UPDATE OF PAD LITERATURE REVIEW: Inhaled medications for treatment of chronic obstructive pulmonary disease (COPD) Therapeutics Initiative report February 28, 2019
UPDATEOFPADLITERATUREREVIEW:
Inhaledmedicationsfortreatmentofchronicobstructivepulmonarydisease(COPD)
TherapeuticsInitiativereport
February28,2019
2
TableofContents
EXECUTIVESUMMARY 4
BACKGROUND 12
DIAGNOSISANDMANAGEMENTOFCOPD 12
PHARMACEUTICALSERVICESDIVISION(PSD)REQUEST 17
METHODS 18
SEARCHSTRATEGY 18STUDYSELECTION 18DATACOLLECTIONANDANALYSIS 18ASSESSMENTOFRISKOFBIASININCLUDEDSTUDIES 19EVALUATIVEFRAMEWORK 19
RESULTS 20
FINDINGSFROMTHELITERATURE 20SUMMARYOFEXCLUDEDSTUDIES 20DESCRIPTIONOFINCLUDEDSTUDIES 21SUMMARYANDCRITICALAPPRAISALOF1YEARORLONGERSTUDIES 22LABAVS.LAMA 23RISKOFBIASININVIGORATE2013 25OUTCOMESREPORTED 27LABA+LAMAVS.LABA 31RISKOFBIASINDONOHUE2016 34OUTCOMESREPORTED 35LABA+LAMAVS.LAMA 38RISKOFBIASINDYNAGITO2018 40OUTCOMESREPORTED 42RISKOFBIASINSPARK2013 48OUTCOMESREPORTED 50LABA+ICSVS.LABA 53RISKOFBIASINSUMMIT2016 56OUTCOMESREPORTED 60LABA+ICSVS.LAMA 64RISKOFBIASINSARAC2016 65
3
OUTCOMESREPORTED 66LABA+LAMAVS.LABA+ICS 67LABA+LAMA+ICSVS.LABA+LAMA 67LABA+LAMA+ICSVS.LABA+ICS 67
SUMMARY 67
CONCLUSION 69
REFERENCES 70
4
UpdateofPADliteraturereviewofinhaledtherapiesfortreatmentofadultpatientswithchronicobstructivepulmonarydisease(COPD)
ExecutiveSummary
1. BackgroundChronicobstructivepulmonarydisease(COPD)isaprogressiveanddisablingdiseasecharacterizedbyairwayinflammationandairflowlimitationthatisnotfullyreversible.Airflowobstructionleadstosymptomsofwheezing,shortnessofbreath,chesttightness,coughingandproductionofexcessmucus.ThemaintreatmentoptionsforCOPDbelongtoanumberofpharmacologicalclasses:bronchodilators(short-actingbeta2agonists[SABA],long-actingbeta2agonists[LABA],short-actingmuscarinicantagonists[SAMA],andlong-actingmuscarinicantagonists[LAMA]),inhaledcorticosteroids[ICS],andinhibitorsoftheenzymephosphodiesterase-4[PDE4inhibitors].InCanada,approximately20inhaledmedicationsareapprovedtotreatCOPD.DrugstotreatCOPDarelicensedbyregulatoryauthoritiesbasedonshort-termrandomizedtrials(typically12weeksinduration)thatshowanimprovementinthesurrogatemarkerFEV1,whichistheprimaryoutcomemeasureinmosttrials.However,thegoaloftreatingCOPDistopreventacutemoderatetosevereexacerbations,improvequalityoflifeandreducesymptomssuchasdyspneaInTherapeuticsLetter#109publishedinFebruary2018wereportedsystematicreviewsoftheclinicalefficacyofinhaledLABAdrugslicensedforCOPD(formoterol,arformoterolandsalmeterol)ascomparedtoplacebo.Weconcludedthatthe3inhaledlongactingβ2agonists(formoterol,arformoterolandsalmeterol)donotprolongsurvivalorreducetheriskofhospitalization(totalseriousadverseevents)inpatientswithCOPD.Evidenceforimprovementofsymptomscoresisoflowqualityandinsufficienttojustifylong-termuse.Asubsetofpatientsmayderiveclinicallyimportantsymptomaticrelief.Suchpatientscanbeidentifiedonacase-by-casebasisusingashorttherapeutictrial.
TheGlobalInitiativeforChronicObstructiveLungDisease(GOLD)reportandotherclinicalpracticeguidelinesrecommendstepwiseintensificationofdrugtherapyinpeoplewithpersistentbreathlessnessorexacerbations.
In2017,theProvincialAcademicDetailing(PAD)servicecompletedaliteraturereviewandsynopsisofrandomizedcontrolledtrial(RCT)evidencefrom5CochranereviewsthataddressspecificGOLDguidelinerecommendations.ThePADreviewpresentstheoverallgradingofevidenceandtreatmenteffectestimatesasreportedfromtheCochranereviewsinsummarytablesforthefollowingoutcomes:all-causemortality;St.George’sRespiratoryQuestionnaire
5
(SGRQ)totalscore,ameasureofqualityoflife;andnumberofpeoplewith1ormoremoderatetosevereCOPDexacerbation.
2. PharmaceuticalServicesDivision(PSD)Request
ThePAD2017reviewincludedcomparativeeffectivenessfrom5Cochranesystematicreviewspublishedfrom2012to2017.Thesereviewsincludedparallelgroupdesignrandomizedcontrolledtrialsfrom4weeks(Horita2017)toatleast12weeksduration(Chong2012;Farne2015;Welsh2013)inpatientswithstableCOPD.Nannini2012didnotspecifyminimumdurationcriteriaforstudyinclusion.Anyformulationofthedrugusedwithineachdrugclasswasincluded.Thereviewsalsoincludedalldosesofdrugswithineachdrugclassthatwereusedinclinicaltrials.Forcombinationtherapiestreatmentscouldbeadministeredviasinglecombineddeviceorviatwoseparatedevices.ParticipantswereallowedICSandotherco-medicationsprovidedtheywerenotpartoftherandomizedtreatment.
PSDrequestedanupdatedsearchofthescientificliteraturetoidentifyanynewRCTevidencepublishedsincethecompletionofthe2017PADliteraturereviewonthecomparativeeffectsofLAMAandLABAasmonotherapy,aswellascombinationtherapies(LAMA+LABA,LABA+ICS,orLAMA+LABA+ICS)onall-causemortality,changeinSGRQtotalscore,andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbation.PSDagreedthatthePADliteraturereviewupdatewillbelimitedtoRCTsofatleast24weeksdurationandwillexclude:1)studieswithcomparatorsnotcommercializedinCanada;and2)studieswherebothcomparatorsareusedatnonapproveddosagesinCanada.InadditiontoaddingtotheclinicaltrialevidencesummarizedbyPAD,theTIofferedtosummarizeandcriticallyappraisenewRCTsofatleast1yearindurationthatevaluatedtheimpactofCOPDtherapyonoutcomemeasuresrelevanttoPSDfundingdecisions,includingmortality,seriousadverseevents,allcausehospitaladmissions(includingthoseduetosevereacuteexacerbations),andacutemoderatetosevereexacerbations.
3. MethodsWesearchedOvidMEDLINE,MEDLINEIn-Process,MEDLINEAheadofPrint,OvidEmbase,theCochraneCentralRegisterofControlledTrials(CENTRAL),andEBSCOCINAHLfromtheenddatesoftheCochranereviewsincludedinthePADliteraturereviewuntilNovember7,2018.Wealsosearchedclinicaltrials.govforallrelevantRCTreports.ForallnewlyidentifiedstudiesdatawereabstractedforoutcomesofinterestinthePADliteraturereview.ThesummarytablesofthePADreviewwereamendedtoshowtheupdatedestimatesoftreatmenteffectsthatreflectthefindingsofnewRCTevidence.Fornewstudieswithaminimumdurationof1year,dataabstractionwasperformedaccordingtoahierarchyofoutcomesdevelopedbytheTIandriskofbiaswasassessedusingtheCochraneRiskofBiasToolandhelpedtoinformconclusions.
4. SummaryofAvailableEvidenceAtotalof6studiesevaluating7comparisonsofinterestmetthecriteriaforcriticalappraisal.TheTIpreviouslycriticallyappraised1study(IMPACT2018)evaluating3
6
comparisonsofinterest(LABA+LAMAvs.LABA+ICS;LABA+LAMA+ICSvs.LABA+LAMA;LABA+LAMA+ICSvs.LABA+ICS),whichisavailableintheTIreportonTrelegyEllipta,datedSeptember12,2018.Criticalappraisalofthe5remainingstudiesevaluatingLABAvs.LAMA(1study:INVIGORATE2013),LABA+LAMAvs.LABA(1study:Donohue2016),LABA+LAMAvs.LAMA(2studies:DYNAGITO2018;SPARK2013),andLABA+ICSvs.LAMA(1study:Sarac2013)isprovidedinthisreport.Oneadditionalstudy(SUMMIT2016),whichisexcludedfromthePADliteraturereviewupdatebecauseitusesaLABAcomparator(vilanterol25mcg)thatisnotcommercializedinCanada,wascriticallyappraisedsinceitisthelargeststudytodatecomparingLABA+ICSwithLABA.LABAvs.LAMA:OnlyINVIGORATE2013,adoubleblindRCTin3444COPDpatientswithadocumentedhistoryofexacerbationwithinayearbeforeenrolment,mettheinclusioncriteriaforcriticalappraisal.Thisstudycomparedindacaterol150mcg(n=1721)withtiotropium18mcg(n=1718),bothadministeredoncedaily.TheHealthCanadarecommendeddoseofindacaterolis75mcginhalationoncedaily.Patientswhohadbeenusingastabledoseofinhaledcorticosteroidforatleastamonthbeforestudyentrywereinstructedtocontinuethisregimenforthedurationofthestudy.Themeanageofstudypatientswas64.0years,77%weremales,and65%wereformersmokers.Post-bronchodilatorFEV1was40.5%ofpredictednormalvalueandameanSGRQtotalscoreof48.3(17.6)atscreening.Seventyninepercenthadahistoryof1COPDexacerbationinthepreviousyear.SeventytwopercentwerereceivingICSatrandomization.Overall,2711patients(79%)completedthetrial.Thisstudyanalyzed3072(89.3%)intheper-protocolsetforexacerbationsand3013(87.6%)intheper-protocolsetforspirometry.DataforexacerbationsandFEV1werenolongercollectedoncepatientsdiscontinuedfromthestudy.Deathswererecordedforallrandomizedpatientsduringstudyparticipationandfor30daysafterstudydrugdiscontinuation.INVIGORATE2013isjudgedtohaveahighriskofbiasaccordingtotheCochraneRiskofBiasToolwithrespecttoattrition,selectivereportingandsourceoffunding.
LABA+LAMAvs.LABA:Donohue2016isadoubleblindRCTinvestigatingthelong-termsafetyandtolerabilityoftwice-dailyaclidinium/formoterol(ACL/FOR)400/12mcg(n=392)versusformoterol(FOR)12mcg(n=198)in590patientswithsymptomaticCOPD.ICSandoralorparenteralcorticosteroidsatdoses≤10mg/day,theophyllineandH1-antihistaminewerepermittedforchronicuseprovidedthedosagewasstablefor≥4weekspriortoscreeningandthroughoutthetrial.Patientswerepermittedtreatmentwithalbuterolasneeded,butnotwithin6hbeforeavisit.Chronicuseofoxygentherapywasalsopermittedprovidedthedosagewasstablefor≥4weekspriortoscreening.Themeanageofstudypatientswas64.2years,55%weremales,54%wereformersmokersand23.9%had≥1exacerbationinthepreviousyear.Post-bronchodilatorFEV1was51.4%ofpredictednormalvalueatscreening.Fiftytwopercentand46%wereclassifiedasGOLDStageII(moderate)andStageIII(severe),respectively.TherewerenodifferencesinuseofconcomitantCOPDdrugspriortoandcontinuingduringstudywith38%usingdrugsfromanycategoryand35%receivingICS.Only398patients(68%)completedthetrial.All590patientswereincludedinthesafetyanalysis.Patientswhodiscontinuedthestudyprematurelydidnotcomeinforfurtherevaluation.Thestudyreportdoesnotstateifdataforexacerbationsandother
7
efficacyoutcomeswerecollectedfollowingdiscontinuationfromthestudy.VitalstatuswasavailableforthetotalstudypopulationatWeek52.Donohue2016isjudgedtohaveahighriskofbiasaccordingtotheCochraneRiskofBiasToolwithrespecttodetectionbias,attritionbiasandsourceoffunding.Thereisahighriskofbiasrelatedtoefficacyoutcomesgiventhatthisstudyislikelynotblindedforefficacyoutcomesandthehighwithdrawalrates(32%)willleadtoattritionbiasforexacerbationoutcomes.Thestudyisjudgedtohaveunclearriskofbiaswithrespecttoblindingandselectivereporting.
LABA+LAMAvs.LAMA:Twostudies(DYNAGITO2018andSPARK2016)wereidentifiedintheupdatesearchthatfollowedpatientsforatleastayear:1. DYNAGITO2018isadoubleblindRCTcomparingdualbronchodilatortherapywith
olodaterol/tiotropium(OLO/TIO)5/5mcg(n=3939)withtiotropium(TIO)5mcg(n=3939),bothadministeredoncedailyviatheRespimatdevice,in7880patientswithsymptomaticCOPDandahistoryofmoderateorsevereexacerbationintheprecedingyear.PatientstakingICSatbaselinecontinuedthistreatment.Open-labelsalbutamolwasprovidedforas-neededrescuemedicationuse,butothershort-actingbeta-agonists,LAMAsandLABAswerenotpermittedduringthestudy.Themeanageofstudypatientswas66.4years,72%weremales,and63%wereformersmokers.Post-bronchodilatorFEV1was44.6%ofpredictednormalvalueandameanSGRQtotalscoreof47.8(17.7)atscreening.Fortyfivepercenthadahistoryof≥2exacerbationsor≥1severeexacerbationintheprecedingyear.Nearly40%ofthepatientswerereceivingtripletherapy(LABA+LAMA+ICS),and26%werereceivingLABA+ICSatrandomization.Itisnotreportedwhetherdualtherapy(LAMA/LABAorLABA/ICS)actuallyfailedinthosepatientsreceivingtripletherapyatscreening.Overall,6742patients(86%)completedthetrial.FewerpatientsreceivingOLO/TIO(12.4%)withdrewfromthestudyascomparedtoTIOalone(16.5%)[OR0.72(95%CI0.63,0.81);p<0.00001].Thisstudyanalyzedexacerbationdataduringthe“actualtreatmentperiod”,definedasthetimefromfirstdoseofmedicationuntil1dayafterthelastdoseofmedication.Patientswhopermanentlydiscontinuedstudytreatmentdidnotcomeinforfurtherevaluationsothereislossofinformationonexacerbationeventsfollowingprematurediscontinuationofstudytreatment.Vitalstatuswasavailablefor99.6%ofthetotalstudypopulationattheendofthestudy.Also,thefactthatmorepatientsreceivingTIOalonewithdrewfromthestudycouldbiastheanalysisofexacerbationsdata.
2. SPARK2016isadoubleblindRCTin2224patientswithsevereandverysevereCOPDandahistoryofatleast1moderateexacerbationintheprecedingyear.Thisstudycompareddualbronchodilatortherapywithindacaterol/glycopyrronium(IND/GLY)110/50mcg(n=741)withglycopyrronium(GLY)50mcg(n=741),bothadministeredoncedailyviatheBreezhalerdevice.Approximately75%ofpatients,withsimilarproportionsacrosstreatmentgroups,wereusingICSeitherasfixeddosecombinationorasmonotherapyatbaseline.PatientsusingICSatbaselinecontinuedthistreatmentatthesameorequivalentdoseandregimenduringthestudy.Salbutamolwaspermittedasrescuemedicationuse.Longactingbronchodilatorswerediscontinuedwithawashoutofupto7days(fortheophylline,indacaterolandtiotropium)beforescreening.Themeanageofstudypatientswas63.1years,75%weremales,and62%wereformer
8
smokers.PostbronchodilatorFEV1was37.2%ofpredictednormalvalueandameanSGRQtotalscoreof53(18)atscreening.Seventysevenpercentand22%hadahistoryof1COPDexacerbationand2ormoreexacerbations,respectively,inthepreviousyear.SeventyfivepercentwerereceivingICSatrandomization.Useofotherrespiratorymedicationsatbaselineisnotreported.Overall,1108patients(75%)completedthetrial.Atotalof171(23.1%)patientsintheIND/GLYgroupand203(27.4%)intheGLYgrouppermanentlydiscontinuedthestudy.ThehighernumberofwithdrawalsintheGLYgroupascomparedtoIND/GLYwasnotstatisticallysignificant[OR1.26(95%CI0.99,1.59)].DataforexacerbationsandFEV1werenolongercollectedoncepatientsdiscontinuedfromthestudy.Patientswhoprematurelydiscontinuedwerefollowedforsurvivaltotheendofthestudy.SPARK2013isjudgedtohaveahighriskofbiasaccordingtotheCochraneRiskofBiasToolwithrespecttoattritionandsourceoffunding.Selectivereportingisjudgedtohaveanunclearriskofbias.
LABA+ICSvs.LABA:SUMMIT2016isadoubleblindRCTin16,590patientswithsymptomaticCOPDandahistoryofcardiovasculardisease.Thiswasanevent-drivestudyinwhichfollow-upcontinueduntilatleast1000deathshadoccurred(medianstudyexposurewas1.8years).Thisstudycomprisedof4treatmentarms.Thecomparisonofinterestforthisreportisfluticasonefuroate/vilanterol(FF/VI)110/50mcg(n=4145)versusvilanterol(VI)25mcg(n=4146),bothadministeredoncedailyviatheElliptadevice.Vilanterol25mcgmonotherapyisnotcommerciallyavailableinCanada.Theuseofallinhaledcorticosteroidsandinhaledlongactingbronchodilatorswasdiscontinued≤48hoursbeforestudyentry,althoughotherCOPDmedicationssuchastheophyllineswereallowed.Themeanageofstudypatientswas65years,75%weremales,and54%wereformersmokers.PostbronchodilatorFEV1was59.7%ofpredictednormalvalueatscreening.Thirtyeightpercenthadahistoryof1ormoreCOPDexacerbationsinthepreviousyear.Approximately33%werereceivingICSatrandomization.Overall,6250patients(76%)receivingFF/VIandVIcompletedthetrial.DataforexacerbationsandFEV1werenolongercollectedoncepatientsdiscontinuedfromthestudy.Vitalstatuswasknownfor99.97%ofpatientsintheintentiontotreat(ITT)population.SUMMIT2016isjudgedtohaveahighriskofbiasaccordingtotheCochraneRiskofBiasToolwithrespecttoattritionbias,selectivereportingandsourceoffunding.Therearealsootherbiaseswithrespecttostudydesignandthepresenceofconfoundingthatmisrepresentthetreatmenteffect.
LABA+ICSvs.LAMA:OnlySarac2016,asingle-centreopenRCTwasidentified.Thissmallstudyrandomized44COPDpatientswithahistoryof≥1exacerbationintheprecedingyeartotwice-dailysalmeterol/fluticasone50/500mcg(n=22)oronce-dailytiotropium18mcg(n=22).Alllong-actingbronchodilatorsandinhaledsteroidswerestoppedduringthewashoutperiodandtheywereonlyallowedtotakeshort-actingbronchodilators(salbutamol-ipratropiumcombinationMDI).Duringthetreatmentperiodthepatientswereallowedtouseshort-actingbronchodilatorswhenneeded,butwerenotallowedtouseanyotherbronchodilatorsorinhaledsteroids.Themeanageofstudypatientswas66.6yearsand91%weremales.Smokingstatusatscreeningisnotreported,CATscorewas9.3and39%had≥2exacerbationsinthepreviousyear.Post-bronchodilatorFEV1was65.4%of
9
predictednormalvalueatscreening.Respiratorymedicationuseatscreeningisalsonotreported.All44patientscompletedthetrialandnoadverseeventswerereported.AccordingtotheCochraneRiskofBiasTool,Sarac2016isjudgedtohaveahighriskofselectionbias,allocationbias,performancebiasanddetectionbias,andanunclearriskofselectivereportingandsourceoffunding.
5. ResultsandInterpretationAllstudieswithadurationof1yearorlongerarejudgedtohaveahighriskofbiasaccordingtotheCochraneRiskofBiasToolwithrespecttoattrition.However,vitalstatuswasavailablefor>99%ofrandomizedpatientsinallstudies.Thereforetheoverallqualityofevidenceislowforalloutcomesexceptmortality.Nostudyshowedadifferenceintotalmortalitybetweenanyofthecomparatorgroups.
TotalSAEsprovidesthebestsummarystatisticoftherapeuticimpactaccountingforallknownandunknownseriousimpact(benefitandharm)fromtherapy,andincludesoutcomessuchasfatalCOPDexacerbation,hospitalizationduetosevereexacerbationandhospitalizationduetoseverepneumonia.NostudiesshowedadifferenceintotalSAEsforanycomparison.Theeffectofinhaledmedicationsonmoderatetosevereexacerbationsneedstobereportedastheproportionofpatientswithoneormoreexacerbations.Only2studies(Donohue2016;SPARK2013)reportedthenumberofpatientswith1ormoremoderatetosevereexacerbationandbothstudiesshowednodifferencesbetweentheirrespectivetreatmentgroups.Otherstudiesreportedrateofmoderateorsevereexacerbation(DYNAGITO2018;SUMMIT2016)andtime-to-firsteventanalysisofmoderateorsevereexacerbation(INVIGORATE2013).DYNAGITO2018andSUMMIT2016claimednodifferenceinexacerbationratesbetweentreatmentarms.INVIGORATEclaimedthattimetofirstmoderateorsevereexacerbationwaslongerwithtiotropiumversusindacaterol[HR1.20(95%CI1.07to1.33;p=0.0012)].Time-to-first-eventanalysisisusefulonlywhenitisknownhowmanypatientshadmorethanoneexacerbationthroughoutthestudyinbothtreatmentgroups.Furthermore,thereportedeventsandratesareuncertainduetothehighwithdrawalratesinthestudiesandnoattemptwasmadetoreduceattritionbiasbyadequatelyaccountingforthepatientswhowithdrewprematurelyinthecalculationofeventandannualratesofmoderateorsevereexacerbations.
Twostudies(INVIGORATE2013;SPARK2013)reportedqualityoflife(SGRQ)and1study(INVIGORATE2013)reporteddyspneasymptoms(TDI).EstimatesforcomparativetreatmenteffectsonSGRQandTDIareuncertainduetothehighwithdrawalratesinbothstudiesandinadequateaccountingofpatientswhowithdrewprematurely.Furthermore,INVIGORATE2013reportedonasubset(approx.75%)oftotalrandomizedpatients.Therefore,theresultsarenotconsideredvalidduetomissingdata.
