.

DISCUSSIONIn this, the largest study of its kind and the only one withcontemporaneous healthy controls, we have shown that, by 3months of age, many NBS infants with CF have reduced forcedexpired flows and volumes, abnormal gas mixing and hyperin-flation, despite early diagnosis and protocol-driven management,which included prophylactic oral flucloxacillin from time ofdiagnosis.

Strengths and weaknessesThe major strengths of this study include the fact that all lungfunction measurements were performed to international stan-dards in a single centre by a highly experienced team, therebyminimising methodological or analytical bias. Accurate identifi-cation of the extent to which abnormalities in LF were presentin individual infants was facilitated by expressing results asz-scores. These were derived from equipment-specific referenceequations based on a large group of healthy infants studied inour department with the same equipment and methods over thepast decade.31 32 The ability to recruit a large group ofcontemporaneous healthy controls specifically for this studyfurther strengthened the confidence with which we could detectchanges due to CF lung disease after adjusting for other relevantdeterminants such as body size, age and sex.34 35

Table 1 Comparison of background characteristics in infants with cystic fibrosis* and healthy controls

Infants with CF* (n[71) Healthy controls (n[54) D (95% CI), CF e controls p Value

Boy, n (%) 33 (46) 27 (50) e4% (e21% to 13%) 0.696

Gestational age, weeks 39.3 (1.5) 40.2 (1.2) e0.9 (e1.4 to e0.5) <0.001

Birth weight, z-scorey e0.34 (0.96) 0.01 (0.84) e0.35 (e0.67 to e0.03) 0.035

Birth weight below 10th percentile,y n (%) 16 (23) 2 (4) 19% (7% to 30%) 0.003

White mother, n (%) 62 (87) 47 (87) 0% (e12% to 13%) 0.962

Maternal smoking during pregnancy, n (%) 8 (11) 4 (7) 4% (e7% to 14%) 0.456

Current maternal smoking, n (%) 9 (13) 5 (9) 4% (e8% to 15%) 0.540

Mother in non-manual occupation 49 (69) 43 (80) e11% (e25% to 5%) 0.171

Fathers in non-manual occupationx 41 (59) 40 (75) e15% (e30% to 2%) 0.055

Maternal asthma, n (%) 13 (18) 10 (19) 0% (e14% to 13%) 0.976

Data shown as n (%) for categorical variables and mean (SD) for continuous variables.*Limited to those without meconium ileus.yCalculated using UK WHO algorithms.25

xn¼69 CF and n¼53 control infants.CF, cystic fibrosis; D, difference between groups.

Table 2 Infant details at time of lung function test

Infants with CF (n[71) Healthy controls (n[54) D (95% CI) CF e controls p Value

Age at test*, weeks 11.4 (2.3) 12.2 (2.0) e0.82 (e1.59 to e0.06) 0.043

Weight, kg 5.32 (0.84) 6.05 (0.78) e0.73 (e1.02 to e0.45) <0.001

Weight, z-scorey e0.75 (1.08) 0.06 (1.93) e0.81 (e1.17 to e0.45) <0.001

Length, cm 59.2 (2.7) 61.6 (2.3) e2.4 (e3.3 to e1.7) <0.001

Length, z-scorey e0.14 (1.04) 0.72 (0.86) e0.86 (e1.19 to e0.52) <0.001

Body mass index 15.1 (1.6) 15.9 (1.4) e0.8 (e1.3 to e0.3) 0.005

Body mass index, z-scorey e0.94 (1.08) e0.47 (0.98) e0.47 (e0.84 to e0.10) 0.013

Change in weight z-score (3 month e birth) e0.42 (1.06) 0.05 (1.00) e0.46 (e0.83 to e0.10) 0.014

Unless stated otherwise, data presented as mean (SD).*Postnatal age at test corrected for gestational age.yCalculated using UK WHO algorithms.25

CF, cystic fibrosis; D, mean difference between groups.

Cystic fibrosis


Inevitably, it was necessary to approach a large number offamilies of healthy infants to recruit appropriate numbers forcomparison, with only 25% of those eligible and whom we couldcontact actually attending for the tests. This could potentiallyintroduce some bias, especially as due to data protection issues,it was not possible to document background data from thosenot responding or agreeing to participate. However, it wasreassuring to find no significant differences with respect topotential determinants of infant LF such as ethnicity, maternalhistory of asthma, socioeconomic status (based on parentaloccupation) or exposure to tobacco smoke between the controlsand infants with CF, suggesting that the group were represen-tative of the local population. Furthermore, when expressed asz-scores, both anthropometry and LF were very similar in thiscurrent healthy control group as in those previously recruited inLondon by our team.15 20 Travel expenses were reimbursed, butno incentives were provided to encourage attendance for eithergroup, thereby removing a potential source of bias. Infants withCF were, on average, born 1 week earlier than the controls,a pattern that we have noted in previous studies.11 Anypotential group differences due to developmental changes wereavoided by studying all infants within a narrow age range; meandifference at time of test being less than a week.

Of the 110 infants screening positive for CF over the 2.5-yearrecruitment period, 12% were excluded due to coexistentmorbidity or psychosocial reasons, with tests not possible ina further five infants before 4 months of age due to recentexacerbations and deferment of appointments. If anything, theresults from this study therefore underestimate the degree ofmorbidity by 3 months of age. Nevertheless, with only 13% ofparents declining participation (including one family thatwithdrew after initial consent), we were able to test 79/110(72%) of the entire cohort and 81% of those eligible.

Since lung function was assessed in the first few months oflife, we excluded results from those presenting with meconium

ileus to preclude any adverse effects related to surgery, althoughsubsequent sub-analysis indicated that the results were notaffected by including such children. All infants were tested atleast 3 weeks after a respiratory illness when free of symptoms,in an attempt to minimise impact of acute exacerbations. Thisdid mean that we had to exclude several healthy controls inwhom it was impossible to rearrange appointments within thedesignated age range for this study of ‘early LFT’ (figure E1,online supplement). While it could be argued that the observeddifferences in LF between CF and controls may simply representpost-respiratory tract infection changes in those with CF, this isunlikely given the lack of association between LF and respiratorysymptoms in those with CF, except for a significant reduction inz-FEV0.5 when infants had received additional antibiotics (tableE3, online supplement).To optimise recruitment to this observational study and to

ensure initial LFTs could be undertaken within the first monthsof life, routine bronchoscopy and bronchoalveolar lavage werenot included within this study protocol but, together with chesthigh-resolution CT scans, will be included when these childrenare reassessed at 1 year of age.A further strength of this study was the wide range of tests

applied, minimising the chance that early changes in LF wouldbe missed. As in a previous study of infants,20 but in contrast tofindings in older children,13 16 17 36 we found that assessments oflung disease based on forced expiratory manoeuvres and venti-lation inhomogeneity were necessary to identify early lungdisease during the first year of life, with additional usefulinformation regarding hyperinflation being obtained fromplethysmographic lung volumes. The slightly lower absolutemean LCI observed in infants with CF in this study (table E2,online supplement) than in our previous study20 probablyreflects the fact that these infants were diagnosed by NBS ratherthan clinically, and were therefore assessed earlier in the diseaseprocess. Had assessments been limited to a single technique,

Table 3 Comparison of lung function in 71 infants with cystic fibrosis and 54 healthy controls

Variables according to test order n Infants with CF n Healthy controls D (95% CI) CF e controls p Values

Multiple breath washout

LCI, z-score* 70 0.90 (1.38) 51 0.39 (0.88) 0.51 (0.10 to 0.91) 0.016

FRCMBW, z-score* 70 0.15 (0.95) 51 �0.23 (0.86) 0.38 (0.05 to 0.71) 0.023

Tidal breathing

Respiratory rate, z-scorey 71 �0.01 (1.43) 54 0.02 (1.21) �0.03 (�0.50 to 0.44) 0.897

Tidal volume, z-scorey 71 0.20 (1.19) 54 �0.22 (1.09) 0.42 (0.01 to 0.82) 0.044

tPTEF/tE, z-scorey 71 0.22 (1.05) 54 0.41 (1.0) �0.19 (�0.56 to 0.18) 0.312

Passive mechanics

Crs, z-scorey 47 0.22 (1.48) 41 0.14 (0.96) 0.08 (�0.44 to 0.60) 0.763

Rrs, z-scorey 47 0.52 (0.94) 41 0.28 (1.04) 0.24 (�0.18 to 0.66) 0.237

Plethysmography

FRCpleth, z-scorey 56 0.77 (1.15) 47 �0.08 (1.03) 0.85 (0.43 to 1.28) <0.001

DFRC z-scores (pleth � MBW) 55 0.68 (1.10) 45 0.19 (0.92) 0.48 (0.08 to 0.88) 0.018

Raised volume technique

FEV0.5, z-scorez 68 �1.17 (1.20) 52 �0.25 (0.80) �0.92 (�1.29 to �0.56) <0.001

FVC, z-scorez 68 �0.45 (1.07) 52 0.15 (0.71) �0.60 (�0.92 to �0.28) <0.001

FEF25e75, z-scorez 68 �1.11 (1.38) 52 �0.46 (1.08) �0.66 (�1.10 to �0.21) 0.004

FEF75, z-scorez 68 �0.74 (1.19) 52 �0.22 (1.11) �0.53 (�0.95 to �0.11) 0.015

For comparison with previous publications, results expressed as absolute values, or in relation to body weight, are presented in table E2, online supplement.Data shown as mean (SD).*According to Lum et al.31

yAccording to Nguyen et al.32

zAccording to Lum et al.34

CF, cystic fibrosis; CI, confidence interval of the difference; D, mean difference between groups.CF, cystic fibrosis; Crs, total respiratory compliance; FEF75, forced expiratory flow at 75% of vital capacity; FEF25-75, forced expiratory flow between 25% and 75% of vital capacity; FEV0.5,forced expiratory volume in 0.5 s; FRC, functional residual capacity; FVC, forced vital capacity; LCI, lung clearance index; MBW, multiple breath washout; pleth, plethysmographic; Rrs, totalrespiratory resistance; tPTEF/tE, ratio of the time to reach peak tidal expiratory flow in relation to expiratory time.

Cystic fibrosis


abnormalities would only have been detected in w25% ofinfants with CF, this proportion rising to 35% when using eitherof the two primary outcomes (LCI and FEV0.5). The relativelypoor correlation between these various outcomes suggests thatthey are reflecting different aspects of underlying pathophysi-ology. Despite their relative simplicity, assessments of tidalbreathing variables and passive mechanics were far lessdiscriminatory than the other techniques and could be usefullyomitted when investigating infants with CF.37 While Crs resultswere lower and Rrs higher in those with CF when expressed asabsolute values (table E2, online supplement), this was largelyrelated to body size at the time of testing; Crs increasing and Rrs

decreasing with somatic growth.32 38 In contrast to all otheroutcome measures investigated, after adjusting for length and/orage, no significant differences were observed with respect toeither Crs or Rrs (table 3).

Clinical statusDespite early diagnosis and commencement of pancreaticenzyme replacement therapy, vitamin supplement and prophy-lactic antibiotics, infants with CF experienced significantly

slower growth during the first few months and were signifi-cantly lighter and shorter than their healthy peers by the time ofthe 3-month LFTs. Furthermore, by this age, 61% of the screenedinfants had had some respiratory symptoms (52% mild, 9%severe), 23% a positive cough swab and 73% had received anti-biotics in addition to their routine prophylactic medication.Pulmonary involvement is known to be present early, with someinfants with CF having evidence of inflammation in the bron-choalveolar lavage fluid as early as 4 weeks of age.39e41 With theexception of a significantly lower FEV0.5 in those who hadreceived additional antibiotics for symptoms or positive coughswab, there was no significant association between LF outcomesand the infant’s genotype, clinical status, growth trajectory ortreatment prior to the LFTs at 3 months of age. Of note, manyinfants who had been treated aggressively for respiratory exac-erbations in the first few months had entirely normal LF by3 months, whereas others with no prior symptoms or cause forconcern had evidence of early lung disease.

Comparison with the literatureThe commonest characteristics of LF abnormalities described inCF lung disease during the first years of life have been airwayobstruction detected using the raised volume technique,10e12 15 20

hyperinflation indicated by elevated resting lung volumes,42

increased ventilation inhomogeneity20 and, in infants andslightly older children, gas trapping,17 18 43 44 all of which wereobserved in this study by 3 months of age in infants diagnosedby NBS.In a US multicentre evaluation of LF in infants with CF aged

4e24 months (21% diagnosed by screening),42 elevated lungvolumes and diminished forced expiratory flows, but no reduc-tion in FEV0.5, were reported when compared with historicalcontrols. The variability in skill mix and experience of thelaboratories participating in that study, together with the lack ofcontemporaneous healthy controls and different age rangestudied, may have contributed to differences in findingscompared with the current results.35

The Australian AREST-CF study recently published LF resultsfrom infants with CF diagnosed by NBS.9 10 When using a lunginflation pressure of 20 cmH2O during forced expiratorymanoeuvres, LF was reported to be normal during the first6 months of life, but thereafter declined at a rapid rate.9

However, the number of infants studied during the first monthsof life was limited. By contrast, a more recent publication fromthis group,9 10 reporting measurements from 28 NBS infantswith CF within the first 6 months of life when using an inflationpressure of 30 cmH2O as recommended by the AmericanThoracic Society/European Respiratory Society guidelines19

showed diminished LF at the time of the first test, withcontinued deterioration over the next 2 years of life when theresults were compared with published reference equations.45

Clinical implicationsThe results from this study indicate that despite early diagnosisand rapid implementation of therapy, including prophylacticantibiotics, a substantial number of NBS infants with CF haveabnormalities of LF within the first 3 months of life. Theapparent wellness of the cohort should not lead to complacency,and prompt and aggressive treatment of any abnormal symp-toms or signs is surely vital. Follow-up of this cohort will beessential to ascertain the extent to which early changes in LFpersist throughout the first year of life; if there is catch-upgrowth with conventional treatment, then novel, molecular-

Figure 3 Comparison of lung function outcomes between infants withcystic fibrosis and healthy controls. As expected, approximately 95% ofhealthy controls had values within 61.96 z-scores, with none having anabnormal FEV0.5 z-score, 2/51 (4%) having an elevated LCI z-score, and2/47 (4%) an elevated FRCpleth z-score. FEF25-75, forced expiratory flowbetween 25% and 75% of vital capacity; FEV0.5, forced expiratory volumein 0.5 s; FRCpleth, plethysmographic functional residual capacity; LCI,lung clearance index; MBW, multiple breath washout; Rrs, totalrespiratory resistance.

Cystic fibrosis


based therapies46 47 may be safely deferred; if not, there will bea compelling case for initiating treatment early in these infants,using physiological endpoints to detect benefit.

CONCLUSIONSDespite early diagnosis of CF by NBS and protocol-driventreatment in specialist centres, abnormal LF, with increasedventilation inhomogeneity and hyperinflation and diminishedairway function, is evident in many infants with CF diagnosedthrough NBS by 3 months of age. CF clinicians should not belulled into thinking that babies with CF identified by NBS havegood pulmonary health in the first few months of life.

Acknowledgements We thank the infants and parents who participated in thisstudy and contributions by members of the London CF Collaboration (including JanetStocks, Andy Bush, John Price, Colin Wallis, Ranjan Suri, Paul Aurora, Ian Balfour-Lynn, Siobhan Carr, Caroline Pao, Hilary Wyatt, Gary Ruiz, Richard Chavesse, AnuShanker, Wanda Kozlowska, Ah-Fong Hoo, Sooky Lum, The Thanh Diem Nguyen,Lena Thia, Jane Chudleigh, and Ammani Prasad). We also thank Per Gustafsson foradvice and assistance regarding the multiple breath inert gas washout technique.

Contributors Conception and design of the study: JS, AB. Supervision of the study:JS. Research governance issues including ethics committee approval: JS, JC. Settingup of recruitment process: AFH, JC. Technical training, supervision and audit of datacollection/analyses: AFH. Recruitment of infants with CF: AB, IBL, SBC, HAW, JP, RJC,AS. Recruitment of healthy control infants: KLC, AFH, DA, LPT, TDN, JC. Datacollection/analysis and interpretation of results: JS, AFH, LPT, TDN, JC, SL. Statisticalanalyses: AFH, JS, AW. Drafting the manuscript: AFH, JS, LPT, IBL, AB, CW. Approvalfor intellectual content: all authors.

Funding This study is supported by grants from the Cystic Fibrosis Trust, UK; SpecialTrustees: Great Ormond Street Hospital for Children, London, UK; Smiths Medical Ltd,UK; Comprehensive Local Research Network, UK. It was also supported by the NIHRRespiratory Disease Biomedical Research Unit at the Royal Brompton and HarefieldNHS Foundation Trust and Imperial College London.



Ethics approval Ethics approval was granted by the North Thames Multi-centreResearch Ethics Committee (REC) (#09/HO71/314) and local REC of the participatingspecialist centres.


REFERENCES1. Farrell PM, Lai HJ, Li Z, et al. Evidence on improved outcomes with early diagnosis

of cystic fibrosis through neonatal screening: enough is enough! J Pediatr 2005;147:S30e6.

2. Koscik RL, Farrell PM, Kosorok MR, et al. Cognitive function of children with cysticfibrosis: deleterious effect of early malnutrition. Pediatrics 2004;113:1549e58.

3. Chatfield S, Owen G, Ryley HC, et al. Neonatal screening for cystic fibrosis in Walesand the West Midlands: clinical assessment after five years of screening. Arch DisChild 1991;66:29e33.

4. Siret D, Bretaudeau G, Branger B, et al. Comparing the clinical evolution of cysticfibrosis screened neonatally to that of cystic fibrosis diagnosed from clinicalsymptoms: a 10-year retrospective study in a French region (Brittany). PediatrPulmonol 2003;35:342e9.

5. Dankert-Roelse JE, te Meerman GJ, Martijn A, et al. Survival and clinical outcomein patients with cystic fibrosis, with or without neonatal screening. J Pediatr1989;114:362e7.

6. Dankert-Roelse JE, Merelle ME. Review of outcomes of neonatal screening forcystic fibrosis versus non-screening in Europe. J Pediatr 2005;147:S15e20.

7. Merelle ME, Schouten JP, Gerritsen J, et al. Influence of neonatal screening andcentralized treatment on long-term clinical outcome and survival of CF patients. EurRespir J 2001;18:306e15.

8. Dijk FN, McKay K, Barzi F, et al. Improved survival in cystic fibrosis patientsdiagnosed by newborn screening compared to a historical cohort from the samecentre. Arch Dis Child 2011;96:1118e23.

9. Linnane BM, Hall GL, Nolan G, et al. Lung function in infants with cysticfibrosis diagnosed by newborn screening. Am J Respir Crit Care Med2008;178:1238e44.

10. Pillarisetti N, Williamson E, Linnane B, et al. Infection, inflammation and lungfunction decline in infants with cystic fibrosis. Am J Respir Crit Care Med2011;184:75e81.

11. Ranganathan S, Dezateux CA, Bush A, et al. Airway function in infants newlydiagnosed with cystic fibrosis. Lancet 2001;358:1964e5.

12. Ranganathan SC, Bush A, Dezateux C, et al. Relative ability of full and partial forcedexpiratory maneuvers to identify diminished airway function in infants with cysticfibrosis. Am J Respir Crit Care Med 2002;166:1350e7.

13. Aurora P, Stanojevic S, Wade A, et al. Lung clearance index at 4 years predictssubsequent lung function in children with cystic fibrosis. Am J Respir Crit Care Med2011;183:752e8.

14. Kozlowska WJ, Bush A, Wade A, et al. Lung function from infancy to the preschoolyears after clinical diagnosis of cystic fibrosis. Am J Respir Crit Care Med2008;178:42e9.

15. Ranganathan SC, Stocks J, Dezateux C, et al. The evolution of airway function inearly childhood following clinical diagnosis of cystic fibrosis. Am J Respir Crit CareMed 2004;169:928e33.

16. Aurora P, Bush A, Gustafsson P, et al. Multiple-breath washout as a marker of lungdisease in preschool children with cystic fibrosis. Am J Respir Crit Care Med2005;171:249e56.

17. Gustafsson PM, Aurora P, Lindblad A. Evaluation of ventilation maldistribution as anearly indicator of lung disease in children with cystic fibrosis. Eur Respir J2003;22:972e9.

18. Kraemer R, Blum A, Schibler A, et al. Ventilation inhomogeneities in relation tostandard lung function in patients with cystic fibrosis. Am J Respir Crit Care Med2005;171:371e8.

19. American Thoracic Society, European Respiratory Society. ATS/ERS statement:Raised volume forced expirations in infants: guidelines for current practice. Am JRespir Crit Care Med 2005;172:1463e71.

20. Lum S, Gustafsson P, Ljungberg H, et al. Early detection of cystic fibrosis lungdisease: multiple-breath washout versus raised volume tests. Thorax2007;62:341e7.

21. Borowitz D, Robinson KA, Rosenfeld M, et al. Cystic Fibrosis Foundationevidence-based guidelines for management of infants with cystic fibrosis. J Pediatr2009;155:S73e93.

22. Balfour-Lynn IM. Newborn screening for cystic fibrosis: evidence for benefit. ArchDis Child 2008;93:7e10.

23. Report of the UK Cystic Fibrosis Trust Antibiotic Working Group. AntibioticTreatment for Cystic Fibrosis. 3rd edn. Cystic Fibrosis Trust UK. 2009. http://www.cftrust.org.uk/aboutcf/publications/consensusdoc/Antibiotic_treatment_for_Cystic_Fibrosis.pdf (accessed Feb 2012).

24. Equi AC, Pike SE, Davies J, et al. Use of cough swabs in a cystic fibrosis clinic. ArchDis Child 2001;85:438e9.

25. Cole TJ, Wright CM, Williams AF. Designing the new UK-WHO growth charts toenhance assessment of growth around birth. Arch Dis Child Fetal Neonatal Ed2012;97:F219e22.

26. Stocks J, Godfrey S, Beardsmore C, et al. Standards for infant respiratory functiontesting: plethysmographic measurements of lung volume and airway resistance. EurRespir J 2001;17:302e12.

27. Stocks J, Marchal F, Kraemer R, et al. Plethysmographic assessment offunctional residual capacity and airway resistance. In: Stocks J, Sly PD, Tepper RS,et al, eds. Infant Respiratory Function Testing. New York: John Wiley & Sons,1996:191e240.

28. Goetz I, Hoo A-F, Lum S, et al. Assessment of passive respiratory mechanics ininfants: double versus single occlusion? Eur Respir J 2001;17:449e55.

29. Aurora P, Kozlowska W, Stocks J. Gas mixing efficiency from birth to adulthoodmeasured by multiple-breath washout. Respir Physiol Neurobiol 2005;148:125e39.

30. Ranganathan SC, Hoo AF, Lum SY, et al. Exploring the relationship between forcedmaximal flow at functional residual capacity and parameters of forced expiration fromraised lung volume in healthy infants. Pediatr Pulmonol 2002;33:419e28.

31. Lum S, Sonnappa S, Gustafsson PM, et al. Lung growth and ventilationinhomogeneity in health. Eur Respir J 2011;38:351s.

32. Nguyen TTD, Hoo A-F, Lum S, et al. New reference equations to improveinterpretation of infant lung function. Pediatr Pulmonol. In press.

33. Cohen J. Statistical Power Analysis for the Behavioural Sciences. 2nd edn. Hillsdale,New Jersey: Erlbaum Associates, 1988.

34. Lum S, Hoo AF, Hulskamp G, et al. Potential misinterpretation of infant lung functionunless prospective healthy controls are studied. Pediatr Pulmonol 2010;45:906e13.

35. Stocks J, Modi N, Tepper R. Need for healthy controls when assessing lung functionin infants with respiratory disease. Am J Respir Crit Care Med 2010;182:1340e2.

36. Gustafsson PM. Inert gas washout in preschool children. Paediatr Respir Rev2005;6:239e45.

37. Ranganathan S, Goetz I, Hoo AF, et al. Assessment of tidal breathing parameters ininfants with cystic fibrosis. Eur Respir J 2003;22:761e6.

38. Stocks J, Lum S. Pulmonary function tests in infants and preschool children. In:Chernick V, Bush A, eds. Kendig’s disorders of the respiratory tract in children.Philadelphia, PA, USA: Elsevier, 2012:169e210.

39. Armstrong DS, Grimwood K, Carlin JB, et al. Lower airway inflammation in infantsand young children with cystic fibrosis. Am J Respir Crit Care Med1997;156:1197e204.

40. Armstrong DS, Hook SM, Jamsen KM, et al. Lower airway inflammation in infantswith cystic fibrosis detected by newborn screening. Pediatr Pulmonol2005;40:500e10.

41. Khan TZ, Wagener JS, Bost T, et al. Early pulmonary inflammation in infants withcystic fibrosis. Am J Respir Crit Care Med 1995;151:1075e82.

Cystic fibrosis


42. Davis SD, Rosenfeld M, Kerby GS, et al. Multicenter evaluation of infant lungfunction tests as cystic fibrosis clinical trial endpoints. Am J Respir Crit Care Med2010;182:1387e97.

43. Gustafsson PM, Kallman S, Ljungberg H, et al. Method for assessment of volume oftrapped gas in infants during multiple-breath inert gas washout. Pediatr Pulmonol2003;35:42e9.

44. Gustafsson PM, de Jong PA, Tiddens HA, et al. Multiple-breath inert gas washoutand spirometry versus structural lung disease in cystic fibrosis. Thorax2008;63:129e34.

45. Jones M, Castile R, Davis S, et al. Forced expiratory flows and volumes ininfants: normative data and lung growth. Am J Respir Crit Care Med2000;161:353e9.

46. Kerem E, Hirawat S, Armoni S, et al. Effectiveness of PTC124 treatment of cysticfibrosis caused by nonsense mutations: a prospective phase II trial. Lancet2008;372:719e27.

47. Ramsey BW, Davies J, McElvaney NG, et al. A CFTR potentiator inpatients with cystic fibrosis and the G551D mutation. N Engl J Med2011;365:1663e72.


Cystic fibrosis

881

Journal club

Restoration of function of the ΔF508 mutation in cysticfibrosisThe first nucleotide binding domain (NBD1) of the of ΔF508 cystic fibrosis transmembraneconductance regulator (CFTR), is considered a potential drug target in cystic fibrosis but thelinks between this and CFTR misfolding remain unclear. NBD1 is one of the cytosolicdomains of CFTR and is found in at least one allele of 90% of cystic fibrosis patients,significantly diminishing the folding efficiency of CFTR. The aim of this study was to clarifythe role of NBD1 and structural consequences of the ΔF508 mutation in CFTR assembly.The isolated domains of thermodynamic and kinetic destabilisation from isolated NBD1

variants by the ΔF508 mutation were studied along with conformational stabilisation ofΔF508 NBD1. Thermodynamic and kinetic destabilisations in combination were shown to beresponsible for the ΔF508 NBD1 misfolding. In full-length CFTR, the stability of NBD1 andNBD1-CL interface is needed for normal channel function. As ΔF508 causes impairment ofboth these domains, energetic and interface defects (NBD1-MSD2) need to be correctedsimultaneously in ΔF508 CFTR for wild-type function.This study explains the limitations of some corrector molecules when targeted against

a particular structural abnormality in CFTR. The understanding of interface mutation in suchcomplicated multi-domain membrane proteins may lead to structure based combinationcorrector therapies.

< Rabeh WM, Bossard F, Xu H, et al. Correction of both NBD1 energetics and domain interface is required to restore ΔF508 CFTRfolding and function. Cell 2012;148:150e63.

Stephen Rowan

Correspondence to Dr Stephen Rowan, Belfast City Hospital, Belfast, UK; [email protected]





Progress in cystic fibrosis and the CF TherapeuticsDevelopment NetworkSteven M Rowe,1 Drucy S Borowitz,2 Jane L Burns,3 John P Clancy,4

Scott H Donaldson,5 George Retsch-Bogart,5 Scott D Sagel,6

Bonnie W Ramsey3

▸ Additional supplementaryfigures are published onlineonly. To view these files pleasevisit the journal online (http://dx.doi.org/10.1136/thoraxjnl-2012-202550)1Department of Medicine,University of Alabama atBirmingham, Birmingham,Alabama, USA2Department of Pediatrics,State University of New York atBuffalo, Buffalo, New York, USA3Seattle Children’s Hospital andthe University of Washington,Seattle, Washington, USA4Department of Pediatrics,Cincinnati Children’s HospitalMedical Center, and theUniversity of Cincinnati,Cincinnati, Ohio, USA5University of North Carolina atChapel Hill, Chapel Hill, NorthCarolina, USA6Department of Pediatrics,Children’s Hospital Coloradoand University of Colorado,Aurora, Colorado, USA

Correspondence toDr Steven M Rowe, Universityof Alabama at Birmingham,1819 University Boulevard(MCLM 768), Birmingham,AL 35294, USA;[email protected]

Received 9 August 2012Accepted 17 August 2012

ABSTRACTCystic fibrosis (CF), the most common life-shorteninggenetic disorder in Caucasians, affects approximately70 000 individuals worldwide. In 1998, the Cystic FibrosisFoundation (CFF) launched the CF TherapeuticsDevelopment Network (CF-TDN) as a central element ofits Therapeutics Development Programme. Designed toaccelerate the clinical evaluation of new therapiesneeded to fulfil the CFF mission to control and cure CF,the CF-TDN has conducted 75 clinical trials since itsinception, and has contributed to studies as varied asinitial safety and proof of concept trials to pivotalprogrammes required for regulatory approval. This reviewhighlights recent and significant research efforts of theCF-TDN, including a summary of contributions to studiesinvolving CF transmembrane conductance regulator(CFTR) modulators, airway surface liquid hydrators andmucus modifiers, anti-infectives, anti-inflammatories, andnutritional therapies. Efforts to advance CF biomarkers,necessary to accelerate the therapeutic goals of thenetwork, are also summarised.

STRUCTURE OF THE NETWORK, CFFPARTNERSHIP, INDUSTRY PARTNERSHIPS,OPERATIONAL ISSUES THAT FOSTER SUCCESSCystic fibrosis (CF), the most common life-shortening genetic disorder in Caucasians, affectsapproximately 70 000 individuals worldwide andmeets the US Food and Drug Administration (FDA)definition of an orphan disease.1 Historically, thelimited number of patients at each clinical site ledto underpowered research studies2 and was a criticalbarrier to development and regulatory approval ofnew therapies. Hence, the Cystic FibrosisFoundation (CFF) launched the CF TherapeuticsDevelopment Network (CF-TDN) in 1998 to accel-erate the clinical evaluation of new CF therapies.3 4

From its inception, the CF-TDN was organised toprovide necessary infrastructure to promote effi-cient study conduct. In addition, patient safety hasalways been the highest priority.3 From the begin-ning, the programme was composed of TherapeuticDevelopment Centers (TDCs) across the USA, aclinical and data coordinating centre located atSeattle Children’s Research Institute, and a numberof resource centres with expertise in interpretationor performance of specialised outcomes measures(microbiology, imaging, infant lung functiontesting, inflammatory mediators, sweat testing, CFtransmembrane conductance regulator (CFTR)detection and cytology interpretation) (see onlinesupplementary figure S1). To maintain scientific

integrity and safety of CF-TDN studies, there is aSteering Committee, protocol review of all multi-center trials, and review of presentations and publi-cations from CF-TDN studies (see onlinesupplementary figure S2). The CFF-supported DataSafety Monitoring Board located at the Universityof Arizona has oversight of clinical trials at allcentres. The TDCs are selected by a competitive,peer-reviewed application process from the 120 USCF care centres4 5 and receive infrastructure supportfor a research coordinator and other necessary staff.As a result, TDCs have been able to retain welltrained, experienced research staff teamed with siteinvestigators. Over the past 15 years, the CF-TDNhas successfully completed over 75 clinical trialsalong the entire developmental path from phase Ito phase III registration trials. The goal to bringnew therapies to the CF community has beenachieved with FDA approval of several therapies(table 1).6 A key factor in the success of theCF-TDN has been its partnership with industry,providing a full range of consultative services fromthe earliest discussions about protocol developmentthrough critical protocol review to feedback on theindustry/contract research organisation experiencefrom the study site perspective. Above all, industrysponsors value the CF-TDN’s experience in theconduct of CF clinical research in the USA.The coordinated efforts of the CF-TDN also

include the development of novel outcome measures(below), use of banked specimens and data toconduct ancillary studies,7 8 and development ofnovel approaches to study design and analyses.9 Inaddition, the network has focused on a qualityimprovement (QI) initiative based upon ground-breaking work of the CF care centre programme.5 10

The need for improved communications and a QIprogramme became essential when the CF-TDNexpanded from 8 to 77 TDC sites from 1998 to2009, encompassing a population of 19 000 patients.Communication was improved using quarterlynewsletters, conference calls, annual in-person meet-ings, and the creation of the CF-TDN website, CFClinical Research Net, which tracks site study activ-ity, provides tools for site training and programmeenhancement, hosts committee work, posts newsand information regarding studies and network pro-grammes, and provides access to the network data-base. A commitment to QI for all aspects of clinicalstudies has paralleled the growth in network size.Here we summarise recent progress of the networkacross key therapeutic areas (see CF-TDN Pipelinedepicted on the front cover).


Cystic fibrosis

http://dx.doi.org/10.1136/thoraxjnl-2012-202550




CFTR MODULATORSCFTR modulators have brought a personalised approach toaltering the basic defect in the disease using gene class specifictherapy (see reviews11–15). By grouping CFTR mutations intocausative classes, specific approaches can be developed thataddress the underlying molecular defect on an individualpatient level (figure 1), and serves as a paradigm for the treat-ment of other genetic diseases.16 The evaluation of these novelagents required expertise in proof-of-concept clinical trialdesign, biomarkers of pharmacological activity, and clinicaloutcome measures in an entirely new therapeutic area, andwere promoted by close partnerships between the pharmaceut-ical industry and the CFF. Recent results have firmly establishedthat rescue of CFTR-mediated anion transport can result inmarked clinical benefit if CFTR functional improvement is suf-ficient,6 a concept likely to be a major contributor to CF-TDNefforts in the future.

CFTR modulator therapies have been directed towardsspecific disease-causing mutations and the molecular pathwaysthat underlie their cause.12 16 A crucial phase II trial that testedthe CFTR potentiator ivacaftor (Kalydeco, VX-770, VertexPharmaceuticals, Boston, MA, USA) in 40 patients with CFwith at least one copy of the G551D mutation, a relativelycommon class 3 gating mutant, in a two-part randomised

placebo-controlled design was the first to definitively establishthat CFTR functional rescue detected by simultaneous improve-ment in nasal potential difference (NPD) and sweat chloridecould result in meaningful changes in lung function.17 This wasrapidly followed by two phase III trials in patients with G551DCF to establish the clinical effect of ivacaftor during a sustainedperiod. In a trial in older children and adults (aged 12 andabove), forced expiratory volume in 1 s (FEV1) improved 10.5%at 24 weeks, an effect durable at 48 weeks and accompanied by a55% reduction in the probability of experiencing a pulmonaryexacerbation, a 3.1 kg weight gain (compared with 0.9 kg in theplacebo group), and improved respiratory symptoms as assessedby the CFQ-R, a patient reported quality-of-life index.6 As withphase II testing, sweat chloride improved by around 48 mEq/litre, and mean concentration was around 55 mEq/litre, a levelbelow the traditional diagnostic threshold of CF (60 mEq/litre).Similar results were reported in a study of patients with G551Daged 6–12,18 enabling prompt approval of the drug for patientswith G551D CF aged 6 and above by the FDA and Europeanregulatory authorities. The CF-TDN will support additionalstudies to evaluate the safety and pharmacokinetics of ivacaftorin younger patients with CF (aged 2–5), which could also beginto assess the effects of early CFTR rescue on the pancreas.The rapid approval and availability of ivacaftor has also enabledthe G551D observational study (GOAL), which will assessthe clinical effects of ivacaftor in patients with G551D CFpost-approval, including the response of several mechanistic bio-markers such as sweat rate, mucociliary clearance imaging,gastrointestinal pH, and sputum measures of inflammation andmicrobiology; these studies are intended to advance our under-standing of the effects of efficacious CFTR modulation ondisease-relevant biomarkers and pathways. Since ivacaftor alsoexhibits robust activity in other rare but clinically relevantCFTR gating mutations, the drug is also being studied inpatients with CFTR gating mutations other than G551D.Ivacaftor will also be tested in the archetype conductance muta-tion R117H, which could establish whether potentiation ofCFTR gating is sufficient to partially ameliorate non-gatingmutations, setting the stage for other studies involving raremutations localised to the cell surface.

Table 1 CF therapeutics evaluated by the TDN and approved by theFDA

Medication Therapeutic areaFDAapproval

Pancrelipase Pancreatic enzyme replacementtherapy

(Creon, Zenpep) 2009(Pancreaze) 2010(Pertzye) 2012(Ultresa) 2012

Aztreonam lysate for inhalation(Cayston)

Inhaled antibiotic 2010

Ivacaftor (Kalydeco) CFTR potentiator 2012

CF, cystic fibrosis; TDN, Therapeutics Development Network.

Figure 1 Cystic fibrosistransmembrane conductance regulatorgene mutation classes and therapeuticapproaches under study by the cysticfibrosis Therapeutics DevelopmentNetwork (CF-TDN). Note: approachesto class V (splicing mutations that altersurface CFTR levels), and class VI(mutations in the PDZ binding domainthat decrease surface stability and/oralter recycling) CFTR mutations areunder development, but have not yetentered clinical testing through theTDN.


Cystic fibrosis

Since ivacaftor alone is not sufficient to meaningfully alterCFTR activity in patients with CF homozygous for F508del,19

a class 2 mutant and the most common CFTR mutation,‘correctors’ of CFTR misfolding are required to augment localisa-tion of the F508del CFTR mutation to the cell surface. TheCFTR correctors lumacaftor (VX-809, Vertex Pharmaceuticals,Cambridge, Massachusetts, USA) and VX-661 (VertexPharmaceuticals) have entered clinical testing within theCF-TDN for this purpose. While monotherapy with lumacaftorin adults homozygous for the F508del mutation only modestlyreduced sweat chloride (8 mEq effect vs placebo at the highestdose tested (200 mg)) and did not improve lung function orNPD, evidence of sweat chloride improvement, albeit small inmagnitude, suggested correction of the F508del CFTR proteinwas possible.20 To improve upon the effects of a corrector alone,lumacaftor is presently being tested at higher doses in a three-part study alone and in combination with ivacaftor, therebyevaluating the effects of augmenting CFTR channel gating fol-lowing partial restoration of CFTR to the cell surface, a strategyjustified by preclinical research21 22 and the known effects of theF508 deletion on gating in addition to cellular processing.23

Interim results through part 2 of the study have indicated thathigh doses of lumacaftor combined with ivacaftor improvesFEV1 by 7.4%, and have recently provided the basis for a pivotalprogramme to evaluate combination therapy in patients withCF homozygous for F508del that will necessitate involvementin a many sites, including those beyond the network infrastruc-ture. TDCs are also involved in testing the CFTR correctorVX-661 alone and in combination with ivacaftor as an alternateCFTR corrector to lumacaftor. In total, the combination ofCFTR potentiators and correctors could result in effective ther-apies in a large proportion of patients with CF (table 2).

