PDE4inibitors Cochrane 2011.pdf

Phosphodiesterase 4 inhibitors for chronic obstructivepulmonary disease (Review)Chong J, Poole P, Leung B, Black PNThis is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library2011, Issue 5http://www.thecochranelibrary.comPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.

T A B L E O FC O N T E N T SHEADER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1PLAIN LANGUAGE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Figure 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Figure 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10AUTHORS’ CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12CHARACTERISTICS OF STUDIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17DATA AND ANALYSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Analysis 1.1. Comparison 1 Lung function, Outcome 1 FEV 1 (by drug). . . . . . . . . . . . . . . . 56Analysis 1.2. Comparison 1 Lung function, Outcome 2 FEV 1 (by mean COPD severity). . . . . . . . . . . 57Analysis 1.3. Comparison 1 Lung function, Outcome 3 FEV 1 (Roflumilast 500 µg by mean COPD severity). . . . 59Analysis 1.4. Comparison 1 Lung function, Outcome 4 FEV 1 (by study duration). . . . . . . . . . . . . 60Analysis 1.5. Comparison 1 Lung function, Outcome 5 FEV 1 (additional medication). . . . . . . . . . . . 61Analysis 1.6. Comparison 1 Lung function, Outcome 6 FEV 1 (published vs. unpublished). . . . . . . . . . 63Analysis 1.7. Comparison 1 Lung function, Outcome 7 FEV 1 (random effects model). . . . . . . . . . . . 64Analysis 1.8. Comparison 1 Lung function, Outcome 8 FEV 1 (Roflumilast 500 µg vs 250 µg). . . . . . . . . 65Analysis 1.9. Comparison 1 Lung function, Outcome 9 FVC. . . . . . . . . . . . . . . . . . . . 66Analysis 1.10. Comparison 1 Lung function, Outcome 10 PEF. . . . . . . . . . . . . . . . . . . . 67Analysis 1.11. Comparison 1 Lung function, Outcome 11 FEV 1 (Evidence quality). . . . . . . . . . . . . 68Analysis 2.1. Comparison 2 Quality of life, Outcome 1 SGRQ total score. . . . . . . . . . . . . . . . 69Analysis 2.2. Comparison 2 Quality of life, Outcome 2 SGRQ total score (by mean COPD severity). . . . . . . 70Analysis 2.3. Comparison 2 Quality of life, Outcome 3 SGRQ total score (by published vs unpublished). . . . . 71Analysis 2.4. Comparison 2 Quality of life, Outcome 4 SGRQ total score (by duration). . . . . . . . . . . 72Analysis 3.1. Comparison 3 Exacerbations, Outcome 1 No. of subjects (by drug). . . . . . . . . . . . . 73Analysis 3.2. Comparison 3 Exacerbations, Outcome 2 Exacerbation rate. . . . . . . . . . . . . . . . 75Analysis 3.3. Comparison 3 Exacerbations, Outcome 3 Exacerbation rate (Inverse variance). . . . . . . . . . 76Analysis 3.4. Comparison 3 Exacerbations, Outcome 4 Exacerbation rate (Roflumilast 500 µg vs 250 µg). . . . . 77Analysis 4.1. Comparison 4 Symptom score, Outcome 1 SGRQ symptom score. . . . . . . . . . . . . . 77Analysis 4.2. Comparison 4 Symptom score, Outcome 2 Borg Scale. . . . . . . . . . . . . . . . . . 78Analysis 4.3. Comparison 4 Symptom score, Outcome 3 Summary symptom score. . . . . . . . . . . . . 78Analysis 4.4. Comparison 4 Symptom score, Outcome 4 Shortness of breath questionnaire. . . . . . . . . . 79Analysis 5.1. Comparison 5 Exercise tolerance, Outcome 1 6-minute walk test. . . . . . . . . . . . . . 79Analysis 6.1. Comparison 6 Adverse effects, Outcome 1 No of patients experiencing an adverse effect. . . . . . 80Analysis 6.2. Comparison 6 Adverse effects, Outcome 2 No of patients experiencing an adverse event (Roflumilast 500 µgvs 250 µg). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82Analysis 6.3. Comparison 6 Adverse effects, Outcome 3 Diarrhoea. . . . . . . . . . . . . . . . . . 82Analysis 6.4. Comparison 6 Adverse effects, Outcome 4 Nausea. . . . . . . . . . . . . . . . . . . 84Analysis 6.5. Comparison 6 Adverse effects, Outcome 5 Headache. . . . . . . . . . . . . . . . . . 85Analysis 6.6. Comparison 6 Adverse effects, Outcome 6 Vomiting. . . . . . . . . . . . . . . . . . . 87Analysis 6.7. Comparison 6 Adverse effects, Outcome 7 Dyspepsia. . . . . . . . . . . . . . . . . . 88Analysis 6.8. Comparison 6 Adverse effects, Outcome 8 Abdominal pain. . . . . . . . . . . . . . . . 89Analysis 6.9. Comparison 6 Adverse effects, Outcome 9 Weight Loss. . . . . . . . . . . . . . . . . . 90Analysis 6.10. Comparison 6 Adverse effects, Outcome 10 Influenza-like symptoms. . . . . . . . . . . . 91Analysis 6.11. Comparison 6 Adverse effects, Outcome 11 Upper Respiratory Tract Infection. . . . . . . . . 92Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.i

Analysis 6.12. Comparison 6 Adverse effects, Outcome 12 Withdrawals due to adverse effects. . . . . . . . . 93Analysis 6.13. Comparison 6 Adverse effects, Outcome 13 Non fatal serious adverse events. . . . . . . . . . 95Analysis 6.14. Comparison 6 Adverse effects, Outcome 14 Mortality. . . . . . . . . . . . . . . . . . 96Analysis 6.15. Comparison 6 Adverse effects, Outcome 15 No. of subjects (additional medication). . . . . . . 98APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99HISTORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99CONTRIBUTIONS OF AUTHORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99DECLARATIONS OF INTEREST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99DIFFERENCES BETWEEN PROTOCOL AND REVIEW . . . . . . . . . . . . . . . . . . . . . 100Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.ii

[Intervention Review]Phosphodiesterase 4 inhibitors for chronic obstructivepulmonary diseaseJimmy Chong 2 , Phillippa Poole 1 , Bonnie Leung 2 , Peter N Black 31 Department of Medicine, University of Auckland, Auckland, New Zealand. 2 University of Auckland, Auckland, New Zealand.3 (Deceased), University of Auckland, Auckland, New ZealandContact address: Phillippa Poole, Department of Medicine, University of Auckland, Private Bag 92019, Auckland, New Zealand.p.poole@auckland.ac.nz.Editorial group: Cochrane Airways Group.Publication status and date: New, published in Issue 5, 2011.Review content assessed as up-to-date: 29 September 2010.Citation: Chong J, Poole P, Leung B, Black PN. Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease. CochraneDatabase of Systematic Reviews 2011, Issue 5. Art. No.: CD002309. DOI: 10.1002/14651858.CD002309.pub3.Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.BackgroundA B S T R A C TChronic obstructive pulmonary disease (COPD) affects symptoms, lung function, quality of life and life expectancy. Apart fromsmoking cessation, there are no other treatments that slow lung function decline. Roflumilast and cilomilast are oral phosphodiesterase4 (PDE 4 ) inhibitors proposed to reduce the airway inflammation and bronchoconstriction seen in COPD.ObjectivesTo evaluate the efficacy and safety of PDE 4 inhibitors in the management of people with stable COPD. Outcomes included lungfunction, quality of life, symptoms, exacerbations and adverse effects.Search strategyWe identified randomised controlled trials (RCTs) from the Cochrane Airways Group Specialised Register of trials (date of last search6 August 2010). We found other trials from web-based clinical trial registers.Selection criteriaWe included RCTs if they compared oral PDE 4 inhibitors with placebo in people with COPD. We allowed co-administration ofstandard COPD therapy.Data collection and analysisOne review author extracted data and a second review author checked the data, before entry into The Cochrane Collaboration softwareprogramme (RevMan version 5.1). We reported pooled data as mean differences (MD), standardised mean differences (SMD), or oddsratios (OR).Main resultsTwenty-three separate RCTs studying roflumilast (nine trials, 9211 patients) or cilomilast (fourteen trials, 6457 patients) met theinclusion criteria. None of the trials exceeded a year in duration.Treatment with a PDE 4 inhibitor was associated with a significant improvement in FEV 1 over the trial period compared with placebo(MD 45.59 mL; 95% confidence interval (CI) 39.15 to 52.03), regardless of COPD severity or concomitant COPD treatment. TherePhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.1

were some small improvements in quality of life (St George’s Respiratory Questionnaire MD -1.04; 95% CI -1.66 to -0.41) and COPDrelatedsymptoms, but no change in exercise tolerance. Treatment with a PDE 4 inhibitor was associated with a reduced likelihood ofCOPD exacerbation (OR 0.78; 95% CI 0.72 to 0.85). More participants in the treatment groups experienced non-serious adverseevents compared with controls, particularly gastrointestinal symptoms and headache. Roflumilast was associated with weight loss duringthe trial period.Authors’ conclusionsIn people with COPD, PDE 4 inhibitors offered benefit over placebo in improving lung function and reducing likelihood of exacerbations,however, they had little impact on quality of life or symptoms. Gastrointestinal adverse effects and weight loss were common.The optimum place of PDE 4 inhibitors in COPD management remains to be defined. Longer-term trials are needed to determinewhether or not PDE 4 inhibitors modify FEV 1 decline, healthcare utilisation or mortality in COPD.P L A I NL A N G U A G E S U M M A R YPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseNew treatments are needed for chronic obstructive pulmonary disease (COPD). This review looked at the effectiveness of a new classof oral medicines, phosphodiesterase 4 (PDE 4 ) inhibitors (roflumilast and cilomilast), that have actions on airways affected by COPD.We identified reports of 23 separate trials. These medicines improve lung function and reduce the likelihood of a flare-up of COPD,however they have little effect on symptoms or quality of life over and above existing treatments. This may be due to side effects,although these are not serious. Longer studies are needed to see if these medicines improve survival for people with COPD.B A C K G R O U N DDescription of the conditionChronic obstructive pulmonary disease (COPD) is one of theleading causes of global morbidity and mortality, resulting in agrowing social and economic burden (GOLD 2005). In 2002, itwas estimated to be the fifth leading cause of death, responsible forapproximately 4.8% of total deaths worldwide and it is projectedto rise to fourth position by the year 2030 (Mathers 2005).COPD is defined by the Global Initiative for Chronic ObstructiveLung Disease (GOLD) as a “preventable and treatable diseasewith some significant extrapulmonary effects that may contributeto the severity in individual patients. Its pulmonary componentis characterized by airflow limitation that is not fully reversible.The airflow limitation is usually progressive and associated withan abnormal inflammatory response of the lung to noxious particlesor gases” (GOLD 2009). Diagnosis is based on a history ofexposure to risk factors for this disease and symptoms of cough,sputum production, or dyspnoea (shortness of breath). Spirometryis required for diagnosis, with airflow obstruction being confirmedby a post-bronchodilator forced expiratory volume in 1second (FEV 1 )/forced vital capacity (FVC) ≤ 0.7 (Celli 2004).Life expectancy is reduced in people diagnosed with COPD, and,although prognosis is variable, age and FEV 1 are the strongest predictorsof mortality.The predominant risk factor for COPD is tobacco smoking,with other environmental pollutants also known to contribute.Cigarette smoke leads to the activation of macrophages and CD8T lymphocytes that release inflammatory mediators and cytokines.The process also involves neutrophil attraction and cell apoptosis(Barnes 2000). To date, smoking cessation is the only interventionknown to slow the decline in lung function associated withCOPD (GOLD 2009).Pharmacotherapy is commonly used to treat people with COPD,with effects on symptoms, quality of life, or frequency and severityof exacerbations (Celli 2004; GOLD 2009). Mainstays of treatmentinclude short- and long-acting inhaled beta-2 agonists andanticholinergics, corticosteroids, and methylxanthines. New approachesto treatment are needed, as no individual agent slowsthe decline in lung function, or survival. In the TORCH study(Calverley 2007) a combination of salmeterol 50 µg and fluticasone500 µg twice daily reduced the risk of death by 17% comparedwith placebo over the three-year trial period; however, thisdid not reach statistical significance.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.2

Description of the interventionThe intervention is an oral medicine which is a selective inhibitorof the isoenzyme phosphodiesterase 4 (PDE 4 ). This isoenzyme hasa role in airway inflammation and bronchoconstriction, both ofwhich are pathological features of COPD (Boswell-Smith 2006).How the intervention might workCyclic adenosine monophosphate (cAMP) is a secondary messengerthat suppresses the activity of inflammatory cells and mediatesthe process of smooth muscle relaxation in the airways. Phosphodiesterases,in turn, hydrolyse and turn off the biological activity ofcAMP (Boswell-Smith 2006). Therefore, inhibitors of phosphodiesteraseaction should theoretically provide improvements in theextent of airway narrowing and damage from inflammation.Non-selective phosphodiesterase (PDE) inhibitors, such as theophylline,a methylxanthine, have been used in the managementof people with COPD for years. These are recommendedby current international guidelines as part of third-line therapy(GOLD 2005). Limitations to their use include a narrowtherapeutic margin, and the frequency of adverse effects, whichmay occur even when the plasma level is within the therapeuticrange (Boswell-Smith 2006). Common adverse effects associatedwith theophylline include headache, nausea, vomiting, diarrhoea,restlessness, nervousness, insomnia, and gastrointestinaleffects (Barnes 2003). Less common, but more serious, are theincreased risks of cardiac arrhythmias and seizures (Barnes 2003).Some of the adverse effects associated with theophylline have beenattributed to its non-selective PDE inhibition and concurrentadenosine receptor antagonism (Barnes 2005).The isoenzyme PDE 4 is the predominant isoenzyme involved withmetabolising cAMP in immune and inflammatory immune cells,such as neutrophils, macrophages, T cells, and endothelial cells inCOPD, in airway smooth muscle and pulmonary nerves (Agusti2005; Boswell-Smith 2006; Torphy 1998; Vignola 2004). Inhibitionof PDE 4 leads to elevation of cAMP in inflammatory andimmunomodulatory cells, resulting in suppression of inflammatorycell function, relaxation of airways smooth muscle, and modulationof pulmonary nerves (Boswell-Smith 2006; Essayan 2001;Torphy 1999). Thus, PDE 4 is an attractive target for inhibition inCOPD. Furthermore, the central nervous system (CNS) and cardiovascularadverse effects experienced in patients treated with thenon-selective PDE inhibitor, theophylline, are a result of adenosinereceptor antagonism. This feature is not present with PDE 4specific inhibitors (Vignola 2004).Why it is important to do this reviewThe development of selective PDE 4 inhibitors offers new hopefor therapy offering both anti-inflammatory and bronchodilatoryeffects in COPD, with fewer of the adverse effects encounteredwith non-selective inhibitors. Additionally, PDE 4 inhibitors maybe easier to use because there is less pharmacokinetic variabilityand lower potential for drug interactions compared with theophylline(Barnes 2005). A number of PDE 4 inhibitors have beendeveloped, with some progressing to Phase III clinical trials. Theseinclude the second generation PDE 4 inhibitors roflumilast (Nycomed,formerly Altana) and cilomilast (GlaxoSmithKline).This review focuses on the effect of PDE 4 inhibitors in the managementof people with stable COPD, using clinically importantoutcomes. Collating this evidence into a systematic review allowsan assessment as to whether or not the theoretical benefits of PDE 4inhibitors translate into useful clinical effects, and may suggest thepotential place of PDE 4 inhibitors within the increasing pharmacopeiaof COPD treatments.O B J E C T I V E STo evaluate the efficacy and safety of oral PDE 4 inhibitors in themanagement of stable COPD.M E T H O D SCriteria for considering studies for this reviewTypes of studiesWe included randomised controlled trials (RCTs) that comparedorally-administered PDE 4 inhibitors with placebo. We includedboth short- and long-term trials. We excluded single-dose trials,as well as trials in acute exacerbations of COPD.Types of participantsAdults (> 18 years of age) with COPD, as defined by the AmericanThoracic Society, or European Respiratory Society or GOLD,with airflow obstruction evident by spirometry with post-bronchodilatorFEV 1 of < 80% of the predicted value and an FEV 1 /FVC ≤ 0.7 (GOLD 2009). We considered trials that includedpatients with both COPD and asthma only if data from patientswith COPD could be extracted separately from the study report,or through correspondence with the authors. We excluded ex-vivoexperiments and trials with patients requiring mechanical ventilationon presentation.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.3

Types of interventionsWe included trials if they compared outcomes for participants whoreceived an orally-administered PDE 4 inhibitor with those of controlparticipants who received placebo. In most trials, participantsin both groups received standard or their usual COPD therapy,or were given the trial intervention (PDE 4 inhibitor or placebo)additionally with another COPD therapy. We permitted the useof rescue medication.Types of outcome measuresPrimary outcomes• Changes in lung function from baseline including forcedexpiratory volume in 1 second (FEV 1 ), forced vital capacity(FVC), or peak expiratory flow rate (PEF)• Quality of life (e.g. total score on St George’s RespiratoryQuestionnaire (SGRQ))Secondary outcomes• Incidence of COPD exacerbations• Symptoms (breathlessness on Borg and other scales andShortness of Breath Questionnaire; composite measures(summary symptom score))• Exercise tolerance (six-minute walk test)• Adverse effects (e.g. gastrointestinal, central nervous system(CNS) and cardiovascular adverse events, change in weight,withdrawal rates)• Serious adverse events and mortalitySearch methods for identification of studiesElectronic searchesWe identified trials using the Cochrane Airways Group SpecialisedRegister of trials, which is derived from systematic searches ofbibliographic databases including the Cochrane Central Registerof Controlled Trials (CENTRAL), MEDLINE, EMBASE,CINAHL, AMED, and PsycINFO, and handsearching of respiratoryjournals and meeting abstracts (please see the Airways GroupModule for further details). There was no restriction on the basisof language. We searched all records in the Specialised Registercoded as ’COPD’ using the following terms:“Phosphodiesterase 4*” OR “PDE4*” OR Rolipram OR RoflumilastOR Cilomilast OR Ariflo OR SB207499 OR Telomilastor ORIC485 OR Oglemilast OR GRC-3886 OR QAK423 ORArofylline OR AWD12-281We conducted the most recent search on 6 August 2010. Theoriginal strategy for this review, which was more sensitive but lessspecific, is in Appendix 1.Searching other resourcesWe reviewed the reference lists of all primary trials and reviewarticles for additional references. We contacted investigators inthis area and pharmaceutical companies manufacturing PDE 4 inhibitorsto ascertain if they could provide any potentially relevanttrial data. We also searched the US Food and Drug Administration(FDA) website and US National Institutes of Health (NIH)clinical trials registry (last accessed 14 Aug 2010).Data collection and analysisSelection of studiesFrom a search in October 2006, we identified 477 abstracts of potentiallyrelevant trials, which was narrowed to 71 abstracts usinga less sensitive search strategy of the Airways Group SpecialisedRegister.We searched again in December 2008 and located 79 abstracts,many of which had been identified in the original search.From these two searches, we sought the full text for 53 trials. Weconducted two searches in January and August 2010, during whichwe identified another 17 abstracts as being potentially eligible. Wedecided whether or not to seek the full text based on a review of theabstract text, title and MeSH headings of all identified citations.Search date:No. of references for which we sought full textDecember 2008 53January 2010 5August 2010 12Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.4

Thus, we sought full papers for 70 abstracts; however, many ofthe abstracts referred to the same trials and some were publishedonly in abstract format. Three review authors (JC, PB and PP)assessed the full text versions of the trials to determine whether ornot they met the inclusion criteria. We resolved any differences bydiscussion. We then assessed trials that met the inclusion criteriafor methodological quality.We found a further 11 completed trials for roflumilast in patientswith COPD in a search of the online clinical trials registry ofthe US National Institutes of Health (NIH). After attemptingto correlate each unique NIH study number with trials alreadyidentified in the above searches of the Airways registry, we foundfour studies on the NIH clinical trials registry that had not beenpublished elsewhere in any form.Furthermore, for cilomilast, we found a summary of the studydesign, along with a report of results, for 12 individual Phase II orIII trials on the GlaxoSmithKline (GSK) website. We identifieda further trial, Compton 2001 from the literature search and weobtained the full text.We identified two trials, both through the literature search andfrom the GSK website (Cilomilast 039; Cilomilast 076). Gamble2003 is the primary published reference for the latter study.Data extraction and managementOne review author (JC) extracted data from the eligible studiesand a second review author (BL) checked the data. We entered datainto RevMan 5. In some cases, we had to estimate informationregarding outcomes from graphs. Where standard errors (SE) werereported, we converted them to standard deviations (SD).We extracted the following data.• Methods: trial design, duration of follow-up.• Participants: age, gender, smoking status, study setting,inclusion and exclusion criteria.• Intervention: drug name, dose, duration of treatment,control and / or standard therapy.• Outcome measures.We categorised references according to the trial name (by drugname and number, or by author and year). We obtained data onadditional outcomes from other references to the same trial, andmore detailed descriptions of study populations from informationsubmitted by the drug manufacturer to the FDA website.Where there were data from more than one report of the sametrial, we considered the data from the primary published referencewith the most complete data on the primary outcome(s) to takepriority. Also, for primary outcomes, we used intention-to-treatanalysis in preference to per-protocol analysis. On the other hand,the trial data on the company website for Cilomilast 076 providedmore complete details of adverse events than Gamble 2003, andwe used this preferentially.Our initial plan had been to perform meta-analysis on thechange from baseline in post-bronchodilator FEV 1 . Only prebronchodilatorvalues were reported for the majority of cilomilasttrials, with pre- and post- bronchodilator values available formost trials of roflumilast, with the exception of Roflumilast FK1101 and Roflumilast FK1 103 for which only post-bronchodilatormeasures were available.In order to increase the number of trials in the meta-analysis, wedecided to use the change in the pre-bronchodilator FEV 1 for alltrials, except for the two just mentioned where it was not available.Between them, these two trials contributed only 3% to the meandifference (MD), with the mean change seen from baseline inFEV 1 similar to other trials.Lung function is reported in millilitres (mL). Exercise tolerance isreported in metres (m).The outcome ’total adverse events’ included the participants ineach group experiencing greater than one or more adverse event,including an acute exacerbation of COPD. Similarly, serious adverseevents included conditions requiring hospital level treatment,and encompassed more serious COPD exacerbations.Assessment of risk of bias in included studiesTrials were assessed by the risk of bias methods outlined in Chapter8 of the Cochrane Handbook for Systematic Reviews of Interventions(Higgins 2008) as Low, Unclear or High.Data synthesisWhere appropriate, we combined data from trials using RevMan5. We expressed results for continuous variables using a fixed-effectMD, or standardised mean difference (SMD), with 95% confidenceintervals (CI). We expressed results for pooled outcomeswith dichotomous variables using a fixed-effect odds ratio (OR)with 95% CI. We considered a P value of < 0.05 statistically significant.Subgroup analysis and investigation of heterogeneityWe planned the following subgroup analyses a priori.• Severity of airflow obstruction at baseline according to theGOLD classification (FEV 1 % predicted GOLD Stage II 50% to80%. Stage III 30% to 50%, Stage IV < 30%).• Drug (e.g. roflumilast, cilomilast).• Dose (e.g. roflumilast 250 µg and 500 µg).• Duration of therapy (12 weeks or less; 24 to 26 weeks; 52weeks).• Concomitant therapy (inhaled or oral corticosteroids,inhaled beta-2 agonists, or anti-cholinergics, or both).Sensitivity analysisWe planned the following sensitivity analyses a priori.• Fixed-effect versus random-effects models for studies withunexplained heterogeneity.• Methodological quality.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.5

We did not anticipate the large number of unpublished trials at theprotocol stage. Consequently, we undertook a sensitivity analysisof the effect sizes of the primary outcomes seen in published andunpublished trials.R E S U L T SDescription of studiesSee: Characteristics of included studies; Characteristics of excludedstudies.Twenty-three separate RCTs studying roflumilast (nine trials with9211 patients) or cilomilast (fourteen trials with 6457 patients)met the inclusion criteria. Twelve of these 23 studies have beenpublished in full in a peer-reviewed journal.Those for roflumilast included the following.• An early dose selection study comparing patients givenroflumilast 250 µg and 500 µg (Roflumilast M2-107) for 24weeks. The subsequent studies all used 500 µg of roflumilast inthe intervention group and this is the dose that appears in theanalyses except where otherwise stipulated.• The first published one-year study of PDE 4 inhibitortreatment in COPD (Roflumilast M2-112). This was the onlytrial included in this review that allowed concomitantcorticosteroid use during the treatment period.• This was followed by two, year-long studies (RoflumilastM2-124; Roflumilast M2-125) investigating the effect in aspecific subgroup - severe to very severe COPD associated withchronic bronchitis in patients at risk of exacerbations.• We found two studies (Roflumilast M2-127; RoflumilastM2-128) that evaluated the add-on use of roflumilast with longactingbronchodilator agents, the first with salmeterol and thesecond with tiotropium. Both studies ran for 24 weeks.• Grootendorst 2007 was a cross-over study of roflumilastaimed at investigating the reduction in sputum neutrophil andeosinophil numbers. Each study period ran for four weeks.Additionally, there were two trials reported only as conferenceposters: Roflumilast FK1 101 and Roflumilast FK1 103. The firstcompared roflumilast 500 µg, roflumilast 250 µg and placebo for26 weeks; the second compared roflumilast 500 µg once daily for24 weeks with roflumilast 500 µg once daily for 12 weeks, thenplacebo once daily for the following 12 weeks.For cilomilast, data were mostly from Phase III clinical trials withone Phase II/III trial. These included unpublished studies. All useda 15 mg dose twice daily except for Compton 2001.The trials fell into four groups.• Compton 2001 was a parallel, six-week, dose-ranging studycomparing placebo with 5 mg, 15 mg and 15 mg doses ofcilomilast, with FEV 1 as the primary outcome.• Pivotal efficacy studies (Cilomilast 039; Cilomilast 042;Cilomilast 091; Cilomilast 156), all of which were 24 weeks induration. Studies Cilomilast 039 and Cilomilast 156 wereconducted in North America, while Cilomilast 042 andCilomilast 091 were conducted in the European Union. Here theprimary study outcomes were change in FEV 1 , lung function,and SGRQ quality of life score.• Supporting studies (Cilomilast 076; Cilomilast 110;Cilomilast 111; Cilomilast 168), all of which lasted for less than24 weeks, with an average trial duration of 10.8 weeks, whereneither FEV 1 lung function nor SGRQ were the primaryoutcomes.• Other studies were Cilomilast 121 (Phase II/III 24 weeks),Cilomilast 157 (52 weeks), and Cilomilast 103657 (24 weeks),which followed the pivotal efficacy studies and were smaller insample size. Cilomilast 180 (18 weeks) had a primary lungfunction endpoint, functional residual capacity, and Cilomilast181 (13 weeks) assessed the number of inflammatory cells in abronchial biopsy.Risk of bias in included studiesWe considered that the methodological quality of the 12 publishedtrials was acceptable overall. In the roflumilast trials, there wereadequate descriptions of allocation concealment in seven out ofnine trials, and the method of blinding in six out of nine trials.Information about allocation concealment and blinding was onlyprovided for one of the 14 cilomilast trials, while further detailsfor the remaining trials were unclear. The risk of bias for eachparameter in the 23 studies is shown in Figure 1 and Figure 2.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.6

Figure 1.Methodological quality graph: review authors’ judgments about each methodological quality itempresented as percentages across all included studies.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.7

The percentage of participant withdrawals varied among the 21trials that reported this outcome. As discussed below under adverseeffects, withdrawals were significantly higher in the PDE 4inhibitor treated groups compared with control groups. The meanwithdrawal rate in the treatment groups was 23% (range 3% to37%). In the control groups, a mean of 18% withdrew (range 3%to 26%).Effects of interventionsChange in lung function from baseline (Analysis 1)Based on 19 trials that reported this outcome, there was a statisticallysignificant improvement in FEV 1 from baseline in the PDE 4inhibitor-treated participants compared with controls (MD 45.59mL; 95% CI 39.15 to 52.03, P < 0.00001), over the study period(Analysis 1.1).The effect on FEV 1 was seen for roflumilast 500 µg (MD 54.32mL; 95% CI 44.39 to 64.25), roflumilast 250 µg (MD 55.03 mL;95% CI 22.10 to 87.96), and for cilomilast 15 mg twice daily(MD 38.15 mL; 95% CI 29.41 to 46.90). Based on data from twotrials, Roflumilast FK1 101 and Roflumilast M2-107, there wasno significant difference in change in FEV 1 between roflumilast250 µg and 500 µg (MD 22.75; 95% CI -6.08 to 51.58).A moderate level of heterogeneity existed for this outcome whenall trials were pooled (Chi² = 41.09, df = 20, P = 0.004; I² = 51%).This is discussed further in the sensitivity analysis below.We conducted subgroup analyses. In trials where informationabout the mean percent predicted FEV 1 was available, the changesin FEV 1 observed (Analysis 1.2 and Analysis 1.3) were statisticallysignificant for both participants categorised as GOLD stage I or II(FEV 1 ≥ 50% predicted) and GOLD stage III or IV (FEV 1

