www.revportcardiol.org
Revista
Portuguesa
de
Cardiologia
Portuguese
Journal
of
Cardiology
ORIGINAL
ARTICLE
Optimizing
risk
stratification
in
heart
failure
and
the
selection
of
candidates
for
heart
transplantation
Tiago
Pereira-da-Silva
a,∗,
Rui
M.
Soares
a,
Ana
Luísa
Papoila
b,c,
Iola
Pinto
d,e,
Joana
Feliciano
a,
Luís
Almeida-Morais
a,
Ana
Abreu
a,
Rui
Cruz
Ferreira
aaDepartmentofCardiology,HospitaldeSantaMarta,CentroHospitalardeLisboaCentral,Lisbon,Portugal bDepartmentofBiostatisticsandInformatics,NOVAMedicalSchool,UniversidadeNOVAdeLisboa,Lisbon,Portugal cResearchUnit,CentroHospitalardeLisboaCentral,Lisbon,Portugal
dDepartmentalAreaofMathematics,InstitutoSuperiordeEngenhariadeLisboa,UniversidadeNOVAdeLisboa,Lisbon,Portugal eCenterforMathematicsandApplications,FacultyofSciencesandTechnology,UniversidadeNOVAdeLisboa,Lisbon,Portugal
Received22January2017;accepted18June2017 Availableonline5March2018
KEYWORDS Cardiopulmonary exercisetesting; Heartfailure; Heart transplantation; Peakoxygen consumption; Riskstratification; Ventilatoryefficiency slope Abstract
IntroductionandAims:Selectingpatientsforheart transplantationischallenging. Weaimed
toidentifythemostimportantriskpredictorsinheartfailureandanapproachtooptimizethe
selectionofcandidatesforhearttransplantation.
Methods:Ambulatory patients followed in our center with symptomatic heart failure and
left ventricular ejection fraction ≤40% prospectively underwent a comprehensive baseline
assessment includingclinical,laboratory, electrocardiographic, echocardiographic,and
car-diopulmonary exercise testing parameters. All patients were followed for 60 months. The
combined endpointwas cardiacdeath, urgentheart transplantationorneedfor mechanical
circulatorysupport,upto36months.
Results:In the263enrolledpatients (75%male,age54±12years),54eventsoccurred.The
independent predictorsofadverseoutcomewereventilatoryefficiency(VE/VCO2)slope(HR
1.14,95%CI1.11-1.18),creatininelevel(HR2.23,95%CI1.14-4.36),andleftventricular
ejec-tionfraction(HR0.96,95%CI0.93-0.99).VE/VCO2slopewasthemostaccurateriskpredictor
atanyfollow-uptimeanalyzed(upto60months).Thethresholdof39.0yieldedhighspecificity
(97%),discriminatedaworseorbetterprognosisthanthatreportedforpost-heart
transplan-tation,andoutperformedpeakoxygenconsumptionthresholdsof10.0or12.0ml/kg/min.For
low-riskpatients(VE/VCO2slope<39.0),sodiumandcreatininelevelsandvariationsinend-tidal
carbondioxidepartialpressureonexerciseidentifiedthosewithexcellentprognosis.
Conclusions: VE/VCO2slopewasthemostaccurateparameterforriskstratificationinpatients
withheartfailureandreducedejectionfraction.ThosewithVE/VCO2slope≥39.0maybenefit
fromhearttransplantation.
© 2018SociedadePortuguesade Cardiologia.Publishedby ElsevierEspa˜na,S.L.U.Allrights
reserved.
∗Correspondingauthor.
E-mailaddress:[email protected](T.Pereira-da-Silva).
https://doi.org/10.1016/j.repc.2017.06.018
PALAVRAS-CHAVE Provadeesforc¸o cardiorrespiratória; Insuficiência cardíaca; Transplantac¸ão cardíaca; Consumodeoxigénio depico; Estratificac¸ãode risco; Declivedaeficiência ventilatória
Aprimoramentodaestratificac¸ãoderisconainsuficiênciacardíacaedaselec¸ão decandidatosatransplantac¸ãocardíaca
Resumo
Introduc¸ãoeobjetivos: Aselec¸ãodedoentesparatransplantac¸ãocardíacaédifícil.Procurámos
identificarospreditoresderiscomaisrelevantesnainsuficiênciacardíacaeumaabordagem
paraaprimoraraselec¸ãodecandidatosatransplantac¸ão.
Métodos: Doentes sintomáticos com insuficiência cardíaca e frac¸ão de ejec¸ão ventricular
esquerda≤40%,ambulatórios,seguidosnonossocentro,completaramprospetivamenteuma
avaliac¸ãobasalabrangente,inclusiveparâmetrosclínicos,laboratoriais,eletrocardiográficos,
ecocardiográficoseprovadeesforc¸ocardiorrespiratória;foramseguidospor60meses.Endpoint
combinado:mortedecausacardíaca,transplantac¸ãourgente ounecessidadedeassistência
mecânica,atéaos36meses.
Resultados: Nos263doentesincluídos(75%homens,54±12 anos)ocorreram54 eventos.O
declivedaeficiênciaventilatória(decliveVE/VCO2)(HR1,14,IC95%1,11-1,18),acreatinina
(HR2,23,IC95%1,14-4,36)eafrac¸ãodeejec¸ãoventricularesquerda(HR0,96,IC95%
0,93-0,99)forampreditoresindependentesdeeventos.OdecliveVE/VCO2foiomelhorpreditorem
qualquerperíodo analisado(atéaos60meses).Olimiar39,0apresentou elevada
especifici-dade(97%),discriminouumprognósticomelhoroupiordoqueoreportadonopós-transplante
cardíacoesuperouoslimiares10,0ou12,0mL/kg/mindeconsumodeoxigéniodepico.Em
doentesdebaixorisco(decliveVE/VCO2<39,0)osódio,acreatininaeavariac¸ãono
exercí-ciodapressãoparcialdedióxido decarbonoexpirado identificaramaqueles comexcelente
pronóstico.