The CFTR modulator class also includes treatment strategiesdirected towards premature termination codons (class 1 muta-tions). Following the discovery that aminoglycoside antibioticscan induce readthrough premature termination codons resultingin full-length functional protein,24 an initial TDN networkstudy focused on topical administration of aminoglycosides25;while bioactivity was not detectable and conflicted with othersingle-centre studies,26–28 this trial provided experience evaluat-ing CFTR modulators helpful for subsequent studies, includingthose evaluating the small molecule ataluren (PTC124, prema-ture termination codon therapeutics). Based on efficacy invitro29 and in animal models,30 ataluren was studied in a seriesof conflicting phase II trials,31–33 one of which involved the

TDN network.32 Results of a long-term study showed noimprovement in FEV1 % predicted, the primary endpoint, butdid demonstrate a small effect on lung function in a predefinedsubset of individuals who were not treated with inhaled anti-biotics, which can alter the efficacy of ataluren.34 While theclinical status of ataluren remains uncertain at present, giventhe strong foundation of translational readthrough as a thera-peutic strategy for CF and many other genetic diseases, studiesinvolving this compound and other readthrough approacheswithin the CF-TDN are likely in the future.

AIRWAY SURFACE LIQUID DEPTH AND MUCUSMODIFICATIONThe rationale supporting the development of therapeutics tar-geting airway surface liquid (ASL) volume and mucus proper-ties comes from basic research on the pathogenesis of CF lungdisease.35 Numerous experimental models support the centralrole of CFTR as both a chloride channel and as a regulator ofother ion channels in airway epithelia.36 37 Disruption of CFTRactivity, in turn, has been linked to the development of defect-ive mucus clearance via depletion of ASL volume and mucusdehydration.38 These and perhaps other distinct effects onairways defence may ultimately explain the initial, profoundvulnerability to chronic bacterial airways infection that existsin CF, while also contributing to disease progression.

Approaches to increase fluid secretion and/or reduce fluidreabsorption to augment impaired mucociliary clearance in CFrepresent a major line of therapeutics that address fundamentalphysiological deficits in CF, and are not restricted to specificCFTR mutations nor the uncertainties of gene transfer. Amultifaceted approach via collaborations between scientists,industry, the CF Foundation, and the Therapeutic DevelopmentNetwork, utilising different therapeutic targets, has beenpursued to maximise the chance of success. These alternativeASL restoring therapeutics can be grouped into those thatpromote increases in ASL volume and mucus hydration viaosmotic mechanisms (eg, hypertonic saline; Bronchitol,Pharmaxis, Frenchs Forest, Australia), by reducing fluid reabsorp-tion from the airway lumen through inhibition of the epithelialsodium channel (eg, GS9411, Gilead Sciences, Foster City,California, USA) or by increasing chloride secretion via variousnon-CFTR pathways (eg, denufosol, Inspire Pharmceuticals,Raleigh, North Carolina, USA; Moli 1901, Lantibio, Chapel Hill,North Carolina, USA; SPI-881, Sucampo, Bethesda, Maryland,USA). Within the osmotic hydrator class of therapies, hyper-tonic saline is already available,39 40 and successful completionof phase III trials of dry-powder mannitol in the USA andabroad led to the recent approval for use in Australia and theEU, whereas FDA review is presently underway in the USA.41–43

Addressing abnormal CF mucus, marked by the approval ofrecombinant human DNAse (Pulmozyme, Genentech,San Francisco, California, USA),44 has also had renewed interest,and is likely to lead to new lines of investigation supported bythe network.

The concept of targeting sodium hyperabsorption, andthereby improving ASL hydration, first emerged more thanthree decades ago.45 46 The earliest attempts at manipulatingsodium transport in the airways of patients with CF utilisedinhaled amiloride.47–49 Treatment with amiloride was ultim-ately unsuccessful, likely due to inadequate potency and dur-ation of action. Since then, newer agents with improvedpharmacokinetic and pharmacodynamic properties have beendeveloped, and early phase trials have been conducted to test

Table 2 CFTR-based therapies completed or in progress within theTherapeutics Development Network

CFTR mutation

Proportion of patientswith causativemutation (%)

Therapeuticapproach Status

G551D/other 4 Ivacaftor FDA approved,age ≥6Age 3–5 planned

Non-G551Dgating/other

1 Ivacaftor Phase II/III

R117H/other 5 Ivacaftor Phase IIIF508del/F508del

49 Lumacaftor+ivacaftor

Phase II; phase IIIplanned

VX-661+ivacaftor

Phase II

PTC/other 10 Ataluren Phase III (primaryendpoint negative)

CF, cystic fibrosis; CFTR, CF transmembrane conductance regulator; PTC, prematuretermination codon.


Cystic fibrosis

their safety.50 Additional research with the latest moleculesshould be available soon for testing within the CF-TDN.

Stimulation of chloride secretion in the absence of CFTR canbe accomplished in vitro and in vivo using agents that stimu-late alternative chloride channels, including TMEM16a andothers activated by calcium.51 Testing of an orally dosed pros-tone (cobiprostone, Sucampo) was conducted by the CF-TDNnearly 10 years ago. Although this study did not demonstrate aclear chloride secretory response, a related compound was even-tually approved as a chloride secretagogue for the treatment ofchronic constipation, and this study along with other earlyefforts informed the network on the use of NPD and othertechniques now commonly used in CF-TDN networkstudies.25 52 Moli1901 (Lantibio) showed early promise as achloride transport activator in US and European studies,53 buthas not been pursued further due to formulation issues.Denufosol (Inspire Pharmaceuticals), a P2Y2 agonist designedto restore chloride transport and increase mucociliary clearance,also yielded promising results in early phase studies,54 55 butultimately failed after the conduct of two phase III trials.56 57

Reflection on lessons learned and research into the potentialcause of this failure has been undertaken and will educate thesearch for additional ASL modifying agents.

ANTI-INFECTIVESChronic bacterial infections of the CF airways with resistantorganisms, including multidrug resistant (MDR) Pseudomonasaeruginosa and other intrinsically resistant organisms such asBurkholderia cepacia complex, Stenotrophomonas maltophilia andAchromobacter species remain a major therapeutic challenge inCF.58 For example, the current rate of MDR P aeruginosa amongpatients with CF is 9.8%.59 Because of the severity and relent-less nature of CF airway infections and the need to developnew treatment approaches to alleviate the development ofresistance, an important focus of TDN-supported studies hasbeen in the area of anti-infectives. Unfortunately, few new anti-microbials directed against P aeruginosa have been developed inthe past two decades. Thus, recent anti-infective strategieshave used currently available antimicrobials in new ways. Inaddition, novel approaches to treatment have been developed,including suppression of chronic infections and early therapyto eradicate infections prior to irreversible infection.

A sentinel agent in the development of new antimicrobialstrategies for treatment of P aeruginosa in CF was tobramycininhalation solution (TOBI, Novartis Pharmaceuticals, EastHanover, New Jersey, USA). Inhalation of antibiotics allowshigh concentrations of active drug at the site of infection,while minimising side effects. Inhaled tobramycin was testedin two trials including a total of 520 subjects over the age of6 years. The trials demonstrated improvement in lung functionand decrease in bacterial colony counts, and supported thedrug’s FDA approval in 1994. Although early in the formationof the network, the concept of inhaled suppressive antibioticsand the actual study design of the inhaled tobramycin trialshave served as a model for other antimicrobial agents tested byTDCs. Aztreonam lysinate for inhalation (AZLI, Cayston,Gilead Sciences) was initially developed through support of theCFF and the clinical trials were performed through the TDN.Since its approval by the FDA, AZLI has been rapidly used bythe CF population, demonstrating the intense need for newantimicrobial therapies. To extend efficacy to alternative organ-isms, a phase IV trial of AZLI for patients with CF andB cepacia complex infections was recently completed andGilead is conducting an ongoing phase IV CFF Patient

Registry-based study of antibiotic resistance. Other agents thathave been formulated for airway administration and are cur-rently in the CF Foundation Therapeutics Pipeline includetobramycin inhaled powder (TOBI Podhaler, NovartisPharmaceuticals), inhaled levofloxacin (Aeroquin AxcanPharmaceuticals, San Diego, California, USA), and inhaled lipo-somal amikacin (Arikace, Insmed, Monmouth Junction, NewJersey, USA), all of which are in phase III trials and offerimproved convenience, delivery, and durability. Similarly, cipro-floxacin inhaled powder (Bayer Schering Pharma AG,Berlin-Wedding, Germany), has recently completed a phase IItrial in CF, but has not yet begun phase III testing.

Another example of using currently available antimicrobialsis the novel use of azithromycin in the management of CFairway infections. Although active in vitro againstStaphylococcus aureus and Haemophilus influenzae, azithromycinis not considered active against the most typical gram-negativeCF pathogens. However, studies developed through CysticFibrosis Foundation Therapeutics have demonstrated clinicalresponse to three times a week administration of azithromycin.This was initially shown in a large clinical trial in patientswith CF and P aeruginosa infection who were over the age of 6and subsequently confirmed in younger children and thoseuninfected with P aeruginosa.60 61 The exact mechanism ofaction has not been demonstrated but patients responded withan improvement in FEV1 and a decreased frequency of pulmon-ary exacerbations.

Studies within the TDN have examined the role of earlyantibiotic therapy in eradicating early infection and delayingthe establishment of chronic infections in young children.62–64

The Early Pseudomonas Infection Control (EPIC) trial was amulticentre randomised trial comparing one-drug or two-drugtherapy (TOBI with or without oral ciprofloxacin) with eithercycled or culture-dependent administration to eradicate early Paeruginosa infection. It demonstrated equal efficacy and safetyin all arms without emergence of multiple antibiotic resistantorganisms.65 66

A new antimicrobial soon to begin phase II testing is KB001(KaloBios Pharmaceuticals, South San Francisco, California,USA), a monoclonal antibody fragment directed against thePcrV protein of the type 3 secretion system of P aeruginosa.67

Other antimicrobial therapies on the horizon include trials forthe management of methicillin-resistant S aureus and non-tuberculous mycobacteria, and intravenous administration ofgallium, an agent that is not a conventional antibiotic but hasantimicrobial properties and is currently used by radiologistsfor nuclear medicine scans.

ANTI-INFLAMMATORY THERAPIESNeutrophil-dominated airway inflammation is a hallmark ofCF lung disease (see previous reviews68–71); thus, evaluatingdrugs that target inflammation is a major emphasis of thenetwork. Corticosteroids and high-dose ibuprofen, each broadspectrum inhibitors of pro-inflammatory signalling, were twoof the first anti-inflammatory drugs studied in CF. Both demon-strated clinical benefit,72–74 but side effects and other considera-tions have limited their use.75–77 However, the evidence thatthese anti-inflammatory agents can slow the progression of CFlung disease, particularly in children, suggests that strategies tomodulate lung inflammation can be beneficial.

The CF-TDN has devoted significant resources to evaluatinganti-inflammatory treatments. Among those successful includeazithromycin, hydroxychloroquine which has been shown toreduce pulmonary exacerbations and improve lung function in


Cystic fibrosis

patients with CF, without significantly affecting lower airwaybacterial density60 61 (see Anti-infectives section). The evidencesuggests that azithromycin may act as an anti-inflammatory orimmunomodulatory agent, which may also explain its efficacyin chronic obstructive pulmonary disease.78 In a trial of azithro-mycin in patients with CF chronically infected with P aerugi-nosa, there was a modest reduction in sputum neutrophilelastase activity in favour of the azithromycin group.61 In asubsequent clinical trial in patients with CF, 6–18 years of ageand uninfected with P aeruginosa, azithromycin significantlyreduced circulating neutrophil counts and systemic markers ofinflammation, including C-reactive protein, serum amyloid Aand calprotectin.79

Several CF-TDN-supported studies evaluating more targetedanti-inflammatory agents have been largely unsuccessful, and afew were associated with detrimental side effects. A notableexample of the latter was a trial testing the leukotriene B4 recep-tor antagonist amelubant (Boehringer Ingelheim, Bracknell, UK)in patients with CF and mild to moderate lung disease.80

The study was terminated early due to a significant increase inthe frequency of pulmonary exacerbations in adults receivingamelubant. This trial gave the CF community pause, providinga potent reminder that modulating inflammation is not withoutrisks. Based on preclinical models and associated clinical data,81

it was hypothesised that treatment with interferon γ (IFNγ), apleiotropic cytokine with immunomodulatory and antimicrobialactivities, would be beneficial in CF. However, a clinical trialwith inhaled IFN-γ1b (Actimmune, InterMune, Brisbane,California, USA) did not significantly improve lung function,sputum bacterial density or sputum biomarkers of inflamma-tion.82 The CF-TDN network also sponsored studies of‘low-hanging fruit’ anti-inflammatory compounds (ie, availablemedications used for other inflammatory disorders), includingthe 3-hydroxy-3-methyl coenzyme A reductase inhibitor simvas-tatin, the peroxisome proliferator-activated receptor γ agonistpioglitazone, and the immunosuppressants methotrexate andhydroxychloroquine. Data from these small, short studies werelargely negative, while methotrexate was associated with anintolerable adverse event profile.83–86 These compounds are nolonger being actively investigated, and indicate the complexitiesof a targeted anti-inflammatory approach in CF lung disease.Additionally, recent 6-month antioxidant trials of inhaled gluta-thione and oral N-acetylcysteine, a glutathione prodrug, did notdemonstrate positive effects either on inflammation, oxidativestress, or clinical outcomes in patients with CF.87 88

Despite these disappointments, there are several ongoingstudies of potential therapeutic agents which address airwayinflammation. Numerous studies have demonstrated an excessburden of neutrophil-derived proteases in the CF airway andprovide justification for interventions that neutralise free neu-trophil elastase activity or augment local antiprotease levels.There has been renewed interest by the pharmaceutical indus-try to develop α1-antitrypsin therapies in CF and clinical trialsusing more potent and efficacious formulations are once againbeing planned. Antioxidant deficiencies remain a viable thera-peutic target in CF. A trial of a reformulated oral multivitaminsupplement (AquADEKs), enriched with several non-enzymaticantioxidants including β-carotene, mixed carotenoids, coen-zyme Q10, mixed tocopherols and selenium, will evaluate theeffects of this antioxidant ‘cocktail’ on antioxidant levels andmarkers of oxidative stress. Other novel approaches include thephosphodiesterase inhibitor sildenafil on airway inflammationand the effects of sulforaphane (found in broccoli sprouts) onNrf2 activation. Finally, a crucial question in the new era of

CFTR modulation is whether improving CFTR activity willhave a detectable impact on airway inflammation, a questionaddressed in the GOAL study (see CFTR modulation section).The breadth of these studies using the CF-TDN networksuggest targeting inflammation in CF remains an attractivetherapeutic approach, but optimising anti-inflammatory effectswhile minimising any detrimental impact on host defenceremains a key challenge.

NUTRITIONAL THERAPIESGood nutrition is vital in CF,89 thus the TDN has investigatedinterventions to support normal growth. Almost 90% ofpatients with CF are pancreatic insufficient90 and need life-longtreatment with pancreatic enzyme replacement therapy (PERT)to prevent malnutrition. Because some pancreatic enzymes pro-ducts were available before passage of the 1937 Pure Food andDrug Act, drug companies had been allowed to manufactureand sell PERT without prior FDA approval.91 In 1994 casereports of fibrosing colonopathy (FC) began to appear; a subse-quent analysis of CF Foundation Patient Registry data showeda strong correlation between dose of PERT and the appearanceof FC.92 Concerns regarding PERToveruse, and other issues (eg,enteric coating and high-dose capsules) were raised as potentialcontributing factors. As a result of these and other problems, in2004 the FDA issued notice that they would require a newdrug application for PERT with a deadline of April 2008,93

which was subsequently extended to 2010. The CF-TDNplayed a central role in developing study protocols and imple-menting studies of the efficacy of PERT in this short buturgent time window. Based on these efforts, the FDA hasapproved five oral delayed-release PERTs, Creon (Creon AbbottLaboratories. Abbott Park, IL, USA),94 Zenpep (Zenpep AptalisPharma US, Inc., Birmingham, AL, USA),95 96 Pancreaze(Pancreaze Janssen Pharmaceuticals, Titusville, NJ, USA),97 98

Pertzye (Pertzye Digestive Care, Inc., Bethlehem, PA, USA) 99

and Ultresa (Aptalis Pharma US, Inc., Birmingham, AL, USA).A longitudinal observational study of nutritional status in

infants (BONUS) is underway, cosponsored by the NationalInstitute of Health. With newborn screening for CF available inall 50 states as of 2010, PERT is being used in a larger numberof infants, yet dose response has never been defined. A nestedsub-study of BONUS will explore the efficacy of higher andlower doses of PERT in infants.

All current PERTs are biological products derived from hogpancreas. Although current FDA-approved products are nolonger overfilled, have improved stability and are free of envel-oped viruses, non-enveloped viruses cannot be removed fromthe drug substance without loss of efficacy. The CF-TDN hasparticipated in studies of liprotamase, a novel formulation ofrecombinant bacterial lipase, fungal protease and amorphousamylase that is a non-biological PERT.100–103 The crystallinenature of some of its components confers stability in an acidenvironment and makes a liquid preparation possible, providinga potential advantage.

Progression of CF lung disease is linked to oxidative stress.104

Pancreatic insufficiency and a diminished bile acid pool causemalabsorption of important antioxidants including fat-solublevitamins (vitamins A, D, E and K), carotenoids, tocopherols, sel-enium and coenzyme Q10.105 A number of studies have shownthat in patients with CF, supplementation can correct manymicronutrient deficiencies, although the evidence of clinicalbenefit has been hampered by inconsistent study design.106 TheCF-TDN participated in developing an oral antioxidant-enrichedmultivitamin supplement that safely increased systemic


Cystic fibrosis

antioxidant levels107 and continues efforts to explore the bene-fits of this nutritional approach to modify CF disease.

BIOMARKERSSince the inception of the CF-TDN, the development of bio-markers to enhance CF clinical trial planning and conduct hasbeen at the forefront of network activity since alternatives tohard clinical endpoints such as mortality and hospitalisationare not viable endpoints for phase II and many phase III pro-grammes.4 Biomarkers are ‘a characteristic that is objectivelymeasured and evaluated as an indicator of a normal biologicprocess, pathogenic process, or pharmacologic response to atherapeutic intervention’.108 As such, CF-TDN-supportedresearch has concentrated on the testing and validation of newbiomarkers, linking changes in biomarkers to established surro-gate or clinical outcome measures, and refining biomarkerassessment techniques largely in the context of drug develop-ment. In addition, the acquisition of high-quality longitudinalbiospecimens linked to clinical data has been incorporated intoseveral CF-TDN clinical trials, leading to the development ofrich specimen banks for biomarker discovery. For example,these biobanks have been leveraged by the CF research commu-nity to help define relationships between sputum inflammatorymarkers (eg, interleukin-8, human neutrophil elastase) and clin-ical outcome measures (eg, lung function),8 109 and to relatemacrolide therapy to anti-inflammatory activity.79 The capacityto interrogate samples obtained from multicentre studies thathave been collected with common standard operating proce-dures also increases the probability that results can be success-fully recapitulated in the clinic.

With the development of CFTR modulators, there has been aneed to develop and refine biomarkers related to augmentedCFTR function. From these efforts, sweat chloride has emergedas a robust, standardised and validated method to monitor thebioactivity of CFTR potentiators and correctors.6 17 19 20 110

NPD has also performed reasonably well as a CFTR bio-marker,111–113 successfully detecting ivacaftor activity inpatients with G551D CF. For unclear reasons, neither CFTR bio-marker has correlated with clinical outcomes (eg, lungfunction, patient-reported outcomes (PROs), weight) on an indi-vidual patient basis. Ongoing CF-TDN trials are underway tobetter understand this gap; one example is the GOAL study,which includes the evaluation of functional biomarkers (seeCFTR modulation section), banking of serum, sputum, andurine linked to clinical data, and may help future biomarkerdiscovery.

A critical barrier that can potentially limit the extension ofnew therapies to the treatment of infants and toddlers with CF(and other patients with very mild obstructive disease) is therelative lack of biomarkers sensitive to small changes in thesepopulations. It is a widely held opinion that the greatest bene-fits conferred by many CF therapies would be expected in thehealthiest patients, prior to the development of irreversiblelung injury. As one example of success in this area, theCF-TDN developed standard operating procedures for theconduct of infant pulmonary function tests for multicentreperformance and analysis.114 These efforts were used to dem-onstrate improved lung function (change in FEV in 0.5 s) ininfants and toddlers with CF treated with 7% hypertonicsaline compared with normal saline, despite no detectable dif-ference in pulmonary exacerbation rates between the two treat-ment groups.115 Other CF-TDN studies in progress are gearedtowards accelerating drug development in the infant popula-tion, including the Baby Observational and Nutritional Study

which also includes a biospecimen collection programme (seealso above). Finally, TDCs have participated in non-invasivetechniques to monitor mild lung disease and responses to inter-vention, including CT imaging55 116–119 and Lung ClearanceIndex measurements in ivacaftor-treated patients with G551DCF. Further studies that link sensitive disease biomarkers toclinical endpoints in the setting of new and established therap-ies could help accelerate the development path required tobring therapies to the healthiest patients with CF.

The TDN has played an instrumental role in advancing CFPROs and helping define their role in drug development. Thiswork was particularly instrumental during the development ofAZLI.120–122 PROs have also served as primary and secondaryendpoints in pivotal trials involving CF therapeutics; establish-ing the minimal clinically important difference for theRespiratory Domain of the CF Questionnaire-Revised (CFQ-R)was critical to this effort.123 This work has paved the way forthe incorporation of the CFQ-R into clinical trials of manynew CF treatment strategies.6 17 19 20 124 125 Future studies areplanned to further refine CF PROs, including the developmentof PROs for non-pulmonary disease manifestations, correlationof PRO results with other biomarkers, and the capture of infor-mation more frequently (eg, daily diaries compared with2-week recall that is a limitation of the current CFQ-R).

FUTURE DIRECTIONS FOR THE NETWORKAs novel therapies emerge and evolve, clinical trials infrastruc-tures must respond to the expanding need for studies. TheCF-TDN and its partner network, the European Society forCystic Fibrosis (ESCF) Clinical Trials Network, must maintainhigh-quality study conduct, which will be optimised by provid-ing the structure and tools for improvement through web-based, QI programmes. These programmes are available to allsites with the requirement that they participate in an assess-ment–prioritisation–implementation–reflection cycle and reporttheir results annually. A competitive application process willcontinue to sustain an engaged network of participating sitesthat allows new sites to enter the network when they

Table 3 Lessons learned by the TDN and changes implemented toaddress themLessons learned Actions to address challenges

In an expanding network, different siteshave variable experience and knowledgebase regarding conduct of clinicalresearch

Defined education programmes includingself-education modules, publishedstandard operating procedures, etc

Variable performance in study start-uptime and enrolment rate across TDN sites

Established a central database to trackperformance, provide sites real-timefeedback, annual peer comparisons ofmetrics and a web-based qualityimprovement programme

Variability in important clinical endpointsand biomarkers can adversely impactinterpretation of results

Established National Resource Centres toprovide expertise in particular endpointsand provide quality assurance toindividual TDCs. Established andpublished standard operating proceduresfor outcomes

Increase in number of interestedsponsors and therapeutics to be testedcreates stress on recruitment capabilitiesand study throughput

Steady expansion of clinical networkthrough competitive renewal programmethat emphasises performance metricsand quality study conduct

Variable experience and understanding ofCF by industry sponsors

Developed TDN sponsored consultingprogramme to provide consistent adviceto sponsors and to enable rapiddevelopment of clinical protocols

CF, cystic fibrosis; TDC, Therapeutic Development Centre; TDN, Therapeutics DevelopmentNetwork.


Cystic fibrosis

demonstrate interest and capability, and permits adjustingnetwork size to meet enrolment projections for studies linkedto the research pipeline.

The CF-TDN will continue to work cooperatively with theESCF Clinical Trials Network and the Canadian and AustralianCF research communities to facilitate the conduct of uniformprotocols across multiple countries, to enrol larger sample sizesin a shorter time period, and permit sufficient recruitment ofsubsets of patients with unique characteristics, such as age,genotype, infecting organism or comorbidities. Collaborationacross networks also exists for combined protocol reviews,development of industry and investigator initiated studies, andworking groups developing consensus around novel outcomemeasures and endpoint definitions. Similarly, the network hasmade a consistent effort to learn from challenges experienced,and address these areas in the future; some important examplesare noted in table 3.

The development of novel or improved outcome measuresremains a major focus for CF-TDN working groups; in particu-lar, assessing outcome measures applicable for infants andyoung children as new therapies become available and decisionsregarding efficacy in younger populations must be made, forexample.126 The existence of the CF Patient Registry enhancesthe conduct of clinical research in many ways, and will be apivotal tool for implementation of comparative effectivenessresearch (CER) in CF. The tool has already been used in long-term observational studies conducted through the CF-TDN.127

As this capability grows, it is conceivable that it will allow theexpansion of future CER and observational studies to all CF carecentres that participate in the registry, maximising investigativepower. Combined with the robust pipeline of therapeutics, thenetwork is well poised to make continued contributions to therapid evolution of CF care.

Acknowledgements The authors acknowledge grant support includingCLANCY09Y0, CLANCY05Y2, and UL1 TR000165 (SMR), BOROWI03CS0 (DSB),RETSCH09Y0 and UL1TR000083 (GRB and SHD), AMIN09YO (JPC), RAMSEY03Y0,1P30DK089507 and UL1TR000423-06 ( JLB and BMR) and SAGEL07B0 (SDS).

Contributors All named authors wrote and edited the manuscript. SMR takes finalresponsibility for the content. No ghost writers were affiliated with the work.

Funding CFF and NIH.



REFERENCES1. US Food and Drug Administration. FDA definition ‘orphan disease’.

http://www.fda.gov/ForIndustry/DevelopingProductsforRareDiseasesConditions/HowtoapplyforOrphanProductDesignation/DefinitionofDiseasePrevalence/default.htm(accessed 23 Aug 2012)

2. Pattishall EN. Negative clinical trials in cystic fibrosis research. Pediatrics1990;85:277–81.

3. Goss CH, Mayer-Hamblett N, Williams J, et al. The cystic fibrosis foundationtherapeutics development network: a national effort by the cystic fibrosisfoundation to build a clinical trials network. Child Health Care 2008;37:5–20.

4. Goss CH, Mayer-Hamblett N, Kronmal RA, et al. The cystic fibrosis therapeuticsdevelopment network (CF TDN): a paradigm of a clinical trials network for geneticand orphan diseases. Adv Drug Deliv Rev 2002;54:1505–28.

5. Marshall BC, Penland CM, Hazle L, et al. Cystic fibrosis foundation: achieving themission. Respir Care 2009;54:788–95; discussion 95.

6. Ramsey BW, Davies J, McElvaney NG, et al. A CFTR potentiator in patients withcystic fibrosis and the G551D mutation. N Engl J Med 2011;365:1663–72.

7. Emond MJ, Louie T, Emerson J, et al. Exome sequencing of extreme phenotypesidentifies DCTN4 as a modifier of chronic Pseudomonas aeruginosa infection incystic fibrosis. Nat Genet 2012;44:886–9.

8. Mayer-Hamblett N, Aitken ML, Accurso FJ, et al. Association betweenpulmonary function and sputum biomarkers in cystic fibrosis. Am J Respir Crit CareMed 2007;175:822–8.

9. Mayer-Hamblett N, Kronmal RA. Improving the estimation of change frombaseline in a continuous outcome measure in the clinical trial setting. ContempClin Trials 2005;26:2–16.

10. Quon BS, Goss CH. A story of success: continuous quality improvement in cysticfibrosis care in the USA. Thorax 2011;66:1106–8.

11. Clancy JP, Jain M. Personalized medicine in cystic fibrosis: dawning of a new era.Am J Respir Crit Care Med 2012.

12. Sloane PA, Rowe SM. Cystic fibrosis transmembrane conductance regulatorprotein repair as a therapeutic strategy in cystic fibrosis. Curr Opin Pulm Med2010;16:591–7.

13. Davis PB, Yasothan U, Kirkpatrick P. Ivacaftor. Nat Rev Drug Discov 2012;11:349–50.14. Corbyn Z. Promising new era dawns for cystic fibrosis treatment. Lancet

2012;379:1475–6.15. Kaiser J. Personalized medicine. New cystic fibrosis drug offers hope, at a price.

Science 2012;335:645.16. Rowe SM, Miller S, Sorscher EJ. Cystic fibrosis. N Engl J Med

2005;352:1992–2001.17. Accurso FJ, Rowe SM, Clancy JP, et al. Effect of VX-770 in persons with cystic

fibrosis and the G551D-CFTR mutation. N Engl J Med 2010;363:1991–2003.18. Aherns R, Rodriguez S, Yen K, et al. VX-770 in subjects 6 to 11 years with cystic

fibrosis and the G551D-CFTR mutation. Ped Pulm Suppl 2011;46:A198.19. Flume PA, Liou TG, Borowitz DS, et al. Ivacaftor in subjects with cystic fibrosis

who are homozygous for the F508del-CFTR mutation. Chest 2012. Published OnlineFirst: 1 March 2012. doi: 10.1378/chest.11-2672

20. Clancy JP, Rowe SM, Accurso FJ, et al. Results of a phase IIa study of VX-809,an investigational CFTR corrector compound, in subjects with cystic fibrosishomozygous for the F508del-CFTR mutation. Thorax 2011;67:12–18.

21. Van Goor F, Hadida S, Grootenhuis PD, et al. Correction of the F508del-CFTRprotein processing defect in vitro by the investigational drug VX-809. Proc NatAcad Sci U S A 2011;108:18843–8.

22. Van Goor F, Hadida S, Grootenhuis PD, et al. Rescue of CF airway epithelial cellfunction in vitro by a CFTR potentiator, VX-770. Proc Natl Acad Sci U S A2009;106:18825–30.

23. Wang F, Zeltwanger S, Hu S, et al. Deletion of phenylalanine 508 causesattenuated phosphorylation-dependent activation of CFTR chloride channels.J Physiol 2000;524(Pt 3):637–48.

24. Bedwell DM, Kaenjak A, Benos DJ, et al. Suppression of a CFTR premature stopmutation in a bronchial epithelial cell line. Nat Med 1997;3:1280–4.

25. Clancy JP, Rowe SM, Bebok Z, et al. No detectable improvements in cystic fibrosistransmembrane conductance regulator by nasal aminoglycosides in patients withcystic fibrosis with stop mutations. Am J Respir Cell Mol Biol 2007;37:57–66.

26. Clancy JP, Bebok Z, Ruiz F, et al. Evidence that systemic gentamicin suppressespremature stop mutations in patients with cystic fibrosis. Am J Respir Crit CareMed 2001;163:1683–92.

27. Wilschanski M, Yahav Y, Yaacov Y, et al. Gentamicin-induced correction of CFTRfunction in patients with cystic fibrosis and CFTR stop mutations. N Engl J Med2003;349:1433–41.

28. Sermet-Gaudelus I, Renouil M, Fajac A, et al. In vitro prediction of stop-codonsuppression by intravenous gentamicin in patients with cystic fibrosis: a pilot study.BMC Med 2007;5:5–14.

29. Welch EM, Barton ER, Zhuo J, et al. PTC124 targets genetic disorders caused bynonsense mutations. Nature 2007;447:87–91.

30. Du M, Liu X, Welch EM, et al. PTC124 is an orally bioavailable compound thatpromotes suppression of the human CFTR-G542X nonsense allele in a CF mousemodel. Proc Natl Acad Sci U S A 2008;105:2064–9.

31. Kerem E, Hirawat S, Armoni S, et al. Effectiveness of PTC124 treatment of cysticfibrosis caused by nonsense mutations: a prospective phase II trial. Lancet2008;372:719–27.

32. Clancy JP, Konstan MW, Rowe SM, et al. A phase II study of PTC124 in CFpatients harboring premature stop mutations. Ped Pulmonol Suppl 2006;41(Suppl29):abstract 269.

33. Sermet-Gaudelus I, Boeck KD, Casimir GJ, et al. Ataluren (PTC124) inducescystic fibrosis transmembrane conductance regulator protein expression andactivity in children with nonsense mutation cystic fibrosis. Am J Respir Crit CareMed 2010;182:1262–72.

34. Konstan M, Accurso F, De Boeck K, et al. Results of the phase 3 study of atalurenin nonsense mutation cystic fibrosis (nmCF). J Cyst Fibros 2012;11(Suppl 1).

35. Matsui H, Grubb BR, Tarran R, et al. Evidence for periciliary liquid layer depletion,not abnormal ion composition, in the pathogenesis of cystic fibrosis airwaysdisease. Cell 1998;95:1005–15.

36. Stutts MJ, Canessa CM, Olsen JC, et al. CFTR as a cAMP-dependent regulator ofsodium channels [see comments]. Science 1995;269:847–50.

37. Tarran R, Loewen ME, Paradiso AM, et al. Regulation of murine airway surfaceliquid volume by CFTR and Ca2+-activated Cl− conductances. J Gen Physiol2002;120:407–18.

38. Tarran R, Button B, Picher M, et al. Normal and cystic fbrosis airway surfaceliquid homeostasis: the effects of phasic shear stress and viral infections. J BiolChem 2005;280:35751–9.


Cystic fibrosis

http://www.fda.gov/ForIndustry/DevelopingProductsforRareDiseasesConditions/HowtoapplyforOrphanProductDesignation/DefinitionofDiseasePrevalence/default.htm



39. Elkins MR, Robinson M, Rose BR, et al. A controlled trial of long-term inhaledhypertonic saline in patients with cystic fibrosis. N Engl J Med 2006;354:229–40.

40. Donaldson SH, Bennett WD, Zeman KL, et al. Mucus clearance and lung functionin cystic fibrosis with hypertonic saline. N Engl J Med 2006;354:241–50.

41. Bilton D, Robinson P, Cooper P, et al. Inhaled dry powder mannitol in cysticfibrosis: an efficacy and safety study. Eur Respir J 2011;38:1071–80.

42. Aitken ML, Bellon G, De Boeck K, et al. Long-term inhaled dry powder mannitolin cystic fibrosis: an international randomized study. Am J Respir Crit Care Med2012;185:645–52.

43. Minasian C, Wallis C, Metcalfe C, et al. Comparison of inhaled mannitol, dailyrhDNase and a combination of both in children with cystic fibrosis: a randomisedtrial. Thorax 2010;65:51–6.

44. Fuchs HJ, Borowitz DS, Christiansen DH, et al. Effect of aerosolized recombinanthuman DNase on exacerbations of respiratory symptoms and on pulmonaryfunction in patients with cystic fibrosis. The Pulmozyme Study Group. N Engl JMed 1994;331:637–42.

45. Willumsen NJ, Boucher RC. Transcellular sodium transport in cultured cysticfibrosis human nasal epithelium. Am J Physiol 1991;261(2 Pt 1):C332–41.

46. Knowles M, Gatzy J, Boucher R. Increased bioelectric potential differenceacross respiratory epithelia in cystic fibrosis. N Engl J Med 1981;305:1489–95.

47. Knowles MR, Church NL, Waltner WE, et al. A pilot study of aerosolizedamiloride for the treatment of lung disease in cystic fibrosis. N Engl J Med1990;322:1189–94.

48. App EM, King M, Helfesrieder R, et al. Acute and long-term amiloride inhalation incystic fibrosis lung disease. A rational approach to cystic fibrosis therapy. Am RevRespir Dis 1990;141:605–12.

49. Graham A, Hasani A, Alton EW, et al. No added benefit from nebulized amiloridein patients with cystic fibrosis. Eur Respir J 1993;6:1243–8.

50. Quijana F, Doran J, Smith R, et al. Safety and pharmacokinetics of a single doseof 552-02 following aerosolized administration to patients with cystic fibrosis.Pediatr Pulmonol 2005;(Suppl 28):A216.

51. Mason SJ, Paradiso AM, Boucher RC. Regulation of transepithelial ion transportand intracellular calcium by extracellular ATP in human normal and cystic fibrosisairway epithelium. Br J Pharmacol 1991;103:1649–56.

52. Ahrens RC, Standaert TA, Launspach J, et al. Use of nasal potential differenceand sweat chloride as outcome measures in multicenter clinical trials in subjectswith cystic fibrosis. Pediatr Pulmonol 2002;33:142–50.

53. Zeitlin PL, Boyle MP, Guggino WB, et al. A phase I trial of intranasal Moli1901 forcystic fibrosis. Chest 2004;125:143–9.

54. Deterding R, Retsch-Bogart G, Milgram L, et al. Safety and tolerability ofdenufosol tetrasodium inhalation solution, a novel P2Y2 receptor agonist: results ofa phase 1/phase 2 multicenter study in mild to moderate cystic fibrosis. PediatrPulmonol 2005;39:339–48.

55. Deterding RR, Lavange LM, Engels JM, et al. Phase 2 randomized safety andefficacy trial of nebulized denufosol tetrasodium in cystic fibrosis. Am J Respir CritCare Med 2007;176:362–9.

56. Accurso FJ, Moss RB, Wilmott RW, et al. Denufosol tetrasodium in patients withcystic fibrosis and normal to mildly impaired lung function. Am J Respir Crit CareMed 2011;183:627–34.

57. Ratjen F, Durham T, Navratil T, et al. Long term effects of denufosol tetrasodium inpatients with cystic fibrosis. J Cyst Fibros 2012. Published Online First: 8 June2012. doi: 10.1016/j.jcf.2012.05.003

58. Mogayzel PJ Jr, Flume PA. Update in cystic fibrosis 2010. Am J Respir Crit CareMed 2011;183:1620–4.

59. Cystic Fibrosis Foundation Patient Registry. 2010 Annual Data Report, 2011.60. Saiman L, Anstead M, Mayer-Hamblett N, et al. Effect of azithromycin on

pulmonary function in patients with cystic fibrosis uninfected with Pseudomonasaeruginosa: a randomized controlled trial. JAMA 2010;303:1707–15.

61. Saiman L, Marshall BC, Mayer-Hamblett N, et al. Azithromycin in patients withcystic fibrosis chronically infected with Pseudomonas aeruginosa: a randomizedcontrolled trial. JAMA 2003;290:1749–56.

62. Gibson RL, Emerson J, Mayer-Hamblett N, et al. Duration of treatment effectafter tobramycin solution for inhalation in young children with cystic fibrosis. PediatrPulmonol 2007;42:610–23.

63. Gibson RL, Emerson J, McNamara S, et al. Significant microbiological effect ofinhaled tobramycin in young children with cystic fibrosis. Am J Respir Crit CareMed 2003;167:841–9.

64. Treggiari MM, Rosenfeld M, Retsch-Bogart G, et al. Approach to eradication ofinitial Pseudomonas aeruginosa infection in children with cystic fibrosis. PediatrPulmonol 2007;42:751–6.

65. Mayer-Hamblett N, Kronmal RA, Gibson RL, et al. Initial Pseudomonasaeruginosa treatment failure is associated with exacerbations in cystic fibrosis.Pediatr Pulmonol 2012;47:125–34.

66. Treggiari MM, Retsch-Bogart G, Mayer-Hamblett N, et al. Comparative efficacyand safety of 4 randomized regimens to treat early Pseudomonas aeruginosainfection in children with cystic fibrosis. Arch Pediatr Adolesc Med2011;165:847–56.

67. Francois B, Luyt CE, Dugard A, et al. Safety and pharmacokinetics of an anti-PcrVPEGylated monoclonal antibody fragment in mechanically ventilated patientscolonized with Pseudomonas aeruginosa: a randomized, double-blind,placebo-controlled trial. Crit Care Med 2012;40:2320–6.

68. Chmiel JF, Berger M, Konstan MW. The role of inflammation in thepathophysiology of CF lung disease. Clin Rev Allergy Immunol 2002;23:5–27.

