Complement activation is associated with poor outcome after out-of-hospital cardiac arrest

(1)

Clinical paper

Complement activation is associated with poor outcome after out-of-hospital cardiac arrest

Viktoriia Chaban

^a

, Espen R. Nakstad

^b

, Henrik Stær-Jensen

^c

, Camilla Schjalm

^a

, Ingebjørg Seljeflot

^d,e

, Jarle Vaage

^e,f,g

, Christofer Lundqvist

^e,h,i

,

Jurat e _ altyt e _ Benth

^e,i

, Kjetil Sunde

^c,e

, Tom Eirik Mollnes

^a,e,j,k

, Geir Ø. Andersen

^d

, Søren Erik Pischke

^a,c,e,f,

*

aDept.ofImmunology,UniversityofOsloandOsloUniversityHospital,Oslo,Norway

bDept.ofAcuteMedicine,OsloUniversityHospital,Oslo,Norway

cDept.ofAnaesthesiology,DivisionofEmergenciesandCriticalCare,OsloUniversityHospital,Oslo,Norway

dDept.ofCardiology,OsloUniversityHospital,Oslo,Norway

eInstituteofClinicalMedicine,UniversityofOslo,Oslo,Norway

fDept.ofResearchandDevelopment,DivisionofEmergenciesandCriticalCare,OsloUniversityHospital,Oslo,Norway

gSectionofPhysiology,Dept.ofMolecularMedicine,InstituteofBasicMedicalSciences,UniversityofOslo,Oslo,Norway

hDept.ofNeurology,AkershusUniversityHospital,Oslo,Norway

iHealthServicesResearchUnit,AkershusUniversityHospital,Oslo,Norway

jResearchLaboratory,NordlandHospitalBodø,andK.G.JebsenTREC,UniversityofTromsø,Norway

kCentreofMolecularInflammationResearch,DepartmentofClinicalandMolecularResearch,NorwegianUniversityofScienceandTechnology, Trondheim,Norway

Abstract

Background:Cardiopulmonaryresuscitationaftercardiacarrestinitiatesawhole-bodyischemia-reperfusioninjury,whichmayactivatetheinnate immunesystem,includingthecomplementsystem.Wehypothesizedthatcomplementactivationandsubsequentreleaseofsolubleendothelial activationmarkerswereassociatedwithcerebraloutcomeincludingdeath.

Methods:Outcomewasassessedatsixmonthsanddefinedbycerebralperformancecategoryscale(1 2;goodoutcome,3 5;pooroutcome includingdeath)in232resuscitatedout-of-hospitalcardiacarrestpatients.Plasmasamplesobtainedatadmissionanddaythreewereanalysedfor complementactivationproductsC3bc,thesolubleterminalcomplementcomplex(sC5b-9),andsolubleCD14.Endothelialcellactivationwasmeasured bysolublemarkerssyndecan-1,sE-selectin,thrombomodulin,andvascularcelladhesionmolecule.

Results:Forty-ninepercentofthepatientshadgoodoutcome.C3bcandsC5b-9weresignificantlyhigheratadmissioncomparedtodaythree (p<0.001forboth)andinpatientswithpoorcomparedtogoodoutcome(p=0.03andp<0.001,respectively).Unadjusted,highersC5b-9atadmission wasassociatedwithpooroutcome(oddsratio1.08(95%CI1.01 1.14),p=0.024).Adjusted,sC5b-9wasstillassociatedwithoutcome,butthe associationbecamenon-significantwhentimetoreturn-of-spontaneous-circulationabove25minwasincludedasacovariate.Endothelialcell activationmarkersincreasedfromadmissiontodaythree,butonlysE-selectinandthrombomodulinweresignificantlyhigherinpatientswithpoorversus goodoutcome(p=0.004andp=0.03,respectively)andcorrelatedtosCD14andsC5b-9/C3bc,respectively.

* Correspondingauthorat:OsloUniversityofOslo,OsloUniversityHospital—Rikshospitalet,Dept.forImmunologyandTransfusionMedicine, ComplementGroup,Sognsvannsveien20,0372Oslo,Norway.

E-mailaddress:[email protected](S.E.Pischke).

https://doi.org/10.1016/j.resuscitation.2021.05.038

Received23February2021;Receivedinrevisedform4May2021;Accepted30May2021

by/4.0/).

Availableonlineatwww.sciencedirect.com

Resuscitation

j our na lho me pa g e :ww w. e l s e v i e r. c om/ l o ca t e / re s usc i ta t i on

(2)

Conclusion:Complementsystemactivation,reflectedbysC5b-9atadmission,leadingtosubsequentendothelialcellactivation,wasassociatedwith pooroutcomeinout-of-hospitalcardiacarrestpatients.

Keywords:Out-of-hospitalcardiacarrest,Outcome,SC5b-9proteincomplex,Cardiopulmonaryresuscitation,Returnofspontaneouscirculation, Immunesystem,Endothelialcells

Introduction

Out-of-hospitalcardiacarrest(OHCA)remainsanimportantpublic healthmatterwithahighmortalityrateandisthethirdleadingcauseof deathinEurope.¹Aftersuccessfulresuscitation,themainreasonfor mortalityandmorbidityisthedurationofhypoxiaandawhole-body ischemia/reperfusioninjury.² Thebrainismostsusceptibletothis reperfusioninjury,³andtwo-thirdsofhospitaldeathsareduetothe neurologicalinjury,⁴