Conclusões:OdecliveVE/VCO2foiomelhorpreditorderiscoemdoentescominsuficiência
cardíacaefrac¸ãodeejec¸ãoreduzida.DoentescomdecliveVE/VCO2≥39,0poderãobeneficiar
detransplantac¸ãocardíaca.
©2018SociedadePortuguesadeCardiologia.PublicadoporElsevierEspa˜na,S.L.U.Todosos
direitosreservados.
Listofabbreviations
PetCO2 variationofend-tidalcarbondioxidepartial pressure
CI confidenceinterval
CPET cardiopulmonaryexercisetesting HF heartfailure
HTX hearttransplantation
IDI integrateddiscriminationimprovement ISHLT International Society for Heart and Lung
Transplantation
LVEF leftventricularejectionfraction NRI netreclassificationimprovement VO2max peakoxygenconsumption VE/VCO2slope ventilatoryefficiencyslope
Introduction
Awide variety of predictors of adverse outcome in heart failure(HF)havebeen describedanditcanbedifficult to choose the most appropriate tools in clinical practice.1,2
Riskstratification should be asaccurate aspossible, par-ticularlywhenselectingpatients forheart transplantation (HTX), as procedure-related morbidity and mortality are
non-negligibleanditcannotbeofferedtoallcandidatesdue totheshortageof donors.3Forambulatory patients,both
theAmericanHeartAssociationandtheInternational Soci-etyforHeartandLungTransplantation(ISHLT)recommend theuseofpeakoxygenconsumption(VO2max)achievedin
cardiopulmonaryexercisetesting(CPET),withoptionaluse ofrisk scoresingrayzones;VO2max<10-12ml/kg/min is
consideredalistingcriterionforHTX.4,5Webelievethatrisk
stratificationandreferralcriteriaforHTXcanbeimproved using simple parameters. Additional CPET variables have showntobeaccurateforriskstratification,particularlythe ventilatory efficiency (VE/VCO2) slope.6---8 However, most
studiesthathighlightthevalueofergospirometric parame-tershavefocusedmainlyonclinicalandCPETdata,without comprehensive assessment of other parameters, and few had long-term follow-up.6---9 Identifyingrobust criteriafor
selectingpatientsforHTXshouldbebasedona comprehen-siveprospectiveclinicalandcomplementaryassessment.
Ouraimsweretoidentifythemostaccuratepredictors ofadverseeventsinnon-transplantedpatientswithHFand anapproachtooptimizetheselectionofpatientsforHTX.
Methods
Theinvestigationconformstotheprinciplesoutlinedinthe DeclarationofHelsinki.Theinstitutionalethicscommittee approvedthestudyprotocol.
Patientselectionandcomplementaryassessment Thissingle-centeranalysisincludedallpatientswithHFwith leftventricularejectionfraction(LVEF)≤40%,inNewYork HeartAssociation classII or III,followed intheheart fail-ure clinic of our institution between 2000 and 2009. All patients referred to the heart failure clinic underwent a comprehensive complementaryassessment. Clinical, labo-ratory,electrocardiographic,echocardiographic, andCPET data were prospectively collected; all these exams were performedwithinaperiodofonemonthineachpatient.
Patients aged <18 years and those with planned per-cutaneous coronary revascularization or cardiac surgery, exercise-limiting comorbidities (cerebrovascular disease, musculoskeletalimpairment,orsevereperipheralvascular disease),previousHTX,orfailuretoachievetheanaerobic thresholdwereexcluded.
Cardiopulmonaryexercisetesting
Maximal symptom-limited treadmill CPET was performed usingthemodifiedBruceprotocol(GEMarquetteSeries2000 treadmill).Gasanalysiswasprecededbycalibrationofthe equipment.Minuteventilation,oxygen uptakeand carbon dioxideproductionwereacquiredbreath-by-breath,usinga SensorMedicsVmax229gasanalyzer.Patientswere encour-agedtoexerciseuntiltherespiratoryexchangeratio(ratio between carbon dioxideproduction and oxygen consump-tion)was≥1.10. VO2maxwasdefinedasthehighest30-s
average achieved duringexercise and wasnormalized for body mass, corrected for fat-free mass in obesepatients (bodymassindex>30kg/m2).Theventilatorythresholdwas
determinedbycombiningthestandardmethods(V-slopeand ventilatory equivalents).10 The VE/VCO
2 slope was
calcu-latedbyleastsquareslinearregression,usingdataacquired throughouttheexercisesession.10 SeveralcompositeCPET
parameterswerealsocalculated. Follow-upandendpoint
Allpatientswerefollowedfor 60monthsfromthedateof completionoftheabove-mentionedcomplementaryexams. Patients were assessed for the occurrence of death, HTX ortheneedformechanicalcirculatorysupport.Datawere obtainedfromoutpatientclinicvisitsandreviewofmedical charts, and were complemented by a standardized tele-phoneinterviewwithallpatientsat12,36and60months offollow-up.
The combined endpoint of cardiac death, urgent HTX (occurringduringanunplannedhospitalizationfor worsen-ingofHF, requiring inotropes)or theneed formechanical circulatorysupportwasanalyzed.
Statisticalanalysis
Categoricaldataarepresentedasfrequencies(percentages) and continuous variables asmean (standard deviation) or median(25th-75thpercentile),asappropriate.Continuous variableswere analyzedusingthe Student’st test, or the Mann-Whitneytestwhennormalitywasnotverifiedbythe
Kolmogorov-Smirnovtest; categorical variables were ana-lyzedusingthechi-squaredorFisher’sexacttests.