Therewerenodifferencesintotaladverseeventsbetweenanyofthecomparatorsinthesestudies.DYNAGITO2018istheonlystudythatdemonstratedadifferenceinwithdrawaldue
10
toadverseeventsbetweencomparatorgroupswithsignificantlymorepatientsreceivingTIO(16.5%)thanOLO/TIO(12.4%)whowithdrewduetoanadverseevent(ARI4.1%,NNH=24for1year).
ReducedneedforrescuemedicationisamarkerofimprovedcontrolofCOPDsymptoms.Threestudies(INVIGORATE2013;Donohue2016;SPARK2013)reportedneedforrescuesalbutamolduringthetreatmentperiod.UseofrescuetreatmentdidnotsignificantlydifferbetweenACL/FORandFORgroupsinDonohue2016.INVIGORATE2013reportedinasubset(91%)ofrandomizedpatientsthatpatientsintheINDgroupneededrescuetreatmentlessoftenascomparedtothosewhoreceivedTIO[LSmeandifference-0.62(95%CI-0.79,-0.45);p<0.0001)indailynumberofpuffs;LSmeandifference8.0%(95%CI5.9,10.2);p<0.0001inproportionofdayswithnorescueuse).ThisfindingisinconsistentwithindacaterolshowingnodifferenceversustiotropiumforSGRQtotalscoreandTDIscore.Alsoitisdifficulttounderstandhowindacaterolreducedtheneedforrescuemedicationwhentimetofirsteventanalysisofexacerbationrevealedthatindacaterolincreasedtheriskofmoderateorsevereexacerbationduringtreatmentversustiotropium.SPARK2013reportedareductionindailypuffsofrescuesalbutamolintheIND/GLYgroupascomparedtoGLYgroup[LSmeandifference-0.81;p<0.0001).TheclinicalrelevanceofareducedneedofrescuesalbutamolisunclearconsideringtheSGRQeffectestimateisuncertain(i.e.athighriskofattritionbias)andTDIscoreisnotreported.
COPDrelatedhealthcareutilization,whichincludesphysicianvisits/ERvisitsandhospitalizations,isanendpointthatwasnotreportedinanyofthestudies.Thesefindingswouldcorroboratethefindingsofdecreasedrateofacutemoderatetosevereexacerbation.
Fivestudies(Donohue2016;INVIGORATE2013;Sarac2016;SPARK2013;SUMMIT2016)reportedtroughFEV1,ofwhich4studies(Donohue2016;INVIGORATE2013;Sarac2016;SPARK2013)showedstatisticallysignificantbutnotclinicallyrelevantbetween-groupdifferences.SUMMIT2016didnotstatisticallycompareFF/VIandVIgroupsforon-treatmentrateofdeclineinFEV1.FEV1isasurrogateoutcomethathasvalidityinestimatingtheriskofdyingfromCOPDbutlittleuseinassessingtheimpactofinhaleddrugtherapyonCOPDsymptoms.
6. ConclusionBasedonthenewlyidentifiedRCTsofatleast1yearduration,thereisinsufficientscientificallyvalidevidencethatanyofthesecomparisons(LABAvs.LAMA,LABA+LAMAvs.LABA,LABA+LAMAvs.LAMA,andLABA+ICSvs.LAMA)providesatherapeuticadvantageintermsofmoderateorsevereexacerbation,qualityoflife(SGRQ),reporteddyspneasymptoms(TDI),needforrescuemedication,andCOPDrelatedhealthcareutilization.
BasedonthenewlyidentifiedRCTsofatleast1yearduration,thereissufficientscientificallyvalidevidencedemonstratingthatnoneofthesecomparisons(LABAvs.LAMA,LABA+LAMAvs.LABA,LABA+LAMAvs.LAMA,andLABA+ICSvs.LAMA)provideadifferenceintermsofall-causemortality,totalseriousadverseevents(whichincludesallcause
11
hospitalizationandhospitalizationduetosevereexacerbation),andtotaladverseeventsinthetreatmentofCOPD.
12
UpdateofPADliteraturereviewofinhaledtherapiesfortreatmentofadultpatientswithchronicobstructivepulmonarydisease(COPD)
Background
DiagnosisandmanagementofCOPDChronicobstructivepulmonarydisease(COPD)isaprogressiveanddisablingdiseasecharacterizedbyairwayinflammationandairflowlimitationthatisnotfullyreversible.Airflowobstructionleadstosymptomsofwheezing,shortnessofbreath,chesttightness,coughingandproductionofexcessmucus.COPDoccursasaconsequenceofexposuretonoxiousparticlesorgases.Exposuretocigarettesmokeisthemostcommonriskfactor.DrugstotreatCOPDarelicensedbyregulatoryauthoritiesbasedonshort-termrandomizedtrials(typically12weeksinduration)thatshowanimprovementinthesurrogatemarkerFEV1,whichistheprimaryoutcomemeasureinmosttrials.However,thegoaloftreatingCOPDistopreventacutemoderatetosevereexacerbations,improvequalityoflifeandreducesymptomssuchasdyspnea.(1)ThemaintreatmentoptionsforCOPDbelongtoanumberofpharmacologicalclasses:bronchodilators(short-actingbeta2agonists[SABA],long-actingbeta2agonists[LABA],short-actingmuscarinicantagonists[SAMA],andlong-actingmuscarinicantagonists[LAMA]),inhaledcorticosteroids[ICS],andinhibitorsoftheenzymephosphodiesterase-4[PDE4inhibitors].InCanada,approximately20inhaledmedicationsareapprovedtotreatCOPD(Table1).
Table1:COPDInhaledMedications
Class Medication (Brand Name, Inhaler Device)SABA salbutamol(VentolinHFAMDI,AiromirMDI,VentolinDiskus)
terbutaline(BricanylTurbuhaler)SAMA ipratropium(AtroventHFAMDI)SAMA+SABA ipratropium+salbutamol(CombiventRespimat)LABA formoterol(ForadilAerolizer)
indacaterol(OnbrezBreezhaler)salmeterol(SereventDiskus,SereventDiskhaler)
LAMA aclidinium(TudorzaGenuair)glycopyrronium(SeebriBreezhaler)tiotropium(SpirivaHandiHaler,SpirivaRespimat)umeclidinium(IncruseEllipta)
LAMA+LABA aclidinium+formoterol(DuaklirGenuair)glycopyrronium+indacaterol(UltibroBreezhaler)tiotropium+olodaterol(InspioltoRespimat)umeclidinium+vilanterol(AnoroEllipta)
ICS+LABA budesonide+formoterol(SymbicortTurbuhaler)
13
Class Medication (Brand Name, Inhaler Device)fluticasonefuroate+vilanterol(BreoEllipta)fluticasonepropionate+salmeterol(AdvairDiskus)
ICS+LAMA+LABA fluticasonefuroate+umeclidinium+vilanterol(TrelegyEllipta)SABAshortactingbeta2adrenergicagonist;SAMAshortactingmuscarinicantagonist,LABAlongactingbeta2adrenergicagonist;LAMAlongactingmuscarinicantagonist,ICSinhaledcorticosteroid
InTherapeuticsLetter#109publishedinFebruary2018wereportedsystematicreviewsoftheclinicalefficacyofinhaledLABAdrugslicensedforCOPD(formoterol,arformoterolandsalmeterol)ascomparedtoplacebo.(2)Weidentified22RCTsforformoteroland2RCTsforarformoterol(N=13,958),and17RCTsforsalmeterol(N=10,115).Durationofformoteroltrialsrangedfrom4-26weeks(14RCTs)to48-52weeks(8RCTs).Arformoteroltrialsrangedfrom12to52weeks(2RCTs).Salmeteroltrialsrangedfrom4-24weeks(14RCTs)to52weeks(2RCTs),withasinglemuchlongertriallasting156weeks.Thesetrialsexcludedpatientswithotherconcurrentrespiratorydiseases,includingasthma.Mostparticipantsweremenwithameanagerangingfrom60to67years.ThedosesmoststudiedinCOPDclinicaltrialswereformoterol9and12μgtwicedailyandsalmeterol50μgtwicedaily.Patientswereallowedtocontinueusingthefollowingatstabledosesthroughoutthestudies:ICSororalcorticosteroids,inhaledSAMAsorLAMAs,PDE4inhibitorsandshort-actingsalbutamolforrescuetherapy.Weconcludedthat,ascomparedtoplacebo,the3inhaledlongactingβ2agonists(formoterol,arformoterolandsalmeterol)donotprolongsurvivalorreducetheriskofhospitalization(totalseriousadverseevents)inpatientswithCOPD.Evidenceforimprovementofsymptomscoresisoflowqualityandinsufficienttojustifylong-termuse.Asubsetofpatientsmayderiveclinicallyimportantsymptomaticrelief.Suchpatientscanbeidentifiedonacase-by-casebasisusingashorttherapeutictrial.
NumerousclinicalpracticeguidelinesaddressingthemanagementofCOPDrecommendstepwiseintensificationofdrugtherapyinpeoplewithpersistentbreathlessnessorexacerbations.
TheGlobalInitiativeforChronicObstructiveLungDisease(GOLD)reporthasprovidedtherapeuticrecommendationsforpatientswithCOPDbasedon“ABCD”groupsderivedexclusivelyfrompatientsymptomsandtheirassessmentofexacerbation:(3)
1. GroupAhasmodifiedMRCdyspneascale(mMRC)0-1,CAT<10and0or1exacerbationnotleadingtohospitaladmission.
2. GroupBhasmMRCscore≥2,CAT≥10and0or1exacerbationnotleadingtohospitaladmission.
3. GroupChasmMRC0-1,CAT<10and≥2exacerbationor≥1exacerbationleadingtohospitaladmission.
4. GroupDhasmMRCscore≥2,CAT≥10and≥2exacerbationor≥1exacerbationleadingtohospitaladmission.
TheGOLDreportprovidesapharmacologicaltreatmentalgorithmbasedonwhichgroupthepatientbelongs:
14
GroupA:Startwithashortorlongactingbronchodilator;evaluateeffect;thencontinue,stoportryalternateclassofbronchodilator.Thisshouldbecontinuedifsymptomaticbenefitisdocumented.GroupB:Startwithalongactingbronchodilator(LABAorLAMA).Longactingbronchodilatorsaresuperiortoshortactingbronchodilatorstakenasneeded.Thereisnoevidencetorecommendoneclassoflongactingbronchodilatoroveranotherforinitialreliefofsymptomsinthisgroupofpatients.Intheindividualpatient,thechoiceshoulddependonpatient’sperceptionofsymptomrelief.Forpatientswithpersistentbreathlessnessonmonotherapytheuseof2bronchodilatorsisrecommended(LAMA/LABA).Forpatientswithseverebreathlessnessinitialtherapywith2bronchodilatorsmaybeconsidered.Ifadditionofthesecondbronchodilatordoesnotimprovesymptomstreatmentshouldbesteppeddowntoasinglebronchodilator.GroupC:Startwithasinglelongactingbronchodilator.LAMAwassuperiortoLABAregardingexacerbationpreventionsostartwithLAMAinthisgroup.Patientswithpersistentexacerbationmaybenefitfromaddingasecondlongactingbronchodilator(LAMA/LABA)or(LABA/ICS).AnICSincreasesriskofdevelopingpneumoniainsomepatientssoprimarychoiceisLAMA/LABA.GroupD:StartwithaLAMA/LABAcombination,asitissuperiortomonotherapywitheachclassofdrug.IfsinglebronchodilatorischosenthenLAMAispreferredforexacerbationpreventionascomparedtoLABA.LAMA/LABAcombinationwassuperiortoLABA/ICSinpreventingexacerbationandotherpatientreportedoutcomes.AlsoGroupDpatientsareathigherriskofdevelopingpneumoniawhenreceivingtreatmentwithICS.InsomepatientsLABA/ICSmaybefirstchoice(historyand/orfindingsofasthma-COPD;orinpatientswithhigheosinophilcounts).InpatientswhodevelopfurtherexacerbationsonLAMA/LABAtherapytheneitherswitchtoLABA/ICSoraddICStoLAMA/LABA.Ifpatientsontripletherapystillhaveexacerbationsthen:a. AddroflumilastinpatientswithFEV1<50%predictedvalueandchronicbronchitis
particularlyiftheyexperiencedatleastonehospitalizationforanexacerbationinthepreviousyear.
b. Addmacrolide(azithromycin)c. StopICS.
InFebruary10,2017theProvincialAcademicDetailing(PAD)servicecompletedaliteraturereviewandsynopsisofrandomizedcontrolledtrial(RCT)evidencefromexistingCochranereviewsthataddressspecificGOLDguidelinerecommendations.ThePADreviewpresentstheoverallgradingofevidenceandtreatmenteffectestimatesasreportedfromtheCochranereviewsassummarytablesforthefollowingoutcomes:all-causemortality;St.George’sRespiratoryQuestionnaire(SGRQ)totalscore,ameasureofqualityoflife;andnumberofpeople
15
with1ormoreacutemoderatetosevereCOPDexacerbation.FindingsofthePADreviewaresummarizedbelow.
LABAvs.Placebo
• KewKMetal2017CochranereviewsummarybyPAD:Evidenceisbasedfrom26RCTsin14,939adultpatientswithCOPD;Mortality(Moderatequalityevidence0.90(0.75,1.08);26RCTs;N=14,179);SGRQtotalscore(Moderatequalityevidence-2.32(-3.09,-1.54);17RCTs;N=11,397);andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbation(Moderatequalityevidence0.88(0.76,1.02);7RCTs;N=3,968).
• TheTIletter#109publishedinFebruary2018(www.ti.ubc.ca)on“Inhaledlongactingbeta2agonistforCOPD”concludedthatthefourinhaledlongactingβ2agonists(indacaterol,formoterol,arformoterolandsalmeterol)donotprolongsurvivalorreducetheriskofhospitalization(totalseriousadverseevents)inpatientswithCOPD.Evidenceforimprovementofsymptomscoresisoflowqualityandinsufficienttojustifylong-termuse.Asubsetofpatientsmayderiveclinicallyimportantsymptomaticrelief.Suchpatientscanbeidentifiedonacase-by-casebasisusingashorttherapeutictrial.
Tiotropium(LAMA)vs.Placebo
• KarnerCetal2014CochranereviewsummarybyPAD:Evidenceisbasedfrom22RCTsin23,309adultpatientswithCOPD;Mortality(Moderatequalityevidence0.98(0.86,1.11);22RCTs;N=23,309);SGRQtotalscore(Highqualityevidence-2.89(-3.35,-2.44);9RCTs;N=13,304);andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbation(Highqualityevidence0.78(0.70,0.87);22RCTs;N=23,309).
• TIletter#60publishedinSeptember2006on“ClinicalPearlsfromPrescrire”reportedinMay2006thattiotropiumisame-toodrugforCOPD.Forpatientswithchronicobstructivepulmonarydisease,tiotropiumhasmoreadverseeffectsthanthebronchodilatorswithwhichithasbeencompared,andithasnotbeenshowntobemoreeffective.Ipratropiumseemstobethebestchoiceforpatientsneedinginhaledantimuscarinictherapy.
Tiotropium(LAMA)vs.LABA
• ChongDetal2012CochranereviewsummarybyPAD:Evidenceisbasedfrom7RCTsin12,223adultpatientswithCOPD;Mortality(Verylowqualityevidence0.82(0.60,1.13);6RCTs;N=12,123);SGRQtotalscore(ungradedRCTSwerenotpooled);andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbation(Moderatequalityevidence0.86(0.79,0.93);6RCTs;N=12,123).
[Tiotropium(LAMA)+LABA]vs.Tiotropium(LAMA)
• FarneDetal2015CochranereviewsummarybyPAD:Evidenceisbasedfrom10RCTsin10,894adultpatientswithCOPD;Mortality(Lowqualityevidence1.24(0.81,1.90);8RCTs;N=9,633);SGRQtotalscore(moderatequalityevidence;-1.34(-1.87to-0.80);5RCTs;N=6,709);andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbation(UngradedqualityevidencefromRCTswerenotpooled).
16
[Tiotropium(LAMA)+LABA]vs.LABA
• FarneDetal2015CochranereviewsummarybyPAD:Evidenceisbasedfrom4RCTsin3,378adultpatientswithCOPD;Mortality(Lowqualityevidence1.15(0.62,2.13);3RCTs;N=3,514);SGRQtotalscore(moderatequalityevidence;-1.25(-2.14to-0.37);4RCTs;N=3,378);andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbation(notreported).
[LABA+ICS]vs.LABA
• NanniniLetal2012CochranereviewsummarybyPAD:Evidenceisbasedfrom14RCTsin11,794adultpatientswithCOPD;Mortality(Moderatequalityevidence0.92(0.76,1.11);10RCTs;N=10,618);SGRQtotalscore(Ungradedqualityevidence-1.58(-2.15,-1.01);6RCTs;N=10,681);andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbation(Moderatequalityevidence0.83(0.70,0.98);6RCTs;N=3,357).
[LABA+LAMA]vs.[ICS+LABA]
• HoritaNetal2017CochranereviewsummarybyPAD:Evidenceisbasedfrom11RCTsin9,839adultpatientswithCOPD;Mortality(Lowqualityevidence1.01(0.61,1.67);8RCTs;N=8,200);SGRQtotalscore(Ungradedqualityevidence-1.22(-2.52,+0.07);6RCTs;N=6,055);andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbation(Lowqualityevidence0.82(0.70,0.96);9RCTs;N=8,922).
[Tiotropium(LAMA)+LABA+ICS]vs.[Tiotropium(LAMA)+LABA]
KarnerCetal2011CochranereviewsummarybyPAD:Evidenceisbasedfrom1RCTsin293adultpatientswithCOPD;Mortality(Ungradedqualityevidence1.02(0.32,3.24);1RCT;N=293);SGRQtotalscore(Ungradedqualityevidence-1.02(-5.10,+3.06);1RCT;N=293);andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbation(Ungradedqualityevidence0.81(0.51,1.30);1RCT;N=293).
2017PADliteraturereviewconcluded:
1.Thereisinsufficientevidencetoestimatetheeffectoftripletherapy(LAMA+LABA+ICS)onhealth-relatedqualityoflife,theriskofexacerbations,ortheriskofdeath.
2.Thereisabsenceofhighqualityevidenceregardingtheeffectofintensifyinginhaledtherapy(i.e.progressingtoLAMA+LABAandLAMA+LABA+ICS)onhealth-relatedqualityoflifeandontheriskofexacerbationanddeathinpeoplewithCOPD.Thetrueeffectcannotbefirmlyestablished.
3.ConsiderriskfactorsforpneumoniawhenweighingthesuitabilityofinhaledcorticosteroidtherapyinpeoplewithCOPD.Theseinclude:COPDexacerbationinthepreviousyear,FEV1<50%predicted,priorhistoryofpneumonia,lowbodymassindex,advancingage,andcurrentsmoker.
17
4.Ina2016comparativeeffectivenessreviewofinhaledtherapiesforCOPD,noconsistentdifferenceswereidentifiedinbenefitorharmoutcomeswithintheclassesofLABA,LAMAorICS+LABAtherapies.
5.Whenevaluatingpatient’ssymptomaticresponsetoinhaledtherapiesusingagoal-settingapproach,giveattentiontotheirabilitytouseinhaledtherapydevices.
PharmaceuticalServicesDivision(PSD)Request
PSDrequestedanupdatesearchofthescientificliteraturetoidentifyanynewRCTevidencepublishedsincethecompletionofthePADliteraturereviewonthecomparativeeffectsofLAMAandLABAasmonotherapy,aswellascombinationtherapies(LAMA+LABA,LABA+ICS,orLAMA+LABA+ICS)onall-causemortality,changeinSGRQtotalscore,andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbationinthetreatmentofCOPD.
ThePADreviewincludedcomparativeeffectivenessfrom5Cochranesystematicreviewspublishedfrom2012to2017.(4-8)Thesereviewsincludedparallelgroupdesignrandomisedcontrolledstudiesfrom4weeks(Horita2017)toatleast12weeksduration(Chong2012;Farne2015;Welsh2013)inpatientswithstableCOPD.Nannini2012didnotspecifyminimaldurationcriteriaforstudyinclusion.Anyformulationofthedrugusedwithineachdrugclasswasallowed.Alsothereviewsincludedvariousrangeofdosesofdrugswithineachdrugclassthatwereusedinclinicaltrials.Forcombinationtherapiestreatmentscouldbeadministeredviasinglecombineddeviceorviatwoseparatedevices.Thevariousdosesofdrugswithineachdrugclasswhetherofficiallyindicatedornotthatwereusedinclinicaltrialswereincluded.ParticipantswereallowedICSandotherco-medicationsprovidedtheywerenotpartoftherandomizedtreatment.Chong2012Cochranereview(LAMAvsLABA):TheonlyLAMAincludedwastiotropium18mcgasSpirivaHandihaler.LABAsincludedweresalmeterol50to100mcgandindacaterol150to300mcg.(4)
Farne2015Cochranereview(LAMA+LABAvsLAMA):TheonlyLAMAincludedwastiotropiumatdosesrangingfrom2.5to5mcgasSpirivaRespimatand18mcgasSpirivaHandihaler.LABAsincludedwereformoterol20to24mcg,indacaterol150mcg,salmeterol50mcg,andolodaterol5mcg.(5)
Nannini2012Cochranereview(LABA+ICSvsLABA):LABAsincludedwereformoterol9to18mcgandsalmeterol100mcg.ICSincludedwerebudesonide160to640mcgandfluticasone500to1000mcg.(6)
Welsh2013Cochranereview(LABA+ICSvsLAMA):TheonlyLAMAincludedwastiotropium18mcgasSpirivaHandihaler.LABAincludedwassalmeterol100mcg.ICSincludedwasfluticasone500to1000mcg.(7)
18
Horita2017Cochranereview(LABA+LAMAvsLABA+ICS):LAMAsincludedwereaclidinium800mcg,glycopyrronium50mcg,tiotropiumatdosesrangingfrom2.5to5mcgasSpirivaRespimatand18mcgasSpirivaHandihaler,andumeclidinium62.5mcg.LABAsincludedwereformoterol24mcg,indacaterol110to150mcg,olodaterol5mcg,salmeterol100mcg,andvilanterol50mcg.ICSusedwasfluticasone500mcg.(8)
ThePSDagreedthePADliteraturereviewupdatewillbelimitedtoRCTsofatleast24weeksdurationandwillexclude:1)studieswithcomparatorsnotcommercializedinCanada;and2)studieswherebothcomparatorsareusedatnonapproveddosagesinCanada.