3.2). Furthermore, the 500 µg dose had a statistically significantlygreater effect on exacerbations than the 250 µg dose when the exacerbationrate was compared directly in a single trial (RoflumilastM2-107; Analysis 3.4).In the Roflumilast FK1 101 trial, it was reported that the probabilityof experiencing an exacerbation was reduced by 8% with250 µg of roflumilast and 48% with 500 µg, although the absolutevalue was not reported, nor whether this was statisticallysignificant.Other secondary outcomes (Analyses 4 (symptoms)and 5 (exercise tolerance))Overall, the mean difference in change from baseline with PDE 4inhibitor treatment compared with controls on COPD-relatedsymptoms was small, regardless of the scale used to measure it.The only statistically significant effect was seen in one trial ofcilomilast, on breathlessness scored using a Borg scale (Analysis4.2 MD -0.19, 95% CI -0.33, -0.05, P = 0.007).Exercise tolerance using the six-minute walk test was measured infour cilomilast trials. There was no significant difference in walktest distance between treatment and controls (Analysis 5.1 MD1.92m; 95% CI -7.58 to 11.41).Adverse events (Analysis 6)The likelihood of a participant experiencing an adverse event washigher with PDE 4 inhibitor treatment than with placebo (Analysis6.1; OR 1.20; 95% CI 1.11 to 1.28, P < 0.00001). This effectwas seen for both roflumilast and cilomilast, with no significantheterogeneity (Chi² = 19.65, df = 22, P = 0.60; I² = 0%).We noted a range of adverse effects more frequently in PDE 4inhibitor treated participants. Diarrhoea was more common inPDE 4 inhibitor treated groups than in controls (Analysis 6.3; OR2.81; 95% CI 2.42 to 3.26), as was nausea (Analysis 6.4; OR4.02; 95% CI 3.35 to 4.84), headache (Analysis 6.5; OR 1.60;95% CI 1.35 to 1.89), vomiting (Analysis 6.6; OR 4.01; 95% CI2.80 to 5.74), dyspepsia (Analysis 6.7; OR 3.13; 95% CI 2.30to 4.27) and abdominal pain (Analysis 6.8; OR 1.97; 95% CI1.55 to 2.49). Weight loss was reported more frequently in fourroflumilast trials than in controls (Analysis 6.9; OR 4.62; 95% CI3.38 to 6.31), but had not been studied in the cilomilast trials.There were no significant differences in the incidence of eitherinfluenza-like symptoms Analysis 6.10 or upper respiratory tractinfections Analysis 6.11 between treatment and control groups.The lower dose of roflumilast was not associated with more adverseeffects than placebo; however, this was based on only two trialsand confidence intervals were wide (Analysis 6.2). In the trialof Compton 2001 that studied the effects of varying doses ofcilomilast, adverse effects were seen in 52% of the placebo group,48% of the cilomilast 5 mg group, 47% of the 10 mg group,and 61% of the 15 mg group, with rates of serious adverse eventssimilar across the groups.An increase in withdrawals attributed to adverse effects wasrecorded for both roflumilast and cilomilast treatment groups(Analysis 6.12; OR 1.76; 95% CI 1.58 to 1.96). There was, however,no effect of treatment on non-fatal serious adverse events(Analysis 6.13; OR 0.93; 95% CI 0.83 to 1.04) or mortality(Analysis 6.14; OR 0.85; 95% CI 0.62 to 1.17), although mortalitywas a relatively rare event during the trials.Sensitivity analysesA moderate level of heterogeneity existed for the change in FEV 1outcome when all trials were pooled (Chi² = 41.09, df = 20, P= 0.004; I² = 51%). Using a random-effects model made no differenceto the levels of statistical significance or degree of heterogeneityfor the change in FEV 1 (Analysis 1.7). The effect size wassignificant regardless of whether evidence quality was consideredto be low or unclear risk of bias (Analysis 1.11). There was a significanttreatment effect seen in the 12 published trials (MD 55.85mL; 95% CI 46.66 to 65.03) as well as the seven unpublishedtrials (MD 34.60; 95% CI 25.21 to 43.98) (Analysis 1.6).Taking out the two roflumilast trials in which the change in lungfunction was determined on change in post-bronchodilator FEV 1 ,rather than pre-, there was no change in the MD, and heterogeneityonly increased slightly.The heterogeneity appeared to be a function of the cilomilast treatmentstudies (Chi² = 23.75, df = 9, P = 0.005; I² = 62%) as therewas no significant heterogeneity in the roflumilast trials. By sequentialelimination, the six-week Compton 2001 cilomilast trialwas identified as a significant contributor towards the heterogeneityof pooled results.There was a difference in effect size of the total SQRQ score betweenpublished and unpublished trials (Analysis 2.3 MD -1.93;95% CI -3.02 to -0.83 versus MD -0.43; 95% CI -1.26 to 0.40).D I S C U S S I O NThis systematic review evaluated RCTs that assessed the efficacyand safety of oral PDE 4 inhibitors in people with COPD. Thefirst major finding, based on data from 19 trials, was that bothroflumilast and cilomilast led to greater improvements in lungfunction from baseline, as measured by FEV 1 , FVC or PEF, thanplacebo. Furthermore, the improvement in lung function was seenregardless of the severity of the disease. Significant improvementoccurred even when treatment was given in addition to other standardCOPD treatments such as beta-2 agonists, anti-cholinergics,and inhaled corticosteroids (ICS).The mean change in FEV 1 was, however, below what is usuallyconsidered a minimum clinically important difference (100 mL)Donohue 2005, but comparable to those seen with other COPDtreatments in recent large RCTs. For example, the mean improvementin FEV 1 of 46 mL with treatment over controls seen in thisPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.10

eview is of similar magnitude to that seen with fluticasone (47mL), salmeterol (42 mL), and fluticasone and salmeterol combined(92 mL) in the TORCH 2007 study in people with severeCOPD.A second major finding, based on data from 17 trials, was thatparticipants were more likely to be exacerbation-free while beingtreated with PDE 4 inhibitors compared with controls. Overall,participants were about 25% less likely to have an exacerbation,translating to a number needed to treat to benefit of around five,for one person to be exacerbation-free in the study period. Whilethe likelihood of an individual experiencing an exacerbation waslowered with PDE 4 inhibitor treatment, the decrease in the rateof exacerbations was less marked, only just reaching statistical significance.Taken together, these two major findings suggest PDE 4 inhibitorsin patients with COPD are acting independently of other treatments,particularly bronchodilators. This is an encouraging findingthat could be consistent with a broad anti-inflammatory effect(Fabbri 2009). In support of this is that the improvementin FEV 1 and effect on exacerbations was less marked in the onetrial Roflumilast M2-112 where patients were already on inhaledsteroids, which are also anti-inflammatory. On the other hand,short-duration studies showed more favourable results than thelonger studies, but the reasons for this are unclear. Significant heterogeneitywas noted among the former set of trials but not thelatter, suggesting unmeasured differences between the trials maybe having an impact.A third major finding of the review was that, despite significantimprovements in lung function and a reduction in exacerbationsin participants treated with PDE 4 inhibitors, there was only asmall improvement in quality of life as assessed by SGRQ totalscore. The average change in SGRQ total score was 1.04 units(over a duration of between six and 12 months), and was of similarmagnitude among trials of participants with milder or more severeCOPD. Although this improvement was statistically significant,a change of greater than four units is usually regarded as the minimumclinically important difference (Jones 2005). While symptomscores were marginally better in the treatment groups, therewas no change seen in exercise tolerance, suggesting that improvementsin respiratory symptoms may not necessarily correlate withoverall enhancement of physical functioning. There were, though,smaller number of trials assessable for these outcomes, raising thepossibility of Type 1 or Type 2 errors. Quality of life has beenchosen as a primary outcome because of concerns as to whetheror not the adverse effects of PDE 4 inhibitors might outweigh anybeneficial COPD-related effects.This review found that adverse effects were greater in the roflumilastand cilomilast treated participants than in those receivingplacebo, particularly gastrointestinal-related effects such as diarrhoea,nausea, vomiting and dyspepsia. Participants in the treatmentgroups were also more likely to withdraw from the trials becauseof adverse effects; on average 15% in the treatment groupswithdrew compared with 9% in the control groups. Nevertheless,there is only a slight excess in the total number of participants inthe treatment groups experiencing any adverse effect, comparedwith controls (Analysis 6.1). As this analysis included symptoms aswell as exacerbations, which were reduced in the treatment groups,this analysis will tend to under-estimate the excess of non COPDrelatedadverse effects occurring with PDE 4 inhibitor treatment.It was notable that treatment with roflumilast was associated with asignificant chance of weight loss. Whether this was due to anorexiafrom gastrointestinal adverse effects, or another effect, is not yetclear. Also not clear is whether cilomilast has the same effect as ithas not been studied. As maintenance of weight is important inthe later stages of COPD (GOLD 2009). This adverse effect isconcerning and warrants further investigation. Reassuringly, therewas no increase in serious adverse effects or mortality, althoughtrials were of relatively short duration, and analyses underpoweredto report on the latter outcome.While a lower dose (250 µg) of roflumilast did improveFEV 1 significantly, it was not associated with fewer adverse effectsthan the larger dose (Analysis 6.2), or with a significant reductionin exacerbations in the one trial that reported this (Analysis 3.4).Moreover, it has not been studied as add-on therapy to other bronchodilators.On the other hand, lower doses of cilomilast did notimprove FEV 1 . Thus, the doses of PDE 4 inhibitors chosen by thepharmaceutical companies for subsequent trials appear appropriate,based on the evidence reported to date.A U T H O R S ’ C O N C L U S I O N SImplications for practicePhosphodiesterase 4 inhibitors are oral medicines that may betaken in combination with other standard COPD treatments. Themost evidence exists for roflumilast at a dose of 500 µg daily andcilomilast at 15 mg twice daily.Phosphodiesterase 4 inhibitors join an increasing list of treatmentsfor COPD that improve short-term lung function and reduce exacerbations,but have not been shown to increase life expectancy.Trials to date have been one year or less in duration. In contrast tolong-acting bronchodilators, PDE 4 inhibitors have minimal benefitson symptoms on a day-to-day basis or quality of life, and areoften associated with adverse effects, especially of the gastrointestinalsystem, and headaches. Encouraging people with COPD tocontinue to take these medicines in the in the absence of symptomaticrelief may be challenging.Implications for researchThis review has highlighted several possible areas for further study:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.11

• longer-duration studies to look at the effect of PDE 4inhibitors on FEV 1 decline and mortality;• a direct comparison of PDE 4 inhibitors and ICS, whenused as add-on therapies to either tiotropium or long acting beta-2 agonists, or both;• a direct comparison of either tiotropium or long-actingbeta-2 agonists, or both, as add-on therapies to PDE 4 inhibitors(+/- ICS);• effect of roflumilast on quality of life;• better characterisation of the weight loss seen with PDE 4inhibitors in COPD;• better description of the nature of the effect on theexacerbations that do occur;• use of PDE 4 inhibitors in acute exacerbations;• effect of PDE 4 inhibitors on healthcare utilisation,including hospitalisation (incidence and bed days);• cost-effectiveness of PDE 4 inhibitors;• ascertaining exercise tolerance data for roflumilast;• using the effects of PDE 4 inhibitors to better understandthe pathophysiology of COPD.A C K N O W L E D G E M E N T SThis review is dedicated to Professor Peter Black (deceased January2010) who led the development of the protocol and the earlypart of the review. Peter made significant contributions throughresearch, teaching and clinical practice to the furthering of evidence-basedmanagement of airways diseases.We thank Claire Arandjus for her contribution to protocol development.To Nycomed and Forest Laboratories for confirming study detailsand results extracted from published articles and abstracts.GlaxoSmithKline (GSK) for study summaries available via theGSK online clinical study register.R E F E R E N C E SReferences to studies included in this reviewCilomilast 039 {published data only}Edelson JD, Compton C, Nieman R, Robinson CB, Amit O,Bagchi I, et al.Cilomilast (Ariflo) a potent selectivephosphodiesterase 4 inhibitor, reduces exacerbations in COPDpatients: Results of a 6 month trial. American Journal of Respiratoryand Critical Care Medicine 2001;163(5 Suppl):A771.Edelson JD, Compton C, Nieman R, Robinson CB, Watt R, AmitO, et al.Cilomilast (Ariflo) improves health status in patients withCOPD: Results of a 6-month trial. American Journal of Respiratoryand Critical Care Medicine 2001;163(5 Suppl):A277.GSK CTR-039. A randomized, 24-week, double-blind, placebocontrolled,parallel-group study to evaluate the efficacy, safety andtolerability of cilomilast (15 mg twice daily) in patients withchronic obstructive pulmonary disease. http://www.gskclinicalstudyregister.com/files/pdf/24043.pdf.Accessed 25/05/2010.Rennard SI, Schachter N, Strek M, Rickard K, Amit O. Cilomilastfor COPD: Results of a 6-month, placebo controlled study of apotent, selective inhibitor of phosphodiesterase 4. Chest 2006;129(1):55–66.Cilomilast 042 {unpublished data only}∗ GSK CTR-042. A randomized, 24-week, double-blind, placebocontrolled,parallel-group study to evaluate the efficacy, safety andtolerability of cilomilast (15 mg twice daily) in patients withchronic obstructive pulmonary disease. http://www.gskclinicalstudyregister.com/files/pdf/24046.pdf.Accessed 25/05/2010.Cilomilast 076 {published and unpublished data}Gamble E, Grootendorst DC, Brightling CE, Troy S, Qiu Y, Zhu J,et al.Antiinflammatory effects of the phosphodiesterase-4 inhibitorcilomilast (Ariflo) in chronic obstructive pulmonary disease.American Journal of Respiratory and Critical Care Medicine 2003;168:976–82.∗ GSK CTR-076. A 12-week, multicentre, double-blind, placebocontrolled,parallel-group study to evaluate the anti-inflammatoryactivity of SB207499 15 mg twice daily in patients with chronicobstructive pulmonary disease. http://www.gskclinicalstudyregister.com/files/pdf/24047.pdf.Accessed 25/05/2010.Cilomilast 091 {unpublished data only}∗ GSK CTR-091. A randomized, 24-week, double-blind, placebocontrolled,parallel-group study followed by a 2-week, randomized,double-blind, run-out phase to evaluate the efficacy, safety,tolerability and discontinuation of SB207499 (15 mg twice daily)in patients with chronic obstructive pulmonary disease. http://www.gsk-clinicalstudyregister.com/files/pdf/24048.pdf. Accessed25/05/2010.Cilomilast 103657 {unpublished data only}∗ GSK CTR-657. A randomized, 24-week, double-blind, placebocontrolled,parallel-group study to evaluate the efficacy, safety andtolerability of cilomilast (15 mg BID) in patients with ChronicObstructive Pulmonary Disease (COPD). http://www.gskclinicalstudyregister.com/files/pdf/20593.pdfAccessed 25/05/10.Cilomilast 110 {unpublished data only}∗ GSK CTR-110. A 12-week, multicenter, double-blind, placebocontrolled,parallel-group study to evaluate the anti-inflammatoryPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.12

activity of cilomilast 15 mg twice daily in patients with chronicobstructive pulmonary disease. http://www.gskclinicalstudyregister.com/files/pdf/24049.pdf.Accessed 25/05/2010.Cilomilast 111 {published and unpublished data}∗ GSK CTR-111. A 12-week, randomized, double-blind, placebocontrolled,parallel-group study to investigate the effect ofcilomilast (15 mg twice daily) on trapped gas volume in patientswith chronic obstructive pulmonary disease. http://www.gskclinicalstudyregister.com/files/pdf/24050.pdf.Accessed 25/05/2010.Zamel N, McClean P, Zhu J, Schryver B, Madan A, Robinson CB,et al.Effect of cilomilast (Ariflo) on trapped gas volume and indicesof hyperinflation in patients with chronic obstructive pulmonarydisease. American Journal of Respiratory and Critical CareMedicine 2002; Vol. 165(Suppl 8):A226.Cilomilast 121 {unpublished data only}∗ GSK CTR-121. A randomized, 24-week, double-blind, placebocontrolled,parallel-group study to evaluate the efficacy, safety andtolerability of cilomilast (15 mg BID) in patients with chronicobstructive pulmonary disease. http://www.gskclinicalstudyregister.com/files/pdf/24042.pdf.Accessed 25/05/2010.Cilomilast 156 {unpublished data only}∗ GSK CTR-156. A randomized, 24-week, double-blind, placebocontrolled,parallel-group study to evaluate the efficacy, safety andtolerability of cilomilast (15 mg BID) in patients with chronicobstructive pulmonary disease. http://www.gskclinicalstudyregister.com/files/pdf/24051.pdf.Accessed 25/05/2010.Cilomilast 157 {unpublished data only}∗ GSK CTR-157. A randomised, double-blind, placebo-controlled,parallel-group study to evaluate the efficacy, safety and tolerabilityof oral cilomilast (15 mg bd) when given as maintenance treatmentfor 12 months to subjects with chronic obstructive pulmonarydisease. http://www.gsk-clinicalstudyregister.com/files/pdf/24053.pdf. Accessed 25/05/2010.Cilomilast 168 {published and unpublished data}∗ GSK CTR-168. A randomized, 12-week, double-blind, placebocontrolled,parallel-group study to evaluate the safety andtolerability of cilomilast 15 mg twice daily in patients with chronicobstructive pulmonary disease. http://www.gskclinicalstudyregister.com/files/pdf/24054.pdf.Accessed 25/05/2010.Reisner C, Zhu J, Morris A, Lim J, Knobil K. Assessment of cardiacevents via 24-hour electrocardiographic (Holter) monitoring withcilomilast in chronic obstructive pulmonary disease. EuropeanRespiratory Journal 2003;22(Suppl 45):Abstract No:P522.Reisner C, Zhu J, Morris A, Lim J, Knobil K. Assessment of cardiacevents via 24-hour electrocardiographic (Holter) monitoring withcilomilast in chronic obstructive pulmonary disease [Abstract].American Thoracic Society 99th International Conference. 2003:Poster D86, A035.Cilomilast 180 {unpublished data only}∗ GSK CTR-180. An 18-week randomized, double-blind, placebocontrolled,multicenter study designed to compare treatment withcilomilast to that with placebo for changes in ventilatory mechanicsand function (both at rest and during exercise), as well as relatedexertional dyspnea and exercise performance, in hyperinflatedpatients with stable COPD. http://www.gskclinicalstudyregister.com/files/pdf/24052.pdfAccessed 25/05/2010.Cilomilast 181 {unpublished data only}∗ GSK CTR-181. A 13-week randomised, double-blind, parallelgroup, multicentre study to compare the bronchial antiinflammatoryactivity of oral cilomilast (15 mg bd) with placebotwice daily in subjects with chronic obstructive pulmonary disease.http://www.gsk-clinicalstudyregister.com/files/pdf/24055.pdf.Assessed 25/05/10.Compton 2001 {published and unpublished data}Compton CH, Gubb J, Cedar E, Bakst A, Nieman RB, Amit O, etal.SB 207499, a second generation, oral PDE4 inhibitor, improveshealth status in patients with COPD. European Respiratory Society1999 Madrid, Spain:[P2237].∗ Compton CH, Gubb J, Nieman R, Edelson J, Amit O, Bakst A,et al.Cilomilast, a selective phosphodiesterase-4 inhibitor fortreatment of patients with chronic obstructive pulmonary disease: arandomised, dose-ranging study. Lancet 2001;358(9278):265–270.Grootendorst 2007 {published data only}∗ Grootendorst DC, Gauw SA, Verhoosel RM, Sterk PJ, HospersJJ, Bredenbröker D, et al.Reduction in sputum neutrophil andeosinophil numbers by the PDE4 inhibitor roflumilast in patientswith COPD. Thorax 2007;62(12):1081–7.Roflumilast FK1 101 {published and unpublished data}Bredenbroker D, Syed J, Leichtl S, Rathgeb F, Wurst W.Roflumilast, a new, orally active phosphodiesterase 4 inhibitor, iseffective in the treatment of chronic obstructive pulmonary disease[Abstract]. European Respiratory Society Annual Congress. 2002:Abstract no:2330.∗ Bredenbroker D, Syed J, Leichtl S, Rathgeb F, Wurst W. Safety ofonce-daily roflumilast, a new, orally active, selectivephosphodiesterase 4 inhibitor, in patients with COPD. AmericanJournal of Respiratory and Critical Care Medicine. 2002; Vol. 165(Suppl 8):A595.Leichtl S, Syed J, Bredenbroker D, Rathgeb F, Wurst W.Roflumilast, a new, orally active, selective phosphodiesterase 4inhibitor, is safe and well tolerated in patients with chronicobstructive pulmonary disease [Abstract]. European RespiratorySociety Annual Congress. 2002:Abstract no: P1907.Leichtl S, Syed J, Bredenbröker D, Rathgeb F, Wurst W. Efficacy ofonce-daily roflumilast, a new, orally active, selectivephosphodiesterase 4 inhibitor, in chronic obstructive pulmonarydisease. American Journal of Respiratory and Critical Care Medicine2002;165(Suppl 8):A229.Roflumilast FK1 103 {unpublished data only}Boszormenyi-Nagy G, Pieters WR, Steffen H, Timar M, Vinkler I,Teichmann P, et al.The effect of roflumilast treatment andsubsequent withdrawal in patients with COPD. American ThoracicSociety International Conference 2005 San Diego, California;B93:Poster 323.Roflumilast M2-107 {published and unpublished data}Bateman ED, Holmes M, Muir JF, Andrae K, Witte S,Bredenbroeker D. Safety profile of roflumilast, a novel, selectivephosphodiesterase 4 inhibitor, in patients with moderate to severePhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.13

COPD [Abstract]. American Thoracic Society 100th InternationalConference. 2004 Orlando:C43;Poster F17.O’Donnell D, Muir JF, Jenkins C, Plit P, Brockhaus F, Witte S, etal.Roflumilast, a novel selective phosphodiesterase 4 inhibitor,improves quality of life and lowers exacerbation rate in patients withmoderate to severe COPD [Abstract]. American Thoracic Society100th International Conference. 2004 Orlando:C44;Poster J58.Rabe F, O’Donnell D, Muir F, Jenkins C, Witte S, BredenbroekerD, et al.Roflumilast an oral once daily PDE4 inhibitor improveslung function and reduces exacerbation rates in patients withCOPD [Abstract]. European Respiratory Journal 2004;24 (Suppl48):21s.∗ Rabe KF, Bateman ED, O’Donnell D, Witte S, Bredenbröker D,Bethke TD. Roflumilast - an oral anti-inflammatory treatment forchronic obstructive pulmonary disease: a randomised controlledtrial. Lancet 2005;36(9485):563–71.Rabe KF, Chapman KR, Joubert J, Vetter N, Witte S,Bredenboecker D. Roflumilast, a novel, selective phosphodiesterase4 inhibitor, improves lung function in patients with moderate tosevere COPD [Abstract]. American Thoracic Society 100thInternational Conference. 2004 Orlando:C22;Poster 509.Rabe KF, O’Donnell D, Bateman ED, Andrae K, Witte S,Bredenbroeker D. Roflumilast improves lung function and qualityof life in chronic obstructive pulmonary disease [Abstract]. Chest2004;126(4 Suppl):709S–a.Roflumilast M2-112 {published and unpublished data}Calverley PM, Fabbri LM, Teichmann P, Bredenbroeker D. Effectof roflumilast on lung function and exacerbations in patients withchronic obstructive pulmonary disease: results of a one year study[Abstract]. Thorax 2005;2(Suppl II):ii42.∗ Calverley PM, Sanchez-Toril F, McIvor A, Teichmann P,Bredenbroeker D, Fabbri LM. Effect of 1-year treatment withroflumilast in severe chronic obstructive pulmonary disease.American Journal of Respiratory and Critical Care Medicine 2007;176(2):154–61.Calverley PM, Sanchez-Toril F, McIvor RA, Teichmann P,Bredenbroeker D, Fabbri LM. Effect of roflumilast on lungfunction: a 1-year study in patients with severe to very severeCOPD [Abstract]. Proceedings of the American Thoracic Society.2006:A725.Fabbri LM, Sanchez-Toril F, McIvor RA, Teichmann P,Bredenbroeker D, Calverley PM. Effect of roflumilast onexacerbations: a 1-year study in patients with severe to very severeCOPD [Abstract]. American Thoracic Society. 2006:A841;Poster615.Mclvor RA, Calverley PM, Sanchez-Toril F, Teichmann P,Bredenbroeker D, Fabbri LM. Effect of roflumilast on quality oflife: a 1-year study in patients with severe to very severe COPD.American Thoracic Society 2006;3:A850.Rutten-van Molken M, Van Nooten F, Lindermann M, Caser M.The 1-year cost effectiveness of roflumilast for the treatment ofsevere to very severe COPD patients [Abstract]. EuropeanRespiratory Journal 2007;30(Suppl 51):194s, P1188.Roflumilast M2-124 {published and unpublished data}∗ Calverley PM, Rabe KF, Goehring U-M, Kristiansen S, FabbriLM, Martinez FJ, et al.Roflumilast in symptomatic chronicobstructive pulmonary disease: two randomised clinical trials.Lancet 2009;374(9691):685–94.Martinez F, Hanania N, AURA Study Team. Efficacy and safety ofthe phosphodiesterase-4 inhibitor roflumilast in patients withsymptomatic chronic obstructive pulmonary disease in the M2-124study. Chest 2009;136(4):3S–e.Roflumilast M2-124+M2-125 {published data only}Calverley P, Fabbri L, Rabe K, Goehring UM, Martinez F. Efficacyof the PDE4 inhibitor roflumilast in COPD patients with chronicbronchitis [Abstract]. European Respiratory Society AnnualCongress. 2009 Vienna, Austria:1629.∗ Calverley PM, Rabe KF, Goehring U-M, Kristiansen S, FabbriLM, Martinez FJ, et al.Roflumilast in symptomatic chronicobstructive pulmonary disease: two randomised clinical trials.Lancet 2009;374(9691):685–94.Hanania NA, Brose M, Larsson T, Rabe KF. Efficacy of roflumilastin patients receiving concomitant treatments for chronicobstructive pulmonary disease over 12 months [Abstract].American Journal of Respiratory and Critical Care Medicine. 2010:181(Meeting Abstracts);A4435.Martinez F, Fabbri L, Rabe K, Goehring U-M, Calverley P. Safetyof the PDE4 inhibitor roflumilast in COPD patients with chronicbronchitis [Abstract]. European Respiratory Society AnnualCongress. 2009 Vienna, Austria:1630.Roflumilast M2-125 {published data only}Andrew M, Fernando J, HERMES Study Team. Efficacy and safetyof the phosphodiesterase 4 inhibitor roflumilast in patients withsymptomatic chronic obstructive pulmonary disease in the M2-125study. Chest 2009;136(4):93S–b,94.∗ Calverley PM, Rabe KF, Goehring U-M, Kristiansen S, FabbriLM, Martinez FJ, et al.Roflumilast in symptomatic chronicobstructive pulmonary disease: two randomised clinical trials.Lancet 2009;374(9691):685–94.Roflumilast M2-127 {published data only}Chapman K R, McIvor A, Maltais F, EOS Study Team. Additionalclinical benefit in patients with chronic obstructive pulmonarydisease treated with roflumilast and salmeterol. Chest 2009;136(4):3S–f.∗ Fabbri LM, Calverley PM, Izquierdo-Alonso JL, Bundschuh DS,Brose M, Martinez FJ, et al.Roflumilast in moderate-to-severechronic obstructive pulmonary disease treated with long actingbronchodilators: two randomised clinical trials. Lancet 2009;374(9691):695–703.Izquierdo JL, MacNee W, Biermann E, Goehring U-M, McIvor A.The PDE4 inhibitor roflumilast provides additional clinical benefitin COPD patients receiving salmeterol [Abstract]. EuropeanRespiratory Society Annual Congress. 2009 Vienna, Austria:1627.Roflumilast M2-128 {published data only}∗ Fabbri LM, Calverley PM, Izquierdo-Alonso JL, Bundschuh DS,Brose M, Martinez FJ, et al.Roflumilast in moderate-to-severechronic obstructive pulmonary disease treated with long actingbronchodilators: two randomised clinical trials. Lancet 2009;374(9691):695–703.Paggiaro P, Foden A. Improvements in breathlessness in patientswith chronic obstructive pulmonary disease treated with roflumilastand tiotropium. Chest 2009;136(4):3S–g,4.Rabe K, Paggiaro P, Bernabeu L, Brose M, Geohring U-M, FabbriPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.14