UnivariateandmultivariateCoxregressionmodelswere appliedtotime untilthe combined endpoint, considering thefollow-uptimes of 12,36and 60months.A complete listof the tested variables and a detailed description of the univariate and multivariate analysis arepresented as supplementarymaterial(SupplementaryTables1and2).
The 36-month follow-up period was analyzed to iden-tifyapossibleapproachforselectingpatientsforHTX.7,8,11
Toidentifythemostaccurateindividualpredictorfromthe multivariatemodel,thevariablewiththehighestareaunder thecurve(AUC)onreceiveroperatingcharacteristic(ROC) analysis(VE/VCO2slope, posthoc)wasselected.Toassess
the additive value of other predictors to VE/VCO2 slope,
the AUC of VE/VCO2 slope was compared with the AUC
ofthemodel includingallpredictors ofadverse outcome. TheDeLongtest wasusedtocomparetwocorrelatedROC curves.Inaddition,continuousnetreclassification improve-ment(NRI)andintegrateddiscriminationimprovement(IDI) measuresforcensoreddatawerecalculated.12
Thebestcut-offvalue ofVE/VCO2 slopeforidentifying
high-riskpatientswascalculatedusingthemartingale resid-uals.SinceHTXcanonlybeofferedtoasmallsubgroupofHF patients,weidentifiedanothercut-offvaluethatminimized the rate of misclassified high-risk patients, even if sensi-tivitydecreased to reasonable levels.3 We thus identified
aVE/VCO2slope thresholdthatprovidedahighspecificity
(atleast90%)withatleast50%sensitivity,usingtheinverse probabilityofcensoringweightingapproach.13Subgroupsof
high-andlow-riskpatientswerecreatedaccordingly. Event-freesurvivalratesofbothsubgroupswereestimatedusing theKaplan-Meiermethodandcomparedusingthelog-rank test.ToassesswhethertheidentifiedVE/VCO2slope
thresh-oldissuitableasapotentialindicationforHTX,twoanalyses werecarried out. Firstly, survival rates of high- and low-risksubgroups ofour cohortwere comparedwithsurvival rates after HTX reportedby the ISHLT transplantregistry (quarterlydatareportonsurvivalratesfororthotopicHTX performedinEuropebetweenOctober1,2011and Septem-ber30,2015),usingapreviouslyreportedmethod.9,14Since
theISHLTreportsoverallsurvival,westudied(forthis anal-ysisonly)timetodeathfromanycause,rightcensoringin theeventofurgentor non-urgentHTX.15 Ifthe95%
confi-denceintervals(CI)of estimatedsurvival 36monthsafter HTXreportedbytheISHLTdidnotoverlapwiththoseof high-andlow-risksubgroupsofourcohort,basedontheVE/VCO2
slopethreshold,thiscanbetakenasevidenceofsignificant difference.7,15Secondly,theNRIandIDIwereusedto
com-parediscretized VE/VCO2 slope andVO2 max, considering
thecombinedendpointfortheanalysis.
To further stratify prognosis in low-risk patients, high-riskpatients wereexcluded from subsequentanalysisand univariateandmultivariateCoxregressionmodelswere fit-tedtothelow-risksubgroup.Martingaleresidualswereused toidentify cut-off valuesfor the variablesthat remained inthemodel andsubgroupsof high-andlow-risk patients werecreated accordingly.Event-freesurvival rates ofthe subgroupswereestimated usingthe Kaplan-Meiermethod andcomparedusingtheGehan-Breslow-Wilcoxontest.
The level of significance considered was ␣=0.05. Data wereanalyzed usingSPSS for Windows,version 20.0 (IBM
SPSSInc.,Chicago,IL)andthestatisticalprogramR Devel-opmentCoreTeam(2014),R:Alanguageandenvironment forstatisticalcomputing(RFoundationforStatistical Com-puting,Vienna,Austria).
Results
Atotalof 263patientswereincluded.The combined end-pointoccurredin54patients(20.5%)within36monthsand in69(26.2%)within60months.Themainbaselinedataare presented in Table 1, in which patients withand without eventsupto36monthsoffollow-uparecomparedandthe mostimportantriskpredictorsidentifiedinunivariateCox regressionarepresented.Completedataonbaseline char-acteristicsand univariateCox regression arepresented in
SupplementaryTables1and2,respectively.
Cumulativeadverseeventsoccurringindifferent follow-up periods are presented in Table 2. The independent predictorsofadverseevents identifiedinmultivariateCox regressionwithfollow-uptimesof12,36and60monthsare presentedinTable3.At36monthsof follow-up,VE/VCO2
slope,creatininelevelsandLVEFwereindependent predic-torsofadverseevents.TheVE/VCO2slopehadthehighest
Waldchi-squarevaluein univariateandmultivariate anal-ysisat36monthsandwastheparameterwiththehighest AUCinall multivariatemodels,considering 12,36 and60 monthsoffollow-up.Specificallyforthe36-monthfollow-up period,the AUCoftheoverall modelincludingall predic-torsofadverseoutcomedidnotdiffersignificantlyfromthe AUCofthe modelwithVE/VCO2slope alone (DeLongtest
p=0.103).Inaddition,theoverallNRIwas63.4%(95%CI 33.5-93.4%)andtheIDIwas0.019(95%CI[-0.01]-[0.046]),when creatininelevelsandLVEFwereaddedtoVE/VCO2slope;as
the95%CIofIDIincludesthenullvalue,theimprovement inmodelperformanceisnegligible.Sincetheother predic-torsincluded inthemodel(creatininelevelsandLVEF)did notaddsignificantly toVE/VCO2slope for risk prediction,
onlyVE/VCO2slope wasselectedtoobtain acut-offvalue
forclinicaluse.