ThesummarytablesofthePADreviewwillbeamendedtoshowtheupdatedestimatesoftreatmenteffectsthatreflectthefindingsofnewRCTevidenceforall-causemortality,meanchangeinSGRQtotalscore,andnumberofpeoplewith1ormoreacutemoderatetosevereexacerbationinthetreatmentofCOPD.
InadditiontoaddingtotheclinicaltrialevidencesummarizedbyPAD,theTIofferedtosummarizeandcriticallyappraisenewRCTsofatleast1-yearindurationthatevaluatedtheimpactofCOPDtherapyonoutcomemeasuresrelevanttoPSDfundingdecisions,includingmortality,seriousadverseevents,allcausehospitaladmissions(includingthoseduetosevereacuteexacerbations),andacutemoderatetosevereexacerbations.TheFDAalsorecommendsminimum1-yeardurationstudiestosupportclaimsofmodificationorpreventionofexacerbationwithCOPDdrugs.(9)
Methods
SearchstrategyWesearchedOvidMEDLINE,MEDLINEIn-Process,MEDLINEAheadofPrint,OvidEmbase,theCochraneCentralRegisterofControlledTrials(CENTRAL),andEBSCOCINAHLfromtheenddatesoftheCochranereviewsincludedinthePADliteraturereviewuntilNovember7,2018.Wealsosearchedclinicaltrials.govforallrelevantRCTreports.
StudyselectionTheinitialsearchofallthedatabaseswasperformedtoidentifycitationsofpotentialrelevance.Theinitialscreenoftheseabstractsexcludedarticleswhosetitlesand/orabstractsareclearlyirrelevant.Thefulltextsofremainingarticleswerethenretrieved(andtranslatedintoEnglishwhererequired).Twoindependentreviewersassessedtheeligibilityofthetrialsusingastandardizedtrialselectionform.Athirdreviewerresolvedanydiscrepancies.
DatacollectionandanalysisDataabstractionwasdonebytwoindependentreviewers.ReviewManager5.3softwareoftheCochraneCollaborationwasusedtometa-analyzedata.Resultsarepresentedasrelativerisks(RR)with95%confidenceintervalsfordichotomousoutcomesandasweightedmeandifference(WMD)with95%confidenceintervalforcontinuousoutcomes.ForallnewlyidentifiedstudiesdatawereabstractedforoutcomesofinterestinthePADliteraturereview(all-causemortality,
19
changeinSGRQtotalscore,andnumberofpeoplewith1ormoreacutemoderatetosevereCOPDexacerbation).DataabstractionwasperformedaccordingtoahierarchyofoutcomesdevelopedbytheTIforstudieswithaminimumdurationof1year(seeEvaluativeframeworksectionbelow).
AssessmentofriskofbiasinincludedstudiesRiskofbiasforeachincludedtrialofatleast1-yeardurationwasassessedusingtheCochraneRiskofBiasToolwhichincludessevendomains:Randomization;allocationconcealment;blindingofparticipantandphysician;blindingofoutcomeassessor;attritionbias;selectivereportingbias;andotherbias(e.g.conflictofinterestbias-fundingofstudybythemanufactureroremployeeofthemanufacturerisauthorofthestudy).Eachdomainwasassessedas“Low”,“Unclear”or“High”riskofbias.(10)
EvaluativeframeworkEvidencefromvarioussourcesisorganizedandsituatedwithinahealthoutcomeandevidencehierarchy.Theprincipleisthathealthoutcomeshigheronthehierarchyaremoreimportantthanthoseloweronthehierarchy.Recognizingthatnotalloutcomesareofequivalentvalueandnotallevidencehasuniformprotectionagainstbias,theoverallframeworkforthesummaryandcriticalappraisalofnewstudieswithaminimumdurationof1yearwasbasedonahierarchyofoutcomesdevelopedbytheTI.Asmuchaspossible,thehierarchywascompletedforeachincludedRCTwithaminimum1-yearduration.
Outcomehierarchy:
1. Totalmortality2. Totalseriousadverseevents(includingtotalhospitalizations)3. Numberofpatientswithoneormoreacutemoderateorsevereexacerbation4. QualityoflifemeasuredbySaintGeorge’sRespiratoryQuestionnaire(SGRQ)totalscore
(≥4pointchangeintotalscoreisconsideredaminimalclinicallyimportantdifferenceinclinicaltrials;andameanchangeintotalscorefrombaseline)
5. Timetofirstmoderateorsevereexacerbation6. ImprovementinsymptomssuchasdyspneameasuredbyTransitionDyspneaIndex(TDI)
score(≥1pointimprovementisconsideredMCIDinclinicaltrials;ameanchangeinTDIscore)
7. Decreasedneedforrescuemedications(anadditionalmeasureofsymptomimprovement)
8. Totaladverseevents9. Totalwithdrawals10. Withdrawalduetoadverseevents11. COPDrelatedhealthcareutilization(physicianvisits/ERvisitsandhospitalization)12. EndofstudytroughFEV1(WeacceptthereisanincreaseinFEV1–asurrogateoutcome
measure.WewillproviderangeofimprovementinFEV1.Meta-analysisofthisoutcomewillnotbeperformed.)
20
Results
FindingsfromtheliteratureThesortingandinclusionprocessisdocumentedusingthePRISMAflowdiagram(Figure1).
Figure1:PRISMAflowdiagramofstudyselection
SummaryofexcludedstudiesReasonsforexclusionofthe6excludedstudiesareprovidedinTable2.
21
Table2:Excludedstudies(asrequestedandagreedbyPSD)
ClinicalStudyID/Reference ReasonforExclusionBuhl2017(11) AlreadyincludedinCochranereview(Farne2015);LABAusedas
monotherapyisnotapprovedinCanada(olodaterol5mcg)Celli2014(12) LABAusedasmonotherapyisnotapprovedinCanada(vilanterol25mcg)Donohue2014(13) ComparesanunapproveddoseofLABA/LAMAcombination(vilanterol25
mcg+umeclidinium125mcg)withLAMAmonotherapyatanunapproveddose(umeclidinium125mcg)
Dransfield2013(14) ComparesLABA/ICScombinationapprovedinCanada(vilanterol25mcg+fluticasonefuroate100mcg)withLABAmonotherapynotapprovedinCanada(vilanterol25mcg)
Ferguson2018(15) LABA/LAMAcombinationusedascomparatorisnotapprovedinCanada(formoterol9.6mcg+glycopyrronium18mcg)
Sharafkhneh2012(16) All3treatmentarmsusedosesofformoterolandbudesonidethatarenotapprovedinCanada
SUMMIT2016(17) LABAusedasmonotherapyisnotapprovedinCanada(vilanterol25mcg)
DescriptionofincludedstudiesAtotalof14studiesmettheinclusioncriteriaforthisPADliteraturereviewupdate.(16-29)Studydurationrangedfrom24weeksto64weeks.Table3liststhenewlyidentifiedstudiesaccordingtoinhaledCOPDtreatmentcomparison.Thestudiesinboldfonthaveaminimumdurationof1year.Atotalof10studiesreportSGRQmeandifference,ofwhichonly3studiesofatleast1-yeardurationreportthisoutcome.Only3studiesreportthenumberofpatientswithatleast1moderateorsevereexacerbation.Twoofthesestudieshaveaminimum1-yearfollow-up.All14studiesreportall-causemortality.Table3:NewstudiesaccordingtoinhaledCOPDtreatmentcomparison
2017PADReviewOutcomesofInterestReported?(YorN)
SGRQMeanDifference
Patientswith≥1Moderate/SevereExacerbation
All-CauseMortality
LAMAvs.LABA(4studies) Bateman2013(18) Y N YD’Urzo2014(19) Y N YINVIGORATE2013(20) Y N YSingh2014(21) Y Y YLABA+LAMAvs.LABA(4studies) Bateman2013(18) Y N YD’Urzo2014(19) Y N YDonohue2016(22) N Y YSingh2014(21) Y Y YLABA+LAMAvs.LAMA(8studies)Bateman2013(18) Y N Y
22
2017PADReviewOutcomesofInterestReported?(YorN)
SGRQMeanDifference
Patientswith≥1Moderate/SevereExacerbation
All-CauseMortality
DYNAGITO2018(23) N N YD’Urzo2014(19) Y N YDecramer2014(24) Y N YDonohue2013(25) Y N YMaleki-Yazdi2014(26) Y N YSingh2014(21) Y Y YSPARK2013(27) Y Y YLABA+ICSvs.LABA(1study) Ohar2014(28) N Y YLABA+ICSvs.LAMA(1study) Sarac2013(29) N N YLABA+LAMAvs.LABA+ICS(1study)IMPACT2018(30) Y N YLABA+LAMA+ICSvs.LABA+LAMA(1study) IMPACT2018(30) Y N YLABA+LAMA+ICSvs.LABA+ICS(2studies) FULFIL2017(31) Y N YIMPACT2018(30) Y N YNewdatafromthe14identifiedstudieswereincorporatedinthedataandanalysessectionsofthecorrespondingCochranereviews.Theestimatesoftreatmenteffectsonall-causemortality,SGRQtotalscoreandexacerbationswereupdatedtoreflectthefindingsofnewevidence.Theevidencesummarytablesandconclusionsofthe2017PADliteraturereviewwereamendedaccordinglyandareavailableasaseparatedocument[PAD2017COPDInhaledMedicationsBOOKLET(updatedFebruary2019)].
Summaryandcriticalappraisalof1yearorlongerstudiesAtotalof6studiesevaluating7comparisonsofinterestmetthecriteriaforcriticalappraisal.Thestudypopulationconsistedofpredominantlymiddle-agedwhitemales.Onestudy(IMPACT2018)thatevaluates3comparisonsofinterestwascriticallyappraisedpreviouslyaspartofasystematicreviewofTrelegyElliptasingleinhalertripletherapyfortreatmentofpatientswithmoderate-to-severeCOPD.PleaseseetheTIreportdatedSeptember12,2018formoreinformationonIMPACT2018.
SUMMIT2016isexcludedfromthePADliteraturereviewupdatebecauseitusesaLABAcomparator(vilanterol25mcg)thatisnotcommercializedinCanada.ThisstudyisnotincludedintheupdatedPADsummarytablesavailableseparately[PAD2017COPDInhaledMedicationsBOOKLET(updatedFebruary2019)].However,thisisthelargeststudytodatecomparingLABA+ICS(vilanterol/fluticasonefuroate25/100mcg)combinationtherapywithLABA(vilanterol25mcg)sotheTIwouldberemissnottoprovideacriticalappraisal.
23
LABAvs.LAMAOnlyINVIGORATE2013,adoubleblindRCTin3444COPDpatientswithadocumentedhistoryofexacerbationwithinayearbeforeenrolment,mettheinclusioncriteriaforcriticalappraisal.(20)Thisstudycomparedindacaterol150mcg(n=1721)withtiotropium18mcg(n=1718),bothadministeredoncedaily.Patientswhohadbeenusingastabledoseofinhaledcorticosteroidforatleastamonthbeforestudyentrywereinstructedtocontinuethisregimenforthedurationofthestudy.AdescriptionofthestudycharacteristicsisprovidedinTable4.Studysamplesizewasbasedonthekeysecondaryobjectiveofshowingnon-inferiorityofindacateroltotiotropiumintermsofrateofCOPDexacerbationsoverthe52-weektreatmentperiod.Onthebasisofavailablestudiesatthetimetheexpectedexacerbationratewas1.1peryearinbothgroups.Theoverdispersionfactorwasestimatedtobe1.6.Thenon-inferioritymarginselectedwas12%basedonhalfthereductioninexacerbations(20–25%)seeninpreviousstudiesofactivetreatmentsversusplacebo.Usingtheseassumptionsthesamplesizeneededwas1750patientspergroupfor80%power,includinganallowancefordropout.
Table4:INVIGORATE2013studycharacteristics
Participants N=3444COPDpatients≥40yearsofagewith:1)smokinghistoryof≥10pack-years;2)post-bronchodilatorFEV1between30%and<50%predictedvalue;3)post-bronchodilatorFEV1/FVC<0.7;and4)documentedhistoryof≥1moderateorsevereexacerbationinprevious12months.Exclusioncriteria:1)BMI<15kg/m2or>40kg/m2;2)respiratorytractinfectionorCOPDexacerbationneedingsystemiccorticosteroidswithin6weeksofscreeningvisit;or3)historyofasthma
Intervention Indacaterol150mcgOD[IND](n=1721)AvailableasOnbrezBreezhaler(75mcg)1inhalationOD
Comparator Tiotropium18mcgOD[TIO](n=1718)AvailableasSpirivaHandihaler(18mcg)1inhalationOD
ConcomitantMedications
Useofinhaledcorticosteroidwasallowedifapatient'streatmentregimenhadbeenstableforatleastamonthbeforestudyentry;patientswereinstructedtocontinuethisregimenforthedurationofthestudy.TreatmentswithfixeddosecombinationsofaLABAplusinhaledcorticosteroidwerediscontinuedbeforethestartofthestudy,aswerethosewithSAMAplusaSABAandthosewithLABAplusaLAMA.Treatmentwithafixed-dosecombinationofaLABAplusinhaledcorticosteroidwasreplacedbytheequivalentmonotherapyinhaledcorticosteroidforthedurationofthetrial.AllpatientswereprovidedwithaSABA,salbutamoloralbuterol,whichtheywereinstructedtousethroughoutthestudyasrescuetreatment.
Outcomes PRIMARY:• ChangefrombaselineintroughFEV1atweek12(per-protocolanalysis)
SECONDARY(pre-specified):• Rateofexacerbationsatweek52(per-protocolanalysis)
OTHER:
24
• TroughFEV1valuesatothertimepoints;FVCatweeks12,26and52• SGRQtotalscore• TDItotalscore• Changefrombaselinerescuemedicationandsymptomscores• ProportionofpatientswhoachievedaMCIDinSGRQtotalscore• ProportionofpatientswhoachievedaMCIDinTDItotalscore
StudyDesign Multicentre2-armparallelgroupDBRCTconsistingofa52-weektreatmentperiod
Therewerenosignificantdifferencesbetweentreatmentgroupsatbaselinewithregardtodemographics,COPDexacerbationsandSGRQtotalscore(Table5).Themeanageofstudypatientswas64.0years,77%weremales,and65%wereformersmokers.Post-bronchodilatorFEV1was40.5%ofpredictednormalvalueandameanSGRQtotalscoreof48.3(17.6)atscreening.Seventyninepercenthadahistoryof1COPDexacerbationinthepreviousyear.Useofspecificdrugsatbaselineisnotreported.SeventytwopercentwerereceivingICSatrandomization.
Table5:INVIGORATE2013baselinecharacteristicsofstudyparticipants
IND150mcg(n=1721)
TIO18mcg(n=1718)
Ageinyears,median(range) 64.0(40-91) 64.0(40-87)Males 1344(78%) 1313(76%)Whiterace 1330(77%) 1324(77%)Formersmokers 1125(65%) 1133(66%)NumberofCOPDexacerbationsinpreviousyear0123≥4
6(<0.5%)1365(79%)244(14%)59(3%)47(3%)
7(<0.5%)1342(78%)251(15%)69(4%)49(3%)
Post-bronchodilatorFEV1,mean%predicted(SD) 40.2(6.0) 40.7(6.1)SGRQtotalscoreMean(SD)
n=166447.9(17.4)
n=166948.7(17.8)
BaselineDyspneaIndex(BDI)totalscoreMean(SD)
n=16845.9(2.1)
n=16776.0(2.1)
DailyrescuetreatmentuseMeannumberofpuffs(SD)
n=16683.9(3.6)
n=16383.9(3.7)
InhaledcorticosteroidsuseYesNo
1235(72%)486(28%)
1234(72%)484(28%)
Overall,2711patients(79%)completedthetrial.Thisstudyanalyzed3072(89.3%)intheper-protocolsetforexacerbationsand3013(87.6%)intheper-protocolsetforspirometry.Datafor
25
exacerbationsandFEV1werenolongercollectedoncepatientsdiscontinuedfromthestudy.Deathswererecordedduringstudyparticipationandfor30daysafterstudydrugdiscontinuation.AsummaryofpatientdispositionisprovidedinTable6.
Table6:PatientdispositioninINVIGORATE2013
IND150mcg TIO18mcgRandomized 1721 1718Totaladverseevents 1119(65.0%) 1065(62.0%)Totalwithdrawals 386(22.4%) 342(19.9%)Withdrewconsent 105(6.1%) 108(6.3%)Withdrawalduetoadverseevents 101(5.9%) 96(5.6%)Withdrawalduetolackofefficacy 51(3.0%) 39(2.3%)Losttofollow-up 22(1.3%) 13(0.8%)
RiskofbiasinINVIGORATE2013TheCochraneRiskofBiasToolwasusedtoassessthequalityofINVIGORATE2013.Thisappraisaltoolhighlightsboththestrengthsandweaknessesofincludedstudies.Keyelementsoftrialmethodologyandreportingareassessedusingastandardizedsetofcriteria.Ifthemethodsareinadequatethereisa“highriskofbias”.Iftheriskofbiasis“unclear”usuallythetrialreportdidnotadequatelydescribethemethods.Ifthemethodologyandreportingareadequatethereisalowriskofbias.INVIGORATE2013isjudgedtohaveahighriskofbiasaccordingtotheCochraneRiskofBiasToolwithrespecttoattrition,selectivereportingandsourceoffunding.(Table7).
Table7:CochraneriskofbiassummaryforINVIGORATE2013Domain Judgement SupportforJudgementRandomsequencegeneration(selectionbias)
Lowrisk "Therandomisationsequencewascomputer-generatedbyaninteractivevoiceresponsesystem(IVRS;OracleAmericaInc,RedwoodCity,CA,USA)...Randomisationwasstratifiedbyinhaledcorticosteroiduse,withbalanceoftreatmentsmaintainedatcountrylevel.Balancewasmaintainedatthestratalevelbyusingrandomlypermutedblocks."
Allocationconcealment(selectionbias)
Lowrisk "Therandomisationsequencewascomputer-generatedbyaninteractivevoiceresponsesystem(IVRS;OracleAmericaInc,RedwoodCity,CA,USA)..."
Blindingofparticipantsandpersonnel(performancebias)
Lowrisk "Participants,investigatorsandthoseassessingtheoutcomesweremaskedtotreatmentintervention.Weusedadouble-dummydesignbecausetheidentityofthestudydrugs(indacaterolandtiotropium)couldnotbedisguisedduetotheirdifferentinhalerdevices.Patientsweredispensedthetwodevicesandtheircorrespondingcapsules(eitheractiveor
26
Domain Judgement SupportforJudgementplacebo).Drugpacksweredispensedbyathirdpartywhowasnotinvolvedinanyotheraspectofthestudy."
Blindingofoutcomeassessment(detectionbias)
Lowrisk "...thoseassessingtheoutcomesweremaskedtotreatmentintervention."
Incompleteoutcomedata(attritionbias)
Highrisk
3439patientswererandomised.“Therewere3419patientsintheFASpopulation,3072(89.3%)intheper-protocolsetforexacerbationsand3013(87.6%)intheper-protocolsetforspirometry.”“Analysisfortheprimaryandkeysecondaryendpointswasdoneinper-protocolpopulations.Thesepopulationsincludedallrandomisedpatientswhoreceivedatleastonedoseofstudydruganddidnothaveanymajorprotocoldeviations(aspre-definedintheanalysisplan)thatcouldaffecttheanalysisofspirometryorexacerbationsdata.Allotheranalysesofefficacyendpointsweredoneonthefullanalysisset(FAS),whichconsistedofallrandomisedpatientswhoreceivedatleastonedoseofstudydrug.”“Oncepatientsdiscontinuedfromthestudy,eitherbeinglosttofollow-uporbyundertakingastudydiscontinuationvisit,dataforFEV1andexacerbationswerenolongercollected.”Deathswererecordedduringstudyparticipationandfor30daysafterstudydrugdiscontinuation.Thehighwithdrawalrateswillleadtoattritionbias.
Selectivereporting(reportingbias)
Unclearrisk ThisstudyisregisteredwithClinicalTrials.gov,numberNCT00845728,andtheprotocolliststheprimaryoutcomeandthepre-specifiedsecondaryoutcomeonly.“Werecordedadverseeventsandseriousadverseevents(includingadmissionstohospitalanddeaths),alongwithphysicalexaminationmeasurements,vitalsigns,laboratoryassessments,bloodtestresults,andelectrocardiogramreadings.”Totalhospitalizationsarenotreported.
Otherbias Highrisk “RC,MS,DL,DY,andDMcBareemployeesofthetrialsponsor(Novartis)andcontributedtothedesign,preparation,andconductofthe
27
Domain Judgement SupportforJudgementstudy.Theyalsomadesubstantialcontributionstotheanalysisandinterpretationofthestudy.MDwastheprincipalinvestigatorofthestudywhocriticallyreviewedthefullstudyreportandhadthefinalresponsibilityforthedecisiontosubmitforpublication.Allauthorshadfullaccesstotheresultsofpre-specifiedstatisticalanalyses,wereencouragedtosuggestappropriatepost-hocanalyses,andmadesubstantialcontributionstothecontentofeachdraft.”
OutcomesreportedResultsarepresentedinTable6accordingtotheoutcomehierarchydescribedabove.
Table8:HierarchyofoutcomesinINVIGORATE2013 IND150mcg
(n=1721)TIO18mcg(n=1718)
Totalmortality 28(1.6%) 28(1.6%)TotalSAEsTotalhospitalizationsHospitalizationduetosevereCOPDexacerbationWorseningofCOPDasaSAE
263(15%)
NR
NR
147(8.5%)
255(15%)
NR
NR
121(7.0%)Numberofpatientswith≥1moderateorsevereCOPDexacerbationNumberofpatientswith≥1severeCOPDexacerbation
NR
NR
NR
NR
SGRQtotalscoreatWeek52,basedonsubsetofpatientsPatientsevaluatedChangefrombaseline,mean(SD)PatientsevaluatedSGRQtotalscore,mean(SE)LSmeandifference%Patientswith≥4pointimprovement(MCID)
1281(74.4%)-4.5(15.5)
1273(74.0%)42.3(0.66)
0.2(95%CI-0.8,1.2)p=NS
626/1273(49%)OR1.03(95%CI0.88,
1.21)p=NS
1325(77.1%)-4.9(14.8)
1314(76.5%)42.2(0.65)
646/1314(49%)
Timeto1stmoderateorsevereexacerbation, 1.20(1.07,1.33)
28
IND150mcg(n=1721)
TIO18mcg(n=1718)
HR(95%CI) p=0.0012TDItotalscoreatWeek52,basedonsubsetofpatientsPatientsevaluatedChangefrombaseline,mean(SD)PatientsevaluatedTDItotalscore,mean(SE)LSmeandifference(95%CI)%Patientswith≥1unitimprovement(MCID)
1296(75.3%)2.22(3.53)
1288(74.8%)2.01(0.17)
0.26(95%CI0.04,0.05)p=0.02
745/1288(58%)OR1.12(95%CI0.96,
1.31)p=NS
1332(77.5%)1.92(3.56)
1322(77.0%)1.75(0.16)
728/1322(55%)
Rescuetreatmentover52weeks,basedonsubsetofpatientsPatientsevaluatedChangefrombaselineindailynumberofpuffs,mean(SE)LSmeandifference(95%CI)Patientsevaluated%ofdayswithnorescueuse,mean(SE)LSmeandifference(95%CI)
1584(92.0%)-1.01(0.11)
-0.62(-0.79,-0.45)
p<0.00011561(90.7%)42(1.39)
8.0(5.9,10.2)p<0.0001
1561(90.9%)-0.39(0.11)
1526(88.8%)34(1.40)
TroughFEV1basedonsubsetofpatientsPatientsevaluatedChangefrombaseline(mL)LSmeandifference
1324(76.9%)
73-20;p=0.022
1362(79.3%)
92
1. Totalmortality
Therewasnodifferenceinall-causemortalitybetweenindacateroIandtiotropiumgroups.