The pathophysiology of ischemia/reperfusion injury is very complexwithaplethoraofplayers.Cardiopulmonaryresuscitation (CPR)initiated reperfusiontriggerstheinnateimmunesystemby dangeragentsactingasligandsforpatternrecognitionmoleculesof thevariousbranchesoftheinnateimmunity,includingthecomplementsystem.^5,6Thisinducesasecondaryandbroad-actingsystemic inflammatoryresponsewhich,whenover-activatedordysregulated, leadstotissuedamage,organfailureandinworstcasetodeath.⁷

Complementispresentinplasmaandthusimmediatelyactivated uponinjurythroughtheclassical,lectin,oralternativepathway.All threepathwaysconvergeatthelevelofC3,whichgetscleavedand formsaproteasethatcleavesC5leadingto theformationofthe terminalcomplementcomplex.Theterminalcomplementcomplexis eitherincorporatedincellularmembranes,whereitmayresultinlysis andinflammation,oritisreleasedasasolublemolecule(sC5b-9)to the fluid phase.⁸ All C3 and C5 derived complement activation products,includingtheanaphylatoxinsC3aandC5a,canactivate endothelialcells.

Theaimofthepresentstudywastoinvestigateifcomplement activationwithsubsequentendothelialcellactivationaspartofthe initialischemia/reperfusioninjurywasassociatedwithpoorcerebral outcomeanddeathinpatientsresuscitatedafterOHCA.

Material and methods

The present study is a planned sub-study of the prospective observational Norwegian Cardio-Respiratory Arrest Study (NOR- CAST,NCT01239420).⁹TheaimofNORCASTwastoassessthe abilityofcurrentlyrecommendeddiagnostictoolstoidentifypatients withapoorprognosis.Importantly,resultsofprognostictestswere blindedtoclinicianstoavoidhastywithdrawal decisionsandself- fulfillingprophecies.ThedesignandpatientpopulationinNORCAST hasbeenpreviouslydescribedindetail.⁹Briefly,259comatoseadult OHCApatientsadmitted toOsloUniversity Hospital Ullevål were includedbetweenOctober1st,2010andJanuary30th,2014.Post- resuscitation care was performed according to local standard proceduresincludingtargeted temperaturemanagementto33 C (TTM33) for 24 h and immediate coronary angiography with subsequentpercutaneouscoronaryintervention,ifacardiaccause wassuspected.⁹Hypoxiccauseofcardiacarrestwasdefinedasa non-cardiaccauseofthearrest,wherehypoxiaofdifferentreasons resultedinunresponsiveness,absenceofbreathingandlossofheart

function.TheSequentialOrganFailureAssessment(SOFAscore) wasperformedtoassesstheextentofapatient'sorganfailurewithin thefirst24hofadmission.Neurologicaloutcomeat6monthswas assessedbyNeurologyspecialistswhowereblindedtoallclinicaland paraclinicalscoresduringhospitaladmissionandwellacquaintedwith CerebralPerformanceCategory(CPC,1 2goodoutcome,3 5poor outcomewith5includingdeath)scoring.Theexaminationonwhich CPC scoringwas basedconsistedof structuredneurological and cognitiveexaminationaswellasqueriesoffunctionalleveltopatients and next-of-kin. For complement activation markers, the upper reference limitforeachassaywas settothe95th percentileofa healthyreferencepopulationconsistingof20femaleand20male blood donors without cardiovascular or autoimmune diseases.¹⁰ UpperreferencelimitofsC5b-9is>0.7complementactivationunits (CAU)/mLandforC3bc>9CAU/mL.¹⁰Forendothelialmarkerssuch international reference limit donot existandvenousEDTA-blood obtainedfromtwelvehealthyvolunteerswithaclosetosimilarage (median 59 years,interquartile range 51 68 years) and gender distribution(7female:5male)comparedtothestudypopulationwas usedascontrol.

Bloodsamplingprotocol

At admission (before initiation of TTM33) and after 72 h (after rewarming),peripheralvenousbloodwasobtainedinethylenedia- minetetraaceticacid(EDTA)vacutainertubes(BD,Plymouth,UK).

Sampleswere immediatelykeptoncrushediced andcentrifuged within30minat2500gfor20minat4C.Plasmawascollectedand storedat 80Cuntilanalyses.

Complementactivationmarkers

Complementactivationproducts,C3bc(commonforclassical,lectin andalternativepathways)andsC5b-9(solubleterminalcomplement complex),weremeasuredbyin-houseenzyme-linkedimmunosor- bentassays(ELISA)whichhaspreviouslybeendescribed^11,12and latermodifiedaccordingtotheprotocolpresentedinRef.10.

EndothelialactivationmarkersandsCD14

Soluble endothelial activation markers syndecan-1, sE-selectin (sCD62E),thrombomodulin(sTM),vascularcelladhesionmolecule 1(sVCAM-1)and sCD14weremeasured byDuoSetELISA(R&D Systems,Minneapolis,MN)inEDTAplasma.

Ethicsanddatamanagement

The Regional Committeefor Medical Research Ethics of South- EasternNorwayapprovedNORCAST(ApprovalnumberREKS-O/A- 2010/1116a).Ethicsandstudymanagementhavebeenpreviously reported.⁹Writteninformedconsentwasobtainedfromcloserelatives orguardianswithin24 hafterhospitalizationandfromallpatients regaining consciousness and decision-making capacity withinsix

(3)

months.Clinical,biochemicalandoutcomedatawereprospectively collected from medical records, questionnaires, and paramedic recordsaspresentedintheUtsteincriteria.¹³Bloodsamplesfrom healthy donors were collected into a biobank approved by the RegionalEthicalCommittee(REKS-04114).