Based onthe martingale residuals, the best estimated cut-offvalueforVE/VCO2slopetoidentifypatientsathigher
riskat36monthsoffollow-upwas32.0(specificity80%, sen-sitivity83%)(SupplementaryFigure1).However,athreshold of39.0providedhigherspecificity(97%),with52% sensitiv-ity.Theestimated36-monthsurvivalforhigh-risk(VE/VCO2
slope ≥39.0) andlow-risk (VE/VCO2 slope <39.0) patients
wassignificantlydifferent(Figure1).
Athresholdof39.0forVE/VCO2slopewasassessedasa
potentiallistingcriterionforHTX.Firstly,the95%CIsof esti-matedsurvival (considering all-causedeath) for high-and low-risksubgroupsdidnotoverlap withthoseof post-HTX reportedbytheISHLT.15Secondly,theVE/VCO
2slopehada
higherAUCthanVO2 max, consideringboth ascontinuous
variables(AUCofVO2 max0.79,95%CI0.72-0.87;DeLong
testforcomparisonp=0.009).Moreover,therewasa signifi-cantimprovementinthepercentageofcorrectclassification using the VE/VCO2 slope threshold of 39.0, in
compari-sontodiscretizedVO2 max(Table 4):considering theVO2
maxthresholdof 10.0 ml/kg/min,the overallNRI and IDI were82.2%(95%CI52.3-112.1%) and0.278(95%CI 0.182-0.373),respectively;consideringtheVO2maxthresholdof
12.0ml/kg/min,theoverallNRIandIDIwere93.3%(95%CI 63.4-123.2%)and0.226(95%CI0.141-0.311),respectively. Since the data consistently favored the use of VE/VCO2
slopeasanindicationforHTX,high-riskpatients(VE/VCO2
slope ≥39.0) were excluded from subsequent analysis. In theother 229 patients,27(11.8%) eventsoccurred during 36monthsoffollow-up.Sodiumandcreatininelevels,LVEF, andvariationachievedinCPET(anaerobicthresholdminus baseline)oftheend-tidalcarbondioxidepartialpressure( PetCO2)wereindependentpredictors ofadverseeventsat
36monthsoffollow-up(Table3).The AUCwassimilarfor these predictors. ForLVEF, the flatness of the martingale residualssmootherdidnotenableasuitablecut-offpointto beidentifiedthatdiscriminatedhigh-andlow-riskpatients (SupplementaryFigure2);sodium≤136.0mEq/l,creatinine ≥1.0mg/dlandPetCO2≤0.45kPa(3.4mmHg)were
asso-ciatedwithhigher risk ofadverse events (specificity 63%, 61% and 63%,sensitivity 77%, 74%and 74%, respectively). Eightsubgroupswerecreatedaccordingtothethree men-tioned cut-off values (Supplementary Figure 3).Prognosis wassimilar for patients withup toone variable(sodium, creatinine or PetCO2) with abnormal values (classified
accordingtothethresholdsidentified);prognosiswas simi-larforpatientswithtwovariableswithabnormalvalues,and event-freesurvivalwasworstwhenthethreevariableswere classifiedasabnormal. Threegroups werecreated accord-ingly(Figure2);prognosiswassignificantlydifferentforthe threecategories.
Discussion
Themostaccuratepredictorofadverseoutcomeinpatients withHFwithreducedLVEFwasVE/VCO2slopeandthebest
thresholdforidentifyingpatientswhomaybenefitfromHTX was 39.0. Sodiumlevels, creatinine levels and PetCO2
wereabletoidentifylow-riskpatients withexcellent out-come.
Riskstratification in HF andselecting patients for HTX arechallenging.Mancinietal.18showedinanancillarystudy
thatVO2maxisavaluableparameterforselectingpatients
forHTX,andsubsequentlyrefinedtheircut-offvalues.11,19
Currently,theAmericanHeartAssociationrecommends list-ing ambulatory patients for HTX when VO2 max is <10
ml/kg/min with achievement of anaerobic metabolism, and to defer when VO2 max is >14 ml/kg/min.4 Mancini
etal.19 proposedasimilardecisionalgorithm,also
recom-mendingassessmentusingtheHeartFailureSurvivalScore (HFSS). The ISHLT recommends listing when VO2 max is
<12 ml/kg/minonbeta-blockers, deferringwhenit is >14 ml/kg/min, andusingtheHFSSin thegrayzone.5 Only in
thepresenceof asubmaximalCPETshouldVE/VCO2slope
beconsideredasalistingcriterion,accordingtotheISHLT guidelines.5RiskstratificationandpatientselectionforHTX
couldprobablybeimproved.7However,theuseof
random-izedcontrolledtrialstoaddresstheissueoflistingcriteria is hindered by ethical and social considerations. In this context, data from robust registries on non-transplanted HF patients receiving contemporary pharmacological and devicetherapyareofgreatvalue.
Weevaluatedanextensiverangeofclinical,laboratory, electrocardiographic,echocardiographic,andCPET
param-Table1 BaselinedataandunivariateCoxregressionanalysisforpredictingadverseeventsupto36monthsoffollow-up.