L. Roflumilast, a PDE4 inhibitor, improves lung function inpatients with COPD treated with tiotropium [Abstract]. EuropeanRespiratory Society Annual Congress. 2009 Vienna, Austria:1628.Wouters EFM, Teichmann P, Brose M, Rabe KF, Fabbri LM.Effects of roflumilast, a phosphodiesterase 4 inhibitor, on bodycomposition in chronic obstructive pulmonary disease [Abstract].American Journal of Respiratory and Critical Care Medicine. 2010:181(Meeting Abstracts);A4473.References to studies excluded from this reviewBorker 2003 {published data only}Borker RD, Morris A, Lim J, Zhu J, Reisner C. Effect of cilomilaston quality of life improvement/ deterioration and non-drug costs inpatients with chronic obstructive pulmonary disease. Chest 2003;124(4):170S–b,171.Ferguson 2003 {published data only}Ferguson G, Fischer TL, Morris A, Zhu J, Barnhart F, Reisner C.Cardiovascular safety of cilomilast in patients with chronicobstructive pulmonary disease. Chest 2003;124(4):171S.Fischer 2003 {published data only}Fischer T, Borker R, Barnhart F, Morris A, Zhu J. Effect ofcilomilast on chronic obstructive pulmonary disease patients withimpaired quality of life. Chest 2003;124(4):129S.Grootendorst 2001 {published data only}Grootendorst DC, Gauw SA, Kelly J, Murdoch RD, Sterk PJ, RabeKF. First dose bronchodilating effect of phosphodiesterase- 4 (PDE-4) inhibition by cilomilast (Ariflo) with or without coadministrationof salbutamol and/or ipratrpium in COPD patients.European Respiratory Journal 2001;18(Suppl 33):1:35s.Grootendorst 2002 {published data only}Grootendorst, DC, Gauw, SA, Verhoosel, R, van der Veen H, vander Linden A, Moesker, H, et al.Effect of a PDE4 inhibitor (Bay19-8004) on FEV1 and airway inflammation in patients withCOPD. American Journal of Respiratory and Critical Care Medicine2002;165(8 Suppl):A226.Grootendorst 2003 {published data only}Grootendorst DC, Gauw SA, Baan R, Kelly J, Murdoch RD, SterkPJ, et al.Does a single dose of the phosphodiesterase 4 inhibitor,cilomilast (15mg), induce bronchodilation in patients with chronicobstructive pulmonary disease?. Pulmonary Pharmacology andTherapeutics 2003;16(2):115–20.GSK256066 {published data only}Lazaar AL, Mistry S, Barrett C, Lulic-Burns Z. A four-weekrandomized study of the safety and tolerability of the inhaled PDE4inhibitor GSK256066 in COPD [Abstract]. American Journal ofRespiratory and Critical Care Medicine 2010;181(MeetingAbstracts):A4444.Kelsen 2002 {published data only}Kelsen SG, Rennard SI, Chodosh S, Schryver B, Vleisides C, Zhu J.COPD exacerbation in a 6-month trial of cilomilast (Ariflo) apotent, selective phosphodiesterase 4 inhibitor. American Journal ofRespiratory and Critical Care Medicine 2002;165(Suppl8):A271.Knobil 2003 {published data only}Knobil K, Morris A, Zhu J, Fischer T, Reisner C. Cilomilast isefficacious in chronic obstructive pulmonary disease [Abstract].American Thoracic Society 99th International Conference. 2003:A035;Poster D92.Reisner C, Morris A, Zhu J, Fischer T, Knobil K. Cilomilast isefficacious in chronic obstructive pulmonary disease [Abstract].European Respiratory Journal 2003;22(Suppl 45):Abstract No:P530.Lim 2004 {published data only}Lim S, Zhu J, Lake P. Cilomilast decreases exacerbations andmaintains lung function in patients with poorly reversible COPD.European Respiratory Journal 2004;24(Suppl 48):88s.Nieman 1999 {unpublished data only}Nieman RB, Taneja DT, Amit O, Benincosa LJ, Compton CH,Bethala VK, et al.The effects of low dose SB207499, a secondgeneration, oral PDE4 inhibitor, in patients with COPD. EuropeanRespiratory Society. 1999 Madrid, Spain:P2236.Pascoe 2007 {unpublished data only}Pascoe SJ, Bonner J, Hauffe S, Bohnemeier H. Gradual doseescalation of QAK423, a novel PDE4 inhibitor, significantlyimproves the tolerability [Abstract]. American Thoracic SocietyInternational Conference. 2007 San Francisco, California:PosterC31.Reisner 2002 {published data only}Reisner C, Fischer T, Morris A, Zhu J, Barnhart F. Cilomilastreduces exacerbations in COPD patients. Chest. 2002:S148.Reisner 2003 {published data only}Reisner C, Morris A, Barnhart F, Fischer TL, Acusta A, Darken P.Cilomilast reduces exacerbations in patients with chronicobstructive pulmonary disease. Chest 2003;124(4):Reisner C,Morris A, Barnhart F, Fischer TL, Acusta A, Darken P.Rennard 2008 {published data only}Rennard SI, Calverley PMA, Rempel A, Bredenbroker D, MartinezFJ. The effect of roflumilast treatment on exacerbations in patientswith COPD results of a pooled analysis of two 1-year studies[Abstract]. American Thoracic Society International Conference.2008 Toronto:A963.SB207499/040 {unpublished data only}GSK CTR-040. A multicentre, open-label extension study toevaluate the safety, tolerability and efficacy of oral SB-207499 (15mg twice daily) in patients with chronic obstructive pulmonarydisease. http://www.gsk-clinicalstudyregister.com/files/pdf/24044.pdf. Accessed 25 May 2010.SB207499/041 {unpublished data only}GSK CTR-041. A multicenter open-label extension study toevaluate the safety, tolerability and efficacy of oral cilomilast (15 mgtwice daily) in patients with chronic obstructive pulmonary disease.http://www.gsk-clinicalstudyregister.com/files/pdf/24045.pdf.Accessed 25/05/2010.Song 2005 {published data only}Song Y, Wang C, Liao X, Wang Y, Li Q, Zhao Z, etal.Improvement in lung residual volume in patients with COPDroles of anti-inflammation activity of cilomilast. Respiratory 2005;10(Suppl3):A135.Spencer 2002 {published data only}Spencer MD, Zhu J, Izard D. The direct costs of exacerbations inCOPD and the effect of Cilomilast treatment. EuropeanRespiratory Journal. 2002; Vol. 20(Suppl 38):245s.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.15

Vestbo 2007 {published data only}Vestbo J, Tan L, Atkinson G. A 6 week study of the efficacy andsafety of UK 500,001 dry powder for inhalation (DPI) in adultswith chronic obstructive pulmonary disease [Abstract]. EuropeanRespiratory Journal 2007;30(Suppl 51):612s P3598.Vestbo 2009 {published data only}Vestbo J, Tan L, Atkinson G, Ward J. A controlled trial of 6-weeks’treatment with a novel inhaled phosphodiesterase type-4 inhibitorin COPD. European Respiratory Journal 2009;33(5):1039–1044.Wang 2005 {published data only}Wang C, Song Y, Liao X. Efficacy and anti-inflammation activity ofa selective phospodiesterase-4 inhibitor cilomilast in treatment ofCOPD. Chest 2005;128(4):262S–a.Additional referencesAgusti 2005Agusti A. COPD, a multicomponent disease: implications formanagement. Respiratory Medicine 2005;99(6):670–82.Barnes 2000Barnes P. Medical progress: chronic obstructive pulmonary disease.New England Journal of Medicine 2000;343:269–80.Barnes 2003Barnes P. Theophylline: new perspectives for an old drug. AmericanJournal of Respiratory and Critical Care Medicine 2003;167(6):813–8.Barnes 2005Barnes P. Theophylline in chronic obstructive pulmonary disease:new horizons. Proceedings of the American Thoracic Society 2005;2(4):334–9.Boswell-Smith 2006Boswell-Smith V, Spina D, Page C. Phosphodiesterase inhibitors.British Journal of Pharmacology 2006;147:s252–7.Calverley 2007Calverley PM, Anderson JA, Celli B, Ferguson GT, Jenkins C, JonesPW, et al.for the TORCH investigators. Salmeterol and fluticasonepropionate and survival in chronic obstructive pulmonary disease.New England Journal of Medicine 2007;356(8):775–89.Celli 2004Celli B, MacNee W. Standards for the diagnosis and treatment ofpatients with COPD: a summary of the ATS/ERS position paper.European Respiratory Journal 2004;23(6):932–46.Donohue 2005Donohue J. Minimal clinically important differences in COPDlung function. COPD 2005;2:111–24.Essayan 2001Essayan D. Cyclic nucleotide phosphodiesterases. Journal of Allergyand Clinical Immunology 2001;108(5):671–80.Fabbri 2009Fabbri LM, Calverley PM, Izquierdo-Alonso JL, Bundschuh DS,Brose M, Martinez FJ et a. Roflumilast in moderate-to-severechronic obstructive pulmonary disease treated with long actingbronchodilators: two randomised clinical trials. Lancet 2009;374(9691):695–703.Gamble 2003Gamble E, Grootendorst DC, Brightling CE, Troy S, Qiu Y, Zhu J,et al.Antiinflammatory effects of the phosphodiesterase-4 inhibitorcilomilast (Ariflo) in chronic obstructive pulmonary disease.American Journal of Respiratory and Critical Care Medicine 2003;168:976–82.GOLD 2005Global Initiative for Chronic Obstructive Lung Disease. Globalstrategy for the diagnosis, management, and prevention of chronicobstructive pulmonary disease (GOLD). NHLBI / WHOWorkshop Report. http://www.goldcopd.org. 2003 (updated2005).GOLD 2009Global Initiative for Chronic Obstructive Lung Disease. Globalstrategy for the diagnosis, management, and prevention of chronicobstructive pulmonary disease (GOLD). NHLBI / WHOWorkshop Report. http://www.goldcopd.com.Higgins 2008Higgins JPT, Green S, editors. Cochrane Handbook for SystematicReviews of Interventions Version 5.0.1 [updated September 2008].The Cochrane Collaboration, Available from www.cochranehandbook.org,2008.Jones 2005Jones P. St. George’s Respiratory Questionnaire: MCID. COPD2005;2:75–9.Mathers 2005Mathers C, Loncar D. Updated projections of global mortality andburden of disease, 2002-2030: data sources, methods and results.World Health Organisation 2005 October. Evidence andInformation for Policy Working Paper.RevMan 5The Nordic Cochrane Centre, The Cochrane Collaboration.Review Manager (RevMan). 5.0. Copenhagen: The NordicCochrane Centre, The Cochrane Collaboration, 2008.TORCH 2007Calverley P, Anderson J, Celli B, Ferguson GT, Jenkins C, Jones P,et al.Salmeterol and fluticasone propionate and survival in chronicobstructive pulmonary disease. New England Journal of Medicine2007;356(8):775–89.Torphy 1998Torphy T. Phosphodiesterase isozymes: molecular targets for novelantiasthma agents. American Journal of Respiratory and Critical CareMedicine 1998;157(2):351–70.Torphy 1999Torphy T, Barnette M, Underwood D, Griswold DE, ChristensenSB, Murdoch RD, et al.Ariflo (SB 207499), a second generationphosphodiesterase 4 inhibitor for the treatment of asthma andCOPD: from concept to clinic. Pulmonary Pharmacology andTherapeutics 1999;12(2):131–5.Vignola 2004Vignola A. PDE4 inhibitors in COPD--a more selective approachto treatment. Respiratory Medicine 2004;98(6):495–503.∗ Indicates the major publication for the studyPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.16

C H A R A C T E R I S T I C S O F S T U D I E SCharacteristics of included studies [ordered by study ID]Cilomilast 039MethodsParticipantsInterventionsOutcomesParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 24 weeks.Intention-to-treat analysis: Stated.1) Setting:102 centres in Canada, Mexico and the US.2) Participants: 647 (15 mg cilomilast: 431, Placebo: 216).3) Baseline characteristics: Mean age: 65 years, 62% male, mean FEV 1 % predicted of49.7%, mean smoking history of 59.9 pack years for cilomilast and 56.1 pack years forplacebo) or current smokers (44% and 47% respectively).4) Inclusion criteria: FEV 1 /FVC ≤ 0.7, FEV 1 30% to 70% with smoking history > 10pack years or current smokers.5) Exclusion criteria: active tuberculosis, lung cancer and bronchiectasis.6) Total number of participant withdrawals: 137 (32%) and 52 (24%) from treatmentand control groups respectively.Run in: 4 weeks, single blind. Placebo tablets to assess suitability.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: “The only other permitted medications for thetreatment of airways disease were stable doses of Ipratropium, via a metered-doseinhaler, and mucolytic agents.”• Short acting β2 agonist: “the short-acting ß 2 -agonist Albuterol, which wasadministered via a metered- dose inhaler, was supplied for the relief of acute respiratorysymptoms.”• Corticosteroid: None.• Long acting β2 bronchodilator: None.Primary Outcomes: Lung Function; FEV 1 , SGRQ averaged over 24 weeksSecondary Outcomes: Incidence rate of COPD exacerbations, Adverse events, FVC atthe trough, 6-min walk test, post exercise dyspnoea.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.17

Cilomilast 039(Continued)Randomised? Low risk “Eligible subjects were randomised in a 2:1ratio to receive oral cilomilast, 15 mg bid,or placebo for 24 weeks.”Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk“The primary reasons for the withdrawal ofsubjects from the study prior to randomisationwere the failure to meet inclusion/exclusion criteria (15.4%) and the presenceof adverse effects, including COPD exacerbations(8.5%). More subjects receivingcilomilast than placebo withdrew from thedouble-blind phase of study (31.8% (n =137) versus 24.1% (n = 52).”Baseline profile: Anticholingeric use High risk 54% in cilomilast; 58% placebo used Ipratropium.Baseline profile: β2 agonist use High risk 99% in cilomilast; 100% placebo used Albuterol.9% in cilomilast; 12% placeboused Salmeterol.Baseline profile: Corticosteroid use High risk 7% in cilomilast; 8% placebo used Triamcinolone.6% in cilomilast; 7% placebo usedBeclomethasoneCilomilast 042MethodsParticipantsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 24 weeks.Intention-to-treat analysis: Stated.1) Setting: 98 centres in Australia and New Zealand, Germany, Spain, South Africa andthe UK.2) Participants: 700 (15 mg Cilomilast: 474, Placebo: 226).3) Baseline characteristics: Mean age: 64.6 years, 80% male, mean FEV 1 % predictedof 49% with 5.1% reversibility, DLCO was 71% predicted, also with higher rates ofchronic bronchitis 80.1%. 45% active smokers.4) Inclusion criteria: Aged 40 to 80 years, FEV 1 /FVC ≤ 0.7, FEV 1 30% to 70% withsmoking history > 10 pack years.5) Exclusion criteria: active tuberculosis, lung cancer and bronchiectasis.6) Total number of participant withdrawals: 122 (26%) and 51 (23%) from treatmentPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.18

Cilomilast 042(Continued)and control groups respectively.InterventionsOutcomesRun in: 4 weeks, single blind with placebo.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: 2% in cilomilast; 3% placebo used Salbutamol. 3% incilomilast; 1% placebo used Ipratropium.• Short acting β2 agonist: “Albuterol MDI was used as rescue medication”.• Corticosteroid: None.• Long acting β2 bronchodilator: None.Primary Outcomes: Lung Function; FEV 1 , SGRQ averaged over 24 weeks.Secondary Outcomes: Incidence rate of COPD exacerbations, summary symptom score,FVC at the trough, 6-min walk test, post exercise dyspnoea.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Randomised? Low risk Patients were randomised in a 2:1 ratio toreceive oral cilomilast, 15 mg twice daily,or placebo for 24 weeks.Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk Total number of participants withdrawn 51(23%) placebo, 122 (26%) cilomilast, primarilydue to adverse events, of which mostwere not from COPD exacerbations.Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.19

Cilomilast 076MethodsParticipantsInterventionsOutcomesParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 12 weeks.Analysis was done per-protocol population.1) Setting: Not stated.2) Participants: 59 (15 mg Cilomilast: 29, Placebo: 30).3) Baseline characteristics: Mean age: 61 to 62 years, 81% male, 53% active smokers,mean 46 pack-years, 53% to 58% FEV 1 predicted.4) Inclusion criteria: Aged 40 to 80 years, fixed airflow obstruction, smoking history >10 pack years.5) Exclusion criteria: Not stated.6) Total number of participant withdrawals: 4 (14%) and 2 (7%) from treatment andcontrol groups respectively.Run in: 4 weeks, single blind with placebo.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: “14 of 59 used ipratropium bromide at a constantdosage (eight in the placebo group, six in the cilomilast group).”• Short acting β2 agonist: “All patients were given albuterol for use as required”.• Corticosteroid: None.• Long acting β2 bronchodilator: None.Used alongside short acting ß2 agonists (available to all) and anticholingeric drugs (offeredto 24%).Primary Outcomes: change in neutrophil percentage in induced sputum.Secondary Outcomes: FEV 1 , numbers of subepithelial CD8+ cells, CD 68+ cells, epithelialand subepithelial neutrophils.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Randomised? Low risk Randomised to a ratio of 1:1.Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?High riskPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.20

Cilomilast 076(Continued)Description of Withdrawals and Dropouts?Low risk “One patient was lost to follow-up 3days after randomisation and another withdrawnfor non-compliance 32 days afterrandomisation. Four Patients were withdrawnafter adverse events.”Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Cilomilast 091MethodsParticipantsInterventionsOutcomesParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 24 weeks.Intention-to-treat analysis: Stated.1) Setting: 110 centres in Belgium, Finland, France, Italy, the Netherlands, Norway,Portugal, Spain and the UK.2) Participants: 711 (15 mg Cilomilast: 469, Placebo: 242).3) Baseline characteristics: Mean age: 64.6 years, 86% male, mean FEV 1 % predicted of53% with 5.0% reversibility. 38% active smokers.4) Inclusion criteria: FEV 1 /FVC ≤ 0.7 with smoking history > 10 pack years.5) Exclusion criteria: active tuberculosis, lung cancer and bronchiectasis.6) Total number of participant withdrawals: 121 (26%) and 63 (26%) from treatmentand control groups respectively.Run in: 4 weeks, single blind with placebo.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: 0.9% in cilomilast; 4% placebo used Salbutamol. 1%in cilomilast; 3% placebo used Ipratropium.• Short acting β2 agonist: “Albuterol MDI was used as rescue medication”.• Corticosteroid: None.• Long acting β2 bronchodilator: None.Primary Outcomes: Lung Function; FEV 1 , SGRQ averaged over 24 weeks.Secondary Outcomes: Incidence rate of COPD exacerbations, summary symptom score,FVC at the trough, 6-min walk test, post exercise dyspnoea.NotesRisk of biasPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.21

Cilomilast 091(Continued)Bias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Randomised? Low risk “Eligible patients were randomised toreceive either SB 207499 or matchingplacebo ina ratio of 2:1 for 24 weeks.”Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk Total number of participants withdrawn 63(26%) placebo, 121 (26%) cilomilast, primarilydue to adverse events, of which mostwere not from COPD exacerbations.Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Cilomilast 103657MethodsParticipantsInterventionsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 24 weeks.Intention-to-treat analysis: Stated.1) Setting: 103 centres in the US.2) Participants: 613 (15 mg Cilomilast: 296, Placebo: 317).3) Baseline characteristics: Mean age: 63.2 years: placebo and 63.1 years; cilomilast. 47%male: placebo and 46% male: cilomilast. Mean FEV 1 % predicted not available.4) Inclusion criteria: Aged ≥ 40 years of age. FEV 1 /FVC ≤ 0.7 with smoking history >10 pack years.≤70%, post-albuterol reversibility ≤ 15% or ≤ 200 mL (or both), postalbuterolFEV 1 ≤ 70% of predicted normal and at least one COPD exacerbation withinthe 12 months prior to screening.5) Exclusion criteria: not stated.6) Total number of participant withdrawals: 105 (35%) and 76 (24%) from treatmentand control groups respectively.Run in: not stated.1) Cilomilast 15 mg twice daily.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.22

Cilomilast 103657(Continued)2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: No information available.• Short acting β2 agonist: No information available.• Corticosteroid: No information available.• Long acting β2 bronchodilator: No information available.OutcomesPrimary Objectives: Change from baseline to endpoint in trough pre-bronchodilatorFEV 1 . Change in total score of SGRQ averaged over 24 weeks.Secondary Outcomes: Changes from baseline in clinic trough FVC, time to first Level2 or Level 3 COPD exacerbation.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Randomised. No other information given.Randomised? Low risk Participants were randomised to receiveeither 15 mg of cilomilast or matchingplacebo twice daily.Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk Total number of participants withdrawn 76(24%) placebo, 105 (35%) cilomilast.Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Cilomilast 110MethodsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 12 weeks.Analysis was done per-protocol population.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.23

Cilomilast 110(Continued)ParticipantsInterventionsOutcomes1) Setting:10 centres in the US.2) Participants: 65 (15 mg Cilomilast: 31, Placebo: 34).3) Baseline characteristics: Mean age: 64.4 years: placebo and 66.1 years: cilomilast. 67%male: placebo and 84% male: cilomilast. Mean FEV 1 % predicted not available.4) Inclusion criteria: Aged 40 to 80 years. FEV 1 /FVC ≤ 0.7 with smoking history > 10pack years. Post-salbutamol reversibility ≤ 15% or 200 mL and a post-salbutamol FEV 1at least 1.0 L and between 30% and 70% of predicted.5) Exclusion criteria: not stated.6) Total number of participant withdrawals: 1 (3%) and 1 (3%) from treatment andcontrol groups respectively.Run in: not stated.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: No information available.• Short acting β2 agonist: No information available.• Corticosteroid: No information available.• Long acting β2 bronchodilator: No information available.Primary Outcomes: Change from baseline at endpoint in neutrophils as a percentage oftotal cells in induced sputum.Secondary Outcomes: FVC at the trough, sputum macrophages, eosinophils and lymphocytesas a percentage of total cells in induced sputum. Total cell counts in inducedsputum.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Randomised? Low risk All participants were randomised to receiveeither cilomilast 15 mg or matchingplacebo.Method of Randomisation described? High risk No information available.Blinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk Total number of participants withdrawn 1(3%) placebo, 1 (3%) cilomilast.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.24

Cilomilast 110(Continued)Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Cilomilast 111MethodsParticipantsInterventionsOutcomesParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 12 weeks.Intention-to-treat analysis: Stated.1) Setting: 32 centres in the US, Canada and Australia.2) Participants: 156 (15 mg Cilomilast: 79, Placebo: 77).3) Baseline characteristics: Mean age: 64.2 years: placebo and 65 years; cilomilast. 66%male: placebo and 65% male: cilomilast. Mean FEV 1 % predicted not available.4) Inclusion criteria: Aged 40 to 80 . FEV 1 /FVC ≤ 0.7 with smoking history > 10 packyears. Post-salbutamol reversibility less than or equal to 15% or 200 mL and a postsalbutamolFEV 1 at least 1.0 L and between 30% and 70% of predicted. Baseline RV(from plethysmography) ≥ 120% of predicted RV.5) Exclusion criteria: not stated.6) Total number of participant withdrawals: 15 (19%) and 14 (18%) from treatmentand control groups respectively.Run in: not stated.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: No information available.• Short acting β2 agonist: No information available.• Corticosteroid: No information available.• Long acting β2 bronchodilator: No information available.Primary Outcomes: Change from baseline to endpoint in Volume of trapped gas (D).Secondary Outcomes: Lung volume measurements, including SVC and RV, 6-min walktest and exertional dyspnoea.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.25

Cilomilast 111(Continued)Randomised? Low risk Participants were randomised to receiveeither 15 mg of cilomilast or matchingplacebo.Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk Total number of participants withdrawn 14(18%) placebo, 15 (19%) cilomilast.Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Cilomilast 121MethodsParticipantsInterventionsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 24 weeks.Intention-to-treat analysis: Stated.1) Setting: 22 centres in China.2) Participants: 1018 (15 mg Cilomilast: 678, Placebo: 340).3) Baseline characteristics: Mean age:63.9 years: placebo and 64.6 years; cilomilast. 91%male: placebo and 93% male: cilomilast. Mean FEV 1 % predicted not available.4) Inclusion criteria: Aged 40 to 75 years. FEV 1 /FVC ≤ 0.7 with smoking history > 10pack years. Documented history of COPD exacerbations each year for 3 years prior toscreening. At least one exacerbation in the last year that required oral corticosteroids orantibiotics. Post-salbutamol reversibility less than or equal to 15% or 200 mL and a postsalbutamolFEV 1 at least 1.0 L and between 25% and 70% of predicted. % predictedFRC of ≥120% from plethysmography.5) Exclusion criteria: not stated.6) Total number of participant withdrawals: 124 (18%) and 35 (10%) from treatmentand control groups respectively.Run in: not stated.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: No information available.• Short acting β2 agonist: No information available.• Corticosteroid: No information available.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.26

Cilomilast 121(Continued)• Long acting β2 bronchodilator: No information available.OutcomesPrimary Outcomes: Change from baseline to endpoint in trough pre-bronchodilatorFEV 1 .Secondary Outcomes: Time to first Level 2 or Level 3 COPD exacerbation. Level 2is defined as acute worsening of COPD that requires additional treatment or hospitaloutpatient visit. Level 3 is hospital admission for treatment. Change from baseline toendpoint RV and FRC. Change from baseline total score of SGRQ.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Randomised? Low risk Participants were randomised to a 2:1 ratiofor cilomilast 15 mg to placebo.Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk Total number of participants withdrawn 35(10%) placebo, 124 (18%) cilomilast.Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Cilomilast 156MethodsParticipantsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 24 weeks.Intention-to-treat analysis: Stated.1) Setting: 132 centres in US and Canada.2) Participants: 825 (15 mg Cilomilast: 418, Placebo: 407).3) Baseline characteristics: Mean age:64.4 years: placebo and 64.5 years; cilomilast. 62%male: placebo and 56% male: cilomilast. >.50% predicted FEV 1 for both groups.4) Inclusion criteria: Aged 40 to 80 years. FEV 1 /FVC ≤ 0.7 with smoking history > 10Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.27

Cilomilast 156(Continued)pack years. Post-salbutamol reversibility less than or equal to 15% or 200 mL and a postsalbutamolFEV 1 at least 1.0 L and between 30% and 70% of predicted.5) Exclusion criteria: not stated.6) Total number of participant withdrawals: 143 (34%) and 96 (24%) from treatmentand control groups respectively.InterventionsOutcomesRun in: not stated.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: 8.1% in cilomilast; 8.6% placebo used Salbutamol.1.7% in cilomilast; 2% placebo used Ipratropium bromide• Short acting β2 agonist: “Albuterol MDI was used as rescue medication”• Corticosteroid: None• Long acting β2 bronchodilator: NonePrimary Outcomes: Change from baseline to endpoint in trough pre-bronchodilatorFEV 1 . Change in total score of SGRQ averaged over 24 weeks.Secondary Outcomes: Post exercise breathlessness, clinic trough FVC, time to first Level2 or Level 3 COPD exacerbation.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Randomised? Low risk Participants were randomised to receiveeither 15 mg of cilomilast or matchingplacebo twice daily.Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk Total number of participants withdrawn 96(24%) placebo, 143 (34%) cilomilast.Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.28

Cilomilast 157MethodsParticipantsInterventionsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 52 weeks.Intention-to-treat analysis: Stated.1) Setting: 137 centres from 18 countries.2) Participants: 907 (15 mg Cilomilast: 455, Placebo: 452).3) Baseline characteristics: Mean age: 63.3 years: placebo and 64.6 years; cilomilast. 73%male: placebo and 78% male: cilomilast. 42% current smokers.4) Inclusion criteria: Aged 40 to 80 years. FEV 1 /FVC ≤ 0.7 with smoking history >10 pack years. Poor reversibility of airway obstruction, defined by ≤ 10% of predictednormal or≤ 200 mL (or both) increase in FEV 1 after administration of salbutamol 400mcg via MDI at screening; post-salbutamol FEV 1 of between 30% to 70% predictednormal at screening.5) Exclusion criteria: not stated.6) Total number of participant withdrawals: 167 (37%) and 121 (27%) from treatmentand control groups respectively.Run in: not stated.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: No information available.• Short acting β2 agonist: No information available.• Corticosteroid: No information available.• Long acting β2 bronchodilator: No information available.Outcomes Primary Outcomes: Mean change from baseline in trough pre-bronchodilator FEV 1averaged over 52 weeks. Incidence rate of level II (moderate) and level III (severe) COPDexacerbations during the treatment period.Secondary Outcomes: Time to first Level 2 or Level 3 COPD exacerbation. Quality ofLife determined by SGRQ.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Randomised? Low risk Particpants were randomised to receiveeither 15 mg of cilomilast or matchingplacebo twice daily.Method of Randomisation described? Unclear risk A randomisation criteria was used. Noother information available.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.29

Cilomilast 157(Continued)Blinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow riskTotal number of participants withdrawn121 (27%) placebo, 167 (37%) cilomilast.Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Cilomilast 168MethodsParticipantsInterventionsOutcomesParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 12 weeks.Intention-to-treat analysis: Not stated.1) Setting: 42 centres in the US.2) Participants: 306 (15 mg Cilomilast: 203, Placebo: 103).3) Baseline characteristics: Mean age: 64.3 years: placebo and 65.0 years; cilomilast. 64%male: placebo and 70% male: cilomilast.4) Inclusion criteria: pre-albuterol FEV 1 /FVC ≤ 0.7. Post-albuterol FEV 1 between 30%and 70% of predicted.5) Exclusion criteria: not stated.6) Total number of participant withdrawals: 61 (30%) and 14 (14%) from treatmentand control groups respectively.Run in: not stated.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: No information available.• Short acting β2 agonist: No information available.• Corticosteroid: No information available.• Long acting β2 bronchodilator: No information available.Primary Outcomes: No primary efficacy or safety analyses were defined. Descriptivestatistics of change from baseline in minimum and maximum heart rate via 24- hourHolter monitoring reported.Secondary Outcomes: No secondary efficacy or safety analyses were defined.NotesRisk of biasPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.30

Cilomilast 168(Continued)Bias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Randomised? Low risk Patients were randomised at 2:1 ratio ofcilomilast to placebo.Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk Total number of participants withdrawn 14(14%) placebo, 61 (30%) cilomilast.Baseline profile: Anticholingeric use Unclear risk No other information available.Baseline profile: β2 agonist use Unclear risk No other information available.Baseline profile: Corticosteroid use Unclear risk No other information available.Cilomilast 180MethodsParticipantsInterventionsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 18 weeks.Intention-to-treat analysis: Stated.1) Setting: 34 centres in the US, Canada and South America.2) Participants: 199 (15 mg Cilomilast: 97, Placebo: 102).3) Baseline characteristics: Mean age:.64.7 years: placebo and 63.7 years; cilomilast. 76%male: placebo and 69% male: cilomilast.4) Inclusion criteria: Age at least 40 years. FEV 1 /FVC ≤ 0.7 with smoking history > 10pack years. Baseline FEV 1 < 70% predicted normal. Moderate to severe chronic dyspnoeadefined by Baseline Dyspnoea Index focal score 7 or less, evidence of hyperinflationdefined by RFRC at least 140% of predicted. Exercise limitation, defined as peaksymptom limited VO 2 < 75%.5) Exclusion criteria: not stated.6) Total number of participant withdrawals: 24 (25%) and 13 (13%) from treatmentand control groups respectively.Run in: not stated.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medicationPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.31