69. Sagel SD, Chmiel JF, Konstan MW. Sputum biomarkers of inflammation in cysticfibrosis lung disease. Proc Am Thorac Soc 2007;4:406–17.

70. Nichols D, Chmiel J, Berger M. Chronic inflammation in the cystic fibrosis lung:alterations in inter- and intracellular signaling. Clin Rev Allergy Immunol2008;34:146–62.

71. Elizur A, Cannon CL, Ferkol TW. Airway inflammation in cystic fibrosis. Chest2008;133:489–95.

72. Eigen H, Rosenstein BJ, FitzSimmons S, et al. A multicenter study ofalternate-day prednisone therapy in patients with cystic fibrosis. Cystic FibrosisFoundation Prednisone Trial Group. J Pediatr 1995;126:515–23.

73. Konstan MW, Byard PJ, Hoppel CL, et al. Effect of high-dose ibuprofen in patientswith cystic fibrosis. N Engl J Med 1995;332:848–54.

74. Lands LC, Milner R, Cantin AM, et al. High-dose ibuprofen in cystic fibrosis:Canadian safety and effectiveness trial. J Pediatr 2007;151:249–54.

75. Oermann CM, Sockrider MM, Konstan MW. The use of anti-inflammatorymedications in cystic fibrosis: trends and physician attitudes. Chest 1999;115:1053–8.

76. Lai HC, FitzSimmons SC, Allen DB, et al. Risk of persistent growth impairmentafter alternate-day prednisone treatment in children with cystic fibrosis. N Engl JMed 2000;342:851–9.

77. Konstan MW, VanDevanter DR, Rasouliyan L, et al. Trends in the use of routinetherapies in cystic fibrosis: 1995–2005. Pediatr Pulmonol 2010;45:1167–72.

78. Wenzel RP, Fowler AA III, Edmond MB. Antibiotic prevention of acuteexacerbations of COPD. N Engl J Med. [Review] 2012;367:340–7.

79. Ratjen F, Saiman L, Mayer-Hamblett N, et al. Effect of azithromycin on systemicmarkers of inflammation in cystic fibrosis patients uninfected with Pseudomonasaeruginosa. Chest 2012. Published Online First: 17 May 2012. doi:10.1378/chest.12-0628

80. Konstan MW, Doring G, Lands LC, et al. Results of a phase II clinical trial of BIIL284 BS (an LTB4 receptor antagonist) for the treatment of CF lung disease. PedPulmonol Suppl 2005;28(Suppl 7.4):125–6.

81. Chmiel JF, Konstan MW. Anti-inflammatory medications for cystic fibrosis lungdisease: selecting the most appropriate agent. Treat Respir Med 2005;4:255–73.

82. Moss RB, Mayer-Hamblett N, Wagener J, et al. Randomized, double-blind,placebo-controlled, dose-escalating study of aerosolized interferon gamma-1b inpatients with mild to moderate cystic fibrosis lung disease. Pediatr Pulmonol2005;39:209–18.

83. Oermann C, Katz M, Wheeler C, et al. A pilot study evaluating the potential useof low-dose methotrexate as an anti-inflammatory for cystic fibrosis lung disease[abstract]. Pediatr Pulmonol Suppl 2007;30:292–3.

84. Kraynack N, Chmiel J, Xue W, et al. Effect of simvastatin on exhaled nitric oxideand inflammatory markers in sputum in patients with cystic fibrosis [abstract].Pediatr Pulmonol Suppl 2008;31:300.

85. Konstan M, Hilliard K, Bucur C, et al. Effect of pioglitazone on sputum markers ofinflammation in cystic fibrosis [abstract]. Pediatr Pulmonol Suppl 2008;31:310.

86. Williams B, Robinette M, Slovis B, et al. Hydroxychloroquine—pilot study ofanti-inflammatory effects in cystic fibrosis [abstract]. Pediatr Pulmonol Suppl2008;31:314.

87. Tirouvanziam R, Lymp J, Thompson V, et al. A multi-center, phase IIB,randomized, placebo-controlled, double-blind study of the effects ofN-acetylcysteine (NAC) on redox changes and lung inflammation in cystic fibrosispatients [abstract]. Pediatr Pulmonol Suppl 2011:280–1.

88. Griese M, Hector A, Kappler M, et al. Inhaled glutathione in cystic fibrosis[abstract]. J Cyst Fibros 2012;11:S11.

89. Stallings VA, Stark LJ, Robinson KA, et al. Evidence-based practicerecommendations for nutrition-related management of children and adults withcystic fibrosis and pancreatic insufficiency: results of a systematic review. J AmDiet Assoc 2008;108:832–9.

90. Borowitz D, Baker SS, Duffy L, et al. Use of fecal elastase-1 to classifypancreatic status in patients with cystic fibrosis. J Pediatr 2004;145:322–6.

91. Hendeles L, Hochhaus G, Kazerounian S. Generic and alternative brand-namepharmaceutical equivalents: select with caution. Am J Hosp Pharm1993;50:323–9.

92. FitzSimmons SC, Burkhart GA, Borowitz D, et al. High-dose pancreatic-enzymesupplements and fibrosing colonopathy in children with cystic fibrosis. N Engl JMed 1997;336:1283–9.

93. FDA. Exocrine pancreatic insufficiency drug products. Fed Regist2004;69:23410–14.

94. Trapnell BC, Maguiness K, Graff GR, et al. Efficacy and safety of Creon 24 000 insubjects with exocrine pancreatic insufficiency due to cystic fibrosis. J Cyst Fibros2009;8:370–7.


Cystic fibrosis

95. Wooldridge JL, Heubi JE, Amaro-Galvez R, et al. EUR-1008 pancreatic enzymereplacement is safe and effective in patients with cystic fibrosis and pancreaticinsufficiency. J Cyst Fibros 2009;8:405–17.

96. Graff GR, Maguiness K, McNamara J, et al. Efficacy and tolerability of a newformulation of pancrelipase delayed-release capsules in children aged 7 to 11 yearswith exocrine pancreatic insufficiency and cystic fibrosis: a multicenter,randomized, double-blind, placebo-controlled, two-period crossover, superioritystudy. Clin Ther 2010;32:89–103.

97. Trapnell BC, Strausbaugh SD, Woo MS, et al. Efficacy and safety of PANCREAZE(R) for treatment of exocrine pancreatic insufficiency due to cystic fibrosis. J CystFibros 2011;10:350–6.

98. Erratum in J Cyst Fibros 2011;10:491–2.99. Digestive Care. http://www.digestivecare.com/PDFs/NDA%20022175%

20APPROVAL%2017MAY2012_PI&MG.pdf (accessed 23 Aug 2012)100. Borowitz D, Goss CH, Stevens C, et al. Safety and preliminary clinical activity of a

novel pancreatic enzyme preparation in pancreatic insufficient cystic fibrosispatients. Pancreas 2006;32:258–63.

101. Borowitz D, Goss CH, Limauro S, et al. Study of a novel pancreatic enzymereplacement therapy in pancreatic insufficient subjects with cystic fibrosis.J Pediatr 2006;149:658–62.

102. Borowitz D, Stevens C, Brettman LR, et al. Liprotamase long-term safety andsupport of nutritional status in pancreatic-insufficient cystic fibrosis. J PediatrGastroenterol Nutr 2012;54:248–57.

103. Borowitz D, Stevens C, Brettman LR, et al. International phase III trial ofliprotamase efficacy and safety in pancreatic-insufficient cystic fibrosis patients.J Cyst Fibros 2011;10:443–52.

104. Wood LG, Fitzgerald DA, Gibson PG, et al. Oxidative stress in cystic fibrosis:dietary and metabolic factors. J Am Coll Nutr 2001;20(2 Suppl):157–65.

105. Cantin AM, White TB, Cross CE, et al. Antioxidants in cystic fibrosis. Conclusionsfrom the CF antioxidant workshop, Bethesda, Maryland, November 11–12, 2003.Free Radic Biol Med 2007;42:15–31.

106. Shamseer L, Adams D, Brown N, et al. Antioxidant micronutrients for lungdisease in cystic fibrosis. Cochrane Database Syst Rev 2010;(12):CD007020.

107. Sagel SD, Sontag MK, Anthony MM, et al. Effect of an antioxidant-richmultivitamin supplement in cystic fibrosis. J Cyst Fibros 2011;10:31–6.

108. Mayer-Hamblett N, Ramsey BW, Kronmal RA. Advancing outcome measures for thenew era of drug development in cystic fibrosis. Proc Am Thorac Soc 2007;4:370–7.

109. Ordonez CL, Henig NR, Mayer-Hamblett N, et al. Inflammatory and microbiologicmarkers in induced sputum after intravenous antibiotics in cystic fibrosis. AmJ Respir Crit Care Med 2003;168:1471–5.

110. Ahrens R, Rodriguez S, Yen K, et al. VX-770 in subjects 6 to 11 years with cysticfibrosis and the G551D CFTR genotype. Pediatr Pulmonol 2011;34(Suppl):283.

111. Solomon GM, Konstan MW, Wilschanski M, et al. An international randomizedmulticenter comparison of nasal potential difference techniques. Chest 138:919–28.

112. Boyle MP, Diener-West M, Milgram L, et al. A multicenter study of the effect ofsolution temperature on nasal potential difference measurements. Chest2003;124:482–9.

113. Standaert TA, Boitano L, Emerson J, et al. Standardized procedure formeasurement of nasal potential difference: an outcome measure in multicentercystic fibrosis clinical trials. Pediatr Pulmonol 2004;37:385–92.

114. Davis SD, Rosenfeld M, Kerby GS, et al. Multicenter evaluation of infant lungfunction tests as cystic fibrosis clinical trial endpoints. Am J Respir Crit Care Med2010;182:1387–97.

115. Rosenfeld M, Ratjen F, Brumback L, et al. Inhaled hypertonic saline in infants andchildren younger than 6 years with cystic fibrosis: the ISIS randomized controlledtrial. JAMA 2012;307:2269–77.

116. Davis SD, Brody AS, Emond MJ, et al. Endpoints for clinical trials in youngchildren with cystic fibrosis. Proc Am Thorac Soc 2007;4:418–30.

117. Davis SD, Fordham LA, Brody AS, et al. Computed tomography reflects lowerairway inflammation and tracks changes in early cystic fibrosis. Am J Respir CritCare Med 2007;175:943–50.

118. Brody AS. Computed tomography scanning in cystic fibrosis research trials:practical lessons from three clinical trials in the United States. Proc Am Thorac Soc2007;4:350–4.

119. Moss RB, Rodman D, Spencer LT, et al. Repeated adeno-associated virus serotype2 aerosol-mediated cystic fibrosis transmembrane regulator gene transfer to thelungs of patients with cystic fibrosis: a multicenter, double-blind, placebo-controlledtrial. Chest 2004;125:509–21.

120. Retsch-Bogart GZ, Quittner AL, Gibson RL, et al. Efficacy and safety of inhaledaztreonam lysine for airway pseudomonas in cystic fibrosis. Chest2009;135:1223–32.

121. McCoy KS, Quittner AL, Oermann CM, et al. Inhaled aztreonam lysine for chronicairway Pseudomonas aeruginosa in cystic fibrosis. Am J Respir Crit Care Med2008;178:921–8.

122. Oermann CM, Retsch-Bogart GZ, Quittner AL, et al. An 18-month study of thesafety and efficacy of repeated courses of inhaled aztreonam lysine in cysticfibrosis. Pediatr Pulmonol 45:1121–34.

123. Quittner AL, Modi AC, Wainwright C, et al. Determination of the minimalclinically important difference scores for the Cystic Fibrosis Questionnaire-Revisedrespiratory symptom scale in two populations of patients with cystic fibrosisand chronic Pseudomonas aeruginosa airway infection. Chest 2009;135:1610–18.

124. Geller DE, Flume PA, Staab D, et al. Levofloxacin inhalation solution (MP-376) inpatients with cystic fibrosis with Pseudomonas aeruginosa. Am J Respir Crit CareMed 2011;183:1510–16.

125. Konstan MW, Geller DE, Minic P, et al. Tobramycin inhalation powder forP. aeruginosa infection in cystic fibrosis: the EVOLVE trial. Pediatr Pulmonol2011;46:230–8.

126. Rosenfeld M, Ratjen F, Brumback L, et al. Inhaled hypertonic saline in infants andchildren younger than 6 years with cystic fibrosis: the ISIS randomized controlledtrial inhaled hypertonic saline in children with CF. JAMA 2012:1–9.

127. Rosenfeld M, Emerson J, McNamara S, et al. Baseline characteristics and factorsassociated with nutritional and pulmonary status at enrollment in the cystic fibrosisEPIC observational cohort. Pediatr Pulmonol 2010;45:934–44.


Cystic fibrosis

http://www.digestivecare.com/PDFs/NDA%20022175%20APPROVAL%2017MAY2012_PI&MG.pdf



ORIGINAL ARTICLE

Transient receptor potential channels mediate thetussive response to prostaglandin E2 and bradykinin

Megan Grace,1,2 Mark A Birrell,1,2 Eric Dubuis,1 Sarah A Maher,1,2 Maria G Belvisi1,2,3

ABSTRACTBackground Cough is the most frequent reason forconsultation with a family doctor, or with a general orrespiratory physician. Treatment options are limited anda recent meta-analysis concluded that over-the-counterremedies are ineffective and there is increasing concernabout their use in children. Endogenous inflammatorymediators such as prostaglandin E2 (PGE2) and bradykinin(BK), which are often elevated in respiratory diseasestates, are also known to cause cough by stimulatingairway sensory nerves. However, how this occurs is notunderstood.Methods We hypothesised that the transient receptorpotential (TRP) channels, TRPA1 and TRPV1, may havea role as ‘common effectors’ of tussive responses tothese agents. We have employed a range of in vitroimaging and isolated tissue assays in human, murine andguinea pig tissue and an in vivo cough model to supportthis hypothesis.Results Using calcium imaging we demonstrated thatPGE2 and BK activated isolated guinea pig sensoryganglia and evoked depolarisation (activation) of vagalsensory nerves, which was inhibited by TRPA1 andTRPV1 blockers (JNJ17203212 and HC-030031). Thesedata were confirmed in vagal sensory nerves from TRPA1and TRPV1 gene deleted mice. TRPV1 and TRPA1blockers partially inhibited the tussive response to PGE2and BK with a complete inhibition obtained in thepresence of both antagonists together in a guinea pigconscious cough model.Conclusion This study identifies TRPA1 and TRPV1channels as key regulators of tussive responses elicitedby endogenous and exogenous agents, making them themost promising targets currently identified in thedevelopment of anti-tussive drugs.

INTRODUCTIONCough is the most frequent reason for consultationwith a family doctor,1 or with a general or respi-ratory physician. Patients with chronic coughprobably account for 10e38% of respiratoryoutpatient practice in the USA.2 Chronic cough ofvarious aetiologies is a common presentation tospecialist respiratory clinics, and is reported asa troublesome symptom by 7% of the population.3

Treatment options are limited. A recent meta-analysis concluded that over-the-counter (OTC)cough remedies are ineffective4 and there isincreasing concern about the use of OTC therapiesin children. Despite its importance, our under-standing of the mechanisms which provoke coughis poor.

The respiratory tract is innervated by sensoryafferent nerves which are activated by mechanicaland chemical stimuli.5 Activation of capsaicin-sensitive C-fibres and acid-sensitive, capsaicin-insensitive mechanoreceptors innervating thelarynx, trachea, and large bronchi regulate thecough reflex.5 6 Endogenous inflammatory media-tors are often elevated in respiratory disease states.For example, higher concentrations of prosta-glandin E2 (PGE2)7 and bradykinin (BK)8 have beenfound in the airways of patients with asthma andchronic obstructive pulmonary disease. PGE2 andBK are also known to cause cough by stimulatingairway sensory nerves.9 10 Furthermore, increasedPGE2 levels have been found in idiopathic coughand cough associated with post-nasal drip, gastro-oesophageal reflux disease, cough variant asthmaand eosinophilic bronchitis.11 It has previously beendemonstrated that PGE2 activates guinea pig,mouse and human airway sensory nerves andcauses cough via EP3 receptor activation.10 BKactivates guinea pig airway sensory nerves andelicits cough via activation of the B2 receptor, but itis not known if the same process occurs in otherspecies.9 Although we do have some informationregarding which G-protein-coupled receptors(GPCRs) are activated by these endogenous tussiveagents, it is still unclear what post-receptorsignalling pathways are involved.Recently, ion channels of the transient receptor

potential (TRP) class such as TRPV1 have beenimplicated in the afferent sensory loop of thecough reflex12 13 and in the heightened cough sensi-tivity seen in disease.14 TRPA1 is a Ca2+-permeant

Key messages

What is the key question?< Endogenous mediators which are often elevated

in respiratory disease states, such as PGE2 andbradykinin, are also known to cause cough.However, how this occurs is not known.

What is the bottom line?< Here we have elucidated the signaling mecha-

nisms involved in this tussive response andidentified a role for TRPA1 and TRPV1 channels.

Why read on?< These findings could have major implications for

the treatment of cough which currently presentsa significant unmet medical need.

< Additional materials arepublished online only. To viewthese files please visit thejournal online (http://dx.doi.org/10.1136/thoraxjnl-2011-201443).1Respiratory PharmacologyGroup, Pharmacology andToxicology Section, NationalHeart and Lung Institute, Facultyof Medicine, Imperial CollegeLondon, London, UK2Centre for IntegrativePhysiology and Pharmacology,Imperial College London,London, UK3Respiratory Research Group,University of Manchester,Wythenshawe Hospital,Manchester, UK

Correspondence toProfessor Mara Belvisi, ImperialCollege London, Exhibition Road,London SW7 2AZ, UK;[email protected]

Received 29 November 2011Accepted 9 May 2012

This paper is freely availableonline under the BMJ Journalsunlocked scheme, see http://thorax.bmj.com/site/about/unlocked.xhtml

Cough



Figure 1 Establishing concentration responses for prostaglandin (PGE2) and bradykinin (BK) in the in vitro preparations and in vivo cough model.(AeD) Concentration responses showing increases in intracellular calcium ([Ca2+]i) for PGE2 and BK in primary neurons isolated from guinea pigjugular (A, B) and nodose (C, D) ganglia. In each panel, histograms show an increase in [Ca2+]i for increasing concentrations of tussive agent. To takeinto account multiphasic shapes of some responses and their lengths, the calcium flux (area under curve (AUC)) generated by applications of tussive

Cough


non-selective channel with 14 ankyrin repeats in its aminoterminus which also belongs to the larger TRP family. TRPA1channels are activated by a range of natural products such asallyl isothiocyanate, allicin and cannabinol, found in mustardoil, garlic and cannabis15e17 and by environmental irritants (eg,acrolein, present in air pollution, vehicle exhaust and cigarettesmoke),18e20 and is primarily expressed in small diameter,nociceptive neurons where its activation contributes to theperception of noxious stimuli such as itch.18 20 21 It has beendemonstrated that stimulating TRPA1 channels activates vagalbroncho-pulmonary C-fibres in rodent lung,22e24 inducing a lateasthmatic response in sensitised rodents following allergenchallenge25 and causing cough in guinea pig models and innormal human volunteers.26 Although many exogenous stimuliare known to activate TRPA1 and TRPV1, it is still unknownhow cough and other reflexes are elicited in health and disease byendogenous agents, and whether these ion channels areinvolved. We hypothesised that the TRPA1 and TRPV1 ionchannels may have a role as common effectors for such tussiveagents.

METHODSIsolated vagal gangliaIntracellular free calcium ([Ca2+]i) measurements wereperformed in dissociated jugular and nodose neurons. Thesestudies were performed on all isolated vagal neurons (not airwayspecific), with the concentrationeresponse data representing anoverview of responding and non-responding cells. For subse-quent antagonist studies only responding cells were analysed,with the criteria for a ‘responsive cell’ judged as an increase in[Ca2+]i of $10% of the K50 response. In each case, N ¼ numberof animals and n ¼ number of cells tested. Comprehensivemethods are detailed in online supplementary text.

Isolated vagus nerve preparationGuinea pigs or mice (C57BL/6, Trpa1�/� and Trpv1�/�) weresacrificed by injection of sodium pentobarbitone (200 mg/kgintraperitoneal injection). The vagus nerves were removed andexperiments conducted in our fully characterised isolated vaguspreparation, as described in previous publications.10 26 Humanvagal tissue (n¼6, two men, 27e72-year-old donors with norespiratory disease) was obtained from two sourcesdtransplanttissue surplus to requirements (Harefield Hospital, UK); andpurchased from the International Institute for the Advancementof Medicine (Edison, New Jersey, USA). In all cases, the tissuewas consented for use in scientific research and ethics approvalobtained from the Royal Brompton & Harfield Trust. See onlinesupplementary text for full methods.

Conscious guinea pig cough modelConscious unrestrained guinea pigs were placed in individualplastic transparent whole-body plethysmograph chambers(Buxco, Wilmington, North Carolina, USA) and cough detectedas previously described.10 26

Data analysis and statisticsFor imaging, RM is the maximum response observed expressed asa percentage of the K50 response. EC50 values quoted in theimaging studies are the concentrations of drug that produced 50%of the maximum response obtained. Inhibition of agonistresponses in the isolated vagus nerve preparation was analysed bytwo-tailed paired t test, comparing responses to the agonist in theabsence and presence of an antagonist in the same piece of nerve.Inhibition of cough by TRPA1 and TRPV1 antagonists in vivo

was analysed by KruskaleWallis test for multiple comparisonswith Dunn’s post hoc test, comparing responses from eachgroup of antagonist/vehicle combination to the vehicle-onlycontrol. Data are presented as median 6 IQR, with statisticalsignificance set at p<0.05.

RESULTSCharacterising agonist responses in isolated vagal gangliaCapsaicin and acrolein produced concentration-related increasesin [Ca2+]i in sensory neurons (supplementary figure 1AeD).PGE2 stimulation was multiphasic in both ganglia, of which56.3% of jugular and 40% of nodose neurons responded. Overall,PGE2 increased [Ca2+]i in jugular neurons with an RM of 4169%at 10 mM and an EC50 of 5.0761.0 mM (N¼5, n¼16); whereas, innodose neurons RM was only 1162% at 10 mM with an EC50 of3.1160.4 mM (N¼4, n¼15). 52.9% of jugular neurons and 37.5%of nodose neurons responded to BK stimulation. Overall, BKinduced 2264% RM at 10 mM, with an EC50 of 2.3260.36 mM injugular neurons (N¼5, n¼17); and 1763% RM at 30 mM with anEC50 of 2.260.2 mM in nodose neurons (N¼5, n¼24) (figure1AeD).

Characterising agonist responses in vitro and in vivoCapsaicin and acrolein produced concentration-relatedincreases in depolarisation of guinea pig, mouse and humanvagus nerve (online supplementary figure 1EeG). BK and PGE2concentration dependently activated both guinea pig andmouse isolated vagus nerves, whereas the corresponding vehi-cles did not induce depolarisation (figure 1E,F). BK (3 mM) andPGE2 (10 mM) also activated human afferent sensory nerves(n¼5e6, data not shown). The GPCR mediating the tussiveeffects of PGE2 has already been established as the EP3receptor.10 Here, we show that BK activates only the B2

receptor in human and guinea pig, but B1 and B2 receptors in

agents is normalised, and expressed as percentage of response to K50. The tussive agent used is indicated above each set of histograms and theconcentration below each bar in mM (N¼4e5, n¼15e24). The trace in the lower left shows a typical recording of the light intensity over timefollowing exposure to the agonist. Time and duration of drug application are indicated by a black bar above the trace. Time scale is given by the 1 minlength-equivalent black bar shown below the trace. On the bottom right are display images taken during the recording. Time of the snapshot isindicated below each picture with zero being the start of tussive agent application. The pseudo colour code used for light intensity in the pictures isrepresented on the right of each set of images. (E, F) Perfusion for 2 min of BK (black bars) or PGE2 (white bars) activated (E) guinea pig and (F) mouseisolated vagus nerves in a concentration-dependent manner, measured as depolarisation of the nerve in mV (n¼6). (G) The G-protein coupled receptormediating BK-induced depolarisation (3 mM in guinea pig and 1 mM in mouse tissue) was identified as the B2 receptor in human (n¼1e2) and guineapig (n¼6), and a combination of B1 and B2 receptors in the mouse (n¼6) by incubating the nerve with either B1 (R715, 1 mM; checked bars) or B2(WIN 64338, 10 mM; striped bars) selective antagonists for 10 min, measured as % inhibition of agonist responses. (E) BK (filled circles) and PGE2(open circles) also induced concentration-related coughing in the conscious guinea pig, measured as the total number of coughs counted during 10 minof aerosol stimulation (n¼4e8). Data are expressed as mean 6 SEM of n observations (AeG) or median 6 IQR (H). Statistical significance isindicated by *p<0.05 and **p<0.01, calculated as a paired t-test comparing responses in the same piece of nerve (human data were not analyseddue to low numbers). Veh, vehicle. This figure is produced in colour in the online journaldplease visit the website to view the colour figure.

[Continued]

Cough


Figure 2 Characterisation of transient receptor potential channel A1 (TRPA1)-selective and TRPV1-selective antagonists in the in vitro primary ganglia andisolated vagus nerve preparations. The TRPA1 antagonist HC-030031 (HC) or TRPV1 antagonists JNJ17203212 (JNJ) or capsazepine (CAPZ) were assessedfor their ability to inhibit capsaicin (black bars) and acrolein (white bars) responses in isolated guinea pig jugular neurons and guinea pig, mouse or humanisolated vagus nerves. (A) HC concentration-dependently inhibited acrolein-induced (10 mM) increases in [Ca2+]i in guinea pig isolated jugular neurons, but

Cough


the mouse isolated vagus (figure 1G). It is possible that BK isinducing airway sensory afferent activation and cough viaproduction of prostanoids.27 28 However, incubation of thevagus nerve with indomethacin did not alter BK-induced acti-vation of either the guinea pig (20611% inhibition, p>0.05) orwild-type mouse sensory nerves (13610% inhibition; n¼6,p>0.05; data not shown). The magnitude of BK-inducedsensory nerve depolarisation was also similar in wild-typecompared with EP3�/� mouse vagus (n¼6, p>0.05; data notshown), which is the GPCR through which PGE2 causes cough.Depolarisations to BK, PGE2, acrolein and capsaicin wereabolished with the sodium channel blocker tetrodotoxin (n¼3;100% inhibition, data not shown). Further evidence for theobserved depolarisation being mediated via sensory nerveactivation comes from the in vivo experiments, demonstratingthat BK and PGE2 successfully induce concentration-relatedcoughing in conscious guinea pigs (figure 1H).

Characterising antagonist responses in isolated vagal gangliaConcentration responses for the TRPV1-selective antagonistJNJ17203212 (JNJ) and TRPA1-selective antagonist HC-030031(HC) were established in primary jugular cells for their ability toinhibit agonist-induced increases in [Ca2+]i (figure 2A). JNJconcentration-dependently inhibited increases in [Ca2+]i causedby the TRPV1-selective agonist capsaicin, with a maximal effectof 8662% at 10 mM. Alternatively, HC concentration-dependentlyinhibited increases in [Ca2+]i induced by the TRPA1-selectiveagonist acrolein, with a maximal effect of 7668% at 0.1 mM. Atthe concentration which caused maximal inhibition of its ownreceptor, 10 mM JNJ did not inhibit acrolein, and 0.1 mM HC didnot inhibit capsaicin stimulation of jugular cells (figure 2A).

Characterising TRP-selective antagonists in vitroDepolarisation of guinea pig and mouse vagus nerve by acroleinwas concentration-dependently inhibited with the TRPA1-selective antagonist HC. Similarly, capsaicin responses wereconcentration-dependently inhibited by the TRPV1-selectiveantagonists capsazepine (CAPZ) and JNJ (figure 2B,C). At theconcentration which maximally inhibited acrolein, HC (10 mM)did not inhibit capsaicin-induced nerve depolarisation; andequally CAPZ (10 mM) and JNJ (100 mM) did not inhibit acrolein-induced nerve depolarisation (figure 2B,C). This suggests thatthese compounds are not exhibiting off-target actions at theseconcentrations. Subsequently, the effects of HC (10 mM) and JNJ(100 mM) were investigated in human isolated vagus. In theseexperiments (n¼2e3), acrolein responses were abolished by HCbut not affected by JNJ; whereas, capsaicin responses wereabolished by JNJ but not affected by HC (example traces shownin figure 2D). Vehicle control (0.1% dimethyl sulfoxide (DMSO)vol/vol) did not inhibit agonist responses (data not shown).

Determining the role of TRPA1 and TRPV1 in PGE2 and BKinduced vagal ganglia and sensory nerve activation in vitroHaving characterised the available tools, and confirmedselectivity of the antagonists, the role of TRPA1 and TRPV1 in

PGE2-induced and BK-induced vagal ganglia and nerve stimula-tion was established. HC (0.1 mM) or JNJ (10 mM) partiallyinhibited PGE2-induced (5564% and 4069%, respectively) orBK-induced (4565% and 4667%, respectively) increases in [Ca2+]i in guinea pig primary cells isolated from jugular vagal ganglia(p<0.01). Furthermore, when used in combination, HC and JNJinhibited PGE2-induced [Ca2+]i elevation by 8863% and BK by80612% (p<0.0001). In contrast, vehicle incubation had noeffect on [Ca2+]i (�1611% for PGE2 and �468% for BK;p>0.05) (figure 3A,B).TRPA1 antagonism with HC partially inhibited PGE2 and BK

responses in the guinea pig isolated vagus nerve (4465% and4763%, respectively). Additionally, TRPV1 antagonism withCAPZ or JNJ also partially inhibited PGE2 (4563% and 4864%,

had no effect on capsaicin (1 mM) at the concentration which maximally inhibited its own receptor (0.1 mM). Similarly, JNJ concentration-dependentlyinhibited capsaicin-induced responses, but had no effect on acrolein at 10 mM (N¼3e4, n¼5e19). (B, C) HC concentration-dependently inhibited acrolein-induced (300 mM) depolarisation of the guinea pig and wild-type mouse isolated vagus nerves, but had no effect on capsaicin (1 mM) stimulation.Conversely, TRPV1 antagonism with capsazepine or JNJ17203212 concentration-dependently inhibited capsaicin-induced depolarisation in guinea-pig andmouse isolated vagus nerves, but had no effect on acrolein stimulation at 10 mM or 100 mM, respectively (n¼6). (D) Representative traces showinginhibition of human vagus nerve depolarisation with 10 mM HC when stimulated with acrolein (300 mM) but not capsaicin (1 mM). Conversely, 100 mM JNJinhibited capsaicin but not acrolein responses (n¼2e3). Black lines represent agonist incubation (2 min) and grey bars antagonist incubation (10 min). Dataare presented as mean 6 SEM of n observations, calculated as % inhibition of agonist responses. *(p<0.05), **(p<0.01) and ***(p<0.0001) indicatestatistical significance, paired t-test comparing responses in the same piece of nerve. Veh, vehicle for the antagonist (0.1% dimethyl sulfoxide).

[Continued]

Figure 3 Determining the role of transient receptor potential channelA1 (TRPA1) and TRPV1 in prostaglandin E2 (PGE2) and bradykinin (BK)induced isolated primary jugular neurons. The TRPA1 antagonist HC-030031 (HC, 0.1 mM; white bars); TRPV1 antagonist JNJ17203212(JNJ, 10 mM; striped bars); and a combination of HC+JNJ (black bars)were assessed for their ability to inhibit (A) 1 mM PGE2 and (B) 10 mMBK responses in isolated guinea pig jugular neurons. HC or JNJ partiallyinhibited PGE2 and BK responses, whereas HC+JNJ almost completelyabolished increases in [Ca2+]i. Data are presented as mean 6 SEM ofN¼3e5, n¼10e19 observations, calculated as % inhibition of agonistresponses. **(p<0.01) and ***(p<0.0001) indicate statistical signifi-cance, paired t-test comparing responses in the same neuron. Veh,vehicle for the antagonist (0.1% dimethyl sulfoxide).

Cough


Figure 4 Determining the role of transient receptor potential channel A1 (TRPA1) and TRPV1 in prostaglandin E2 (PGE2) and bradykinin (BK) inducedsensory nerve activation. The TRPA1 antagonist HC-030031 (HC 10 mM; white bars), TRPV1 antagonists capsazepine (CAPZ 10 mM; grey bars) andJNJ17203212 (JNJ 100 mM; striped bars), and a combination of HC+JNJ (black bars) were assessed for their ability to inhibit PGE2 (10 mM) and BK(3 mM in guinea pig and human, and 1 mM in mouse tissue) isolated vagus nerve responses. (A, B) HC, CAPZ or JNJ partially inhibited PGE2 and BK

Cough


respectively) and BK responses (5563% and 4968%, respec-tively) (p<0.05). When used in combination, HC+JNJ abolishedvagus nerve responses to PGE2 (9064%) and BK (9564%)(p<0.0001; figure 4A,B). To further confirm these results theexperiments were repeated in a second species. The mouse waschosen because we have access to Trpa1�/� and Trpv1�/� mice,and have previously shown that mouse vagus responds ina similar fashion to human isolated vagus.10 26 Knockdown ofthe TRPA1 or TRPV1 gene was confirmed by genotyping (figure4C). Vagal nerve activation induced by acrolein and capsaicinwere initially assessed to ensure phenotypical loss of TRPA1 andTRPV1 responses (data not shown). The results obtained inguinea pigs were then confirmed by comparing the magnitude ofstimulation of the endogenous tussive agents in wild-type miceto that of Trpa1�/� and Trpv1�/� animals. The Trpa1�/� andTrpv1�/� responses to PGE2 and BK stimulation were approxi-mately half those seen in wild-type mouse vagal tissue (p<0.01;data not shown).

In agreement with the results obtained from guinea pig vagus,wild-type mouse nerve responses to PGE2 and BK were partiallyinhibited by HC (4762% and 5165%, respectively), CAPZ(4964% and 4866%, respectively) and JNJ (5465% and 4665%,respectively) (p<0.05). When used in combination, HC+JNJabolished vagus nerve responses to PGE2 (9464%) and BK(9563%) (p<0.0001; figure 4D,E). And when tissue taken fromgenetically modified animals was tested in combination with thealternative antagonist, responses were again virtually abolished.In Trpa1�/� tissue PGE2 was inhibited 9166% and 9464%; andBK 8767% and 9763% by CAPZ and JNJ, respectively (p<0.01;figure 4D,E). In Trpv1�/� tissue, PGE2 was inhibited 9268% andBK 8966% by HC (p<0.01; figure 4D,E).

We were able to confirm the above results in human vagaltissue (n¼2e3). In these experiments, both PGE2 and BK werepartially inhibited by either HC or JNJ antagonism withcomplete inhibition when the antagonists were used in combi-nation. Vehicle control (0.1% DMSO vol/vol) did not inhibitagonist responses (figure 4F,G).

Determining a role for TRPA1 and TRPV1 in PGE2 and BK inducedguinea pig coughConcentration responses for the selective agonists capsaicin(TRPV1; figure 5A) and acrolein (TRPA1; figure 5B) wereinitially established to determine a submaximal concentrationfor which to test the antagonists (60 mM capsaicin and 100 mMacrolein). Subsequently, the maximally effective doses of theselective antagonists HC (TRPA1) and JNJ (TRPV1) which didnot display off-target effects on the alternative receptor weredetermined. HC dose-dependently inhibited acrolein-inducedcough in vivo, but at 300 mg/kg had no effect on capsaicin; andJNJ concentration-dependently inhibited capsaicin-inducedcough but at 100 mg/kg had no effect on acrolein in consciousguinea pigs (figure 5C,D).

The selective antagonists were then tested against PGE2-induced and BK-induced cough, appropriate concentrations ofwhich had been determined earlier (300 mg/ml PGE2 and 3 mg/ml BK) (figure 1H). The in vivo guinea pig cough results agreewith the in vitro findings. When pretreated with vehicle control,PGE2 induced coughs in response to 10 min aerosol stimulation.This was reduced with either HC or JNJ pretreatment, respec-tively. Similarly, pretreatment with HC or JNJ antagonistsreduced BK-induced coughing compared with vehicle control(figure 5E,F). When pretreated with a combination of HC+JNJ,the cough responses to PGE2 and BK were completely abolished(figure 5E,F).

DISCUSSIONDespite its importance, our understanding of the mechanismswhich provoke cough and the endogenous tussive agentsinvolved in health and disease is poor. Chronic cough is oftenassociated with an underlying inflammatory condition, as inasthma and chronic obstructive pulmonary disease, but theendogenous mediators and signal transduction pathways whichinitiate cough are not known. Inflammatory diseases are asso-ciated with enhanced release of inflammatory mediators in theairways.7 8 Two such mediators are PGE2 and BK, which havebeen shown to induce coughing in humans29 30 and animals.9 10

Interestingly, cough associated with patients who take angio-tensin-converting enzyme (ACE) inhibitors has also beensuggested to be due to the increased levels of bradykinin.9

Furthermore, PGE2 levels have been found to be elevated ininduced sputum of patients with chronic cough.11 It has previ-ously been demonstrated that PGE2 activates guinea pig, mouseand human airway sensory nerves and evokes cough in guineapigs via the EP3 receptor.10 However, BK stimulates guinea pigsensory nerves and elicits cough via activation of the B2

receptor.9 31 In these studies we demonstrate that BK and PGE2are able to activate sensory jugular ganglia; depolarise guinea pig,mouse and human vagal afferents; and evoke cough in a guineapig model in a concentration-related fashion. Interestingly, thebradykinin B2 receptor mediated sensory nerve activation in theisolated guinea pig and human vagal nerve assays but the B1receptor also played a role in the BK-induced activation of themouse vagus, highlighting a species difference.To induce coughing, post-receptor signalling pathways

downstream of GPCR coupling are likely to cause the opening ofmembrane-bound ion channels leading to activation of airwaysensory nerves and subsequent coughing. Previously othergroups have presented data implicating the TRP family of ionchannels in sensory nerve activation and the cough reflex elicitedby BK.32 In these studies we have confirmed these data andextended these findings by confirming a partial inhibition of BK-induced sensory nerve activation by TRPV1 antagonists. Similarresults were obtained with PGE2, with the TRPV1 antagonists

responses in isolated guinea pig vagus tissue, whereas, HC+JNJ almost completely abolished nerve activation. (C) Knockdown of the TRPA1 orTRPV1 gene was verified by genotyping. Bands were expected at 317 bp for wild-type and 184 bp for Trpa1�/�; and 984 bp for wild-type and 600 bpfor Trpv1�/� mice. C, water (negative control); bp, base pair. (D, E) HC, CAPZ or JNJ partially inhibited PGE2 and BK responses in isolated wild-typemouse vagus tissue, whereas, HC+JNJ almost completely abolished nerve activation. In agreement with this, sensory nerves taken from geneticallymodified mice Trpa1�/� or Trpv1�/� tested in combination with the alternative TRPV1 or TRPA1 antagonist also largely eliminated sensory nerveresponses to PGE2 and BK. (F, G) HC and JNJ partially inhibited PGE2 and BK responses in human isolated vagal tissue, whereas, HC+JNJ abolishednerve depolarisation. Example traces are shown above, where black lines represent agonist incubation (2 min), and grey bars represent antagonistincubation (10 min). Scatter plots of % inhibition are shown below and time and magnitude scales for the traces are shown in the top left hand corner.Data are presented as mean 6 SEM of n¼6 observations for guinea pig and mouse experiments, and n¼2e3 for human experiments, calculated as %inhibition of agonist responses. *(p<0.05), **(p<0.01) and ***(p<0.0001) indicate statistical significance, paired t-test comparing responses in thesame piece of nerve. Veh, vehicle for the antagonist (0.1% dimethyl sulfoxide).

