Biomarker prognostication of cognitive impairment may be feasible even in out-of hospital cardical arrest survivors with good neurological outcome

(1)

Clinical paper

Biomarker prognostication of cognitive impairment may be feasible even in out-of hospital cardical arrest survivors with good neurological outcome

Kolbjørn Brønnick

^a,e,

*, Lars Evald

^b

, Christophe Henri Valdemar Duez

^c

, Anders Morten Grejs

^g

, Anni Nørgaard Jeppesen

^c

, Hans Kirkegaard

^c

, Jørgen Feldbæk Nielsen

^b

, Eldar Søreide

^d,f

aDepartmentofPublicHealth,UniversityofStavanger,Stavanger,Norway

bHammelNeurorehabilitationCentreandUniversityResearchClinic,Hammel,Denmark

cResearchCenterforEmergencyMedicine,EmergencyDepartmentandDepartmentofClinicalMedicine,AarhusUniversityHospitalandAarhus University,Aarhus,Denmark

dCriticalCareandAnaesthesiologyResearchGroup,StavangerUniversityHospital,Stavanger,Norway

eCentreforAge-RelatedMedicine(SESAM),HelseStavanger,Stavanger,Norway

fDepartmentofClinicalMedicine,UniversityofBergen,Bergen,Norway

gDepartmentofIntensiveCareMedicine,AarhusUniversityHospital,Aarhus,Denmark

Abstract

Background:Patientssurvivingout-ofhospitalcardicacarrest,withgoodneurologicaloutcomeaccordingtoCerebralPerformanceCategory, frequentlyhaveneuropsychologicalimpairment.Westudiedwhetherbiomarkerdata(S-100bandneuron-specificenolase)obtainedduringtheICU staypredictedcognitiveimpairment6monthsafterresuscitation.

Methods:Patients(N=79)withaCPC-score2wererecruitedfromtwotrialsitestakingpartintheTTH48trialcomparingtargetedtemperature management(TTM)for48hvs.24hat331C.Weassessedpatients6monthsaftertheOHCA.WemeasuredbiomarkersS-100bandNSEatarrival andat24,48and72hafterreachingthetargettemperatureof331C.

Fourcognitivedomainz-scoreswerecalculated,andglobalcognitiveimpairmentwasdefinedasz< 1.67onatleast3outof13cognitivetests.Non- parametriccorrelationswereusedtoassesstherelationshipbetweencognitivedomainandbiomarkers.ROCcurveswereusedtoassesspredictionof cognitiveimpairmentfromthebiomarkers.LogisticregressionwasusedtoinvestigatewhetherTTMdurationmoderatedbiomarkerpredictionof cognitiveimpairment.

Results:Cognitiveimpairmentwaspresentin22%ofthepatientswithmemoryimpairmentbeingthemostcommon.Thebiomarkerscorrelated significantlywithseveralcognitivedomainscoresandNSEat48hpredictedcognitiveimpairmentwith100%sensitivityand56%specificity.The predictivepropertiesofNSEat48hwasunaffectedbydurationofTTM.

Conclusions:EarlybiomarkerprognosticationofcognitiveimpairmentisfeasibleeveninOHCAsurvivorswithgoodneurologicaloutcomeasdefined byCPC.NSEat48hpredictedcognitiveimpairment.

Keywords:Neurologicaloutcome,Outofhospitalcardiacarrest,Neuropsychology,Hypothermictreatment

* Correspondingauthorat:UniversityofStavanger,P.box8600,4036Stavanger,Norway.

E-mailaddress:[email protected](K.Brønnick).

https://doi.org/10.1016/j.resuscitation.2021.02.025

by/4.0/).ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).

Availableonlineatwww.sciencedirect.com

Resuscitation

j o urna lh ome p a ge :w ww . e l se v i e r . com / l oca t e / r e sus ci t a ti o n

(2)

Introduction

InEurope,mostsurvivorsofout-of-hospitalcardiacarrest(OHCA) havea goodneurological outcome, oftenexpressed as Cerebral PerformanceCategory(CPC)of1or2.¹Nevertheless,about50%of theseOHCAsurvivorssufferfromcognitivesequela.²Ascognitive deficitsmayseverelyaffectfunctionalcapacityandrenderreturnto workandself-sufficiencyimpossible,³predictionoffuturecognitive impairment is clinically important as it may help planning the rehabilitationprocess.