Cilomilast 180(Continued)• Short acting anticholingeric: No information available.• Short acting β2 agonist: No information available.• Corticosteroid: No information available.• Long acting β2 bronchodilator: No information available.OutcomesPrimary Outcomes: Change from baseline at endpoint in RFRC.Secondary Outcomes: Change from baseline at endpoint in IC during exercise, exertionaldyspnoea as measured by the modified Borg scale.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Randomised? Low risk Participants were randomised to receiveeither 15 mg of cilomilast or matchingplacebo twice daily.Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk Total number of participants withdrawn 13(13%) placebo, 24 (25%) cilomilast.Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Cilomilast 181MethodsParticipantsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 13 weeks.Analysis was done per-protocol population.1) Setting: 27 centres in Australia, Canada, Finland, Ireland, Lithuania, Norway, Romania,Slovakia, Slovenia, South Africa, Sweden and the UK.2) Participants: 127 (15 mg Cilomilast: 65, Placebo: 62).3) Baseline characteristics: Mean age: 63.4 years: placebo and 61.4 years; cilomilast. 76%Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.32

Cilomilast 181(Continued)male: placebo and 72% male: cilomilast.4) Inclusion criteria: Aged 40 to 80 years. FEV 1 /FVC ≤ 0.7 with smoking history > 10pack years. Post- bronchodilator FEV 1 between 40% and 80% predicted normal. Poorreversibility of less than or equal to 10% or 200 mL increase in FEV 1 .5) Exclusion criteria: not stated.6) Total number of participant withdrawals: 8 (12%) and 6 (10%) from treatment andcontrol groups respectively.InterventionsOutcomesRun in: not stated.1) Cilomilast 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: No information available.• Short acting β2 agonist: No information available.• Corticosteroid: No information available.• Long acting β2 bronchodilator: No information available.Primary Outcome: Change from baseline at endpoint in CD68+ (macrophages) andCD8+ (cytotoxic T-lymphocytes) per unit area of tissue.Secondary Outcome: Change from baseline in numbers of sub-epithelial cells per unitarea in biopsy for neutrophil elastase positive (ne+) cells, CD4+, IL-8 mRNA positivecells, TNF-alpha mRNA positive cells.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Randomised. No other information available.Randomised? Low risk Participants were randomised to receiveeither 15 mg of cilomilast or matchingplacebo twice daily.Method of Randomisation described?High riskBlinding? Low risk Double blinding.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk Total number of participants withdrawn 6(10%) placebo, 8 (12%) cilomilast.Baseline profile: Anticholingeric use Unclear risk No other information available.Baseline profile: β2 agonist use Unclear risk No other information available.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.33

Cilomilast 181(Continued)Baseline profile: Corticosteroid use Unclear risk No other information available.Compton 2001MethodsParticipantsInterventionsOutcomesNotesParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 6 weeks.Intention-to-treat analysis: Stated.1) Setting: 60 centres in Austria, France, Germany, the Netherlands, and the UK.2) Participants: 424 (5 mg Cilomilast: 109, 10 mg Cilomilast: 102, 15 mg Cilomilast:107, Placebo: 106).3) Baseline characteristics: Mean age: 62 to 63 years, 75% to 78% male, mean FEV 1 %predicted of 46.8%, mean smoking history of 36 to 43 (SD 22.4) pack years.4) Inclusion criteria: FEV 1 /FVC ≤ 0.7 with smoking history > 10 pack years.5) Exclusion criteria:asthma, poorly controlled COPD needing hospital visit 6 weeksbefore study, recent COPD exacerbations or recent corticosteroid use.6) Total number of participant withdrawals: 18 (17%) and 17 (16%) from treatmentand control groups respectively.Run in: 2 weeks, single blind. Placebo tablets to assess compliance.1) Cilomilast 5 mg, 10 mg, 15 mg twice daily.2) Placebo twice daily.Concomitant medication• Short acting anticholingeric: 382 (90%) patients were given concomitant treatmentfor COPD during the study; 267 (70%)salbutamol and 115 (30%) ipratropiumbromide.• Short acting β2 agonist: Salbutamol used in 70%.• Corticosteroid: None.• Long acting β2 bronchodilator: None.Primary Outcomes: Lung Function: FEV, SGRQ.Secondary Outcomes: peak expiratory flow and FVC, the first dose effect of activetreatment on FEV 1 .Post bronchodilator results not given so pre bronchodilator values used in analysis.Risk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Described as randomised. No other informationavailable.Randomised? Low risk “Eligible patients were randomly assignedto receive a 5, 10, or 15 mg tablet of cilomilasttwice daily (morning and evening) orPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.34

Compton 2001(Continued)matching placebo for 6 weeks.”Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk“14 patients (13%) taking cilomilast 15 mghad adverse events leading to patient withdrawal,as did 12 each in the 5 and 10mg groups (11 and 12%, respectively) andeight (8%) in the placebo group.”Baseline profile: Anticholingeric use Unclear risk Information not available.Baseline profile: β2 agonist use Low risk 102 (24%) patients had been taking longacting ß 2 - agonists; e.g.. salmeterol, formoterolBaseline profile: Corticosteroid use Low risk 331 (78%) individuals had taken othermedications for their COPD, the mostcommon being inhaled steroids. 229(54%) took beclomethasone,budesonide,or fluticasone.Grootendorst 2007MethodsParticipantsCrossover group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: Two 4-week treatment periods. Washout period of 4 to 6 weeks betweentreatments.Intention-to-treat analysis: Not stated.1) Setting: Not stated.2) Participants: 383) Baseline characteristics: Mean age:63.1 years, 76% male, mean FEV 1 % predicted of61%, mean smoking history of 40 pack years. 53% current smokers.4) Inclusion criteria: Aged 45 to 74 years with history of COPD for at least 1 year.FEV 1 /FVC ≤ 0.7 with smoking history > 10 pack years. Pre-bronchodilator reversibilityless than 12% or 200 mL and a post-bronchodilator FEV 1 between 35% and 75% ofpredicted. Sputum neutrophilia (≥45% non-squamous cells) and no exacerbation orupper respiratory tract infection during the 4 weeks before the start of the study.5) Exclusion criteria: asthma, poorly controlled COPD needing hospital visit 6 weeksbefore study, recent COPD exacerbations or recent corticosteroid use.6) 6 patients withdrew after randomisation had occurred with 3 of these cases attributedto adverse effects during the Roflumilast treatment arm.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.35

Grootendorst 2007(Continued)InterventionsOutcomesNotesRun in: 2 weeks with placebo.1) Roflumilast 500 µg once daily.2) Placebo once daily.Concomitant medication• Short acting anticholingeric: “Patients withheld salbutamol, smoking and caffeinecontainingbeverages for 6 h and anticholinergic agents for 8 h before lung functionmeasurements.”• Short acting β2 agonist: As rescue medication and available to all.• Corticosteroid: None• Long acting β2 bronchodilator: None.Primary Outcomes: Reduction in the percentage of sputum neutrophils.Secondary Outcomes: FEV 1 , absolute number of sputum neutrophils, other inflammatorycell numbers and percentages in sputum, markers of activation of inflammatorycells (IL-8, neutrophil elastase, lactoferrin and ECP), markers of microvascular leakageand systemic markers of inflammation.Adverse events are grouped within categories and can not be separated for data extraction.Risk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Low risk Through a concealed computer generatedrandomisation list.Randomised? Low risk Randomised to receive either roflumilast500 µg or placebo.Method of Randomisation described? Low risk “were randomised to receive either roflumilast500 µg or placebo once daily for 4weeks by means of a concealed computergenerated randomisation list.”Blinding? Low risk Double-blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskLow risk“Six patients dropped out of the studyduring treatment of the following reasons:withdrawal of consent (n=1), exacerbationof COPD requiring additional treatment(n = 2), and adverse events to the studymedication (n = 3).”Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.36

Grootendorst 2007(Continued)Baseline profile: Corticosteroid use Unclear risk No information available.Roflumilast FK1 101MethodsParticipantsInterventionsOutcomesParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 26 weeks.Intention-to-treat analysis: Stated.1) Setting: Not stated.2) Participants: 516 (Roflumilast 250 µg: 175, Roflumilast 500 µg: 169, Placebo: 172)3) Baseline characteristics: Median age: 61 to 62 years. 72% male. Mean 38 to 63 packyears. 53% current smokers.4) Inclusion criteria: Aged 40-75. FEV 1 /FVC ≤0.7 with smoking history > 10 packyears. Reversibility < 12% or 200 mL. Post-brochodilator FEV 1 35% to 75% predicted.5) Exclusion criteria: Not stated.6) Participant withdrawals not stated.Run in: 2 weeks with placebo.1) Roflumilast 500 µg once daily.2) Roflumilast 250 µg once daily.3) Placebo once daily.Concomitant medication• Short acting anticholingeric: Allowed at a constant daily dose for those treatedbefore with anticholinergics on a constant dosage.• Short acting β2 agonist: Salbutamol was allowed as rescue medication.• Corticosteroid: None.• Long acting β2 bronchodilator: None.Primary Outcomes: Post-bronchodilator FEV 1 and FEF between 25% to 75% of vitalcapacity .Secondary Outcomes: Number of moderate or severe COPD exacerbations which requiredtreatment with oral steroids.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk Randomised. No other information available.Randomised? Low risk Patients randomised in either roflumilast250 µg, roflumilast 500 µg or placebogroups.Method of Randomisation described?High riskPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.37

Roflumilast FK1 101(Continued)Blinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskHigh riskBaseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Roflumilast FK1 103MethodsParticipantsInterventionsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 24 weeks.Intention-to-treat analysis: Stated.1) Setting: Not stated.2) Participants: 518 (Roflumilast 500 µg: 200, Placebo: 186)3) Baseline characteristics: Mean age: 60 years. 75% male. 62% current smokers. Averageof 35 pack years.4) Inclusion criteria: Aged 40 to 75 years. FEV 1 /FVC ≤ 0.7. Post-bronchodilator FEV 135% to 75% of predicted. FEV 1 reversibility ≤ 12% and ≤ 200 mL. Pre-bronchodilatorFEV 1 /FVC ≤ 70%.5) Exclusion criteria: not stated6) Participant withdrawals not stated.Run in: 2 weeks with placebo.1) Roflumilast 500 µg once daily for 24 weeks.2) Roflumilast 500 µg once daily for 12 weeks. Placebo once daily for following 12 weeks.Concomitant medication• Short acting anticholingeric: All medications were withdrawn except constant doseshort acting anti-cholinergics.• Short acting β2 agonist: As rescue medication.• Corticosteroid: None.• Long acting β2 bronchodilator: None.Used alongside short acting ß 2 agonists (available to all).Outcomes Primary Outcomes: Results for 12/24 week post bronchodilator FEV 1NotesRisk of biasPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.38

Roflumilast FK1 103(Continued)Bias Authors’ judgement Support for judgementAllocation concealment (selection bias) Unclear risk No information available.Randomised? Low risk ”After randomisation, patients receivedplacebo or roflumilast 500 µg once dailyfor 24 weeks or roflumilast 500 µg for 12weeks followed by placebo for 12 weeks.Method of Randomisation described?High riskBlinding? Low risk Double blinded.Method of Blinding described?Description of Withdrawals and Dropouts?High riskHigh riskBaseline profile: Anticholingeric use Unclear risk No further information available.Baseline profile: β2 agonist use Unclear risk No further information available.Baseline profile: Corticosteroid use Unclear risk No information available.Roflumilast M2-107MethodsParticipantsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 24 weeks.Intention-to-treat analysis: Stated.1) Setting: 159 centres in Australia, Austria, Belgium, Canada, France, Germany, Hungary,Ireland, South Africa, Spain and the UK.2) Participants: 1411 (Roflumilast 250 µg: 576, Roflumilast 500 µg: 555, Placebo: 280).3) Baseline characteristics: Median age: 64 years. 74% male. Postbronchodilator FEV 1is 51% for both groups. Average of 42 pack years. 45% current smokers.4) Inclusion criteria: Aged ≥ 40 with history of COPD > 12 months. FEV 1 < 50%predicted, FEV 1 /FVC ≤ 0.7 with smoking history > 10 pack years. Reversibility < 12%or 200 mL. Mean post-bronchodilator FEV 1 30% to 80% predicted.5) Exclusion criteria: Asthma, lung cancer or bronchiectasis, long term oxygen treatment,recent exacerbation that required a course of systemic corticosteroids, emergency roomtreatment or hospital admission within 4 weeks before the run-in period.6) Total number of participant withdrawals: 124 (22%) and 32 (11%) from treatmentand control groups respectively.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.39

Roflumilast M2-107(Continued)InterventionsOutcomesRun in: 4 weeks with placebo.1) Roflumilast 500 µg once daily.2) Roflumilast 250 µg once daily.3) Placebo once daily.Concomitant medication• Short acting anticholingeric: Used at a constant daily dose.• Short acting β2 agonist: Salbutamol as rescue medication.• Corticosteroid: None.• Long acting β2 bronchodilator: None.Primary Outcomes: Post-bronchodilator FEV 1 and SGRQ total score.Secondary Outcomes: Change from baseline in pre-bronchodilator FEV 1 , post-bronchodilatorFVC, post bronchodilator FEV in 6 seconds and FVC, FEF rate between25% to 75% of vital capacity and number of moderate or severe COPD exacerbations.Notes There is inconsistency in the quoting of statistical errors. Within the text and Table 2,data is quoted as “least squares means and SD”, however in Figures 2 and 3, SE bars areshown. It is more likely that the results represent SE and not SD.Risk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Low risk “No person involved in data analysis hadknowledge of the randomisation sequence”Randomised? Low risk “Treatment was assigned by the investigatorswith sequential study numbers accordingto a block randomisation list in ratiosof 2:2:1 (Roflumilast 250 µg, Roflumilast500 µg, Placebo).”Method of Randomisation described? Low risk “The randomisation sequence was generatedby ALTANA Pharma AG in a blindedmanner;”Blinding? Low risk Double blinded.Method of Blinding described? Low risk “Medication boxes were labelled with thestudy protocol number, randomisationnumber, and visit code; coding preventedthe investigator and people at the studycentre from knowing which medicationwas given.”Description of Withdrawals and Dropouts?Low risk100 patients discontinued from study inthe roflumilast 250 µg, 124 from the roflumilast500 µg and 32 from the placeboPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.40

Roflumilast M2-107(Continued)group.Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.Roflumilast M2-112MethodsParticipantsInterventionsOutcomesParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 52 weeks.Intention-to-treat analysis: Stated.1) Setting:159 centres in 14 countries.2) Participants: 1513 (Roflumilast 500 µg: 760, Placebo: 753).3) Baseline characteristics: Severe COPD according to GOLD criteria stages III and IV.Mean age: 65 years. 75% male.4) Inclusion criteria: Aged ≥ 40 years. Post-bronchodilator FEV 1 < 50% predicted.Reversibility < 15%. Mean post-bronchodilator FEV 1 41%. FEV 1 /FVC ≤ 0.7 withsmoking history > 10 pack years. 37% current smokers, 63% ex-smokers; average 44pack years.5) Exclusion criteria: history of asthma, lung cancer or bronchiectasis, need for long-termoxygen therapy, knownα 1 antitrypsin deficiency or clinically significant cardiopulmonaryco-morbidity.6) Total number of participant withdrawals: 217 (29%) and 163 (22%) from treatmentand control groups respectively.Run in: 4 weeks with placebo.1) Roflumilast 500 µg once daily.2) Placebo once daily.Concomitant medication• Short acting anticholingeric: 891 patients on short acting anticholinergics.• Short acting β2 agonist: Salbutamol as rescue medication.• Corticosteroid: 943 patients continued corticosteroid use.• Long acting β2 bronchodilator: None.Used alongside corticosteroids, anticholinergics and rescue short acting ß 2 agonists 54%overall (available to all).Primary Outcomes: Change from baseline to endpoint in post-bronchodilator FEV 1 andthe number of moderate or severe exacerbations per patient per year.Secondary Outcomes: Change from baseline in SGRQ total score, change from baselinein prebronchial FEV 1 , post bronchodilator FEV in 6 seconds and FVC, FEF rate between25% to 75% of vital capacity and number of moderate or severe COPD exacerbationsrequiring systemic corticosteroid treatment per patient per year.NotesPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.41

Roflumilast M2-112(Continued)Risk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Low risk “Each study participant who qualified wasassigned a number in sequential order.Code labelling prevented the investigatorand the patient from knowing which drugwas administered.”Randomised? Low risk “randomised (1:1).”Method of Randomisation described? Low risk “The randomization list was generated usinga multiplicative congruential pseudorandomnumber generator (program RAN-DOM, based on Fishman and Moore).”Blinding? Low risk Double blinded.Method of Blinding described? Low risk “There was a stratification of patients accordingto smoking status(current smokers/ex-smokers)and treatment with inhaledcorticosteroids(yes/no).”Description of Withdrawals and Dropouts?Low risk“Over 70% of patients completed thestudy. The reasons for withdrawal weresimilar between groups except for adverseevents, which occurred more frequentlywith roflumilast.”“Withdrawal due to COPD exacerbationswas reported in 3.5 and 3.2% of patientsin roflumilast and placebo groups, respectively.”Baseline profile: Anticholingeric use Low risk 739 patients used anticholinergics.Baseline profile: β2 agonist use Low risk 820 patients on short acting β 2 - agonists.Baseline profile: Corticosteroid use Low risk 727 on Beclomethasone dipropionate2000µg or less.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.42

Roflumilast M2-124MethodsParticipantsInterventionsOutcomesNotesParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 52 weeks.Intention-to-treat analysis: Stated.1) Setting: 246 centres in ten countries.2) Participants: 1513 (Roflumilast 500 µg: 760, Placebo: 753).3) Baseline characteristics: Mean age: 64 years. 71% male. Post bronchodilator FEV 137.6% predicted. Average of 47 pack years. 48% current smokers.4) Inclusion criteria: former or current smokers with at least a 20 pack-year history.Aged ≥ 40 years. Post-bronchodilator FEV 1 /FVC ≤ 0.7. Chronic cough and sputumproduction. Post-bronchodilator FEV 1 < 50% predicted. At least one recorded COPDexacerbation requiring systemic glucocorticosteroids or treatment in hospital in previousyear.5) Exclusion criteria: available in the online web appendix (pg11).6) Total number of participant withdrawals: 264 (34%) and 234 (31%) from treatmentand control groups respectively.Run in: 4 weeks with placebo.1) Roflumilast 500 µg once daily.2) Placebo once daily.Concomitant medication• Short acting anticholingeric: 31% of those in the roflumilast group and 32% onplacebo.• Short acting β2 agonist: ”Patients could use short acting β 2 agonists as needed“.• Corticosteroid: None.• Long acting β2 bronchodilator: ”Eligible patients were stratified according to theiruse of long acting β2 agonists and smoking status.“ Roflumilast 49%, placebo 51%”.Primary Outcomes: Mean change in pre-bronchodilator FEV 1 from baseline to eachpost-randomisation visit during the treatment period. Mean rate of COPD exacerbationsrequiring oral or parenteral glucocorticosteroids or requiring hospitalisation or leadingto death, per patient per year.Secondary Outcomes: Mean change in post-bronchodilator FEV 1 from baseline to eachpost-randomisation visit during the treatment period. Time to mortality due to any reason.Natural log-transformed CRP (mg/L), Mean TDI Focal score during the treatmentperiod.Adverse event data is pooled with numbers from study M2-125 that followed an identicalstudy design.Risk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Low risk All individuals involved in the studies wereunaware of treatment assignment.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.43

Roflumilast M2-124(Continued)Randomised? Low risk “Randomly assigned to oral roflumilast 500µg once daily or placebo.”Method of Randomisation described? Low risk “The sponsor generated a randomisationlist of patient random numbers using apseudorandom number generator. The investigatorused an automated, interactivevoice response system to randomly assignpatients.”Blinding? Low risk Double blinded.Method of Blinding described? Low risk “All individuals involved in the studies wereunaware of treatment assignment. Tabletswere identical in appearance. The investigatoror anyone at the study site was preventedfrom knowing the allocation sequencewith code labelling.”Description of Withdrawals and Dropouts?Low risk264 patients discontinued from study inthe roflumilast group and 234 discontinuedfrom the placebo group.Baseline profile: Anticholingeric use Unclear risk No other information available.Baseline profile: β2 agonist use Unclear risk No other information available.Baseline profile: Corticosteroid use High risk Pretreatment of 44% in both roflumilastand placebo groups.Roflumilast M2-124+M2-125MethodsAs described in separate studies above and below.ParticipantsInterventionsOutcomesNotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias)Low riskPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.44

Roflumilast M2-124+M2-125(Continued)Randomised?Low riskMethod of Randomisation described?Blinding?Method of Blinding described?Description of Withdrawals and Dropouts?Baseline profile: Anticholingeric useBaseline profile: β2 agonist useBaseline profile: Corticosteroid useLow riskLow riskLow riskLow riskLow riskLow riskHigh riskRoflumilast M2-125MethodsParticipantsInterventionsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 52 weeks.Intention-to-treat analysis: Stated.1) Setting: 221 centres in eight countries.2) Participants: 1571 (Roflumilast 500 µg: 773, Placebo: 798)3) Baseline characteristics: Mean age: 64 years. 80% male. Average of 48 pack years. 35%current smokers.4) Inclusion criteria: former or current smokers with at least a 20 pack-year history.Aged ≥ 40 years. Post-bronchodilator FEV 1 /FVC ≤ 0.7. Chronic cough and sputumproduction. Post-bronchodilator FEV 1 < 50% predicted. At least one recorded COPDexacerbation requiring systemic glucocorticosteroids or treatment in hospital in previousyear.5) Exclusion criteria: available in the online web appendix (pg11).6) Total number of participant withdrawals: 246 (32%) and 248 (31%) from treatmentand control groups respectively.Run in: 4 weeks with placebo.1) Roflumilast 500 µg once daily.2) Placebo once daily.Concomitant medication• Short acting anticholingeric: 38% of those in the roflumilast group and 41% onplacebo.• Short acting β2 agonist: ”Patients could use short acting β 2 agonists as needed“.• Corticosteroid: None.• Long acting β2 bronchodilator: ”Eligible patients were stratified according to theiruse of long acting β2 agonists and smoking status.“ Roflumilast 48%, placebo 51%”Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.45

Roflumilast M2-125(Continued)OutcomesNotesPrimary Outcomes: Mean change in pre-bronchodilator FEV 1 from baseline to eachpost-randomisation visit during the treatment period. Mean rate of COPD exacerbationsrequiring oral or parenteral glucocorticosteroids or requiring hospitalisation or leadingto death, per patient per year.Secondary Outcomes: Mean change in post-bronchodilator FEV 1 from baseline to eachpost-randomisation visit during the treatment period. Time to mortality due to any reason.Natural log-transformed CRP (mg/L), Mean TDI Focal score during the treatmentperiod.Adverse event data is pooled with numbers from study M2-124 that followed an identicalstudy design.Risk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Low risk All individuals involved in the studies wereunaware of treatment assignment.Randomised? Low risk “Randomly assigned to oral roflumilast 500µg once daily or placebo.”Method of Randomisation described? Low risk “The sponsor generated a randomisationlist of patient random numbers using apseudorandom number generator. The investigatorused an automated, interactivevoice response system to randomly assignpatients.”Blinding? Low risk Double blinded.Method of Blinding described? Low risk “All individuals involved in the studies wereunaware of treatment assignment. Tabletswere identical in appearance. The investigatoror anyone at the study site was preventedfrom knowing the allocation sequencewith code labelling.”Description of Withdrawals and Dropouts?Low risk246 patients discontinued from study inthe roflumilast group and 248 discontinuedfrom the placebo group.Baseline profile: Anticholingeric use Low risk No other information available.Baseline profile: β2 agonist use Low risk No other information available.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.46

Roflumilast M2-125(Continued)Baseline profile: Corticosteroid use High risk Pretreatment of 40% in both roflumilastand placebo groupsRoflumilast M2-127MethodsParticipantsInterventionsOutcomesParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 24 weeks.Intention-to-treat analysis: Stated.1) Setting: 135 centres in ten countries.2) Participants: 1221 (Roflumilast 500 µg: 467, Placebo: 468).3) Baseline characteristics: Mean age: 65 years. 71% male. Post bronchodilator FEV 154.7 and 55.3% predicted (roflumilast and placebo). Average of 43 pack years. 39%current smokers.4) Inclusion criteria: former or current smokers with ( ≥ 1 year smoking cessation) andat least a 10 pack-year history. Aged ≥ 40 years. Post-bronchodilator FEV 1 /FVC ≤ 0.7.Post-bronchodilator FEV 1 40%- 70% predicted. Partial reversibility to albuterol withincrease from baseline FEV 1 of ≤12% or 200ml.5) Exclusion criteria: available in the online web appendix (pg10).6) Total number of participant withdrawals: 107 (23%) and 82 (18%) from treatmentand control groups respectively.Run in: 4 weeks with placebo once a day.1) Roflumilast 500 µg and salmeterol once daily.2) Placebo once daily.Concomitant medication• Short acting anticholingeric: None.• Short acting β2 agonist: Patients used short acting β 2 as rescue medication.• Corticosteroid: None.• Long acting β2 bronchodilator: None.Primary Outcomes: Change in mean pre-bronchodilator FEV 1 from baseline to eachpost-randomisation visit.Secondary Outcomes: Post-bronchodilator FEV 1 and FVC, TDI score, SOBQ, rate ofCOPD exacerbations, and use of rescue medication.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Low risk All individuals involved in the studies wereunaware of treatment assignment.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.47

Roflumilast M2-127(Continued)Randomised? Low risk “Randomly assigned to oral roflumilast 500µg once daily or placebo.”Method of Randomisation described? Low risk “The sponsor generated a randomisationlist of patient random numbers using apseudorandom number generator. The investigatorused an automated, interactivevoice response system to randomly assignpatients.”Blinding? Low risk Double blinded.Method of Blinding described? Low risk “All individuals involved in the studies wereunaware of treatment assignment. The investigatoror anyone at the study site wasprevented from knowing the allocation sequencewith code labelling.Tablets wereidentical in appearance.”Description of Withdrawals and Dropouts?Low risk107 patients discontinued from study inthe roflumilast group and 82 discontinuedfrom the placebo group.Baseline profile: Anticholingeric use Unclear risk No other information available.Baseline profile: β2 agonist use Unclear risk No other information available.Baseline profile: Corticosteroid use Unclear risk No other information available.Roflumilast M2-128MethodsParticipantsParallel group study.Randomisation: Randomised, double blind, placebo-controlled trial.Trial duration: 24 weeks.Intention-to-treat analysis: Stated.1) Setting: 85 centres in seven countries.2) Participants: 910 (Roflumilast 500 µg: 372, Placebo: 372)3) Baseline characteristics: Mean age: 64 years. 71% male. Post bronchodilator FEV 156.0 and 56.2% predicted (roflumilast and placebo). Average of 44 pack years. 40%current smokers.4) Inclusion criteria: former or current smokers with ( ≥ 1 year smoking cessation) andat least a 10 pack-year history. Aged ≥ 40 years. Post-bronchodilator FEV 1 /FVC ≤ 0.7.Post-bronchodilator FEV 1 40% to 70% predicted. Partial reversibility to albuterol withincrease from baseline FEV 1 of ≤ 12% or 200 mL.5) Exclusion criteria: available in the online web appendix (pg10).6) Total number of participant withdrawals: 62 (17%) and 39 (11%) from treatmentPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.48

Roflumilast M2-128(Continued)and control groups respectively.InterventionsOutcomesRun in: 4 weeks with placebo once a day.1) Roflumilast 500 µg and tiotropium once daily.2) Placebo once daily.Concomitant medication• Short acting anticholingeric: None.• Short acting β2 agonist: Patients used short acting β 2 as rescue medication.• Corticosteroid: None.• Long acting β2 bronchodilator: NonePrimary Outcomes: Change in mean pre-bronchodilator FEV 1 from baseline to eachpost-randomisation visit.Secondary Outcomes: Post-bronchodilator FEV 1 and FVC, TDI score, SOBQ, rate ofCOPD exacerbations, and use of rescue medication.NotesRisk of biasBias Authors’ judgement Support for judgementAllocation concealment (selection bias) Low risk All individuals involved in the studies wereunaware of treatment assignment.Randomised? Low risk “Randomly assigned to oral roflumilast 500µg once daily or placebo.”Method of Randomisation described? Low risk “The sponsor generated a randomisationlist of patient random numbers using apseudorandom number generator. The investigatorused an automated, interactivevoice response system to randomly assignpatients.”Blinding? Low risk Double blinded.Method of Blinding described? Low risk “All individuals involved in the studies wereunaware of treatment assignment. The investigatoror anyone at the study site wasprevented from knowing the allocation sequencewith code labelling.Tablets wereidentical in appearance.”Description of Withdrawals and Dropouts?Low risk62 patients discontinued from study inthe roflumilast group and 39 discontinuedfrom the placebo group.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.49