2. SAEs
TherewerenodifferencesintotalSAEsandCOPDasaSAEbetweentreatmentwithindacaterolandtiotropium.All-causehospitalizationwasnotreported.
3.AcutemoderateorsevereCOPDexacerbations
a. Amoderateexacerbationwasdefinedasanexacerbationleadingtotreatmentwithantibioticsorsystemiccorticosteroids.Asevereexacerbationwasdefinedasanexacerbationthatledtoadmissiontohospital(includingavisittotheemergencyroomformorethan24h)inadditiontotreatmentwithsystemiccorticosteroidsorantibiotics.
29
CRITICALAPPRAISALISSUE:Giventhatthismulticentertrialwasconductedin41countriestherewillbevariabilityintreatmentpracticesofmoderateCOPDexacerbationsacrosscentersthatcouldbiasthestudyfindings.
b. Thenumberofpatientswithoneormoreacutemoderateorsevereexacerbationswasnotreported.Insteadtime-to-first-eventanalysisreportedthatindacaterolwasassociatedwithahigherriskofmoderateorsevereexacerbationduringtreatmentthantiotropium.Thehazardratio(HR)onthereportedstudysampleforindacaterolversustiotropiumwas1.20(95%CI1.07to1.33;p=0.0012).
CRITICALAPPRAISALISSUE:Time-to-first-eventanalysisisusefulonlywhenadditionalinformationonthenumberofpatientswhohadmorethanoneexacerbationthroughoutthestudyinbothtreatmentgroupsisprovided.(32)
Infact,attritionbiasinthisstudyandlackofdatacollectiononexacerbationeitherasthefirsteventortotalnumberofexacerbationsthroughoutthestudyperiodpostwithdrawalwillalsoaffecttheanalysis.
4.Health-relatedqualityoflife
SGRQwasusedtomeasurehealth-relatedqualityoflifeinthisstudy.SGRQtotalscorerangesfrom0to100,withlowerscoresindicatingbetterhealth-relatedqualityoflife.Aminimumchangeinscoreof4pointsisconsideredasclinicallyimportant(i.e.MCID).
MeanchangeinSGRQtotalscoreandproportionofpatientswithaMCIDwereevaluatedin2606(76%)patients.Inthissubsetofpatientstherewerenosignificantdifferencesbetweentheindacaterolgroupandthetiotropiumgroupfortheseoutcomemeasures.
CRITICALAPPRAISALISSUE:SGRQtotalscorewasonlyreportedforasubsetof2606(76%)patients.Thedatafor24%ofrandomizedpatientsaremissing.AnalysisoftheeffectoftreatmentonSGRQtotalscoreshouldbebasedonallrandomizedpatientsratherthanincompletedatafromasubsetofpatients.
5.Symptomaticimprovement
TDIscorewasusedtomeasuretheseverityofdyspnea(breathlessness,shortnessofbreath)inthisstudy.TDIscorerangesfrom-9to9,withalowerscoreindicatingmoredeteriorationinseverityofdyspnea.Aminimumimprovementof1pointisconsideredaMCID.
TDIscoreandproportionofpatientsachievingaMCIDwereonlyreportedinasubsetof2628(76%)ofrandomizedpatients.Therewerenosignificantdifferencesbetweentreatmentgroups.
CRITICALAPPRAISALISSUE:TDIscorewasonlyreportedforasubsetof2626(76%)patients.Thedatafor24%ofrandomizedpatientsaremissing.AnalysisoftheeffectoftreatmentonTDIscoreshouldbebasedonallrandomizedpatientsratherthanincompletedatafrom76%ofrandomizedpatients.
30
6.Useofrescuesalbutamol
ReducedneedforrescuemedicationisamarkerofimprovedcontrolofCOPDsymptoms.Rescuetreatmentuse,intermsofchangeindailynumberofpuffsandproportionofdayswithnouseofrescuemedication,wasonlyreportedinasubsetof3145(92%)and3087(90%)ofrandomizedpatients,respectively.Patientsintheindacaterolgroupneededrescuetreatmentlessoftenascomparedtothosewhoreceivedtiotropium[LSmeandifference-0.62(95%CI-0.79,-0.45);p<0.0001)indailynumberofpuffs;LSmeandifference8.0%(95%CI5.9,10.2);p<0.0001inproportionofdayswithnorescueuse).
CRITICALAPPRAISALISSUE:Theclinicalimportanceofasmallreductionintheneedforrescuetreatmentisquestionable.Ifindacaterolactuallyreducestheneedforrescuemedication,asignificantimprovementinSGRQandTDIscoresisalsoexpectedinthisgroup.However,bothSGRQtotalscoreandTDIscorewerenotdifferentfromtiotropiumgroup.Asnotedbytheauthorsofthisstudyinthetimetofirsteventanalysisofexacerbation,indacaterolisclaimedtobeassociatedwithhigherriskofmoderateorsevereexacerbationduringtreatmentversustiotropium.Itisdifficulttounderstandhowthiswouldtranslateintoareducedneedforrescuemedication.
7.COPDrelatedhealthcareutilization
Thisincludesphysicianvisits/ERvisitsandhospitalization.Itisanotheroutcomethatwasnotreportedinthestudy.
8.Adverseevents
a. Adverseeventsoccurredin1119(65%)receivingindacateroland1065(62%)receivingtiotropium.Therewasnodifferencebetweenindacaterolandtiotopriumfortotaladverseevents.
b. Atotalof101(5.9%)and96(5.6%)patientstreatedwithindacaterolandtiotropium,respectively,withdrewduetoanadverseevent.Therewasnobetween-groupdifferenceforwithdrawalduetoadverseevents.
CRITICALAPPRAISALISSUE:Overall,2711patients(79%)completedthetrial.Thestudyreportdoesnotprovidesufficientinformationontheanalysisofharmdataalthoughitdoesstatethatpatientswhopermanentlydiscontinuedstudytreatmentdidnotcomeinforfurtherevaluation.Whetherdatawerecollectedforadverseeventsfollowingdiscontinuationfromthestudyisnotreported.
9.FEV1
In2686(78%)patientsevaluated,themeanchangefrombaselineintroughFEV1intheindacaterolandtiotropiumgroupswas73mland92ml,respectively.Astatisticallysignificant
31
butnotclinicallyrelevantbetween-groupdifferenceinfavouroftiotropiumwasseeninLSmeantroughFEV1atstudyendpoint[-20ml;p=0.02].
CRITICALAPPRAISALISSUE:FEV1isasurrogateoutcomethathasvalidityinestimatingtheriskofdyingfromCOPDbutlittleuseinassessingtheimpactofinhaleddrugtherapyonCOPDsymptoms.(3)
Othercriticalappraisalissues:
1. ThedoseofindacaterolusedinINVIGORATE2013is150mcgODwhentheapproveddosebyHealthCanadais75mcgOD.
2. Thesamplesizewasbasedonnon-inferiorityofindacateroltotiotropiumforrateofallexacerbations.Thestudywasnotpoweredtodetectadifferencebetweentreatmentgroupsinclinicallyrelevantoutcomessuchasreductioninexacerbation,improvementinqualityoflifeordecreaseinmortalityorhospitalizations.
LABA+LAMAvs.LABAOnlyonestudy(Donohue2016)wascriticallyappraised,whichisadoubleblindRCTinvestigatingthelong-termsafetyandtolerabilityoftwice-dailyaclidinium/formoterol(ACL/FOR)400/12mcg(n=392)versusformoterol(FOR)12mcg(n=198)in590patientswithsymptomaticCOPD.(22)ICSandoralorparenteralcorticosteroidsatdoses≤10mg/day,theophyllineandH1-antihistaminewerepermittedforchronicuseprovidedthedosagewasstablefor≥4weekspriortoscreeningandthroughoutthetrial.Patientswerepermittedtreatmentwithalbuterolasneeded,butnotwithin6hbeforeavisit.Chronicuseofoxygentherapywasalsopermittedprovidedthedosagewasstablefor≥4weekspriortoscreening.ThestudycharacteristicsareprovidedinTable9.Theprimaryobjectiveofthisstudywaslong-termsafetyandtolerabilityofACL/FOR400mg/12mgandallefficacyoutcomeswerepre-definedasadditionalendpoints.Originally450patientsweretoberandomizedasitwasconsideredsufficienttomeetthesafetyobjectivesofthestudyandtoobtainlong-termsafetydataforregulatoryrequirements.Itwasnotbasedonstatisticalpowertomeetanefficacyobjective.
Table9:Donohue2016studycharacteristics
Participants N=590COPDpatients≥40yearsofagewith:1)smokinghistoryof≥10pack-years;2)diagnosisofmoderatetosevereCOPD;3)post-bronchodilatorFEV1≥30%and<80%predictedvalue;and4)post-bronchodilatorFEV1/FVC<0.7Exclusioncriteria:AnyrespiratoryinfectionorCOPDexacerbation≤6weeksbeforescreening;pulmonaryrehabilitationwithin3monthsofscreeningoranintentiontostartduringthetrial;clinicallysignificantcardiovascularconditions,includingmyocardialinfarction≤6months;newlydiagnosedarrhythmia≤3months;unstableangina;unstablearrhythmiathathadrequiredchangesinpharmacologicaltherapyorotherinterventions≤6months;useofanautomatedimplantablecardioverter-defibrillator;historyofthoracicsurgery≤1
32
yearofscreening;hospitalization≤12monthsforheartfailure(NYHAclassIII)orhistoryofthoracicsurgery≤1yearofscreeningandNYHAclassIV;QTcB>470msatrest;orBMI≥40kg/m2
Intervention Aclidinium/formoterol400/12mcgBID[ACL/FOR]administeredviaGenuiar/Presairdevice(n=392)AvailableasDuaklirGenuair(400mcg/12mcg)1inhalationBID
Comparators Formoterol12mcgBID[FOR]administeredviaGenuiar/Presairdevice(n=198)AvailableasForadilAerolizer(12mcg1-2)inhalationsBID
Patientswerepermittedtreatmentwithalbuterolasneeded,butnotwithin6hbeforeavisit.ICSandoralorparenteralcorticosteroidsatdoses≤10mg/day,theophyllineandH1-antihistaminewerepermittedforchronicuseprovidedthedosagewasstablefor≥4weekspriortoscreeningandthroughoutthetrial.Chronicuseofoxygentherapywaspermittedforupto15h/dayprovidedthedosagewasstablefor≥4weekspriortoscreening.Indacaterolwasnotallowedwithin15dayspriortoscreeningorduringthetrial.Selectβ1-blockingagents(atenolol,metoprolol,nebivolol)werepermittedforchronicuseifthedosagewasstablefor≥2weekspriortoscreening.
Outcomes PRIMARY:• Proportionofpatientswhoexperienced≥1Treatment-Emergent
AdverseEvent(TEAE)uptoweek56SECONDARY(prespecified):
• ProportionofpatientswhoexperiencedanyPotentiallyClinicallySignificant(PCS)post-baselinechangeinclinicallaboratoryvaluesforhematology,chemistryorurinalysis
• ProportionofpatientswhoexperiencedanyPCSpost-baselinechangeinpulserate,systolicanddiastolicbloodpressure
• ProportionofpatientswhoexperiencedPCSchangesinECGfrombaseline
OTHER:• ChangefrombaselineintroughFEV1at1,12,24,38and52weeks• ChangefrombaselineintroughFVCat1,12,24,38and52weeks• Changefrombaselineintotaldailyrescuemedicationuseover52weeks• RateofCOPDexacerbations,definedasanincreaseinCOPDsymptoms
thatrequiredachangeinCOPDtreatmentStudyDesign Multicentre2-armparallelgroupDBRCTconductedtofulfillFDAsafety
requirementsandcompriseda2-3-weekrun-inperiod,52-weekdouble-blindtreatmentperiod,andafollow-upphonecall4weeksafterlasttreatmentdose
TherewerenosignificantdifferencesamongtheACL/FORandFORtreatmentgroupsinbaselinecharacteristics(Table10).Themeanageofstudypatientswas64.2(9.3)years,55%weremales,54%wereformersmokersand23.9%had≥1exacerbationinthepreviousyear.Post-bronchodilatorFEV1was51.4%ofpredictednormalvalueatscreening.Fiftytwopercent
33
and46%wereclassifiedasGOLDStageII(moderate)andStageIII(severe),respectively.TherewerenodifferencesinuseofconcomitantCOPDdrugspriortoandcontinuingduringstudywith38%usingdrugsfromanycategoryand35%receivingICS.
Table10:Donohue2016baselinecharacteristicsofstudyparticipants
ACL/FOR400/12mcgBID(n=392)
FOR12mcgBID(n=198)
Ageinyears,mean(SD) 63.9(9.3) 64.7(9.4)Males 216(55.1%) 109(55.1%)Whiterace 364(92.9%) 181(91.4%)Formersmokers 208(53.1%) 111(56.1%)≥1COPDexacerbationinpreviousyear 89(22.7%) 52(26.3%)Post-bronchodilatorFEV1,mean%predicted(SD)
51.8(13.0) 50.5(13.5)
COPDGOLDStageII(Moderate)III(Severe)
207(52.8%)181(46.2%)
102(51.5%)92(46.5%)
DailyrescuemedicationuseMeannumberofpuffs(SD)
n=3854.5(3.7)
n=1964.5(3.6)
ConcomitantmedicationspriortoandcontinuingduringstudyAnycategoryICSLABALABA+ICSLAMASystemicantihistaminesTheophylline
150(38.3%)138(35.2%)
000
26(6.6%)4(1.0%)
72(36.6%)68(34.3%)
01(0.5%)2(1.0%)17(8.6%)1(0.5%)
Only398patients(68%)completedthetrial.All590patientswereincludedinthesafetyanalysis.Patientswhodiscontinuedthestudyprematurelydidnotcomeinforfurtherevaluation.Thestudyreportdoesnotstateifdataforexacerbationsandotherefficacyoutcomeswerecollectedfollowingdiscontinuationfromthestudy.VitalstatuswasavailableforthetotalstudypopulationatWeek52.AsummaryofpatientdispositionisprovidedinTable11.
Table11:PatientdispositioninDonohue2016
ACL/FOR400/12mcgBID FOR12mcgBIDRandomized 392 198Totaladverseevents 280(71.4%) 130(65.7%)Totalwithdrawals 127(32.4%) 65(32.8%)Withdrewconsent 31(7.9%) 15(7.6%)Withdrawalduetoadverseevents 26(6.6%) 13(6.6%)Withdrawalduetolackofefficacy 23(5.9%) 11(5.6%)Losttofollow-up 3(0.8%) 5(2.5%)
34
RiskofbiasinDonohue2016Donohue2016isjudgedtohaveahighriskofbiasaccordingtotheCochraneRiskofBiasToolwithrespecttodetectionbias,attritionbiasandsourceoffunding.Thestudyisjudgedtohaveunclearriskofbiaswithrespecttoblindingandselectivereporting(Table12).Themainconcernisthehighriskofbiasrelatedtoefficacyoutcomesgiventhatthisstudyislikelynotblindedforefficacyoutcomesandthehighwithdrawalrates(32%)willleadtoattritionbiasforexacerbationoutcomes.
Table12:CochraneriskofbiassummaryforDonohue2016Domain Judgement SupportforJudgementRandomsequencegeneration(selectionbias)
Lowrisk "RandomizationwascarriedoutbyassigningpatientidentificationnumbersviaaninteractivewebresponsesystemprovidedbyPremierResearchGroupLimited(EastHartford,Connecticut,USA)."
Allocationconcealment(selectionbias)
Lowrisk "RandomizationwascarriedoutbyassigningpatientidentificationnumbersviaaninteractivewebresponsesystemprovidedbyPremierResearchGroupLimited(EastHartford,Connecticut,USA)."
Blindingofparticipantsandpersonnel(performancebias)
Unclearrisk Notdescribedinsufficientdetailasidefrombothinterventionswereadministeredvia“doubleblind”Genuiar/Presairdevice.
Blindingofoutcomeassessment(detectionbias)
Highrisk “Majoradversecardiacevents(MACE)wereevaluatedandclassifiedaccordingtothecriteriaprespecifiedbythreeblindedindependentexpertcardiologistsnotparticipatinginthestudy.”Likelynotblindedforefficacyoutcomes.
Incompleteoutcomedata(attritionbias)
Highrisk
All590patientswereincludedinthesafetyanalysis.Thestudyreportdoesnotstateifdataforexacerbationswerecollectedinpatientswhodiscontinuedfromthestudyandsubsequentlyhadanexacerbation.Thehighwithdrawalrates(32%)willleadtoattritionbiasforexacerbationoutcomesbutnotforsafetyoutcomes.
Selectivereporting(reportingbias)
Unclearrisk ThisstudyisregisteredwithClinicalTrials.gov,numberNCT01437540andthestudyprotocolisprovided.Thestudypublicationreportsalloutcomesspecifiedintheprotocol.Prespecifiedsecondaryoutcomesarereportedinlimiteddetail:“Comparedwithbaseline,noclinicallyrelevantchangesinclinicalhematology,biochemicalandurinalysisparameters,vitalsigns,orECGswereattributed
35
Domain Judgement SupportforJudgementtoaclidinium400mg/formoterol12mgfixed-dosecombination.”Other(efficacy)outcomesthatwerenotpre-specifiedarereported.
Otherbias Highrisk "ThisstudywassupportedbyForestLaboratoriesLLC,awhollyownedsubsidiaryofAllerganplc,andbyAlmirallS.A.Bothfunderswereinvolvedinthestudydesign,andcollection,analysisandinterpretationofthedata.ThedevelopmentofthismanuscriptwassupportedbyForestLaboratories,LLC;thedecisiontosubmitthemanuscriptforpublicationwasmadejointlybythefundersandauthors.""MedicalwritingandeditingwasprovidedbyMaryClareKane,PhD,andKristenA.Andersen,PhD,ofPrescottMedicalCommunicationsGroup(Chicago,IL,US),fundedbyForestLaboratoriesLLC,awhollyownedsubsidiaryofAllerganplc(JerseyCity,NJ,USA),andbyRichardKnight,PhD,ofCompleteMedicalCommunications(Macclesfield,UK),fundedbytheAstraZenecagroupofcompanies."
OutcomesreportedResultsarepresentedinTable13accordingtothehierarchyofoutcomesdescribedabove.
Table13:HierarchyofoutcomesinDonohue2016 ACL/FOR400/12mcgBID
(n=392)FOR12mcg(n=198)
Totalmortality 5(1.3%) 1(0.5%)TotalSAEsTotalhospitalizationsHospitalizationduetosevereCOPDexacerbationWorseningofCOPDasaSAE
38(9.7%)NRNRNR
21(10.6%)NRNRNR
Numberofpatientswith≥1moderateorsevereCOPDexacerbationOddsratio(95%CI)Numberofpatientswith≥1severeCOPDexacerbation
99(25.3%)0.88(0.60,1.29)p=NSNR
55(27.8%)NR
36
ACL/FOR400/12mcgBID(n=392)
FOR12mcg(n=198)
SGRQtotalscoreatWeek64SGRQtotalscore,mean(SE)Patientswith≥4pointimprovement(MCID)
NRNR
NRNR
RateofmoderatetosevereCOPDexacerbationsRateperpatient-year(95%CI)BetweengroupdifferenceTimeto1stmoderateorsevereexacerbation,HR(95%CI)RateofCOPDexacerbationsleadingtohospitalisationTimeto1stexacerbationleadingtohospitalisation,HR(95%CI)
0.52p=NSNRNRNR
0.49NRNRNR
TDItotalscore NR NRRescuetreatmentover52weeks,basedonITTpopulationPatientsevaluatedChangefrombaselineindailynumberofpuffs,LSmean(SE)
n=385-1.9(0.10)
n=196-1.6(0.16)
TroughFEV1atweek64Changefrombaseline(mL)Meandifference(95%CI)
14381.5(12.5,150.5)p<0.05
62
1. Totalmortality
TotalmortalitydidnotsignificantlydifferbetweenACL/FORandFORgroups.
2. SAEs
Therewasnosignificantbetween-groupdifferenceintotalSAEs.All-causehospitalizationwasnotreported.
3.AcutemoderateorsevereCOPDexacerbations
a. Thestudyreportdoesnotdefineamoderateexacerbationorsevereexacerbation.ItsimplystatesthatanexacerbationwasdefinedasanincreaseinCOPDsymptomsthatrequiredachangeinCOPDtreatment
CRITICALAPPRAISALISSUE:Giventhatthismulticentertrialwasconductedin127centersintheUStherewillbevariabilityintreatmentpracticesofmoderateCOPD
37
exacerbationsacrosscentersthatcouldbiasthestudyfindings,especiallyifamoderateexacerbationisnotclearlydefined.
Therewasnosignificantdifferenceinthenumberofpatientswithoneormoreacutemoderateorsevereexacerbationsbetweenthe2treatmentarms[OR0.88(95%CI0.60,1.29)].Thetrialalsoreportstherateofmoderateorsevereexacerbations,whichwas0.52perpatient-yearwithACL/FORversus0.49perpatient-yearwithFORalone(p=NS).
CRITICALAPPRAISALISSUE:Thereportedexacerbationeventsandratesarealsouncertainduetothehighwithdrawalratesinthestudy(32and33%inACL/FORandFOR,respectively).Itisunclearwhetherpatientswhowithdrewprematurelywereappropriatelyaccountedforinthisstudy.