Variable Allpatients Eventsa Noeventsa p Chi-squareHazardratio,
95%CI
p
Clinicalandelectrocardiographicdata
Age(years)b 54(12) 54(12) 53(12) 0.718 0.1 1.00,0.98-1.02 0.781 Malec 197(75) 44(81) 153(73) 0.141 2.3 0.61,0.33-1.15 0.125 Diabetesc 54(21) 8(15) 46(22) 0.248 0.1 1.12,0.60-2.08 0.728 Ischemicetiologyc 97(37) 28(52) 69(33) 0.009 4.3 2.06,1.13-3.75 0.018 NYHAII(vs.III)c 200(76) 28(52) 172(82) <0.001 37.3 0.22,0.12-0.42 <0.001 Sinusrhythmc 214(81) 36(67) 178(85) 0.003 11.7 2.52,1.28-4.96 0.008 Loopdiureticc 234(89) 54(100) 180(86) 0.002 4.2 24.9,1.1-546.0 0.041 ACEi/ARBc 255(97) 54(100) 201(96) 0.309 1.2 0.05,0.01-12.7 0.285 Betablockerc 233(89) 48(89) 185(89) 0.999 1.5 1.41,0.81-2.43 0.224
Aldosteronereceptorblockerc 176(67) 43(80) 133(64) 0.035 5.6 2.00,1.12-3.54 0.018
BaselineCRTc 72(27) 14(26) 58(28) 0.342 1.3 1.72,0.64-3.36 0.462 BaselineICDc 85(32) 18(33) 67(32) 0.117 1.4 1.41,0.79-2.64 0.236 Laboratorydata Creatinine(mg/dl)b 1.1(0.3) 1.2(0.3) 1.1(0.3) <0.001 15.1 3.93,1.66-9.26 0.001 Sodium(mEq/l)b 137(3) 134(4) 137(3) <0.001 62.0 0.73,0.66-0.81 <0.001 NT-proBNP(pg/ml)b 2304(2821) 4269(4724) 1898(2029) <0.001 38.3 1.01,1.00-1.01 <0.001 ElevatedtroponinTc 37(14) 13(24) 24(11) 0.004 14.3 3.54,1.54-8.12 0.003 Echocardiographicdata LVEDD(mm/m2)b 39(6) 42(7) 39(5) 0.006 8.8 0.16,0.04-0.64 0.010 LVEF(%)b 28(7) 23(6) 28(8) <0.001 26.7 0.92,0.90-0.95 <0.001 MRc 42(16) 9(17) 33(16) 0.988 1.5 0.92,0.30-1.55 0.872 RVSDc 42(16) 19(35) 23(11) <0.001 37.9 7.09,3.28-15.33 <0.001
Cardiopulmonaryexercisetestingdata
HRrecovery(1stminute)(min)d 18(13-29) 13(8-18) 19(15-30) <0.001 33.5 0.90,0.85-0.93 <0.001
doubleproduct(mmHg/min)b 11911(5374) 8564(4443)12619(5547) <0.001 26.2 1.00,1.00-1.00 <0.001
PeakRERb 1.10(0.10) 1.13(0.10) 1.09(0.09) 0.008 5.6 45.75,2.59-808.94 0.011
VO2max(ml/kg/min)e,b 20.0(5.3) 15.6(4.7) 21.1(5.7) <0.001 56.8 0.78,0.71-0.85 <0.001
VO2max(%predicted)e,b 67(15) 51(14) 71(16) <0.001 69.1 0.91,0.81-0.94 <0.001
Circulatorypower(mmHg/ml/kg/min)b 3115(1123) 2185(895) 3289(1132) <0.001 48.3 0.99,0.99-0.99 <0.001
Peakoxygenpulse(%predicted)b 91(30) 74(23) 95(30) <0.001 28.7 0.97,0.96-0.98 <0.001
OUESb 1.8(0.6) 1.4(0.4) 1.9(0.6) <0.001 49.9 0.11,0.05-0.27 <0.001
AT-VO2(ml/kg/min)b 16.1(4.0) 12.6(3.6) 17.0(4.1) <0.001 61.1 0.74,0.66-0.81 <0.001
VE/VCO2slopef,b 31(7) 40(9) 29(5) <0.001139 1.27,1.19-1.35 <0.001
PeakVE/VCO2f,b 35(8) 42(10) 33(7) <0.001 76.4 1.14,1.09-1.19 <0.001
ATVE/VCO2f,b 33(7) 40(9) 31(6) <0.001 78.3 1.17,1.08-1.23 <0.001
VE/VCO2slope/VO2max(ml/kg/min)-1 b 1.7(0.9) 2.9(1.4) 1.5(0.6) <0.001130 5.00,3.12-8.31 <0.001
BaselinePetCO2(kPa)b 4.4(0.6) 4.2(0.7) 4.4(0.6) 0.017 8.21 0.52,0.30-0.89 0.018
ATPetCO2(kPa)g,b 4.9(0.8) 4.3(0.8) 5.0(0.7) <0.001 49.4 0.27,0.17-0.44 <0.001
PetCO2(kPa)g,b 0.5(0.4) 0.1(0.4) 0.6(0.4) <0.001 84.3 0.03,0.01-0.17 <0.001 Score
HFSSb 8.7(1.0) 7.9(0.9) 8.8(0.9) <0.001 51.2 0.33,0.22-0.49 <0.001
a Cardiacdeath,urgenthearttransplantationormechanicalcirculatorysupportupto36monthsoffollow-up(n=54). b Valuesexpressedasmean(standarddeviation).
c Valuesexpressedasn(%).
d Valuesexpressedasmedian(25th-75thpercentile). e,f,gOfthese,onlyVO
2max,VE/VCO2slope(entireexercise)andPetCO2wereenteredinthemultivariatemodel,toavoid
multi-collinearity.