Roflumilast M2-128(Continued)Baseline profile: Anticholingeric use Unclear risk No information available.Baseline profile: β2 agonist use Unclear risk No information available.Baseline profile: Corticosteroid use Unclear risk No information available.COPD: chronic obstructive pulmonary diseaseDLCO: diffusing capacity of the lung for carbon monoxideFEF: forced expiratory flowFEV 1 : forced expiratory volume in 1 secondFRC: functional residual capacityFVC: forced vital capacityGOLD: Global Initiative for Chronic Obstructive Lung DiseaseIC: inspiratory capacityL: litreMDI: metered dose inhalermL: millilitreRFRC: resting functional residue capacityRV: residual volumeSD: standard deviationSE: standard errorSGRQ: St Georges Respiratory QuestionnaireSOBQ: Shortness of Breath QuestionnaireSVC: slow vital capacityTDI: transition dyspnea index≤: less than or equal to≥: more than or equal toCharacteristics of excluded studies [ordered by study ID]StudyBorker 2003Ferguson 2003Fischer 2003Grootendorst 2001Grootendorst 2002Grootendorst 2003Reason for exclusionInsufficient data. Only RR of QOL improvement provided.Integrated results from four 24-week Cilomilast trials.Analysis focused on patients with a baseline SGRQ score of ≥ the median SGRQ score only.Endpoint: First dose bronchodilator effects only.Treatment Bayer BAY 19-8004. 11 patients only 1 week in duration.Endpoint: First dose bronchodilator effects only.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.50

(Continued)GSK256066Kelsen 2002Knobil 2003Lim 2004Nieman 1999Pascoe 2007Reisner 2002Reisner 2003Rennard 2008SB207499/040SB207499/041Song 2005Spencer 2002Vestbo 2007Vestbo 2009Wang 2005Phase II trial. No primary outcome measure investigating lung function. Only one trial to date.No study ID or group numbers identified.No SD or SE given.Combining results from two pivotal European phase III Cilomilast trials.Study 038. Insufficient data available for changes in lung function and exacerbation rates.Treatment QAK423 (Novartis), discontinued. Only one trial available.Combined results. Individual studies already included in review.Combined results. Individual studies already included in review.SE not provided for exacerbation rates.Open label study. Males or females with COPD who successfully completed study 042 or 091 where participantsreceived cilomilast 15 mg twice daily or placebo for 24 weeks in study 042 and 26 weeks in study 091 withouttolerability problems. Concomitant COPD medication use allowed, given placebo or placebo/ariflo duringstudy period.Open label study. Males or females with COPD who successfully completed study 039 where participantsreceived cilomilast 15 mg twice daily or placebo for 24 weeks without tolerability problems. ConcomitantCOPD medication use allowed, given placebo or placebo/ariflo during study period.Although quoted as significant, mean and SD figures not provided.No study ID or group numbers identified.Treatment UK-500,001 (Pfizer). Discontinued.Treatment UK-500,001 (Pfizer). Discontinued.Although quoted as significant, mean and SD figures not provided.COPD: chronic obstructive pulmonary diseaseRR: risk ratioQOL: quality of lifeSD: standard deviationSE: standard error≥: more than or equal toPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.51

D A T AA N D A N A L Y S E SComparison 1.Lung functionOutcome or subgroup titleNo. ofstudiesNo. ofparticipants Statistical method Effect size1 FEV 1 (by drug) 19 14307 Mean Difference (IV, Fixed, 95% CI) 45.59 [39.15, 52.03]1.1 Roflumilast 500 µg 9 7782 Mean Difference (IV, Fixed, 95% CI) 54.32 [44.39, 64.25]1.2 Roflumilast 250 µg 2 1203 Mean Difference (IV, Fixed, 95% CI) 55.03 [22.10, 87.96]1.3 Cilomilast 15 mg 10 5322 Mean Difference (IV, Fixed, 95% CI) 38.15 [29.41, 46.90]2 FEV 1 (by mean COPD severity) 13 9972 Mean Difference (IV, Fixed, 95% CI) 49.89 [41.45, 58.33]2.1 GOLD stage I + II (FEV1 7 4016 Mean Difference (IV, Fixed, 95% CI) 50.74 [37.95, 63.52]≥ 50% predicted)2.2 GOLD stage III + IV 6 5956 Mean Difference (IV, Fixed, 95% CI) 49.24 [38.01, 60.48](FEV1 < 50% predicted)3 FEV 1 (Roflumilast 500 µg by 7 7055 Mean Difference (IV, Fixed, 95% CI) 55.08 [44.79, 65.37]mean COPD severity)3.1 GOLD stage I + II (FEV 1 4 2556 Mean Difference (IV, Fixed, 95% CI) 68.03 [51.54, 84.52]≥ 50% predicted)3.2 GOLD stage III + IV 3 4499 Mean Difference (IV, Fixed, 95% CI) 46.83 [33.66, 59.99](FEV 1 < 50% predicted)4 FEV 1 (by study duration) 19 13098 Mean Difference (IV, Fixed, 95% CI) 45.23 [38.66, 51.79]4.1 Duration < 12 weeks 4 379 Mean Difference (IV, Fixed, 95% CI) 99.99 [68.42,131.57]4.2 Duration 24-26 weeks 11 7419 Mean Difference (IV, Fixed, 95% CI) 42.03 [33.86, 50.20]4.3 Duration 52 weeks 4 5300 Mean Difference (IV, Fixed, 95% CI) 44.26 [32.49, 56.03]5 FEV 1 (additional medication) 19 13098 Mean Difference (IV, Fixed, 95% CI) 45.23 [38.66, 51.79]5.1 Long acting2 1645 Mean Difference (IV, Fixed, 95% CI) 60.52 [40.57, 80.46]bronchodilator5.2 Corticosteroids 1 1513 Mean Difference (IV, Fixed, 95% CI) 36.0 [5.52, 66.48]5.3 PDE4i treatment only 16 9940 Mean Difference (IV, Fixed, 95% CI) 43.77 [36.63, 50.91]6 FEV 1 (published vs.19 13098 Mean Difference (IV, Fixed, 95% CI) 45.45 [38.89, 52.02]unpublished)6.1 Published 12 8624 Mean Difference (IV, Fixed, 95% CI) 55.85 [46.66, 65.03]6.2 Unpublished 7 4474 Mean Difference (IV, Fixed, 95% CI) 34.60 [25.21, 43.98]7 FEV 1 (random effects model) 19 14301 Mean Difference (IV, Random, 95% CI) 48.49 [38.13, 58.85]8 FEV 1 (Roflumilast 500 µg vs 2 1475 Mean Difference (IV, Fixed, 95% CI) 22.75 [-6.08, 51.58]250 µg)9 FVC 12 10470 Mean Difference (IV, Fixed, 95% CI) 82.67 [66.10, 99.24]10 PEF 4 3854 Mean Difference (IV, Fixed, 95% CI) 6.10 [3.45, 8.75]10.1 Roflumilast 500 µg 3 3294 Mean Difference (IV, Fixed, 95% CI) 4.92 [2.13, 7.70]10.2 Roflumilast 250 µg 1 347 Mean Difference (IV, Fixed, 95% CI) 7.0 [-4.05, 18.05]10.3 Cilomilast 15 mg 1 213 Mean Difference (IV, Fixed, 95% CI) 34.0 [20.14, 47.86]11 FEV 1 (Evidence quality) 19 13104 Mean Difference (IV, Fixed, 95% CI) 45.22 [38.65, 51.78]11.1 Low risk of bias 7 7055 Mean Difference (IV, Fixed, 95% CI) 55.08 [44.79, 65.37]11.2 Unclear risk of bias 12 6049 Mean Difference (IV, Fixed, 95% CI) 38.45 [29.93, 46.97]Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.52

Comparison 2.Quality of lifeOutcome or subgroup titleNo. ofstudiesNo. ofparticipants Statistical method Effect size1 SGRQ total score 10 7898 Mean Difference (IV, Fixed, 95% CI) -1.04 [-1.66, -0.41]1.1 Roflumilast 500 µg 2 2348 Mean Difference (IV, Fixed, 95% CI) -0.69 [-2.07, 0.69]1.2 Roflumilast 250 µg 1 856 Mean Difference (IV, Fixed, 95% CI) -1.60 [-3.56, 0.36]1.3 Cilomilast 15 mg 8 4694 Mean Difference (IV, Fixed, 95% CI) -1.06 [-1.81, -0.31]2 SGRQ total score (by mean 7 4824 Mean Difference (IV, Fixed, 95% CI) -1.53 [-2.36, -0.70]COPD severity)2.1 GOLD stage I and II 3 2042 Mean Difference (IV, Fixed, 95% CI) -1.62 [-2.80, -0.44]2.2 GOLD stage III and IV 4 2782 Mean Difference (IV, Fixed, 95% CI) -1.44 [-2.61, -0.27]3 SGRQ total score (by published 10 7042 Mean Difference (IV, Fixed, 95% CI) -0.97 [-1.63, -0.31]vs unpublished)3.1 Published 4 3052 Mean Difference (IV, Fixed, 95% CI) -1.93 [-3.02, -0.83]3.2 Unpublished 6 3990 Mean Difference (IV, Fixed, 95% CI) -0.43 [-1.26, 0.40]4 SGRQ total score (by duration) 10 7042 Mean Difference (IV, Fixed, 95% CI) -0.97 [-1.63, -0.31]4.1 Duration < 12 weeks 1 213 Mean Difference (IV, Fixed, 95% CI) -3.90 [-7.50, -0.30]4.2 Duration 24-26 weeks 7 4600 Mean Difference (IV, Fixed, 95% CI) -1.18 [-1.94, -0.42]4.3 Duration 52 weeks 2 2229 Mean Difference (IV, Fixed, 95% CI) 0.26 [-1.18, 1.69]Comparison 3.ExacerbationsOutcome or subgroup titleNo. ofstudiesNo. ofparticipants Statistical method Effect size1 No. of subjects (by drug) 17 13841 Odds Ratio (M-H, Fixed, 95% CI) 0.78 [0.72, 0.85]1.1 Roflumilast 500 µg 7 7457 Odds Ratio (M-H, Fixed, 95% CI) 0.77 [0.69, 0.87]1.2 Roflumilast 250 µg 1 856 Odds Ratio (M-H, Fixed, 95% CI) 1.06 [0.78, 1.43]1.3 Cilomilast 10 5528 Odds Ratio (M-H, Fixed, 95% CI) 0.76 [0.67, 0.85]2 Exacerbation rate 1 1691 Mean Difference (IV, Fixed, 95% CI) -0.25 [-0.48, -0.02]2.1 Roflumilast 500 µg 1 835 Mean Difference (IV, Fixed, 95% CI) -0.38 [-0.70, -0.06]2.2 Roflumilast 250 µg 1 856 Mean Difference (IV, Fixed, 95% CI) -0.10 [-0.44, 0.24]3 Exacerbation rate (Inverse 6 Rate Ratio (Fixed, 95% CI) 0.94 [0.88, 1.00]variance)3.1 Roflumilast 5 Rate Ratio (Fixed, 95% CI) 0.93 [0.87, 1.00]3.2 Cilomilast 1 Rate Ratio (Fixed, 95% CI) 0.98 [0.80, 1.20]4 Exacerbation rate (Roflumilast500 µg vs 250 µg)1 1131 Mean Difference (IV, Fixed, 95% CI) -0.28 [-0.53, -0.03]Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.53

Comparison 4.Symptom scoreOutcome or subgroup titleNo. ofstudiesNo. ofparticipants Statistical method Effect size1 SGRQ symptom score 2 1048 Mean Difference (IV, Fixed, 95% CI) -1.53 [-4.11, 1.06]1.1 Roflumilast 1 835 Mean Difference (IV, Fixed, 95% CI) 1.00 [-3.78, 1.78]1.2 Cilomilast 1 213 Mean Difference (IV, Fixed, 95% CI) -4.80 [-11.73, 2.13]2 Borg Scale 6 2860 Mean Difference (IV, Fixed, 95% CI) -0.19 [-0.33, -0.05]2.1 Cilomilast 6 2860 Mean Difference (IV, Fixed, 95% CI) -0.19 [-0.33, -0.05]3 Summary symptom score 3 1899 Mean Difference (IV, Fixed, 95% CI) -0.06 [-0.25, 0.13]3.1 Cilomilast 3 1899 Mean Difference (IV, Fixed, 95% CI) -0.06 [-0.25, 0.13]4 Shortness of breath questionnaire 2 1633 Mean Difference (IV, Fixed, 95% CI) -1.09 [-2.47, 0.28]Comparison 5.Exercise toleranceOutcome or subgroup titleNo. ofstudiesNo. ofparticipants Statistical method Effect size1 6-minute walk test 4 1948 Mean Difference (IV, Fixed, 95% CI) 1.92 [-7.58, 11.41]1.1 Cilomilast 4 1948 Mean Difference (IV, Fixed, 95% CI) 1.92 [-7.58, 11.41]Comparison 6.Adverse effectsOutcome or subgroup titleNo. ofstudiesNo. ofparticipants Statistical method Effect size1 No of patients experiencing an 21 15277 Odds Ratio (M-H, Fixed, 95% CI) 1.20 [1.11, 1.28]adverse effect1.1 Roflumilast 500 µg 7 7532 Odds Ratio (M-H, Fixed, 95% CI) 1.20 [1.09, 1.32]1.2 Roflumilast 250 µg 2 1203 Odds Ratio (M-H, Fixed, 95% CI) 1.12 [0.88, 1.42]1.3 Cilomilast 15 mg 14 6542 Odds Ratio (M-H, Fixed, 95% CI) 1.21 [1.08, 1.36]2 No of patients experiencing an 2 1475 Odds Ratio (M-H, Fixed, 95% CI) 1.01 [0.82, 1.25]adverse event (Roflumilast 500µg vs 250 µg)3 Diarrhoea 20 13658 Odds Ratio (M-H, Fixed, 95% CI) 2.81 [2.42, 3.26]3.1 Roflumilast 6 7116 Odds Ratio (M-H, Fixed, 95% CI) 3.41 [2.69, 4.33]3.2 Cilomilast 14 6542 Odds Ratio (M-H, Fixed, 95% CI) 2.47 [2.05, 2.98]4 Nausea 20 14514 Odds Ratio (M-H, Fixed, 95% CI) 4.02 [3.35, 4.84]4.1 Roflumilast 500 µg 6 7116 Odds Ratio (M-H, Fixed, 95% CI) 3.32 [2.38, 4.61]4.2 Roflumilast 250 µg 1 856 Odds Ratio (M-H, Fixed, 95% CI) 3.97 [0.91, 17.39]4.3 Cilomilast 15 mg 14 6542 Odds Ratio (M-H, Fixed, 95% CI) 4.37 [3.49, 5.47]5 Headache 18 13107 Odds Ratio (M-H, Fixed, 95% CI) 1.60 [1.35, 1.89]5.1 Roflumilast 500 µg 7 7457 Odds Ratio (M-H, Fixed, 95% CI) 2.53 [1.83, 3.49]5.2 Roflumilast 250 µg 1 347 Odds Ratio (M-H, Fixed, 95% CI) 0.98 [0.24, 3.99]5.3 Cilomilast 15 mg 11 5303 Odds Ratio (M-H, Fixed, 95% CI) 1.32 [1.08, 1.62]Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.54

6 Vomiting 11 5828 Odds Ratio (M-H, Fixed, 95% CI) 4.01 [2.80, 5.74]6.1 Roflumilast 1 835 Odds Ratio (M-H, Fixed, 95% CI) 1.52 [0.06, 37.37]6.2 Cilomilast 10 4993 Odds Ratio (M-H, Fixed, 95% CI) 4.06 [2.83, 5.82]7 Dyspepsia 12 5621 Odds Ratio (M-H, Fixed, 95% CI) 3.13 [2.30, 4.27]7.1 Cilomilast 12 5621 Odds Ratio (M-H, Fixed, 95% CI) 3.13 [2.30, 4.27]8 Abdominal pain 11 5604 Odds Ratio (M-H, Fixed, 95% CI) 1.97 [1.55, 2.49]8.1 Cilomilast 11 5604 Odds Ratio (M-H, Fixed, 95% CI) 1.97 [1.55, 2.49]9 Weight Loss 4 4768 Odds Ratio (M-H, Fixed, 95% CI) 4.62 [3.38, 6.31]9.1 Roflumilast 4 4768 Odds Ratio (M-H, Fixed, 95% CI) 4.62 [3.38, 6.31]10 Influenza-like symptoms 8 7935 Odds Ratio (M-H, Fixed, 95% CI) 1.12 [0.86, 1.47]10.1 Roflumilast 500 µg 6 6622 Odds Ratio (M-H, Fixed, 95% CI) 1.17 [0.87, 1.57]10.2 Roflumilast 250 µg 1 347 Odds Ratio (M-H, Fixed, 95% CI) 1.98 [0.18, 22.00]10.3 Cilomilast 15 mg 2 966 Odds Ratio (M-H, Fixed, 95% CI) 0.88 [0.44, 1.75]11 Upper Respiratory Tract 16 11427 Odds Ratio (M-H, Fixed, 95% CI) 0.88 [0.75, 1.02]Infection11.1 Roflumilast 500 µg 6 5944 Odds Ratio (M-H, Fixed, 95% CI) 0.82 [0.63, 1.07]11.2 Roflumilast 250 µg 2 1203 Odds Ratio (M-H, Fixed, 95% CI) 0.84 [0.54, 1.31]11.3 Cilomilast 15 mg 10 4280 Odds Ratio (M-H, Fixed, 95% CI) 0.92 [0.75, 1.13]12 Withdrawals due to adverse 22 14429 Odds Ratio (M-H, Fixed, 95% CI) 1.76 [1.58, 1.96]effects12.1 Roflumilast 500 µg 8 7537 Odds Ratio (M-H, Fixed, 95% CI) 1.66 [1.43, 1.93]12.2 Roflumilast 250 µg 1 347 Odds Ratio (M-H, Fixed, 95% CI) 1.3 [0.55, 3.05]12.3 Cilomilast 15 mg 14 6545 Odds Ratio (M-H, Fixed, 95% CI) 1.90 [1.61, 2.24]13 Non fatal serious adverse events 19 12748 Odds Ratio (M-H, Fixed, 95% CI) 0.93 [0.83, 1.04]13.1 Roflumilast 500 µg 5 5856 Odds Ratio (M-H, Fixed, 95% CI) 0.95 [0.82, 1.09]13.2 Roflumilast 250 µg 1 347 Odds Ratio (M-H, Fixed, 95% CI) 1.27 [0.56, 2.89]13.3 Cilomilast 15 mg 14 6545 Odds Ratio (M-H, Fixed, 95% CI) 0.87 [0.72, 1.06]14 Mortality 17 11771 Odds Ratio (M-H, Fixed, 95% CI) 0.85 [0.62, 1.17]14.1 Roflumilast 4 5439 Odds Ratio (M-H, Fixed, 95% CI) 0.89 [0.62, 1.28]14.2 Cilomilast 13 6332 Odds Ratio (M-H, Fixed, 95% CI) 0.70 [0.34, 1.45]15 No. of subjects (additional 17 12985 Odds Ratio (M-H, Fixed, 95% CI) 0.77 [0.71, 0.83]medication)15.1 Long acting2 1676 Odds Ratio (M-H, Fixed, 95% CI) 0.69 [0.54, 0.88]bronchodilators15.2 Corticosteroids 1 1513 Odds Ratio (M-H, Fixed, 95% CI) 0.87 [0.71, 1.06]15.3 Treatment only 14 9796 Odds Ratio (M-H, Fixed, 95% CI) 0.76 [0.69, 0.84]Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.55

Analysis 1.1.Comparison 1 Lung function, Outcome 1 FEV 1 (by drug).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome:1 FEV 1 (by drug)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 Roflumilast 500 gGrootendorst 2007 38 61 (98) 38 -18.5 (98) 2.1 % 79.50 [ 35.43, 123.57 ]Roflumilast FK1 101 169 109 (273) 172 57 (302) 1.1 % 52.00 [ -9.08, 113.08 ]Roflumilast FK1 103 200 78 (240) 186 39 (245) 1.8 % 39.00 [ -9.44, 87.44 ]Roflumilast M2-107 555 49 (283) 280 -39 (268) 2.7 % 88.00 [ 48.76, 127.24 ]Roflumilast M2-112 760 9 (303) 753 -27 (302) 4.5 % 36.00 [ 5.52, 66.48 ]Roflumilast M2-124 745 46 (218) 745 8 (218) 8.4 % 38.00 [ 15.86, 60.14 ]Roflumilast M2-125 730 33 (189) 766 -25 (194) 11.0 % 58.00 [ 38.59, 77.41 ]Roflumilast M2-127 456 39 (192) 460 -10 (193) 6.7 % 49.00 [ 24.07, 73.93 ]Roflumilast M2-128 365 65 (229) 364 -16 (229) 3.7 % 81.00 [ 47.75, 114.25 ]Subtotal (95% CI) 4018 3764 42.0 % 54.32 [ 44.39, 64.25 ]Heterogeneity: Chi 2 = 10.74, df = 8 (P = 0.22); I 2 =25%Test for overall effect: Z = 10.72 (P < 0.00001)2 Roflumilast 250 gRoflumilast FK1 101 175 93 (273) 172 57 (302) 1.1 % 36.00 [ -24.60, 96.60 ]Roflumilast M2-107 576 24 (288) 280 -39 (268) 2.7 % 63.00 [ 23.78, 102.22 ]Subtotal (95% CI) 751 452 3.8 % 55.03 [ 22.10, 87.96 ]Heterogeneity: Chi 2 = 0.54, df = 1 (P = 0.46); I 2 =0.0%Test for overall effect: Z = 3.28 (P = 0.0011)3 Cilomilast 15 mgCilomilast 039 378 10 (194) 207 -30 (144) 5.4 % 40.00 [ 12.30, 67.70 ]Cilomilast 042 435 30 (210) 230 0 (296) 2.2 % 30.00 [ -13.04, 73.04 ]Cilomilast 076 21 -50 (183) 23 -70 (192) 0.3 % 20.00 [ -90.83, 130.83 ]Cilomilast 091 435 0 (417) 230 -30 (303) 1.3 % 30.00 [ -25.40, 85.40 ]Cilomilast 103657 296 50 (86) 316 6 (89) 21.5 % 44.00 [ 30.13, 57.87 ]Cilomilast 110 20 10 (179) 26 -60 (204) 0.3 % 70.00 [ -40.92, 180.92 ]Cilomilast 121 622 14 (175) 328 -6 (181) 7.2 % 20.00 [ -3.93, 43.93 ]Cilomilast 156 364 7 (153) 377 -17 (155) 8.4 % 24.00 [ 1.82, 46.18 ]-200 -100 0 100 200Favours controlFavours treatment(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.56

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CICilomilast 157 390 32 (197) 411 -2 (182) 6.0 % 34.00 [ 7.70, 60.30 ]Compton 2001 107 130 (206) 106 -30 (207) 1.3 % 160.00 [ 104.53, 215.47 ]Subtotal (95% CI) 3068 2254 54.2 % 38.15 [ 29.41, 46.90 ]Heterogeneity: Chi 2 = 23.75, df = 9 (P = 0.005); I 2 =62%Test for overall effect: Z = 8.55 (P < 0.00001)Total (95% CI) 7837 6470 100.0 % 45.59 [ 39.15, 52.03 ]Heterogeneity: Chi 2 = 41.09, df = 20 (P = 0.004); I 2 =51%Test for overall effect: Z = 13.89 (P < 0.00001)Test for subgroup differences: Chi 2 = 6.07, df = 2 (P = 0.05), I 2 =67%-200 -100 0 100 200Favours controlFavours treatmentAnalysis 1.2.Comparison 1 Lung function, Outcome 2 FEV 1 (by mean COPD severity).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome:2 FEV 1 (by mean COPD severity)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 GOLD stage I + II (FEV1 ≥ 50% predicted)Cilomilast 076 21 -50 (183) 23 -70 (192) 0.6 % 20.00 [ -90.83, 130.83 ]Cilomilast 091 443 -30 (421) 232 -60 (305) 2.3 % 30.00 [ -25.47, 85.47 ]Cilomilast 156 364 7 (153) 377 -17 (155) 14.5 % 24.00 [ 1.82, 46.18 ]Grootendorst 2007 38 61 (98) 38 -18.5 (98) 3.7 % 79.50 [ 35.43, 123.57 ]Roflumilast M2-107 555 49 (283) 280 -39 (268) 4.6 % 88.00 [ 48.76, 127.24 ]Roflumilast M2-127 456 39 (192) 460 -10 (193) 11.5 % 49.00 [ 24.07, 73.93 ]Roflumilast M2-128 365 65 (229) 364 -16 (229) 6.4 % 81.00 [ 47.75, 114.25 ]Subtotal (95% CI) 2242 1774 43.6 % 50.74 [ 37.95, 63.52 ]Heterogeneity: Chi 2 = 14.72, df = 6 (P = 0.02); I 2 =59%Test for overall effect: Z = 7.78 (P < 0.00001)2 GOLD stage III + IV (FEV1 < 50% predicted)-200 -100 0 100 200Favours controlFavours treatment(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.57

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CICilomilast 039 378 10 (194) 207 -30 (144) 9.3 % 40.00 [ 12.30, 67.70 ]Cilomilast 042 440 30 (210) 219 0 (296) 3.7 % 30.00 [ -13.84, 73.84 ]Compton 2001 107 130 (206) 106 -30 (207) 2.3 % 160.00 [ 104.53, 215.47 ]Roflumilast M2-112 760 9 (303) 753 -27 (302) 7.7 % 36.00 [ 5.52, 66.48 ]Roflumilast M2-124 745 46 (218) 745 8 (218) 14.5 % 38.00 [ 15.86, 60.14 ]Roflumilast M2-125 730 33 (189) 766 -25 (194) 18.9 % 58.00 [ 38.59, 77.41 ]Subtotal (95% CI) 3160 2796 56.4 % 49.24 [ 38.01, 60.48 ]Heterogeneity: Chi 2 = 18.98, df = 5 (P = 0.002); I 2 =74%Test for overall effect: Z = 8.59 (P < 0.00001)Total (95% CI) 5402 4570 100.0 % 49.89 [ 41.45, 58.33 ]Heterogeneity: Chi 2 = 33.73, df = 12 (P = 0.00074); I 2 =64%Test for overall effect: Z = 11.59 (P < 0.00001)Test for subgroup differences: Chi 2 = 0.03, df = 1 (P = 0.86), I 2 =0.0%-200 -100 0 100 200Favours controlFavours treatmentPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.58

Analysis 1.3.Comparison 1 Lung function, Outcome 3 FEV 1 (Roflumilast 500 µg by mean COPD severity).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome:3 FEV 1 (Roflumilast 500 g by mean COPD severity)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean Difference1 GOLD stage I + II (FEV 1 ≥ 50% predicted)N Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CIGrootendorst 2007 38 61 (98) 38 -18.5 (98) 5.5 % 79.50 [ 35.43, 123.57 ]Roflumilast M2-107 555 49 (283) 280 -39 (268) 6.9 % 88.00 [ 48.76, 127.24 ]Roflumilast M2-127 456 39 (192) 460 -10 (193) 17.0 % 49.00 [ 24.07, 73.93 ]Roflumilast M2-128 365 65 (229) 364 -16 (229) 9.6 % 81.00 [ 47.75, 114.25 ]Subtotal (95% CI) 1414 1142 38.9 % 68.03 [ 51.54, 84.52 ]Heterogeneity: Chi 2 = 4.08, df = 3 (P = 0.25); I 2 =26%Test for overall effect: Z = 8.09 (P < 0.00001)2 GOLD stage III + IV (FEV 1 < 50% predicted)Roflumilast M2-112 760 9 (303) 753 -27 (302) 11.4 % 36.00 [ 5.52, 66.48 ]Roflumilast M2-124 745 46 (218) 745 8 (218) 21.6 % 38.00 [ 15.86, 60.14 ]Roflumilast M2-125 730 33 (189) 766 -25 (194) 28.1 % 58.00 [ 38.59, 77.41 ]Subtotal (95% CI) 2235 2264 61.1 % 46.83 [ 33.66, 59.99 ]Heterogeneity: Chi 2 = 2.37, df = 2 (P = 0.31); I 2 =16%Test for overall effect: Z = 6.97 (P < 0.00001)Total (95% CI) 3649 3406 100.0 % 55.08 [ 44.79, 65.37 ]Heterogeneity: Chi 2 = 10.33, df = 6 (P = 0.11); I 2 =42%Test for overall effect: Z = 10.49 (P < 0.00001)Test for subgroup differences: Chi 2 = 3.88, df = 1 (P = 0.05), I 2 =74%-200 -100 0 100 200Favours controlFavours treatmentPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.59