4.Health-relatedqualityoflife
SGRQtotalscorewasnotaprespecifiedoutcomeofthisstudy.
5.Symptomaticimprovement
TDIscorewasnotaprespecifiedoutcomeofthisstudy.
6.Useofrescuesalbutamol
UseofrescuetreatmentdidnotsignificantlydifferbetweenACL/FORandFORgroups.
7.COPDrelatedhealthcareutilization
COPDrelatedhealthcareutilizationwasnotaprespecifiedoutcomeofthisstudy.
8.Adverseevents
a. Adverseeventsoccurredin280(71.4%)receivingACL/FORand130(65.7%)receivingUMEC/VI.Therewasnobetween-groupdifferenceintotaladverseevents[OR1.31(95%CI0.91,1.89)].
b. Atotalof26(6.6%)patientstreatedwithACL/FORversus13(6.6%)patientstreatedwithFORalonewithdrewduetoanadverseevent.Therewasnodifferencebetweentreatmentgroupsforwithdrawalduetoadverseevents.
9.FEV1
ThedifferencebetweenACL/FORandFORgroupsinthemeanchangefrombaselineintroughFEV1atWeek64was143mland62ml,respectively.Themeandifferencebetweengroupswasstatisticallysignificant[81.5(95%CI12.5,150.5);p<0.05)]buttheclinicalrelevanceofthisdifferenceinfavourofACL/FORisunknown.
CRITICALAPPRAISALISSUE:FEV1isasurrogateoutcomethathasvalidityinestimatingtheriskofdyingfromCOPDbutlittleuseinassessingtheimpactofinhaleddrugtherapyonCOPDsymptoms.(3)
38
Othercriticalappraisalissues:
1. Thesamplesizewasbasedontreatmentemergentadverseevents.Thestudywasnotpoweredtodetectadifferencebetweentreatmentgroupsinclinicallyrelevantoutcomessuchasreductioninexacerbation,improvementinqualityoflifeordecreaseinmortalityorhospitalizations.
LABA+LAMAvs.LAMATwostudies(DYNAGITO2018andSPARK2016)wereidentifiedintheupdatesearchthatfollowedpatientsforatleastayear.A. DYNAGITO2018isadoubleblindRCTin7880patientswithsymptomaticCOPDandahistory
ofmoderateorsevereexacerbationintheprecedingyear.(23)Thisstudycompareddualbronchodilatortherapywitholodaterol/tiotropium(OLO/TIO)5/5mcg(n=3939)withtiotropium(TIO)5mcg(n=3939),bothadministeredoncedailyviatheRespimatdevice.PatientstakingICSatbaselinecontinuedthistreatment.Open-labelsalbutamolwasprovidedforas-neededrescuemedicationuse,butothershort-actingbeta-agonists,LAMAsandLABAswerenotpermittedduringthestudy.AdescriptionofthestudycharacteristicsisprovidedinTable14.Asamplesizeof3900patientspergroupwastoprovidesufficientpowertodetecta12%reduction(basedontheresultsofapreviousstudy)inrateofexacerbationswithOLO/TIOcomparedwithTIO.Thissamplesizeallowedfor15%lossofdataduetopatientwithdrawals.
Table14:DYNAGITO2018studycharacteristics
Participants N=7880COPDpatients≥40yearsofagewith:1)smokinghistoryof>10pack-years;2)post-bronchodilatorFEV1<60%predictedvalue;3)post-bronchodilatorFEV1/FVC<0.7;and4)historyof≥1moderateorsevereexacerbationinprecedingyearrequiringtreatmentwithsystemiccorticosteroidsorantibioticsorboth,withorwithouthospitalisationExclusioncriteria:1)currentdiagnosisofasthma;2)severeemphysemarequiringendobronchialinterventionsintheprevious6months;3)treatmentwithantibioticsforanyreasonwithin4weeksofscreening;or4)PDE-4inhibitorsusewithin3monthsofscreening
Intervention Olodaterol/tiotropium5/5mcgOD[OLO/TIO]viaRespimatdevice(n=3939)AvailableasInspioltoRespimat(2.5mcg/2.5mcg)2inhalationsOD
Comparator Tiotropium5mcgOD[TIO]viaRespimatdevice(n=3941)AvailableasSpirivaRespimat(2.5mcg)1inhalationsOD
ConcomitantMedications
PatientstakingICSatbaselinecontinuedthistreatment;thosereceivingICSinafixed-dosecombinationwithaLABAwereswitchedtoanequivalentcorticosteroidmonotherapy.Open-labelsalbutamolwasprovidedforas-neededrescuemedicationuse,butothershort-actingbeta-agonists,LAMAsandLABAs(otherthanthestudymedication)werenotpermittedduringthe
39
study.Outcomes PRIMARY:
• RateofmoderatetosevereCOPDexacerbationsduringthe“actualtreatmentperiod”,definedasthetimefromthefirstdoseofmedicationuntil1dayafterlastdrugadministration
SECONDARY(pre-specified):• Numberofpatientswith≥1moderatetosevereCOPDexacerbation
duringthe“actualtreatmentperiod”• Numberofpatientswith≥1COPDexacerbationleadingto
hospitalizationduringthe“actualtreatmentperiod”• RateofCOPDexacerbationsleadingtohospitalisationduringthe“actual
treatmentperiod”• All-causemortalityduringthe“actualtreatmentperiod”
OTHER(notpre-specified):• Timetoall-causemortalityduringthe“on-treatmentperiod”,definedas
thetimefromfirstdoseofmedicationuntil21daysafterlastdose• TimetofirstactualtreatmentmoderateorsevereCOPDexacerbation• TimetofirstactualtreatmentCOPDexacerbationleadingto
hospitalisation• RateofactualtreatmentCOPDexacerbationstreatedwithantibiotics
RateofactualtreatmentCOPDexacerbationstreatedwithcorticosteroidsandantibiotics
StudyDesign Multicentre2-armparallelgroupDBRCTconsistingofa52-weektreatmentperiod
TherewerenosignificantbaselinedifferencesbetweenOLO/TIOandTIOgroupsintermsofdemographics,COPDexacerbationsandSGRQtotalscore(Table15).Themeanageofstudypatientswas66.4(8.5)years,72%weremales,and63%wereformersmokers.Post-bronchodilatorFEV1was44.6%ofpredictednormalvalueandameanSGRQtotalscoreof47.8(17.7)atscreening.Fortyfivepercenthadahistoryof≥2exacerbationsor≥1severeexacerbationintheprecedingyear.Nearly40%ofthepatientswerereceivingtripletherapy(LABA+LAMA+ICS),and26%werereceivingLABA+ICSatrandomization.Itisnotreportedwhetherdualtherapy(LAMA/LABAorLABA/ICS)actuallyfailedinthosepatientsreceivingtripletherapyatscreening.
Table15:DYNAGITO2018baselinecharacteristicsofstudyparticipants
OLO/TIO5/5mcg(n=3939)
TIO5mcg(n=3941)
Ageinyears,mean(SD) 66.5(8.4) 66.3(8.5)Males 2785(71%) 2841(72%)Whiterace 3134(80%) 3113(79%)Formersmokers 2505(64%) 2462(62%)Patientswith≥2COPDexacerbationsor≥1severe 1754(45%) 1733(44%)
40
OLO/TIO5/5mcg(n=3939)
TIO5mcg(n=3941)
exacerbationinpreviousyear Post-bronchodilatorFEV1,mean%predicted(SD) 44.6(37.5) 44.5(11.5)SGRQtotalscore,mean(SD) 48.1(17.7) 47.4(17.7)GOLDclassABCDMissing
260(7%)
1922(49%)176(4%)
1569(40%)12(<1%)
308(8%)
1895(48%)176(4%)
1547(39%)15(<1%)
RespiratorymedicationLABAonlyLAMAonlyICSonlyLABA+ICSLAMA+ICSLAMA+LABALAMA+LABA+ICSNeither
122(3%)365(9%)107(3%)
1031(26%)78(2%)
461(12%)1555(39%)220(6%)
135(3%)350(9%)93(2%)
1005(26%)88(2%)
478(12%)1577(40%)215(5%)
Overall,6742patients(86%)completedthetrial.FewerpatientsreceivingOLO/TIO(12.4%)withdrewfromthestudyascomparedtoTIOalone(16.5%)[OR0.72(95%CI0.63,0.81);p<0.00001].Thisstudyanalyzedexacerbationdataduringthe“actualtreatmentperiod”,definedasthetimefromfirstdoseofmedicationuntil1dayafterthelastdoseofmedication.Patientswhopermanentlydiscontinuedstudytreatmentdidnotcomeinforfurtherevaluationsothereislossofinformationonexacerbationeventsfollowingprematurediscontinuationofstudytreatment.Vitalstatuswasavailablefor99.6%ofthetotalstudypopulationattheendofthestudy.Table16providesasummaryofpatientdisposition.
Table16:PatientdispositioninDYNAGITO2018
OLO/TIO5/5mcg TIO5mcgRandomized 3939 3941Totaladverseevents 1119(65.0%) 1065(62.0%)Totalwithdrawals 488(12.4%) 650(16.5%)Withdrewconsent 131(3.3%) 184(4.7%)Withdrawalduetoadverseevents 259(6.6%) 348(8.8%)Withdrawalduetolackofefficacy 37(0.9%) 59(1.5%)Losttofollow-up 131(3.3%) 184(4.7%)
RiskofbiasinDYNAGITO2018TheCochraneRiskofBiasToolwasusedtoassessthequalityofDYNAGITO2018.Thisstudyisjudgedtohaveahighriskofbiaswithrespecttoattrition,selectivereportingandsourceoffunding(Table17).Inparticular,analysisofexacerbationendpointswasdoneoneventsthatoccurredduringthe“actualtreatmentperiod”.Patientswhowithdrewprematurelyand
41
subsequentlyhadanexacerbationoutsidethe“actualtreatmentperiod”arenotaccountedforintheanalysis.Thefactthatmorepatientsreceivingtiotropiumalonewithdrewfromthestudycouldbiastheanalysisofexacerbationsdata.
Table17:CochraneriskofbiassummaryforDYNAGITO2018Domain Judgement SupportforJudgementRandomsequencegeneration(selectionbias)
Lowrisk "Aninteractiveresponsetechnologysystemwasusedforrandomisationandallocationoftrialmedication.Therandomisationschemeusedarandomisedblockdesign(blocksize4)andwasgeneratedusingvalidatedrandomisationsoftwarebyBoehringerIngelheim,whichalsopreparedandcodedthemedicationstomaintainthedouble-blind."
Allocationconcealment(selectionbias)
Lowrisk "Aninteractiveresponsetechnologysystemwasusedforrandomisationandallocationoftrialmedication.Therandomisationschemeusedarandomisedblockdesign(blocksize4)andwasgeneratedusingvalidatedrandomisationsoftwarebyBoehringerIngelheim,whichalsopreparedandcodedthemedicationstomaintainthedouble-blind."
Blindingofparticipantsandpersonnel(performancebias)
Lowrisk "Asbothtreatments(tiotropiumandtiotropium-olodaterol)weredeliveredviaidenticalRespimatdevices,treatmentwasmaskedtopatients,investigators,andeveryoneinvolvedinanalysingthetrialdata.Nodummydeviceswererequired."
Blindingofoutcomeassessment(detectionbias)
Lowrisk "Asbothtreatments(tiotropiumandtiotropium-olodaterol)weredeliveredviaidenticalRespimatdevices,treatmentwasmaskedtopatients,investigators,andeveryoneinvolvedinanalysingthetrialdata.Nodummydeviceswererequired."
Incompleteoutcomedata(attritionbias)
Highrisk
“Theprimaryanalysiswasdoneonthetreatedset(allrandomlyassignedpatientswhoreceivedanydoseofstudymedicationandwerenotfromasiteexcludedduetoon-siteprotocolviolations.”“Theprimaryendpointwasanalyzedduringthe“actualtreatmentperiod”,definedasthetimefromfirstdoseofmedicationuntil1dayafterthelastdoseofmedication.Thiswasalsotheperiodusedforexacerbationandotherendpoints.”“Alladverseeventswerecollectedduringthe
42
Domain Judgement SupportforJudgementon-treatmentperiod,definedasupto21daysafterthelastdose(appendixp1).”“Attheendoftheplannedstudyperiod(360plus21daysafterfirstdose),dataonvitalstatuswereavailablefrom99.6%ofpatients.”Analysisfortheprimaryandsecondaryendpointswasdoneoneventsthatoccurredduringthe“actualtreatmentperiod”.Patientswhopermanentlydiscontinuedstudytreatmentandhadanexacerbationoutsidethe“actualtreatmentperiod”arenotaccountedforintheanalysisandthisinformationisnotreported.Thiscouldaffecttheanalysisofexacerbationsdataconsideringmorepatientsreceivingtiotropiumalonewithdrewfromthestudy,whichcouldleadtoattritionbiasexceptformortalitydata.
Selectivereporting(reportingbias)
Highrisk ThisstudyisregisteredwithClinicalTrials.gov,numberNCT02296138andthestudyprotocolisprovided.Thestudypublicationdoesnotreportalloutcomesspecifiedintheprotocol(e.g.numberofpatientswith≥1moderatetosevereexacerbationduringthe“actualtreatmentperiod”;numberofpatientswith≥1exacerbationleadingtohospitalizationduringthe“actualtreatmentperiod”)Otheroutcomesthatwerenotpre-specifiedarereported.
Otherbias Highrisk “RC,MS,DL,DY,andDMcBareemployeesofthetrialsponsor(Novartis)andcontributedtothedesign,preparation,andconductofthestudy.Theyalsomadesubstantialcontributionstotheanalysisandinterpretationofthestudy.MDwastheprincipalinvestigatorofthestudywhocriticallyreviewedthefullstudyreportandhadthefinalresponsibilityforthedecisiontosubmitforpublication.Allauthorshadfullaccesstotheresultsofpre-specifiedstatisticalanalyses,wereencouragedtosuggestappropriatepost-hocanalyses,andmadesubstantialcontributionstothecontentofeachdraft.”
OutcomesreportedResultsarepresentedinTable6accordingtotheoutcomehierarchydescribedabove.
43
Table18:HierarchyofoutcomesinDYNAGITO2018 OLO/TIO5/5mcg
(n=3939)TIO5mcg(n=3941)
Totalmortality 110(2.8%) 123(3.1%)TotalSAEsTotalhospitalizationsHospitalizationduetosevereCOPDexacerbationWorseningofCOPDasaSAE
810(21%)
NR
NR
NR
862(22%)
NR
NR
NRNumberofpatientswith≥1moderateorsevereCOPDexacerbationNumberofpatientswith≥1severeCOPDexacerbation
NR
NR
NR
NR
SGRQtotalscore NR NRRateofmoderatetosevereCOPDexacerbationsAdjustedrateperpatient-yearRateratio(95%CI)Timeto1stmoderateorsevereexacerbation,HR(99%CI)RateofCOPDexacerbationsleadingtohospitalisationAdjustedrateperpatient-yearRateratio(95%CI)Timeto1stexacerbationleadingtohospitalisation,HR(95%CI)
0.90(0.84,0.96)0.93(0.85,1.02)p=NS
0.95(0.87,1.03)p=NS
0.18(0.16,0.20)0.89(0.76,1.03)p=NS
0.93(0.82,1.06)p=NS
0.97(0.90,1.03)
0.20(0.18,0.22)
TDItotalscore NR NRUseofrescuetreatment NR NRTroughFEV1 NR NR
1. Totalmortality
TherewasnodifferenceintotalmortalitywithOLO/TIOcombinationversusTIOalone.
2. SAEs
TherewasnodifferenceintotalSAEsbetweendualtherapywithOLO/TIOandTIO.Hospitalizationduetoanycausewasnotreported.
3.AcutemoderateorsevereCOPDexacerbations
44
a. Amoderateexacerbationwasdefinedasanexacerbationrequiringtreatmentwithoralcorticosteroidsorantibioticsorboth.Asevereexacerbationwasdefinedasanexacerbationthatrequiredhospitalizationoranemergencyroomvisit.
CRITICALAPPRAISALISSUE:Giventhatthismulticentertrialwasconductedin51differentcountriestherewillbevariabilityintreatmentpracticesofmoderateCOPDexacerbationsacrosscentersthatcouldbiasthestudyfindings.
b. Thenumberofpatientswithoneormoreacutemoderateorsevereexacerbationswasnotreported.Insteadthetrialreportstheannualrateofmoderateorsevereexacerbationsduringthe“actualtreatmentperiod”(pre-specifiedprimaryoutcome),whichwas0.90peryearwithdualOLO/TIOtherapyversus0.97peryearwithTIOalone.TherateratiowithdualOLO/TIOtherapyversusTIO,0.93(95%CI0.85,1.02)wasnotstatisticallysignificant.
CRITICALAPPRAISALISSUES:Thetrialreportstheannualrateofmoderateorsevereexacerbationsduringthe“actualtreatmentperiod”,whichwascalculatedbyaddedalltheexacerbationsthattookplaceinatreatmentarmdividedbythestudyduration.Therefore,multipleexacerbationsthatoccurredinasinglepatientarecounted.Interpretationofanannualrateisnotpossiblewithoutknowinghowtodividetheeffectamongindividualpeople.Iftherewasareductionintheproportionofpeoplewhohadoneormoreexacerbation,NNTcalculationscouldbemadeandthetreatmenteffect,ifany,couldbeeasilyinterpreted.
Thereportedratesarealsouncertainduetotheimbalanceinwithdrawalratesamongthetreatmentarms(12.4and16.5%inOLO/TIOandTIO,respectively)andnoattemptwasmadetoreduceattritionbias.Patientswhowithdrewprematurelywerenotadequatelyaccountedforinthecalculationofannualratesofmoderateorsevereexacerbations.
c. Time-to-first-eventanalysisreportednodifferenceintheriskofmoderateorsevere
exacerbationsduringtreatmentwithOLO/TIOversusTIOalone[HR0.95(99%CI0.87,1.03);p=NS].
CRITICALAPPRAISALISSUE:Time-to-first-eventanalysisisusefulonlywhenitisknownhowmanypatientshadmorethanoneexacerbationthroughoutthestudy.(32)
4.Health-relatedqualityoflife
SGRQtotalscorewasnotaprespecifiedoutcomeofthisstudy.
5.Symptomaticimprovement
TDIscorewasnotaprespecifiedoutcomeofthisstudy.
6.Useofrescuesalbutamol
Rescuetreatmentusewasnotaprespecifiedoutcomeofthisstudy.
45
7.COPDrelatedhealthcareutilization
COPDrelatedhealthcareutilization,whichincludesphysicianvisits/ERvisitsandhospitalization,wasnotaprespecifiedoutcome.
8.Adverseevents
a. Adverseeventsoccurredin1119(65%)receivingOLO/TIOand1065(62%)receivingTIO.Therewasnodifferencebetweentreatmentarmsfortotaladverseevents.
b. SignificantlymorepatientsreceivingTIO(16.5%)thanOLO/TIO(12.4%)withdrewduetoanadverseevent(ARI4.1%,NNH=24for1year).
9.FEV1
TroughFEV1wasnotaprespecifiedoutcomeofthisstudy.
B. SPARK2016isadoubleblindRCTin2224patientswithsevereandverysevereCOPDandahistoryofatleast1moderateexacerbationintheprecedingyear.(27)Thisstudycompareddualbronchodilatortherapywithindacaterol/glycopyrronium(IND/GLY)110/50mcg(n=741)withglycopyrronium(GLY)50mcg(n=741),bothadministeredoncedailyviatheBreezhalerdevice.Athirdtreatmentarmreceivedopen-labeltiotropum18mcg(n=742)viatheHandihalerdevicewasexcluded.Approximately75%ofpatients,withsimilarproportionsacrosstreatmentgroups,wereusingICSeitherasfixeddosecombinationorasmonotherapyatbaseline.PatientsusingICSatbaselinecontinuedthistreatmentatthesameorequivalentdoseandregimenduringthestudy.Salbutamolwaspermittedasrescuemedicationuse.Longactingbronchodilatorswerediscontinuedwithawashoutofupto7days(fortheophylline,indacaterolandtiotropium)beforescreening.StudycharacteristicsareprovidedinTable19.Samplesizewascalculatedintermsofthenumberofpatient-yearsneededtodetecta20%reductionintherateofCOPDexacerbations(ontheassumptionofaconstantrateduringthetreatmentperiod)intheIND/GLYgroupcomparedwiththeGLYgroup.Reassessmentofsamplesizeandstatisticalpowerwithinthefirst5monthsofthestudyindicatedthatthestudywasunderpoweredforitsprimaryendpoint.Toensurethatthestudyhadsufficientpower(≥80%)thesamplesizewasincreasedby200patientsandthestudydurationwasincreasedtoaminimumof15months,withanallowanceforadditionalvariableexposuretotreatmentuptoamaximumof18months.Assuminganaverageexposuretreatmentperpatientofapproximately17months,atotalof2198patientsrandomlyassignedtothethreetreatmentgroupswouldgiveapowerof84%.
Table19:SPARK2013studycharacteristics
Participants N=2224COPDpatients≥40yearsofageatriskofexacerbations,definedaspatientswithseveretoverysevereairflowlimitation(StageIIIorIVaccordingtoGOLD2008criteria),with:1)smokinghistoryof≥10pack-years;2)post-bronchodilatorFEV1<50%predictedvalue;3)post-bronchodilatorFEV1/FVC
46
<0.7;and4)historyof≥1moderateorsevereexacerbationinprecedingyearrequiringtreatmentwithsystemiccorticosteroidsorantibioticsorbothExclusioncriteria:1)anyhistoryofasthma;2)treatmentwithantibiotics,systemiccorticosteroids(oralorintravenous),orhospitalisationforCOPDexacerbationwithin6weeksbeforeprescreeningorduringscreening;3)developedaCOPDexacerbationduringprescreeningorscreening;or4)hadarespiratorytractinfectionwithin4weeksbeforeprescreening
Intervention Indacaterol/glycopyrronium110/50mcgOD[IND/GLY]administeredviaBreezhalerdevice(n=741)AvailableasUltibroBreezhaler(110mcg/50mcg)1inhalationOD
Comparators Glycopyrronium50mcgOD[GLY]administeredviaBreezhalerdevice(n=741)AvailableasSeebriBreezhaler(50mcg)1inhalationODTiotropium18mcgOD[TIO]administeredviaopen-labelHandihalerdevice(n=742)AvailableasSpirivaHandhaler(18mcg)1inhalationOD
ConcomitantMedications
Patientsreceivinginhaledcorticosteroidsatbaselinecontinuedtreatment(ortheinhaledcorticosteroidcomponentaloneiftakenasafixedcombinationwithabronchodilator)atthesameorequivalentdoseandregimenduringthestudy.Longactingbronchodilatorswerediscontinuedwithawashoutofupto7days(fortheophylline,indacaterol,andtiotropium)beforescreening.