Moststudiesonneurologicalprognosisintheearlyphaseafter OHCAhave employedcrudemeasuresof functioning suchas theCPC⁴ whichisproblematicregardingsensitivity,asconsiderablecognitive impairmentmaybepresenteveninpatientswithgoodCPCscoresof

<=2.⁵Eventhoughformalneuropsychologicaltestsshouldideallybe employedinordertomeasurecognitiveoutcomeafterOHCA,thisis often notfeasible. Hence, thereis aneed forearly predictors of neuropsychologicaloutcomethatcanbeemployedregardlessofother clinicalvariablesandthatarerelativelyindependentfromTTM.Two such predictors are the biomarkers S-100b and neuron-specific enolase(NSE.⁶S-100bisacalciumbindingproteinthatisfoundin Schwann-cellsandglialcellsandneuronspecificenolase(NSE)is foundinsideneuronsandplaysaroleinaxonaltransport.⁶Survivalas wellassuperficiallyandsubjectivelyevaluatedneurologicaloutcome canbepredictedfromS-100bandNSEinpatientsundergoingTTM.^7,8 However,thesestudiesonNSEandS100busingCPCasoutcome mainly demonstrated prediction of survival, as very few surviving patientshadaCPCscore>2.Further,manypatientswithCPC2may sufferfrom significant cognitive sequelae that are detected when sensitiveneuropsychologicaltestsareused.^5,9

Thus, we analyzed S-100b and NSE regarding prediction of cognitive impairment, using neuropsychological data from asub- studyoftheRCT“TTH48”comparingprolongedtargetedtemperature management(TTM)for48hvs.standard24hat331C,andwhere wefoundthatprolongedTTMwasassociatedwithimprovedmemor.⁹ Theresearchquestionswere whethercognitive impairmentin OHCAsurvivorssixmonthsafterresuscitationcanbepredictedfrom NSEandS-100bobtainedduringtheirICUstayandwhetherlengthof TTMmoderatesthisprediction.Wefirstexploredcognitiveimpairment prevalence as well as cognitive domain impairment. Finally, we evaluatedwhichcognitivedomainsweremoststronglyassociated withNSEandS-100bsixmonthsafterOHCAandwhetherlengthof TTMmoderatedsuchassociations.

Methods

Thisstudyisaposthocfollow-up-studybasedontheTTH48trial,an investigator-initiated,blinded-outcome-assessor,parallelcontrolled multicentertrial inwhichpatientswith OHCAwererandomizedto receiveTTMat331Cforeither24or48h.^10,11

Patients

ThesamplewaspatientsrecruitedfromtheTTH48trialwithaCPC score2. They consented to a follow-up neuropsychological assessmentsixmonthsfollowingresuscitationfromOHCA.Wehave previously describedthe present sample,⁹ hencewe only briefly summarizethestudysamplinghere.

The TTH48 trial enrolled 355 patients, and 159 patients participatedin thesub-studyattheintensivecare units(ICUs)at Aarhus University Hospital, Denmark and Stavanger University Hospital,Norway.EveryunconsciousOHCApatientwasscreened foreligibilityatICUadmission.Alegalsurrogateandarelativegave writteninformedconsentbeforerandomizationofthepatientto24or 48hTTM.Theinclusioncriteriawere:Ageover17andbelow80years, OHCA with a presumed cardiac origin, sustained (>20min) spontaneouscirculationafterresuscitation,GlasgowComaScore¹² (GCS)<8onadmission.Detailedinclusionandexclusioncriteriacan be foundin theoriginal studyprotocol.¹³ Included patients were recruitedforneuropsychologicalassessmentsixmonthspostOHCA.

Thisresultedintheinclusionof79patientsasdescribedinTable1.

ThestudywasapprovedbytheDanishDataProtectionAgency andtheCentralDenmarkRegionCommitteesonHealthResearch Ethics(casenumber20110022)andtheRegionalEthicsCommittee ofWesternNorway(ref2013/1486).

Measures

S-100bandNSE

BloodsamplesformeasuringS-100bandNSEweredrawnupon patientadmission,24h,48hand72hafterthepatientreachedthe targettemperature.Detailsofthelabprocedureshavebeenpublished previously.⁸Measurementrangeswere.3 740

m

^g/l^for^NSE^and^.02

39

m

^g/l^for^S-100b.

Neuropsychologicalvariables

SixmonthsafterOHCA,testsofmemory,attention,andexecutive functions were administered by trained and assessor-blinded research assistants. The following tests were performed: Rey Auditory-VerbalLearningTest(RAVLT),¹²Rey OsterreithComplex FigureTest (ROCFT),^13,14 WAIS-IVDigitSpanandVocabulary,¹⁵ Trail MakingTest A & B (TMT-A& B),^16,17 and D-KEFSVerbal Fluency.¹⁸NormativedatafortheWAIS-IVandD-KEFSsubscales wereobtainedfromthetestmanuals.NormativedataforTMT-A&B, RAVLT and ROCFT were obtained from Mitrushina¹⁹ based on regressionequationsderivedbymeta-analysisofmultiplenormative datasets. Thus, cognitive raw-scores were converted to z-scores basedonnormativedataandadjustedforageasestimatedfromthe norms.Thecut-offlimitforimpairmentbasedonthez-scoreswasset atz<= 1.67asthestandardizedscaledscoresofWAIS-IVandD- KEFSsubscalesjumpinone-thirdSDincrementsandthisz-value mostcloselyresemblesthefifthpercentile(4.75%).Sinceweincluded 13 different test scores in assessing cognitive impairment, this increasestheriskoffalsepositivecases(type1error).Weassessed thenumberofobservedsignificantfindingsandtookintoaccountthe likelihoodofasinglescorebelowcut-offoccurringbychance,e.g.the probability ofone ormore significant findings in 13 comparisons (correspondingtop=.49),twoormorefindings(p=.14),threeormore (p=.03),andsoforth.Consequently,patientswithperformancebelow thecut-offscoreonthreeormorecognitivetestswereconsidered cognitivelyimpaired(correspondingtop<.02).