Analysis 1.4.Comparison 1 Lung function, Outcome 4 FEV 1 (by study duration).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome:4 FEV 1 (by study duration)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 Duration < 12 weeksCilomilast 076 21 -50 (183) 23 -70 (192) 0.4 % 20.00 [ -90.83, 130.83 ]Cilomilast 110 20 10 (179) 26 -60 (204) 0.4 % 70.00 [ -40.92, 180.92 ]Compton 2001 107 130 (206) 106 -30 (207) 1.4 % 160.00 [ 104.53, 215.47 ]Grootendorst 2007 38 61 (98) 38 -18.5 (98) 2.2 % 79.50 [ 35.43, 123.57 ]Subtotal (95% CI) 186 193 4.3 % 99.99 [ 68.42, 131.57 ]Heterogeneity: Chi 2 = 7.61, df = 3 (P = 0.05); I 2 =61%Test for overall effect: Z = 6.21 (P < 0.00001)2 Duration 24-26 weeksCilomilast 039 378 10 (194) 207 -30 (144) 5.6 % 40.00 [ 12.30, 67.70 ]Cilomilast 042 440 30 (210) 219 0 (296) 2.2 % 30.00 [ -13.84, 73.84 ]Cilomilast 091 435 0 (417) 230 -30 (303) 1.4 % 30.00 [ -25.40, 85.40 ]Cilomilast 103657 296 50 (86) 316 6 (89) 22.4 % 44.00 [ 30.13, 57.87 ]Cilomilast 121 622 14 (175) 328 -6 (181) 7.5 % 20.00 [ -3.93, 43.93 ]Cilomilast 156 364 7 (153) 377 -17 (155) 8.8 % 24.00 [ 1.82, 46.18 ]Roflumilast FK1 101 169 109 (273) 172 57 (302) 1.2 % 52.00 [ -9.08, 113.08 ]Roflumilast FK1 103 200 78 (240) 186 39 (245) 1.8 % 39.00 [ -9.44, 87.44 ]Roflumilast M2-107 555 49 (283) 280 -39 (268) 2.8 % 88.00 [ 48.76, 127.24 ]Roflumilast M2-127 456 39 (192) 460 -10 (193) 6.9 % 49.00 [ 24.07, 73.93 ]Roflumilast M2-128 365 65 (229) 364 -16 (229) 3.9 % 81.00 [ 47.75, 114.25 ]Subtotal (95% CI) 4280 3139 64.6 % 42.03 [ 33.86, 50.20 ]Heterogeneity: Chi 2 = 17.33, df = 10 (P = 0.07); I 2 =42%Test for overall effect: Z = 10.08 (P < 0.00001)3 Duration 52 weeksCilomilast 157 390 32 (197) 411 -2 (182) 6.2 % 34.00 [ 7.70, 60.30 ]Roflumilast M2-112 760 9 (303) 753 -27 (302) 4.6 % 36.00 [ 5.52, 66.48 ]Roflumilast M2-124 745 46 (218) 745 8 (218) 8.8 % 38.00 [ 15.86, 60.14 ]Roflumilast M2-125 730 33 (189) 766 -25 (194) 11.4 % 58.00 [ 38.59, 77.41 ]-200 -100 0 100 200Favours controlFavours treatment(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.60

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CISubtotal (95% CI) 2625 2675 31.1 % 44.26 [ 32.49, 56.03 ]Heterogeneity: Chi 2 = 3.10, df = 3 (P = 0.38); I 2 =3%Test for overall effect: Z = 7.37 (P < 0.00001)Total (95% CI) 7091 6007 100.0 % 45.23 [ 38.66, 51.79 ]Heterogeneity: Chi 2 = 40.21, df = 18 (P = 0.002); I 2 =55%Test for overall effect: Z = 13.50 (P < 0.00001)Test for subgroup differences: Chi 2 = 12.17, df = 2 (P = 0.00), I 2 =84%-200 -100 0 100 200Favours controlFavours treatmentAnalysis 1.5.Comparison 1 Lung function, Outcome 5 FEV 1 (additional medication).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome:5 FEV 1 (additional medication)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 Long acting bronchodilatorRoflumilast M2-127 456 39 (192) 460 -10 (193) 6.9 % 49.00 [ 24.07, 73.93 ]Roflumilast M2-128 365 65 (229) 364 -16 (229) 3.9 % 81.00 [ 47.75, 114.25 ]Subtotal (95% CI) 821 824 10.8 % 60.52 [ 40.57, 80.46 ]Heterogeneity: Chi 2 = 2.28, df = 1 (P = 0.13); I 2 =56%Test for overall effect: Z = 5.95 (P < 0.00001)2 CorticosteroidsRoflumilast M2-112 760 9 (303) 753 -27 (302) 4.6 % 36.00 [ 5.52, 66.48 ]Subtotal (95% CI) 760 753 4.6 % 36.00 [ 5.52, 66.48 ]Heterogeneity: not applicableTest for overall effect: Z = 2.31 (P = 0.021)3 PDE4i treatment onlyCilomilast 039 378 10 (194) 207 -30 (144) 5.6 % 40.00 [ 12.30, 67.70 ]Cilomilast 042 440 30 (210) 219 0 (296) 2.2 % 30.00 [ -13.84, 73.84 ]Cilomilast 076 21 -50 (183) 23 -70 (192) 0.4 % 20.00 [ -90.83, 130.83 ]-200 -100 0 100 200Favours controlFavours treatment(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.61

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CICilomilast 091 435 0 (417) 230 -30 (303) 1.4 % 30.00 [ -25.40, 85.40 ]Cilomilast 103657 296 50 (86) 316 6 (89) 22.4 % 44.00 [ 30.13, 57.87 ]Cilomilast 110 20 10 (179) 26 -60 (204) 0.4 % 70.00 [ -40.92, 180.92 ]Cilomilast 121 622 14 (175) 328 -6 (181) 7.5 % 20.00 [ -3.93, 43.93 ]Cilomilast 156 364 7 (153) 377 -17 (155) 8.8 % 24.00 [ 1.82, 46.18 ]Cilomilast 157 390 32 (197) 411 -2 (182) 6.2 % 34.00 [ 7.70, 60.30 ]Compton 2001 107 130 (206) 106 -30 (207) 1.4 % 160.00 [ 104.53, 215.47 ]Grootendorst 2007 38 61 (98) 38 -18.5 (98) 2.2 % 79.50 [ 35.43, 123.57 ]Roflumilast FK1 101 169 109 (273) 172 57 (302) 1.2 % 52.00 [ -9.08, 113.08 ]Roflumilast FK1 103 200 78 (240) 186 39 (245) 1.8 % 39.00 [ -9.44, 87.44 ]Roflumilast M2-107 555 49 (283) 280 -39 (268) 2.8 % 88.00 [ 48.76, 127.24 ]Roflumilast M2-124 745 46 (218) 745 8 (218) 8.8 % 38.00 [ 15.86, 60.14 ]Roflumilast M2-125 730 33 (189) 766 -25 (194) 11.4 % 58.00 [ 38.59, 77.41 ]Subtotal (95% CI) 5510 4430 84.5 % 43.77 [ 36.63, 50.91 ]Heterogeneity: Chi 2 = 35.16, df = 15 (P = 0.002); I 2 =57%Test for overall effect: Z = 12.02 (P < 0.00001)Total (95% CI) 7091 6007 100.0 % 45.23 [ 38.66, 51.79 ]Heterogeneity: Chi 2 = 40.21, df = 18 (P = 0.002); I 2 =55%Test for overall effect: Z = 13.50 (P < 0.00001)Test for subgroup differences: Chi 2 = 2.77, df = 2 (P = 0.25), I 2 =28%-200 -100 0 100 200Favours controlFavours treatmentPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.62

Analysis 1.6.Comparison 1 Lung function, Outcome 6 FEV 1 (published vs. unpublished).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome:6 FEV 1 (published vs. unpublished)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean Difference1 PublishedN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CICilomilast 039 378 10 (194) 207 -30 (144) 5.6 % 40.00 [ 12.30, 67.70 ]Cilomilast 076 21 -50 (183) 23 -70 (192) 0.4 % 20.00 [ -90.83, 130.83 ]Compton 2001 107 130 (206) 106 -30 (207) 1.4 % 160.00 [ 104.53, 215.47 ]Grootendorst 2007 38 61 (98) 38 -18.5 (98) 2.2 % 79.50 [ 35.43, 123.57 ]Roflumilast FK1 101 169 109 (273) 172 57 (302) 1.2 % 52.00 [ -9.08, 113.08 ]Roflumilast FK1 103 200 78 (240) 186 39 (245) 1.8 % 39.00 [ -9.44, 87.44 ]Roflumilast M2-107 555 51 (283) 280 -45 (268) 2.8 % 96.00 [ 56.76, 135.24 ]Roflumilast M2-112 760 9 (303) 753 -27 (302) 4.6 % 36.00 [ 5.52, 66.48 ]Roflumilast M2-124 745 46 (218) 745 8 (218) 8.8 % 38.00 [ 15.86, 60.14 ]Roflumilast M2-125 730 33 (189) 766 -25 (194) 11.4 % 58.00 [ 38.59, 77.41 ]Roflumilast M2-127 456 39 (192) 460 -10 (193) 6.9 % 49.00 [ 24.07, 73.93 ]Roflumilast M2-128 365 65 (229) 364 -16 (229) 3.9 % 81.00 [ 47.75, 114.25 ]Subtotal (95% CI) 4524 4100 51.1 % 55.85 [ 46.66, 65.03 ]Heterogeneity: Chi 2 = 27.47, df = 11 (P = 0.004); I 2 =60%Test for overall effect: Z = 11.92 (P < 0.00001)2 UnpublishedCilomilast 042 440 30 (210) 219 0 (296) 2.2 % 30.00 [ -13.84, 73.84 ]Cilomilast 091 435 0 (417) 230 -30 (303) 1.4 % 30.00 [ -25.40, 85.40 ]Cilomilast 103657 296 50 (86) 316 6 (89) 22.4 % 44.00 [ 30.13, 57.87 ]Cilomilast 110 20 10 (179) 26 -60 (204) 0.4 % 70.00 [ -40.92, 180.92 ]Cilomilast 121 622 14 (175) 328 -6 (181) 7.5 % 20.00 [ -3.93, 43.93 ]Cilomilast 156 364 7 (153) 377 -17 (155) 8.8 % 24.00 [ 1.82, 46.18 ]Cilomilast 157 390 32 (197) 411 -2 (182) 6.2 % 34.00 [ 7.70, 60.30 ]Subtotal (95% CI) 2567 1907 48.9 % 34.60 [ 25.21, 43.98 ]Heterogeneity: Chi 2 = 4.53, df = 6 (P = 0.60); I 2 =0.0%Test for overall effect: Z = 7.22 (P < 0.00001)Total (95% CI) 7091 6007 100.0 % 45.45 [ 38.89, 52.02 ]-200 -100 0 100 200Favours controlFavours treatment(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.63

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceHeterogeneity: Chi 2 = 42.07, df = 18 (P = 0.001); I 2 =57%Test for overall effect: Z = 13.57 (P < 0.00001)Test for subgroup differences: Chi 2 = 10.06, df = 1 (P = 0.00), I 2 =90%N Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI-200 -100 0 100 200Favours controlFavours treatmentAnalysis 1.7.Comparison 1 Lung function, Outcome 7 FEV 1 (random effects model).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome:7 FEV 1 (random effects model)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Random,95% CI IV,Random,95% CICilomilast 039 378 10 (194) 207 -30 (144) 6.2 % 40.00 [ 12.30, 67.70 ]Cilomilast 042 440 30 (210) 219 0 (296) 3.7 % 30.00 [ -13.84, 73.84 ]Cilomilast 076 21 -50 (183) 23 -70 (192) 0.8 % 20.00 [ -90.83, 130.83 ]Cilomilast 091 435 0 (417) 230 -30 (303) 2.7 % 30.00 [ -25.40, 85.40 ]Cilomilast 103657 296 50 (86) 316 6 (89) 9.3 % 44.00 [ 30.13, 57.87 ]Cilomilast 110 20 10 (179) 26 -60 (204) 0.8 % 70.00 [ -40.92, 180.92 ]Cilomilast 121 622 14 (175) 328 -6 (181) 7.0 % 20.00 [ -3.93, 43.93 ]Cilomilast 156 364 7 (153) 377 -17 (155) 7.4 % 24.00 [ 1.82, 46.18 ]Cilomilast 157 390 32 (197) 411 -2 (182) 6.5 % 34.00 [ 7.70, 60.30 ]Compton 2001 107 130 (206) 106 -30 (207) 2.7 % 160.00 [ 104.53, 215.47 ]Grootendorst 2007 38 61 (98) 38 -18.5 (98) 3.7 % 79.50 [ 35.43, 123.57 ]Roflumilast FK1 101 175 93 (273) 172 57 (302) 2.3 % 36.00 [ -24.60, 96.60 ]Roflumilast FK1 101 169 109 (273) 172 57 (302) 2.3 % 52.00 [ -9.08, 113.08 ]Roflumilast FK1 103 200 78 (240) 186 39 (245) 3.2 % 39.00 [ -9.44, 87.44 ]Roflumilast M2-107 555 49 (283) 280 -39 (268) 4.3 % 88.00 [ 48.76, 127.24 ]Roflumilast M2-107 576 24 (288) 280 -39 (268) 4.3 % 63.00 [ 23.78, 102.22 ]-200 -100 0 100 200Favours controlFavours treatment(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.64

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Random,95% CI IV,Random,95% CIRoflumilast M2-112 760 9 (303) 753 -27 (302) 5.7 % 36.00 [ 5.52, 66.48 ]Roflumilast M2-124 745 46 (218) 745 8 (218) 7.4 % 38.00 [ 15.86, 60.14 ]Roflumilast M2-125 730 33 (189) 766 -25 (194) 8.0 % 58.00 [ 38.59, 77.41 ]Roflumilast M2-127 456 39 (192) 460 -10 (193) 6.8 % 49.00 [ 24.07, 73.93 ]Roflumilast M2-128 365 65 (229) 364 -16 (229) 5.2 % 81.00 [ 47.75, 114.25 ]Total (95% CI) 7842 6459 100.0 % 48.49 [ 38.13, 58.85 ]Heterogeneity: Tau 2 = 251.40; Chi 2 = 41.07, df = 20 (P = 0.004); I 2 =51%Test for overall effect: Z = 9.17 (P < 0.00001)-200 -100 0 100 200Favours controlFavours treatmentAnalysis 1.8.Comparison 1 Lung function, Outcome 8 FEV 1 (Roflumilast 500 µg vs 250 µg).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome: 8 FEV 1 (Roflumilast 500 g vs 250 g)Study or subgroup Roflumilast 500mcg Roflumilast 250mcg Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CIRoflumilast FK1 101 169 109 (273) 175 93 (273) 25.0 % 16.00 [ -41.71, 73.71 ]Roflumilast M2-107 555 49 (283) 576 24 (288) 75.0 % 25.00 [ -8.28, 58.28 ]Total (95% CI) 724 751 100.0 % 22.75 [ -6.08, 51.58 ]Heterogeneity: Chi 2 = 0.07, df = 1 (P = 0.79); I 2 =0.0%Test for overall effect: Z = 1.55 (P = 0.12)-100 -50 0 50 100Favours 250mcgFavours 500mcgPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.65

Analysis 1.9.Comparison 1 Lung function, Outcome 9 FVC.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome:9 FVCStudy or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CICilomilast 039 394 -10 (397) 208 -120 (433) 5.5 % 110.00 [ 39.29, 180.71 ]Cilomilast 042 448 30 (635) 220 -20 (445) 4.0 % 50.00 [ -33.16, 133.16 ]Cilomilast 091 443 -20 (631) 232 -70 (609) 2.9 % 50.00 [ -47.95, 147.95 ]Cilomilast 103657 296 25 (447) 316 -2 (409) 5.9 % 27.00 [ -41.02, 95.02 ]Cilomilast 156 377 20 (388) 383 -40 (391) 9.0 % 60.00 [ 4.62, 115.38 ]Compton 2001 107 180 (517) 106 -15 (515) 1.4 % 195.00 [ 56.41, 333.59 ]Roflumilast M2-107 555 39 (495) 280 -75 (452) 6.1 % 114.00 [ 46.93, 181.07 ]Roflumilast M2-112 760 -33 (716) 753 -80 (713) 5.3 % 47.00 [ -25.00, 119.00 ]Roflumilast M2-124 729 76 (405) 736 -25 (407) 15.9 % 101.00 [ 59.42, 142.58 ]Roflumilast M2-125 724 58 (350) 764 -45 (359) 21.2 % 103.00 [ 66.97, 139.03 ]Roflumilast M2-127 452 67 (319) 460 10 (322) 15.9 % 57.00 [ 15.40, 98.60 ]Roflumilast M2-128 364 27 (439) 363 -74 (419) 7.1 % 101.00 [ 38.62, 163.38 ]Total (95% CI) 5649 4821 100.0 % 82.67 [ 66.10, 99.24 ]Heterogeneity: Chi 2 = 12.88, df = 11 (P = 0.30); I 2 =15%Test for overall effect: Z = 9.78 (P < 0.00001)-500 -250 0 250 500Favours controlFavours treatmentPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.66

Analysis 1.10.Comparison 1 Lung function, Outcome 10 PEF.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome:10 PEFStudy or subgroup PDE4i Treatment Control Mean Difference Weight Mean Difference1 Roflumilast 500 gN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CIRoflumilast FK1 101 169 10 (39) 172 2 (52.5) 7.3 % 8.00 [ -1.80, 17.80 ]Roflumilast M2-124 729 8.08 (40.5) 736 3.87 (40.2) 41.1 % 4.21 [ 0.08, 8.34 ]Roflumilast M2-125 724 1.93 (40.1) 764 -3.14 (40.1) 42.2 % 5.07 [ 0.99, 9.15 ]Subtotal (95% CI) 1622 1672 90.6 % 4.92 [ 2.13, 7.70 ]Heterogeneity: Chi 2 = 0.50, df = 2 (P = 0.78); I 2 =0.0%Test for overall effect: Z = 3.46 (P = 0.00054)2 Roflumilast 250 gRoflumilast FK1 101 175 9 (52.5) 172 2 (52.5) 5.7 % 7.00 [ -4.05, 18.05 ]Subtotal (95% CI) 175 172 5.7 % 7.00 [ -4.05, 18.05 ]Heterogeneity: not applicableTest for overall effect: Z = 1.24 (P = 0.21)3 Cilomilast 15 mgCompton 2001 107 25.5 (51.7) 106 -8.5 (51.5) 3.7 % 34.00 [ 20.14, 47.86 ]Subtotal (95% CI) 107 106 3.7 % 34.00 [ 20.14, 47.86 ]Heterogeneity: not applicableTest for overall effect: Z = 4.81 (P < 0.00001)Total (95% CI) 1904 1950 100.0 % 6.10 [ 3.45, 8.75 ]Heterogeneity: Chi 2 = 16.79, df = 4 (P = 0.002); I 2 =76%Test for overall effect: Z = 4.51 (P < 0.00001)Test for subgroup differences: Chi 2 = 16.29, df = 2 (P = 0.00), I 2 =88%-50 -25 0 25 50Favours controlFavours treatmentPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.67

Analysis 1.11.Comparison 1 Lung function, Outcome 11 FEV 1 (Evidence quality).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:1 Lung functionOutcome:11 FEV 1 (Evidence quality)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean Difference1 Low risk of biasN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CIGrootendorst 2007 38 61 (98) 38 -18.5 (98) 2.2 % 79.50 [ 35.43, 123.57 ]Roflumilast M2-107 555 49 (283) 280 -39 (268) 2.8 % 88.00 [ 48.76, 127.24 ]Roflumilast M2-112 760 9 (303) 753 -27 (302) 4.6 % 36.00 [ 5.52, 66.48 ]Roflumilast M2-124 745 46 (218) 745 8 (218) 8.8 % 38.00 [ 15.86, 60.14 ]Roflumilast M2-125 730 33 (189) 766 -25 (194) 11.4 % 58.00 [ 38.59, 77.41 ]Roflumilast M2-127 456 39 (192) 460 -10 (193) 6.9 % 49.00 [ 24.07, 73.93 ]Roflumilast M2-128 365 65 (229) 364 -16 (229) 3.9 % 81.00 [ 47.75, 114.25 ]Subtotal (95% CI) 3649 3406 40.7 % 55.08 [ 44.79, 65.37 ]Heterogeneity: Chi 2 = 10.33, df = 6 (P = 0.11); I 2 =42%Test for overall effect: Z = 10.49 (P < 0.00001)2 Unclear risk of biasCilomilast 039 378 10 (194) 207 -30 (144) 5.6 % 40.00 [ 12.30, 67.70 ]Cilomilast 042 435 30 (210) 230 0 (296) 2.3 % 30.00 [ -13.04, 73.04 ]Cilomilast 076 21 -50 (183) 23 -70 (192) 0.4 % 20.00 [ -90.83, 130.83 ]Cilomilast 091 435 0 (417) 230 -30 (303) 1.4 % 30.00 [ -25.40, 85.40 ]Cilomilast 103657 296 50 (86) 316 6 (89) 22.4 % 44.00 [ 30.13, 57.87 ]Cilomilast 110 20 10 (179) 26 -60 (204) 0.3 % 70.00 [ -40.92, 180.92 ]Cilomilast 121 622 14 (175) 328 -6 (181) 7.5 % 20.00 [ -3.93, 43.93 ]Cilomilast 156 364 7 (153) 377 -17 (155) 8.8 % 24.00 [ 1.82, 46.18 ]Cilomilast 157 390 32 (197) 411 -2 (182) 6.2 % 34.00 [ 7.70, 60.30 ]Compton 2001 107 130 (206) 106 -30 (207) 1.4 % 160.00 [ 104.53, 215.47 ]Roflumilast FK1 101 169 109 (273) 172 57 (302) 1.2 % 52.00 [ -9.08, 113.08 ]Roflumilast FK1 103 200 78 (240) 186 39 (245) 1.8 % 39.00 [ -9.44, 87.44 ]Subtotal (95% CI) 3437 2612 59.3 % 38.45 [ 29.93, 46.97 ]Heterogeneity: Chi 2 = 23.94, df = 11 (P = 0.01); I 2 =54%Test for overall effect: Z = 8.84 (P < 0.00001)Total (95% CI) 7086 6018 100.0 % 45.22 [ 38.65, 51.78 ]-200 -100 0 100 200Favours controlFavours treatment(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.68

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceHeterogeneity: Chi 2 = 40.22, df = 18 (P = 0.002); I 2 =55%Test for overall effect: Z = 13.51 (P < 0.00001)Test for subgroup differences: Chi 2 = 5.95, df = 1 (P = 0.01), I 2 =83%N Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI-200 -100 0 100 200Favours controlFavours treatmentAnalysis 2.1.Comparison 2 Quality of life, Outcome 1 SGRQ total score.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:2 Quality of lifeOutcome:1 SGRQ total scoreStudy or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 Roflumilast 500 gRoflumilast M2-107 555 -3.5 (14.1) 280 -1.8 (13.4) 10.2 % -1.70 [ -3.66, 0.26 ]Roflumilast M2-112 760 -1.7 (19.3) 753 -2 (19.2) 10.4 % 0.30 [ -1.64, 2.24 ]Subtotal (95% CI) 1315 1033 20.6 % -0.69 [ -2.07, 0.69 ]Heterogeneity: Chi 2 = 2.02, df = 1 (P = 0.16); I 2 =51%Test for overall effect: Z = 0.98 (P = 0.33)2 Roflumilast 250 gRoflumilast M2-107 576 -3.4 (14.4) 280 -1.8 (13.4) 10.2 % -1.60 [ -3.56, 0.36 ]Subtotal (95% CI) 576 280 10.2 % -1.60 [ -3.56, 0.36 ]Heterogeneity: not applicableTest for overall effect: Z = 1.60 (P = 0.11)3 Cilomilast 15 mgCilomilast 039 310 -3.7 (12.32) 181 0.4 (10.76) 9.0 % -4.10 [ -6.18, -2.02 ]Cilomilast 042 375 -4.2 (15.5) 190 -4.9 (13.8) 6.2 % 0.70 [ -1.81, 3.21 ]Cilomilast 091 369 -2.7 (21.1) 197 -2.3 (16.8) 3.9 % -0.40 [ -3.58, 2.78 ]Cilomilast 103657 292 -1.8 (10.2) 310 -1.84 (10) 15.0 % 0.04 [ -1.57, 1.65 ]Cilomilast 121 580 -9 (14.7) 320 -8.7 (14.7) 9.7 % -0.30 [ -2.31, 1.71 ]Cilomilast 156 304 -3.2 (10.5) 337 -1.3 (11) 14.1 % -1.90 [ -3.57, -0.23 ]-10 -5 0 5 10Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.69

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CICilomilast 157 347 -1.29 (14.9) 369 -1.49 (14.4) 8.5 % 0.20 [ -1.95, 2.35 ]Compton 2001 107 -3.9 (13.4) 106 0 (13.4) 3.0 % -3.90 [ -7.50, -0.30 ]Subtotal (95% CI) 2684 2010 69.3 % -1.06 [ -1.81, -0.31 ]Heterogeneity: Chi 2 = 17.26, df = 7 (P = 0.02); I 2 =59%Test for overall effect: Z = 2.76 (P = 0.0058)Total (95% CI) 4575 3323 100.0 % -1.04 [ -1.66, -0.41 ]Heterogeneity: Chi 2 = 19.84, df = 10 (P = 0.03); I 2 =50%Test for overall effect: Z = 3.25 (P = 0.0011)Test for subgroup differences: Chi 2 = 0.56, df = 2 (P = 0.75), I 2 =0.0%-10 -5 0 5 10Favours treatmentFavours controlAnalysis 2.2.Comparison 2 Quality of life, Outcome 2 SGRQ total score (by mean COPD severity).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:2 Quality of lifeOutcome:2 SGRQ total score (by mean COPD severity)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 GOLD stage I and IICilomilast 091 369 -2.7 (21.1) 197 -2.3 (16.8) 6.8 % -0.40 [ -3.58, 2.78 ]Cilomilast 156 304 -3.2 (10.5) 337 -1.3 (11) 24.8 % -1.90 [ -3.57, -0.23 ]Roflumilast M2-107 555 -3.5 (14.1) 280 -1.8 (13.4) 17.9 % -1.70 [ -3.66, 0.26 ]Subtotal (95% CI) 1228 814 49.6 % -1.62 [ -2.80, -0.44 ]Heterogeneity: Chi 2 = 0.68, df = 2 (P = 0.71); I 2 =0.0%Test for overall effect: Z = 2.70 (P = 0.0070)2 GOLD stage III and IVCilomilast 039 310 -3.7 (12.32) 181 0.4 (10.76) 15.9 % -4.10 [ -6.18, -2.02 ]Cilomilast 042 375 -4.2 (15.5) 190 -4.9 (13.8) 10.9 % 0.70 [ -1.81, 3.21 ]Compton 2001 107 -3.9 (13.4) 106 0 (13.4) 5.3 % -3.90 [ -7.50, -0.30 ]Roflumilast M2-112 760 -1.7 (19.3) 753 -2 (19.2) 18.3 % 0.30 [ -1.64, 2.24 ]-10 -5 0 5 10Favours experimentalFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.70

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CISubtotal (95% CI) 1552 1230 50.4 % -1.44 [ -2.61, -0.27 ]Heterogeneity: Chi 2 = 13.94, df = 3 (P = 0.003); I 2 =78%Test for overall effect: Z = 2.42 (P = 0.016)Total (95% CI) 2780 2044 100.0 % -1.53 [ -2.36, -0.70 ]Heterogeneity: Chi 2 = 14.66, df = 6 (P = 0.02); I 2 =59%Test for overall effect: Z = 3.62 (P = 0.00030)Test for subgroup differences: Chi 2 = 0.05, df = 1 (P = 0.83), I 2 =0.0%-10 -5 0 5 10Favours experimentalFavours controlAnalysis 2.3.Comparison 2 Quality of life, Outcome 3 SGRQ total score (by published vs unpublished).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:2 Quality of lifeOutcome:3 SGRQ total score (by published vs unpublished)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 PublishedCilomilast 039 310 -3.7 (12.32) 181 0.4 (10.76) 10.0 % -4.10 [ -6.18, -2.02 ]Compton 2001 107 -3.9 (13.4) 106 0 (13.4) 3.4 % -3.90 [ -7.50, -0.30 ]Roflumilast M2-107 555 -3.5 (14.1) 280 -1.8 (13.4) 11.3 % -1.70 [ -3.66, 0.26 ]Roflumilast M2-112 760 -1.7 (19.3) 753 -2 (19.2) 11.6 % 0.30 [ -1.64, 2.24 ]Subtotal (95% CI) 1732 1320 36.3 % -1.93 [ -3.02, -0.83 ]Heterogeneity: Chi 2 = 10.45, df = 3 (P = 0.02); I 2 =71%Test for overall effect: Z = 3.45 (P = 0.00055)2 UnpublishedCilomilast 042 375 -4.2 (15.5) 190 -4.9 (13.8) 6.9 % 0.70 [ -1.81, 3.21 ]Cilomilast 091 369 -2.7 (21.1) 197 -2.3 (16.8) 4.3 % -0.40 [ -3.58, 2.78 ]Cilomilast 103657 292 -1.8 (10.2) 310 -1.84 (10) 16.7 % 0.04 [ -1.57, 1.65 ]Cilomilast 121 580 -9 (14.7) 320 -8.7 (14.7) 10.8 % -0.30 [ -2.31, 1.71 ]Cilomilast 156 304 -3.2 (10.5) 337 -1.3 (11) 15.7 % -1.90 [ -3.57, -0.23 ]-10 -5 0 5 10Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.71

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CICilomilast 157 347 -1.29 (14.9) 369 -1.49 (14.4) 9.4 % 0.20 [ -1.95, 2.35 ]Subtotal (95% CI) 2267 1723 63.7 % -0.43 [ -1.26, 0.40 ]Heterogeneity: Chi 2 = 4.44, df = 5 (P = 0.49); I 2 =0.0%Test for overall effect: Z = 1.02 (P = 0.31)Total (95% CI) 3999 3043 100.0 % -0.97 [ -1.63, -0.31 ]Heterogeneity: Chi 2 = 19.49, df = 9 (P = 0.02); I 2 =54%Test for overall effect: Z = 2.89 (P = 0.0038)Test for subgroup differences: Chi 2 = 4.60, df = 1 (P = 0.03), I 2 =78%-10 -5 0 5 10Favours treatmentFavours controlAnalysis 2.4.Comparison 2 Quality of life, Outcome 4 SGRQ total score (by duration).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:2 Quality of lifeOutcome:4 SGRQ total score (by duration)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 Duration < 12 weeksCompton 2001 107 -3.9 (13.4) 106 0 (13.4) 3.4 % -3.90 [ -7.50, -0.30 ]Subtotal (95% CI) 107 106 3.4 % -3.90 [ -7.50, -0.30 ]Heterogeneity: not applicableTest for overall effect: Z = 2.12 (P = 0.034)2 Duration 24-26 weeksCilomilast 039 310 -3.7 (12.32) 181 0.4 (10.76) 10.0 % -4.10 [ -6.18, -2.02 ]Cilomilast 042 375 -4.2 (15.5) 190 -4.9 (13.8) 6.9 % 0.70 [ -1.81, 3.21 ]Cilomilast 091 369 -2.7 (21.1) 197 -2.3 (16.8) 4.3 % -0.40 [ -3.58, 2.78 ]Cilomilast 103657 292 -1.8 (10.2) 310 -1.84 (10) 16.7 % 0.04 [ -1.57, 1.65 ]Cilomilast 121 580 -9 (14.7) 320 -8.7 (14.7) 10.8 % -0.30 [ -2.31, 1.71 ]Cilomilast 156 304 -3.2 (10.5) 337 -1.32 (11) 15.7 % -1.88 [ -3.55, -0.21 ]Roflumilast M2-107 555 -3.5 (14.1) 280 -1.8 (13.4) 11.3 % -1.70 [ -3.66, 0.26 ]-10 -5 0 5 10Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.72