doubleproduct:productofpeakminusbaselineheartrateandsystolicbloodpressure;PetCO2:anaerobicthresholdminusbaseline
end-tidalcarbondioxidepartialpressure;ACEi:angiotensin-convertingenzymeinhibitor;ARB:angiotensinreceptorblocker;AT: anaer-obicthreshold;Chi-square:Waldchi-squarevalue;CI:confidenceinterval;Circulatorypower:productofpeakoxygenconsumptionand systolicbloodpressure;CRT:cardiacresynchronizationtherapy,withorwithoutdefibrillator;HFSS:HeartFailureSurvivalScore;HR:heart rate;HRrecovery(1stminute):peakheartrateminusheartrateatfirstminuteofrecovery;ICD:implantablecardioverter-defibrillator; LVEDD:leftventricularend-diastolicdiameter;LVEF:leftventricularejectionfraction;MR:moderateorseveremitralregurgitation; NT-proBNP:N-terminalpro-B-typenatriureticpeptide;NYHA:NewYorkHeartAssociation;OUES:oxygenuptakeefficiencyslope16;peak
oxygenpulse:peakoxygenconsumption/heartrateratio;PetCO2:end-tidalcarbondioxidepartialpressure;RER:respiratoryexchange
ratio(ratiobetweencarbondioxideproductionandoxygenconsumption);RVSD:rightventricularsystolicdysfunction;VO2max:peak
oxygenconsumption;VO2max(%predicted):basedonWassermanandHansen’sformula;VCO2:carbondioxideproduction;VE:minute
Table2 Adverseeventsat12,36and60monthsoffollow-up.
12months 36months 60months
n (%) n (%) n (%)
Combinedendpointa 27 10.3 54 20.5 69 26.2
Death 22 8.4 47 17.9 66 25.1
Cardiacdeath 18 6.8 36 13.7 49 18.6
Suddendeath 9 3.4 13 4.9 18 6.8
Worseningofheartfailure 9 3.4 23 8.7 31 11.8
Hearttransplantation 12 4.6 19 7.2 22 8.4
Urgenthearttransplantation 8 3.0 15 5.7 17 6.5
Mechanicalcirculatorysupportb 1 0.4 3 1.1 3 1.1
aCardiacdeath,urgenthearttransplantationormechanicalcirculatorysupport.
b AllpatientswereinInteragencyRegistryforMechanicallyAssistedCirculatorySupport(INTERMACS)profiles1or2.17
Table3 MultivariateCoxregression.
Chi-square Hazardratio,95%CI p AUC,95%CI
12-monthfollow-upa VE/VCO2slope 47.8 1.14,1.10-1.18 <0.001 0.91,0.87-0.95 LVEF 4.1 0.94,0.89-0.99 0.044 0.76,0.68-0.84 OverallAUC - - - 0.91,0.88-0.95 36-monthfollow-upb VE/VCO2slope 89.0 1.14,1.11-1.18 <0.001 0.87,0.81-0.93 Creatininelevels 5.4 2.23,1.14-4.36 0.020 0.69,0.65-0.74 LVEF 4.7 0.96,0.93-0.99 0.030 0.72,0.65-0.78 OverallAUC - - - 0.89,0.84-0.94 60-monthfollow-upc VE/VCO2slope 35.0 1.11,1.08-1.16 <0.001 0.87,0.81-0.92 Creatininelevels 7.8 2.25,1.27-3.96 0.005 0.68,0.61-0.75 PetCO2 5.1 0.39,0.17-0.89 0.024 0.84,0.79-0.90 LVEF 3.9 0.97,0.94-0.99 0.047 0.72,0.66-0.79 OverallAUC - - - 0.89,0.84-0.94
36-monthfollow-up(low-risk)d,e
Sodiumlevels 18.8 0.79,0.71-0.88 <0.001 0.73,0.61-0.85 Creatininelevels 5.7 2.75,1.20-6.29 0.017 0.71,0.62-0.81 LVEF 5.6 0.94,0.89-0.99 0.018 0.72,0.64-0.80 PetCO2 4.3 0.30,0.10-0.94 0.039 0.73,0.63-0.84 aC-index0.90,95%CI0.86-0.94. b C-index0.88,95%CI0.83-0.92. c C-index0.86,95%CI0.82-0.90. d C-index0.83,95%CI0.76-0.91. e ExcludingpatientswithVE/VCO
2slope≥39.0.
AUCforindividualvariablesandfortheoverallmodelarepresented,foreachfollow-uptime.
PetCO2: anaerobicthreshold minusbaseline end-tidalcarbon dioxide partialpressure;AUC: areaunderthereceiving operating
characteristiccurve;Chi-square:Waldchi-squarevalue;CI:confidenceinterval;LVEF:leftventricularejectionfraction;VCO2:carbon
dioxideproduction;VE:minuteventilation.
Table4 Proportionofpatientscorrectlyandincorrectlyclassifiedat36monthsoffollow-up.
High-risk,n(%) Low-risk,n(%)
Correct Incorrect Correct Incorrect
VO2max≤10.0ml/kg/min 7(87.5) 1(12.5) 208(81.6) 47(18.4)
VO2max≤12.0ml/kg/min 10(71.4) 4(28.6) 205(82.3) 44(17.7)
VE/VCO2slope≥39.0 27(79.4) 7(20.6) 202(88.2) 27(11.8)
Figure1 Event-freesurvivalupto36monthsoffollow-upaccordingtoVE/VCO2slopethresholdof39.0,and95%CIofestimated
survivalforeachsubgroup. CI:95%confidenceinterval;ISHLT:InternationalSocietyforHeart andLungTransplantation; VCO2:
carbondioxideproduction;VE:minuteventilation.