Statisticalanalysis

Datawerepresentedasmedians,25th 75thpercentile,orfrequen- ciesandpercentages,asappropriate.Patientswithgoodandpoor outcomewerecomparedbyMann WhitneyUtestforcontinuousand

x

²^-test^or^Fisher’sexacttestforcategoricaldata.Thedifferencein complement or endothelial activation between CPC groups was assessedbyMann WhitneyUtest.Kruskal Wallistestwasusedfor comparison of more than twogroups. Bonferroni correction was appliedto correct for multiple testing. TheWilcoxon signed-rank pairedtestwasusedtocomparethecomplementandendothelial activationatadmissionanddaythree.Associationsbetweencerebral outcome and SOFA score,time from cardiac arrest-to-return-of- spontaneouscirculation(time-to-ROSC),cardiacarrest(unwitnessed vs.witnessed),initialrhythm(shockablevs.non-shockable),bystand- erCPR(yesvs.no),andcauseofcardiacarrestwereassessedby unadjustedandadjusted(multivariable)logisticregressionmodel.To assesswhethersC5b-9andC3bcatadmissionwasassociatedwith the odds for poor outcome separately as well as combined, unadjustedlogisticregressionmodelswereestimated.Themodels werefurtheradjustedforSOFAscore,time-to-ROSC,andcauseof cardiacarrest.InteractionsbetweensC5b-9andtime-to-ROSC,and betweenC3bcandtime-to-ROSCwereincludedintotheadjusted modelifsignificant.Theresultsoflogisticregressionmodelswere presented as odds ratios (ORs) with the corresponding 95%

confidenceintervals(CIs)andp-values.Theregressioncoefficients andstandarderrors(SE)werepresentedforthevariablesincluded intotheinteraction term.Foreasierinterpretation,theinteractions wereillustratedgraphically.ToassesshowwellthesC5b-9andC3bc distinguishesbetweenthosewithgoodandpooroutcome,thearea underthecharacteristic(ROC)curvewiththecorresponding95%CI wascalculated. Thelogisticregressionmodelswereestimatedto assess the association between the CPC score and sC5b-9 at admissionadjustedforendothelialmarkersSyndecan-1,E-selectin, Thrombomodulin,andVCAM,andCD14measuredatdaythree,one atatime.Also,themodelcombiningallendothelialmarkersandCD14 wasestimated.ThemodelswereadjustedforSOFAscore,time-to- return-of-spontaneous circulation (time-to-ROSC), and cause of cardiac arrest. Area under the ROC curve calculated for the unadjusted and eachadjusted model was compared. Correlation between complement and endothelial activation markers was assessedusingSpearman’sranktest.Resultswithp-values<0.05 wereregardedassignificant.Statisticalanalyseswereperformedin StatisticalPackagefortheSocialSciences(SPSS)software(version 26, IBM, Armonk, NY), Stata (version 16.0, StataCorp, College Station, TX), SAS (version 9.4, SAS Institute, Cary, NC) and GraphPadPrism(version8,GraphPadSoftware,SanDiego,CA).

Results

Outof259successfullyresuscitatedpatientsafterOHCA,232were includedintothestudy(Fig.1).Medianagewas63(54 69),83%

weremale,and82%hadcardiaccauseofarrest(Table1).Outcome

definedasbestCPCscorewithinsixmonthsafterOHCA,waspoorfor 49%andgoodfor51%ofpatientswithsignificantdifferencesintime- to-ROSCandSOFAscoreatadmission(Table1).Clinicalvariables significantly associated with poor outcome in adjusted logistic regressionanalysiswerenon-shockableinitialcardiacarrestrhythm, longertime-to-ROSCandhigheroverallorgandysfunctionduringthe first24hafteradmission(Suppl.Table1).Unwitnessedcardiacarrest wasassociatedwithoutcomeinunadjustedlogisticregressionmodel only, while performance of bystander CPR was not significantly associatedwithoutcome(Suppl.Table1).

Complementactivationproducts

C3bcandsC5b-9weremarkedlyhighercomparedtoupperreference limitinallpatients(n=232)bothatadmissionanddaythree(Fig.2A, B). C3bcand sC5b-9wereboth significantlyhigher atadmission comparedtodaythree.Patientswithpooroutcomehadsignificantly higherlevelsofC3bcandsC5b-9atadmissioncomparedtopatients withgoodcerebraloutcome(Fig.2A,B).

Inunadjustedmodels,highersC5b-9atadmissionwasassociated withpooroutcome(OR1.08,95%CI(1.01 1.14),p=0.02,Suppl.

Table2),whileC3bcwasnot(Suppl.Table3).Theareaunderthe ROCcurve(AUC)was0.6595%CI(0.57;0.72)forsC5b-9and0.57 (0.49;0.65)forC3bc.TherewasnoadditiveeffectontheROCcurve bycombiningsC5b-9andC3bc(AUC0.66(0.58;0.73)).Longertime- to-ROSCandhigherSOFAscoresatadmissionwerealsoassociated withpooroutcome(OR1.06(1.05 1.09),p<0.001andOR1.23 (1.06 1.43), p= 0.006, respectively, Suppl.Table 2).Interaction betweensC5b-9andtime-to-ROSCwassignificantinanadjusted model(p=0.015,Suppl.Table2).Posthocanalysisoftheinteraction term showed that odds for poor outcome were increasing with increasingvaluesofsC5b-9whentime-to-ROSCwasbelow30min, whileoddsforpooroutcomeweredependentontime-to-ROSCalone at30minandlonger(Suppl.Table2,Fig.3).Therewasnosignificant interactionbetweenC3bcandtime-to-ROSCinanunadjustedmodel,

Admitted to hospitalafter out-of-hospital cardiac arrest (n=259)

Included with available clinical data (n= 243)

Included with available samples at admission(n=232)and admission + day 3 (n=192)

No blood collected at admission and day 3 (n=3) No blood collected at admission (n=8) No blood collected at day 3 (n=40)

TTM protocol discontinued, earlydeath, early awakening(n=16)

Fig.1–Flow chartofpatientpopulation inthisstudy.