[Continued]

Cough


Figure 5 Determining the role of transient receptor potential channel A1 (TRPA1) and TRPV1 in prostaglandin E2 (PGE2) and bradykinin (BK)-induced coughin conscious guinea pigs. (A) Capsaicin and (B) acrolein concentration-dependently induced coughing in conscious, unrestrained guinea pigs. Tussive agentswere aerosolised for 5 min, the number of coughs was counted during this time and for a further 5 min post stimulation (10 min total). Data are presented asmean 6 SEM of n¼10e12 observations. (C, D) Animals received intraperitoneal injections with a concentration of TRPA1 antagonist HC-030031 (HC),TRPV1 antagonist JNJ17203212 (JNJ) or vehicle (Veh) 1 h prior to 5 min aerosol stimulation with a tussive agonist. The number of coughs was countedduring the 5 min stimulation plus a further 5 min (10 min total). (C) HC concentration-dependently inhibited acrolein-induced coughing (100 mM; opencircles), but had no effect on capsaicin cough (60 mM; filled circles) at 300 mg/kg. (D) Conversely, JNJ concentration-dependently inhibited capsaicin-induced cough, with no effect on acrolein at 100 mg/kg. Data are presented asmean6 SEM of n¼8e10 observations. (E, F) Animals received intraperitonealinjection with HC (300 mg/kg; filled circles), JNJ (100 mg/kg; filled squares), a combination of both antagonists (HC+JNJ; filled triangles), or appropriate Veh(open circles) 1 h prior to stimulation with a tussive agonist. (E) PGE2 (300 mg/ml) or (F) BK (3 mg/ml) were aerosolised for 10 min, during which time thenumber of coughs was counted. Compared with vehicle control, pretreatment with either HC or JNJ significantly inhibited PGE2-induced or BK-inducedcoughing; and pr-treatment with HC+JNJ abolished cough altogether. Data are presented as median 6 IQR of n¼10e12 observations. *(p<0.05),**(p<0.01) and ***(p<0.0001) indicate statistical significance, KruskaleWallis one-way analysis of variance with Dunn’s multiple comparison post-test.

Cough


producing partial inhibition of vagal sensory nerve activation inall species, guinea pig jugular neurons, and in a guinea pig coughmodel. At concentrations which were selective for inhibition ofTRPA1 ligands (and not TRPV1 ligands) we also found that theTRPA1 antagonist, HC-030031 inhibited BK-induced and PGE2-induced [Ca2+]i in jugular neurons, vagal sensory nerve activa-tion and cough. Furthermore, both antagonists together(JNJ17203212 and HC-030031) completely inhibited PGE2 andBK in vitro and in vivo responses. In vitro pharmacologicalsensory nerve studies were confirmed in tissue from Trpv1�/�

and Trpa1�/� gene deleted mice.The mechanisms downstream of GPCR coupling that lead to

either sensitisation or activation of ion channels are not yet fullyunderstood, but phospholipase C (PLC) and protein kinase A(PKA) pathways are thought to be important in the signallingfor a number of TRP channels.32e34 GPCR binding to Gq-coupledreceptors can lead to activation of PLC, hydrolysis of phospha-tidylinositol-(4,5)-biphosphate (PIP2) to yield inositol-(1,4,5)-triphosphate (IP3), production of diacylglycerol (DAG) andactivation of phosphokinase C (PKC). PKC and DAG have beenfound to directly bind the TRPV1 receptor; and IP3-inducedrelease of intracellular calcium stores may be involved in acti-vation of TRPA1. Moreover, PIP2 is thought to constitutivelyinhibit TRP receptors. Therefore, its hydrolysis by PLC maydisinhibit these ion channels, sensitising them to subsequentstimulation.34 Alternatively, PKA-dependent phosphorylationcan occur through activation of Gs-coupled receptors, therebyenhancing ion channel excitability.34

A number of the functional responses elicited by BK arecaused via indirect effects, including the release of otherendogenous mediators downstream of arachadonic acid. Weestablished here that the stimulatory effects of BK on isolatedvagus nerves were not due to subsequent release of prostanoidsby using the general cyclo-oxygenase inhibitor indomethacin.However, a number of studies have also implicated downstreamrelease of lipoxygenase products. For example, BK evokes therelease of 15-HETE from airway epithelial cells.27 Furthermore,12-lipoxygenase and 5-lipoxygenase products have been impli-cated in BK-induced stimulation of airway afferent nerveterminals via TRPV1 channel activation.35 It is therefore plau-sible that BK may be causing cough via the release of lipox-ygenase products downstream of arachadonic acid.

In this paper, we have used in vitro cellular and tissue prep-arations, and an in vivo animal model to investigate the coughreflex. The data generated with these models are useful inattempting to understand cough; however, each model has itslimitations. In the calcium-imaging preparation, we cannotdetermine if there are phenotypical changes induced in theprimary ganglia cells during the isolation process. One of thebenefits of the isolated vagus nerve preparation is that we canparallel our animal experiments in human tissue. However, theagents being tested are applied to the axon of the vagus nerve(not the nerve endings), meaning that the extracellular depo-larisation signal recorded represents a summation of the changein membrane potential of all the nerve fibres being carried by thevagus. In addition, receptor expression and signal transductionmechanisms may differ from those at the peripheral endings.Finally, though the conscious guinea pig cough model is gener-ally considered to be a valid tool for studying the cough reflex,there are a number of reported differences between the actionsof certain drugs in guinea pig and man. These differences couldbe due to strong tachykinin-driven responses via sensory nervesin the guinea pig airways or because potential anti-tussives havebeen trialled in clinical studies in which cough was not the

primary endpoint and where there was no objective coughmonitoring and so any efficacy may have been hard to capture.Furthermore, in guinea pig studies compounds are often not doselimited as they are in the clinic due to safety concerns.The findings presented here are important for our under-

standing of the cough reflex (and in particular inflammatory andACE inhibitor induced cough) and strongly support a role forTRPA1 and TRPV1 as common effectors of the tussive responseto endogenous tussive agents. These studies were conducted intissues and in vivo models under ‘normal’ physiological condi-tions and so a role for these TRP channels has not been estab-lished under pathophysiological conditions. However, it hasbeen shown that patients suffering from chronic cough exhibitan increased TRPV1 expression within the lungs, which wascorrelated with an increase in cough sensitivity to capsaicinchallenge,14 indicating that TRP channels could be commoneffectors of tussive responses in disease and that these channelscould be associated with long-term potentiation of the coughreflex. Studies have not yet been conducted to show if TRPA1 isoverexpressed in pathological cough in man because suitableantibodies are not available but these will be important studiesto perform when appropriate tools are developed. Currentresearch investigating the pathogenesis of cough supports thedevelopment of TRP channel inhibitors as novel and selectivetreatment modalities.

Contributors MGB and MAB conceived and designed the studies; MG, ED, SAM,performed all the experiments and provided intellectual input; MGB wrote themanuscript.

Funding MAB, SAM and MG were funded by project grants from the MedicalResearch Council (MRC, UK) (MAB, G0800196; SAM, MG, G0800195). The humantissue experiments in this study were undertaken with the support of the NIHRRespiratory Disease Biomedical Research Unit at the Royal Brompton and HarefieldNHS Foundation Trust and Imperial College London. ED was funded by a WellcomeTrust project grant (089301/Z/09/Z).


Ethics approval Ethics approval was provided by Royal Brompton & Harefield Trust.


REFERENCES1. McCormick A, Fleming DM, Charlton J; Office of Population Censuses and Surveys.

Morbidity Statistics from General Practice, Fourth National Study 1991e1992 (SeriesMB5 No 3). London: HMSO, 1995.

2. Irwin RS, Corrao WM, Pratter MR. Chronic persistent cough in the adult: thespectrum and frequency of causes and successful outcome of specific therapy. AmRev Respir Dis 1981;123:413e17.

3. Ford AC, Forman D, Moayyedi P, et al. Cough in the community: a crosssectional survey and the relationship to gastrointestinal symptoms. Thorax2006;61:975e9.

4. Schroeder K, Fahey T. Systematic review of randomised controlled trials of over thecounter cough medicines for acute cough in adults. BMJ 2002;324:329e31.

5. Canning BJ, Chou Y-L. Cough sensors. I. Physiological and pharmacologicalproperties of the afferent nerves regulating cough. Handb Exp Pharmacol2009;187:23e47.

6. Nasra J, Belvisi MG. Modulation of sensory nerve function and the cough reflex:understanding disease pathogenesis. Pharmacol Ther 2009;124:354e75.

7. Profita M, Sala A, Bonanno A, et al. Increased prostaglandin E2 concentrations andcyclooxygenase-2 expression in asthmatic subjects with sputum eosinophilia. JAllergy Clin Immunol 2003;112:709e16.

8. Baumgarten CR, Lehmkuhl B, Henning R, et al. Bradykinin and other inflammatorymediators in BAL-fluid from patients with active pulmonary inflammation. AgentsActions Suppl 1992;38:475e81.

9. Fox AJ, Lalloo UG, Belvisi MG, et al. Bradykinin-evoked sensitization of airwaysensory nerves: a mechanism for ACE-inhibitor cough. Nat Med 1996;2:814e17.

10. Maher SA, Birrell MA, Belvisi MG. Prostaglandin E2 mediates cough via the EP3receptor: implications for future disease therapy. Am J Respir Crit Care Med2009;180:293e8.

11. Birring SS, Parker D, Brightling CE, et al. Induced sputum inflammatory mediatorconcentrations in chronic cough. Am J Respir Crit Care Med 2004;169:15e19.

12. Laude EA, Higgins KS, Morice AH. A comparative study of the effects of citric acid,capsaicin and resiniferatoxin on the cough challenge in guinea-pig and man. PulmPharmacol 1993;6:171e5.

Cough


13. Lalloo UG, Fox AJ, Belvisi MG, et al. Capsazepine inhibits cough induced bycapsaicin and citric acid but not by hypertonic saline in guinea pigs. J Appl Physiol1995;79:1082e7.

14. Groneberg DA, Niimi A, Dinh QT, et al. Increased expression of transient receptorpotential vanilloid-1 in airway nerves of chronic cough. Am J Respir Crit Care Med2004;170:1276e80.

15. Story GM, Peier AM, Reeve AJ, et al. ANKTM1, a TRP-like channel expressed innociceptive neurons, is activated by cold temperatures. Cell 2003;112:819e29.

16. Jordt SE, Bautista DM, Chuang HH, et al. Mustard oils and cannabinoids excitesensory nerve fibres through the TRP channel ANKTM1. Nature 2004;427:260e5.

17. Bandell M, Story GM, Hwang SW, et al. Noxious cold ion channel TRPA1 isactivated by pungent compounds and bradykinin. Neuron 2004;41:849e57.

18. Bautista DM, Jordt SE, Nikai T, et al. TRPA1 mediates the inflammatory actions ofenvironmental irritants and proalgesic agents. Cell 2006;124:1269e82.

19. Trevisani M, Siemens J, Materazz S, et al. 4-Hydroxynonenal, an endogenousaldehyde, causes pain and neurogenic inflammation through activation of the irritantreceptor TRPA1. Proc Natl Acad Sci U S A 2007;104:13519e24.

20. Andre E, Campi B, Materazzi S, et al. Cigarette smoke-induced neurogenicinflammation is mediated by alpha, beta-unsaturated aldehydes and the TRPA1receptor in rodents. J Clin Invest 2008;118:2574e82.

21. Wilson SR, Gerhold KA, Bifolck-Fisher A, et al. TRPA1 is required for histamine-independent, Mas-related G protein-coupled receptor-mediated itch. Nat Neurosci2011;14:595e602.

22. Bessac BF, Sivula M, von Hehn CA, et al. TRPA1 is a major oxidant sensor in murineairway sensory neurons. J Clin Invest 2008;118:1899e910.

23. Nassenstein C, Kwong K, Taylor-Clark T, et al. Expression and function of the ionchannel TRPA1 in vagal afferent nerves innervating mouse lungs. J Physiol2008;586:1595e604.

24. Taylor-Clark TE, McAlexander MA, Nassenstein C, et al. Relative contributions ofTRPA1 and TRPV1 channels in the activation of vagal bronchopulmonary C-fibres bythe endogenous autocoid 4-oxononenal. J Physiol 2008;586:3447e59.

25. Raemdonck K, de Alba J, Birrell MA, et al. A role for sensory nerves in the lateasthmatic response. Thorax 2012;67:19e25.

26. Birrell MA, Belvisi MG, Grace M, et al. TRPA1 agonists evoke coughing in guinea pigand human volunteers. Am J Respir Crit Care Med 2009;180:1042e7.

27. Salari H, Chan-Yeung M. Release of 15-hydroxyeicosatetaenoic acid (15-HETE) andprostaglandin E2 (PGE2) by cultured human bronchial epithelial cells. Am J Respir CellMol Biol 1989;1:245e50.

28. Ellis K, Fozard J. Species differences in bradykinin receptor-mediated responses ofthe airways. Auton Autacoid Pharmacol 2002;22:3e16.

29. Fuller RW, Dixon CM, Cuss FM, et al. Bradykinin-induced bronchoconstriction inhumans. Mode of action. Am Rev Respir Dis 1987;135:176e80.

30. Costello JF, Dunlop LS, Gardiner PJ. Characteristics of prostaglandin induced coughin man. Br J Clin Pharmacol 1985;20:355e9.

31. Fox AJ, Barnes PJ, Urban L, et al. An in vitro study of the properties of single vagalafferents innervating guinea-pig airways. J Physiol 1993;469:21e35.

32. Chuang HH, Prescott ED, Kong H, et al. Bradykinin and nerve growth factor release thecapsaicin receptor from PtdIns(4,5)P2-mediated inhibition. Nature 2001;411:957e62.

33. Wang S, Dai Y, Fukuoka T, et al. Phospholipase C and protein kinase A mediatebradykinin sensitisation of TRPA1: a molecular mechanism of inflammatory pain.Brain 2008;131:1241e51.

34. Bessac BF, Jordt SE. Breathtaking TRP channels: TRPA1 and TRPV1 in airwaychemosensation and reflex control. Physiology (Bethesda) 2008;23:360e70.

35. Carr MJ, Kollarik M, Meeker SN, et al. A role for TRPV1 in bradykinin-inducedexcitation of vagal airway afferent nerve terminals. J Pharmacol Exp Ther2003;304:1275e9.


Cough


ORIGINAL ARTICLE

Chronic obstructive pulmonary disease amongresidents of an historically industrialised area

Anthony C Darby,1 Judith C Waterhouse,2,3 Vivien Stevens,4 Clare G Billings,3

Catherine G Billings,2,3 Clare M Burton,1 Charlotte Young,5 Jeremy Wight,6

Paul D Blanc,7 David Fishwick1

ABSTRACTObjective To assess the contribution of workplaceexposures to chronic obstructive pulmonary disease(COPD) risk in a community with a heavy burden of pastindustrial employment.Methods A random population sample of Sheffield, UKresidents aged over 55 years (n¼4000), enriched witha hospital-based supplemental sample (n¼209), wasapproached for study. A comprehensive self-completedquestionnaire elicited physician-made diagnoses, currentsymptoms, and past workplace exposures. The latterwere defined in three ways: self-reported exposure tovapours, gases, dusts and fumes (VGDF); response toa specific exposure checklist; and through a job exposurematrix (JEM) assigning exposure risk likelihood based onjob history independent of respondent-reportedexposure. A subset of the study group underwent lungfunction testing. Population attributable risk fractions(PAR%), adjusted for age, sex and smoking, werecalculated for association between workplace exposureand COPD.Results 2001 (50%) questionnaires were returned fromthe general population sample and 60 (29%) by thehospital supplement. Among 1754 with completeoccupational data, any past occupational exposure toVGDF carried an adjusted excess risk for report ofa physician’s diagnosis of COPD, emphysema, or chronicbronchitis (ORs 3.9; 95% CI 2.7 to 5.8), witha corresponding PAR% value of 58.7% (95% CI 45.6% to68.7%). The PAR% estimate based on JEM exposurewas 31%. From within the subgroup of 571 thatunderwent lung function testing, VGDF exposure wasassociated with a PAR% of 20.0% (95% CI �7.2 to40.3%) for Global initiative for chronic Obstructive LungDisease (GOLD) 1 (or greater) level of COPD.Conclusion This heavy industrial community-basedpopulation study has confirmed significant associationsbetween reported COPD and both generic VGDF andJEM-defined exposures. This study supports thepredominantly international evidence-based notion thatworkplace conditions are important when consideringthe current and future respiratory health of theworkforce.

INTRODUCTIONChronic obstructive pulmonary disease (COPD) isa common disease, associated with substantialmorbidity, mortality, direct and indirect healthcarecosts, including COPD-related absence fromwork.1e7 While the primary contributor to COPD

risk overall is tobacco smoking, other importantcauses are also recognised, including harmful occu-pational and environmental exposures.8 Indeed, the‘cause’ of COPD is likely to be multifactorial inmany cases, reflecting complex exposureehostinteractions. Thus, a more comprehensive under-standing of the relative contribution to COPDcausation from occupational exposures offers animportant platform on which to construct targetedand effective interventions to reduce the burden ofdisease.The role of occupational exposures in the devel-

opment of COPD (including chronic bronchitis)has long been identified; landmark reports from thenineteenth century explored this relationship.9e11

This concept was developed further in the twen-tieth century, with research establishing a linkbetween dusty work and the development ofchronic bronchitis12 and consolidating the viewthat occupational exposures were an important riskfactor for developing obstructive airway disease.13

Many more recent studies have also identified thatworkplace exposures to vapours, gases, dusts andfumes (VGDF) are potentially harmful to lunghealth and contribute to the overall burden ofCOPD.14e17 These workplace-based exposures havealso been shown to interact with tobacco exposure,in certain studies increasing risk to a degree thatmay be more than simply additive.14 15

Consistent estimates place the size of this occu-pational contribution to COPD at approximately15% of the total burden of the disease.18 19 Eventhough this estimate varies among individual

Key messages

What is the key question?< To what extent does occupation contribute to

chronic obstructive pulmonary disease (COPD)causation in an industrialised area of the UK.

What is the bottom line?< In this population-based study, a significant

association was found between COPD andprevious occupational exposures.

Why read on?< As a cause of significant morbidity and

mortality, all contributing factors to COPDcausation should be considered.

< Additional materials arepublished online only. To viewthese files please visit thejournal online (http://dx.doi.org/10.1136/thoraxjnl-2011-200543).1Centre for Workplace Health,Health and Safety Laboratory,Buxton, UK2Centre for Workplace Health,University of Sheffield, UK3Respiratory Function Unit,Royal Hallamshire Hospital,Sheffield, UK4Scientific Computing andInformatics, Medical Imagingand Medical Physics, RoyalHallamshire Hospital, Sheffield,UK5Mathematical Sciences Unit,Health and Safety Laboratory,Buxton, UK6Sheffield NHS Primary CareTrust, Sheffield, UK7Department of Medicine,University of California, SanFrancisco, USA

Correspondence toDr Anthony Darby, Centre forWorkplace Health, Health andSafety Laboratory, Buxton SK179JN, UK;[email protected]

Received 30 June 2011Accepted 11 April 2012

Chronic obstructive pulmonary disease



studies, this figure gives a broad sense of the future COPD burdenthat might be avoidable were all harmful occupational exposuresremoved. This represents an important public health issue, withthe future possibility to reduce incident cases, and potentially, toslow progression in those with already established COPD.

Very few modern analyses have examined UK populations.20

We therefore describe the results of an epidemiological study toassess this contribution, based in Sheffield, UK.

METHODSInitial questionnaire phaseA random population sample of 4000 Sheffield residents, agedover 55 (with no upper age limit or specific exclusion criteria), wasidentified from health records in a specific area within the city. Inmultiple waves over a 12-month period, postal questionnaireswere sent to potential participants covering demographics, health,and occupational history, including work exposures. Work expo-sure was categorised in three ways: based on exposure to vapours,gas, dust, or fumes, exposure to a checklist of specific exposures,and lastly, by assignment of exposure likelihood to COPD-causingagents based on a job exposure matrix (JEM) independent ofthe respondent’s self-reported exposures. Social deprivationwas assessed based on the proportion of individuals withina participant’s post (zip) code receiving income support (%IS).

Airway disease case definition and follow-up assessment‘Probable’ cases of airway disease required report of a physician’sdiagnosis of COPD (including emphysema or chronic bronchitis)or asthma or, alternatively, Medical Research Council21 grade 3dyspnoea and another respiratory symptom (wheeze, chesttightness, winter cough/phlegm). ‘Possible’ cases lacked a physi-cian’s diagnosis but reported dyspnoea or respiratory symptoms.We retained such possible cases in the ‘No reported diagnosisgroup’ and did not analyse them separately, but did use this asa basis of exclusion for recruitment for spirometry. Thus, onlyprobable cases and those without symptoms or a diagnosis(referents) were recruited for home-based spirometric assessment,with an option for testing at a hospital-based lung functionlaboratory.

Supplemental casesWe enriched the study population with additional cases ofCOPD (n¼209) assessed at the same hospital-based lung func-tion laboratory that the population-based sample had the optionof attending in lieu of home visit spirometry. These participantscompleted the same questionnaire as noted above.

Data analysisWe used logistic regression analysis to test the associations amongsmoking, occupational exposures and COPD by calculating OddsRatios (OR) unadjusted and adjusted for age, sex and smoking.We defined COPD based on self-reported COPD, emphysema orchronic bronchitis (then repeated these excluding bronchitisalone). Separate analyses were conducted on the subpopulationwith lung function, COPD being defined according to the Globalinitiative for chronic Obstructive Lung Disease (GOLD) criteria.1

We derived population attributable risk fractions (PAR%) esti-mates and their 95% CIs from adjusted ORs.

Additional details of study methods are provided in an onlinesupplement.

RESULTSA total of 2001 (50%) of 4000 questionnaires were returned fromthe random population sample, 1587 (39.7%) from the first mail

out, and a further 414 (10.4%) from the second. The 50%response rate is conservative, as the non-responders include atleast 260 known by the second mailing to have moved or to havedied and 95 others were otherwise determined to be non-eligible(together indicating a minimum response rate of 55% among theeligible).Despite the obvious absence of clinical data for non-responders

from the population sample, age, gender, and deprivation datawere available. Responders (mean age at randomisation 68.5,SD 8.8) were significantly younger (p<0.001) than the 1999 non-responders (69.9, SD 10.0). There was no statistically significantdifference by gender; responders had significantly less (p<0.001)social deprivation: responders %IS of 20.5% (SD 16.8) comparedwith non-responders (25.7%, SD 18.2).Sixty (29%) of the 209 patients in the supplemental sample

participated. As cases were identified from their physiologyrequest card prior to attendance, reasons for non-inclusion werevaried and included non-attendance or refusal. Figure 1 showsa more detailed breakdown of the study numbers within eachCOPD group.Of the total 2061 (population and enriched) participants, 1579

reported no doctor ’s diagnosis of chronic airway disease. Of theremaining, 119 (5.8%) reported a doctor ’s diagnosis of COPD, 50of whom also reported concomitant asthma. Eighty-three (4.0%)reported emphysema and 123 (6.0%) chronic bronchitis. Table 1presents demographics, smoking and symptom reporting databy reported diagnosis.Table 2 provides a detailed occupational exposure profile,

including JEM-derived exposure risk assignments, across themain diagnostic categories. Excluding those with missing orinsufficient occupational or smoking data, 1798 participantsremained for this analysis. Significant differences are presentamong the diagnostic groups for self-reported and JEM-derivedexposure assessment. Comparing self-reported VGDF exposureand JEM estimates, 75.9% of those who reported ever havingVGDF exposure received a JEM rating compatible with work-placed exposures of high or intermediate COPD risk.Table 3 presents the ORs and PAR% values relating occupa-

tional exposures and smoking to COPD. Ever exposure to VGDFwas associated with an excess risk of COPD with ORs of 3.7 and3.9 (with or without the exclusion of chronic bronchitis in theCOPD definition, respectively) following adjustment for age, sexand smoking (corresponding PAR% values of 57.6 and 58.7%). Aseparate sensitivity analysis, excluding all those with concomi-tant asthma from the any COPD category, yielded a PAR% of60.4%. The JEM-associated PAR% (combining intermediate andhigh-likelihood exposure jobs) was 30.8% for broadly definedCOPD. Excluding chronic bronchitis, this estimate was 12.1%(discounting intermediate JEM risk which was not statisticallysignificant).The smoking-associated PAR% estimates adjusted for VDGF

and demographics were 44.7% and 52.8% for any COPD and anyCOPD excluding chronic bronchitis alone, respectively.Adjusting for JEM instead of VGDF, slightly higher smoking-associated PAR% values were obtained, as shown in table 3.Certain differences emerged in sex-stratified analyses. For

example, using the broader any COPD definition, the effect ofVGDF-associated PAR% for men was estimated to be 59.3%,while for women it was 50.0%.In addition to broadly defined VGDF exposure, adjusted ORs

also manifested an excess risk for any COPD for the majority ofthe specific occupational exposures analysed. ORs of 2.0 orgreater were identified for: cadmium fumes, batteries or silversolder (n exposed¼66, OR 2.8, 95% CI 1.5 to 5.1); incinerators,



boilers or oil refineries (n exposed¼118, OR 2.3 95%, CI 1.4 to3.6); irritant gases, for example, chlorine or ammonia (nexposed¼147, OR 2.1, 95% CI 1.4 to 3.3); coal dust or powder (nexposed¼197, OR 2.0 95%, CI 1.3 to 2.9); wheat, flour or othergrain dusts (n exposed¼99, OR 2.0, 95% CI 1.2 to 3.4) and wooddust or sawdust (n exposed¼168, OR 2.0, 95% CI 1.3 to 3.1).The more broadly defined exposure categories of organic dust(OR 2.0), inorganic dust (OR 2.5), and combustion by-products(OR 1.4) also were each associated with significantly (p<0.05)increased odds of any COPD. Steel industry work increased theodds of any COPD by 25% (OR 1.26; 95% CI 0.9 to 1.7); and

although this was not statistically significant, the frequency ofthis occupation in the study population yielded a PAR% of10.2% (95% CI �4.5% to 22.8%).A further analysis was carried out excluding the enriched

population from the dataset. Again, the association between anyCOPD and VGDF exposure retained its significance (OR 4.1,95% CI 2.7 to 6.3). Very similar results were observed for anyCOPD excluding chronic bronchitis alone. Further adjustedsupplementary analysis using %IS as a continuous variable inthe VGDF and any COPD analysis identified %IS to be a signif-icant independent predictor of disease (p<0.0001), but its

Figure 1 Subject sampling schematic. COPD, chronic obstructive pulmonary disease; MRC3, Medical Research Council grade 3 shortness of breath.

Table 1 Demographic, smoking and symptom reporting data by predominant airway diagnosis among 2061 participants

Any COPD,* n (%)Any COPD excluding chronicbronchitis alone, n (%) Asthma only, n (%)

No reporteddiagnosis, n (%)

Number 252 (12.2) 165 (8.0) 230 (11.2) 1579 (76.6)

Mean age (SD) 70.7 (8.3) 70.7 (7.7) 68.1 (8.1) 69.3 (8.9)

(95% CI) (69.6 to 70.7) (69.5 to 71.8) (67.1 to 69.2) (68.7 to 70.2)

Men 145 (57.5) 96 (58.2) 90 (39.1) 765 (48.4)

Mean %IS (SD)y 27.1 (16.5) 26.7 (16.0) 22.6 (17.2) 19.2 (16.5)

Current smoker 48 (19.0) 36 (21.8) 20 (8.7) 152 (9.6)

Former smoker 164 (65.1) 111 (67.3) 111 (48.3) 639 (40.5)

Never smoker 38 (15.1) 18 (10.9) 92 (40.0) 762 (48.3)

Smoking status unknown 2 (0.8) 0 7 (3.0) 26 (1.6)

Pack year, mean (SD) 30.5 (28.7) 36.0 (27.8) 15.6 (21.8) 11.7 (17.9)

(n¼243) (n¼162) (n¼214) (n¼1516)

Pack year > 20 136 (56.0) 110 (67.9) 61 (28.5) 347 (22.9)

Shortness of breath (MRC3z) 190 (75.4) 136 (82.4) 122 (53.0) 273 (17.3)

Winter morning cough 170 (67.5) 110 (66.7) 107 (46.5) 218 (13.8)

Winter morning sputum 148 (58.7) 95 (57.6) 78 (33.9) 200 (12.7)

Chest tightness or difficulty breathing 201 (79.8) 141 (85.5) 162 (70.4) 256 (16.2)

Wheezing in last year 194 (77.0) 137 (83.0) 137 (59.6) 244 (15.5)

The differences among the categories are statistically significant (p<0.05) in three-way comparisons (among any chronic obstructive pulmonary disease (COPD) (or any COPD excluding chronicbronchitis), asthma, and no reported diagnosis).*COPD or emphysema or chronic bronchitis, and may include concomitant asthma.yMissing data for mean %Income Support (%IS), n¼132, distributed similarly across all groups.zMedical Research Council grade 3 shortness of breath.



inclusion did not alter the estimate of the effect of VGDFexposure (OR 4.1, PAR% 2.6 to 6.3). A limited sensitivity anal-ysis correcting for late return of the questionnaire was alsocarried out by introducing a dichotomous variable for lateresponse. Addition of this new variable made no significantdifferences to the results as shown in table 3. For any COPD, forexample, the effect of VGDF was retained at an OR of 3.9 (95%CI 2.7 to 5.8).

Table 4 delineates the interaction between the effects ofsmoking in pack years (low (#20) versus high (>20)) andoccupational exposures by calculating adjusted ORs for theseeffects in isolation and in combination. The overall step-up inestimated OR values, at most, approximates the product of thepoint estimate of each risk factor OR, a relationship numericallyconsistent with an additive rather than synergistic relationship,although the confidence limits for these point estimates arewide. It is evident that for any COPD the OR for low pack yearsalone and VGDF alone were approximately equal, and the effect

of both exposures combined was more than additive. The samegeneral pattern was seen for COPD or emphysema.Of respondents from the community population offered

spirometry, 268 (43.0%) of those with probable COPD accepted,compared with 290 (28.1%) of those with no airways disease.Those who underwent spirometry testing tended to be younger(mean age 67.5 compared with 70.1, p<0.001); men (57.5%compared with 39.4%, p<0.001); have less evidence of depriva-tion (mean %IS 18.6 compared with 21.7, p¼0.001), be morelikely to be ever smokers (59.5% compared with 49.1%,p<0.001), be more likely ever to have worked in a job exposed toVGDF (58.2% compared with 36.3%, p<0.001), and have hada job classified as intermediate or high exposure likelihood byJEM assignment (51.6% compared with 44.8%, p¼0.009). Themean values for all lung function values were; forced expiratoryvolume in 1 s (FEV1) 2.20 litres (range 0.42e4.55), forced vitalcapacity (FVC) 3.1 litres (range 0.9e6.0), and FEV1/FVC ratio70.9% (range 25.8e93.1).

Table 2 Occupational factors by predominant diagnosis among 1798 participants with completeoccupational and smoking data

Any COPD*,n (% exposed)

Asthma only,n (% exposed)

No reported diagnosis,n (% exposed)

Number 222 193 1383

Self-reported VGDF exposuredeverexposed

176 (79.3) 110 (57.0) 617 (44.6)

Steel industrydever exposed 110 (49.5) 74 (38.3) 499 (36.1)

Combustion by-productsdever exposed 84 (37.8) 58 (30.1) 361 (26.1)

Inorganic dusts or fumesdever exposed 130 (58.6) 74 (38.3) 471 (34.1)

Organic dustsdever exposed 60 (27.0) 32 (16.6) 196 (14.2)

JEM leveldnot exposedy 64 (28.8) 92 (47.7) 669 (48.4)

JEM leveldintermediate exposure 75 (33.8) 61 (31.6) 416 (30.1)

JEM leveldhigh exposure 83 (37.4) 40 (20.7) 298 (21.5)

The differences among categories are statistically significant (p<0.01) across the three groups.*COPD or emphysema or chronic bronchitis; may also include concomitant asthma.yNon-exposed (also includes those never employed (n¼8)).COPD, chronic obstructive pulmonary disease; JEM, job exposure matrix; VGDF, vapours, gases, dusts or fumes.

Table 3 Risk of COPD related to smoking and VGDF exposure

Exposed cases Exposed non-cases Adjusted OR* (95% CI) PAR% (95% CI)

Any COPD, emphysema or chronic bronchitis, with or without concomitant asthma (216 cases)

Exposure measure 1

VGDF exposure 170 704 3.94 (2.68 to 5.78) 58.7 (45.6 to 68.7)

20 Pack years 185 812 1.74 (1.54 to 1.96) 44.7 (35.3 to 52.7)

Exposure measure 2

JEM exposure

Not exposed 1.0 e

Intermediate 74 462 1.45 (1.00 to 2.11) 10.6 (�1.2 to 20.9)

High 80 328 2.20 (1.45 to 3.35) 20.2 (9.5 to 29.7)

20 Pack years 185 812 1.82 (1.61 to 2.05) 46.2% (37.2 to 53.9)

Any COPD excluding chronic bronchitis alone (149 cases)

Exposure measure 1

VGDF exposure 118 756 3.66 (2.31 to 5.79) 57.6 (40.5 to 69.7)

20 Pack years 133 864 1.90 (1.66 to 2.18) 52.8 (42.2 to 61.4)

Exposure measure 2

JEM exposure

Not exposed 1.0 e

Intermediate 51 485 1.16 (0.75 to 1.81) 4.7 (�11.1 to 18.3)

High 51 357 1.55 (0.94 to 2.54) 12.1 (�2.6 to 24.7)

20 Pack years 133 864 2.01 (1.75 to 2.30) 54.2 (44.1 to 62.4)

A total of 1754 people are included in the analysis shown, with complete VGDF, smoking and JEM code data.*All ORs adjusted for age and sex. Occupational exposure-associated ORs (VGDF or JEM) adjusted for pack years of smoking. Pack years of smoking associated (continuous variable) ORadjusted for occupational exposure (VGDF or JEM) and presented per 20 pack years of exposure.COPD, chronic obstructive pulmonary disease; JEM, job exposure matrix (944 participants classified as intermediate or high exposure); PAR, population attributable risk; VGDF, vapours, gases,dusts or fumes (874 participants exposed).



Table 5 provides PAR% values for COPD defined using GOLD1 level or higher from within the subgroup of the 618 partici-pants who underwent physiology testing (also includessupplemental population). Elevated PAR% values are seen forVGDF exposure (20%; 95% CI �7.2 to 40.3%) and smoking inpack years (45.5%). Of note, the unadjusted increased ORassociated with JEM exposure is attenuated after correction forsmoking, age and gender. A similar analysis to that shown inTable 5, but adjusted for the effects of smoking using a dichot-omous (ever/never) smoking instead of pack years, yieldeda similar OR for VGDF of 1.5 (95% CI 0.99 to 2.3) with anassociated PAR% of 24. Using a more restricted definition ofCOPD (GOLD 2 or worse), the PAR% for the effect of VGDFexposure was estimated to be 14.0%.

DISCUSSIONThese findings add to the accumulating evidence supportinga causal relationship between inhaled, potentially harmfulexposures at work and COPD, and indicate a high PAR% esti-mate relative to a 15% median from other studies14 19 22

Historically, it is likely that there have been heavy levels of VGDFexposure in this study area. A comparable investigation from

Newcastle, UK reported an occupational exposure-associated ORfor COPD of 3.0 with half of the population exposed20; the PAR% extrapolated from these data is 33e50% depending on thespecific formula applied, similar to our estimates. Our findings,along with those of others, suggest that a meaningful proportionof COPD could be prevented in the future by addressing harmfulexposures both directly and by attenuating interactions (additiveor supra-additive) with cigarette smoking.Although the primary aim of this study was not to address

the specific contribution made by steel work exposures to thedevelopment of COPD, this relationship is worthy of mention,given that Sheffield has a historic and current tradition for suchindustry. An earlier iron and steel foundry study identifiedincreased symptoms and decreased airflow in foundry workers,although interpretation is complicated by concomitantpneumoconiosis and asthma.23 Iron foundry workers havea moderate (but non-significant) mortality excess for emphy-sema in the UK24 and in Denmark.25 The European Coaland Steel Community research programme also found a work-associated increase in chronic bronchitis, although lung functiondata did not show an exposure-associated FEV1 deficit.26 Otherstudies of steel workers have observed longitudinal declines in

Table 4 Smoking and occupational exposure as independent and joint associations with COPD

Cigarette smoking/occupational VGDFexposure n Risk of COPD Excess risk Unadjusted OR Adjusted OR* (95% CI)

Any COPD, emphysema or chronic bronchitis, with or without concomitant asthma (231 cases)

Never/no 530 0.02 e 1.00 e

Never/yes 302 0.08 0.06 4.29 5.63 (2.60 to 12.20)

Pack years low/no 248 0.07 0.05 3.59 3.96 (1.77 to 8.89)

Pack years low/yes 279 0.18 0.16 11.63 15.68 (7.62 to 32.28)

Pack years high/no 186 0.15 0.13 8.83 10.44 (4.91 to 22.20)

Pack years high/yes 338 0.31 0.29 23.11 32.04 (15.92 to 64.47)

Any COPD excluding chronic bronchitis alone (155 cases)

Never/no 525 0.01 e 1.00 e

Never/yes 290 0.04 0.03 4.10 5.47 (1.85 to 16.16)

Pack years low/no 239 0.03 0.02 3.14 3.50 (1.10 to 11.18)

Pack years low/yes 253 0.10 0.09 11.40 15.93 (5.85 to 43.34)

Pack years high/no 180 0.12 0.11 13.74 16.47 (6.07 to 44.74)

Pack years high/yes 320 0.27 0.26 38.22 54.11 (20.94 to 139.87)

This table includes only the 1883 respondents who gave complete smoking and VGDF data (participants with missing data for JEM classification were included). For the any COPD excludingchronic bronchitis alone analysis, those with chronic bronchitis have been completely excluded (not included with referents) to avoid misclassification bias (n¼1807).*ORs adjusted for age and sex.JEM, job exposure matrix; pack years low, 20 pack years or less; Pack years high, more than 20; VGDF, vapours, gases, dusts or fumes.

Table 5 Risk of COPD related to smoking and VGDF exposure for spirometry group

Unadjusted OR(95% CI) model 1

Adjusted OR(95% CI) model 2

Adjusted OR(95% CI) model 3 PAR% (95% CI)*

Spirometry alonedGOLD 1 and above (197 cases)

Exposure measure 1

VGDF exposure 2.00 (1.38 to 2.89) 1.84 (1.22 to 2.77) 1.40 (0.91 to 2.15) 20.0 (�7.2 to 40.3)

20 Pack years 2.00 (1.67 to 2.40) 1.94 (1.61 to 2.34) 1.90 (1.57 to 2.29) 45.5 (34.7 to 54.4)

Exposure measure 2

JEM exposure

Not exposed 1.00 1.00 1.00 e

Intermediate or high exposure 1.37 (0.97 to 1.95) 1.11 (0.75 to 1.64) 0.88 (0.58 to 1.34) e

20 Pack years 2.00 (1.67 to 2.40) 1.94 (1.61 to 2.34) 1.96 (1.62 to 2.36) 46.4 (36.1 to 55.1)

Includes 571 who had spirometry with complete occupational exposure data (VGDF and JEM) and smoking data. The OR for smoking is expressed per 20 pack years of exposure.Model 1: unadjusted ORs for the association between COPD and VGDF, JEM-based risk, or cigarette exposure (the identical univariate smoking-associated risk is presented twice).Model 2: occupational exposure (VGDF or JEM based) OR adjusted for age and sex but not pack years smoking. Pack years smoking OR adjusted for age and sex but not for occupationalexposure.Model 3: Occupational exposure OR adjusted for age, sex and pack years smoking. Pack years smoking OR, adjusted for age, sex and occupational exposure (VGDF in upper row, JEM in lowerrow).*All PARs calculated from the results in model 3.GOLD, Global initiative for chronic Obstructive Lung Disease; JEM, job exposure matrix; PAR, population attributable risk; VGDF, vapours, gases, dusts or fumes.