Outcomes PRIMARY:• RateofmoderatetosevereCOPDexacerbationsatweek64forIND/GLY
vs.GLYSECONDARY(prespecified):
• RateofmoderatetosevereCOPDexacerbationsatweek76forIND/GLYvs.TIO
• TimetofirstmoderatetosevereCOPDexacerbationatweek64forIND/GLYvs.GLYvs.TIO
• RateofmoderatetosevereCOPDexacerbationsrequiringuseofbothsystemicglucocorticoidsandantibioticsatweek64
• Numberofdayswithmoderateorsevereexacerbationthatrequiredtreatmentwithsystemicglucocorticoidsandantibioticsatweek64
• Timetostudywithdrawalorprematurediscontinuationforanyreason• Proportionofpatientswithstudywithdrawalorpremature
discontinuationforanyreason• Cumulativeratesofmoderateorsevereexacerbationsformultiple
COPDexacerbationat26,52,64,and76weeks• TroughFEV1at4,12,26,38,52and64weeks• TroughFVCat4,12,26,38,52and64weeks• Changefrombaselineof%ofdayswithoutrescuetherapyuse• SGRQtotalscoreat12,26,38,52and64weeks
StudyDesign Multicentre3-armparallelgroupDBRCTconsistingofa2-weekrun-inperiod,64-weekdouble-blindtreatmentperiod,andoptiontoextenddouble-blind
47
treatmentperiodtoatotalof76weekstoensurestudyachievedexacerbationrateprespecifiedforanalysis
Therewerenosignificantdifferencesbetweentreatmentgroupsatbaselinewithregardtodemographics,COPDexacerbationsandSGRQtotalscore(Table20).Themeanageofstudypatientswas63.1years,75%weremales,and62%wereformersmokers.Post-bronchodilatorFEV1was37.2%ofpredictednormalvalueandameanSGRQtotalscoreof53(18)atscreening.Seventysevenpercentand22%hadahistoryof1COPDexacerbationand2ormoreexacerbations,respectively,inthepreviousyear.SeventyfivepercentwerereceivingICSatrandomization.Useofotherrespiratorymedicationsatbaselineisnotreported.
Table20:SPARK2013baselinecharacteristicsofstudyparticipants
IND/GLY110/50mcg(n=729)
GLY50mcg(n=740)
Ageinyears,mean(SD) 63.1(8.1) 63.1(8.0)Males 556(76%) 542(73%)Whiterace 594(81%) 605(82%)Formersmokers 452(62%) 457(62%)NumberofCOPDexacerbationsinpreviousyear01≥2
8(1%)
557(76%)164(22%)
13(2%)
572(77%)155(21%)
Post-bronchodilatorFEV1,mean%predicted(SD) 37.0(8.1) 37.3(8.1)SGRQtotalscoreMean(SD)
n=72753(18)
n=73352(18)
DailyrescuetreatmentuseMeannumberofpuffs(SD)
n=7165.7(4.6)
n=7375.7(5.0)
InhaledcorticosteroidsuseOtherrespiratorymedicationuse
546(75%)NR
557(75%)NR
Overall,1108patients(75%)completedthetrial.Atotalof171(23.1%)patientsintheIND/GLYgroupand203(27.4%)intheGLYgrouppermanentlydiscontinuedthestudy.ThehighernumberofwithdrawalsintheGLYgroupascomparedtoIND/GLYwasnotstatisticallysignificant[OR1.26(95%CI0.99,1.59)].DataforexacerbationsandFEV1werenolongercollectedoncepatientsdiscontinuedfromthestudy.Patientswhoprematurelydiscontinuedwerefollowedforsurvivaltotheendofthestudy.AsummaryofpatientdispositionisprovidedinTable21.
Table21:PatientdispositioninSPARK2013
IND/GLY110/50mcg GLY50mcgRandomized 741 741Totaladverseevents 678(93.0%) 694(93.8%)Totalwithdrawals 171(23.1%) 203(27.4%)
48
IND/GLY110/50mcg GLY50mcgWithdrewconsent 33(4.5%) 50(6.8%)Withdrawalduetoadverseevents 59(8.0%) 67(9.0%)Withdrawalduetolackofefficacy 18(2.4%) 32(4.3%)Losttofollow-up 5(0.7%) 6(0.8%)
RiskofbiasinSPARK2013SPARK2013isjudgedtohaveahighriskofbiasaccordingtotheCochraneRiskofBiasToolwithrespecttoattritionandsourceoffunding.Selectivereportingisjudgedtohaveanunclearriskofbias(Table22).
Table22:CochraneriskofbiassummaryforSPARK2013Domain Judgement SupportforJudgementRandomsequencegeneration(selectionbias)
Lowrisk "Investigatorscontactedaninteractivevoiceresponsesystemorwebsystem,whichgeneratedarandomisationnumber(notcommunicatedtothecaller)thatlinkedthepatienttoatreatmentgroup.Patientsrandomlyallocatedtoopen-labeltiotropiumwerenotassignedamedicationnumberbecausethistreatmentwassuppliedlocally.Randomisationwasstratifiedbycurrentorex-smokerstatusandinhaledcorticosteroiduse."
Allocationconcealment(selectionbias)
Lowrisk "Investigatorscontactedaninteractivevoiceresponsesystemorwebsystem,whichgeneratedarandomisationnumber(notcommunicatedtothecaller)thatlinkedthepatienttoatreatmentgroup.Patientsrandomlyallocatedtoopen-labeltiotropiumwerenotassignedamedicationnumberbecausethistreatmentwassuppliedlocally.”
Blindingofparticipantsandpersonnel(performancebias)
Lowrisk "Patients,investigatorstaff,peopleperformingassessments,anddataanalystsweremaskedtotreatment(QVA149orglycopyrronium)fromrandomisationuntildatabaselock.""Thedouble-blindstudydrugswereidenticalinpackaging,labelling,scheduleofadministration,appearance,taste,andodour.Unmaskingoccurredinthecaseofemergenciesandatconclusionofthestudy."
Blindingofoutcomeassessment(detectionbias)
Lowrisk "Patients,investigatorstaff,peopleperformingassessments,anddataanalystsweremaskedtotreatment(QVA149orglycopyrronium)fromrandomisationuntildatabaselock."
Incompleteoutcomedata(attritionbias)
Highrisk
"Patientswerefollowedupforseriousadverseeventsupto30daysaftertheirlaststudy
49
Domain Judgement SupportforJudgementdose.""Patientswhoprematurelydiscontinuedwerefollowedforsurvivaltotheendofthestudy."“Efficacyvariableswereanalysedforallpatientsrandomlyassignedtotreatmentgroupswhoreceivedatleastonedoseofstudydrug,analysedaccordingtothetreatmenttheywererandomlyassignedtoreceive(thefullanalysisset)…”“Exacerbationratesduringtreatmentwereanalysedwiththenegativebinomialmodel.Becausethismodelincludesthelengthoftimethepatientwasinthestudyasanoffsetvariable,whichautomaticallyaccountsforpatientswhodiscontinuedprematurely,theprimaryanalysiswasdonewithoutimputation.”Thestudyreportdoesnotstateifdataforexacerbationswerecollectedinpatientswhodiscontinuedfromthestudyandsubsequentlyhadanexacerbation.Thehighwithdrawalrates,withmorepatientsreceivingglycopyrroniumalonewithdrawingfromthestudy,willleadtoattritionbiasexceptformortalitydata.
Selectivereporting(reportingbias)
Unclearrisk ThisstudyisregisteredwithClinicalTrials.gov,numberNCT01120691andthestudyprotocolisprovided.Thestudypublicationdoesnotreportalloutcomesspecifiedintheprotocol(e.g.timetofirstmoderatetosevereCOPDexacerbation;numberofdayswithmoderateorsevereexacerbationthatrequiredtreatmentwithsystemicglucocorticoidsandantibioticsatweek64;changefrombaselineof%ofdayswithoutrescuetherapyuse)
Otherbias Highrisk "AFT,PD,CA,HC,andDB,asemployeesofthesponsor(NovartisPharmaAG),contributedtothedesignandpreparation,conduct,analysis,andinterpretationofthestudy.JAWandDENcontributedtothedesignofthestudy.DBwastheresponsiblemedicalofficerforthesponsor.JAWwastheprincipalinvestigatorofthestudy,readandcommentedonthefullstudyreport,andhadfinalresponsibilityforthedecisiontosubmitforpublication.Allauthorshadfullaccesstorawdata,contributedtothewritingofeachdraftofthereport,andwere
50
Domain Judgement SupportforJudgementresponsibleforthedecisiontosubmitforpublication.Norestrictionswereplacedonauthorsregardingthestatementsmadeinthereport."“Symptomsconstitutinganexacerbationwereidentified…basedonthepresenceoftwomajorsymptoms(dyspnoea,sputumvolume,sputumpurulence)foratleast2consecutivedaysoraworseningofonemajorsymptomtogetherwithanincreaseinanyoneminorsymptom(sorethroat,cold,feverwithoutothercause,cough,wheeze)foratleast2consecutivedays.”DefinitionofexacerbationisnotspecifictoCOPD.
OutcomesreportedResultsarepresentedinTable23accordingtotheoutcomehierarchydescribedabove.
Table23:HierarchyofoutcomesinSPARK2013 IND/GLY110/50mcg
(n=729)GLY50mcg(n=740)
Totalmortality 23(3%) 22(3%)TotalSAEsTotalhospitalizationsHospitalizationduetosevereCOPDexacerbationWorseningofCOPDasaSAE
167(23%)NRNR107(15%)
179(24%)NRNR116(16%)
Numberofpatientswith≥1moderateorsevereCOPDexacerbationOddsratio(95%CI)Numberofpatientswith≥1severeCOPDexacerbation
419(57.5%)0.96(0.78,1.18)95(13.0%)
426(57.7%)108(14.6%)
SGRQtotalscoreatWeek64SGRQtotalscore,mean(SE)Changefrombaseline,meanLSmeandifferencePatientswith≥4pointimprovement(MCID)Oddsratio(95%CI)
43.4(0.8)-9.2-2.1p=0.006756.7%1.28(0.99,1.66)p=NS
45.5(0.8)-7.151.5%
RateofmoderatetosevereCOPDexacerbations
51
IND/GLY110/50mcg(n=729)
GLY50mcg(n=740)
Adjustedrateperpatient-year(95%CI)Rateratio(95%CI)Timeto1stmoderateorsevereexacerbation,HR(95%CI)RateofCOPDexacerbationsleadingtohospitalisationAdjustedrateperpatient-yearRateratio(95%CI)Timeto1stexacerbationleadingtohospitalisation,HR(95%CI)
0.84(0.75,0.94)0.88(0.77,0.99)p=0.038NR0.09(0.07,0.13)0.81(0.60,1.10)p=NS0.79(0.60,1.05)p=NS
0.95(0.85,1.06)NR0.12(0.09,0.16)
TDItotalscore NR NRRescuesalbutamolovertreatmentperiodChangefrombaselineindailynumberofpuffs,mean(SE)LSmeandifference
-2.3(0.13)-0.81p<0.0001
-1.5(0.13)
TroughFEV1atweek64Changefrombaseline(mL)Meandifference
15170;p<0.0001
81
1. Totalmortality
Therewasnobetween-groupdifferenceinall-causemortality.
2. SAEs
TherewerenodifferencesintotalSAEsandCOPDasaSAEbetweentreatmentwithIND/GLYandGLYalone.All-causehospitalizationwasnotreported.
3.AcutemoderateorsevereCOPDexacerbations
a. Amoderateexacerbationwasdefinedasanexacerbationleadingtotreatmentwithantibioticsorsystemiccorticosteroidsorboth.Asevereexacerbationwasdefinedasanexacerbationthatrequiredhospitaladmissionoremergencytreatment.
CRITICALAPPRAISALISSUE:Giventhatthismulticentertrialwasconductedin27countriestherewillbevariabilityintreatmentpracticesofmoderateCOPDexacerbationsacrosscentersthatcouldbiasthestudyfindings.
b. IND/GLYandGLYgroupsshowednosignificantdifferenceinthenumberofpatientswithoneormoreacutemoderateorsevereexacerbation[OR0.96(95%CI0.78,1.18)].Thestudyalsoreportsthenumberofpatientswithoneormoresevereexacerbationandtherewasnodifferencebetweentreatmentgroups.AscomparedtoGLYalone,thereisa
52
significantreductionwithdualIND/GLYtherapyintherateofmoderateorsevereexacerbationstherapy[rateratio0.88(95%CI0.77,0.99);p=0.038]butnotintherateofsevereCOPDexacerbationsleadingtohospitalization.
CRITICALAPPRAISALISSUE:Thereportedexacerbationeventsandratesareuncertainduetothehighwithdrawalratesinthestudy(23and27%inIND/GLYandGLY,respectively).Patientswhowithdrewprematurelywerenotadequatelyaccountedforinthisstudy.
4.Health-relatedqualityoflife
MeanchangeinSGRQtotalscoreandproportionofpatientswithaMCIDwereevaluatedinthisstudy.TherewasnosignificantdifferencebetweentheIND/GLYandGLYalonefortheproportionofpatientswhoachievedaMCID.ChangefrombaselineinSGRQtotalscorewasgreaterintheIND/GLYgroupascomparedtothosewhoreceivedGLYalone[LSmeandifference-2.1;p<0.0067].
CRITICALAPPRAISALISSUE:EstimatesformeanchangeinSGRQtotalscoreandproportionofpatientswithaMCIDareuncertainduetothehighwithdrawalratesinthestudy(23and27%inIND/GLYandGLY,respectively).Patientswhowithdrewprematurelywerenotadequatelyaccountedforinthisstudy.
5.Symptomaticimprovement
TDIscorewasnotaprespecifiedoutcomeofthisstudy.
6.Useofrescuesalbutamol
ReducedneedforrescuemedicationisamarkerofimprovedcontrolofCOPDsymptoms.ChangeinthedailynumberofpuffsofrescuemedicationwasreportedandpatientsintheIND/GLYgrouprequiredlessdailypuffsofrescuesalbutamolascomparedtoGLYgroup[LSmeandifference-0.81;p<0.0001).
CRITICALAPPRAISALISSUE:Theclinicalrelevanceofareducedneedofrescuesalbutamolbylessthan1dailypuffisunclear,especiallyconsideringtheSGRQeffectestimateisuncertain(i.e.athighriskofattritionbias)andTDIscoreisnotreported.
7.COPDrelatedhealthcareutilization
Thisincludesphysicianvisits/ERvisitsandhospitalization.Itisanotheroutcomethatwasnotreportedinthestudy.
8.Adverseevents
a. Adverseeventsoccurredin678(93%)receivingIND/GLYand694(94%)receivingGLY.Therewasnodifferencebetweentreatmentgroupsfortotaladverseevents.
53
b. Atotalof59(8%)and67(9%)patientstreatedwithIND/GLYandGLY,respectively,withdrewduetoanadverseevent.Therewasnobetween-groupdifferenceforwithdrawalduetoadverseevents.
CRITICALAPPRAISALISSUE:Overall,1108patients(75%)completedthetrial.ThestudyreportdoesnotprovidesufficientinformationontheanalysisofharmdataalthoughitdoesstatethatinformationaboutadverseeventswascollectedatclinicvisitsandpatientswerefollowedupforSAEsupto30daysaftertheirlaststudydose.Whetherdatawerecollectedforadverseeventsfollowingdiscontinuationfromthestudyisnotreported.
9.FEV1
FortheIND/GLYandGLYgroupsthemeanchangefrombaselineintroughFEV1was151mland81ml,respectively.Astatisticallysignificantbutnotclinicallyrelevantbetween-groupdifferenceinfavourofIND/GLYwasseen[LSmeandifference70ml;p<0.0001].
CRITICALAPPRAISALISSUE:FEV1isasurrogateoutcomethathasvalidityinestimatingtheriskofdyingfromCOPDbutlittleuseinassessingtheimpactofinhaleddrugtherapyonCOPDsymptoms.(3)
LABA+ICSvs.LABASUMMIT2016isadoubleblindRCTin16,590patientswithsymptomaticCOPDandahistoryofcardiovasculardisease.(17)Thiswasanevent-drivestudyinwhichfollow-upcontinueduntilatleast1000deathshadoccurred(medianstudyexposurewas1.8years).Thisstudycomprisedof4treatmentarms.Thecomparisonofinterestforthisreportisfluticasonefuroate/vilanterol(FF/VI)110/50mcg(n=4145)versusvilanterol(VI)25mcg(n=4146),bothadministeredoncedailyviatheElliptadevice.(Thestudyalsorandomizedpatientstofluticasonefuroate(FF)100mcgODaswellasplacebo.TheresultsfortheFFtreatmentarmarenotincludedinthisreportsincethecomparisonwithICSisoutofscopeforthisliteraturereviewupdate.Theresultsfortheplaceboarmareincludedforinformationalpurposesonly.)Theuseofallinhaledcorticosteroidsandinhaledlongactingbronchodilatorswasdiscontinued≤48hoursbeforestudyentry,althoughotherCOPDmedicationssuchastheophyllineswereallowed.AdescriptionofthestudycharacteristicsisprovidedinTable24.Thestudywasdesignedtohave90%powertodetecta30%reductioninall-causemortality(hazardratio=0.70)oncombinationtherapycomparedwithplacebo.Tocontrolformultiplicityoftestingofcombinationtreatmentversusplaceboacrossendpoints,aclosedtestingprocedure(gatekeeper)approachwasplanned.ThehierarchywastheprimaryendpointfollowedbytherateofdeclineinFEV1followedbythecompositecardiovascularendpoint.Ifsignificanceatthe5%levelwasnotachievedfortheprimaryendpointforthecomparisonofcombinationtreatmentwithplacebo,thenthetestsforthesecondaryandotherefficacyendpointswouldbeinterpretedasdescriptiveonly.
54
Table24:SUMMIT2016studycharacteristics
Participants N=16,590COPDpatientsaged40-80yearswith:1)smokinghistoryof≥10pack-years;2)post-bronchodilatorFEV150-70%predictedvalue;3)post-bronchodilatorFEV1/FVC≤0.7;4)scoreof≥2onmMRCdyspnoeascale;and5)historyofcardiovasculardisease(CVD),definedascoronaryarterydisease,peripheralarterialdisease,stroke,MIordiabetesmellituswithtargetorgandisease,oratincreasedCVDrisk,definedasaged≥60yearsandreceivingmedicationsfor>2ofthefollowing:hypercholesterolaemia,hypertension,diabetesmellitus,orperipheralarterialdiseaseExclusioncriteria:1)respiratorydisordersotherthanCOPD;2)lungreductionsurgery;3)receivinglong-termoxygen,ororalcorticosteroidtherapy;4)severeheartfailure(NYHAClassIV)orejectionfraction<30%;5)lifeexpectancylessthan3years;or6)end-stagechronicrenaldisease
Intervention Fluticasonefuroate/vilanterol100/25mcgOD[FF/VI]administeredviadrypowderinhaler(Ellipta)device(n=4145)AvailableasBreoEllipta(100mcg/25mcg)1inhalationOD
Comparators Vilanterol25mcgOD[VI]administeredviaElliptadevice(n=4146)VilanterolmonotherapynotavailableinCanadaFluticasonefuroate100mcgOD[FF]administeredviaElliptadevice(n=4158)NotindicatedforCOPDpatients.AvailableasArnuityEllipta(at100mcgor200mcg)foruseinasthmapatients.Placebo[PBO](n=4141)
ConcomitantMedications
Theuseofallinhaledcorticosteroidsandinhaledlongactingbronchodilatorswasdiscontinued≤48hoursbeforestudyentry,althoughotherCOPDmedicationssuchastheophyllineswereallowed.Patientsunabletotoleratewithdrawaloftherapywereexcludedfromstudyentry.
Outcomes PRIMARY:• Timetoall-causemortality
SECONDARY(prespecified):• On-treatmentrateofdeclineinFEV1forFF/VIvs.PBO• On-treatmentcompositecardiovascularendpointofcardiovascular
death,myocardialinfarction,stroke,unstableangina,andtransientischaemicattackforFF/VIvs.PBO
OTHER(prespecified):• Allprimary,secondary,exploratoryandotherend-pointsforFF/VIvs.
FF,FF/VIvs.VI,FFvs.PBO,andVIvs.PBO• RateofmoderatetosevereCOPDexacerbationsforFF/VIvs.PBO• COPD-relatedmortalityforFF/VIvs.PBO• ArterialstiffnessinasubsetofsubjectsforFF/VIvs.PBO• Health-relatedqualityoflifemeasuredwiththeSGRQ-Cinasubsetof
subjectsforFF/VIvs.PBO• Quality-adjustedlifeyearsbytreatmentgroupusinghealthstatusdata
55
collectedfromEuroQolQuestionnaireinasubsetofsubjects• Healthcareresourceutilisation(measuredbynumberofdays
hospitalisedforCOPD)forFF/VIvs.PBOStudyDesign Multicentre4-armparallelgroupDBRCTconsistingofa4-10dayrun-inperiod,
median1.8yearsdouble-blindtreatmentphase(maximum4yearsfollow-up),and1-weekfollow-upphase.Thiswasanevent-drivenstudyinwhichfollow-upcontinueduntilatleast1000deathshadoccurred.
Therewerenosignificantdifferencesbetweentreatmentgroupsatbaselinewithregardtodemographics,COPDexacerbations,pre-studyCOPDtherapyandcardiovasculardiseaseorcardiovascularrisk(Table25).Themeanageofstudypatientswas65years,75%weremales,and54%wereformersmokers.Post-bronchodilatorFEV1was59.7%ofpredictednormalvalueatscreening.Thirtyeightpercenthadahistoryof1ormoreCOPDexacerbationsinthepreviousyear.Thereisnobetween-groupdifferenceinuseofcardiovasculartherapyatbaseline.Approximately33%werereceivingICSatrandomization.