Enrolmentinstudyandexclusions,availabilityofclinical dataandbloodsamplesateachtimepoint.TTM;target temperaturemanagement.

(4)

andC3bcalonewasnotassociatedwithoutcome(Suppl.Table3).

Hypoxiaasthecauseofcardiacarrestincreasedtheoddsforpoor outcome,withincreasedlevelsofsC5b-9andC3bcatadmissionin comparisontoacutemyocardialinfarction(OR6.78(1.90 24.16), p=0.003,andOR5.97(1.75 20.36),p=0.004,respectively,Suppl.

Tables2and3).

sCD14

sCD14,aco-receptorofTLR,wassignificantlyhigherincardiacarrest patients(n =192) comparedto healthycontrolsand significantly higheratdaythreecomparedtoadmission(Suppl.Fig.1).However, sCD14wasnotassociatedwithoutcomealoneorinadjustedmodels (Suppl.Table4).

Markersofendothelialactivationanddamage

Glycocalyx damage marker sSyndecan-1 was higher in OHCA patients(n=192)comparedtohealthycontrolsatadmissionandday three(Fig.4A).Itwassignificantlyhigheratadmissioncomparedto daythreeinOHCApatients(Fig.4A).Markersofendothelialactivation sE-selectin,sVCAM-1andsTMwereconsistentlyhigherinOHCA patientsandconcentrationsatdaythreeweresignificantlyincreased comparedtoadmission(Fig.4B D).Amongthefourmarkers,only sE-selectinandsTMweresignificantlyhigherinpatientswithpoor outcomecomparedtopatientswithgoodoutcome(Fig.5A,B).C3bc correlated significantly with sSyndecan-1 and sTM at admission (R=0.26,p<0.001andR=0.17,p=0.02,respectively)anddaythree (R=0.19,p=0.007andR=0.2,p=0.007,respectively,Suppl.Fig.2).

Likewise,sC5b-9correlatedsignificantlywithsSyndecan-1andsTM atadmission(R=0.25,p<0.001andR=0.23,p=0.001,respectively) andatdaythree(R=0.16,p=0.03andR=0.23,p=0.002,Suppl.

Fig.2).sCD14wassignificantlycorrelatedwithsE-selectin atday three(R=0.44,p<0.001).Nootherimportantcorrelationswere

observed (Suppl. Fig. 2). sE-selectin was the only endothelial parameter significantlyassociated withoutcomein an unadjusted model(OR1.02 [1.00 1.05],p=0.038),butdidnotimprovethe discriminativeabilityofthemodelwithonlysC5b-9(Suppl.Table4).

AdjustmentdidnotimprovethediscriminativeabilityofsC5b-9in combination with endothelial markers except for E-selectin with additiveeffectontheROCcurvewhensC5b-9wascombinedwithsE- selectin(AUC0.79(0.72 0.85),p=0.047,Suppl.Tables4and5).

Discussion

Awhole-bodyinflammatoryreactionfollowsresuscitationafterOHCA.

In thepresentstudy,we showthat initial strongactivation of the complementsystem,inparticularreflectedbytheterminalpathway activation product sC5b-9, was associated with poor long-term outcome.Followingtheinitialrapidcomplementactivation,endothe- lialactivationoccurredafteradmission.Whiletheendothelialdamage markerSyndecan-1peakedatadmission,allothermeasuredmarkers ofendothelialactivationshowedhighestlevelatdaythreeincontrast tocomplementactivation,whichthenhadmarkedlydeclined.

Complementactivation

Complementactivationatadmissionwasassociatedwithcerebral outcome.Complementhasbeenshowntogetactivatedimmediately afterinitiationofresuscitationinOHCApatientswithaswiftdecline afterROSC.¹⁴Thismightbeduetoreleaseofdamageassociated molecularpatterns(DAMPs)frominjuredcellsfollowedbyachainof injurious events¹⁵ leading to activation of complement signalling pathways, inparticular thelectinpathway.¹⁶While theaforemen- tioned studyshowed no relation oflectin pathwayactivation and outcome, thisstudyconfirms thattheend-productofcomplement systemactivationsC5b-9isassociatedwithpooroutcome.TTM33 Table1–Demographicsandclinicalcharacteristicsatadmission.

Total(n=232) Goodoutcome(CPC1 2) (n=118)

Pooroutcome(CPC3 5) (n=114)

p-value

Age(years) 63(54 70) 62(54 69) 64(55 71) 0.3^a

Sex 0.5^b

Female 39(17) 14(36) 25(64)

Male 193(83) 104(54) 89(46)

Causeofcardiacarrest(CA)

Acutemyocardialinfarction 89(38) 48(54) 41(46) 0.5^b

Chroniccardiacdisease 71(31) 40(56) 31(44) 0.3^b

Arrythmia(VF) 29(13) 18(62) 11(38) 0.2^b

HypoxiainducedCA 31(13) 9(29) 22(71) 0.01^b

Othercauses 5(2) 2(40) 3(60) 0.7^c

Unknown 7(3) 1(14) 6(86) 0.6^c

Survivors 123(53) 114(93) 9(7) <0.001^b

Scoresatadmission

SOFAscore 11(9.5 12) 10(9 12) 11(10 12) <0.001^a

Time-to-ROSC(min) 25(16 33) 19(12 28) 30(24 40) <0.001^a

Datapresentedasmedian(25th 75thpercentile)orn(%).CerebraloutcomeisbestCPCwithin6monthsaftercardiacarrest.CPC;cerebralperformance category,VF;ventricularfibrillation,CA;cardiacarrest,SOFA;SequentialOrganFailureAssessment,Time-to-ROSC;timetoreturnofspontaneouscirculation, min;minutes.

aMann WhitneyUtest.

bx²^-test^or

cFisher’sexacttest.