(... Continued)Study or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CISubtotal (95% CI) 2785 1815 75.7 % -1.18 [ -1.94, -0.42 ]Heterogeneity: Chi 2 = 13.81, df = 6 (P = 0.03); I 2 =57%Test for overall effect: Z = 3.05 (P = 0.0023)3 Duration 52 weeksCilomilast 157 347 -1.29 (14.9) 369 -1.49 (14.4) 9.4 % 0.20 [ -1.95, 2.35 ]Roflumilast M2-112 760 -1.7 (19.3) 753 -2 (19.2) 11.6 % 0.30 [ -1.64, 2.24 ]Subtotal (95% CI) 1107 1122 21.0 % 0.26 [ -1.18, 1.69 ]Heterogeneity: Chi 2 = 0.00, df = 1 (P = 0.95); I 2 =0.0%Test for overall effect: Z = 0.35 (P = 0.73)Total (95% CI) 3999 3043 100.0 % -0.97 [ -1.63, -0.31 ]Heterogeneity: Chi 2 = 19.44, df = 9 (P = 0.02); I 2 =54%Test for overall effect: Z = 2.88 (P = 0.0039)Test for subgroup differences: Chi 2 = 5.62, df = 2 (P = 0.06), I 2 =64%-10 -5 0 5 10Favours treatmentFavours controlAnalysis 3.1.Comparison 3 Exacerbations, Outcome 1 No. of subjects (by drug).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:3 ExacerbationsOutcome:1 No. of subjects (by drug)Study or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 Roflumilast 500 gRoflumilast FK1 101 19/169 25/172 1.6 % 0.74 [ 0.39, 1.41 ]Roflumilast M2-107 157/555 97/280 6.8 % 0.74 [ 0.55, 1.01 ]Roflumilast M2-112 339/760 362/753 14.7 % 0.87 [ 0.71, 1.06 ]Roflumilast M2-124 70/769 82/755 5.5 % 0.82 [ 0.59, 1.15 ]Roflumilast M2-125 87/778 122/790 7.9 % 0.69 [ 0.51, 0.93 ]Roflumilast M2-127 74/466 111/467 6.8 % 0.61 [ 0.44, 0.84 ]Roflumilast M2-128 58/374 67/369 4.2 % 0.83 [ 0.56, 1.22 ]0.5 0.7 1 1.5 2Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.73

(... Continued)Study or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CISubtotal (95% CI) 3871 3586 47.5 % 0.77 [ 0.69, 0.87 ]Total events: 804 (PDE4i Treatment), 866 (Control)Heterogeneity: Chi 2 = 4.34, df = 6 (P = 0.63); I 2 =0.0%Test for overall effect: Z = 4.40 (P = 0.000011)2 Roflumilast 250 gRoflumilast M2-107 207/576 97/280 6.1 % 1.06 [ 0.78, 1.43 ]Subtotal (95% CI) 576 280 6.1 % 1.06 [ 0.78, 1.43 ]Total events: 207 (PDE4i Treatment), 97 (Control)Heterogeneity: not applicableTest for overall effect: Z = 0.37 (P = 0.71)3 CilomilastCilomilast 039 95/431 76/216 5.8 % 0.52 [ 0.36, 0.75 ]Cilomilast 042 176/474 83/226 5.2 % 1.02 [ 0.73, 1.41 ]Cilomilast 076 5/29 4/30 0.2 % 1.35 [ 0.33, 5.64 ]Cilomilast 091 151/469 107/242 7.0 % 0.60 [ 0.44, 0.82 ]Cilomilast 111 15/79 15/77 0.9 % 0.97 [ 0.44, 2.15 ]Cilomilast 121 297/678 178/340 9.7 % 0.71 [ 0.55, 0.92 ]Cilomilast 156 114/418 125/407 6.7 % 0.85 [ 0.63, 1.14 ]Cilomilast 157 191/455 209/452 8.9 % 0.84 [ 0.65, 1.09 ]Cilomilast 168 28/203 18/103 1.5 % 0.76 [ 0.40, 1.44 ]Cilomilast 180 3/97 6/102 0.4 % 0.51 [ 0.12, 2.10 ]Subtotal (95% CI) 3333 2195 46.4 % 0.76 [ 0.67, 0.85 ]Total events: 1075 (PDE4i Treatment), 821 (Control)Heterogeneity: Chi 2 = 12.00, df = 9 (P = 0.21); I 2 =25%Test for overall effect: Z = 4.64 (P < 0.00001)Total (95% CI) 7780 6061 100.0 % 0.78 [ 0.72, 0.85 ]Total events: 2086 (PDE4i Treatment), 1784 (Control)Heterogeneity: Chi 2 = 20.66, df = 17 (P = 0.24); I 2 =18%Test for overall effect: Z = 6.06 (P < 0.00001)Test for subgroup differences: Chi 2 = 4.29, df = 2 (P = 0.12), I 2 =53%0.5 0.7 1 1.5 2Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.74

Analysis 3.2.Comparison 3 Exacerbations, Outcome 2 Exacerbation rate.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:3 ExacerbationsOutcome:2 Exacerbation rateStudy or subgroup PDE4 inhibitor Placebo Mean Difference Weight Mean Difference1 Roflumilast 500 gN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CIRoflumilast M2-107 555 0.75 (1.89) 280 1.13 (2.37) 52.7 % -0.38 [ -0.70, -0.06 ]Subtotal (95% CI) 555 280 52.7 % -0.38 [ -0.70, -0.06 ]Heterogeneity: not applicableTest for overall effect: Z = 2.33 (P = 0.020)2 Roflumilast 250 gRoflumilast M2-107 576 1.03 (2.33) 280 1.13 (2.37) 47.3 % -0.10 [ -0.44, 0.24 ]Subtotal (95% CI) 576 280 47.3 % -0.10 [ -0.44, 0.24 ]Heterogeneity: not applicableTest for overall effect: Z = 0.58 (P = 0.56)Total (95% CI) 1131 560 100.0 % -0.25 [ -0.48, -0.02 ]Heterogeneity: Chi 2 = 1.40, df = 1 (P = 0.24); I 2 =29%Test for overall effect: Z = 2.09 (P = 0.036)Test for subgroup differences: Chi 2 = 1.40, df = 1 (P = 0.24), I 2 =29%-2 -1 0 1 2Favours PDE4 inhibitorFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.75

Analysis 3.3.Comparison 3 Exacerbations, Outcome 3 Exacerbation rate (Inverse variance).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:3 ExacerbationsOutcome:3 Exacerbation rate (Inverse variance)Study or subgroup log [Rate Ratio] Rate Ratio Weight Rate Ratio1 Roflumilast(SE) IV,Fixed,95% CI IV,Fixed,95% CIRoflumilast M2-112 -0.02965 (0.075) 19.2 % 0.97 [ 0.84, 1.12 ]Roflumilast M2-124 -0.07058 (0.061) 29.0 % 0.93 [ 0.83, 1.05 ]Roflumilast M2-125 -0.08619 (0.059) 31.0 % 0.92 [ 0.82, 1.03 ]Roflumilast M2-127 -0.10237 (0.128) 6.6 % 0.90 [ 0.70, 1.16 ]Roflumilast M2-128 -0.07572 (0.168) 3.8 % 0.93 [ 0.67, 1.29 ]Subtotal (95% CI) 89.6 % 0.93 [ 0.87, 1.00 ]Heterogeneity: Chi 2 = 0.43, df = 4 (P = 0.98); I 2 =0.0%Test for overall effect: Z = 2.01 (P = 0.044)2 CilomilastCilomilast 157 -0.02054 (0.102) 10.4 % 0.98 [ 0.80, 1.20 ]Subtotal (95% CI) 10.4 % 0.98 [ 0.80, 1.20 ]Heterogeneity: not applicableTest for overall effect: Z = 0.20 (P = 0.84)Total (95% CI) 100.0 % 0.94 [ 0.88, 1.00 ]Heterogeneity: Chi 2 = 0.64, df = 5 (P = 0.99); I 2 =0.0%Test for overall effect: Z = 1.97 (P = 0.049)Test for subgroup differences: Chi 2 = 0.21, df = 1 (P = 0.65), I 2 =0.0%0.5 0.7 1 1.5 2Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.76

Analysis 3.4.Comparison 3 Exacerbations, Outcome 4 Exacerbation rate (Roflumilast 500 µg vs 250 µg).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:3 ExacerbationsOutcome: 4 Exacerbation rate (Roflumilast 500 g vs 250 g)Study or subgroup Roflumilast 500mcg Roflumilast 250mcg Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CIRoflumilast M2-107 555 0.75 (1.89) 576 1.03 (2.33) 100.0 % -0.28 [ -0.53, -0.03 ]Total (95% CI) 555 576 100.0 % -0.28 [ -0.53, -0.03 ]Heterogeneity: not applicableTest for overall effect: Z = 2.22 (P = 0.026)-2 -1 0 1 2Favours 500mcgFavours 250mcgAnalysis 4.1.Comparison 4 Symptom score, Outcome 1 SGRQ symptom score.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:4 Symptom scoreOutcome:1 SGRQ symptom scoreStudy or subgroup PDE4i Treatment Control Mean Difference Weight Mean Difference1 RoflumilastN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CIRoflumilast M2-107 555 -4.6 (21.2) 280 -3.6 (18.4) 86.1 % -1.00 [ -3.78, 1.78 ]Subtotal (95% CI) 555 280 86.1 % -1.00 [ -3.78, 1.78 ]Heterogeneity: not applicableTest for overall effect: Z = 0.70 (P = 0.48)2 CilomilastCompton 2001 107 -7.2 (25.9) 106 -2.4 (25.7) 13.9 % -4.80 [ -11.73, 2.13 ]Subtotal (95% CI) 107 106 13.9 % -4.80 [ -11.73, 2.13 ]Heterogeneity: not applicableTest for overall effect: Z = 1.36 (P = 0.17)Total (95% CI) 662 386 100.0 % -1.53 [ -4.11, 1.06 ]Heterogeneity: Chi 2 = 0.99, df = 1 (P = 0.32); I 2 =0.0%Test for overall effect: Z = 1.16 (P = 0.25)Test for subgroup differences: Chi 2 = 0.99, df = 1 (P = 0.32), I 2 =0.0%-10 -5 0 5 10Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.77

Analysis 4.2.Comparison 4 Symptom score, Outcome 2 Borg Scale.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:4 Symptom scoreOutcome:2 Borg ScaleStudy or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 CilomilastCilomilast 039 357 -0.13 (1.7) 196 0.19 (1.7) 22.8 % -0.32 [ -0.62, -0.02 ]Cilomilast 042 422 -0.29 (2.1) 206 -0.11 (1.9) 18.6 % -0.18 [ -0.51, 0.15 ]Cilomilast 091 417 0.03 (2.9) 223 0.21 (2.2) 12.4 % -0.18 [ -0.58, 0.22 ]Cilomilast 111 73 0.03 (1.6) 69 -0.03 (1.7) 6.8 % 0.06 [ -0.48, 0.60 ]Cilomilast 156 347 -0.16 (1.7) 369 0.02 (1.5) 36.0 % -0.18 [ -0.42, 0.06 ]Cilomilast 180 84 -0.1 (2.7) 97 0 (2.5) 3.4 % -0.10 [ -0.86, 0.66 ]Total (95% CI) 1700 1160 100.0 % -0.19 [ -0.33, -0.05 ]Heterogeneity: Chi 2 = 1.62, df = 5 (P = 0.90); I 2 =0.0%Test for overall effect: Z = 2.68 (P = 0.0075)-1 -0.5 0 0.5 1Favours treatmentFavours controlAnalysis 4.3.Comparison 4 Symptom score, Outcome 3 Summary symptom score.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:4 Symptom scoreOutcome:3 Summary symptom scoreStudy or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 CilomilastCilomilast 039 382 -0.21 (1.8) 202 -0.12 (1.7) 42.4 % -0.09 [ -0.39, 0.21 ]Cilomilast 042 435 -0.41 (2.3) 212 -0.59 (1.9) 33.1 % 0.18 [ -0.15, 0.51 ]Cilomilast 091 437 -0.3 (2.7) 231 0.04 (2.3) 24.4 % -0.34 [ -0.73, 0.05 ]Total (95% CI) 1254 645 100.0 % -0.06 [ -0.25, 0.13 ]Heterogeneity: Chi 2 = 3.99, df = 2 (P = 0.14); I 2 =50%Test for overall effect: Z = 0.63 (P = 0.53)-1 -0.5 0 0.5 1Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.78

Analysis 4.4.Comparison 4 Symptom score, Outcome 4 Shortness of breath questionnaire.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:4 Symptom scoreOutcome:4 Shortness of breath questionnaireStudy or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CIRoflumilast M2-127 454 -0.6 (14.9) 461 -1.1 (15) 50.2 % 0.50 [ -1.44, 2.44 ]Roflumilast M2-128 359 -3.4 (13.3) 359 -0.7 (13.3) 49.8 % -2.70 [ -4.65, -0.75 ]Total (95% CI) 813 820 100.0 % -1.09 [ -2.47, 0.28 ]Heterogeneity: Chi 2 = 5.22, df = 1 (P = 0.02); I 2 =81%Test for overall effect: Z = 1.56 (P = 0.12)-0.2 -0.1 0 0.1 0.2Favours treatmentFavours controlAnalysis 5.1.Comparison 5 Exercise tolerance, Outcome 1 6-minute walk test.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:5 Exercise toleranceOutcome:1 6-minute walk testStudy or subgroup PDE4i Treatment Control Mean Difference Weight Mean DifferenceN Mean(SD) N Mean(SD) IV,Fixed,95% CI IV,Fixed,95% CI1 CilomilastCilomilast 039 356 16.1 (79) 194 9 (77) 48.8 % 7.10 [ -6.49, 20.69 ]Cilomilast 042 423 5.8 (125) 207 22.2 (109) 24.9 % -16.40 [ -35.44, 2.64 ]Cilomilast 091 408 5.4 (178) 218 4.6 (139) 14.1 % 0.80 [ -24.47, 26.07 ]Cilomilast 111 73 16.1 (82) 69 -3.7 (83) 12.2 % 19.80 [ -7.35, 46.95 ]Total (95% CI) 1260 688 100.0 % 1.92 [ -7.58, 11.41 ]Heterogeneity: Chi 2 = 5.79, df = 3 (P = 0.12); I 2 =48%Test for overall effect: Z = 0.40 (P = 0.69)-100 -50 0 50 100Favours controlFavours treatmentPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.79

Analysis 6.1.Comparison 6 Adverse effects, Outcome 1 No of patients experiencing an adverse effect.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:1 No of patients experiencing an adverse effectStudy or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 Roflumilast 500 gGrootendorst 2007 33/38 26/38 0.2 % 3.05 [ 0.95, 9.75 ]Roflumilast FK1 101 82/169 85/172 3.1 % 0.96 [ 0.63, 1.48 ]Roflumilast M2-107 370/555 174/280 5.6 % 1.22 [ 0.90, 1.64 ]Roflumilast M2-112 592/760 584/753 9.4 % 1.02 [ 0.80, 1.30 ]Roflumilast M2-124+M2-125 1040/1537 963/1554 22.3 % 1.28 [ 1.11, 1.49 ]Roflumilast M2-127 294/466 276/467 7.3 % 1.18 [ 0.91, 1.54 ]Roflumilast M2-128 172/374 150/369 5.9 % 1.24 [ 0.93, 1.66 ]Subtotal (95% CI) 3899 3633 53.9 % 1.20 [ 1.09, 1.32 ]Total events: 2583 (PDE4i Treatment), 2258 (Control)Heterogeneity: Chi 2 = 6.10, df = 6 (P = 0.41); I 2 =2%Test for overall effect: Z = 3.74 (P = 0.00019)2 Roflumilast 250 gRoflumilast FK1 101 85/175 85/172 3.2 % 0.97 [ 0.63, 1.47 ]Roflumilast M2-107 382/576 174/280 5.7 % 1.20 [ 0.89, 1.61 ]Subtotal (95% CI) 751 452 8.9 % 1.12 [ 0.88, 1.42 ]Total events: 467 (PDE4i Treatment), 259 (Control)Heterogeneity: Chi 2 = 0.67, df = 1 (P = 0.41); I 2 =0.0%Test for overall effect: Z = 0.89 (P = 0.38)3 Cilomilast 15 mgCilomilast 039 373/431 176/216 2.3 % 1.46 [ 0.94, 2.27 ]Cilomilast 042 340/474 154/226 4.3 % 1.19 [ 0.84, 1.67 ]Cilomilast 076 22/29 20/30 0.3 % 1.57 [ 0.50, 4.91 ]Cilomilast 091 320/469 166/242 5.0 % 0.98 [ 0.70, 1.37 ]Cilomilast 103657 234/296 232/316 3.4 % 1.37 [ 0.94, 1.99 ]Cilomilast 110 24/31 20/34 0.3 % 2.40 [ 0.81, 7.10 ]0.1 0.2 0.5 1 2 5 10Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.80

(... Continued)Study or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CICilomilast 111 63/79 49/77 0.7 % 2.25 [ 1.10, 4.62 ]Cilomilast 121 518/678 245/340 5.6 % 1.26 [ 0.93, 1.69 ]Cilomilast 156 364/418 339/407 3.2 % 1.35 [ 0.92, 1.99 ]Cilomilast 157 351/455 343/452 5.7 % 1.07 [ 0.79, 1.46 ]Cilomilast 168 136/203 76/103 2.4 % 0.72 [ 0.43, 1.22 ]Cilomilast 180 65/97 68/102 1.6 % 1.02 [ 0.56, 1.83 ]Cilomilast 181 41/65 36/62 1.0 % 1.23 [ 0.60, 2.52 ]Compton 2001 64/105 55/105 1.5 % 1.42 [ 0.82, 2.45 ]Subtotal (95% CI) 3830 2712 37.3 % 1.21 [ 1.08, 1.36 ]Total events: 2915 (PDE4i Treatment), 1979 (Control)Heterogeneity: Chi 2 = 12.53, df = 13 (P = 0.48); I 2 =0.0%Test for overall effect: Z = 3.19 (P = 0.0014)Total (95% CI) 8480 6797 100.0 % 1.20 [ 1.11, 1.28 ]Total events: 5965 (PDE4i Treatment), 4496 (Control)Heterogeneity: Chi 2 = 19.65, df = 22 (P = 0.60); I 2 =0.0%Test for overall effect: Z = 4.95 (P < 0.00001)Test for subgroup differences: Chi 2 = 0.35, df = 2 (P = 0.84), I 2 =0.0%0.1 0.2 0.5 1 2 5 10Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.81

Analysis 6.2.Comparison 6 Adverse effects, Outcome 2 No of patients experiencing an adverse event(Roflumilast 500 µg vs 250 µg).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome: 2 No of patients experiencing an adverse event (Roflumilast 500 g vs 250 g)Study or subgroup Roflumilast 500mcg Roflumilast 250mcg Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CIRoflumilast FK1 101 82/169 85/175 25.6 % 1.00 [ 0.65, 1.52 ]Roflumilast M2-107 370/555 382/576 74.4 % 1.02 [ 0.79, 1.30 ]Total (95% CI) 724 751 100.0 % 1.01 [ 0.82, 1.25 ]Total events: 452 (Roflumilast 500mcg), 467 (Roflumilast 250mcg)Heterogeneity: Chi 2 = 0.00, df = 1 (P = 0.94); I 2 =0.0%Test for overall effect: Z = 0.10 (P = 0.92)0.5 0.7 1 1.5 2Favours 500mcgFavours 250mcgAnalysis 6.3.Comparison 6 Adverse effects, Outcome 3 Diarrhoea.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:3 DiarrhoeaStudy or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 RoflumilastRoflumilast M2-107 50/555 6/280 3.0 % 4.52 [ 1.91, 10.68 ]Roflumilast M2-112 71/760 20/753 7.5 % 3.78 [ 2.27, 6.27 ]Roflumilast M2-124 63/769 26/755 10.0 % 2.50 [ 1.57, 4.00 ]Roflumilast M2-125 67/778 23/790 8.6 % 3.14 [ 1.94, 5.10 ]Roflumilast M2-127 38/466 16/467 6.1 % 2.50 [ 1.37, 4.56 ]Roflumilast M2-128 33/374 2/369 0.8 % 17.76 [ 4.23, 74.57 ]Subtotal (95% CI) 3702 3414 36.0 % 3.41 [ 2.69, 4.33 ]Total events: 322 (PDE4i Treatment), 93 (Control)0.1 0.2 0.5 1 2 5 10Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.82

(... Continued)Study or subgroup PDE4i Treatment Control Odds Ratio Weight Odds RatioHeterogeneity: Chi 2 = 8.47, df = 5 (P = 0.13); I 2 =41%Test for overall effect: Z = 10.15 (P < 0.00001)2 Cilomilastn/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CICilomilast 039 90/431 23/216 10.0 % 2.21 [ 1.36, 3.62 ]Cilomilast 042 38/474 8/226 4.1 % 2.38 [ 1.09, 5.18 ]Cilomilast 076 6/29 4/30 1.3 % 1.70 [ 0.42, 6.77 ]Cilomilast 091 47/469 10/242 4.9 % 2.58 [ 1.28, 5.21 ]Cilomilast 103657 50/296 20/316 6.7 % 3.01 [ 1.74, 5.19 ]Cilomilast 110 4/31 2/34 0.7 % 2.37 [ 0.40, 13.96 ]Cilomilast 111 14/79 3/77 1.0 % 5.31 [ 1.46, 19.31 ]Cilomilast 121 94/678 20/340 9.5 % 2.58 [ 1.56, 4.25 ]Cilomilast 156 82/418 44/407 14.8 % 2.01 [ 1.36, 2.99 ]Cilomilast 157 33/455 13/452 5.0 % 2.64 [ 1.37, 5.09 ]Cilomilast 168 28/203 7/103 3.3 % 2.19 [ 0.92, 5.21 ]Cilomilast 180 9/97 3/102 1.1 % 3.38 [ 0.89, 12.86 ]Cilomilast 181 5/65 2/62 0.8 % 2.50 [ 0.47, 13.39 ]Compton 2001 9/105 2/105 0.8 % 4.83 [ 1.02, 22.91 ]Subtotal (95% CI) 3830 2712 64.0 % 2.47 [ 2.05, 2.98 ]Total events: 509 (PDE4i Treatment), 161 (Control)Heterogeneity: Chi 2 = 4.44, df = 13 (P = 0.99); I 2 =0.0%Test for overall effect: Z = 9.40 (P < 0.00001)Total (95% CI) 7532 6126 100.0 % 2.81 [ 2.42, 3.26 ]Total events: 831 (PDE4i Treatment), 254 (Control)Heterogeneity: Chi 2 = 15.84, df = 19 (P = 0.67); I 2 =0.0%Test for overall effect: Z = 13.74 (P < 0.00001)Test for subgroup differences: Chi 2 = 4.37, df = 1 (P = 0.04), I 2 =77%0.1 0.2 0.5 1 2 5 10Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.83

Analysis 6.4.Comparison 6 Adverse effects, Outcome 4 Nausea.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:4 NauseaStudy or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 Roflumilast 500 gRoflumilast M2-107 27/555 2/280 1.8 % 7.11 [ 1.68, 30.11 ]Roflumilast M2-112 38/760 10/753 6.8 % 3.91 [ 1.93, 7.91 ]Roflumilast M2-124 41/769 15/755 10.1 % 2.78 [ 1.52, 5.06 ]Roflumilast M2-125 21/778 15/790 10.2 % 1.43 [ 0.73, 2.80 ]Roflumilast M2-127 25/466 1/467 0.7 % 26.42 [ 3.56, 195.79 ]Roflumilast M2-128 11/374 4/369 2.8 % 2.77 [ 0.87, 8.76 ]Subtotal (95% CI) 3702 3414 32.4 % 3.32 [ 2.38, 4.61 ]Total events: 163 (PDE4i Treatment), 47 (Control)Heterogeneity: Chi 2 = 11.85, df = 5 (P = 0.04); I 2 =58%Test for overall effect: Z = 7.11 (P < 0.00001)2 Roflumilast 250 gRoflumilast M2-107 16/576 2/280 1.9 % 3.97 [ 0.91, 17.39 ]Subtotal (95% CI) 576 280 1.9 % 3.97 [ 0.91, 17.39 ]Total events: 16 (PDE4i Treatment), 2 (Control)Heterogeneity: not applicableTest for overall effect: Z = 1.83 (P = 0.067)3 Cilomilast 15 mgCilomilast 039 82/431 10/216 7.6 % 4.84 [ 2.46, 9.54 ]Cilomilast 042 56/474 9/226 7.6 % 3.23 [ 1.57, 6.65 ]Cilomilast 076 3/29 2/30 1.2 % 1.62 [ 0.25, 10.45 ]Cilomilast 091 34/469 9/242 7.8 % 2.02 [ 0.95, 4.29 ]Cilomilast 103657 63/296 12/316 6.5 % 6.85 [ 3.61, 13.00 ]Cilomilast 110 4/31 1/34 0.6 % 4.89 [ 0.52, 46.36 ]Cilomilast 111 13/79 4/77 2.4 % 3.59 [ 1.12, 11.57 ]Cilomilast 121 39/678 3/340 2.7 % 6.86 [ 2.10, 22.35 ]Cilomilast 156 109/418 27/407 14.3 % 4.96 [ 3.17, 7.76 ]Cilomilast 157 44/455 14/452 9.0 % 3.35 [ 1.81, 6.20 ]Cilomilast 168 28/203 2/103 1.6 % 8.08 [ 1.89, 34.63 ]0.001 0.01 0.1 1 10 100 1000Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.84

(... Continued)Study or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CICilomilast 180 11/97 2/102 1.2 % 6.40 [ 1.38, 29.65 ]Cilomilast 181 6/65 4/62 2.6 % 1.47 [ 0.40, 5.50 ]Compton 2001 12/105 1/105 0.6 % 13.42 [ 1.71, 105.19 ]Subtotal (95% CI) 3830 2712 65.8 % 4.37 [ 3.49, 5.47 ]Total events: 504 (PDE4i Treatment), 100 (Control)Heterogeneity: Chi 2 = 14.16, df = 13 (P = 0.36); I 2 =8%Test for overall effect: Z = 12.87 (P < 0.00001)Total (95% CI) 8108 6406 100.0 % 4.02 [ 3.35, 4.84 ]Total events: 683 (PDE4i Treatment), 149 (Control)Heterogeneity: Chi 2 = 29.21, df = 20 (P = 0.08); I 2 =32%Test for overall effect: Z = 14.82 (P < 0.00001)Test for subgroup differences: Chi 2 = 1.84, df = 2 (P = 0.40), I 2 =0.0%0.001 0.01 0.1 1 10 100 1000Favours treatmentFavours controlAnalysis 6.5.Comparison 6 Adverse effects, Outcome 5 Headache.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:5 HeadacheStudy or subgroup Favours treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 Roflumilast 500 gRoflumilast FK1 101 6/169 4/172 1.7 % 1.55 [ 0.43, 5.58 ]Roflumilast M2-107 10/555 1/280 0.6 % 5.12 [ 0.65, 40.19 ]Roflumilast M2-112 47/760 18/753 7.7 % 2.69 [ 1.55, 4.68 ]Roflumilast M2-124 26/769 17/755 7.5 % 1.52 [ 0.82, 2.82 ]Roflumilast M2-125 25/778 8/790 3.5 % 3.25 [ 1.45, 7.24 ]Roflumilast M2-127 14/466 5/467 2.2 % 2.86 [ 1.02, 8.01 ]Roflumilast M2-128 8/374 0/369 0.2 % 17.14 [ 0.99, 298.03 ]Subtotal (95% CI) 3871 3586 23.4 % 2.53 [ 1.83, 3.49 ]0.01 0.1 1 10 100Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.85