Table25:SUMMIT2016baselinecharacteristicsofstudyparticipants(ITTpopulation)
FF/VI100/25mcg(n=4121)
VI25mcg(n=4118)
PBO(n=4111)
Age(years),mean(SD) 65(8) 65(8) 65(8)Males 3112(76%) 3053(74%) 3071(75%)Whiterace 3332(81%) 3339(81%) 3328(81%)Formersmokers 2253(55%) 2189(53%) 2175(53%)Pre-studyexacerbationsinpreviousyear01≥2
2528(61%)998(24%)595(14%)
2500(61%)988(24%)630(15%)
2447(60%)1044(25%)620(15%)
Pre-studyCOPDtherapyLABALAMAICS
1456(35%)638(15%)1394(34%)
1464(36%)634(15%)1374(33%)
1417(34%)659(16%)1349(33%)
PostbronchodilatorFEV1(%predictednormalvalue),mean(SD)
59.7(6.1) 59.7(6.1) 59.7(6.1)
CardiovasculardiseaseCoronaryarterydiseasePeripheralarterialdiseasePreviousstrokePreviousMIDiabeteswithtargetorgandisease
2113(51%)807(20%)386(9%)730(18%)397(10%)
2044(50%)817(20%)387(9%)722(18%)377(9%)
2103(51%)766(19%)404(10%)658(16%)374(9%)
AtcardiovascularriskHypercholesterolaemiaHypertensionDiabetesmellitus
2125(66%)2882(90%)886(28%)
2191(67%)2900(89%)874(27%)
2112(66%)2861(89%)850(27%)
56
FF/VI100/25mcg(n=4121)
VI25mcg(n=4118)
PBO(n=4111)
Peripheralarterialdisease 310(10%) 301(9%) 279(9%)BaselinecardiovasculartherapyAnyAntithromboticmedicationLipid-loweringmedicationRAASinhibitortherapyBetablockersCalciumchannelblockersNitratesDiuretics
4021(98%)2384(58%)2829(69%)2932(71%)1444(35%)1593(39%)556(13%)1550(38%)
3996(97%)2295(56%)2797(68%)2862(69%)1376(33%)1569(38%)569(14%)1549(38%)
3996(97%)2292(56%)2751(67%)2887(70%)1389(34%)1551(38%)613(15%)1508(37%)
Overall,6250patients(76%)receivingFF/VIandVIcompletedthetrial.DataforexacerbationsandFEV1werenolongercollectedoncepatientsdiscontinuedfromthestudy.Vitalstatuswasknownfor99.97%ofpatientsintheintentiontotreat(ITT)population.AsummaryofpatientdispositionisprovidedinTable26.
Table26:PatientdispositioninSUMMIT2016
FF/VI VI PBORandomized(ITTpopulation) 4121 4118 4111Totaladverseevents 2780(67%) 2809(68%) 2782(67%)Totalwithdrawals 950(23%) 1039(25%) 1192(29%)Withdrawalduetolackofefficacy
46(1.2%) 65(1.6%) 98(2.4%)
Withdrawalduetoadverseevents
329(8.0%) 366(8.9%) 387(9.4%)
Losttofollow-up 0 1 0
RiskofbiasinSUMMIT2016SUMMIT2016isjudgedtohaveahighriskofbiasaccordingtotheCochraneRiskofBiasToolwithrespecttoattritionbias,selectivereportingandsourceoffunding,andunclearriskofbiaswithrespecttoblinding(Table27).Thestudyreportdoesnotstateifdataforexacerbationsandotherefficacyoutcomeswerecollectedinpatientswhodiscontinuedfromthestudyandsubsequentlyhadanexacerbation.Thehighwithdrawalrates(23-29%)acrosstreatmentgroupswillleadtoattritionbiasforallefficacyoutcomesincludingexacerbationsbutnotformortality.Therearealsootherbiaseswithrespecttostudydesignandthepresenceofconfoundingthatmisrepresentthetreatmenteffect(seeResultsandCriticalAppraisalfollowingTable28).
Table27:CochraneriskofbiassummaryforSUMMIT2016Domain Judgement SupportforJudgementRandomsequencegeneration(selectionbias)
Lowrisk "Patientswererandomlyassigned(1:1:1:1)throughacentralisedrandomisationservice
57
Domain Judgement SupportforJudgementinpermutedblockstoreceiveeitherplacebo,fluticasonefuroate,vilanterol,orthecombinationoffluticasonefuroateandvilanterol.TherandomisationschedulewasgeneratedusingtheGSKvalidatedrandomisationsoftwareRANDALL.Aseparaterandomisationschedulewasproducedforeachcountry."
Allocationconcealment(selectionbias)
Lowrisk "Patientswererandomlyassigned(1:1:1:1)throughacentralisedrandomisationserviceinpermutedblockstoreceiveeitherplacebo,fluticasonefuroate,vilanterol,orthecombinationoffluticasonefuroateandvilanterol.TherandomisationschedulewasgeneratedusingtheGSKvalidatedrandomisationsoftwareRANDALL.Aseparaterandomisationschedulewasproducedforeachcountry."“…withonlythedatabaseadministratorshavingknowledgeoftreatmentassignment."
Blindingofparticipantsandpersonnel(performancebias)
Unclearrisk "Treatmentwasdoubleblind(maskingwasachievedwithElliptainhalersofidenticalappearance)withonlythedatabaseadministratorshavingknowledgeoftreatmentassignment."Dysphoniaisa common local side effect of ICS. WithdrawalofICSmayleadtounblindinginpatientswhowerepreviouslyonICSthenrandomizedtonon-ICStreatmentinthestudy.Successofblindingwasnotreported.
Blindingofoutcomeassessment(detectionbias)
Lowrisk "Treatmentwasdoubleblind(maskingwasachievedwithElliptainhalersofidenticalappearance)withonlythedatabaseadministratorshavingknowledgeoftreatmentassignment."
Incompleteoutcomedata(attritionbias)
Highrisk
“16590underwentrandomisation…Ofthese,22participantsnevertookstudymedicationandthesafetypopulationthereforeconsistsof16568patients[4131intheplacebogroup,4157inthefluticasonefuroategroup,4140inthevilanterolgroup,and4140inthecombinationgroup].Datafromfivecentres(83patients)wereexcludedfromtheefficacyanalysisbecauseoffailuretomeetthestandardsofGoodClinicalPracticeandethical
58
Domain Judgement SupportforJudgementpractice,andwereclosedbeforethestudyended.Thus,atotalof16485patientswereincludedintheintention-to-treatefficacy(ITT)population;4111intheplacebogroup,4135inthefluticasonefuroategroup,4118inthevilanterolgroup,and4121inthecombinationgroup(table1).”"Afterrandomisation,patientswereseenevery3monthstoconfirmvitalstatusandrecordadverseevents.Postbronchodilatorspirometrywasdoneevery3monthsandhealthstatuswasassessedat3monthsthenevery6months.Anindependentdatamonitoringcommitteeundertooksafetyreviewsevery6months,andonepredefinedinterimefficacyanalysiswasdoneafterabout500deathshadoccurred.”“Toensurenobiasintheascertainmentofsurvivalstatus,a“commonenddate”wasdeterminedseveralmonthsinadvance.Thiscommonenddatewasselectedsothattherewouldbeatleast1000deathsbythisdate.ThecommonenddatewassetatJan25,2015,andsiteswererequiredtoascertainthesurvivalstatusoftheirpatientsonorafterthisdate.”“Qualityoflifequestionnaireswerecollectedinasubsetofpatients(4443[27%]).”“Morepatientswithdrewfromstudymedicationintheplacebogroup(29%)thaninthethreeothergroups:thelowestwithdrawalrateswereseenwithcombinationtherapy(23%).”“Vitalstatuswasknownfor16480(99.97%)of16485patientsintheITTpopulation.”“PatientswithworseningCOPDstatusorprogressiveCVDwhileonstudytreatmentcanreceiveothermedications,orbewithdrawnifintheinvestigator’sopinionthepatient’sdeteriorationpreventsongoingparticipation.Thereasonforwithdrawalwillberecordedandpatientswillbefollowedupuntilstudytermination.”Thestudyreportdoesnotstateifdataforexacerbationswerecollectedinpatientswhodiscontinuedfromthestudyandsubsequentlyhadanexacerbation.
59
Domain Judgement SupportforJudgementThehighwithdrawalrates(23-29%)acrosstreatmentgroups,withmorepatientsintheplacebogroupwithdrawingfromthestudy,willleadtoattritionbiasexceptformortalitydata.Alsotimingofwithdrawalindifferenttreatmentgroupsandhowmanydaystheywerefollowedfollowingwithdrawalineachtreatmentgroup,torecordsubsequentoutcomesneedtobereported.
Selectivereporting(reportingbias)
Highrisk ThisstudyisregisteredwithClinicalTrials.gov,numberNCT013130676andthestudyprotocolwaspublishedpreviously.Thestudypublicationdoesnotreportalloutcomesspecifiedintheprotocol(e.g.COPD-relatedmortality;health-relatedqualityoflifemeasuredwiththeSGRQ-Cinasubsetofsubjects;quality-adjustedlifeyearsbytreatmentgroupusinghealthstatusdatacollectedfromEuroQolQuestionnaireinasubsetofsubjects;healthcareresourceutilisation[measuredbynumberofdayshospitalisedforCOPD]).Qualityoflifeoutcomewascollectedonlyin27%oftotalrandomisedpatients.
Otherbias Highrisk "Thestudywasdesignedbythefunder(GlaxoSmithKline)incollaborationwiththeacademicmembersofthesteeringcommittee.Thesponsorwasresponsiblefortherunningofthetrial,datacollection,andstatisticalanalysis.Statisticalanalysesweredonebyacontractresearchorganisation(VeramedLtd,Twickenham,UK;fundedbyGSK)onbehalfof,andwithoversightfrom,employeesofthefunder.Thefirstdraftofthereportwaswrittenbytheprimaryacademicauthor,andalltheauthorsworkedcollaborativelytopreparethefinalcontent.Allauthorsmadethedecisiontosubmitthemanuscriptforpublication.Alltheauthorshadfullaccesstothedataandvouchfortheaccuracyandcompletenessofalldataandanalyses,andforthefidelityofthestudytotheprotocol.Thecorrespondingauthorhadaccesstoallthedataandhadfinalresponsibilityforthedecisiontosubmitforpublication."Exacerbationdefinitiondidnotincludespecific
60
Domain Judgement SupportforJudgementsymptoms(nopre-definedclinicalcriteria);temporalindependenceofexacerbationeventswasnotestablished;noblindedadjudicationofexacerbationevents.“Aboutathirdofpatientsstoppedinhaledcorticosteroidsbeforestudyentry,withasimilarproportionstoppinglong-actingβagonists.6464(39%)patientsreportedhavinghadaCOPDexacerbationintheyearbeforeentry.”AbruptwithdrawalofICSmayleadtoincreaseinexacerbationspostrandomization.Thenumberofpatientswith≥1exacerbationhasnotbeenreportedinthisstudy.
OutcomesreportedInthisstudypublicationstatisticalanalysisofmortalityandallefficacyendpointswasincomparisonwithplacebo.However,thestudyprotocoldoesspecifythatcomparisonsofFF/VIandVIwillbeperformedforallprimary,secondary,exploratoryandotherend-pointsbutthesedataarenotreported.Ifstudyendpointdatawereavailableinthepublication,comparisonsofFF/VIversusVIweredoneusingCochrane’sReviewManager5.3softwareandpresentedasoddsratios(italicized)inTable28.
ResultsarepresentedinTable28accordingtotheoutcomehierarchydescribedabove.
Table28:HierarchyofoutcomesinSUMMIT2016 FF/VI
(n=4121)VI
(n=4118)PBO
(n=4111)TotalmortalityPvaluevs.PBOOddsratio(95%CI)vs.VI[calculatedinRevMan]Timeto1steventanalysisPvaluevs.PBOPvaluevs.VI
275(6.7%)NS1.04(0.87,1.24)HR0.88(0.74,1.04)NSNR
265(6.4%)NSHR0.96(0.81,1.14)NS
246(6.0%)
TotalSAEsTotalhospitalizationsHospitalizationduetosevere
961(23%)NRNR
972(23%)NRNR
918(22%)NRNR
61
FF/VI(n=4121)
VI(n=4118)
PBO(n=4111)
COPDexacerbationSAEofpneumoniaAEofpneumoniaPvaluevs.PBOOddsratio(95%CI)vs.VI[calculatedinRevMan]
NR237(6%)NS1.48(1.21,1.82)
NR163(4%)NS
NR214(5%)
Numberofpatientswith≥1moderateorsevereCOPDexacerbationNumberofpatientswith≥1severeCOPDexacerbation
NRNR
NRNR
NRNR
SGRQtotalscore–basedonsubsetof4443(27%)patients
NR NR NR
AnnualrateofmoderatetosevereexacerbationPvaluevs.PBOPvaluevs.VIAnnualrateofsevereexacerbationPvaluevs.PBOPvaluevs.VI
0.25<0.0001NR0.050.0004NR
0.310.0170.060.013
0.350.07
TransitionDyspneaIndex NR NR NRUseofrescuesalbutamol NR NR NRCOPDrelatedhealthcareutilization
NR NR NR
On-treatmentrateofdeclineinFEV1(mL/year),mean(SE)Difference(95%CI)Pvaluevs.PBOPvaluevs.VI
38(2.4)8(1,15)0.019NR
47(2.4)-2(-8,5)NS
46(2.5)
1. Totalmortality
TherewasnodifferenceintotalmortalitybetweenFF/VIcombinationtherapyandVIalone.
2. SAEs
TherewasnodifferenceintotalSAEsbetweendualtherapywithFF/VIandVImonotherapy.All-causehospitalizationwasnotreported.
3.AcutemoderateorsevereCOPDexacerbations
62
a. Amoderateexacerbationwasdefinedasanexacerbationleadingtotreatmentwithantibioticsorsystemiccorticosteroids.Asevereexacerbationwasdefinedasanexacerbationthatrequiredhospitalization.
CRITICALAPPRAISALISSUE:Giventhatthismulticentertrialwasconductedin43differentcountriestherewillbevariabilityintreatmentpracticesofmoderateCOPDexacerbationsacrosscentersthatcouldbiasthestudyfindings.
b. Thenumberofpatientswithoneormoreacutemoderateorsevereexacerbationswasnotreported.
CRITICALAPPRAISALISSUES:Thetrialreportstheannualrateofmoderateorsevereexacerbations,whichwas0.25peryearwithdualFF/VItherapyversus0.31peryearwithVIalone.Theannualratesofsevereexacerbationwere0.05peryearand0.06peryearwithFF/VIandVI,respectively.Therateswerenotstatisticallycompared(i.e.rateratio)betweenFF/VIandVIgroups.Rateswerecalculatedbyaddingalltheexacerbationsthattookplaceinatreatmentarmdividedbythedurationofthestudy.Therefore,multipleexacerbationsthatoccurredinasinglepatientarecounted.
Interpretingadifferencebetweentreatmentgroupsinanannualrateisnotpossiblewithoutknowinghowtodividetheeffectamongindividualpeople.Ifthisratereductionwasareductionintheproportionofpeoplewhohadoneormoreexacerbation,NNTcalculationscouldbemade.
AlsothereportedratesareuncertainduetothewithdrawalratesintheFF/VI(23%)andVI(25%)groups.Itisunclearhowannualratesofmoderateorsevereexacerbationswerecalculatedandwhetherpatientswhowithdrewprematurelywereappropriatelyaccountedforinthiscalculation.
c. Thirtyfivepercent,34%and15%ofrandomizedpatientswerealreadyreceivingLABA,ICSandLAMAatscreeningandwererequiredtodiscontinuethesemedications≤48hoursbeforestudyentry.
CRITICALAPPRAISALISSUE:SuddenICSorLAMAwithdrawalorbothatrandomizationinthosepatientsassignedtoVImonotherapymayexplainthenumericallyhigherrateofexacerbationsinthisgroupascomparedtodualtherapy(althoughthetwotreatmentgroupswerenotstatisticallycompared).Evidencefromdoubleblind,placebocontrolled,parallelgroupRCTsrangingfrom26to52weeksdurationinpatients(N=244-373)withmoderatetosevereCOPDandahistoryofexacerbationsreportedthatabruptwithdrawalofICSincreasedtheproportionofpatientswithoneormoresevereexacerbations(33-35).Ofthe244patientsinthe6-monthstudy,69(57%)intheplacebo(i.e.ICSdiscontinuation)groupand58(47%)intheICSgroupexperiencedatleastonemoderateexacerbation[HR1.5(95%CI1.1,2.1)],definedasworseningofrespiratorysymptomsthatrequiredtreatmentwithashort
63
courseoforalcorticosteroidsorantibiotics.(33)Ina1-yearpragmaticRCTin260primarycareCOPDpatientstherelativeriskofexperiencingamoderate(i.e.requiringoralcorticosteroidsorantibiotics)orsevereexacerbation(i.e.resultinginhospitalization)wasgreaterwithplaceboversuscontinuedICS[RR1.6(95%CI1.2,2.2);P<0.001].(34)Theeffectsof1-yearwithdrawalofICSaftera3-monthrun-inwithICS/LABAwerestudiedin373COPDpatients.(35)
4.Health-relatedqualityoflife
Health-relatedqualityoflifemeasuredwiththeSGRQ-Cwasnotreportedinthestudypublicationdespitebeinglistedasaprespecifiedstudyendpointintheprotocol.
5.Symptomaticimprovement
TDIscorewasnotaprespecifiedoutcomeofthisstudy.
6.Useofrescuesalbutamol
Useofrescuemedicationwasnotaprespecifiedoutcomeofthisstudy.
7.COPDrelatedhealthcareutilization
Healthcareresourceutilization,measuredbynumberofdayshospitalisedforCOPD,isanotheroutcomethatwasnotreportedinthestudypublicationdespitebeinglistedasaprespecifiedstudyendpointintheprotocol.However,thisstudyendpointdoesnotcapturephysicianvisitsandERvisits.
8.Adverseevents
a. Adverseeventsoccurredin2780(67%)receivingdualtherapywithFF/VIand2809(68%)receivingVIalone.TherewasnodifferencebetweenFF/VIandVIcomparatorgroupsfortotaladverseevents.SignificantlymorepatientswhoreceivedFF/VI(6%)versusVIalone(4%)hadanAEofpneumonia[OR1.48(95%CI1.21,1.82)].
b. Atotalof329(8%)and366(9%)patientstreatedwithFF/VIandVI,respectively,withdrewduetoanadverseevent.Therewasnobetween-groupdifferenceforwithdrawalduetoadverseevents.
CRITICALAPPRAISALISSUE:Overall,5589(68%)patientsreceivingFF/VIorVIcompletedthetrial.Informationonadverseeventsmaybeincompletegiventhatpatientswhopermanentlydiscontinuedstudytreatmentdidnotcomeinforfurtherevaluation.
9.FEV1
On-treatmentrateofdeclineinFEV1isreportedbutthedifferenceof10ml/yearbetweenFF/VIandVIgroupsisnotstatisticallycompared.
64
CRITICALAPPRAISALISSUE:FEV1isasurrogateoutcomethathasvalidityinestimatingtheriskofdyingfromCOPDbutlittleuseinassessingtheimpactofinhaleddrugtherapyonCOPDsymptoms.(3)
LABA+ICSvs.LAMAOnlySarac2016,asingle-centreopenRCTwasidentified.(29)Thissmallstudyrandomized44COPDpatientswithahistoryof≥1exacerbationintheprecedingyeartotwice-dailysalmeterol/fluticasone50/500mcg(n=22)oronce-dailytiotropium18mcg(n=22).Alllong-actingbronchodilatorsandinhaledsteroidswerestoppedduringthewashoutperiodandtheywereonlyallowedtotakeshort-actingbronchodilators(salbutamol-ipratropiumcombinationMDI).Duringthetreatmentperiodthepatientswereallowedtouseshort-actingbronchodilatorswhenneeded,butwerenotallowedtouseanyotherbronchodilatorsorinhaledsteroids.ThestudycharacteristicsareprovidedinTable29.
Withasamplesizeofonly44patientsthisstudywasnotpoweredtodetectadifferencebetweentreatmentgroupsinclinicallyrelevantoutcomessuchasreductioninexacerbation,improvementinqualityoflifeordecreaseinmortalityorhospitalizations.
Table29:Sarac2016studycharacteristics
Participants N=44moderateCOPDpatientsaged35-80yearswith:1)smokinghistoryof≥10pack-years;2)FEV150-80%predictedvalue;and3)historyof≥1exacerbationinprecedingyearExclusioncriteria:1)priordiagnosisofasthma;2)previousdocumentationofbronchialhyperreactivity;3)historyofallergyand/oratopy;or4)presenceofcongestiveheartfailureoranyothercardiopulmonarydiseasethatmightinterferewithpatients’follow-up
Intervention Salmeterol/fluticasone50/500mcgBID[SF]viaDiskusinhaler(n=22)Comparator Tiotropium18mcgOD[TIO]viaHandihalerdevice(n=22)ConcomitantMedications
Alllong-actingbronchodilatorsandinhaledsteroidswerestoppedduringthewashoutperiodandtheywereonlyallowedtotakeshort-actingbronchodilators(salbutamol-ipratropiumcombinationMDI).Duringthetreatmentperiodthepatientswereallowedtouseshort-actingbronchodilatorswhenneeded,butwerenotallowedtouseanyotherbronchodilatorsorinhaledsteroids.
Outcomes PRIMARY:• RateofCOPDexacerbations
OTHER:• RateofCOPDexacerbationsleadingtohospitalization• Post-bronchodilatorspirometryandlungvolumes• COPDAssessmentTest(CAT)score• Arterialbloodgasanalysis• BODEindex• Six-minutewalkdistance
65
StudyDesign Singlecentre2-armparallelgroupopenRCTconsistingofa2-weekwashoutperiodfollowed52-weekopen-labeltreatmentperiod
Themeanageofstudypatientswas66.6(10.2)yearsand91%weremales(Table30).Smokingstatusatscreeningisnotreported,CATscorewas9.3and39%had≥2exacerbationsinthepreviousyear.Post-bronchodilatorFEV1was65.4%ofpredictednormalvalueatscreening.Respiratorymedicationuseatscreeningisalsonotreported.
Table30:Sarac2016baselinecharacteristicsofstudyparticipants
SF50/500mcg(n=22)
TIO18mcg(n=22)
Ageinyears,mean(SD) 65.7(10.6) 67.4(9.7)Males 20(91%) 20(91%)Whiterace NR NRFormersmokers NR NRPatientswith≥2exacerbationsinpreviousyear 9(41%) 8(36%)No.ofexacerbationsinpreviousyear,mean(SD) 2.2(2.1) 1.9(1.4)Post-bronchodilatorFEV1,mean%predicted(SD) 63.5(9.5) 67.2(9.1)SGRQtotalscore,mean(SD) NR NRCATscore 8.7(5.5) 9.6(7.2)Respiratorymedicationuse NR NR
All44patientscompletedthetrialandnoadverseeventswerereported(Table31).
Table31:PatientdispositioninSarac2016
OLO/TIO5/5mcg TIO5mcgRandomized 22 22Totaladverseevents 0 0Totalwithdrawals 0 0
RiskofbiasinSarac2016AccordingtotheCochraneRiskofBiasTool,Sarac2016isjudgedtohaveahighriskofselectionbias,allocationbias,performancebiasanddetectionbias,andanunclearriskofselectivereportingandsourceoffunding(Table32).