(5)

treatment ofOHCApatients leadstosuppression ofcomplement activationwithreturntolevelsabovereferencelimitafterrewarming¹⁷ whileonestudyreportsthatTTM33leadtoreductionoftheregulatory protein Map19 of the complement lectin pathway compared to TTM36.¹⁸AllpatientsinthisstudyweretreatedwithTTM33andit mightthusbespeculatedthattheinitialcomplementactivationisof specialimportanceforoutcomeinOHCApatients.Inlinewiththisare findingsthatthedegreeofcomplementactivationcorrelateswiththe severityofheartfailureinpatientsdevelopingcardiogenicshockafter acutemyocardialinfarction.¹⁹Furthermore,blockadeofC5reduced infarctsizeandimprovedcardiacfunctioninanexperimentalporcine modelofseveremyocardialinfarction.²⁰

Animportantfactorinfluencingthereperfusioninjuryanddegreeof neurologicalinjuryisthedurationofhypoxiai.e.,thetime-to-ROSC.²¹ Asinpreviousstudies,wealsoverifiedanassociationbetweentime- to-ROSCandpoorcerebraloutcome.²²Weaddthatforthesame time-to-ROSC, odds for poor outcome increased with increasing levelsofsC5b-9,suggestingthatincreasedcomplementactivation

persehadaneffectonoutcomeandmightbeaneffect-modifier.The effectofassociationbetweensC5b-9andoutcomewasabsentin patients with time-to-ROSC 30 min, indicating that theimpact of inflammationon cerebraloutcomemaybe reducedwhentime-to- ROSCislongandischemicbraindamagebecomestheonlydecisive factor.DespitetheassociationbetweensC5b-9andpooroutcome, complement activation cannot be used asa decisive prognostic marker,giventhelargeinter-patientvariationsofabsolutevalues, whichwereinthisstudyalmostallaboveaninternationallyaccepted upperreference limit.¹⁰ However,complement inhibitioninOHCA patientsmightbeexploredinfurtherstudiesasinhibitionofC5has beenshowntoreduceischemia/reperfusioninjuryingeneral,andour dataonsC5b-9supportthishypothesis.

sCD14

sCD14isamultifunctional molecule,whichrecognisesandbinds endogenousandexogenousdangersignals.²³Itisalsoaco-factorfor several TLRs which are mainly expressed by monocytes and macrophages. In the present study, sCD14was increasedupon admissionandhighestondaythree,suggestingamaintainedrelease, whichmightbeduetosheddingofmembrane-boundCD14proteinor secretion via intracellular vesicles by inflammatory cells.²⁴ The prolongedreleaseofsCD14mayindicateprolongedactivationofthe immunesystemwhichisinagreementwiththepatternofcomplement andendothelialactivation.

Endothelialactivation

Theendotheliumisanactivecomponentofinnateimmunity.Itgets injuredduringischemia/reperfusioninjury,²⁵e.g.bycomplement.This againleadstoretrogradecomplementactivationresultinginavicious circlewithexaggeratedinflammatoryreactioncausingtissuedamage andorganfailure.²⁶

A

B

sC5b-9 (CAU/ml)C3bc (CAU/ml)

0 5 10

15 p<0.001

p<0.001 p<0.001

0.7 0 50 100

150 p=0.03

p<0.001

Good cerebral outcome

Poor cerebral outcome Admission Day 3 Admission Day 3

Good cerebral outcome

Poor cerebral outcome Admission Day 3 Admission Day 3

p<0.001

9

Fig.2–Patients(n=232)withsuccessfulresuscitation after out-of-hospital-cardiac-arrest. Plasma levels of C3bcandsC5b-9atadmission(bluelines)wereabove upper reference limit (line) in most of the patients, significantlyhigheratadmissioncomparedtodaythree (redlines)andhigherinpatientswithpoorcomparedto goodcerebraloutcome(A,B).

Cerebraloutcomewasdefinedbycerebralperformance category(CPC)asgood;CPC 1 2andpoor; CPC3 5.

Dataisshownasboxplotswithmedianaslineandbox indicating25th 75thpercentileandwhiskersrepresent- ing10th 90thpercentiles.Wilcoxonsigned-rankpaired testandnon-pairedMann WhitneyUtest.

CAU;complementarbitraryunit.

0.00 0.05 0.10 0.15 0.20 0.25

1 2 3 4 5 6 sC5b-9(CAU/ml) Odds for poor outcome for different value of Time-to ROSC

Time-to-ROSC 30min 25min 20min 15min 5min

*

Fig.3–InteractionbetweensC5b-9andtime-to-ROSCis associated with poor outcome. Increasing values of sC5b-9wereassociatedwithpooroutcomewhentime- to-ROSCwas25minbutnotwhentime-to-ROSCwas30 min or higher (asterix). Odd ratiosadjusted for cova- riatesSOFA,Time-to-ROSCandcauseofcardiacarrest.

CAU;complementarbitraryunit,Time-to-ROSC;time-to- return-of-spontaneous-circulation, SOFA; Sequential OrganFailureAssessment.