FEV1, but changes have been difficult to separate out fromconcomitant restrictive disease.27e29 Indeed, because of itsprevalence and associated morbidity and mortality, silicosisrather than COPD historically has been the chief focus of non-malignant respiratory disease research in the steel industry.30 Inour study, half of those with a physician’s diagnosis of COPDhad worked in the steel industry, with an associated PAR%indicating that more than 1 in 10 cases were attributable to thisrisk factor, even taking into account cigarette smoking. Werecognise, however, that the CIs surrounding this estimate arewide and should temper the interpretation of this specificfinding. Moreover, our study design does not allow furtherassessment of a more accurate clinical diagnosis, as no chestradiographs were available, and no other investigations toconfirm or exclude asthma specifically were undertaken.

Although this study broadly supports the overall associationof COPD with ‘dusty trades’, as well as the specific contribu-tions to the risk of disease from certain types of exposure, thelimitations of this analysis should be considered. The relativelylow response rate may have introduced selection bias: althoughthere was no sex difference between respondents and non-respondents, non-respondents were significantly older (albeit,only a 1 year difference). Additionally, correcting analyses forlater response did not significantly alter the main study findings,and the term included in the analysis to represent late responsedid not have a significant influence as judged by its OR. Thedifferences in estimated associations with COPD comparing self-reported exposure (VGDF) and assigned exposure (JEM) may, inpart, reflect reporting bias in the former measure, although thisis counter-balanced against random misclassification biasingtowards the null in the latter.

Systematic classification error in disease assignment based onsubject report of a physician’s diagnosis should also be consid-ered. If such misreporting was associated with occupationalexposure, this could lead to a false association between exposureand disease. Because the link between COPD and occupation (asopposed to smoking) is not generally appreciated by the laypublic, this kind of systematic (as opposed to random)misclassification would not be anticipated. The weaker associ-ations between exposure and disease in the spirometry-definedanalysis, however, also warrant further discussion in thiscontext. This sub-analysis was subject to potential furtherselection biases (including higher overall exposure rates), andlimitations in study power. More importantly, the burden ofexposure (even by JEM assignment) is such in this subset thatthere may be unmeasured risk in the presumed ‘unexposed’referent category (including ‘relatively ’ clean occupations ingenerally contaminated workplaces, as well as neighbourhood-level factory-driven ambient pollution). Thus, there could bemultiple factors accounting for the attenuated risk estimates weobserved in this lung-function-based study subset, although theelimination by spirometry-based disease classification of a falseassociation based on systematic misreporting of physiciandiagnosis cannot be excluded as one possible factor. Of note, analternative analysis of the spirometry-based subset adjusting forusing ever versus never smoking rather than pack years asa continuous variable narrowed the CI of the OR, suggestingthat how smoking is quantified and how work-related exposureis categorised can affect the estimates of occupational risk forCOPD.

The use of population attributable risk as an estimate of thereduction in average disease risk over a specified time intervalthat would be achieved by eliminating the exposures is a wellapplied metric in these circumstances. Definitions of PAR% can

differ among studies, however, and these statistics can bemisinterpreted. We used a derivation for this value based onaccurate knowledge of the proportion of cases exposed, and theadjusted relative risks associated with various risk factors. It istherefore unlikely that the PAR% estimates are unhelpful ormisleading.In summary, this study has identified a significant contribu-

tion from workplace exposures to COPD prevalence, witha particularly heavy burden as a legacy from a highly industri-alised area dominated by the steel industry. These findings mustbe placed into the context of cigarette smoking still being themost important overall risk factor in COPD causation, whilealso lending further evidence to international data showing thatworkplace conditions must be considered in the larger aetio-logical picture of this disease.

Contributors AD performed data collection, entry and analysis, coded the jobexposure matrix and wrote the manuscript. DF and PB designed the study and wrotethe manuscript. DF is also the study guarantor. JW designed the study, stratifiedparticipants, performed the mail out and performed spirometry, data collection anddata entry. JWight contributed to the design of the study, provided access to studyparticipants and approved the final article. CB performed job exposure matrix codingand approved the final article. CY checked the statistical analysis. VS designed andmanaged the secure database and assisted with mail outs. CGB performed data entryand contributed to initial study design. CathGB recruited supplemental patients,performed some spirometry and contributed to questionnaire design. All authors hadaccess to the data if required and read the final article.

Funding The study was funded by internal research monies, and no external fundingwas obtained.


Ethics approval The study was approved by the Sheffield Research EthicsCommittee, the Sheffield Health and Social Research Consortium and by the SheffieldTeaching Hospitals NHS Foundation Trust Research Department. All participantsreceived written information concerning the study and gave informed consent.


REFERENCES1. Rabe K, Hurd S, Anzueto A, et al. Global strategy for the diagnosis, management,

and prevention of chronic obstructive pulmonary disease: GOLD executive summary.Am J Respir Crit Care Med 2007;176:532e55.

2. World Health Organization. COPD. http://www.who.int/respiratory/copd/en(accessed 25 Apr 2012).

3. Murray CJ, Lopez AD. Alternative projections of mortality and disability by cause1990e2020: Global Burden of Disease Study. Lancet 1997;349:1498e504.

4. European Lung Foundation. http://www.european-lung-foundation.org/63-european-lung-foundation-elf-burden-in-europe.htm (accessed 25 Apr 2012).

5. Darkow T, Chastek BJ, Shah H, et al. Health care costs among individuals withchronic obstructive pulmonary disease within several large, multi-state employers.J Occup Environ Med 2008;50:1130e8.

6. Halpin D, Miravitlles M. Chronic obstructive pulmonary disease: the disease and itsburden to society. Proc Am Thorac Soc 2006;3:619e23.

7. Ferrer M, Alonso J, Morera J, et al. Chronic obstructive pulmonary disease stageand health-related quality of life. Ann Intern Med 1997;127:1072e9.

8. Chapman K, Mannino D, Soriano J, et al. Epidemiology and costs of chronicobstructive pulmonary disease. Eur Respir J 2006;27:188e207.

9. Greenhow E. Chronic Bronchitis Especially as Connected with Gout, Emphysema,and Diseases of the Heart. London: Longmans, 1868.

10. Greenhow EH. Report of the Medical Officer of the Privy Council. Appendix VI.London: HM Stationary Office, 1861.

11. Thackrah CT. The Effects of Arts, Trades and Professions and of Civic States andHabits of Living on Health and Longevity. 2nd edn. London: Longman, 1832.

12. Fletcher C. Disability and mortality from chronic bronchitis in relation to dustexposure. AMA Arch Ind Health 1958;18:368e73.

13. Becklake M, Dupreez L, Lutz W. Lung function in silicosis of the Witwatersland goldminer. Am Rev Tuberc Pulm Dis 1958;77:400e12.

14. Trupin L, Earnest G, San Pedro M, et al. The occupational burden of chronicobstructive pulmonary disease. Eur Respir J 2003;22:462e9.

15. Blanc PD, Iribarren C, Trupin L, et al. Occupational exposures and the risk of COPD:dusty trades revisited. Thorax 2009;64:6e12.

16. Blanc PD, Menezes AM, Plana E, et al. Occupational exposures and COPD: anecological analysis of international data. Eur Respir J 2009;33:298e304.

17. Hnizdo E, Sullivan PA, Bang KM, et al. Association between chronic obstructivepulmonary disease and employment by industry and occupation in the US population:a study of data from the Third National Health and Nutrition Examination Survey. AmJ Epidemiol 2002;156:738e46.



18. Balmes J, Becklake M, Blanc P, et al. American Thoracic Society Statement:Occupational contribution to the burden of airway disease. Am J Respir Crit CareMed 2003;167:787e97.

19. Blanc P, Toren K. Occupation in chronic obstructive lung disease and chronicbronchitis: an update. Int J Tuberc Lung Dis 2007;11:122e33.

20. Melville A, Pless-Mulloli T, Afolabi O, et al. COPD prevalence and its associationwith occupational exposures in a general population. Eur Respir J 2010;36:488e93.

21. Fletcher CM, Elmes PC, Fairbairn AS, et al. The significance of respiratorysymptoms and the diagnosis of chronic bronchitis in a working population. BMJ1959;2:257e66.

22. Blanc PD, Eisner MD, Earnest G, et al. Further exploration of the links betweenoccupational exposure and chronic obstructive pulmonary disease. J Occup EnvironMed 2009;51:804e10.

23. Johnson A, Chan-Yeung M, MacLean L, et al. Respiratory abnormalities amongworkers in an iron and steel foundry. Br J Ind Med 1985;42:94e100.

24. Andjelkovich DA, Mathew RM, Richardson RB, et al. Mortality of iron foundryworkers: I. Overall findings. J Occup Med 1990;32:529e40.

25. Hansen ES. A cohort mortality study of foundry workers. Am J Ind Med1997;32:223e33.

26. Scotti PG, Arossa W, Bugiani M, et al. Chronic bronchitis in the iron and steelindustry: prevalence study. Med Lav 1989;80:123e31.

27. Wang ML, McCabe L, Hankinson JL, et al. Longitudinal and cross-sectionalanalyses of lung function in steelworkers. Am J Respir Crit Care Med1996;153:1907e13.

28. Wang ML, McCabe L, Petsonk EL, et al. Weight gain and longitudinal changes inlung function in steel workers. Chest 1997;111:1526e32.

29. Kuo HW, Chang CL, Liang WM, et al. Respiratory abnormalities among male foundryworkers in central Taiwan. Occup Med (Lond) 1999;49:499e505.

30. McLaughlin AI. Industrial Lung Disease of Iron and Steel Workers. London: HisMajesties Stationery Office, 1950.



907

Journal club

Autoimmune disorders increase the risk of developingpulmonary embolismThis large retrospective study analysed the effect of 33 autoimmune disorders on the risk ofdeveloping pulmonary emboli (PE). Five hundred and fifteen thousand one hundred andthirty-seven patients in Sweden with an autoimmune disorder initially diagnosed on hospitaladmission were identified and retrospectively analysed for PE, between 1964 and 2008.Risk of PE increased across all age groups in the first year post admission. Risk gradually

decreased after hospitalisation but remained above the control group at 10 years postadmission. Length of stay did not affect the risk. The thrombotic risk may have been relatedto active inflammation, side effects of the autoimmune treatment and/or immobilisation. Thestudy postulated that the fall in risk over time was attributable to the inflammatory activityof autoimmune conditions decreasing with effective treatment. Overall risk of PE was lowerduring 1989e2008 than during 1964e1988. Interestingly, no difference was seen after theintroduction of low molecular weight heparin in the 1990s.This was a large study, comparing autoimmune disorders to the general population.

However, acute hospital data was used, potentially skewing the study population towardsseverer cases. There was also no data on thrombo-prophylaxis or treatment.The study suggests that autoimmune disorders are associated with a particularly high risk

of PE and should be considered hypercoagulable disorders; this risk decreases but persists overtime. Thrombo-prophylaxis should be considered in these patients, potentially for some timepost discharge. However, further studies to assess its effectiveness are warranted.

< Zoller B, Li X, Sundquist J, et al. Risk of pulmonary embolism in patients with autoimmune disorders: a nationwide follow-upstudy from Sweden. Lancet 2012;379:244e9.

Helen Umpleby

Correspondence to Dr Helen Umpleby, ST3 at East Surrey Hospital, Canada Avenue, Redhill RH1 5RH, UK;[email protected]





ORIGINAL ARTICLE

Cigarette smoke and platelet-activating factorreceptor dependent adhesion of Streptococcuspneumoniae to lower airway cells

Jonathan Grigg,1 Haydn Walters,2 Sukhwinder Singh Sohal,2 Richard Wood-Baker,2

David W Reid,2 Cang-Bao Xu,3 Lars Edvinsson,3 Mathieu C Morissette,4

Martin R Stampfli,4 Michael Kirwan,1 Lee Koh,1 Reetika Suri,1 Naseem Mushtaq1

ABSTRACTBackground Exposure to cigarette smoke (CS) isassociated with increased risk of pneumococcalinfection. The mechanism for this association isunknown. We recently reported that the particulatematter from urban air simulates platelet-activating factorreceptor (PAFR)-dependent adhesion of pneumococci toairway cells. We therefore sought to determine whetherCS stimulates pneumococcal adhesion to airway cells.Methods Human alveolar (A549), bronchial (BEAS2-B),and primary bronchial epithelial cells (HBEpC) wereexposed to CS extract (CSE), and adhesion ofStreptococcus pneumoniae determined. The role of PAFRin mediating adhesion was determined using a blocker(CV-3988). PAFR transcript level was assessed byquantitative real-time PCR, and PAFR expression by flowcytometry. Lung PAFR transcript level was assessed inmice exposed to CS, and bronchial epithelial PAFRexpression assessed in active-smokers byimmunostaining.Results In A549 cells, CSE 1% increased pneumococcaladhesion (p<0.05 vs control), PAFR transcript level(p<0.01), and PAFR expression (p<0.01).Pneumococcal adhesion to A549 cells was attenuated byCV-3988 (p<0.001). CSE 1% stimulated pneumococcaladhesion to BEAS2-B cells and HBEpC (p<0.01 vscontrol). CSE 1% increased PAFR expression in BEAS2-B(p<0.01), and in HBEpC (p<0.05). Lung PAFR transcriptlevel was increased in mice exposed to CS in vivo(p<0.05 vs room air). Active smokers (n¼16) had anincreased percentage of bronchial epithelium withPAFR-positive cells (p<0.05 vs never smokers, n¼11).Conclusion CSE stimulates PAFR-dependentpneumococcal adhesion to lower airway epithelial cells.We found evidence that CS increases bronchial PAFR invivo.

BACKGROUNDEpidemiological studies suggest that exposure tocigarette smoke (CS) is a major risk factor forinfection with Streptococcus pneumoniae. Nuortiet al1 reported that in active smokers, the adjustedrisk for invasive pneumococcal disease is 4.1 (95%CI 2.4 to 7.3), and for adults exposed to environ-mental tobacco smoke (ETS) the adjusted risk is 2.5(95% CI 1.2 to 5.1). In young children, Suzuki et al2

reported that exposure to ETS is associated withhospital admissions with pneumonia (adjusted OR

1.55, 95% CI 1.25 to 1.92), and estimated that 28%of pneumonia admissions in Vietnamese childrenare attributable to ETS. Since S pneumoniae is themajor cause of pneumonia,3 these data suggest thatrisk of pneumococcal pneumonia is increased byactive smoking and exposure to ETS. Biologicalplausibility for this association is provided bya study that found chronic exposure of mice to CSfollowed by pneumococcal infection increasesmorbidity and the amount of colony forming unit(CFU) counts of pneumococci in the lung.4

To date, the mechanism whereby CS increasesvulnerability to pneumococcal pneumonia isunclear. A prerequisite step for the development ofpneumococcal infection is the adhesion of patho-genic bacteria to lower airway epithelial cells.3 Thecapacity of pneumococci to adhere to airway cells isenhanced by acid,5 respiratory viral infection,6 andinterleukin (IL)-1a.7 Adhesion of virulent strains ofS pneumoniae to bronchial and alveolar airwayepithelial cells is mediated, in part, by phosphor-ylcholine in the bacterial cell wall binding to thereceptor for platelet activating factor receptor(PAFR) on host cells.5e7 During the normalprocesses of PAFR internalisation, pneumococcibound to the receptor are transported into cellsdthat is, pneumococci use PAFR as aTrojan horse.7Wepreviously reported that exposure of airway cells tourban air particulate matter (PM) in vitro increasesthe adhesion of S pneumoniae via the PAFR.8 Since

Key messages

What is the key question?< Does exposure to cigarette smoke increase the

adhesion of pneumococci to lower airway cells?

What is the bottom line?< Cigarette smoke extract stimulates platelet-

activating factor receptor (PAFR)-dependentpneumococcal adhesion to lower airway cells.Increased PAFR was found in mice exposed tocigarette smoke, and in active smokers.

Why read on?< We have defined a mechanism underlying the

epidemiological association between exposure totobacco smoke and pneumococcal pneumonia.

< Additional materials arepublished online only. To viewthese files please visit thejournal online (http://dx.doi.org/10.1136/thoraxjnl-2011-200835).1Centre for Paediatrics, BlizardInstitute, Barts and the LondonSchool of Medicine andDentistry, Queen MaryUniversity of London, London,UK2NHMRC National Centre forResearch Excellence in ChronicRespiratory Disease, MenziesResearch Institute, Hobart,Australia3Division of ExperimentalVascular Research, Institute ofClinical Science in Lund, LundUniversity, Lund, Sweden4Departments of Pathology andMolecular Medicine, McMasterImmunology Research Centre,McMaster University, Hamilton,Ontario, Canada

Correspondence toProfessor Jonathan Grigg,Blizard Institute, Barts and theLondon School of Medicine andDentistry, Queen Mary,University of London, 4 NewarkStreet, London E1 2AT, UK;[email protected]

Received 21 July 2011Accepted 5 April 2012

Respiratory infection



both CS and urban PM share pollutants, for example, carbona-ceous PM, we hypothesised that CS upregulates pneumococcaladhesion to lower airway cells via PAFR. In this study, we soughtto assess the effect of cigarette smoke extract (CSE) on pneu-mococcal adhesion in a human lower airway cells. Using storedtissue from mouse and human studies, we then sought evidencethat CS exposure increases PAFR expression in vivo.

METHODSA549, BEAS2-B, primary bronchial epithelial cellsand S pneumoniaeA549, a human type II pneumocyte cell line, was obtained fromSigma Aldrich (Poole, UK). BEAS2-B cells were kindly donatedby Dr Nicolas Mercardo (National Heart and Lung Institute,Imperial College London, UK). Human primary bronchialepithelial cells (HBEpC) were purchased from Promocell(Heidelberg, Germany). Cells were maintained in Dulbecco’sModified Eagle’s Medium with 10% fetal bovine serum and 1%L-glutamine/penicillin-streptomycin (Lonza, Basel, Switzerland).Passage number was <10 for bronchial epithelial cells, and <15for A549 cells. We aimed to fully define the effect of CSE onpneumococcal adhesion and PAFR in A549 cells, then replicatekey findings in BEAS2-B and primary human bronchial epithelialcells. The virulent type 2 S pneumoniae encapsulated strain D39(NCTC 7466) was purchased from the National Collection ofType Cultures (Central Public Health Laboratory, London, UK)and grown to mid-log phase (optical density 600¼0.5) prior toadding to A549 and primary bronchial cells at 200 ml, 23108

mid-logarithmic phase, and 100 ml for BEAS2-B cells.

Generation of CSECSE in filter material was prepared at the Institute of ClinicalScience (Lund, Sweden), as previously described.9 Briefly, threecigarettes (0.8 mg nicotine per cigarette; Marlboro; Philip MorrisUSA, Pittsburgh, Pennsylvania, USA) were ‘smoked’ by a wateraspirator, and CS aspirated through a cotton wool filter. CSE(100%) was obtained by vortexing the cotton wool filter in 2 mldimethyl sulfoxide (DMSO).9 DMSO was stored at �208C as100% stock solution. Stock solution was then diluted in mediumand used for all experiments at #1%. We previously determinedthat the nicotine content of CSE was 0.1e0.15 mg/ml.9 CSEwas diluted in medium and used in experiments at #1%. DMSO#1% did not alter pneumococcal adhesion.

Cell viability in A549 cells exposed to CSE in vitroThe effect of CSE on cell viability was assessed in A549 cells.Lactate dehydrogenase (LDH) release, a marker of cell membraneintegrity, was done according to the manufacturer ’s protocol(Sigma Aldrich, Gillingham, UK). The assay included LDHreleased after total cell lysis (‘total’ LDH control). Opticaldensity of LDH was assessed by spectroscopy. Cell cytotoxicitywas assessed by the MTT (3-(4,-dimethylthiazol-2-yl)-2, 5-diphenyl-tetrazolium) assay.10 This assay assesses the conver-sion of the MTT reagent to formazan which then accumulatesin healthy cells. Briefly, 20 ml MTT (5 mg/ml, Sigma Aldrich) inphosphate-buffered saline was added to A549 cells, and incu-bated for 1 h. Medium was removed and 100 ml of DMSO(Sigma Aldrich) added for 30 min. Absorbance (optical densityunits) was assessed at 550 nm. In the MTT assay, cytotoxicityreduces the optical density.

PAFR transcript level in airway cells exposed to CSE in vitroTranscript level of PAFR was assessed in A549 cells by quanti-tative real-time PCR. For A549 cells, 23105 cells were harvested

for each sample and RNA was prepared using the RNeasy minikit (Qiagen, Crawley, UK). Random hexamers (Qiagen) andM-MLV reverse transcriptase (Invitrogen, Paisley, UK) were usedto synthesise first strand cDNA according to manufacturer ’sinstructions. mRNA transcript level of PAFR and the house-keeping gene glyceraldehyde-3-phosphate dehydrogenase(GAPDH) were determined by quantitative real-time PCR usingan ABI 7500 real-time PCR system (Applied Biosystems,Warrington, UK) with TaqMan primer and probe setsHS00265399_S1 (PAFR), and Hs99999905_m1 (GAPDH)respectively. mRNA transcript level was normalised to GAPDH,and the relative change in expression between samples calcu-lated using the comparative CT method according to themanufacturer ’s instructions (Applied Biosystems). Data arerepresentative of two separate experiments with six replicates.

PAFR expression in airway cells exposed to CSE in vitroPAFR expression in A549, BEAS2-B and HBEpC was determinedby flow cytometry (details in online supplement). Specific PAFRmedian florescence intensity was determined for each concen-tration of CSE by subtracting the median florescence intensity ofthe isotypic control monoclonal antibody (mAb) from themedian florescence intensity of the PAFR mAb. Data areexpressed as fold-change in PAFR over control, and are from fourexperiments done at different times for A549 and three experi-ments done at different times for BEAS2-B cells. Due to limitedcell number, PAFR expression in HBEpC is from one experiment.

Pneumococcal adhesion to airway cells exposed to CSE in vitroAdhesion of pneumococci to airway cells in vitro was done usinga standard bacterial adhesion assay.7 8 11 The adhesion assayreflects adherent and intracellular bacteria. Briefly, A549 cells andHBEpC at 23105/ml were seeded in 24-well plates (Costar,Sigma Aldrich). CSE up to 1% was added to cell monolayers andincubated for 4 h at 378C. CSE was removed by washing twicewith Dulbecco’s Modified Eagle Medium (Lonza) and pneumo-cocci added and incubated for 2 h.5 Cell monolayers werewashed five times and cells removed from the tissue cultureplate by trypsin-EDTA, and lysed with ice-cold sterile distilledwater for 10 min. Lysates were plated out for colony formingunits (CFU) /mL.12 For BEAS2-B cells, the adhesion assay wasmodified. Briefly, cells were maintained in DMEM and thenseeded at 23105 cells/ml in Promocell Bronchial Epithelial CellMedia (Promocell) for 48 h. The last washing step reduced tothree times before plating. The functional relevance of PAFR wasassessed by co-incubating cells with the competitive PAFRantagonist CV-398813 (Sigma Aldrich). A stock solution of CV-3988 was prepared at 1 mM in ethanol then diluted in mediumto a final concentration of 10 mMda concentration that inhibitsPAF-induced cortisol release from adrenal glands.14 We estab-lished that CSE per se does not stimulate pneumococcal prolif-eration, and that ethanol at the concentration in 10 mM CV-3988 does not affect pneumococcal adhesion. Adhesion data arerepresentative of at least three separate experiments done ondifferent days for A549 and BEAS2-B cells, and two experimentsdone on different days for HBEpC.

Lung PAFR transcript level in mice exposed to CS in vivoSix to eight-week-old female C57BL/6 mice (purchased fromCharles River Laboratories, Montreal, Canada) were exposed tomainstream CS using a whole body CS exposure system (SIU48,Promech Lab AB, Vintrie, Sweden), as previously reported indetail.15 Briefly, mice (n¼5) were exposed to the mainstream CSof 123 2R4F reference cigarettes with filter removed (Tobacco



and Health Research Institute, University of Kentucky,Lexington, Kentucky, USA) over 50 min, twice daily. Controlanimals (n¼5) were exposed to room air only. Following 4 daysof CS exposure, lungs were removed and snap frozen in liquidnitrogen. Samples were sent to the UK on dry ice for PAFRmRNA analysis. Unfrozen tissue samples were homogenised andthe RNA extracted as described above. First strand cDNAsynthesis was carried out using the high-capacity RNA-to-cDNA master mix (Applied Biosystems), according to themanufacturer ’s instructions. mRNA transcript level of PAFR andhouse-keeping gene, b2 microglobulin, were determined byquantitative real-time PCR as described above, with TaqManprimer and probe sets Mm02621061_m1 and Mm00437762_m1respectively. Mouse PAFR mRNA transcript level was normal-ised to the housekeeping gene b2 microglobulin. McMasterUniversity ’s Animal Research Ethics Board approved all animalprocedures.

Bronchial PAFR expression in adults exposed to CS in vivoBronchial biopsies were obtained for active smokers and healthynever smokers as previously described.16 Subjects gave written,informed consent approved by the Human Research EthicsCommittee (Tasmania) Network (approval number: H0007017).Subjects with a history suggestive of asthma, other respiratorydisorders and uncontrolled comorbidities were excluded. Neversmokers had no history of respiratory illness or smoking. Forsmokers, the inclusion criteria were a minimum 10-pack-yearhistory of cigarette smoking with spirometry within normallimits post bronchodilator (forced expiratory volume in a 2(FEV1)>80% of predicted, and FEV1/forced vital capacity>70%).Endobronchial biopsies were taken from subsegmental carinae ofthe right lower lobe of each patient, using alligator forceps (FB-15C; Olympus, Tokyo, Japan). Bronchial biopsies were fixed in4% neutral buffered formalin for 2 h and subsequently processedinto paraffin through graded alcohol and xylene using a LeicaASP 200 tissue processor. Sections from stored specimens werecut at 3 mm from individual paraffin blocks, stained with H&Eand morphologically assessed for immunostaining (details inonline supplement). Expression of PAFR was measured onrandomised and coded slides by an operator blinded to smokingstatus and expressed as the percentage of epithelium withPAFR-positive cells for each subject.

Statistical analysisNormally distributed data are summarised as mean (SEM) andanalysed using GraphPad Prism version 5.03 (GraphPad Soft-ware, La Jolla, California, USA), with analyses done by unpairedt test. For non-normally distributed data, comparisons weredone by ManneWhitney U test. For mRNA transcript number,the mean of raw data of control samples was assigned 100% andthe SEM converted to a percentage of the raw value. In humantissue, PAFR mRNA data were normalised to mRNA from thehousekeeping gene GAPDH, and in mouse tissue normalised tob2 microglobulin. A p value <0.05 was considered significant.

RESULTSExposure of airway cells to CSE 1% for 4 h in vitro resulted incell contraction, cytoplasmic vacuolation, and a tendency forcells to detach from the culture plate on washing (online figures1 and 2). Brown PM was visible on cells cultured with CSE 1%,but at a relatively low density compared with the concentrationof urban PM (50 mg/ml) reported to stimulate pneumococcaladhesion8 (online figure 1). Culture of A549 cells with CSE 1%

for 4 h was not cytotoxic as assessed by the MTT assay(figure 1). A moderate increase in LDH release by A549 cells wasstimulated by CSE 1% (p<0.01 vs medium control, figure 2).CSE at 0.5% and 1% increased pneumococcal adhesion to

A549 cells (p<0.05 vs control, figure 3A). Co-incubation witha specific PAFR antagonist (CV-3988) attenuated CSE-stimu-lated adhesion to A549 cells (p<0.001; CSE 0.5%+antagonist vsCSE 0.5%, figure 3B), but had no effect on basal adhesion (datanot shown). CSE 1% increased PAFR transcript level (p<0.01 vscontrol, figure 3C), and PAFR expression in A549 cells (p<0.01 vscontrol, figure 3D). CSE 1% increased pneumococcal adhesion toBEAS2-B cells and HBEpC (p<0.01 vs control, figure 4A,B), andincreased PAFR expression in BEAS2-B cells (1.960.07-foldincrease, p<0.05 vs control), and HBEpC (3.460.8-fold increasevs control).In mice, 4-day exposure to CS increased lung PAFR transcript

level (n¼5 per group, p<0.05 vs room air, figure 5). In humans,6/16 active smokers had PAFR-positive bronchial cells (0.05, 0.1,0.58, 2.12, 2.25, 4.82% of bronchial epithelium with positive

Figure 1 Conversion of MTT (3-(4,-dimethylthiazol-2-yl)-2, 5-diphenyl-tetrazolium) by A459 cells stimulated with cigarette smoke extract (CSE)for 4 h. Optical density (OD) of the formazin product is detected byspectroscopy at 550 nm. CSE does not attenuate in the ability of cells toconvert MTT, indicating an absence of cytotoxicity. CSE increasedconversion at 0.5 and 1% (*p<0.05 vs medium control).

Figure 2 Release of lactate dehydrogenase (LDH) by A549 cellsstimulated with cigarette smoke extract (CSE) for 4 h. Compared withLDH release after lysis of all cultured cells (total LDH), there is a small,but significant, increase in LDH by cells stimulated by CSE 1% (*p<0.01vs medium control).



cells respectively, figure 6). In contrast, all 11 never smokers hadoccasional very low levels of PAFR staining that were alwaysbelow the positive cutoff threshold (active smokers vs neversmokers, p<0.05, ManneWhitney U test). There was nosignificant difference in sex and age between smokers and non-smokers (Table 1) and no difference in age, pack-years or lungfunction between the PAFR-positive and PAFR-negative activesmokers. Non-specific staining levels were low in both activeand never smokers (online figure 3).

DISCUSSIONIn this study, we found that in vitro exposure to CSE upregu-lates the adhesion of pneumococci to lower airway epithelialcells. A major role for PAFR in CSE-stimulated adhesion issuggested by increased PAFR mRNA transcript level and PAFRexpression in CSE-exposed A549 cells. Furthermore, we foundthat adhesion was attenuated by a specific PAFR antagonist

(CV-3988). Compatible with previous studies in cytokine-stim-ulated A549 cells,7 blocking PAFR attenuated stimulated adhe-sion, but not unstimulated adhesion of pneumococci to A549cells. The effect of CSE on adhesion is not limited to the mostdistal epithelial cells, since CSE stimulated adhesion to a bron-chial epithelial cell line (BEAS2-B) and to human primarybronchial cells (HBEpC). The clinical relevance of upregulationof PAFR in vivo is suggested by several animal models. Forexample, interleukin 1 stimulates PAFR in the rabbit airway,which in turn increases vulnerability to pneumococcal pneu-monia.7 Similarly, influenza infection increases PAFR in themouse lung which increases lethality to pneumococcal infec-tion,17 and increased lung PAFR in mice transplanted withsickle-cell bone marrow causes hypersusceptibility to pneumo-coccal infection.18 Conversely, PAFRe/e mice with reduced PAFRexpression in airway cells are resistant to lethal pneumococcalpneumonia.19 In this study, we found evidence that CS upre-gulates PAFR in vivo, since PAFR transcript level was increasedin the lungs of mice exposed to CS, and bronchial epithelialPAFR expression was increased in active smokers.The mechanism whereby CSE upregulates airway epithelial

PAFR is unclear. Cell death was excluded since, compatible withother studies,20 CSE 1% did not attenuate cell function assessedby the MTT assay. CSE did however stimulate a moderateincrease in LDH release. Previous studies report that tobaccosmoke, via induction of oxidative stress, stimulates bronchialepithelial cell proliferation,21 activates stress-activated proteinkinase,22 and disrupts airway cell cytoskeleton proteins.23 In thisstudy, the tendency of CSE-exposed cells to contract, detachfrom the tissue culture plate, and release LDH is compatiblewith a response to oxidative stress. Since CSE contains thou-sands of bioactive chemical compoundsdincluding carbona-ceous PM, nicotine, alkaloids, metals, nitrosamines,polyaromatic hydrocarbons, aromatic amines, organiccompounds,24 and endotoxin25dthe component stimulatingpneumococcal adhesion remains unclear. We previously reported

Figure 3 Effect of cigarette smokes extract (CSE) on A549 cells. (A) Incubation with CSE 0.5% and 1% increases Streptococcus pneumoniae colony-forming units (CFU), indicating increased adhesion (*p<0.05 vs medium control). Data are represented as mean and SEM (n¼8, and n¼6respectively). (B) Coincubation with a platelet-activating factor receptor (PAFR) blocker (10 mM CV-3988) attenuates CSE-stimulated adhesion(*p<0.001 vs without PAFR blocker). Dimethyl sulfoxide (DMSO) at 1% does not affect adhesion. Data are represented as mean and SEM (n¼8), andare representative of four separate experiments. (C) CSE 1% increases PAFR transcript level (*p<0.01 vs medium control). PAFR was assessed byquantitative real-time PCR and normalised for the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Data are representative oftwo experiments with six replicates. (D) CSE 1% increases PAFR expression. Cells were immunostained with either mouse antihuman PAFR, orisotypic control and expression determined by flow cytometry. Data are described as fold change in median florescence intensity (MFI) over mediumcontrol, adjusted for the isotypic control. Data are expressed as mean (SEM) fold change, and are from four separate experiments (*p<0.01 vscontrol).

Figure 4 Effect of cigarette smoke extract (CSE) on the adhesion ofStreptococcus Pneumoniae to; (A) BEAS2-B cells, and (B) humanprimary bronchial epithelial cells (HBEpC). CSE 1% increases adhesion toboth cell types (*p<0.05, and **p<0.01 vs control, n$5). Data areexpressed as mean (SEM) and are representative of four separateexperiments for BEAS2-B, and two separate experiments for primarybronchial cells. CFU, colony forming unit.



that fossil-fuel PM per se, albeit at high density, stimulatespneumococcal adhesion to lower airway cells (online figure 1A toF).8 However, the density of PM on cells exposed to CSE 1% wasrelatively low (online figure 1A to F). We therefore speculate thatthe PM and soluble fractions of CSE stimulate adhesion.Evidence for the bioactivity of soluble components of CS isprovided by our previous study in vascular and bronchial smoothmuscle cells in which CSE, but not a suspension of PM from CS,upregulated immune receptor expression.9 Furthermore, Sohnet al26 assessed the effect of three soluble CS components(nicotine, benzopyrene and napthylamine) on immune response-related genes in A549 cells, and reported that PAFR was one ofonly 39/1152 genes upregulated by nicotine.26

Previous studies support the hypothesis that inhalation of CSupregulates pneumococcal adhesion to lower airway cells invivo. For example, Riise et al27 reported a non-significant trendfor increased adhesion of S pneumoniae to fixed airway cells fromsmokers with chronic bronchitis and El Ahmer et al28 reportedincreased binding of S pneumoniae to buccal epithelial cells fromactive smokers compared with non-smokers. To date, data onPAFR in the human lung are conflicting. On one hand, Ishizukaet al5 reported PAFR expression by cultured ciliated trachealepithelial cells obtained postmortem. On the other hand, Shir-asaki et al29 found no PAFR mRNA in airway epithelial cellsfrom heartelung transplant donor lungs, and lower levels ofPAFR mRNA in whole lung tissue from smokers (n¼4)compared with non-smokers (n¼8).29 In contrast, we foundincreased PAFR transcript level in mouse lung after controlledexposure to CS, and evidence of increased PAFR expression inbronchial biopsies from active smokers. The reason why wefound PAFR-positive cells in only a subgroup of smokers isunclear. We speculate that immunostaining of stored, fixedtissue is a relatively insensitive technique for assessing PAFRexpression. Alternatively PAFR expression may be determinedby variables that we did not record, for example, time from lastcigarette.

There are limitations to this study. First, we did not model theantioxidant and surfactant-rich layer of epithelial lining fluidthat provides protection against inhaled pollutants in vivo.30

Second, there are very likely to be several other factors, apartfrom adhesion, contributing to CS-induced vulnerability tobacterial infection in vivo. For example, CS attenuates humanb-defensin-2 production in airway cells,31 and in murine modelsreduces complement-mediated phagocytosis of S pneumoniae byairway macrophages,4 and reduces ciliary beat frequency.32

Third, CSE in solution may not fully reflect airway deposition atan airetissue interface.33 Finally, additional mechanisms may beinvolved in mediating pneumococcal adhesion in vivo. In upper

Figure 5 Exposure of mice to cigarette smoke for 50 min twice a dayfor 4 days increases pulmonary platelet-activating factor receptor(PAFR) transcript level (*p<0.05 vs room air control). PAFR wasassessed by quantitative real-time PCR and transcript level normalised tob2 microglobulin. Data are from a single in vivo exposure experimentwith five animals in each group.

Figure 6 Active smoking and bronchial platelet-activating factorreceptor (PAFR) expression. (A) Section of a bronchial biopsy from oneof the six active smokers with PAFR-positive cells. Ten active smokershad no cells that reached the positive threshold. PAFR staining is brownand is most pronounced in the apical part of the bronchial epithelium(arrowed). (B) A representative bronchial biopsy from a never smokershowing low PAFR staining that is below the cutoff threshold forpositivity. Original magnification 3400. Scale bar ¼ 50 mm. The isotypiccontrol is shown in online figure 3.

Table 1 Demographics and lung function data for adults undergoingbronchoscopy and bronchial biopsy

Active smokers Never smokers

n 16 11

Men/women 12/4 4/7

Age (years) 50 (30 to 66) 54 (41 to 68)

Smoking (pack-years) 32 (10 to 57) 0

FEV1 (% predicted)* 99 (78 to 125)y 114 (113 to 125)

FEV1/FVC ratio (% predicted)* 77 (70 to 96) 82 (71 to 88)

Data expressed as median (range).There was no difference in FEV1/FVC ratio between the two groups.*After 400 mg inhaled salbutamol.yp<0.01 versus never smokers.FEV1, forced expiratory volume in one second; FVC, forced vital capacity.



airway epithelial cells the polymeric immunoglobulin receptorcontributes to pneumococcal adhesion,34 although this receptoris not expressed by A549 cells (S. Hammerschmidt, personalcommunication).

In conclusion, exposure of lower airway epithelial cells to CSEin vitro upregulates PAFR-dependent pneumococcal adhesion.Increased lung PAFR mRNA in mice exposed to CS, andincreased PAFR expression in the bronchial epithelium of activesmokers, suggests that active and passive inhalation of CSupregulates airway PAFR in vivo.

Contributors JG: devised overall study, supervised and audited in vitro work, wrotemanuscript. EHW: devised bronchial biopsy study, contributed to final draft. SSS:collated demographic data, performed immunohistochemical analysis of bronchialbiopsies, contributed to final draft. RWB: performed bronchial biopsies and contributedto final draft. DR: performed bronchial biopsies and contributed to final draft. ID:performed mRNA extraction, and contributed to quantitative PCR, and to final draft.MK: performed quantitative PCR and analysis, contributed to final draft. CBX:generated cigarette smoke extract, and contributed to final draft. LE: devised andsupervised generation of CSE, and contributed to final draft. NM: performed adhesionassays and contributed to in vitro study design, and contributed to final draft. LK:performed the flow cytometry experiments and contributed to the revised draft. MS:devised and supervised the mouse exposure experiments. MM: did the mouseexposure experiments. RS: did the BEAS2-B receptor expression experiments.