Table32:CochraneriskofbiassummaryforSarac2016Domain Judgement SupportforJudgementRandomsequencegeneration(selectionbias)
Highrisk “Therandomizatonwasdoneaccordingtoalistpreparedpriortotheinitiationofthestudy.”
Allocationconcealment(selectionbias)
Highrisk “Therandomizatonwasdoneaccordingtoalistpreparedpriortotheinitiationofthestudy.”
Blindingofparticipantsand Highrisk Opentrial
66
Domain Judgement SupportforJudgementpersonnel(performancebias)Blindingofoutcomeassessment(detectionbias)
Highrisk Opentrial
Incompleteoutcomedata(attritionbias)
Lowrisk
“Allpatientscompletedthestudy.”
Selectivereporting(reportingbias)
Unclearrisk ThisstudyisnotregisteredwithClinicalTrials.gov.Astudyprotocolwasnotfoundsoitisunknownifthestudypublicationreportsalloutcomesspecifiedintheprotocol.Otheroutcomesthatwerenotpre-specifiedmaybereported.
Otherbias Unclearrisk “ThestudymedicationswerekindlyprovidedbyGlaxoSmithKlineandBoehringerIngelheimcompanies.Thesecompanieshadnootherinvolvementintheplanning,designandexecutionofthestudyandintheanalysisofthedata.”Sourceoffundingisnotreported.
OutcomesreportedResultsarepresentedinTable33accordingtotheoutcomehierarchydescribedabove.
Table33:HierarchyofoutcomesinSarac2016 SF50/500mcg
(n=22)TIO18mcg(n=22)
Totalmortality 0 0TotalSAEsTotalhospitalizationsHospitalizationduetosevereCOPDexacerbationWorseningofCOPDasaSAE
NR
NR
NR
NR
NR
NR
NR
NRNumberofpatientswith≥1moderateorsevereCOPDexacerbationNumberofpatientswith≥1severeCOPDexacerbation
NR
NR
NR
NR
SGRQtotalscore NR NRRateofmoderatetosevereCOPDexacerbationsAdjustedrateperpatient-yearRateratio(95%CI)
NR
NR
67
SF50/500mcg(n=22)
TIO18mcg(n=22)
Timeto1stmoderateorsevereexacerbation,HR(99%CI)RateofCOPDexacerbationsleadingtohospitalisationRateperpatient-yearRateratio(95%CI)Timeto1stexacerbationleadingtohospitalisation,HR(95%CI)
NR
0.6(1.0)p=NS
NR
NR
1.1(1.4)
NR
TDItotalscore NR NRUseofrescuetreatment NR NRChangefrombaselineinFEV1(ml) 34.8;p=NS 16.1;p=NS
Sarac2016doesnotreportanyoutcomeslistedintheoutcomehierarchyotherthanthatnodeathsoccurredineithertreatmentgroupandtherewasnodifferencebetweengroupsinchangefrombaselineinFEV1.SeveraloutcomemeasuresincludingrateofCOPDexacerbationsleadingtohospitalizationandCATscorewerenotreporteddespitebeingmentionedintheMethodssection.
LABA+LAMAvs.LABA+ICSIMPACT2018wascriticallyappraisedpreviously.PleaseseeTIreportonTrelegyElliptadatedSeptember12,2018.
LABA+LAMA+ICSvs.LABA+LAMAIMPACT2018wascriticallyappraisedpreviously.PleaseseeTIreportonTrelegyElliptadatedSeptember12,2018.
LABA+LAMA+ICSvs.LABA+ICSIMPACT2018wascriticallyappraisedpreviously.PleaseseeTIreportonTrelegyElliptadatedSeptember12,2018.
Summary
Atotalof6studiesevaluating7comparisonsofinterestmetthecriteriaforcriticalappraisal.TheTIpreviouslycriticallyappraised1study(IMPACT2018)evaluating3comparisonsofinterest(LABA+LAMAvs.LABA+ICS;LABA+LAMA+ICSvs.LABA+LAMA;LABA+LAMA+ICSvs.LABA+ICS),whichisavailableintheTIreportonTrelegyEllipta,datedSeptember12,2018.Criticalappraisalofthe5remainingstudiesevaluatingLABAvs.LAMA(1study:INVIGORATE2013),LABA+LAMAvs.LABA(1study:Donohue2016),LABA+LAMAvs.LAMA(2studies:DYNAGITO2018;SPARK2013),andLABA+ICSvs.LAMA(1study:Sarac2013)isprovidedinthisreport.Oneadditional
68
study(SUMMIT2016),whichisexcludedfromthePADliteraturereviewupdatebecauseitusesaLABAcomparator(vilanterol25mcg)thatisnotcommercializedinCanada,wascriticallyappraisedsinceitisthelargeststudytodatecomparingLABA+ICSwithLABA.
Allstudieswithadurationof1yearorlongerarejudgedtohaveahighriskofbiasaccordingtotheCochraneRiskofBiasToolwithrespecttoattrition.However,vitalstatuswasavailablefor>99%ofrandomizedpatientsinallstudies.Thereforetheoverallqualityofevidenceislowforalloutcomesexceptmortality.
Nostudyshowedadifferenceintotalmortalitybetweenanyofthecomparatorgroups.
TotalSAEsprovidesthebestsummarystatisticoftherapeuticimpactaccountingforallknownandunknownseriousimpact(benefitandharm)fromtherapy.NostudiesshowedadifferenceintotalSAEs(whichincludesallcausehospitalizationandhospitalizationduetosevereexacerbation)foranycomparison.Theeffectofinhaledmedicationsonmoderatetosevereexacerbationsneedstobereportedastheproportionofpatientswithoneormoreexacerbations.Only2studies(Donohue2016;SPARK2013)reportedthenumberofpatientswith1ormoremoderatetosevereexacerbationandbothstudiesshowednodifferencesbetweentheirrespectivetreatmentgroups.Otherstudiesreportedrateofmoderateorsevereexacerbation(DYNAGITO2018;SUMMIT2016)andtime-to-firsteventanalysisofmoderateorsevereexacerbation(INVIGORATE2013).DYNAGITO2018andSUMMIT2016claimednodifferenceinexacerbationratesbetweentreatmentarms.INVIGORATEclaimedthattimetofirstmoderateorsevereexacerbationwaslongerwithtiotropiumversusindacaterol[HR1.20(95%CI1.07to1.33;p=0.0012)].Time-to-first-eventanalysisisusefulonlywhenitisknownhowmanypatientshadmorethanoneexacerbationthroughoutthestudyinbothtreatmentgroups.Furthermore,thereportedeventsandratesareuncertainduetothehighwithdrawalratesinthestudiesandnoattemptwasmadetoreduceattritionbiasbyadequatelyaccountingforthepatientswhowithdrewprematurelyinthecalculationofeventandannualratesofmoderateorsevereexacerbations.
Twostudies(INVIGORATE2013;SPARK2013)reportedqualityoflife(SGRQ)and1study(INVIGORATE2013)reporteddyspneasymptoms(TDI).EstimatesforcomparativetreatmenteffectsonSGRQandTDIareuncertainduetothehighwithdrawalratesinbothstudiesandinadequateaccountingofpatientswhowithdrewprematurely.Furthermore,INVIGORATE2013reportedonasubset(approx.75%)oftotalrandomizedpatients.Therefore,theresultsarenotconsideredvalidduetomissingdata.
Therewerenodifferencesintotaladverseeventsbetweenanyofthecomparatorsinthesestudies.DYNAGITO2018istheonlystudythatdemonstratedadifferenceinwithdrawalduetoadverseeventsbetweencomparatorgroupswithsignificantlymorepatientsreceivingTIO(16.5%)thanOLO/TIO(12.4%)whowithdrewduetoanadverseevent(ARI4.1%,NNH=24for1year).
ReducedneedforrescuemedicationisamarkerofimprovedcontrolofCOPDsymptoms.Threestudies(INVIGORATE2013;Donohue2016;SPARK2013)reportedneedforrescuesalbutamol
69
duringthetreatmentperiod.UseofrescuetreatmentdidnotsignificantlydifferbetweenACL/FORandFORgroupsinDonohue2016.INVIGORATE2013reportedinasubset(91%)ofrandomizedpatientsthatpatientsintheINDgroupneededrescuetreatmentlessoftenascomparedtothosewhoreceivedTIO[LSmeandifference-0.62(95%CI-0.79,-0.45);p<0.0001)indailynumberofpuffs;LSmeandifference8.0%(95%CI5.9,10.2);p<0.0001inproportionofdayswithnorescueuse).ThisfindingisinconsistentwithindacaterolshowingnodifferenceversustiotropiumforSGRQtotalscoreandTDIscore.Alsoitisdifficulttounderstandhowindacaterolreducedtheneedforrescuemedicationwhentimetofirsteventanalysisofexacerbationrevealedthatindacaterolincreasedtheriskofmoderateorsevereexacerbationduringtreatmentversustiotropium.SPARK2013reportedareductionindailypuffsofrescuesalbutamolintheIND/GLYgroupascomparedtoGLYgroup[LSmeandifference-0.81;p<0.0001).TheclinicalrelevanceofareducedneedofrescuesalbutamolisunclearconsideringtheSGRQeffectestimateisuncertain(i.e.athighriskofattritionbias)andTDIscoreisnotreported.
COPDrelatedhealthcareutilization,whichincludesphysicianvisits/ERvisitsandhospitalizations,isanendpointthatwasnotreportedinanyofthestudies.Thesefindingswouldcorroboratethefindingsofdecreasedrateofacutemoderatetosevereexacerbation.
Fivestudies(Donohue2016;INVIGORATE2013;Sarac2016;SPARK2013;SUMMIT2016)reportedtroughFEV1,ofwhich4studies(Donohue2016;INVIGORATE2013;Sarac2016;SPARK2013)showedstatisticallysignificantbutnotclinicallyrelevantbetween-groupdifferences.SUMMIT2016didnotstatisticallycompareFF/VIandVIgroupsforon-treatmentrateofdeclineinFEV1.FEV1isasurrogateoutcomethathasvalidityinestimatingtheriskofdyingfromCOPDbutlittleuseinassessingtheimpactofinhaleddrugtherapyonCOPDsymptoms.
Conclusion
BasedonthenewlyidentifiedRCTsofatleast1yearduration,thereisinsufficientscientificallyvalidevidencethatanyofthesecomparisons(LABAvs.LAMA,LABA+LAMAvs.LABA,LABA+LAMAvs.LAMA,andLABA+ICSvs.LAMA)providesatherapeuticadvantageintermsofmoderateorsevereexacerbation,qualityoflife(SGRQ),reporteddyspneasymptoms(TDI),needforrescuemedication,andCOPDrelatedhealthcareutilization.
BasedonthenewlyidentifiedRCTsofatleast1yearduration,thereissufficientscientificallyvalidevidencedemonstratingthatnoneofthesecomparisons(LABAvs.LAMA,LABA+LAMAvs.LABA,LABA+LAMAvs.LAMA,andLABA+ICSvs.LAMA)provideadifferenceintermsofall-causemortality,totalseriousadverseevents(whichincludesallcausehospitalizationandhospitalizationduetosevereexacerbation),andtotaladverseeventsinthetreatmentofCOPD.
70
References
1. TherapeuticsInitiative.Indacaterolforchronicobstructivepulmonarydisease.TherapeuticsLetter.2016Sept-Oct;102:1-2.URL:http://ti.ubc.ca/letter102.
2. TherapeuticInitiative.Inhaledlongactingbeta2agonistsforCOPD.TherapeuticsLetter.2018Feb;109:1-2.URL:http://ti.ubc.ca/letter109.
3. GlobalInitiativeforChronicObstructiveLungDisease.Globalstrategyfordiagnosis,management,andpreventionofchronicobstructivepulmonarydisease.2018Edition.Availablewww.goldcopd.org.
4. ChongJ,KarnerC,PooleP.Tiotropiumversuslong-actingbeta-agonistsforstablechronicobstructivepulmonarydisease.CochraneDatabaseofSystematicReviews2012,Issue9.Art.No.:CD009157.
5. FarneHA,CatesCJ.Long-actingbeta2-agonistinadditiontotiotropiumversuseithertiotropiumorlong-actingbeta2-agonistaloneforchronicobstructivepulmonarydisease.CochraneDatabaseofSystematicReviews2015,Issue10.Art.No.:CD008989.
6. NanniniLJ,LassersonTJ,PooleP.Combinedcorticosteroidandlong-actingbeta2-agonistinoneinhalerversuslong-actingbeta2-agonistsforchronicobstructivepulmonarydisease.CochraneDatabaseofSystematicReviews2012,Issue9.Art.No.:CD006829.
7. WelshEJ,CatesCJ,PooleP.Combinationinhaledsteroidandlong-actingbeta2-agonistversustiotropiumforchronicobstructivelungdisease.CochraneDatabaseofSystematicReviews2013,Issue5.Art.No.:CD007891.
8. HoritaN,GotoA,ShibataY,OtaE,NakashimaK,NagaiK,KanekoT.Long-actingmuscarinicantagonist(LAMA)pluslong-actingbeta-agonist(LABA)versusLABAplusinhaledcorticosteroid(ICS)forstablechronicobstructivepulmonarydisease(COPD).CochraneDatabaseofSystematicReviews2017,Issue2.Art.No.:CD012066.
9. ChronicObstructivePulmonaryDisease:DevelopingDrugsforTreatmentGuidanceforIndustryDRAFTGUIDANCE.May13,2016Clinical/MedicalRevision1.URL:http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htmorhttp://www.regulations.gov
10. HigginsJPT,AltmanDG,SterneJAC(editors).Chapter8:Assessingriskofbiasinincludedstudies.In:HigginsJPT,GreenS(editors).CochraneHandbookforSystematicReviewsofInterventionsVersion5.1.0(updatedMarch2011).TheCochraneCollaboration,2011.Availablefromwww.handbook.cochrane.org.
11. BuhlR,MagderS,BothnerU,TetzlaffK,VobF,LoaizaL,VogelmeierCF,McGarveyL.Long-termgeneralandcardiovascularsafetyoftiotropium/olodaterolinpatientswithmoderatetoveryseverechronicobstructivepulmonarydisease.RespiratoryMedicine.2017;122:58-66.
12. CelliB,CraterG,KilbrideS,MehtaR,TabbererM,KalbergCJ,ChurchA.Once-dailyumeclidinium/vilanterol125/25mcgtherapyinCOPD:Arandomized,controlledstudy.Chest.2014;145(5):981-91.
13. DonohueJF,NiewoehnerD,BrooksJ,O’DellD,ChurchA.Safetyandtolerabilityofonce-dailyumeclidinium/vilanterol125/25mcgandumeclidinium125mcginpatientswithchronicobstructivepulmonarydisease:resultsfroma52-week,randomized,double-blind,placebo-controlledstudy.RespiratoryResearch.2014;15:78.
71
14. DransfieldMT,BourbeauJ,JonesPW,HananiaNA,MahlerDA,VestboJ,WachtelA,MartinezFJ,BarnhartF,SanfordL,LettisS,CrimC,CalverleyPMA.Once-dailyinhaledfluticasonefuroateandvilanterolversusvilanterolonlyforpreventionofexacerbationsofCOPD:tworeplicatedouble-blind,parallel-group,randomisedcontrolledtrials.LancetRespiratoryMedicine.2013;1:210-23.
15. FergusonGT,RabeKF,MartinezFJ,FabbriLM,WangC,IchinoseM,BourneE,BallalS,DarkenP,DeAngelisK,AurivilliusM,DorinskyP,ReisnerC.Tripletherapywithbudesonide/glycopyrrolate/formoterolfumaratewithco-suspensiondeliverytechnologyversusdualtherapiesinchronicobstructivepulmonarydisease(KRONOS):adouble-blind,parallel-group,multicentre,phase3randomisedcontrolledtrial.LancetRespiratoryMedicine.2018;6:747-58.
16. SharafkhanehA,SouthardJG,GoldmanM,UryniakT,MartinUJ.Effectofbudesonide/formoterolpMDIonCOPDexacerbations:Adouble-blind,randomizedstudy.RespiratoryMedicine.2012;106:257-68.
17. VestboJ,AndersonJA,BrookRD,CalverleyPMA,CelliBR,CrimC,MartinezF,YatesJ,NewbyDE.Fluticasonefuroateandvilanterolandsurvivalinchronicobstructivepulmonarydiseasewithheightenedcardiovascularrisk(SUMMIT):adouble-blindrandomisedcontrolledtrial.Lancet.2016;387:1817-26.
18. BatemanED,FergusonGT,BarnesN,GallagherN,GreenY,HenleyM,BanerjiD.DualbronchodilationwithQVA149versussinglebronchodilatortherapy:theSHINEstudy.EuropeanRespiratoryJournal.2013;42:1484-94.
19. D’UrzoAD,RennardSI,KerwinEM,MergelV,LeselbaumAR,CaractaCF.Efficacyandsafetyoffixed-dosecombinationsofaclidiniumbromide/formoterolfumarate:the24-week,randomized,placebo-controlledAUGMENTCOPDstudy.RespiratoryResearch.2014;15:123.
20. DecramerML,ChapmanKR,DahlR,FrithP,DevouassouxG,FritscherC,CameronR,ShoaibM,LawrenceD,YoungD,McBryan.Once-dailyindacaterolversustiotropiumforpatientswithseverechronicobstructivepulmonarydisease(INVIGORATE):arandomised,blindedparallel-groupstudy.2013;1:524-33.
21. SinghD,JonesPW,BatemanED,KornS,SerraC,MolinsE,CaractaC,GilEG,LeselbaumA.Efficacyandsafetyofaclidiniumbromide/formoterolfumaratefixed-dosecombinationscomparedwithindividualcomponentsandplaceboinpatientswithCOPD(ACLIFORM-COPD):amulticentre,randomisedstudy.2014;14:178.
22. DonohueJF,SoongW,WuX,ShresthaP,LeiA.Long-termsafetyofaclidiniumbromide/formoterolfumaratefixed-dosecombination:Resultsofarandomized1-yeartrialinpatientswithCOPD.2016;116:41-8.
23. CalverleyPMA,AnzuetoAR,CarterK,GronkeL,HallmannC,JenkinsC,WedzichaJ,RabeKF.Tiotropiumandolodaterolinthepreventionofchronicobstructivepulmonarydiseaseexacerbations(DYNAGITO):adouble-blind,randomised,parallel-group,active-controlledtrial.2018;6:337-44.
24. DecramerM,AnzuetoA,KerwinE,KaelinT,RichardN,CraterG,TabbererM,HarrisS,ChurchA.Efficacyandsafetyofumeclidiniumplusvilanterolversustiotropium,vilanterol,orumeclidiniummonotherapiesover24weeksinpatientswithchronicobstructivepulmonarydisease:resultsfromtwomulticentre,blinded,randomisedcontrolledtrials.2014;2:472-86.
72
25. DonohueJF,Maleki-YazdiMR,KilbrideS,MehtaR,KalbergC,ChurchA.Efficacyandsafetyofonce-dailyumeclidinium/vilanterol62.5/25mcginCOPD.RespiratoryMedicine.2013;107:1538-46.
26. Maleki-YazdiMR,KaelinT,RichardN,ZvarichM,ChurchA.Efficacyandsafetyofumeclidinium/vilanterol62.5/25mcgandtiotropium18mcginchronicobstructivepulmonarydisease:Resultsofa24-week,randomized,controlledstudy.RespiratoryMedicine.2014;108:1752-60.
27. WedzichaJA,DecramerM,FickerJH,NiewoehnerDE,SandstromT,TaylorAF,D’AndreaP,ArrasateC,ChenH,BanerjiD.AnalysisofchronicobstructivepulmonarydiseaseexacerbationswiththedualbronchodilatorQVA149comparedwithglycopyrroniumandtiotropium(SPARK):arandomised,double-blind,parallel-groupstudy.LancetRespiratoryMedicine.2013;1:199-209.
28. OharJA,CraterGD,EmmettA,FerroTJ,MorrisAN,RaphiouI,SriramPS,DransfieldMT.Fluticasonepropionate/salmeterol250/50mcgversussalmeterol50mcgafterchronicobstructivepulmonarydiseaseexacerbation.RespiratoryResearch.2014;15:105.
29. SaracP,SayinerA,Comparetheefficacyandsafetyoflong-actinganticholinergicandacombinationofinhaledsteroidsandlong-actingbeta-2agonistinmoderatechronicobstructivepulmonarydisease.TuberkToraks.2016;64(2):112-8.
30. LipsonDA,BarnhartF,BrealeyNetal.Once-dailysingle-inhalertripleversusdualtherapyinpatientswithCOPD.NewEnglandJournalofMedicine.2018;378(18):1671-80.
31. LipsonDA,BarnacleH,BirkR,BrealeyN,LocantoreN,LomasDA,Ludwig-SengpielA,MohindraR,TabbererM,ZhuCQ,PascoeSJ.FULFILTrial:Once-dailytripletherapyforpatientswithchronicobstructivepulmonarydisease.AmericanJournalofRespiratoryandCriticalCareMedicine.2017;196(4):438-46.
32. AaronSD,FergussonD,MarksGB,SuissaS,VandemheenKL,DoucetteS,MaltaisF,BourbeauJF,GoldsteinRS,BalterM,O’DonnellD,FitzGeraldM,fortheCanadianThoracicSociety/CanadianRespiratoryClinicalResearchConsortium.Counting,analysingandreportingexacerbationsofCOPDinrandomisedcontrolledtrials.Thorax.2008;63:122-8.
33. vanderValkP,MonninkhofE,vanderPalenJ,ZielhuisG,vanHerwaardenC.Effectofdiscontinuationofinhaledcorticosteroidsinpatientswithchronicobstructivepulmonarydisease:theCOPEstudy.AmericanJournalofRespiratoryandCriticalCareMedicine.2002;166:1358-63.
34. ChoudhuryAB,DawsonCM,KilvingtonHE,EldridgeS,JamesWY,WedzichaJA,FederGS,GriffithsCJ.WithdrawalofinhaledcorticosteroidsinpeoplewithCOPDinprimarycare:arandomisedcontrolledtrial.RespiratoryResearch.2007;8:93.
35. WoutersEF,PostmaDS,FokkensB,HopWC,PrinsJ,KuipersAF,PasmaHR,HensingCAJ,CreutzbergEC.Withdrawaloffluticasonepropionatefromcombinedsalmeterol/fluticasonetreatmentinpatientswithCOPDcausesimmediateandsustaineddiseasedeterioration:arandomisedcontrolledtrial.Thorax.2005;60:480-7.