(6)

Syndecan-1wastheonlyendothelialmarkerthatpeakedonthe dayofadmission,whichisin-linewithpreviousfindingsdescribing sheddingofglycocalyxasaninitialeventinOHCApatients,where Syndecan-1 is an integral part.²⁷ While sSyndecan-1 correlated significantlywithcomplementactivation,itwasnotassociatedwith outcome.Thus,ourfindingsconfirmresultsofapreviousstudyin163 comatose patients after OHCA, where sSyndecan-1 was not associatedwith mortality,²² whichmayimplythatsSyndecan-1is anendothelialdamagemarker,butnotnecessarilyassociatedwith prolongedinflammatoryreperfusioninjury.

Complementsystemandendothelialcellactivation havebeen showntocorrelatewithsystemiccytokinereleaseandhaemodynamic statusearlypost-cardiacarrest.Themagnitudeofthisresponsehas been associated with the severity of this post-cardiac arrest syndrome,^5,8independentonTTM.²⁸Likewise,endothelialactivation anddamagemarkershavebeenshowntobeindependentofTTM, exceptforsE-Selectin,whichislowerinTTM36treatedpatients.²² The present study confirms this previously observed pattern of

complementandendothelialactivationandthatendothelialactivation reflectedbythrombomodulinisassociatedwithoutcome,²²although individualcorrelationsofsingleparametersarerelativelyweak.Ina study on 163 patients with one-time blood sampling within approximately two hours after OHCA, Cox proportional-hazard analysesfoundsE-selectintobeaunivariatepredictorofmortality.²⁹ OurfindingsaddthatsE-selectinisassociatedwithoutcome,also whenassessedthreedaysafterOHCA.Inaddition,weshowthatsE- selectiniscorrelatedwithsCD14atdaythreeaftercardiacarrestand not with the complement system, highlighting the multifactorial pathogenesis of the post-cardiac arrestsyndrome. Unfortunately, duetolargeinter-patientvariationsseeninourandtheaforemen- tioned studies, thrombomodulin and sE-selectin are not reliable prognostic markers. Especially sE-selectin levels are affected by body-massindex.³⁰However,bothsE-selectinandthrombomodulin couldbeusedinfutureclinicalinterventionstudiesassessingeffectof aninterventionbyreductionoftheseendothelialactivationmarkers.

DualinhibitionofboththecomplementsystemandCD14inpost- OHCApatientsmaybeageneralapproachtoattenuatetheinnate immunesystembroadly,preventingdownstreamoverreactionofthe

A B

C D

sSyndecan-1(ng/ml) sE -selsction (ng/ml)

Admission Day 3 Control Admission Day 3 Control

0 100 200 300

p=.005 p<0.001 p<0.001

0 25 50 75

p<0.001 p<0.001 p<0.001

0 5 10 155

p<0.001 p<0.001

0 500 1000 1500500

sVCAM (ng/mL) sTM (ng/mL)

p=.003 p<0.001 p<0.001

Fig.4–Patients(n=192)withsuccessfulresuscitation afterout-of-hospital-cardiac-arrestand bloodsamples both at admission and day three. Plasma levels of sSyndecan-1(A),sE-selectin(B),and sVCAM(C)were significantlyhighercomparedtohealthycontrols(black lines)atadmission(bluelines)anddaythree(redlines), whilesTM(D)wassignificantlyhigheratdaythree,only.

sSyndecanwassignificantlyhigheratadmissioncom- paredtodaythree,whilesE-selectin,sVCAMandsTM werehigheratdaythreecomparedtoadmission.Datais shown as box plots with median as line and box indicating 25th 75th percentile and whiskers repre- senting 10th 90th percentiles. Wilcoxon signed-rank pairedtestandnon-pairedMann WhitneyUtest.sTM;

sThrombomodulin.

A

B

sE-selectin(ng/ml)sTM (ng/ml)

0 5 10

15 p=0.03

p<0.001 p<0.001 0

20 40 60

p<0.001 p<0.001 p=0.004

Good cerebral

outcome outcome

Poor cerebral

Good cerebral

outcome outcome

Poor cerebral Admission Day 3 Admission Day 3

Admission Day 3 Admission Day 3

Fig.5–sE-Selectin(A)andsTM(B)inpatients(n=192) withsuccessfulresuscitationafterout-of-hospital-cardiac-arrest.Inpatientswithpoorcerebraloutcome,sE- selectinandsTMweresignificantlyhigheratdaythree (redlines)comparedtoadmission(bluelines).

Cerebraloutcomewasdefinedbycerebralperformance category (CPC)as good;CPC1 2 andpoor; CPC3 5.

Dataisshownasboxplotswithmedianaslineandbox indicating25th 75thpercentileandwhiskersrepresent- ing10th 90thpercentiles.Wilcoxonsigned-rankpaired test and non-paired Mann Whitney U test. sTM;

sThrombomodulin.

(7)

immunesystem including activation of endothelialcells, cytokine storm,andadaptiveimmunesystem,^31,32andoughttobeinvestigated further.

Limitations

Thiswasapost-hocanalysisofasingle-centerstudy,whichlimits generalizability.Despitearigoroussamplingroutine,bloodwasnot acquired from 40 patients at day three, decreasing power of especially endothelial markers, which are known to be time- sensitiveandincreaselaterthancomplementactivationmarkers.All patients were treated with the same TTM33 protocol reducing generalizationto internationally recommended post resuscitation care, which includes TTM36. However, by focusing on TTM33, confoundersbased on quality of care were minimized. Time-to- ROSCisapurequantitative parameter,andthequalityof CPR, whichisaknownpredictorofoutcome,wasnotassessed.³³This couldthereforeimpactonthedata.Asthisisanobservationalstudy, nocause-effectconclusionscanbemade,thestudyshouldthusbe regarded ashypothesis generating,and may facilitatedesignof future studies investigating interventions in the immediate post- cardiacarrestphase.