Figure2 Event-freesurvivalupto36monthsoffollow-up(excludingpatientswithVE/VCO2slope≥39.0)and95%CIofestimated
survivalforeachsubgroup,accordingtothecombinationofsodium(136.0mEq/l),creatinine(1.0mg/dl)andvariationof
end-tidalcarbondioxide partialpressure(0.45kPa)cut-offvalues.Group1:uptooneabnormalparameter; group2:twoabnormal
parameters;group3: threeabnormal parameters.CI:95%confidenceinterval; ISHLT:InternationalSocietyfor Heartand Lung
Transplantation;VCO2:carbondioxideproduction;VE:minuteventilation.
etersaspotentialpredictorsofadverseoutcome.Thesingle best parameter of all those studied was VE/VCO2 slope, yieldingthehighestWaldchi-squarevalueandthehighest AUCat12, 36and60monthsof follow-up,andwitheven morediscriminativepowerthantheHFSS,whichcombines differentvariables.Ofnote,VO2maxdidnotremaininany multivariatemodelin ourcohort.The 60-monthfollow-up periodwasnotused toidentifyan approachfor selecting patientsfor HTX,sinceprognosticreassessmentshouldbe undertakenearlier.7,8 Nevertheless, the long-term
follow-upcarriedout confirmedthe consistentlyhigheraccuracy ofVE/VCO2slopeoveralongperiod,incomparisontoother
parameters.This finding hasnot been properlyaddressed in previous studies.6---8 Arena et al.6 showed that the risk
ofadverseeventsincreasescontinuouslyoverdifferent cat-egories of VE/VCO2 slope. However,thresholds areuseful
forclinicalpractice,andVE/VCO2slopevaluesof34.0and
35.0havebeenproposedasoptimalcriteriaforclassifying patientswithHFashigh-andlow-risk.5,14Inlinewiththese
thresholds,theriskofadverseeventsbegantoriseforvalues
above32.0inourcohort,asshownbyanalysisofthe mar-tingaleresiduals.Althoughthespecificityforthisthreshold wasnotlow(80%),a non-negligibleproportionofpatients wouldbeclassifiedashigh-riskeventhoughtheywouldnot experiencean adverse outcome. If this was considered a listingcriterion for HTX,issuesrelatedtothe shortageof donors and tothe morbidity and mortality following HTX mightarise, bylistingpatients whowere at aless severe stageofHF.Therefore,thethresholdof39.0,whichprovides very high specificity with reasonable sensitivity, may be moreappropriateforselectingpatientsforHTXthanlower cut-offvalues.5,14 Freedomfromthecombinedendpointat
36months of follow-up wasverylow (19.3%) for patients withVE/VCO2slope≥39.0.Eventhoughthisthresholdwas
notprimarilyidentifiedtopredicttotalmortality,the95%CI ofestimatedoverallsurvivalofhigh-andlow-risksubgroups did not overlap with those of post-HTX reported by the ISHLT.15Thisfindingsuggeststhat,forourcohort,survivalof
hypotheticallytransplantedpatients wouldbebetterthan survival of non-transplanted high-risk patients and worse
thansurvivalofnon-transplantedlow-riskpatients.In addi-tion, the threshold of 39.0 for VE/VCO2 slope was more
accurate than the cut-off values of 10 or 12 ml/kg/min forVO2max,whicharerecommendedaslistingcriteria.4,5
VE/VCO2slopehaspreviouslybeenreportedasprovidinga
discriminativepoweratleastasgoodasVO2maxfor
predict-ingadverseevents.6,9Nevertheless,themajorityofstudies
thathighlightedthevalueofVE/VCO2 slopedidnotassess
thisparameterinthelightofacomprehensiveassessment of clinical, laboratory, electrocardiographic, echocardiog-raphic and CPET parameters, and few had a long-term follow-up.6---8Comparedtopreviousstudies,wecarriedouta
morecomprehensive(andprospective)baselineassessment, withalong-termfollow-up,andemployingarobust statis-ticalanalysiswithconsistentresults.Basedonourresults, patientswithVE/VCO2slope≥39.0maybenefitfromHTX.
Forlow-risk non-transplantedpatientswithsodium lev-els>136.0mEq/l,creatininelevels<1.0mg/dlandPetCO2
>0.45kPa(3.4mmHg),orpatientswithuptooneofthese variablesclassifiedasabnormal,theprognosiswasexcellent andtotalmortalitywaslowerthanthatreportedfor post-HTX.15Sodiumandcreatininelevelsareknownindependent
predictorsofadverse eventsinHFandareincludedinrisk scoressuchastheHFSSandtheMeta-AnalysisGlobalGroup inChronicHeartFailurescore.1,2PetCO
2atrestandatthe
anaerobicthresholdwereshowntostratifyriskbeyondthe VE/VCO2 slope, and combiningthem into a single
param-eter( PetCO2)may bemore practicaland accurate.20,21
We suggest that particular attention should be paid to sodiumlevels,creatininelevelsandPetCO2,inadditionto
VE/VCO2slope,particularlyforidentifyinglow-riskpatients
inclinicalpractice.
Somelimitations ofthestudyshouldbeacknowledged. Firstly,theanalyzedcohortwasnotlarge.However,itwas possibletoidentifythemostimportantindependent predic-torsofadverseoutcomeinHFandastrategyforoptimizing theselectionofpatientsforHTX,andtheresultswere con-sistentusingdifferentstatisticalanalyses.Thesamplesize issimilartothoseofotherstudiesthathighlightedthevalue ofCPETparameters,includingfortheselectionofpatients forHTX.6,14,18,20,21 Secondly,thiswasasingle-centerstudy.
Nevertheless,thismeant thattheCPETprotocolwas con-sistentin allcases,andmay havereducedthe numberof physiciansresponsiblefor the interpretationof theexam, reducinginterobservervariability.Thirdly,listingforHTXis acomplexandmultidisciplinarydecisionandshouldnotrely solelyon‘magicnumbers’ofspecificparametersfrom com-plementaryexams;however,thresholdsareusefulinclinical practice,aspointedoutabove.Inaddition,theaimwasnot toreplacebuttopotentiallyoptimizecurrentlistingcriteria forHTX.Thedecision thresholdwe proposeisinline with current practicein differentcenters,where clinical deci-sions aresupported by VE/VCO2 slope data inaddition to
VO2max,eventhoughcurrentguidelinesdonotaddressthis
approach.4,5,7Afurthervalidationstudywouldcertainlybe
useful.