(... Continued)Study or subgroup Favours treatment Control Odds Ratio Weight Odds RatioTotal events: 136 (Favours treatment), 53 (Control)Heterogeneity: Chi 2 = 5.81, df = 6 (P = 0.44); I 2 =0.0%Test for overall effect: Z = 5.66 (P < 0.00001)2 Roflumilast 250 gn/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CIRoflumilast FK1 101 4/175 4/172 1.8 % 0.98 [ 0.24, 3.99 ]Subtotal (95% CI) 175 172 1.8 % 0.98 [ 0.24, 3.99 ]Total events: 4 (Favours treatment), 4 (Control)Heterogeneity: not applicableTest for overall effect: Z = 0.02 (P = 0.98)3 Cilomilast 15 mgCilomilast 039 36/431 14/216 7.7 % 1.32 [ 0.69, 2.49 ]Cilomilast 042 44/474 14/226 7.8 % 1.55 [ 0.83, 2.89 ]Cilomilast 076 5/29 1/30 0.4 % 6.04 [ 0.66, 55.30 ]Cilomilast 091 21/469 15/242 8.6 % 0.71 [ 0.36, 1.40 ]Cilomilast 103657 47/296 34/316 12.5 % 1.57 [ 0.98, 2.51 ]Cilomilast 156 46/418 33/407 13.5 % 1.40 [ 0.88, 2.24 ]Cilomilast 157 35/455 22/452 9.2 % 1.63 [ 0.94, 2.82 ]Cilomilast 168 8/203 4/103 2.3 % 1.02 [ 0.30, 3.45 ]Cilomilast 180 16/97 14/102 5.2 % 1.24 [ 0.57, 2.70 ]Cilomilast 181 9/65 12/62 4.8 % 0.67 [ 0.26, 1.72 ]Compton 2001 7/105 7/105 3.0 % 1.00 [ 0.34, 2.96 ]Subtotal (95% CI) 3042 2261 74.8 % 1.32 [ 1.08, 1.62 ]Total events: 274 (Favours treatment), 170 (Control)Heterogeneity: Chi 2 = 8.82, df = 10 (P = 0.55); I 2 =0.0%Test for overall effect: Z = 2.69 (P = 0.0071)Total (95% CI) 7088 6019 100.0 % 1.60 [ 1.35, 1.89 ]Total events: 414 (Favours treatment), 227 (Control)Heterogeneity: Chi 2 = 24.43, df = 18 (P = 0.14); I 2 =26%Test for overall effect: Z = 5.43 (P < 0.00001)Test for subgroup differences: Chi 2 = 11.63, df = 2 (P = 0.00), I 2 =83%0.01 0.1 1 10 100Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.86

Analysis 6.6.Comparison 6 Adverse effects, Outcome 6 Vomiting.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:6 VomitingStudy or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ratio1 Roflumilastn/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CIRoflumilast M2-107 1/555 0/280 1.7 % 1.52 [ 0.06, 37.37 ]Subtotal (95% CI) 555 280 1.7 % 1.52 [ 0.06, 37.37 ]Total events: 1 (PDE4i Treatment), 0 (Control)Heterogeneity: not applicableTest for overall effect: Z = 0.26 (P = 0.80)2 CilomilastCilomilast 039 37/431 6/216 19.1 % 3.29 [ 1.37, 7.91 ]Cilomilast 042 17/474 0/226 1.7 % 17.33 [ 1.04, 289.44 ]Cilomilast 076 4/29 0/30 1.1 % 10.76 [ 0.55, 209.55 ]Cilomilast 091 22/369 1/242 3.0 % 15.28 [ 2.05, 114.12 ]Cilomilast 103657 28/296 8/316 18.3 % 4.02 [ 1.80, 8.98 ]Cilomilast 156 33/418 11/407 26.8 % 3.09 [ 1.54, 6.19 ]Cilomilast 157 21/455 8/452 20.0 % 2.69 [ 1.18, 6.13 ]Cilomilast 168 9/203 1/103 3.3 % 4.73 [ 0.59, 37.87 ]Cilomilast 180 4/97 1/102 2.4 % 4.34 [ 0.48, 39.57 ]Cilomilast 181 5/65 1/62 2.5 % 5.08 [ 0.58, 44.81 ]Subtotal (95% CI) 2837 2156 98.3 % 4.06 [ 2.83, 5.82 ]Total events: 180 (PDE4i Treatment), 37 (Control)Heterogeneity: Chi 2 = 4.95, df = 9 (P = 0.84); I 2 =0.0%Test for overall effect: Z = 7.60 (P < 0.00001)Total (95% CI) 3392 2436 100.0 % 4.01 [ 2.80, 5.74 ]Total events: 181 (PDE4i Treatment), 37 (Control)Heterogeneity: Chi 2 = 5.24, df = 10 (P = 0.87); I 2 =0.0%Test for overall effect: Z = 7.59 (P < 0.00001)Test for subgroup differences: Chi 2 = 0.36, df = 1 (P = 0.55), I 2 =0.0%0.01 0.1 1 10 100Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.87

Analysis 6.7.Comparison 6 Adverse effects, Outcome 7 Dyspepsia.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:7 DyspepsiaStudy or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 CilomilastCilomilast 039 42/431 8/216 17.1 % 2.81 [ 1.29, 6.09 ]Cilomilast 042 29/474 8/226 18.1 % 1.78 [ 0.80, 3.95 ]Cilomilast 076 5/29 1/30 1.4 % 6.04 [ 0.66, 55.30 ]Cilomilast 103657 14/296 5/316 8.2 % 3.09 [ 1.10, 8.68 ]Cilomilast 110 4/31 0/34 0.7 % 11.29 [ 0.58, 218.85 ]Cilomilast 111 10/79 2/77 3.1 % 5.43 [ 1.15, 25.68 ]Cilomilast 121 40/678 4/340 8.9 % 5.27 [ 1.87, 14.84 ]Cilomilast 156 38/418 10/407 16.4 % 3.97 [ 1.95, 8.08 ]Cilomilast 157 18/455 6/452 10.3 % 3.06 [ 1.20, 7.79 ]Cilomilast 168 15/203 6/103 13.1 % 1.29 [ 0.49, 3.43 ]Cilomilast 180 4/97 1/102 1.7 % 4.34 [ 0.48, 39.57 ]Cilomilast 181 3/65 0/62 0.9 % 7.00 [ 0.35, 138.35 ]Total (95% CI) 3256 2365 100.0 % 3.13 [ 2.30, 4.27 ]Total events: 222 (PDE4i Treatment), 51 (Control)Heterogeneity: Chi 2 = 8.48, df = 11 (P = 0.67); I 2 =0.0%Test for overall effect: Z = 7.26 (P < 0.00001)0.01 0.1 1 10 100Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.88

Analysis 6.8.Comparison 6 Adverse effects, Outcome 8 Abdominal pain.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:8 Abdominal painStudy or subgroup PDE4i Treamtnet Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 CilomilastCilomilast 039 61/431 20/216 21.0 % 1.62 [ 0.95, 2.76 ]Cilomilast 042 37/474 10/226 11.5 % 1.83 [ 0.89, 3.75 ]Cilomilast 076 3/29 0/30 0.4 % 8.06 [ 0.40, 163.21 ]Cilomilast 091 59/469 9/242 9.5 % 3.73 [ 1.81, 7.65 ]Cilomilast 110 3/31 1/34 0.8 % 3.54 [ 0.35, 35.93 ]Cilomilast 111 6/79 0/77 0.4 % 13.71 [ 0.76, 247.65 ]Cilomilast 121 36/678 10/340 11.6 % 1.85 [ 0.91, 3.78 ]Cilomilast 156 52/418 38/407 31.0 % 1.38 [ 0.89, 2.15 ]Cilomilast 157 13/455 5/452 4.5 % 2.63 [ 0.93, 7.44 ]Cilomilast 168 19/203 6/103 6.6 % 1.67 [ 0.65, 4.32 ]Compton 2001 8/105 3/105 2.5 % 2.80 [ 0.72, 10.88 ]Total (95% CI) 3372 2232 100.0 % 1.97 [ 1.55, 2.49 ]Total events: 297 (PDE4i Treamtnet), 102 (Control)Heterogeneity: Chi 2 = 9.58, df = 10 (P = 0.48); I 2 =0.0%Test for overall effect: Z = 5.63 (P < 0.00001)0.01 0.1 1 10 100Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.89

Analysis 6.9.Comparison 6 Adverse effects, Outcome 9 Weight Loss.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:9 Weight LossStudy or subgroup Experimental Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 RoflumilastRoflumilast M2-124 92/769 24/755 46.4 % 4.14 [ 2.61, 6.56 ]Roflumilast M2-125 65/778 20/790 39.6 % 3.51 [ 2.10, 5.85 ]Roflumilast M2-127 40/466 5/467 9.9 % 8.68 [ 3.39, 22.19 ]Roflumilast M2-128 21/374 2/369 4.1 % 10.92 [ 2.54, 46.90 ]Total (95% CI) 2387 2381 100.0 % 4.62 [ 3.38, 6.31 ]Total events: 218 (Experimental), 51 (Control)Heterogeneity: Chi 2 = 4.40, df = 3 (P = 0.22); I 2 =32%Test for overall effect: Z = 9.64 (P < 0.00001)0.01 0.1 1 10 100Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.90

Analysis 6.10.Comparison 6 Adverse effects, Outcome 10 Influenza-like symptoms.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:10 Influenza-like symptomsStudy or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ratio1 Roflumilast 500 gn/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CIRoflumilast FK1 101 7/169 1/172 0.9 % 7.39 [ 0.90, 60.72 ]Roflumilast M2-112 40/760 34/753 32.3 % 1.17 [ 0.74, 1.88 ]Roflumilast M2-124 27/769 18/755 17.5 % 1.49 [ 0.81, 2.73 ]Roflumilast M2-125 12/778 20/790 19.5 % 0.60 [ 0.29, 1.24 ]Roflumilast M2-127 9/466 11/467 10.8 % 0.82 [ 0.34, 1.99 ]Roflumilast M2-128 3/374 0/369 0.5 % 6.96 [ 0.36, 135.26 ]Subtotal (95% CI) 3316 3306 81.6 % 1.17 [ 0.87, 1.57 ]Total events: 98 (PDE4i Treatment), 84 (Control)Heterogeneity: Chi 2 = 8.79, df = 5 (P = 0.12); I 2 =43%Test for overall effect: Z = 1.02 (P = 0.31)2 Roflumilast 250 gRoflumilast FK1 101 2/175 1/172 1.0 % 1.98 [ 0.18, 22.00 ]Subtotal (95% CI) 175 172 1.0 % 1.98 [ 0.18, 22.00 ]Total events: 2 (PDE4i Treatment), 1 (Control)Heterogeneity: not applicableTest for overall effect: Z = 0.55 (P = 0.58)3 Cilomilast 15 mgCilomilast 076 2/29 1/30 0.9 % 2.15 [ 0.18, 25.07 ]Cilomilast 157 14/455 17/452 16.5 % 0.81 [ 0.40, 1.67 ]Subtotal (95% CI) 484 482 17.4 % 0.88 [ 0.44, 1.75 ]Total events: 16 (PDE4i Treatment), 18 (Control)Heterogeneity: Chi 2 = 0.55, df = 1 (P = 0.46); I 2 =0.0%Test for overall effect: Z = 0.36 (P = 0.72)Total (95% CI) 3975 3960 100.0 % 1.12 [ 0.86, 1.47 ]Total events: 116 (PDE4i Treatment), 103 (Control)Heterogeneity: Chi 2 = 10.00, df = 8 (P = 0.27); I 2 =20%Test for overall effect: Z = 0.86 (P = 0.39)Test for subgroup differences: Chi 2 = 0.75, df = 2 (P = 0.69), I 2 =0.0%0.01 0.1 1 10 100Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.91

Analysis 6.11.Comparison 6 Adverse effects, Outcome 11 Upper Respiratory Tract Infection.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:11 Upper Respiratory Tract InfectionStudy or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 Roflumilast 500 gRoflumilast FK1 101 26/169 27/172 6.5 % 0.98 [ 0.54, 1.75 ]Roflumilast M2-107 23/555 14/280 5.1 % 0.82 [ 0.42, 1.62 ]Roflumilast M2-124 16/769 21/755 5.9 % 0.74 [ 0.38, 1.43 ]Roflumilast M2-125 33/778 38/790 10.3 % 0.88 [ 0.54, 1.41 ]Roflumilast M2-127 9/466 19/467 5.3 % 0.46 [ 0.21, 1.04 ]Roflumilast M2-128 4/374 2/369 0.6 % 1.98 [ 0.36, 10.90 ]Subtotal (95% CI) 3111 2833 33.7 % 0.82 [ 0.63, 1.07 ]Total events: 111 (PDE4i Treatment), 121 (Control)Heterogeneity: Chi 2 = 3.46, df = 5 (P = 0.63); I 2 =0.0%Test for overall effect: Z = 1.47 (P = 0.14)2 Roflumilast 250 gRoflumilast FK1 101 22/175 27/172 6.8 % 0.77 [ 0.42, 1.42 ]Roflumilast M2-107 27/576 14/280 5.1 % 0.93 [ 0.48, 1.81 ]Subtotal (95% CI) 751 452 11.9 % 0.84 [ 0.54, 1.31 ]Total events: 49 (PDE4i Treatment), 41 (Control)Heterogeneity: Chi 2 = 0.17, df = 1 (P = 0.68); I 2 =0.0%Test for overall effect: Z = 0.76 (P = 0.45)3 Cilomilast 15 mgCilomilast 039 68/431 33/216 10.6 % 1.04 [ 0.66, 1.63 ]Cilomilast 042 29/474 11/226 4.0 % 1.27 [ 0.62, 2.60 ]Cilomilast 076 4/29 1/30 0.2 % 4.64 [ 0.49, 44.27 ]Cilomilast 091 14/469 7/242 2.6 % 1.03 [ 0.41, 2.59 ]Cilomilast 103657 52/296 49/316 11.2 % 1.16 [ 0.76, 1.78 ]Cilomilast 110 3/31 4/34 1.0 % 0.80 [ 0.17, 3.91 ]Cilomilast 111 8/79 5/77 1.3 % 1.62 [ 0.51, 5.20 ]Cilomilast 156 46/418 61/407 15.7 % 0.70 [ 0.47, 1.06 ]Cilomilast 168 18/203 18/103 6.2 % 0.46 [ 0.23, 0.93 ]Cilomilast 180 2/97 6/102 1.6 % 0.34 [ 0.07, 1.71 ]0.005 0.1 1 10 200Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.92

(... Continued)Study or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CISubtotal (95% CI) 2527 1753 54.4 % 0.92 [ 0.75, 1.13 ]Total events: 244 (PDE4i Treatment), 195 (Control)Heterogeneity: Chi 2 = 12.11, df = 9 (P = 0.21); I 2 =26%Test for overall effect: Z = 0.78 (P = 0.44)Total (95% CI) 6389 5038 100.0 % 0.88 [ 0.75, 1.02 ]Total events: 404 (PDE4i Treatment), 357 (Control)Heterogeneity: Chi 2 = 16.18, df = 17 (P = 0.51); I 2 =0.0%Test for overall effect: Z = 1.68 (P = 0.093)Test for subgroup differences: Chi 2 = 0.52, df = 2 (P = 0.77), I 2 =0.0%0.005 0.1 1 10 200Favours treatmentFavours controlAnalysis 6.12.Comparison 6 Adverse effects, Outcome 12 Withdrawals due to adverse effects.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:12 Withdrawals due to adverse effectsStudy or subgroup PDE4i Treatment Control Odds Ratio Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 Roflumilast 500 gGrootendorst 2007 3/38 0/38 7.59 [ 0.38, 152.19 ]Roflumilast FK1 101 12/169 10/172 1.24 [ 0.52, 2.95 ]Roflumilast M2-107 84/555 23/280 1.99 [ 1.23, 3.24 ]Roflumilast M2-112 103/760 56/753 1.95 [ 1.39, 2.75 ]Roflumilast M2-124 119/766 78/759 1.61 [ 1.18, 2.18 ]Roflumilast M2-125 101/773 83/798 1.29 [ 0.95, 1.76 ]Roflumilast M2-127 77/466 45/467 1.86 [ 1.25, 2.75 ]Roflumilast M2-128 33/371 20/372 1.72 [ 0.97, 3.05 ]Subtotal (95% CI) 3898 3639 1.66 [ 1.43, 1.93 ]Total events: 532 (PDE4i Treatment), 315 (Control)Heterogeneity: Chi 2 = 5.69, df = 7 (P = 0.58); I 2 =0.0%Test for overall effect: Z = 6.70 (P < 0.00001)0.01 0.1 1 10 100Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.93

(... Continued)Study or subgroup PDE4i Treatment Control Odds Ratio Odds Ratio2 Roflumilast 250 gn/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CIRoflumilast FK1 101 13/175 10/172 1.30 [ 0.55, 3.05 ]Subtotal (95% CI) 175 172 1.30 [ 0.55, 3.05 ]Total events: 13 (PDE4i Treatment), 10 (Control)Heterogeneity: not applicableTest for overall effect: Z = 0.60 (P = 0.55)3 Cilomilast 15 mgCilomilast 039 94/431 35/216 1.44 [ 0.94, 2.21 ]Cilomilast 042 71/474 22/226 1.63 [ 0.98, 2.71 ]Cilomilast 076 4/29 0/30 10.76 [ 0.55, 209.55 ]Cilomilast 091 78/469 32/242 1.31 [ 0.84, 2.04 ]Cilomilast 103657 46/296 17/316 3.24 [ 1.81, 5.79 ]Cilomilast 110 0/31 0/34 0.0 [ 0.0, 0.0 ]Cilomilast 111 14/79 8/77 1.86 [ 0.73, 4.72 ]Cilomilast 121 49/678 8/340 3.23 [ 1.51, 6.91 ]Cilomilast 156 80/418 42/407 2.06 [ 1.38, 3.07 ]Cilomilast 157 80/455 46/452 1.88 [ 1.28, 2.78 ]Cilomilast 168 36/203 8/103 2.56 [ 1.14, 5.73 ]Cilomilast 180 9/97 6/102 1.64 [ 0.56, 4.78 ]Cilomilast 181 6/65 3/62 2.00 [ 0.48, 8.38 ]Compton 2001 14/107 8/106 1.84 [ 0.74, 4.60 ]Subtotal (95% CI) 3832 2713 1.90 [ 1.61, 2.24 ]Total events: 581 (PDE4i Treatment), 235 (Control)Heterogeneity: Chi 2 = 11.81, df = 12 (P = 0.46); I 2 =0.0%Test for overall effect: Z = 7.72 (P < 0.00001)Total (95% CI) 7905 6524 1.76 [ 1.58, 1.96 ]Total events: 1126 (PDE4i Treatment), 560 (Control)Heterogeneity: Chi 2 = 19.17, df = 21 (P = 0.57); I 2 =0.0%Test for overall effect: Z = 10.18 (P < 0.00001)Test for subgroup differences: Chi 2 = 1.90, df = 2 (P = 0.39), I 2 =0.0%0.01 0.1 1 10 100Favours treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.94

Analysis 6.13.Comparison 6 Adverse effects, Outcome 13 Non fatal serious adverse events.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:13 Non fatal serious adverse eventsStudy or subgroup PDE4i Treatment Control Odds Ratio Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 Roflumilast 500 gGrootendorst 2007 0/38 0/38 0.0 [ 0.0, 0.0 ]Roflumilast FK1 101 9/169 11/172 0.82 [ 0.33, 2.04 ]Roflumilast M2-107 53/555 21/280 1.30 [ 0.77, 2.21 ]Roflumilast M2-112 137/760 132/753 1.03 [ 0.79, 1.35 ]Roflumilast M2-124+M2-125 301/1537 336/1554 0.88 [ 0.74, 1.05 ]Subtotal (95% CI) 3059 2797 0.95 [ 0.82, 1.09 ]Total events: 500 (PDE4i Treatment), 500 (Control)Heterogeneity: Chi 2 = 2.55, df = 3 (P = 0.47); I 2 =0.0%Test for overall effect: Z = 0.77 (P = 0.44)2 Roflumilast 250 gRoflumilast FK1 101 14/175 11/172 1.27 [ 0.56, 2.89 ]Subtotal (95% CI) 175 172 1.27 [ 0.56, 2.89 ]Total events: 14 (PDE4i Treatment), 11 (Control)Heterogeneity: not applicableTest for overall effect: Z = 0.58 (P = 0.56)3 Cilomilast 15 mgCilomilast 039 24/431 25/216 0.45 [ 0.25, 0.81 ]Cilomilast 042 28/474 13/226 1.03 [ 0.52, 2.03 ]Cilomilast 076 2/29 1/30 2.15 [ 0.18, 25.07 ]Cilomilast 091 23/469 19/242 0.61 [ 0.32, 1.13 ]Cilomilast 103657 17/296 23/316 0.78 [ 0.41, 1.48 ]Cilomilast 110 1/31 0/34 3.39 [ 0.13, 86.43 ]Cilomilast 111 8/79 2/77 4.23 [ 0.87, 20.58 ]Cilomilast 121 38/678 10/340 1.96 [ 0.96, 3.98 ]Cilomilast 156 25/418 38/407 0.62 [ 0.37, 1.04 ]Cilomilast 157 53/455 55/452 0.95 [ 0.64, 1.42 ]Cilomilast 168 7/203 4/103 0.88 [ 0.25, 3.09 ]Cilomilast 180 6/97 7/102 0.89 [ 0.29, 2.76 ]0.002 0.1 1 10 500Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.95

(... Continued)Study or subgroup PDE4i Treatment Control Odds Ratio Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CICilomilast 181 2/65 1/62 1.94 [ 0.17, 21.91 ]Compton 2001 4/107 3/106 1.33 [ 0.29, 6.11 ]Subtotal (95% CI) 3832 2713 0.87 [ 0.72, 1.06 ]Total events: 238 (PDE4i Treatment), 201 (Control)Heterogeneity: Chi 2 = 19.12, df = 13 (P = 0.12); I 2 =32%Test for overall effect: Z = 1.34 (P = 0.18)Total (95% CI) 7066 5682 0.93 [ 0.83, 1.04 ]Total events: 752 (PDE4i Treatment), 712 (Control)Heterogeneity: Chi 2 = 22.95, df = 18 (P = 0.19); I 2 =22%Test for overall effect: Z = 1.31 (P = 0.19)Test for subgroup differences: Chi 2 = 1.01, df = 2 (P = 0.60), I 2 =0.0%0.002 0.1 1 10 500Favours treatmentFavours controlAnalysis 6.14.Comparison 6 Adverse effects, Outcome 14 Mortality.Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:14 MortalityStudy or subgroup PDE4i Treatment Control Odds Ratio Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 RoflumilastRoflumilast M2-107 5/555 2/280 1.26 [ 0.24, 6.55 ]Roflumilast M2-112 12/760 20/753 0.59 [ 0.29, 1.21 ]Roflumilast M2-124 17/765 17/758 0.99 [ 0.50, 1.96 ]Roflumilast M2-125 25/772 25/796 1.03 [ 0.59, 1.81 ]Subtotal (95% CI) 2852 2587 0.89 [ 0.62, 1.28 ]Total events: 59 (PDE4i Treatment), 64 (Control)Heterogeneity: Chi 2 = 1.80, df = 3 (P = 0.62); I 2 =0.0%Test for overall effect: Z = 0.63 (P = 0.53)2 CilomilastCilomilast 039 2/431 2/216 0.50 [ 0.07, 3.57 ]0.005 0.1 1 10 200Favour treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.96

(... Continued)Study or subgroup PDE4i Treatment Control Odds Ratio Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CICilomilast 042 2/474 2/226 0.47 [ 0.07, 3.39 ]Cilomilast 076 0/29 0/30 0.0 [ 0.0, 0.0 ]Cilomilast 091 3/469 0/242 3.64 [ 0.19, 70.73 ]Cilomilast 103657 0/296 0/316 0.0 [ 0.0, 0.0 ]Cilomilast 110 0/31 0/34 0.0 [ 0.0, 0.0 ]Cilomilast 111 2/79 0/77 5.00 [ 0.24, 105.86 ]Cilomilast 121 1/678 2/340 0.25 [ 0.02, 2.76 ]Cilomilast 156 0/418 0/407 0.0 [ 0.0, 0.0 ]Cilomilast 157 4/455 8/452 0.49 [ 0.15, 1.65 ]Cilomilast 168 0/203 0/103 0.0 [ 0.0, 0.0 ]Cilomilast 180 0/97 0/102 0.0 [ 0.0, 0.0 ]Cilomilast 181 0/65 0/62 0.0 [ 0.0, 0.0 ]Subtotal (95% CI) 3725 2607 0.70 [ 0.34, 1.45 ]Total events: 14 (PDE4i Treatment), 14 (Control)Heterogeneity: Chi 2 = 4.08, df = 5 (P = 0.54); I 2 =0.0%Test for overall effect: Z = 0.95 (P = 0.34)Total (95% CI) 6577 5194 0.85 [ 0.62, 1.17 ]Total events: 73 (PDE4i Treatment), 78 (Control)Heterogeneity: Chi 2 = 6.49, df = 9 (P = 0.69); I 2 =0.0%Test for overall effect: Z = 0.98 (P = 0.33)Test for subgroup differences: Chi 2 = 0.33, df = 1 (P = 0.56), I 2 =0.0%0.005 0.1 1 10 200Favour treatmentFavours controlPhosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.97

Analysis 6.15.Comparison 6 Adverse effects, Outcome 15 No. of subjects (additional medication).Review:Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary diseaseComparison:6 Adverse effectsOutcome:15 No. of subjects (additional medication)Study or subgroup PDE4i Treatment Control Odds Ratio Weight Odds Ration/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI1 Long acting bronchodilatorsRoflumilast M2-127 74/466 111/467 7.4 % 0.61 [ 0.44, 0.84 ]Roflumilast M2-128 58/374 67/369 4.5 % 0.83 [ 0.56, 1.22 ]Subtotal (95% CI) 840 836 11.9 % 0.69 [ 0.54, 0.88 ]Total events: 132 (PDE4i Treatment), 178 (Control)Heterogeneity: Chi 2 = 1.47, df = 1 (P = 0.23); I 2 =32%Test for overall effect: Z = 2.93 (P = 0.0034)2 CorticosteroidsRoflumilast M2-112 339/760 362/753 16.0 % 0.87 [ 0.71, 1.06 ]Subtotal (95% CI) 760 753 16.0 % 0.87 [ 0.71, 1.06 ]Total events: 339 (PDE4i Treatment), 362 (Control)Heterogeneity: not applicableTest for overall effect: Z = 1.35 (P = 0.18)3 Treatment onlyCilomilast 039 95/431 76/216 6.3 % 0.52 [ 0.36, 0.75 ]Cilomilast 042 176/474 83/226 5.6 % 1.02 [ 0.73, 1.41 ]Cilomilast 076 5/29 4/30 0.3 % 1.35 [ 0.33, 5.64 ]Cilomilast 091 151/469 107/242 7.6 % 0.60 [ 0.44, 0.82 ]Cilomilast 111 15/79 15/77 1.0 % 0.97 [ 0.44, 2.15 ]Cilomilast 121 297/678 178/340 10.6 % 0.71 [ 0.55, 0.92 ]Cilomilast 156 114/418 125/407 7.3 % 0.85 [ 0.63, 1.14 ]Cilomilast 157 192/455 212/452 9.8 % 0.83 [ 0.64, 1.07 ]Cilomilast 168 28/203 18/103 1.6 % 0.76 [ 0.40, 1.44 ]Cilomilast 180 3/97 6/102 0.4 % 0.51 [ 0.12, 2.10 ]Roflumilast FK1 101 19/169 25/172 1.7 % 0.74 [ 0.39, 1.41 ]Roflumilast M2-107 113/555 65/280 5.5 % 0.85 [ 0.60, 1.20 ]Roflumilast M2-124 70/769 82/755 6.0 % 0.82 [ 0.59, 1.15 ]Roflumilast M2-125 87/778 122/790 8.5 % 0.69 [ 0.51, 0.93 ]Subtotal (95% CI) 5604 4192 72.1 % 0.76 [ 0.69, 0.84 ]0.5 0.7 1 1.5 2Favours treatmentFavours control(Continued ...)Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.98

(... Continued)Study or subgroup PDE4i Treatment Control Odds Ratio Weight Odds RatioTotal events: 1365 (PDE4i Treatment), 1118 (Control)Heterogeneity: Chi 2 = 12.82, df = 13 (P = 0.46); I 2 =0.0%n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CITest for overall effect: Z = 5.50 (P < 0.00001)Total (95% CI) 7204 5781 100.0 % 0.77 [ 0.71, 0.83 ]Total events: 1836 (PDE4i Treatment), 1658 (Control)Heterogeneity: Chi 2 = 16.52, df = 16 (P = 0.42); I 2 =3%Test for overall effect: Z = 6.20 (P < 0.00001)Test for subgroup differences: Chi 2 = 2.23, df = 2 (P = 0.33), I 2 =10%0.5 0.7 1 1.5 2Favours treatmentFavours controlA P P E N D I C E SAppendix 1. Airways Group Trials Register search strategy (sensitive search)PDE* or phosphodiesterase* or isoenzyme* or theophylline or rolipram or pentoxifylline or papaverine or milrinone or etazolate oretazolate or dyphylline or dipyridamole or caffeine or amrinone or aminophylline or isobutylxanthine or cilomilast or ariflo or cilostazolor enoximone or milrinone or olprinone or roflumilast or sb207499 or zardaverine or cilostamide or enoximone or trequinsin orTelomilast or IC485 or Oglemilast or QAK423 or GRC-3886 or Arofylline or AWD12-281H I S T O R YProtocol first published: Issue 4, 2000Review first published: Issue 5, 2011C O N T R I B U T I O N S O F A U T H O R SJimmy Chong: Study selection, data extraction, review write up.Phillippa Poole: Protocol initiation and development, study selection, data extraction, review write up.Peter Black: Protocol initiation and development, study selection, data extraction and early review write up.Bonnie Leung: Study selection, data extraction, review write up.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.99

D E C L A R A T I O N S O FNil.I N T E R E S TD I F F E R E N C E S B E T W E E N P R O T O C O L A N D R E V I E WWe added the comparison between published and unpublished results when we discovered the large number of unpublished studies,but before we extracted the data from the studies and carried out the analysis.Phosphodiesterase 4 inhibitors for chronic obstructive pulmonary disease (Review)Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.100