Conclusion

Incomatose,resuscitatedOHCApatients,activationofthecomple- mentsystemispresentinthemajorityofpatientsatadmission.In particular,increasedsC5b-9wasassociatedwithpooroutcome.The whole-body inflammation included subsequent endothelial cell activationandsCD14releasethreedaysafterOHCA.

Authorship contribution statement

Authorshavecontributedasfollowsto(1)Conceptionanddesignofthe study (all authors),or acquisitionof data (Nakstad, Stær-Jensen, Seljeflot,Lundqvist,Sunde,Andersen),oranalysisandinterpretationof data(Chaban,Schjalm,Vaage,Benth,Mollnes,Pischke),(2)drafting thearticleorrevisingitcriticallyforimportantintellectualcontent(all authors),(3)finalapprovaloftheversiontobesubmitted(allauthors).

Funding

Personalfundingwasreceivedasfollows:VC(TheNorwegianHealth Association),SEP(TheResearchCouncilofNorway,grant274352).

Generalfundingwasreceived fromtheSimonFougnerHartmann FamilyFund(rewardedtoTEMandSEP).

Data statement

Allanonymiseddataisavailableuponrequestfromthecorresponding author.

Conflict of interests

None.

Acknowledgments

Wesincerelythankparamedics/physiciansatOslo-AkershusEMS, andnurses,physicians,andlaboratorypersonnelinvolvedinOHCA treatmentatOUHU.

Appendix A. Supplementary data

Supplementarymaterialrelatedtothisarticlecanbefound,inthe online version, at doi:https://doi.org/10.1016/j.resuscitation.2021.

05.038.

REFERENCES

1.GräsnerJT,WnentJ,HerlitzJ,etal.Survivalafterout-of-hospital cardiacarrestinEurope—resultsoftheEuReCaTWOstudy.

Resuscitation2020;148:218 26.

2.NolanJP,NeumarRW,AdrieC,etal.Post-cardiacarrestsyndrome:

epidemiology,pathophysiology,treatment,andprognostication.A ScientificStatementfromtheInternationalLiaisonCommitteeon Resuscitation;theAmericanHeartAssociationEmergency CardiovascularCareCommittee;theCouncilonCardiovascular SurgeryandAnesthesia;theCouncilonCardiopulmonary, Perioperative,andCriticalCare;theCouncilonClinicalCardiology;

theCouncilonStroke.Resuscitation2008;79:350 79.

3.KalogerisT,BainesCP,KrenzM,KorthuisRJ.Cellbiologyof ischemia/reperfusioninjury.IntRevCellMolBiol2012;229 317 Elsevier.

4.LaverS,FarrowC,TurnerD,NolanJ.Modeofdeathafteradmissionto anintensivecareunitfollowingcardiacarrest.IntensiveCareMed 2004;30:2126 8.

5.Bro-JeppesenJ,JohanssonPI,KjaergaardJ,etal.Levelofsystemic inflammationandendothelialinjuryisassociatedwithcardiovascular dysfunctionandvasopressorsupportinpost-cardiacarrestpatients.

6.SekhonMS,AinsliePN,GriesdaleDE.Clinicalpathophysiologyof hypoxicischemicbraininjuryaftercardiacarrest:a"two-hit"model.

CritCare2017;21:90.

7.MentzelopoulosSD,ZakynthinosSG.Post-cardiacarrestsyndrome:

pathologicalprocesses,biomarkersandvasopressorsupport,and potentialtherapeutictargets.Resuscitation2017;121:A12-a4.

8.BottigerBW,MotschJ,BraunV,MartinE,KirschfinkM.Marked activationofcomplementandleukocytesandanincreaseinthe concentrationsofsolubleendothelialadhesionmoleculesduring cardiopulmonaryresuscitationandearlyreperfusionaftercardiac arrestinhumans.CritCareMed2002;30:2473 80.

9.NakstadER,Staer-JensenH,WimmerH,etal.Lateawakening, prognosticfactorsandlong-termoutcomeinout-of-hospitalcardiac arrest—resultsoftheprospectiveNorwegianCardio-Respiratory ArrestStudy(NORCAST).Resuscitation2020;149:170 9.

10.BergsethG,LudviksenJK,KirschfinkM,GiclasPC,NilssonB,Mollnes TE.Aninternationalserumstandardforapplicationinassaystodetect humancomplementactivationproducts.MolImmunol2013;56:232 9.

11.GarredP,MollnesTE,LeaT.Quantificationinenzyme-linked immunosorbentassayofaC3neoepitopeexpressedonactivated humancomplementfactorC3.ScandJImmunol1988;27:329 35.

12.MollnesTE,LeaT,HarboeM,TschoppJ.Monoclonalantibodies recognizinganeoantigenofpoly(C9)detectthehumanterminal complementcomplexintissueandplasma.ScandJImmunol 1985;22:183 95.