Funding This work was supported by the Barts and the London Charity, and grantsfrom the Swedish Research Council (number 05958), and the Royal Hobart HospitalResearch Foundation, and NHMRC Australia.


Ethics approval Human Research Ethics Committee (Tasmania) Network (approvalnumber: H0007017).


Data sharing statement Raw data will be provided on request.

REFERENCES1. Nuorti JP, Butler JC, Farley MM, et al. Cigarette smoking and invasive

pneumococcal disease. Active Bacterial Core Surveillance Team. N Engl J Med2000;342:681e9.

2. Suzuki M, Thiem VD, Yanai H, et al. Association of environmental tobacco smokingexposure with an increased risk of hospital admissions for pneumonia in childrenunder 5 years of age in Vietnam. Thorax 2009;64:484e9.

3. van der Poll T, Opal SM. Pathogenesis, treatment, and prevention of pneumococcalpneumonia. Lancet 2009;374:1543e56.

4. Phipps JC, Aronoff DM, Curtis JL, et al. Cigarette smoke exposure impairspulmonary bacterial clearance and alveolar macrophage complement-mediatedphagocytosis of Streptococcus pneumoniae. Infect Immun 2010;78:1214e20.

5. Ishizuka S, Yamaya M, Suzuki T, et al. Acid exposure stimulates the adherence ofStreptococcus pneumoniae to cultured human airway epithelial cells: effects onplatelet-activating factor receptor expression. Am J Respir Cell Mol Biol2001;24:459e68.

6. Avadhanula V, Rodriguez CA, Devincenzo JP, et al. Respiratory viruses augment theadhesion of bacterial pathogens to respiratory epithelium in a viral species- and celltype-dependent manner. J Virol 2006;80:1629e36.

7. Cundell DR, Gerard NP, Gerard C, et al. Streptococcus pneumoniae anchor toactivated human cells by the receptor for platelet-activating factor. Nature1995;377:435e8.

8. Mushtaq N, Ezzati M, Hall L, et al. Adhesion of Streptococcus pneumoniae tohuman airway epithelial cells exposed to urban particulate matter. J Allergy ClinImmunol 2011;127:1236e42.e2.

9. Granstrom BW, Xu CB, Nilsson E, et al. Smoking particles enhance endothelin Aand endothelin B receptor-mediated contractions by enhancing translation in ratbronchi. BMC Pulm Med 2006;6:6.

10. Mosmann T. Rapid colorimetric assay for cellular growth and survival: application toproliferation and cytotoxicity assays. J Immunol Methods 1983;65:55e63.

11. Hammerschmidt S, Wolff S, Hocke A, et al. Illustration of pneumococcalpolysaccharide capsule during adherence and invasion of epithelial cells. InfectImmun 2005;73:4653e67.

12. Okada N, Tatsuno I, Hanski E, et al. Streptococcus pyogenes protein F promotesinvasion of HeLa cells. Microbiology 1998;144:3079e86.

13. Terashita Z, Imura Y, Nishikawa K. Inhibition by CV-3988 of the binding of[3H]-platelet activating factor (PAF) to the platelet. Biochem Pharmacol1985;34:1491e5.

14. Shimada T, Hirose T, Matsumoto I, et al. Platelet-activating factor acts on cortisolsecretion by perfused guinea-pig adrenals via calcium-/phospholipid-dependentmechanisms. J Endocrinol 2005;184:381e91.

15. Botelho FM, Gaschler GJ, Kianpour S, et al. Innate immune processes are sufficientfor driving cigarette smoke-induced inflammation in mice. Am J Respir Cell Mol Biol2010;42:394e403.

16. Soltani A, Reid DW, Sohal SS, et al. Basement membrane and vascular remodellingin smokers and chronic obstructive pulmonary disease: a cross-sectional study.Respir Res 2010;11:105.

17. van der Sluijs KF, van Elden LJ, Nijhuis M, et al. Involvement of the platelet-activating factor receptor in host defense against Streptococcus pneumoniaeduring postinfluenza pneumonia. Am J Physiol Lung Cell Mol Physiol 2006;290:L194e9.

18. Miller ML, Gao G, Pestina T, et al. Hypersusceptibility to invasive pneumococcalinfection in experimental sickle cell disease involves platelet-activating factorreceptor. J Infect Dis 2007;195:581e4.

19. Rijneveld AW, Weijer S, Florquin S, et al. Improved host defense againstpneumococcal pneumonia in platelet-activating factor receptor-deficient mice.J Infect Dis 2004;189:711e16.

20. Richter PA, Li AP, Polzin G, et al. Cytotoxicity of eight cigarette smoke condensatesin three test systems: comparisons between assays and condensates. Regul ToxicolPharmacol 2010;58:428e36.

21. Lemjabbar H, Li D, Gallup M, et al. Tobacco smoke-induced lung cell proliferationmediated by tumor necrosis factor alpha-converting enzyme and amphiregulin. J BiolChem 2003;278:26202e7.

22. Kaushik G, Kaushik T, Khanduja S, et al. Cigarette smoke condensate promotes cellproliferation through disturbance in cellular redox homeostasis of transformed lungepithelial type-II cells. Cancer Lett 2008;270:120e31.

23. Das A, Bhattacharya A, Chakrabarti G. Cigarette smoke extract induces disruption ofstructure and function of tubulin-microtubule in lung epithelium cells and in vitro.Chem Res Toxicol 2009;22:446e59.

24. Hammond D, O’Connor RJ. Constituents in tobacco and smoke emissions fromCanadian cigarettes. Tob Control 2008;17 Suppl 1:i24e31.

25. Hasday JD, Bascom R, Costa JJ, et al. Bacterial endotoxin is an active componentof cigarette smoke. Chest 1999;115:829e35.

26. Sohn SH, Kim KN, Kim IK, et al. Effects of tobacco compounds on gene expressionin fetal lung fibroblasts. Environ Toxicol 2008;23:423e34.

27. Riise GC, Larsson S, Andersson BA. Bacterial adhesion to oropharyngeal andbronchial epithelial cells in smokers with chronic bronchitis and in healthynonsmokers. Eur Respir J 1994;7:1759e64.

28. El Ahmer OR, Essery SD, Saadi AT, et al. The effect of cigarette smoke onadherence of respiratory pathogens to buccal epithelial cells. FEMS Immunol MedMicrobiol 1999;23:27e36.

29. Shirasaki H, Nishikawa M, Adcock IM, et al. Expression of platelet-activating factorreceptor mRNA in human and guinea pig lung. Am J Respir Cell Mol Biol1994;10:533e7.

30. Cantin AM, Fells GA, Hubbard RC, et al. Antioxidant macromolecules in the epitheliallining fluid of the normal human lower respiratory tract. J Clin Invest1990;86:962e71.

31. Herr C, Beisswenger C, Hess C, et al. Suppression of pulmonary innate host defencein smokers. Thorax 2009;64:144e9.

32. Simet SM, Sisson JH, Pavlik JA, et al. Long-term cigarette smoke exposure ina mouse model of ciliated epithelial cell function. Am J Respir Cell Mol Biol2010;43:635e40.

33. Gowadia N, Dunn-Rankin D. A transport model for nicotine in the tracheobronchialand pulmonary region of the lung. Inhal Toxicol 2010;22:42e8.

34. Asmat TM, Agarwal V, Rath S, et al. Streptococcus pneumoniae infection of hostepithelial cells via polymeric immunoglobulin receptor transiently induces calciumrelease from intracellular stores. J Biol Chem 2011;286:17861e9.

PAGE fraction trail=6



ORIGINAL ARTICLE

Smoking, acute mountain sickness and altitudeacclimatisation: a cohort study

Tian-Yi Wu,1,2,3 Shou-Quan Ding,4 Jin-Liang Liu,4 Jian-Hou Jia,4 Zuo-Chun Chai,4

Rui-Chen Dai,4 Ji-Zhui Zhao,5 Qi De Tang,5 Bengt Kayser6

ABSTRACTRationale The relationship between cigarette smokingand acute mountain sickness (AMS) is not clear.Objective To assess AMS risk and altitudeacclimatisation in relation to smoking.Methods 200 healthy non-smokers and 182 cigarettesmokers were recruited from Han lowland workers.These were men without prior altitude exposure,matched for age, health status and occupation, whowere transported to an altitude of 4525 masl.Measurements AMS, smoking habits, arterialsaturation (SpO2), haemoglobin (Hb), lung function andmean pulmonary artery pressure (PAPm) were assessedupon arrival and after 3 and 6 months.Main results Compared with non-smokers, smokershad a lower incidence of AMS and lower AMS scoresthan non-smokers upon arrival; higher Hb and PAPmassociated with lower SpO2 at 3 and 6 months ataltitude; and lower forced expiratory volume in 1 s andmaximal voluntary ventilation at 3 and 6 months.Conclusions Smoking slightly decreases the risk ofAMS but impairs long-term altitude acclimatisation andlung function during a prolonged stay at high altitude.

INTRODUCTIONIn China in 2010, 53% of men and 28% overallsmoked tobacco.1 Apart from its general healthrisks, smoking may influence altitude hypoxiatolerance. According to some it aggravates hypo-xaemia and hence increases the risk for acutemountain sickness (AMS) (Hultgren, p.469),2 butmountaineers find that smoking decreases AMSrisk.3 However, sound epidemiological data on theeffects of smoking on risk and disease course ofAMS are lacking. During the construction of theQinghaieTibet railroad from 2001 to 2005, >78 000lowland workers ascended to work and live ataltitude. Since 34% of the employed Han maleworkforce smoked cigarettes, this presenteda unique occasion to directly investigate the effectsof smoking on AMS risk. We therefore recruitedconstruction workers ascending from low altitudeto work and live at the highest construction sites atan average altitude of 4552 masl. We measured AMSincidence and progression, and acclimatisation insmokers and non-smokers.

METHODSThree hospitals participated (4779 masl, barometricpressure (Pb) w417 mm Hg; 4505 masl, Pbw440 mm Hg; 4292 masl, Pb w447 mm Hg). The

highest work site was at 4905 masl. The study wasapproved by the China National Science Founda-tion and the Qinghai High Altitude MedicalResearch Institute Committee on Human Research.In 2003, a first group of 4683 workers was recruited.All prospective workers filled out a questionnaireproviding information on age, sex, ethnicity, occu-pation, place of birth, altitude exposure, personaland family medical history, smoking and drinkingbehaviour. Subjects were interviewed and under-went a physical exam. Subjects in good health andphysical condition were offered a job. The subjectswere then asked to participate in a study on thehealth effects of altitude exposure. Subjects werekept unaware of the study objective, were notgiven information on smoking, received no incen-tives, were informed about procedures, knew theycould withdraw at any time and gave signedconsent. We sequentially recruited 200 lowlandsmoking and 200 non-smoking apparently healthynon-acclimatised male first-time ascenders, basedon capacity. Three smokers and four non-smokersrefused to participate. Groups had similar age, bodymass index, working altitude and work (semi-mechanised, laying out tracks). Subjects travelledfor 2 days by train to 2261 masl, stayed there for2 days, and then travelled on for 12 h by train to2808 masl where they stayed for 3 days. The finalaltitude was reached after a further 6e8 h bus ride.A smoker was someone who smoked 10 or more

cigarettes/day for >6 months. Non-smokers hadnever smoked; occasional smokers were excluded.Smoking was classified as mild (<1 pack/day, ie,10e20 cigarettes/day), moderate (1 pack/day) orheavy (>1 pack/day). Smoking duration was shortterm (6 months to 2 years), medium term

Key messages

What is the key question?< Are smokers really better off at altitude?

What is the bottom line?< Smokers may be better off initially at altitude,

but not in the long term.

Why read on?< This study allows us to advise smokers on

altitude exposure using solid epidemiologicaldata and suggests new avenues for research onacute mountain sickness pathophysiology.

1Physiological Research Groupof the Ministry of Railroad,Beijing, People’s Republic ofChina2National Key Laboratory of HighAltitude Medicine, High AltitudeMedical Research Institute,Qinghai, People’s Republic ofChina3High Altitude Medical ResearchCenter, University of Tibet,Lhasa, Tibet, People’s Republicof China4Qingai-Tibet Railroad Hospitalsof the Construction Company,Golmud, People’s Republic ofChina5China Railroad ConstructionCompany, Beijing, People’sRepublic of China6Institut des sciences dumouvement et de la medecinedu sport, Faculte de medicine,Universite de Geneve, Geneva,Switzerland

Correspondence toDr Tian Yi Wu, National KeyLaboratory of High AltitudeMedicine, High Altitude MedicalResearch Institute, Xining,Zhuanchang Rd, No. 7, Qinghai810012, People’s Republic ofChina;[email protected]

Received 15 June 2011Accepted 18 April 2012

Smoking



(2e5 years), and long term (>5 years). There were only cigarettesmokers.

Arterial oxygen saturation (SpO2, finger oximetry, Ohmeda,Louisville, CO, USA) was measured in a seated position after 30min of rest. Mean pulmonary artery pressure (PAPm) was esti-mated by Doppler. With a 3.5 MHz transducer (HP-Sonos 1000or 1500, Palo-Alto, CA, USA) data were obtained from theparasternal short-axis or apical position, the subject lying inslight left oblique rotation. Recordings were stored on videotapefor post hoc analysis by two independent cardiologists, unawareof smoking or altitude status. PAPm was estimated using theKitabatake formula. In our institute correlation with directlymeasured pressure during right-heart catheterisation is high(R2¼0.90). PAPm $25 mm Hg was considered pulmonaryhypertension. Vital capacity (VC), forced expiratory volume in 1s (FEV1), forced expiratory flow between 25% and 75% of vitalcapacity (FEF25e75%) and 20 s maximal voluntary ventilation(MVV) were measured with a portable spirometer (COSMED,Italy). Haemoglobin (Hb) was measured on venous blood(Au-400, Olympus, Shinjuku, Tokyo, Japan). Measurementswere done at low altitude, upon arrival (first hour, except PAPm,next day, and lung function, upon arrival and after 3 days), andagain after 1 week, 3 months and 6 months.

AMS was assessed with Lake Louise Scoring (LLS),4 whichconsists of self-reported assessment of symptoms (headache,dizziness/light-headedness, fatigue, gastrointestinal upset(anorexiaenauseaevomiting) and difficulty sleeping), eachscored from 0 to 3 (nil, mild, moderate, severe). It was completedwith three clinical signs (change in mental status (0e4), ataxiadetermined with heel-to-toe test (0e4), and peripheral oedema(0e2)). We used two cut-offs for AMS defining it as headacheand a cumulative score $3 or $4. Severity of AMS was definedas mild for a score of 3e5 and severe for a score of 5 or more.AMS was assessed on the evening of arrival at 4525 masl and thefollowing evenings, for 1 week.

Data were analysed with SAS version 8.1 and are presented asmean6SD. Significance was set at p<0.05. AMS incidence wascalculated as cumulative case rate. Frequencies were comparedby c2 test. Means were compared by t test. Pearson correlationwas used for relationships between AMS scores and SpO2, Hb,PAPm, and lung function measures. Lung function changes wereanalysed by two-way repeated measures analysis of variance,Tuckey’s post hoc test and t test for group comparisons. CrudeORs with 95% CIs were calculated to quantify the associationbetween smoking and AMS. Univariate logistic regression

analysis was used to estimate AMS risk for smoking versuscontrol and to examine relationships between individual vari-ables and presence of AMS. Multiple logistic regression analysiswas performed to test for the effects of independent variablesand identify the main effects. Significant risk factors wereentered into forward regression using the likelihood ratio test.The dichotomous dependent variable was AMS (0 ¼ no AMS,1¼ AMS, LLS cutoff score $4). Independent variables wereSpO2, Hb, PAPm, VC, FEV1, FEF25e75% and MVV. SpO2 wasrecoded into 0¼at least 90%, 1¼86e89% and 2¼up to 85%. Hbconcentration was recoded into 0¼up to 16 g/dl and 1¼greaterthan 16 g/dl. PAPm was recoded into 0¼up to 20 mm Hg and1¼greater than 20 mm Hg. VC, FEV1, FEF25e75% and MVV wasrecoded as 0¼normal low altitude value and 1¼abnormal, thatis, increased or decreased by >2 SD from the low altitude value.Smoking behaviour was coded as 0¼no smoking, 1¼less than 1pack/day, 2¼about 1 pack/day and 3¼more than 1 pack/day.Smoking history was coded as 0¼no smoking, 1¼short-tem,2¼medium-term and 3¼long-term smoking.

RESULTSSubjectsFour smokers withdrew before ascent and 14 were lost to follow-up at altitude for non-medical reasons; all non-smokerscompleted the study. We obtained data from 182 smokers (SMO,age 3867 years, range 25e54 years) and 200 non-smokers(CON, 3866 years, 24e56 years).

SmokingThe SMO group comprised 18% mild, 45% moderate and 37%heavy smokers. Smoking habit was 23% short term, 35%medium term and 42% long term. At high altitude, packs/daysmoked remained similar to low-altitude smoking (p>0.05).

Acute mountain sicknessAMS incidence in SMO was lower than in CON (LLS$3: 45% vs56%, c2¼4.57, p¼0.039; LLS$4: 39% vs 51%, c2¼5.53, p¼0.013;LLS$5: 3.4% vs 8.5%, c2¼4.56, p¼0.038). Five per cent ofsubjects with LLS$5 were hospitalised, more from CON thanfrom SMO (17 vs 6 cases, c2¼4.56, p¼0.038). On arrival, SMOhad a lower LLS score than CON (1.660.6 vs 1.860.7, p¼0.004).SMO with LLS$3 had lower scores than CON (3.860.5 vs4.060.6, p<0.001). At 1 week SMO still had lower scores thanCON (1.460.8 vs 1.660.5, p¼0.005). There was no altitudecerebral or pulmonary oedema. SpO2 correlated negatively withLLS score (CON: R¼�0.192, p¼0.005; SMO: R¼�0.174,p¼0.019; no difference between groups, p¼0.095). PAPm

Table 1 Symptoms and signs of acute mountain sickness in non-smokers at 4525 masl

LLS symptom intensity

n (%)

0 1 2 3 Total

Headache 46 (23) 70 (35) 56 (28) 28 (14) 154 (77)

Dizziness or light-headedness 158 (79) 22 (11) 12 (6) 8 (4) 42 (21)

Weakness or fatigue 87 (43) 68 (34) 42 (21) 3 (2) 113 (57)

Anorexia, nausea or vomiting 102 (51) 52 (26) 36 (18) 10 (5) 98 (49)

Difficulty sleeping 58 (29) 72 (36) 58 (29) 12 (6) 142 (71)

Reduction in activity 112 (56) 71 (36) 17 (9) 0 88 (44)

Change in mental status 196 (98) 3 (1.5) 1 (0.5) 0 4 (2)

Ataxia 194 (97) 6 (3) 0 0 6 (3)

Peripheral oedema 172 (86) 21 (11) 7 (3) 0 28 (14)

Comparison between control group and smoking group for headache, c2¼4.66, p¼0.031;for anorexia, nausea or vomiting, c2¼3.85, p¼0.049; for difficulty sleeping c2¼13.517,p<0.001; for all other symptoms differences were non-significant. Total: the sum ofscores >0.

Table 2 Symptoms and signs of acute mountain sickness in smokersat 4525 masl

LLS symptom intensity

n (%)

0 1 2 3 Total

Headache 104 (57) 42 (23) 22 (12) 14 (8) 78 (43)

Dizziness or light-headedness 144 (79) 22 (12) 13 (7) 3 (2) 38 (21)

Weakness or fatigue 81 (45) 50 (27) 42 (23) 6 (3) 101 (55)

Anorexia, nausea or vomiting 111 (61) 39 (21) 26 (14) 6 (3) 71 (39)

Difficulty sleeping 86 (47) 42 (23) 45 (25) 9 (5) 96 (53)

Reduction in activity 107 (59) 58 (32) 15 (8) 2 (1) 75 (41)

Change in mental status 180 (99) 2 (1) 0 0 2 (1)

Ataxia 178 (98) 4 (2) 0 0 4 (2)

Peripheral oedema 158 (87) 18 (10) 6 (3) 0 24 (13)

See table 1.

Smoking


correlated negatively with LLS score (CON: R¼�0.147,p¼0.044; SMO: R¼�0.156, p¼0.048; no difference betweengroups p¼0.075). There were no significant correlations withother variables. SMO suffered less from headache, anorexiaenauseaevomiting or sleep disturbances than CON but reportedsimilar frequency and intensity for the other LLS symptoms(tables 1 and 2). Average peak scores for separate AMSsymptoms differed significantly for headache, anorexiaenauseaevomiting and difficulty sleeping (table 3).

Lung functionOn arrival at 4525 masl VC tended to be lower in both groups(table 4). On day 3 the mean decrease was 4% and 6% in SMOand CON respectively. VC had normalised after 3 and 6 monthsin CON, but not in SMO. FEV1 and FEF25e75% were increased inCON and SMO upon arrival. They remained higher in CON,whereas they decreased in SMO over time. A similar pattern wasobserved for MVV.

Oxygen saturationLow-altitude SpO2 values were similar (CON: 9767%, SMO:9766%, p¼0.816). Upon arrival, SpO2 was lower (CON:8366%, SMO: 8365%, p¼0.001 vs low altitude, no differencebetween groups, p¼0.164). With time spent at altitude, SMOdeveloped a lower SpO2 than CON (3 months: 8565% vs8666%, p¼0.004; 6 months: 8566% vs 8666%, p¼0.002, table5). This difference was due to improvement of SpO2 in CON by3.8% and 4.1% after 3 and 6 months respectively, whereas SMOSpO2 only increased by 2.8% and 2.5% at 3 and 6 monthsrespectively (p¼0.035 and p¼0.002).

Haemoglobin concentrationInitially both groups had similar Hb (low altitude, SMO:15.862.1 g/dl, CON: 15.561.4 g/dl, p¼0.164; on arrival, SMO:16.061.8 g/dl, CON: 15.861.6 g/dl, p¼0.189). After 3 monthsthe groups differed (SMO: 16.261.8 g/dl, CON: 15.861.5 g/dl,p¼0.021). This difference was more marked after 6 months

(SMO: 17.461.6 g/dl, CON: 16.261.5 g/dl, p<0.001, seetable 5). Hb increased with packs/day (R¼0.22, p¼0.005) andyears of smoking (R¼0.23, p<0.001). At 6 months, Hb washigher in heavy and long-term smokers (17.262.1 g/dl and18.162.3 g/dl respectively) than in mild and moderate smokers(crude OR 1.1, 95% CI 1.01 to 1.26, p¼0.035) as well as short-term or medium-term smokers (crude OR 1.1, 95% CI 1.11 to1.87, p¼0.011, see table 6).

Pulmonary artery pressureAt low altitude PAPm was similar (SMO: 15.663.1 mm Hg,CON: 15.162.8 mm Hg, p¼0.101). Both groups increasedPAPm upon arrival and SMO had higher PAPm than CON(17.564.5 mm Hg vs 16.263.6 mm Hg, p¼0.005). Over timePAPm increased further (3 months, SMO: 22.464.4 mm Hg,CON: 21.563.8 mm Hg, p¼0.005; 6 months, SMO:23.164.8 mm Hg, CON: 21.764.1 mm Hg, p¼0.023, table 6).PAPm correlated with packs/day (R¼0.17, p¼0.008) andyears smoking (R¼0.19, p¼0.005). At 6 months, PAPm inheavy and long-term smokers was 24.265.2 mm Hg and24.065.7 mm Hg respectively, significantly higher than that ofmild or moderate smokers (crude OR 1.1, 95% CI 1.05 to 1.68,p¼0.048) and short-term and medium-term smokers (crude OR1.2, 95% CI 1.01 to 1.71, p¼0.031, see table 5). SpO2 correlatedpositively with PAPm in CON (R¼0.158, p¼0.019) andnegatively in SMO (R¼�0.163, p¼0.023).

Logistic regressionAt altitude, subjects with SpO2 #85% were 2.6 times morelikely to have AMS than those with SpO2 $90% (table 7). Hb,PAPm, and lung function variables did not show significanteffects. Crude ORs of FEV1, FEF25-75% and MVV were similar tothose of VC (not shown). Heavy smoking and medium-term orlong-term smoking history decreased AMS risk (table 7). Inmultivariate logistic regression only SpO2, smoking habits andsmoking history had significant effects (table 8).

DISCUSSIONAcute mountain sicknessWe found an 11e12% (20e24% relative) lower incidence of AMSfor LLS cut-off scores $3 and $4 respectively in smokerscompared with non-smokers. This contrasts with studies onAMS risk in tourists and climbers,5 6 but confirms a tendencyfound in a prospective cohort study (crude OR 0.66, 95% CI 0.41to 1.07, p¼0.09).7 Hultgren2 (p. 469) hypothesised that smokerswould have more AMS and have problems acclimatising because

Table 3 Mean (6SD) peak scores of Lake Louise Scoring symptoms

Symptom CON SMO p Value

Headache 1.3360.56 0.7060.42 <0.001

Dizziness or light-headedness 0.3560.30 0.3160.18 0.192

Weakness or fatigue 0.8860.26 0.8960.28 0.755

Anorexia, nausea or vomiting 0.7760.42 0.6060.36 <0.001

Difficulty sleeping 1.4860.45 0.9560.38 <0.001

CON, control group; SMO, smoking group.

Table 4 Pulmonary function for SMO versus CON

Parameters Low altitude After arrival Day 3 3 months 6 months Pa Pb

VC (l) SMO 4.4860.46 SMO 4.3660.63 SMO 4.2460.60 SMO 4.1260.62 SMO 4.1060.38 0.046 G: 0.110

CON 4.5460.44 CON 4.4360.45 CON 4.1260.63 CON 4.5260.32* CON 4.5060.26y 0.013 I: 0.027

FEV1 (l) SMO 3.9260.82 SMO 3.9860.74 SMO 4.0260.84 SMO 3.9660.78 SMO 3.8260.74 0.321 G: 0.164

CON 4.1060.48 CON 4.2160.51 CON 4.1860.62 CON 4.2860.46z CON 4.2260.50z 0.388 I: 0.465

FEF25e75% (litres/s) SMO 4.0861.05 SMO 4.0160.85 SMO 4.0661.06 SMO 4.0261.12 SMO 3.9260.92 0.465 G: 0.044

CON 4.1660.85 CON 4.2460.66 CON 4.2860.72 CON 4.8160.63x CON 4.9360.67{ 0.006 I: 0.048

MVV (litres/min) SMO 108.064.5 SMO 110.264.4 SMO 111.664.8 SMO 106.365.2 SMO 107.465.3 0.035 G: 0.002

CON 111.365.2 CON 115.565.6** CON 117.464.6** CON 118.363.8** CON 117.864.5** 0.002 I: 0.001

Data are presented as mean 6 SD. Pa: ANOVA for repeated measures within each group separately. Pb: ANOVAdprobabilities between groups (G), and interaction (I). Group comparisons:*SMO versus CON, p<0.001; ySMO versus CON p<0.001, zSMO versus CON, p¼0.048 and p¼0.036; xSMO versus CON, p¼0.048; {SMO versus CON, p¼0.021; **SMO versus CON,p<0.001.Because of technical problems only a subset of subjects had pulmonary function tests: at low altitude 40 (SMO) and 42 (CON), after arrival 36 (SMO) and 28 (CON), at 3 days 36 (SMO) and 28(CON), at 3 months 32 (SMO) and 34 (CON) and at 6 months 28 (SMO) and 25 (CON). The measurements reported concern the same subjects over time.SMO, smoking group; CON, control group; VC, vital capacity; FEV1, forced expiratory volume in 1 s; FEF25-75% forced expiratory flow between 25% and 75% of vital capacity; MVV, maximalvoluntary ventilation.

Smoking


of aggravated hypoxaemia through diminished blood oxygen-carrying capacity from carboxyhaemoglobin (COHb), decreasedoxygen uptake due to the respiratory effects of smoking, andimpaired peripheral oxygen extraction. One study reported thata combination of smoking and alcohol impeded altitude accli-matisation to 3200 masl in lowland workers, but did not reportAMS.8

Headache, gastrointestinal upset and sleep disturbanceDifferences in AMS incidence and severity were small butstatistically highly significant. Of limited clinical relevance theyare of interest for AMS pathophysiology. Smokers had lessheadache, anorexiaenauseaevomiting and sleep disturbance.AMS headache may result from hypoxia-induced cerebral vaso-dilatation or its effectors, such as nitric oxide (NO), perhapsthrough activation of the trigeminovascular system9 and cere-bral venous hypertension.10 At low altitude NO plays a role intension type headache and NO prodrugs are associated withheadache and nausea.11 Nitroglycerin causes headache andexacerbates AMS12 as does sildenafil.13 Smoking impairs endo-thelial function, decreasing NO formation and increasing NOdegradation14 and smokers expire less NO.15 We speculatethat decreased NO levels protected smokers somewhat fromheadache and gastrointestinal upset.

Smokers reported fewer sleep problems. Altitude exposureinduces a periodic breathing pattern.16 The oscillations resultfrom high ventilatory sensitivity to carbon dioxide (CO2) andhypoxia in the presence of narrowed CO2 reserve and inducefrequent arousals from sleep. Nicotine, NO and carbonmonoxide (CO) influence the regulation of breathing.17 18 Wespeculate that smokers slept better because of less breathinginstability through higher nicotine and CO, and lower NOlevels.

Pulmonary arterial pressureSmokers tended to have higher PAPm at low altitude, which isexpected since smoking increases PAP.19 At altitude, both groupshad increased PAPm, which was expected since hypoxiaincreases PAPm.20 Smokers had higher PAPm, with a smallsignificant difference between smokers and non-smokers at 3and 6 months, which was more pronounced in heavy smokers,

suggesting a doseeresponse effect. Increased PAP at altitude isassociated with high altitude pulmonary oedema (HAPE).21

People prone to HAPE exhale less NO. PAP can be lowered byinhaling NO, and increasing NO with tadalfil prevents HAPE.21

Since smoking impairs NO bioavailability14 19 and lowersexhaled NO levels15 we explain our findings of higher PAPm insmokers in part from decreased NO bioavailability in thepulmonary circulation.

SaturationIncreased SpO2 with time in non-smokers reflects ventilatoryacclimatisation to altitude.22 At low arterial oxygen pressure(PaO2) peripheral chemoreceptor activation induces hyperven-tilation. The sensitivity of this pathway increases with time.22

Smokers showed less increase in SpO2 at 3 and 6 months. Thissuggests that smoking hampers ventilatory acclimatisation to

Table 5 Oxygen saturation, haemoglobin concentration and meanpulmonary artery pressure

N SpO2 (%) Hb (g/dl) PAPm (mm Hg)

Low altitude

SMO 182 96.566.4 15.862.1 15.663.1

CON 200 97.266.8 15.561.4 15.162.8

After arrival at altitude

SMO 182 82.565.2 16.061.8 17.564.5

CON 200 83.265.8 15.861.6 16.263.6

After 3 months

SMO 182 84.864.6* 16.261.8z 22.464.4{CON 200 86.465.7 15.861.5 21.563.8

After 6 months

SMO 182 84.666.3* 17.461.6x 23.164.8**

CON 200 86.665.7 16.261.5 21.764.1

Data are presented as mean6SD.*SMO versus CON: p¼0.004.ySMO versus CON: p¼0.001.zSMO versus CON: p¼0.021.xSMO versus CON: p<0.001.{SMO versus CON: p¼0.044.**SMO versus CON: p¼0.004.SMO, smoking group; CON, control group; Hb, haemoglobin concentration; PAPm, meanpulmonary artery pressure; SpO2, arterial oxygen saturation.

Table 6 Oxygen saturation, haemoglobin concentration and meanpulmonary artery pressure at 6 months and intensity/history of smoking

Smoking intensity/smoking history n SpO2 Hb (g/dl)

PAPm(mm Hg)

Mild 33 84.861.7 16.361.6 22.563.6

Moderate 82 84.262.2 16.561.7 22.864.4

Heavy 67 83.162.5* 17.262.1z 24.265.2{Short term 41 85.262.4 16.661.6 22.163.5

Medium term 65 84.761.6 17.261.8 22.564.5

Long term 76 84.262.1y 18.162.3x 24.065.7**

Data are presented as mean6SD.*Heavy levels of smoking versus moderate (p¼0.002) and mild levels of smoking(p<0.001).yLong-term smoking versus medium-term (p¼0.009) and short-term smoking (p<0.001).zHeavy levels of smoking versus moderate (p¼0.036) and mild levels of smoking(p<0.001).xLong-term smoking versus medium-term (p¼0.008) and short-term smoking (p<0.001).{Heavy levels of smoking versus moderate (p¼0.048) and mild levels of smoking(p¼0.040).**Long-term smoking versus medium-term (p¼0.057) and short-term smoking (p¼0.044).Hb, haemoglobin concentration; PAPm, mean pulmonary artery pressure; SpO2, arterialoxygen saturation.

Table 7 Results of multiple univariate regression analysis (unadjusted)for the variables in the left column

Variables AMS, n (%) Crude OR (95% CI) p Value

SpO2 (%)

$90 32 (33) 1 (ref)

86e89 65 (40) 0.986 (0.647 to 1.545) 0.069

#85 76 (62) 2.630 (2.156 to 3.274) <0.001

Hb (g/dl)

#16 142 (46) 1 (ref)

>16 31 (42) 0.745 (0.504 to 0.762) 0.238

PAPm (mm Hg)

#20 164 (45) 1 (ref)

>20 9 (45) 0.645 (0.446 to 0.672) 0.164

VC

Normal 168 (46) 1 (ref)

Less by $2 SD 5 (39) 0.211 (0.096 to 0.747) 0.770

Smoking

No smoking 102 (51) 1 (ref)

<1 pack/day 16 (49) 0.860 (0.674 to 0.901) 0.755

1 pack/day 30 (37) 0.786 (0.652 to 0.810) 0.035

>1 pack/day 25 (37) 0.627 (0.335 to 0.856) 0.039

Smoking history

No smoking 102 (51) 1 (ref)

Short term 17 (40) 0.864 (0.520 to 0.978) 0.465

Medium term 25 (39) 0.818 (0.465 to 1.075) 0.044

Long term 29 (38) 0.654 (0.358 to 0.861) 0.027

Smoking


high altitude. Chemoreceptor function is modulated by NO andCO.17 Nicotine increases peripheral chemoreflex sensitivity toreductions in arterial oxygen content in non-smokers but not insmokers.18 In people who live at altitude all their lives, a decreasein ventilation may eventually develop.23 The reduced ventilatorydrive results from less sensitivity of central chemoreceptors forCO2 and of peripheral chemoreceptors for hypoxia, and leads topolycythaemia.23 Since smoking is a risk factor for thissyndrome24 we speculate that smokers showed reduced venti-latory acclimatisation from reduced chemoreceptor sensitivity.

Most oximeters, including ours, interpret carboxy-haemoglogin as O2 saturation of Hb (HbO2) and thus indicatean erroneously high SpO2 in smokers.25 Since at altitude alveolaroxygen pressure (PAO2) and PaO2 decrease while alveolar carbondioxide pressure (PACO) remains similar (assuming CO exposurefrom smoking invariable), competition between CO and O2

increases COHb.26 Since increased COHb in smokers displacesthe HbO2 dissociation curve leftward, smokers likely had lowerPaO2, in line with reduced peripheral chemoreceptor sensitivityin smokers. Brewer et al26 indeed found lower PaO2 in smokersat 3100 masl than in non-smokers (53.465.8 mm Hg vs58.664.2 mm Hg).

Smoking polycythaemiaSmoking causes polycythaemia. The tendency for higher Hb insmokers at low altitude became significant at 3 and 6 months ataltitude. This increased blood oxygen carrying capacity, correctingfor decreased saturation, as previously reported.26 Smoking-induced and hypoxia-induced erythropoiesis increased Hb more insmokers, placing them at higher risk of developing chronicmountain sickness if they remained at altitude for years.23 24

Lung functionVC tended to decrease upon arrival at altitude and normalisedwith time in non-smokers but not in smokers. Previous studiesreported a decrease in VC during the first 12e24 h of altitudeexposure.27 This fall may be caused by increased pulmonaryblood volume and mild interstitial oedema.28 VC changes uponarrival were not related to SpO2, but subjects with AMS hadgreater decreases in VC than those without (4.060.7 vs4.260.5%, p<0.002).

Since air density decreases with altitude, increases in FEV1 andFEF25e75% were expected. But data in the literature areconflicting. FEV1 was found to increase,29 decrease30 or remain

unchanged at altitude.31 After arrival at altitude, FEV1 wasslightly higher in smokers and non-smokers. This increasepersisted over time in non-smokers but had decreased after6 months in smokers, suggesting a decrease in lung functionfrom smoking. FEF25e75% was increased in non-smokers to 115%and 118% of low altitude values at 3 and 6 months, respectively.In smokers FEF25e75% was similar to low altitude at 3 monthsand had decreased by 4% at 6 months. As expected, MVVincreased in non-smokers and remained elevated at 3 monthsand 6 months. By contrast, MVV decreased throughout thealtitude stay in smokers and overall smokers showed loss of lungfunction while at altitude.

Strengths and limitationsThe main strength of our study is the inclusion of almost 200smokers, allowing effects to be identified that were previouslyundiscovered. We did not measure exhaled levels of CO and NO,or blood gases or COHb levels, to relate these to AMS symptomscores. Sleep quality measured with actimetry, and quantifica-tion of the ventilatory response to hypoxia and hypercapniamight have provided further insight too. Since smoking wasreported to reduce pain perception, we cannot fully exclude thefact that the perception of severity of symptoms of AMS wasless in smokers compared with non-smokers.32

Smoking and healthPresenting ‘positive’ effects of smoking is uncomfortable;smoking must be strongly discouraged. We do not recommendsmoking to prevent AMS. First, we did not study the effects innon-smokers but investigated habitual smokers. Second,smoking is strongly addictive and increases the risk of cardio-respiratory and other diseases, including cancer.33 Third, altitudeis accompanied by cold exposure and smoking increases the riskof frostbite.34 Fourth, smoking decreases exercise capacity.35

Fifth, smoking represents risk for others because of secondhandsmoke.33 And finally, the effect on AMS risk and severity wassmall. Gradual ascent and sufficient time for acclimatisation arebest for AMS prevention.9

CONCLUSIONWe found that non-acclimatised smokers are at slightly reducedrisk for AMS at altitude but acclimatise less well. We do notrecommend smoking as a preventive measure for AMS buthighlight the effects of smoking on NO metabolism and the

Table 8 Results of multiple logistic regression analysis (adjusted) with the variables in the left columnretained in the final regression (all other variables not significant)

Variables b SE Wald Adjusted OR (95% CI) p Value

SpO2 #85%

Non-smoking �0.048 0.481 14.12 2.343 (2.121 to 2.895) 0.001

Smoking �0.062 0.474 15.86 2.584 (1.984 to 3.365) 0.001

Smoking

No smoking

<1 pack/day 0.084 0.382 3.84 0.865 (0.583 to 1.014) 0.075

1 pack/day �0.051 0.257 4.17 0.662 (0.424 to 0.898) 0.031

>1 pack/day �0.056 0.236 4.28 0.646 (0.328 to 0.856) 0.020

Smoking history

No smoking

Short term �0.086 0.408 3.86 0.748 (0.482 to 1.083) 0.057

Medium term �0.068 0.252 4.04 0.674 (0.412 to 0.767) 0.027

Long term �0.082 0.186 4.16 0.636 (0.318 to 0.825) 0.021

SpO2, arterial oxygen saturation.