Conclusions
Amongalargevarietyofpredictorsofadverseoutcomein ambulatory patientswithHF withreducedLVEF, themost
accuratewasVE/VCO2slope. PatientswithVE/VCO2slope
of39.0orhighermaybenefitfromHTX.BeyondtheVE/VCO2
slope,sodiumlevels,creatinine levelsandPetCO2were
abletoidentifypatientswithexcellentoutcome.
Funding
None.
Conflicts
of
interest
Theauthorshavenoconflictsofinteresttodeclare.
Appendix
A.
Supplementary
material
Supplementary material associated with this article can be found in the online version at doi:10.1016/ j.repc.2017.06.018.
References
1.PocockSJ,AritiCA,McMurrayJJ,etal.,Meta-AnalysisGlobal GroupinChronicHeartFailure.Predictingsurvivalinheart fail-ure:ariskscorebasedon39372patientsfrom30studies.Eur HeartJ.2013;34:1404---13.
2.Zugck C, Krüger C, Kell R, et al. Risk stratification in middle-agedpatientswithcongestiveheartfailure: prospec-tivecomparisonoftheHeartFailureSurvivalScore(HFSS)anda simplifiedtwo-variablemodel.EurJHeartFail.2001;3:577---85.
3.TaylorDO,EdwardsLB,AuroraP,etal.Registryofthe Interna-tionalSocietyforHeartandLungTransplantation:twenty-fifth officialadulthearttransplantreport---2008.JHeartLung Trans-plant.2008;27:943---56.
4.FrancisGS,GreenbergBH,HsuDT,etal.,ACCF/AHA/ACPTask Force. ACCF/AHA/ACP/HFSA/ISHLT2010 clinicalcompetence statementonmanagementofpatientswithadvancedheart fail-ureandcardiactransplant:areportoftheACCF/AHA/ACPTask ForceonClinicalCompetenceandTraining.JAmCollCardiol. 2010;56:424---53.
5.MehraMR,KobashigawaJ,StarlingR,etal.Listingcriteriafor hearttransplantation:InternationalSocietyforHeartandLung Transplantationguidelines for thecare ofcardiac transplant candidates---2006.JHeartLungTransplant.2006;25:1024---42.
6.ArenaR,MyersJ,AbellaJ,etal.Developmentofaventilatory classificationsysteminpatientswithheartfailure.Circulation. 2007;115:2410---7.
7.FerreiraAM,TabetJY,FrankensteinL, etal.Ventilatory effi-ciencyandtheselectionofpatientsforhearttransplantation. CircHeartFail.2010;3:378---86.
8.MyersJ,OliveiraR,DeweyF,etal.Validationofa cardiopul-monaryexercise testscore in heartfailure. CircHeart Fail. 2013;6:211---8.
9.MezzaniA, Corrà U, BosiminiE, et al. Contribution ofpeak respiratoryexchangeratiotopeakVO2prognosticreliabilityin
patientswithchronicheartfailureandseverelyreduced exer-cisecapacity.AmHeartJ.2003;145:1102---7.
10.Guazzi M, Adams V, Conraads V, et al., European Associa-tionforCardiovascularPrevention&Rehabilitation;American Heart Association. EACPR/AHA Scientific Statement. Clini-cal recommendations for cardiopulmonary exercise testing dataassessment in specific patient populations. Circulation. 2012;126:2261---74.
11.O’NeillJO,YoungJB,PothierCE,etal.Peakoxygen consump-tion as a predictor of death in patients with heart failure receivingbeta-blockers.Circulation.2005;111:2313---8.
12.PencinaMJ,D’AgostinoRBSr,D’AgostinoRBJr,etal. Evalu-atingtheaddedpredictiveabilityofanewmarker:fromarea undertheROCcurvetoreclassificationandbeyond.StatMed. 2008;27:157---72.
13.HarrellFE.CasestudyinCoxregression.In:HarrellFE, edi-tor.Regressionmodelingstrategies:withapplicationstolinear models,logisticregression,andsurvivalanalysis.1sted.New York,NY:Springer;2001.p.509---22.
14.Chua TP, Ponikowski P, Harrington D, et al. Clinical corre-latesand prognosticsignificance of theventilatory response to exercise in chronic heart failure. J Am Coll Cardiol. 1997;29:1585---90.
15.ISHLT. Transplant Registry Quarterly Reports for Heart in Europe. Available at: http://www.ishlt.org/registries/ quarterlyDataReport.asp[accessed15.01.17].
16.BabaR,NagashimaM,GotoM,etal.Oxygenuptakeefficiency slope: a new index of cardiorespiratory functional reserve
derivedfromtherelationbetweenoxygenuptakeandminute ventilation during incremental exercise. J Am Coll Cardiol. 1996;28:1567---72.
17.StevensonLW,PaganiFD,YoungJB,etal.INTERMACSprofiles ofadvancedheartfailure:thecurrentpicture.JHeartLung Transplant.2009;28:535---41.
18.ManciniDM,EisenH,KussmaulW,etal.Valueofpeakexercise oxygen consumptionfor optimaltimingofcardiac transplan-tationinambulatorypatientswithheartfailure.Circulation. 1991;83:778---86.
19.ManciniD,LietzK.Selectionofcardiactransplantation candi-datesin2010.Circulation.2010;122:173---83.
20.ArenaR,PeberdyMA,MyersJ,etal.Prognosticvalueofresting end-tidalcarbon dioxideinpatientswithheartfailure.IntJ Cardiol.2006;109:351---8.
21.ArenaR,GuazziM,MyersJ.Prognosticvalueofend-tidalcarbon dioxideduringexercisetestinginheartfailure.IntJCardiol. 2007;117:103---8.