13.PerkinsGD,JacobsIG,NadkarniVM,etal.Cardiacarrestand cardiopulmonaryresuscitationoutcomereports:updateoftheUtstein

(8)

ResuscitationRegistryTemplatesforOut-of-HospitalCardiacArrest:

astatementforhealthcareprofessionalsfromataskforceofthe InternationalLiaisonCommitteeonResuscitation(AmericanHeart Association,EuropeanResuscitationCouncil,AustralianandNew ZealandCouncilonResuscitation,HeartandStrokeFoundationof Canada,InterAmericanHeartFoundation,ResuscitationCouncilof SouthernAfrica,ResuscitationCouncilofAsia);andtheAmerican HeartAssociationEmergencyCardiovascularCareCommitteeand theCouncilonCardiopulmonary,CriticalCare,Perioperativeand Resuscitation.Circulation2015;132:1286 300.

14.BöttigerBW,MotschJ,BraunV,MartinE,KirschfinkM.Marked activation of complement and leukocytes and an increase intheconcentrationsofsolubleendothelialadhesion

moleculesduringcardiopulmonaryresuscitationandearly reperfusionaftercardiacarrestinhumans.CritCareMed 2002;30:2473 80.

15.RicklinD,ReisES,LambrisJD.Complementindisease:adefence systemturningoffensive.NatRevNephrol2016;12:383 401.

16.HaugaardSF,JeppesenAN,TroldborgA,KirkegaardH,ThielS,AM Hvas.Thecomplementlectinpathwayaftercardiacarrest.ScandJ Immunol2018;88:e12680.

17.BisschopsLL,vanderHoevenJG,MollnesTE,HoedemaekersCW.

Seventy-twohoursofmildhypothermiaaftercardiacarrestis associatedwithaloweredinflammatoryresponseduringrewarmingin aprospectiveobservationalstudy.CritCare2014;18:546.

18.Bro-JeppesenJ,JeppesenAN,HaugaardS,etal.Thecomplement lectinpathwayproteinMAp19andout-of-hospitalcardiacarrest:

Insightsfromtworandomizedclinicaltrials.EurHeartJAcute CardiovascCare2020;9:S145 52.

19.OrremHL,NilssonPH,PischkeSE,etal.Acuteheartfailurefollowing myocardialinfarction:complementactivationcorrelateswiththe severityofheartfailureinpatientsdevelopingcardiogenicshock.ESC HeartFail2018;5:292 301.

20.PischkeSE,GustavsenA,OrremHL,etal.Complementfactor5 blockadereducesporcinemyocardialinfarctionsizeandimproves immediatecardiacfunction.BasicResCardiol2017;112:20.

21.WuW,ChopraA,ZieglerC,McLeodSL,LinS.Predictivevalueof hospitaldischargeneurologicaloutcomescoresforlong-term neurologicalstatusfollowingout-of-hospitalcardiacarrest:a systematicreview.Resuscitation2020;151:139 44.

22.Bro-JeppesenJ,JohanssonPI,HassagerC,etal.Endothelial activation/injuryandassociationswithseverityofpost-cardiacarrest

syndromeandmortalityafterout-of-hospitalcardiacarrest.

23.Arroyo-EspligueroR,AvanzasP,JefferyS,KaskiJC.CD14andtoll- likereceptor4:alinkbetweeninfectionandacutecoronaryevents?

Heart(BritishCardiacSociety)2004;90:983 8.

24.MarcosV,LatzinP,HectorA,etal.Expression,regulationandclinical significanceofsolubleandmembraneCD14receptorsinpediatric inflammatorylungdiseases.RespirRes2010;11:32.

25.YangQ,HeG-W,UnderwoodMJ,YuC-M.Cellularandmolecular mechanismsofendothelialischemia/reperfusioninjury:perspectives andimplicationsforpostischemicmyocardialprotection.AmJTransl Res2016;8:765.

26.KarpmanD,StåhlA-l,ArvidssonI,etal.Complementinteractionswith bloodcells,endothelialcellsandmicrovesiclesinthromboticand inflammatoryconditions.ImmuneResponsestoBiosurfaces2015;19

42Springer.

27.GrundmannS,FinkK,RabadzhievaL,etal.Perturbationofthe endothelialglycocalyxinpostcardiacarrestsyndrome.Resuscitation 2012;83:715 20.

28.Bro-JeppesenJ,KjaergaardJ,WanscherM,etal.Theinflammatory responseafterout-of-hospitalcardiacarrestisnotmodifiedby targetedtemperaturemanagementat33Cor36C.Resuscitation 2014;85:1480 7.

29.JohanssonPI,Bro-JeppesenJ,KjaergaardJ,WanscherM,Hassager C,OstrowskiSR.Sympathoadrenalactivationandendothelial damageareintercorrelatedandpredictincreasedmortalityinpatients resuscitatedafterout-of-hospitalcardiacarrest.ApostHocsub-study ofpatientsfromtheTTM-trial.PLoSOne2015;10:e0120914.

30.SchumacherA,SeljeflotI,SommervollL,ChristensenB,OtterstadJE, ArnesenH.Increasedlevelsofmarkersofvascularinflammationin patientswithcoronaryheartdisease.ScandJClinLabInvest 2002;62:59 68.

31.Barratt-DueA,PischkeSE,NilssonPH,EspevikT,MollnesTE.Dual inhibitionofcomplementandToll-likereceptorsasanovelapproachto treatinflammatorydiseases-C3orC5emergetogetherwithCD14as promisingtargets.JLeukocBiol2017;101:193 204.

32.Barratt-DueA,PischkeSE,BrekkeOL,etal.Brideandgroomin systemicinflammation–thebellsringforcomplementandTollin cooperation.Immunobiology2012;217:1047 56.

33.NolanJP,BergRA,CallawayCW,etal.Thepresentandfutureof cardiacarrestcare:internationalexpertsreachouttocaregiversand healthcareauthorities.IntensiveCareMed2018;44:823 32.