Smoking


potential roles for CO, nicotine or other active compoundsfound in cigarette smoke in adaptation to altitude.

DISCLOSURESince it is well documented that the tobacco industry has beenmanipulating science, scientists and the general public fordecades, the present authors declare that none of them has orhas ever had any ties to the tobacco industry and that this studyis independent from any financial or other influence from thetobacco industry.

Contributors TYW conceived the study, analysed the data and participated in writing;SQD, JLL, JHJ, ZCC, RCD, JZZ and QDT collected and analysed data; BK participatedin data analysis, interpretation of the results and writing the final manuscript.

Funding This study was supported by competitive grants NNSF-30393130,973-2012CB518202 and EAF XZ-101, People’s Republic China.


Ethics approval China National Science Foundation (NNSF) and the Qinghai HighAltitude Medical Research Institute Committee on Human Research.


Data sharing statement Any interested scholars can ask the corresponding authorfor an access to the original data.

REFERENCES1. GATS. Global Adult Tobacco Survey (GATS)dChina Fact Sheet 2010. http://www.

who.int/tobacco/surveillance/en_tfi_china_gats_factsheet_2010.pdf2. Hultgren HN. High Altitude Medicine. Stanford: Hultgren Publications, 1997.3. MacLean N. Smoking and acclimatisation to altitude. Br Med J 1979;2:799.4. Roach RC, Bartsch P, Oelz O, et al. The Lake Louise Acute Mountain Sickness

Scoring System. Hypoxia and Molecular Medicine. Burlington, Vermont: Queen CityPress, 1993:272e4.

5. Gaillard S, Dellasanta P, Loutan L, et al. Awareness, prevalence, medication use,and risk factors of acute mountain sickness in tourists trekking around theAnnapurnas in Nepal: a 12-year follow-up. High Alt Med Biol 2004;5:410e19.

6. Schneider M, Bernasch D, Weymann J, et al. Acute mountain sickness: influence ofsusceptibility, preexposure, and ascent rate. Med Sci Sports Exerc2002;34:1886e91.

7. Richalet JP, Larmignat P, Poitrine E, et al. Physiological risk factors of severe highaltitude illness: a prospective cohort study. Am J Respir Crit Care Med2012;185:192e8.

8. Lindgarde F, Lilljekvist R. Failure of long-term acclimatization in smokers moving tohigh altitude. Acta Med Scand 1984;216:317e22.

9. Hackett PH, Roach RC. High-altitude illness. N Engl J Med 2001;345:107e14.10. Wilson MH, Imray CH, Hargens AR. The headache of high altitude and

microgravitydsimilarities with clinical syndromes of cerebral venous hypertension.High Alt Med Biol 2011;12:379e86.

11. Gupta S, Nahas SJ, Peterlin BL. Chemical mediators of migraine: preclinical andclinical observations. Headache 2011;51:1029e45.

12. Mazzuero G, Mazzuero A, Pascariello A. Severe acute mountain sickness andsuspect high altitude cerebral edema related to nitroglycerin use. High Alt Med Biol2008;9:241e3.

13. Ghofrani HA, Reichenberger F, Kohstall MG, et al. Sildenafil increased exercisecapacity during hypoxia at low altitudes and at Mount Everest base camp:a randomized, double-blind, placebo-controlled crossover trial. Ann Intern Med2004;141:169e77.

14. Toda N, Toda H. Nitric oxide-mediated blood flow regulation as affected by smokingand nicotine. Eur J Pharmacol 2010;649:1e13.

15. Malinovschi A, Janson C, Holmkvist T, et al. Effect of smoking on exhaled nitricoxide and flow-independent nitric oxide exchange parameters. Eur Respir J2006;28:339e45.

16. Bloch KE, Latshang TD, Turk AJ, et al. Nocturnal periodic breathing duringacclimatization at very high altitude at Mount Muztagh Ata (7,546 m). Am J Resp CritCare 2010;182:562e8.

17. Prabhakar NR. NO and CO as second messengers in oxygen sensing in the carotidbody. Respir Physiol 1999;115:161e8.

18. Argacha JF, Xhaet O, Gujic M, et al. Nicotine increases chemoreflex sensitivity tohypoxia in non-smokers. J Hypertens 2008;26:284e94.

19. Wright JL, Levy RD, Churg A. Pulmonary hypertension in chronic obstructivepulmonary disease: current theories of pathogenesis and their implications fortreatment. Thorax 2005;60:605e9.

20. Moudgil R, Michelakis ED, Archer SL. Hypoxic pulmonary vasoconstriction. J ApplPhysiol 2005;98:390e403.

21. Scherrer U, Rexhaj E, Jayet PY, et al. New insights in the pathogenesis of high-altitude pulmonary edema. Prog Cardiovasc Dis 2010;52:485e92.

22. Joseph V, Pequignot JM. Breathing at high altitude. Cell Mol Life Sci2009;66:3565e73.

23. Leon-Velarde F, Villafuerte FC, Richalet JP. Chronic mountain sickness and theheart. Prog Cardiovasc Dis 2010;52:540e9.

24. Pei SX, Chen XJ, Si Ren BZ, et al. Chronic mountain sickness in Tibet. Q J Med1989;71:555e74.

25. Tremper KK. Pulse oximetry. Chest 1989;95:713e15.26. Brewer GJ, Eaton JW, Grover RF, et al. Cigarette smoking as a cause of hypoxemia

in man at altitude. Chest 1971;59:Suppl: 30S+.27. Basu CK, Selvamurthy W, Bhaumick G, et al. Respiratory changes during initial

days of acclimatization to increasing altitudes. Aviat Space Environ Med1996;67:40e5.

28. Pellegrino R, Pompilio P, Quaranta M, et al. Airway responses to methacholineand exercise at high altitude in healthy lowlanders. J Appl Physiol2010;108:256e65.

29. Gautier H, Peslin R, Grassino A, et al. Mechanical properties of the lungs duringacclimatization to altitude. J Appl Physiol 1982;52:1407e15.

30. Pollard AJ, Mason NP, Barry PW, et al. Effect of altitude on spirometric parametersand the performance of peak flow meters. Thorax 1996;51:175e8.

31. Mason NP, Barry PW, Pollard AJ, et al. Serial changes in spirometry duringan ascent to 5,300 m in the Nepalese Himalayas. High Alt Med Biol 2000;1:185e95.

32. Girdler SS, Maixner W, Naftel HA, et al. Cigarette smoking, stress-inducedanalgesia and pain perception in men and women. Pain 2005;114:372e85.

33. Edwards R. The problem of tobacco smoking. Br Med J 2004;328:217e19.34. Ervasti O, Juopperi K, Kettunen P, et al. The occurrence of frostbite and its risk

factors in young men. Int J Circumpolar Health 2004;63:71e80.35. Forte VA, Muza SR, Fulco CS, et al. Smoking accentuates the decrement in maximal

oxygen uptake at high altitude. FASEB J 1996;10:3723.


Smoking


ORIGINAL ARTICLE

The effect of continuous positive airway pressureusage on sleepiness in obstructive sleep apnoea: realeffects or expectation of benefit?

Megan R Crawford,1,2 Delwyn J Bartlett,1 Steven R Coughlin,3,4 Craig L Phillips,1,5

Alister M Neill,6 Colin A Espie,1,2 George C Dungan II,1 John P H Wilding,3,4

Peter M A Calverley,3,4 Ronald R Grunstein,1,7 Nathaniel S Marshall1

ABSTRACTRationale Placebo responses are complexpsychobiological phenomena and often involve patientexpectation of benefit. With continuous positive airwaypressure (CPAP) treatment of obstructive sleep apnoea,greater hours of CPAP use are associated with reducedsleepiness. However, these open-label studies have notcontrolled for patient expectation of benefit derived fromtheir knowledge of hours of device use.Objectives To investigate the relative effectiveness ofthe use of real or placebo CPAP on daytime sleepiness.Methods Patient-level meta-analysis combining data onsleepiness measured by the Epworth Sleepiness Scalefrom three randomised placebo-controlled crossovertrials. Mixed model analysis of variance was used toquantify the effects of real versus placebo devicetreatment, usage, their interaction and regression to themean.Measurements and main results Duration of real andplacebo CPAP use was correlated within patients(r¼0.53, p<0.001). High use of real CPAP reducedsleepiness more than high use of placebo (difference 3.0points; 95% CI 1.7 to 4.3, p<0.001) and more than lowuse of real CPAP (difference 3.3; 95% CI 1.9 to 4.7,p<0.0001). High use of placebo was superior to low useof placebo (difference 1.5; 95% CI 0.1 to 2.8, p¼0.03).Twenty-nine per cent of the effect of high usage of CPAP(4.2 points; 95% CI 3.3 to 5.1) was explained by theexpectation of benefit effect associated with high use ofplacebo (1.2 points ; 95% CI 0.2 to 2.3).Conclusions A clinically significant proportion of theeffectiveness of high CPAP use in reducing sleepiness isprobably caused by patient expectation of benefit.

Poor compliance with medical treatments isa major barrier to clinical effectiveness for manychronic conditions.1 2 Recent studies suggest thatplacebo effects may involve crucial psychobiolog-ical factors influencing treatment effectiveness andclinical practice.3 Placebo treatments are notnecessarily inert; one principal component is theexpectation of future benefit responses followingadministration of a placebo.3 4 Patients who accepttreatment are expecting that it will help them.These observations have primarily been made ina range of pharmaceutical and other therapeuticinterventions. However, the availability of real andplacebo continuous positive airway pressure

(CPAP) for obstructive sleep apnoea (OSA) providesan almost unique opportunity to examine therelationships between treatment usage, expectationof benefit, relief of symptoms and the placeboeffect, including expectation of benefit.Unblinded cohort studies have shown

a doseeresponse relationship in which greaterusage of CPAP was associated with better symp-tomatic outcomes for patients.5e8 However,because these studies were open label, patientswere largely aware of how many hours a night theyhad been using CPAP. Therefore it is possible thatsome of the symptomatic benefit associated withgreater use might have arisen from an expectationof benefit.Randomised placebo CPAP controlled crossover

trials provide a method to determine whether thesymptomatic response to higher use of CPAP isrelated to a real effect or expectation of benefitfrom their high use. Given the complexity ofhuman psychobiology, crossover trials offer a bettercomparison as patients act very well as their owncontrols because usage of real and placebo CPAP

Key messages

What is the key question?< How much of the benefit associated with high

compliance with continuous positive airwaypressure (CPAP) for obstructive sleep apnoeais caused by real treatment effects and howmuch by patients’ expectations of benefit?

What is the bottom line?< High use of CPAP does confer real benefits but

about 29% of the benefit felt by highly adherentpatients in regular clinical care is due toexpectation of benefit.

Why read on?< Clinical care is a complex psychobiological

environment so the benefits that seem toaccrue to patients who effectively use treat-ments are partly caused by real effects of thosetreatments and partly by the expectations thatthese highly adherent patients bring with theminto the clinical environment.

< An additional figure ispublished online only. To viewthis file please visit the journalonline (http://dx.doi.org/10.1136/thoraxjnl-2012-201622).1NHMRC Centre for IntegratedResearch and Understanding ofSleep (CIRUS), WoolcockInstitute of Medical Research,Sydney Medical School,University of Sydney, Sydney,Australia2University of Glasgow SleepCentre, University of Glasgow,Glasgow, UK3Department of Obesity andEndocrinology, Clinical SciencesCentre, University of Liverpool,Liverpool, UK4Department of RespiratoryMedicine, Clinical SciencesCentre, University of Liverpool,Liverpool, UK5Department of Respiratory andSleep Medicine, Royal NorthShore Hospital Sydney,Australia6WellSleep, Department ofMedicine, University of Otago atWellington, Wellington, NewZealand7Department of Respiratory andSleep Medicine, Royal PrinceAlfred Hospital, Sydney,Australia

Correspondence toDr Nathaniel S Marshall, SydneyNursing School, University ofSydney, Blg MO2, University ofSydney, Sydney, NSW 2006,Australia; [email protected]

A version of this study has beenpresented in poster format atthe 20th Congress of theEuropean Sleep ResearchSociety, Lisbon, Portugal,September 2010, at theAustralasian Sleep Associationmeeting in Christchurch, NewZealand, October 2010 and atthe World Sleep Federationmeeting in Kyoto, Japan,October 2011.

Received 11 January 2012Accepted 23 April 2012

Sleep



seems to be highly correlated.9 Placebo control via an almostidentical sham device also helps to quantify the size of theexpectation of benefit effect associated with using CPAPtherapy. We combined three crossover trials9e11 in an individualpatient-level meta-analysis to quantify the relative effects of realand placebo CPAP compliance on sleepiness measured using theEpworth Sleepiness Scale (ESS).

METHODSParticipantsWe combined data from the 91 patients who completed one ofthe crossover trials. Detailed descriptions of the study designs,patients’ characteristics, and primary outcome findings of thestudies can be found in the original publications.9e11 Table 1briefly describes the patients included in the analysis.

Study design and procedureTrials were combined as they had congruent study designs withall patients receiving in random order standard individuallytitrated real CPAP and placebo CPAP (sometimes called ‘shamCPAP’ and simply described here as placebo). In all trials theplacebo was identical to the CPAP machine in terms of noise,mask temperature, mask humidity and airflow through theexhalation port. To create a sub-therapeutic treatment, theplacebo machine was set to 8 cm H2O, yet delivered <1.0 cmH2O pressure. The datasets were merged and we analysed theeffects of real and placebo treatment and the effects of greateruse of both of these devices on sleepiness as measured by theESS.12 The ESS was the only outcome in common among thetrials.

Data preparation and statistical analysesStatistical analyses were performed using SAS (V.9.2) and SPSSfor Windows (V.17). Continuous variables were presented asmean 6 SD or 95% CIs, and p values of <0.05 were consideredstatistically significant. We considered the interaction ofcompliance and treatment to be significant when p<0.1 becausethis was the primary screening method for our hypothesis.The specific group-by-group analyses arising from such aninteraction were then judged by the p<0.05 criteria.

We used mixed model analyses of variance (using variancecomponents structure of covariance) to quantify the effects ofusage (high vs low cut at 4 h/night) and the interaction betweentreatment and compliance. These models also included theeffects of treatment (CPAP or placebo) and regression to themean (the baseline severity in ESS). A priori we specificallytested the following: the superiority of high CPAP use comparedwith high placebo use; the superiority of high CPAP usecompared with low CPAP use; and the superiority of highplacebo use compared with low placebo use (all using p<0.05 asthe critical threshold level). Inter-trial and inter-individual vari-ability were classified as random effects; all previous variables

(treatment, compliance, regression to the mean, and the inter-action between treatment and compliance) were fixed effects.We also estimated the proportion of improvement associatedwith high use in clinical practice that is probably attributable toexpectation of benefit by dividing the effect associated with highplacebo use by the effect associated with high CPAP use. Effectsizes were calculated by dividing the mean effects by the SD ofthe ESS, which is often around 4 points in clinical and popula-tion samples. Small effect sizes are between 0.20 and 0.50,medium between 0.50 and 0.80, and large over 0.80.13

These main analyses were conducted using compliance asa dichotomous variable based on compliance cut-off at thestandard 4 h/night. In a second model we investigated therelationship between outcome and compliance as a continuousvariable using the mixed model analysis described above. Insensitivity analyses we tested different dichotomous cut-offpoints and whether the order the treatment was given may haveinfluenced our final conclusions by using an order and an orderby treatment interaction in our final model. We also examinedthe correlation between CPAP and placebo use.

RESULTSCorrelation between placebo and real CPAP complianceReal and placebo device usage was correlated (r2¼0.53, p<0.001;see figure 1). Individuals who used placebo for more than 4 h/night were also very likely to be high CPAP users with only threeof those patients having compliance below 4 h when on CPAP(see figure 1).

Epworth Sleepiness Scale (ESS)High use was associated with superior improvement in the ESSacross both treatments (mean difference 2.2 points; 95% CI 1.0to 3.3, p<0.001; effect size (ES) based on a SD of 4 points 0.55).Real CPAP improved sleepiness more than placebo irrespective ofusage (2.1; 95% CI 1.1 to 3.0, p<0.001; ES 0.53). The interactionbetween hours of usage and type of treatment (real or placebo)was significant (p¼0.056), using a significance threshold level of0.1 for investigating the specific comparisons of interest. Theeffect on ESS of high use of real CPAP was greater than higherusage of placebo CPAP (difference 3.0 points; 95% CI 1.7 to 4.3,p<0.001; ES¼0.75) and low use of CPAP (3.3; 95% CI 1.9 to 4.7,p<0.0001; ES¼0.83). Additionally, high placebo use was betterthan low placebo use (1.5; 95% CI 0.1 to 2.8, p¼0.03; ES¼0.33).These effects were not influenced by order or order by treatmentinteraction. These results are presented in table 2 and figure 2.High use of CPAP was associated with a 4.2-point reduction in

Epworth score (95% CI 3.3 to 5.1 points, p<0.0001) and high useof placebo was associated with a 1.2-point reduction in Epworthscore (95% CI 0.2 to 2.3 points, p¼0.03). As such, we estimateabout 29% of the improvement in Epworth scores associatedwith high CPAP compliance seen in clinical practice is probablyattributable to expectation of benefit in patients who are high

Table 1 Patient characteristics across the three trials

Trial 1 (Marshall et al)9 Trial 2 (Coughlin et al)11 Trial 3 (Phillips et al)10

Gender (women) 7/29 (24%) 0/34 (0%) 3/28 (11%)

Age, median (years) 50.6 (range 25e67) 49.0 (SD 8.3) 48.8 (range 25e72)

Apnoea hypopnoea index (SD) 21.6 (7.5) 39.7 (13.8) 38.68 (24.04)

Body mass index (kg/m2) (SD) 31.5 (6.0) 36.1 (7.6) 31.7 (4.1)

Baseline Epworth score (SD) 12.5 (4.1) 13.8 (4.9) 10.3 (4.8)

Prescribed CPAP pressure (cm H2O) 7 (range 5e10) 10.0 (IQR 8e10) 11 (range 7e18.5)

CPAP, continuous positive airway pressure.

Sleep


users of CPAP (ie, 1.2 points on high-use placebo divided by 4.2points on high-use CPAP).

We also investigated usage as a continuous linear variable. Inthese models, both treatment and usage were significantlyrelated to Epworth improvement in the expected direction(mean estimate 2.1 points (95% CI 1.2 to 2.9) and 0.6 points perhour (95% CI 0.3 to 0.9), respectively, both p<0.0001). Incontrast to the findings in the dichotomous model, there was nosignificant interaction between treatment type and usage dura-tion (p¼0.54). In other words, the effect of duration of use,measured as a continuous linear variable, on ESS did notsignificantly differ between CPAP and placebo devices.

In sensitivity analyses we analysed the cut-off points for highand low use using 5, 6.5 and 7.5 h/night. The combination ofthese analyses, plotting raw data and fitting curves to the dataindicated the likely existence of no additional benefit beyondabout 5.5 h, but with significant uncertainty about the exactlocation of the asymptote (see online supplementary figure).

DISCUSSIONUsing data from three randomised placebo CPAP controlledcrossover trials, we estimate that 29% of the reduction insleepiness seen in patients with OSA who use CPAP for morethan 4 h a night is probably caused by an expectation of benefit.The expectation of benefit is one of the components of theplacebo effect and is caused by patients inferring benefit fromtreatments they choose to use. Conversely, patients who do notuse treatments do not expect to feel better. Because clinicalinteraction is such a complex psychobiological phenomenon,these effects are best quantified in randomised crossover trials in

which patients act as their own controls and tend to use theactive treatment (CPAP) as often as they use the placebo.Previous open-label cohort studies5e7 and parallel design

clinical trials14 15 have shown that there is a dose-dependentassociation between greater use of CPAP and better patientoutcomes.16 We analysed three placebo CPAP controlled rando-mised crossover trials to distinguish between symptomimprovement resulting from real treatment use from improve-ment related to placebo-like expectation of benefit for eachpatient (see figure 2 and table 2). As expected, real CPAP hada beneficial effect over placebo as did high use of either devicecompared with low use. Fortunately, not all of the benefits ofhigh use were driven by expectation of benefit as patients usingCPAP for more than 4 h a night had greater reduction in sleep-iness than patients using placebo for more than 4 h/night. Weinfer expectation of benefit effects because patients usingplacebo for more than 4 h a night exhibited greater reduction insleepiness than patients who used placebo for less than 4 ha night, despite this device having no physiological effect onOSA. These high users of placebo were also 94% likely to be highusers of CPAP (46 out of 49, see figure 1) indicating the existenceof a compliant patient phenotype who will use the deviceregardless of its ability to control OSA. Conversely, there werealso patients who used placebo CPAP less than normal CPAP(see the lower right quadrant of figure 1). This may be furtherevidence that patients who feel benefit from CPAP treatmenttend to use it more.5

The main effect reported here cannot be observed in clinicalpractice because patients in routine clinical care are not givenplacebo CPAP therapy. Additionally, placebo devices are notcommonly employed in crossover trials due to a fear of

Figure 1 Patient use of treatment is stable between a real continuouspositive airway pressure (CPAP) device and a placebo version of thesame device (r2¼0.53, p<0.001). X and Y axes are in units of averagehours per night of use.

Table 2 Effects of high and low CPAP and placebo use on sleepiness

High vs low use CPAP vs placeboTreatment byuse interaction

High-CPAP vshigh-placebo use

High-placebo vslow-placebo use

ESS (24 points) 2.2 (1.0 to 3.3), p<0.001 2.1 (1.1 to 3.0), p<0.001 p¼0.056 3.0 (1.7 to 4.3), p<0.001 1.5 (0.1 to 2.8), p¼0.03

Data are presented as mean points of improvement from baseline (95% CI), p value. The effects of compliance, treatment and the interaction between treatment and compliance are presented.The compliance effect tests whether high use is better than low use; the treatment effect tests whether CPAP was better than placebo; and the interaction tests whether the effect of high usediffers between CPAP and placebo.CPAP, continuous positive airway pressure; ESS, Epworth Sleepiness Scale.

Figure 2 Effects of different treatment types and use on improvementsmeasured by the Epworth Sleepiness Scale (ESS; y axis is in points ofimprovement from baseline in the 24-point ESS). Diamonds indicate theestimated mean effects (bars the 95% CI). High use (>4 h/night) ofeither machine confers greater benefits but the symptomatic benefitsaccruing to high users are greater when using real CPAP than placeboCPAP. CPAP, continuous positive airway pressure.

Sleep


unblinding,17 so the number of clinical trials available for thissort of analysis is limited.

The uncertainty about whether the dose effect is linear,stepped or asymptotic led us to treat compliance as a dichoto-mous (the widely used 4 h/night threshold value) and a contin-uous hours of use variable.6 When compliance is treated asa continuous linear variable in the same models, treatment withCPAP and high usage confer significant benefits. However, thenon-significant interaction term in that linear model indicatesthat the use effects of CPAP and placebo do not differ in a waythat is consistent with a linear-dose response. In sensitivityanalyses we dichotomised hours-of-use data at different cut-offpoints (5, 6.5 and 7.5 h) and additionally visually plotted theraw data (see online supplemental figure). These analyses raisedquestions about the linearity of the relationship betweencompliance and symptomatic improvement. Thus we can alsoconfirm previous reports that after controlling for placeboeffects, the association between compliance and better patientoutcome is weaker than expected5 and may not be linear6 asbenefit ceases to accrue at an asymptote of about 5.5e6.5 h pernight.

Figure 1 shows patients using the real or placebo device forsimilar lengths of time. Whether this ‘high-use’ trait mightgeneralise between CPAP and pharmacological compliance is thesubject of conflicting reports.18e20 Additionally, the possibilityof modifying psychological characteristics of non-compliancewith CPAP treatment should continue to be the subjectof clinical trials (eg, Richards et al21 and ACTRN:12606000065594). The characterisation of patients into ‘high’and ‘low’ users is defensible because CPAP compliance is oftenbimodally distributed (see figure 1 and the relative paucity ofdata points centred around 4 h compared with 0e2 hours and6 h). However this classification may still be flawed as recentinvestigations of both pharmacological treatment of epilepsyand CPAP indicate there may be more than two compliancephenotypes.22 23

The use of crossover trials with placebo control offers onlya partial solution to the problem that patients are aware of(unblinded) their ‘dose’ of treatment. A trial in which patientsare randomised to 2, 4, 6 or 8 h/night of effective treatmentmight offer superior data to the data presented here. Ourapproach of using each patient with their highly correlated9 useof CPAP and placebo to act as their own control offers anadvance on previous analyses. However, such highly selected andmotivated patients may not be generalisable to normal clinicalpopulations. Data from clinical trials and unselected cohortsprovide useful clinically applicable information when usedtogether. However, we were limited by the availability of onlyone subjective outcome variable and by the small numbers ofpatients (n¼91) participating in these technically challengingclinical trials. Two of the trials did not have second-arm baselinemeasurements10 11 and one did not have a washout period,11

which might have affected the results. The Australian-basedstudy10 also had lower baseline ESS scores than would beexpected for a symptomatic clinical sample and this may havemade it more difficult to detect treatment and use differencesbecause of a possible floor effect. One of the implications of thiseffect is that in patients in clinical practice who have very highESS scores, the doseeeffect relationship may be stronger thanreported here and the doseebenefit relationship may extend wellbeyond the asymptote reported here. More accurate treatmentestimates may have been possible if we had measured percentageof sleep with CPAP rather than the crude measure of hours pernight, however we lacked a good objective measure of sleep

duration. Nevertheless, these data provide an opportunity forunderstanding aspects of CPAP compliance that have notpreviously been investigated.Longstanding efforts to improve compliance to CPAP by

employing technological solutions aimed at reducing pressuremay be questionable as figure 1 also suggests compliance haslittle relationship to pressure. This also explains why the twomost recent meta-analyses of pressure modification approachesto improving CPAP compliance show little or no compliancebenefit and no symptomatic benefit.24 25 In this context it is notsurprising that the largest effect of any treatment intended toincrease use of CPAP therapy tested in a randomised trial hasbeen cognitive behavioural therapy.21

SUMMARYThis patient-level meta-analysis of three randomised placebo-CPAP-controlled crossover trials for patients with OSA confirmsthat high use of CPAP provides greater sleepiness reductionbenefits than high use of a placebo device. Interventions thatimprove CPAP use from low levels (ie, less than 4 h/night) arelikely to result in real benefits for patients.

Contributors MRC: conception and design, acquisition of data, analysis andinterpretation of data and drafting and final approval of article. DJB, JPHW and PC:acquisition of data, revision and final approval of article. SRC, CLP, AN and RRG:acquisition of data, revision of article, final approval of article. CAE and GD: revisionand final approval of article. NSM: conception and design, acquisition of data, analysisand interpretation of data, drafting and revising article and final approval of article.

Funding The Health Research Council of New Zealand (grant 00/285); The NationalHealth and Medical Research Council of Australia (grants 202916; 301936; 457355;571179; 571421 and 512498); British Heart Foundation (grant 2001).



REFERENCES1. Choudhry NK, Fischer MA, Avorn J, et al. The implications of therapeutic complexity

on adherence to cardiovascular medications. Arch Intern Med 2011;171:814e22.2. Osterberg L, Blaschke T. Adherence to medication. N Eng J Med

2005;353:487e97.3. Finniss DG, Kaptchuk TJ, Miller F, et al. Biological, clinical, and ethical advances of

placebo effects. Lancet 2010;375:686e95.4. Kirsch I. Response expectancy as a determinant of experience and behavior. Am

Psychol 1985;40:1189e202.5. Kingshott RN, Vennelle M, Hoy CJ, et al. Predictors of improvements in daytime

function outcomes with CPAP therapy. Am J Respir Crit Care Med2000;161:866e71.

6. Weaver TE, Maislin G, Dinges DF, et al. Relationship between hours of CPAP useand achieving normal levels of sleepiness and daily functioning. Sleep2007;30:711e19.

7. Antic NA, Catcheside P, Buchan C, et al. The effect of CPAP in normalizing daytimesleepiness, quality of life, and neurocognitive function in patients with moderate tosevere OSA. Sleep 2011;34:111e19.

8. Stepnowsky CJ, Dimsdale JE. Doseeresponse relationship between CPAPcompliance and measures of sleep apnea severity. Sleep Med 2002;3:329e34.

9. Marshall NS, Neill AM, Campbell AJ, et al. Randomised controlled crossover trial ofhumidified continuous positive airway pressure in mild obstructive sleep apnoea.Thorax 2005;60:427e32.

10. Phillips CL, Yee BJ, Marshall NS, et al. Continuous positive airway pressure reducespost prandial lipidemia in obstructive sleep apnea. A randomised, placebo controlledcrossover trial. Am J Respir Crit Care Med 2011;184:355e61.

11. Coughlin SR, Mawdsley L, Mugarza JA, et al. Cardiovascular and metabolic effectsof CPAP in obese males with OSA. Eur Respir J 2007;29:720e7.

12. Johns MW. A new method for measuring daytime sleepinessdthe EpworthSleepiness Scale. Sleep 1991;14:540e5.

13. Kazis L, Anderson J, Meenan R. Effect sizes for interpreting changes in healthstatus. Med Care 1989;27:S178e89.

14. Engleman HM, Kingshott RN, Wraith PK, et al. Randomized placebo-controlledcrossover trial of continuous positive airway pressure for mild sleep apnea/hypopneasyndrome. Am J Respir Crit Care Med 1999;159:461e7.

15. Stradling JR, Davies RJO. Is more NCPAP better? Sleep 2000;23:S150e3.16. Weaver TE, Grunstein RR. Adherence to continuous positive airway pressure

therapy: the challenge to effective treatment. Proc Am Thorac Soc 2008;5:173e8.

Sleep


17. Karlawish JH, Pack AI. Addressing the ethical problems of randomized andplacebo-controlled trials of CPAP. Am J Respir Crit Care Med 2001;163:809e10.

18. Platt AB, Kuna ST. To adhere or not to adheredpatients selectively decide. Sleep2009;32:583e4.

19. Platt AB, Kuna ST, Field SH, et al. Adherence to sleep apnea therapy and use oflipid-lowering drugs. Chest 2010;137:102e8.

20. Villar I, Izuel M, Carrizo S, et al. Medication adherence and persistence in severeobstructive sleep apnea. Sleep 2009;32:623e8.

21. Richards D, Bartlett DJ, Wong K, et al. Increased adherence to CPAP with a groupcognitive behavioral treatment intervention: a randomized trial. Sleep2007;30:635e40.

22. Modi AC, Rausch JR, Glauser TA. Patterns of nonadherence to antiepileptic drugtherapy in children with newly diagnosed epilepsy. JAMA 2011;305:1669e76.

23. Aloia MS, Goodwin MS, Velicer WF, et al. Time series analysis of treatmentadherence patterns in individuals with obstructive sleep apnea. Ann Behav Med2008;36:44e53.

24. Smith I, Lasserson TJ. Pressure modification for improving usage of continuouspositive airway pressure machines in adults with obstructive sleep apnoea. CochraneDatabase Syst Rev 2009;(4):CD003531.

25. Bakker JP, Marshall NS. Flexible pressure delivery modification of continuouspositive airway pressure for obstructive sleep apnea does not improve compliancewith therapy: systematic review and meta-analysis. Chest 2011;139:1322e30.


Sleep

924

Journal club

Salbutamol infusion worsens outcomes in ARDSSalbutamol infusion has previously been shown to significantly reduce extravascular lungwater in mechanically ventilated patients with acute respiratory distress syndrome (ARDS).This double-blind randomised placebo-controlled study investigated the effects of earlysalbutamol infusion on the clinical outcome of patients with ARDS. The authors recruiteda total of 326 patients from 46 UK intensive care units and randomly assigned them to receiveeither intravenous salbutamol or placebo for 7 days. Exclusion criteria included patientsrequiring continuous or regular aerosolized b2 agonists and those receiving b-adrenergicantagonists. All other treatment measures were carried out according to local practice. Thetrial was terminated early due to safety concerns.When compared with those in the placebo arm, patients receiving continuous salbutamol

infusion demonstrated significantly higher mortality rates at 28 days (34% vs 23%, RR 1.47,95% CI 1.03 to 2.08). There was also a trend towards increased adverse rates necessitatingdrug withdrawal, including tachycardia (14% vs 1%), arrhythmias (9% vs 2%) and lacticacidosis (6% vs <1%). This effect maintained statistical significance even after adjustment forage, cause, PaO2/FiO2 ratio and sex. Secondary outcome measures, including organ failure-freedays and ventilator-free days, were also adversely affected by salbutamol infusion.In conclusion, the results suggest that salbutamol infusion in patients with ARDS leads to

detrimental outcomes. The drug was poorly tolerated by patients, resulting in an increasedincidence of drug side effects, serious adverse effects or death. Routine use of b2 agonisttherapy in ARDS cannot be recommended.

< Gao Smith F, Perkins GD, Gates S, et al. Effect of intravenous b2 agonist therapy on clinical outcomes in acute respiratorydistress syndrome (BALTI-2): a multicentre, randomised controlled trial. Lancet 2012;379:229e35.

Osita Okafor

Correspondence to Dr Osita Okafor, CMT 1, Queen Elizabeth Hospital Birmingham, Mindelsohn Way, Edgbaston,Birmingham B15 2WB, UK; [email protected]





CORRESPONDENCE

Necrotising pneumonia,pneumatoceles and thepneumococcusWe commend Principi and Esposito for theirtimely review of severe community-acquiredpneumonia in children.1 They state that theleading cause of necrotising pneumonia andpneumatoceles within the context of com-munity acquired pneumonia isStaphylococcus aureus. We must disagree withthis claim.Historically, S aureus has been the princi-

pal organism associated with necrotic orcavitary pneumonia in children, but recentdata demonstrate Streptococcus pneumoniae isnow a much more important cause.2–4

Pneumococcal infection was the leadingcause of necrotic or cavitary pneumonia intwo recent studies by Sawicki et al in Bostonand Mckee et al in London, accounting for89% (28/32) of cases in the latter;2 3 thisfigure will underestimate the true incidenceof pneumococcal infection as over 90% ofinvasive pneumococcal disease is culturenegative, and culture negative techniques fordetermining pneumococcal infection andserotype are still not widely available.The occurrence of severe pneumococcal

necrotising pneumonia in children was firstreported by Kerem et al in 1994.5 Thisproblem has become progressively morecommon globally over the last 2 decades.2 3

These changes are linked to increases ininfection with certain pneumococcal sero-types, specifically serotypes 1, 3, 14 and19A.3 4 6 Hsieh et al have demonstrated thatthe lung necrosis and subsequent cavitarychanges are microangiopathic resulting fromthrombosis of intrapulmonary blood vesselsand subsequent pulmonary gangrene.6 It istempting to speculate that earlier recogni-tion of these haematological changes couldoffer the opportunity for therapeutic inter-vention to minimise lung damage.Worryingly, the rise in 19A infection in

the USA has been associated with thespread of several multi-antibiotic resistantclones. We have also recently observed anincrease in severe 19A disease in the UKwithin the UK paediatric empyema surveil-lance programme, but we have not foundany evidence of antibiotic resistance.Staphylococcal disease due to methicillin-

resistant strains and Panton–Valentine leu-kocidin related disease remain important,but most reports are of sporadic outbreaksand the overall incidence of staphylococcalpneumonia in children in developed coun-tries is now very low. These findings havesignificant implications for patient manage-ment. For many years, our first choice anti-biotic has been Clindamycin which hasexcellent antipneumococcal activity andantistaphylococcal activity. When managingpatients with proven staphylococcal pneu-monia, it is also now our practice to

exclude important underlying immuno-logical causes. Paediatricians need to be everaware of the rapidly changing epidemiologyof pneumococcal disease and that the con-sequences of pneumococcal infection can beextremely serious.

Matthew F Thomas,1,2 David A Spencer1

1School of Biology, Newcastle University, Newcastle-Upon-Tyne, UK; 2Department of Respiratory Paediatrics,Great North Children’s Hospital, Newcastle-Upon-Tyne,UK

Correspondence to Dr David Spencer, Department ofRespiratory Paediatrics, Great North Children’s Hospital,Newcastle-Upon-Tyne NE1 4LP, UK; [email protected]

Contributors MFT and DAS contributed equally to thedesign and writing of the manuscript.


Provenance and peer review Not commissioned;internally peer reviewed.

Accepted 23 January 2012Published Online First 16 February 2012

Thorax 2012;67:925.doi:10.1136/thoraxjnl-2011-201308

REFERENCES1. Principi N, Esposito S. Management of severe

community acquired pneumonia of children indeveloping and developed countries. Thorax2011;66:815–22.

2. Sawicki GS, Lu FL, Valim C, et al. Necrotisingpneumonia is an increasingly detected complication ofpneumonia in children. Eur Respir J2008;31:1285–91.

3. McKee AJ, Ives A, Balfour-Lynn IM. Increasedincidence of bronchopulmonary fistulas complicatingpediatric pneumonia. Pediatr Pulmonol2011;46:717–21.

4. Bender JM, Ampofo K, Korgenski K, et al.Pneumococcal necrotizing pneumonia in Utah: doesserotype matter? Clin Infect Dis 2008;46:1346–52.

5. Kerem E, Bar Ziv Y, Rudenski B, et al. Bacteremicnecrotizing pneumococcal pneumonia in children. AmJ Respir Crit Care Med 1994;149:242–4.

6. Hsieh YC, Hsiao CH, Tsao PN, et al. Necrotizingpneumococcal pneumonia in children: the role ofpulmonary gangrene. Pediatr Pulmonol2006;41:623–9.

Authors’ responseIn their letter, Thomas and Spencer1 claimthat our assertion that Staphylococcus aureusis the most important cause of necrotisingpneumonia2 is wrong on the basis of themost recent evidence and that Streptococcuspneumoniae is currently the major cause.1

However, although the studies cited bythem document an increase in the incidenceof necrotising pneumococcal pneumonia,1

we still believe that S aureus is themost important pathogen clinically andtherapeutically.It has recently been found that the inci-

dence of complicated pneumonia in childrenis increasing and that there is a concurrentincrease in the incidence of community-associated methicillin-resistant S aureus(CA-MRSA) infections.3–5 The emergence

of CA-MRSA was initially reported in theUSA and in subjects with skin infections,but paediatric necrotising pneumonia dueto S aureus has been reported in healthysubjects of different ages and patients withunderlying diseases such as cystic fibrosis.4 5

A high level of suspicion is required inpatients with severe community-acquiredpneumonia, and CA-MRSA needs to be con-sidered early in its differential diagnosisbecause it can lead to rapid deteriorationand death unless it is appropriately andimmediately treated.Moreover, the specific antimicrobial

treatment of CA-MRSA is different fromthat using the traditional antimicrobialagents currently prescribed for community-acquired pneumonia. Antimicrobial agentsthat specifically inhibit exotoxin produc-tion, such as clindamycin and linezolid,should be preferred.6

Finally, no prophylaxis against CA-MRSAis yet available. On the contrary, necrotisingpneumonia due to S pneumoniae is mainlydue to serotype 3 and, in a minority ofcases, serotypes 1 and 19A.2 All of these areincluded in the 13-valent pneumococcalconjugate vaccine that has been recentlylicensed and included in the immunisationschedules approved for children in mostindustrialised countries.7 This means thatthe incidence of infections due to S pneumo-niae may be significantly reduced in thefuture, whereas those associated withCA-MRSA can only remain stable or pro-portionally increase.

Nicola Principi, Susanna Esposito

Department of Maternal and Pediatric Sciences,Università degli Studi di Milano, Fondazione IRCCS Ca’Granda Ospedale Maggiore Policlinico, Milan, Italy

Correspondence to Professor Nicola Principi,Department of Maternal and Pediatric Sciences,Università degli Studi di Milano, Fondazione IRCCS“Ospedale Maggiore Policlinico, Mangiagalli e ReginaElena”, Via Commenda 9, 20122 Milano, Italy; [email protected]

Contributors NP and SE wrote the paper together aswell as this response letter.






REFERENCES1. Thomas MF, Spencer DA. Necrotising pneumonia,

pneumatoceles and the pneumococcus. Thorax2012;67:925.

2. Principi N, Esposito S. Management of severecommunity-acquired pneumonia of children indeveloping and developed countries. Thorax2011;66:815–22.

3. Schwartz KL, Nourse C. Panton-Valentineleukocidin-associated Staphylococcus aureusnecrotizing pneumonia in infants: a report of four


PostScript

cases and review of the literature. Eur J Pediatr2012;171:711–7.

4. Hidron AI, Low CE, Honig EG, et al. Emergence ofcommunity-acquired methicillin-resistantStaphylococcus aureus strain USA300 as a cause ofnecrotising community-onset pneumonia. Lancet InfectDis 2009;9:384–92.

5. Carrillo-Marquez MA, Hulten KG, Hammerman W,et al. Staphylococcus aureus pneumonia in children inthe era of community-acquired methicillin-resistanceat Texas Children’s Hospital. Pediatr Infect Dis J2011;30:545–50.

6. Lin YC, Peterson ML. New insights into theprevention of staphylococcal infections and toxicshock syndrome. Expert Rev Clin Pharmacol2010;3:753–67.

7. Paradiso PR. Advances in pneumococcal diseaseprevention: 13-valent pneumococcal conjugate vaccinefor infants and children. Clin Infect Dis2011;52:1241–7.

Exposing children to secondhandsmokeIn the October issue of Thorax, Leonardi-Beeet al reviewed the effects of parentalsmoking on the uptake of smoking by chil-dren.1 They conclude that exposure tosmoking within the family is a significantdeterminant of subsequent smoking.One area that this study did not address

was the effect of the site of the smoking onexposure.We have studied urinary cotinine levels in

children attending an asthma clinic and cor-related them with the self-reported smokinghabits of their carers. Thirty-six girls and 64boys aged between 30 and 164 monthsattending our asthma clinic were recruitedif at least one carer smoked. The carerscompleted a questionnaire about theirsmoking habits and the child’s urine wastested for cotinine using NicAlert, a com-mercially available reagent strip. Levels of0–10 ng/ml correspond to minimal or nonicotine exposure, 10–100 ng/ml to moder-ate to high SHS exposure and >100 ng/mlto active smoking. Ten further children wererecruited from the same clinic from non-smoking families and acted as controls.Seventy-two ‘smoking’ families (72%)

reported only smoking outside the house(‘outside’ smokers) while 28 parents (28%)did not take any harm reduction strategies(‘everywhere’ smokers). Three (4%) of thechildren in outside smoking carers’ group hadlevels indicative of active smoking and wereexcluded from analysis. Urine testing showedmeasureable cotinine (10–100 ng/ml) in 27 of28 (96%) and 61of 69 (88%) everywhere andoutside smokers’ children, respectively(p=0.20). Only one child in the controlgroup (10%) had detectable cotinine(p<0.0001, compared with both everywhereand outside smokers’ children) and it tran-spired that she was exposed to SHS at school.Our findings suggest that there is little or

no difference between nicotine exposure inthose children of outside smokers and those

in direct contact with smokers (everywheresmokers). Even if a proportion of ouroutside smokers were in fact smokingindoors, the very high percentage of thisgroup that had raised cotinine levels indi-cates that most of the ‘outdoor ’ smokers areindeed exposing their children to SHS. Thisis consistent with reports previously pub-lished, showing that nicotine can be trans-mitted through a number of third handroutes.2

Harm reduction strategies (effectivelylimiting the exposure) are reported to havesome benefit in reducing nicotine exposurealthough our data suggest that reductionmay have been overestimated in the past.3

Using urinary cotinine levels may also beuseful in helping advice on smoking cessa-tion.4 In our study we were able to use theresults which were shown to parents at thetime of the clinic, and they did seem tofocus parental attention on the problem.We found that a number of parents reportedcessation of smoking following the test andthat their children were free of urinary coti-nine when tested.We suggest that measurement of urinary

cotinine can be a useful adjunct when dis-cussing parental smoking and that outsidesmoking may not protect children fromSHS exposure as much as previouslythought.

Jenny Pool,1 Natalya Petrova,2 RobertRoss Russell1

1Department of Paediatrics, Cambridge UniversityHospitals NHS Foundation Trust, Cambridge, UK;2Department of Neonatal, Almazov Federal Heart, Bloodand Endocrinology Centre, St Petersburg, RussianFederation

Correspondence to Jenny Pool, Department ofPaediatrics, Cambridge University Hospitals NHSFoundation Trust, Cambridge, UK; [email protected]

Contributors JP and RIRR developed the concept forthe study, recruitment and data collection. The letterwas composed by all three authors.


Ethics approval Ethics approval was provided byCambridge Local Research Ethics Committee.

Provenance and peer review Not commissioned;internally peer reviewed.

Data sharing statement All data from this study areavailable to anyone interested. Please contact theauthors for further details.



REFERENCES1. Leonardi-Bee J, Jere ML, Britton J. Exposure to

parental and sibling smoking and the risk of smokinguptake in childhood and adolescence: a systematicreview and meta-analysis. Thorax 2011;66:847–55.

2. Matt GE, Quintana PJ, Hovell MF, et al. Householdscontaminated by environmental tobacco smoke: sourcesof infant exposures. Tob Control 2004;13:29–37.

3. Spencer N, Blackburn C, Bonas S, et al. Parentreported home smoking bans and toddler (18-30month) smoke exposure: a cross-sectional survey.Arch Dis Child 2005;90:670–4.

4. Irvine L, Crombie IK, Clark RA, et al. Whatdetermines levels of passive smoking in children withasthma? Thorax 1997;52:766–9.

Authors’ response: exposingchildren to secondhand smokePool and colleagues make an importantpoint about exposure to tobacco smoke1

but we are uncertain of the relevance to ourpaper,2 which reported the effects ofsmoking by parents as a behaviour ratherthan source of passive exposure. While wecannot discount the possibility that chil-dren of smokers are more likely to becomesmokers as a consequence of passive smokeexposure, our interpretation is that it is thebehaviour that is responsible.

John Britton, Jo Leonardi-Bee

UK Centre for Tobacco Control Studies, University ofNottingham, Nottingham, UK

Correspondence to Dr Jo Leonardi-Bee, UK Centre forTobacco Control Studies, University of Nottingham,Nottingham NG5 1PB, UK; [email protected]


Provenance and peer review Commissioned;externally peer reviewed.

Accepted 30 January 2012Published Online First 1 March 2012


REFERENCES1. Pool J, Petrova N, Russell RR. Exposing children to

second hand smoke. Thorax 2012;67:926.2. Leonardi-Bee J, Jere ML, Britton J. Exposure to

parental and sibling smoking and the risk of smokinguptake in childhood and adolescence: a systematicreview and meta-analysis. Thorax 2011;66:847–55.

Inequalities in outcomes for non-small cell lung cancer: the roleof the MDTRich et al1 report that non-small cell lungcancer patients first seen in a hospital whichhas on-site thoracic surgical services aremore likely to have surgical treatment oftheir tumour. However, it is not clear whataspects of ‘being a surgical centre’ are crucialto increasing resection rates. Numerousreports have documented a volume–outcome relationship for complex surgicaland medical care and one hypothesis toexplain this relationship is higher case loadsince surgical units tend to be located inlarge hospitals serving large populations.To assess this, we analysed all 129 052

cases submitted to the NLCA (England


PostScript

only) between 2006 and 2010 inclusive,based on the ‘place first seen’ as a surrogatemarker of the decision-making multidiscip-linary team (MDT). After excluding caseswith ‘negative survival’, no recorded placefirst seen and trusts with <100 cases overthe 5-year time span, we divided the MDTsinto quintiles based on the number of casesmanaged by each MDT (quintile 1 thesmallest through to quintile 5 the largest).We used logistic regression to calculate theodds ratio (OR) for receipt of surgery, andalso calculated survival by the Kaplan-Meiermethod.In a multivariate model (adjusted for age,

sex, performance status and stage, but notcomorbidity), the ORs and CIs for surgicaltreatment across the quintiles were 1.0, 1.11(0.99 to 1.25), 1.03 (0.93 to 1.15), 0.94 (0.85to 1.04) and 0.91 (0.82 to 1.0). Survival ana-lyses are shown in table 1.These results do not support the hypoth-

esis that MDTs managing more patientsdeliver better outcomes. It is likely that thebest outcomes are delivered by functionalMDTs having efficient pathways deliveringbest practice diagnostic and treatment regi-mens. Further work is necessary to under-stand the variables which influencevariation in treatment and outcomes forlung cancer patients in England.

Paul Beckett,1 Ian Woolhouse2

1Burton Hospitals NHS Foundation Trust, Burton-on-Trent, England, UK; 2Department of Respiratory

Table 1 Survival data by MDT quintile

QuintileMediansurvival (days)

1 Year(%)

2 Years(%)

3 Years(%)

4 Years(%)

5 Years(%)

1 (SmallestMDTs)

226 37.2 22.6 17.2 14.5 12.9

2 216 36.3 21.4 16.2 13.7 12.23 211 35.8 21.2 16.1 13.6 11.84 204 34.9 20.4 15.3 12.6 10.95 (LargestMDTs)

209 35.3 20.8 15.6 13 11.6

MDTs, multidisciplinary teams.

Authors’ responseBeckett and Woolhouse have investigatedthe inequality in access to surgery for indi-viduals with non-small cell lung cancer(NSCLC) in England.1 2 They have used thenumber of patients entered into theNational Lung Cancer Audit (NLCA) as asurrogate marker for the ‘specialist’ qualitiesof the multi-disciplinary team (MDT) andconcluded that this marker does not influ-ence surgical resection rate or survival.

An MDTwith a high number of patientsmay reflect a high level of experience, but thismay not be the case for every MDT. It wouldbe helpful to see the actual numbers ofpatients per MDT in each quintile and theraw data for survival. Only the largest ofMDTs may link to specialist qualities. MDTsreviewing a large number of patients, butoutside the top quintile, may reflect those thatare under-resourced and struggling to cope.Other factors that may influence the out-

comes of individuals with lung cancer inEngland may be: access to positron emissiontomography scanning, cardiovascular assess-ment, lung function testing and on-siteintensive treatment unit beds. Furtherresearch looking at the composition of theMDT, and also the exact nature of investiga-tive and ‘support’ facilities available atevery NHS Trust is needed.

Anna Rich,1 David Baldwin,1 Richard Hubbard2

1Department of Respiratory Medicine, NottinghamUniversity Hospitals, David Evans building, City campus,Nottingham, UK; 2BRU (Biomedical Research Unit),Department of Epidemiology and Public Health,University of Nottingham, Nottingham, UK

Correspondence to Dr Anna Rich, Department ofRespiratory Medicine, Nottingham University Hospitals,David Evans building, City campus, NG5 1PB,Nottingham, UK; [email protected]



Accepted 20 January 2012Published Online First 14 March 2012


REFERENCES1. Beckett P, Woolhouse I. Inequalities in outcomes for

non-small cell lung cancer: the role of the MDT.Thorax 2012; doi:10.1136/thoraxjnl-2012-201582

2. Rich AL, Tata LJ, Free CM, et al. Inequalities inoutcomes for non-small cell lung cancer: the influenceof clinical characteristics and features of the local lungcancer service. Thorax 2011;66:1078–84.

Medicine, University Hospitals Birmingham, Birmingham,UK

Correspondence to Dr Paul Beckett, Burton HospitalsNHS Foundation Trust, Queens Hospital Belvedere Road,Burton-on-Trent, England DE13 0RB, UK; [email protected]


Provenance and peer review Not commissioned;externally peer reviewed.



REFERENCE1. Rich AL, Tata LJ, Free CM, et al. Inequalities in

outcomes for non-small cell lung cancer: the influenceof clinical characteristics and features of the local lungcancer service. Thorax 2011;66:1078–84.


PostScript




AUDIT, RESEARCH AND GUIDELINE UPDATE

British Thoracic Society national bronchiectasis audit2010 and 2011

Adam T Hill,1 Sally Welham,2 Kerry Reid,2 C E Bucknall,3 On behalf of the BritishThoracic Society

ABSTRACTThere have been two national British Thoracic Society(BTS) bronchiectasis audits from 1 October to 30November in 2010 and 2011 in patients with non-cysticfibrosis attending secondary care. The first audit wassoon after the publication of the BTS guidelines in July2010 and both audits were based on the BTS guidelinerecommendations. We had 1460 and 2404 records in the2 years respectively. The national audits highlight thatthe majority of guideline recommendations were notcurrently being adhered to and demonstrate the need fornational quality standards, which are currently inpreparation.

INTRODUCTIONThere have been two national British ThoracicSociety (BTS) bronchiectasis audits from 1 Octoberto 30 November in 2010 and 2011 in patients withbronchiectasis attending secondary care. The firstaudit was soon after the publication of the BTSbronchiectasis guideline in July 2010 and bothaudits were based on key recommendations fromthe BTS guideline for non-CF bronchiectasis.1

A total of 1460 and 2404 records were submittedfrom 59 and 93 institutions in the 2 years respec-tively. The audit results for 2010 and 2011 werevery similar.

PATIENT PROFILEIn keeping with international studies the majorityof patients were women (61e62%) and the meanage was (64e66 years). From the sputum microbi-ology in the last year, Pseudomonas aeruginosa wasisolated in 21% of patients in the 2010 audit. In the2011 audit P aeruginosa, methicillin-resistant Staph-ylococcus aureus (MRSA) or enteric gram-negativeorganisms (if isolated on two or more occasions, eg,Escherichia coli) were isolated in 26% of patients. Inthe group with Paeruginosa or enteric gram-negativeorganisms, 18% were resistant to ciprofloxacin and9% to gentamicin (see figure 1).The mean (SD) number of exacerbations was 2.6

(2.5) for both audits. This was lower than expectedin a cohort attending secondary care and with21e26% of patients infected with P aeruginosa,MRSA or enteric gram-negative organisms, groupsthat would be expected to have multiple exacer-bations per year. The accuracy is not known as datawere not crosschecked with primary or secondarycare records.

A total of 16e17% had received intravenousantibiotics in the past year, 27e33% used long-term(>28 days) oral antibiotics and 9e10% usednebulised antibiotics (76% nebulised colomycin,12e13% gentamicin and 5e6% tobramycin; 6%used other agents not specified, see figure 1).

Treatment snapshotAs part of the audits, we requested information oncurrent treatment patients were receiving forchronic management of their bronchiectasis. Weasked about use of inhaled corticosteroids, short-acting and long-acting bronchodilators, agentsthat improve mucociliary clearance and long-termantibiotics.A total of 78e81% were on inhaled corticoste-

roids with a mean dose of 1094e1252 mg/day. Thisis despite the guidelines not recommending thelong-term use of inhaled steroids unless there isclear clinical benefit or in those with coexistentasthma or COPD. The audit did not provide anydata on these.Regarding bronchodilators, 66e67% were on

a short-acting b2 agonist and 9e11% wereon a short-acting anticholinergic; 62e65% were ona long-acting b2 agonist and 29e30% were ona long-acting anticholinergic.For agents that improve mucociliary clearance

27e30% used carbocysteine, 6e8% used nebulisedsaline (of these, 37e38% used 0.9% saline and theremainder used higher concentrations varying from3% to 7%), 0e0.4% used inhaled mannitol and0e0.2% used nebulised DNAase.Regarding long-term antibiotics, 27e33% used

long-term (>28 days) oral antibiotics and 9e10%used nebulised antibiotics (76% nebulised colo-mycin, 12e13% gentamicin, 5e6% tobramycin; 6%used other agents not specified).These treatments may have all been influenced if

patients had coexistent asthma, COPD or othercomorbid illness and the audit has no informationabout this. In clinical practice, however, manyclinicians use a variety of treatments despite a lackof evidence base to support these. Trials to improvethe evidence base will guide clinicians on theoptimum therapy for their patients.

AUDIT OF RECOMMENDATIONSThere were seven recommendations drawn fromthe initial bronchiectasis guidelines 1 and the auditswere designed to assess whether there was adher-ence to these recommendations.

1Department of RespiratoryMedicine, Royal Infirmary andUniversity of Edinburgh,Edinburgh, UK2British Thoracic Society,London, UK3Department of RespiratoryMedicine, Stobhill Hospital,Glasgow, UK

Correspondence toDr Adam T Hill, RespiratoryMedicine, Royal Infirmary ofEdinburgh, 51 Little FranceCrescent, Edinburgh EH16 4SA,UK; [email protected]

Received 30 March 2012Accepted 15 June 2012

Chest clinic

Chest

clinic


Published Online First17 July 2012

Recommendation 1 is that 90% of patients diagnosed withbronchiectasis should have had the diagnosis confirmed witha chest CTscan. In the audits, 93%were diagnosedwith a CTscanof the chest, 1% by bronchogram, 2e3% had a clinical diagnosisonly and for 3e4% there were no data. The first recommendationwas met.

Recommendation 2 is that 90% of patients diagnosed withbronchiectasis should see a respiratory physiotherapist sopatients are taught chest clearance techniques. In the audits,65e69% said their patient had seen a respiratory physiothera-pist, 18e23% said they had not and in 12e14% there were nodata. The second recommendation was not met which is inkeeping with clinical experience that not all patients withbronchiectasis have seen a respiratory physiotherapist to betaught chest clearance techniques.

Recommendation 3 is that all patients being seen should havea record of cough, sputum purulence, estimated or measured24 h sputum volume and breathlessness when clinically stable.In the audits, 70e73% had a record of cough, 66e68% sputumcolour, 49e54% 24 h sputum volume and 55e62% breathless-ness. The third recommendation was not met. Most patientswith significant bronchiectasis have these symptoms. Esti-mating their severity will allow consistent assessment to helpthe ongoing management of such patients.

Recommendation 4 is that all patients diagnosed with bron-chiectasis should have immunoglobulins and protein electro-phoresis checked along with total IgE, IgE to Aspergillus or skinprick testing to Aspergillus, and for those aged <40 years, tests toexclude CF. In the audits, 73e77% of patients had their immu-noglobulins checked and 58e60% had serum sent for proteinelectrophoresis. Although protein electrophoresis is advocated inthe guidelines in addition to measurement of immunoglobulins,the additive value of protein electrophoresis needs to be exploredas the primary aim is to identify immunodeficiency. A total of62e75% had IgE measured and 53e54% had A fumigates radio-allergosorbent test or skin prick testing to Aspergillus. Thecombination of total IgE and IgE to Aspergillus or skin pricktesting to Aspergillus is advocated as the results vary dependingon the activity of Aspergillus hypersensitivity. For those aged

<40, 30e33% had CF gene analysis and 46% had a sweat testcarried out. The fourth recommendation was not met. These arerecommended investigations for secondary care as the long-termmanagement may differ if an immunodeficiency, allergic bron-chopulmonary aspergillosis or CF is identified.Recommendation 5 is that all children who are old enough

(usually age over 5 years) and adults should have measuresof forced expiratory volume in 1 s (FEV1), forced vital capacity(FVC) and peak expiratory flow (PEF) in primary and secondarycare. Repeat assessment of FEV1, FVC and PEF should be madeat least annually in patients attending secondary care. FEV1

and FVC should be measured before and after intravenous anti-biotic therapy as this may give objective evidence of improve-ment. Spirometry should be measured in all patients beforeand after commencing long-term oral or nebulised antibiotictherapy.In the audits, 30% had PEF measured and 55e60% had

spirometry measured on the day of consultation. For patientswho received intravenous antibiotics, 22% had spirometryassessed before and after a course of intravenous antibiotics, 56%did not have this assessment and for 22% there were no data. Ofthose receiving nebulised antibiotics, 61e63% had spirometrychecked at the start of treatment and later on during the courseof treatment, 13e20% did not have these checks and for17e26% there were no data. Of those taking nebulised antibi-otics, 82e83% had spirometry checked at least 6 monthly. Thefifth recommendation was not met. Spirometry may be useful tomonitor disease progression and response to treatments.Recommendation 6 is that 90% of patients with an exacer-

bation should have a sputum sample sent for bacteriologyculture prior to empirical antibiotic treatment. In the audits,55e57% had a sputum sample sent for bacteriology culture,33%e42% did not and for 3e10% there were no data. The sixthrecommendation was not met. Monitoring sputum microbi-ology is key to providing appropriate antimicrobial prescribing.Recommendation 7 is that pulmonary rehabilitation should

be offered to individuals who have Medical Research Councilgrade 3 or worse breathlessness affecting their activities of dailyliving. In the audits, 12e15% had been referred, 13e17% had notbeen referred, 5% were unable to participate in pulmonaryrehabilitation, 52e54% did not meet the criteria for pulmonaryrehabilitation and for 13e16% there were no data. The seventhrecommendation was not met. Pulmonary rehabilitation in suchpatients has the potential to improve their exercise capacity andgeneral wellbeing.

CONCLUSIONSThis national audit over a 2-year period has provided a snapshotof how bronchiectasis is managed in secondary care. Themajority of recommendations were not met. The reasons fora lack of adherence to the guidelines are not clear and are likelyto be multifactorial: lack of knowledge of current guidelines;lack of evidence base to support recommendations; patientswith coexisting asthma, COPD or other comorbidities; patientsestablished on therapy; patient and healthcare professionalissues; resource issues in primary and secondary care; accessi-bility to specialist healthcare professionals. Issues raised fromthe national audits along with new evidence will be used toformulate future guidelines.In addition, quality standards for bronchiectasis are to be

published by the BTS in 2012, with the aim of improving care forpatients with bronchiectasis. The BTS has also produceda template self-management plan for patients with bronchiectasis

Figure 1 Snapshot of microbiology and antibiotic treatments given.In 2010, 21% had Pseudomonas aeruginosa (PA) but do not havethe combined data for methicillin-resistant Staphylococcus aureus(MRSA) and enteric gram-negative organisms. IV, intravenous;LT, long term.

Chest clinic

Chest

clinic


which may be adapted for local use (available at http://www.brit-thoracic.org.uk/guidelines/bronchiectasis-guideline-(non-cf).aspx). Results from future cycles of the BTS bronchiectasis auditwill allow the impact of the introduction of quality standards tobe monitored.



REFERENCE1. Pasteur MC, Bilton D, Hill AT; British Thoracic Society Non-CF Bronchiectasis

Guideline Group. British Thoracic Society guideline for non-CF bronchiectasis. Thorax2010;65 Suppl 1:i1e58.


Chest clinic

Chest

clinic

930

Journal club

High doses of vitamin D may reduce exacerbations ofchronic obstructive pulmonary diseaseThere has been much research providing inconsistent evidence for a correlation between lowlevels of vitamin D and chronic conditions, such as coronary artery disease, multiple sclerosisand diabetes. It is known that vitamin D plays an integral role in the cathelicidinantimicrobial peptide, which serves a critical function in mammalian innate immune defenceagainst invasive bacterial infection. As such, it has been hypothesised that low levels ofvitamin D may be important in chronic obstructive pulmonary disease (COPD), in whichthere is an abnormal inflammatory reaction to inhaled particles and a reduced immuneresponse.In this study, the authors recruited 182 patients into a double-blind, randomised, placebo-

controlled trial exploring vitamin D supplementation and COPD exacerbations. One hundredand fifty patients completed the study. Patients recruited were either current or formersmokers over the age of 50 years, with a diagnosis of moderate to severe COPD. They wererandomised to receive 100 000 IU of vitamin D or placebo, and were monitored for a year. Theprimary outcome was time to first exacerbation. While mean vitamin D levels weresignificantly higher in the treatment group there was no statistically significant difference formedian time to first exacerbation. Secondary outcomes, including exacerbation rate andquality of life were also not significantly different between groups.However, there was a significant reduction in exacerbations in the vitamin D group in those

patients with a severe vitamin D deficiency at baseline, suggesting that supplementation maybe useful in a subset of patients.

< Lehouck A, Mathieu C, Carremans C, et al. High doses of vitamin D to reduce exacerbations in chronic obstructive pulmonarydisease. Ann Intern Med 2012;156:105e14.

Emily Heiden

Correspondence to Dr Emily Heiden, CT2, Maidstone and Tunbridge Wells NHS Trust, Tunbridge Wells Hospital, TonbridgeRoad, Pembury, Tunbridge Wells, Kent TN2 4QJ, UK; [email protected]




CASE BASED DISCUSSION

A therapeutic conundrum: recurrentcystic-fibrosis-related haemoptysis complicated byacute pulmonary embolism

William G Flight,1,2 Rowland J Bright-Thomas,1,2 Stephen Butterfield,3

Andrew M Jones,1,2 A Kevin Webb1,2

ABSTRACTThe authors present the case of an older patient withcystic fibrosis (CF) and recurrent haemoptysiscomplicated by acute pulmonary embolism. The patientwas treated successfully with a combination ofanticoagulation and bronchial artery embolisation. Themanagement of CF-related haemoptysis, the impact ofan ageing CF population and the incidence ofthromboembolic disease in CF are discussed.

CASE REPORTDr Flight (WGF): A 72-year-old man with cysticfibrosis (CF) was admitted to hospital with recur-rent haemoptysis. He was a compound heterozy-gote for the F508del and R117H-7T mutations ofthe cystic fibrosis transmembrane conductanceregulator (CFTR) gene. CF had been diagnosed atthe age of 65 prior to which he had been labelled ashaving asthma and non-CF bronchiectasis. Baselineforced expiratory volume in 1 s was 36% predictedand he was chronically infected with Pseudomonasaeruginosa and Burkholderia multivorans. His regulartherapy included long-term oral corticosteroids forallergic bronchopulmonary aspergillosis and nebul-ised colistin. In addition to multiple complications ofCF he also had a history of cataracts, age-relatedmacular degeneration, hypertension, peptic ulcerdisease, colonic polyps, benign prostatic hypertrophyand basal cell carcinomas of the skin.On the day of admission he had expectorated

approximately a third of a cup of bright red blood.Intravenous tranexamic acid was administered ata dose of 1 g four times daily alongside intravenoustemocillin, tobramycin and co-trimoxazole. Thehaemoptysis gradually resolved and he wasdischarged 2 weeks later to continue oraltranexamic acid 1 g twice daily.Professor Webb (AKW): First, this case illustrates

an increasingly important phenomenon in the careof patients with CF: the ageing patient. Thesurvival of people with CF has steadily increasedover recent decades1 and patients are frequentlyliving into their sixth decade and beyond. Much ofthis improved life expectancy is due to multidisci-plinary specialist care but a proportion is accountedfor by greater numbers of late diagnoses afforded byidentification of rarer cystic fibrosis transmembraneconductance regulator mutations. Compoundheterozygosity for the F508del and R117H-7T

mutations has a recognised association withdelayed diagnosis of CF.2 As patients with CF startto live into old age, CF physicians will increasinglyneed to be alert to diseases of ageing neverpreviously considered relevant to CF.Haemoptysis is a common complication of CF

with a 5-year incidence of 9.1% in one study.3 Thepathophysiology of haemoptysis in CF is poorlyunderstood but characteristically involves thedevelopment of tortuous, dilated vessels inthe bronchial circulation. Treatment has been thesubject of consensus guidance4 and centres onappropriate antibiotic therapy with bronchialartery embolisation (BAE) recommended as thetreatment of choice for massive haemoptysis.Tranexamic acid is an anti-fibrinolytic drug that iseffective in reducing bleeding related to post-partum haemorrhage and several surgical proce-dures. Although tranexamic acid is commonly usedin CF-related haemoptysis there is no robustevidence to support its efficacy or safety in thissetting. Concern has previously been raised over anincreased risk of venous thromboembolism withtranexamic acid but this has not been resolvedconclusively as yet.WGF: Following discharge, the patient continued

to experience streaks of blood in his sputum ona daily basis. He was re-admitted 2 weeks later withhaemoptysis and symptoms of a viral upper respi-ratory tract infection. On the day of admission hehad expectorated half a cupful of fresh red blood. Onarrival to hospital he was comfortable at rest withnormal vital signs. Oxygen saturations were 97% onair. An ECG revealed normal sinus rhythm anda chest radiograph was unchanged from previously.Two hours after presentation, he suddenly

developed breathlessness and chest tightness.Examination revealed him to be tachypnoeic,tachycardic and hypotensive. Oxygen saturationshad dropped to 78% on air. A repeat ECG showednew right bundle branch block. An urgent CT-pulmonary angiogram demonstrated a largethrombus involving the left main, left upper andleft lower lobe pulmonary arteries (see figure 1A).Serum troponin I was elevated at 0.51 mg/litre.

AKW: The patient has developed two potentiallyfatal pulmonary emergencies: persistent majorhaemoptysis and concurrent acute pulmonaryembolism (PE). This scenario is a genuine thera-peutic conundrum. Anticoagulation is required toprevent further thrombosis while, simultaneously,

1Manchester Adult CysticFibrosis Centre, Manchester, UK2Respiratory Research Group,University of Manchester,Manchester, UK3Department of Radiology,University Hospital of SouthManchester, Manchester, UK

Correspondence toDr William G Flight, ManchesterAdult Cystic Fibrosis Centre,University Hospital of SouthManchester NHS FoundationTrust, Southmoor Road,Manchester M23 9LT, UK;[email protected]

Received 17 April 2012Accepted 4 May 2012

Chest clinic

Chest

clinic



further bleeding must be avoided. When there is a risk ofsignificant haemorrhage, anticoagulation is best achieved withintravenous unfractionated heparin and close monitoring of theactivated partial thromboplastin time. In the event of life-threatening bleeding, the intravenous heparin can be quicklydiscontinued. If possible, the cause of the bleeding should beidentified and reversed. In the event of bleeding preventing safeanticoagulation, the insertion of an inferior vena cava filter isa further therapeutic option.

WGF: Intravenous heparin was commenced and the patientwas also treated with intravenous temocillin and tobramycin.High-flow oxygen was administered to achieve target saturationsof 94e98%. Fluid resuscitation was given and tranexamic acidwas discontinued. An inferior vena cava filter was considered butwas not pursued as the CT examination showed no evidence ofvenous thrombosis in the legs or pelvis. Despite the abovemeasures, the patient experienced ongoing substantial haemopt-ysis. Advice from the interventional radiology service was sought.

Dr Butterfield (Interventional Radiology): My first step was toreview the original CT pulmonary angiogram study. CT recon-structions were performed which revealed a dilated bronchialartery arising from the descending aorta (see figure 1B). Weelected to perform a bronchial artery angiogram with the aim ofembolising any potential sources of bleeding from the bronchialcirculation. The dilated bronchial artery identified on the CTscan was supra-selectively catheterised and embolised to stasiswith polyvinyl alcohol.

The available evidence suggests that BAE is an effectivetreatment for massive haemoptysis in CF. There have been norandomised trials of BAE in CF but large case series suggest thatacute bleeding is controlled in up to 97% of cases with a long-term recurrence rate of 27e46%.5 Chest pain is the mostcommon side effect following BAE, although potential compli-cations include an acute neurological deficit due to embolisationof a spinal artery or passage of embolic material throughcollateral vessels to the cerebral circulation. In the present casethe patient experienced mild chest pain for 24 h after theprocedure but did not suffer any more serious adverse events.

WGF: Following BAE the patient’s haemoptysis settledcompletely. Intravenous heparin was converted to low molecularweight heparin and oral warfarin was commenced 48 h afterundergoing BAE. He was discharged at day 16, by which timehis symptoms had returned to baseline and he had remained freefrom haemoptysis.

AKW: The development of PE in a patient with ongoing majorbleeding presents a highly challenging clinical situation thatmay be encountered in a wide variety of medical and surgical

specialties. The principles of management in this situation areapplicable to a number of conditions in respiratory medicine,such as the patient with lung cancer, aspergilloma or myriadother causes of haemoptysis.The scenario described is highly unusual in CF, however, and

to my knowledge has not been reported before. Idiopathic PEalone is considered exceptionally rare in patients with CF.Previous reports of PE in CF have all been associated with venousaccess device related thrombosis. In this case it is certainlypossible that the use of the tranexamic acid contributed to thedevelopment of venous thrombosis. The risks associated withsuch pro-thrombotic drugs remain uncertain and need to beaddressed in future studies.As the CF population ages it is reasonable to expect that the

incidence of venous thrombosis in this population will steadilyrise. Since haemoptysis is such a common feature of CF, it standsto reason that we shall also see concurrent haemoptysis andthromboembolic disease with increasing frequency.The key learning points to be taken from this case are:

1. The CF population is getting older and diseases associatedwith ageing are increasingly relevant to the CF physician.

2. Bronchial artery embolisation represents the therapy ofchoice for massive haemoptysis in CF.

3. CF physicians should be alert to the possibility of PE in theirpatients, a diagnosis which is easily overlooked in the contextof severe CF lung disease.

4. The conundrum of acute PE in the context of major bleedingrequires a delicate balance of anticoagulation and correctionof the underlying haemorrhage.

Contributors WGF wrote the first draft. All authors contributed to the finalmanuscript.




REFERENCES1. Dodge J, Lewis P, Stanton M, et al. Cystic fibrosis mortality and survival in the UK:

1947e2003. Eur Respir J 2007;29:522e6.2. Peckham D, Conway SP, Morton A, et al. Delayed diagnosis of cystic fibrosis

associated with R117H on a background of 7T polythymidine tract at intron 8. J CystFibros 2006;5:63e5.

3. Efrati O, Harash O, Rivlin J, et al. Hemoptysis in Israeli CF patientsdprevalence,treatment, and clinical characteristics. J Cyst Fibros 2008;7:301e6.

4. Flume PA, Mogayzel PJ, Robinson KA, et al. Cystic fibrosis pulmonary guidelines:pulmonary complications: hemoptysis and pneumothorax. Am J Respir Crit Care Med2010;182:298e306.

5. Vidal V, Therasse E, Berthiaume Y, et al. Bronchial artery embolization in adults withcystic fibrosis: impact on the clinical course and survival. J Vasc Interv Radiol2006;17:953e8.

Figure 1 (A) CT pulmonary angiogramdemonstrating thrombus in the left mainand upper lobe pulmonary arteries(white arrow). (B) Coronal CTreconstruction illustrating a dilatedbronchial artery arising from thedescending aorta (black arrow).

Chest clinic

Chest

clinic


Images in Thorax

Pulmonary complications ofintravesicular BCG immunotherapy

A 79-year-old man was admitted tohospital with fever, rigours and malaise1 week after completing his seventhintravesicular BCG immunotherapytreatment for carcinoma in situ of thebladder. On admission, respiratory

examination was normal with a peripheral oxygen saturation of94% on air and a normal white cell count but raised C reactiveprotein of 111. His chest radiograph was normal (figure 1). Hewas treated for presumed urinary sepsis with intravenous anti-biotics while awaiting results of microscopy and culture of urineand blood cultures.

By day 5, the patient had become progressively breathlessrequiring high flow oxygen to maintain his oxygen saturations.Urine and blood cultures taken on admission were negative. Arepeat chest radiograph revealed new bilateral pulmonary infil-trates (figure 2). A CT chest was performed which showedbilateral ground glass changes and multiple tiny pulmonarynodules (figure 3). The differential diagnosis was urinary sepsis,miliary tuberculosis or a pulmonary hypersensitivity reactionand due to the severity of his systemic inflammatory responseand deterioration in his clinical condition, he was started onrifampicin, isoniazid and ethambutol as well as oral predniso-lone at a dose of 60 mg daily. He rapidly improved by day 7 andwas discharged on day 10 with repeat chest radiograph showing

a significant improvement (figure 4). Multiple urine, blood andsputum cultures taken during the acute phase were negative formycobacteria.He continued triple antituberculous medication for 2 months,

and then isoniazid and rifampicin for a further 4 months.Steroids were reduced over a 3-month period; this regime wasimplemented based on his clinical response and informationavailable from previous cases.

Bethan Davies, Harpreet Ranu, Mark Jackson

Department of Medicine, Princess Royal Hospital, Haywards Heath, West Sussex, UK

Correspondence to Dr Bethan Davies, Department of Medicine, Princess RoyalHospital, Haywards Heath, West Sussex RH16 4EX, UK;[email protected] 1 Chest radiograph on admission.

Figure 3 CT chest showing bilateral ground glass changes, mostmarked in the upper lobes and apices with multiple tiny nodules.

Figure 2 Chest radiograph 48 h after admission showing bilateralpulmonary infiltrates.

Chest

clinic

Chest clinic





Received 12 September 2011Accepted 6 February 2012

doi:10.1136/thoraxjnl-2011-201088

REFERENCES1. Lamm DL. Efficacy and safety of bacille Calmette-Guerin immunotherapy in superficial

bladder cancer. Clin Infect Dis 2000;(Suppl 31);3:S861.

2. Lamm DL. Complications of bacillus Calmette-Guerin immunotherapy. Urol Clin NorthAm 1992;19:565.

3. LeMense GP, Strange C. Granulomatous pneumonitis following intravesical BCG:what therapy is needed. Chest 1994;106:1624e6.

4. Palayew M, Briedis D, Libman M, et al. Disseminated infection after intravesicualimmunotherapy. Chest 1993;104:307e9.

5. National Institute for Clinical Excellence. CG117 Tuberculosis: Clinical Diagnosisand Management of Tuberculosis, and Measures for its Prevention and Control.National Institute for Clinical Excellence, 2011.

6. Case records of the Massachusetts General Hospital. Weekly clinicopathologicalexercises. Case 29-1998. A 57-year-old man with fever and jaundice after intravesicalinstillation of bacilli Calmette-Guerin for bladder cancer. N Engl J Med 1998;339:831.

Learning points

< Radiological features consistent with interstitial pneumonitishave been reported in 0.7% of patients receiving BCGimmunotherapy for carcinoma of the bladder.1 It is debatedas to whether the interstitial pneumonitis associated withintravesical BCG is predominantly a hypersensitivity reactionor is a manifestation of miliary disease caused by BCG, orboth.

< Based on case reports, it is suggested that patients should begiven antituberculous medication for 6 months;2e4 pyrazina-mide is not included as BCG is resistant to this drug. It shouldbe noted that the National Institute for Clinical Excellenceguidance on the management of tuberculosis recommendstreatment for a total of 9 months if pyrazinamide is not used inthe initial 2 months of therapy.5 If there appears to bea significant hypersensitivity component, steroids should alsobe given.2e4 Although there is no consensus on the optimalduration of steroid therapy, it is recommended thatprednisolone be reduced over a period of weeks to monthsto reduce the risk of recurrence.6 Microbiological sampling isrecommended although our patient was too hypoxaemic toundergo flexible bronchoscopy.

< Bladder carcinoma is the fourth most common cancer in men(11th most common in women) in the UK with an annualincidence of 10 335 in 2008. Within our trust, 45 patients arecommenced on BCG immunotherapy each year, with a total ofat least 270 instillations. It is important for respiratoryphysicians to be aware of the pulmonary complicationsassociated with its use.

Figure 4 Chest radiograph at day 10 showing resolving pulmonaryinfiltrates.

Chest

clinic


Chest clinic


Thorax 2012;67:933–


934

934.

Date post:	20-Nov-2023
Category:	Documents
Upload:	independent
View:	0 times
Download:	0 times

Chronic obstructive pulmonary disease among residents of an historically industrialised area

Documents