Further,inordertoinvestigateimpairmentindifferentcognitive domains,wecalculatedaveragez-scoresin4domainsandvalidated thedomainscalesusingCronbachalphaasameasureofinternal consistency: Episodic memory domain (RAVLT learning, recall, recognition and RCFT immediate and delayed recall;

a

⁼^.748),

Workingmemorydomain(Digitspanforwards,backwards,sequenc- ingandRAVLTtrial1;

a

⁼^.765),^Verbal^fluency^domain^(Category

(3)

fluency,semantic,phonological,categoryshift;

a

⁼^.840)^and^Visuo-

motordomain(RCFTcopy,TMTA,TMTB;

a

⁼^.680).^Impairmentⁱⁿ

eachdomainwasdefinedasanaveragez-score<= 1.67.

Statisticalmethods

S-100bandNSEwereseverelyright-skewed,hencenon-parametric statisticswerechosenthroughout.Continuousdataarepresentedas mediansandinterquartilerangesandcategoricaldataascountsand percentages.

CategoricaldatawereanalyzedusingFisher’sexacttest,assome ofthecell-countswerebelow5.Continuousdatawereanalyzedusing Mann WhitneyUforgroupcomparisons,andSpearmanrhonon- parametriccorrelationsforassessingrelationshipbetweenvariables.

InordertoinvestigatethepredictivevaluesofS-100bandNSE,we performednon-parametricROC-analyses,reportingareasunderthe curveaswellasoptimalcut-offpointswithregardtosensitivityand specificity,choosingthecut-offwiththemaximumYouden'sindex value²⁰andadditionallyshowingtherangeofvaluesfromasensitivity of100%toaspecificityof100%.

Finally,inordertoassesswhetherdurationofTTMaffectedthe predictivepropertiesofS-100bandNSE,sequentiallogisticregression analyseswereperformedwiththebinaryvariableindicatingcognitive impairmentasdependentvariable.Inthefirstblock,thebiomarkerwas enteredaspredictorandinthesecondblock,thevariableindicating TTMconditionwasentered.Finally,inthethirdblock,theinteraction termbetweenTTMconditionandthebiomarkerwasentered.

AlldatawereanalyzedusingSPSS¹Version25forWindows¹.

Results

ThesamplederivedfromtheTTH48trialisthesameaswehave reportedpreviously⁹anddetailsonpatientrecruitmentcanbefound

there.InTable1,wedescribethepatientsenrolledintheTTM24vs TTM48 conditions as well as the total sample with complete neuropsychological data. The Consort flow-chart is included as supplementaryfigureS1.Inaddition,thenumberofpatientswithNSE and/orS100bdataateachmeasurementtimeisshowninTable1.

TheprevalenceofcognitiveimpairmentwashigherintheTTM24 group.ThedifferenceinNSE48hvalueswasnotsignificantaccording toaBonferrroni-correctedalphalimitof.005.

PrognosticationofcognitiveimpairmentamongOHCA survivors

InFig.1wepresentthes100bvaluesatarrivaltothehospital,andat 24,48and72hafterthetargettemperatureof33+ 1Cwasreached.

s100bwaselevatedatadmittance,butrapidlydeclined.

AsseeninTable1,s100bwassignificantly(p=.012)higherin TTM48 than inTTM24 at72h, evenafter applyingaBonferroni- correctedalphalimitofp<.0125(4comparisons).

InFig.2,weshowtheNSEvaluesatarrivaltothehospital,andat 24, 48 and 72h after the target temperature of 33+ 1C was reached.NSEwaselevatedintheimpairedgroupascomparedto thenon-impairedgroup,mostpronouncedintheTTM24groupat 48h.

InTable2,weshowthenon-parametriccorrelationcoefficients betweentheneuropsychologicaldomainscoresandS100b/NSE.

NSEat48hcovariedwiththecognitivedomainscoresandthe effectwasstrongestforverbalfluencyandvisuo-motorperformance.

Noneof theNSEscores correlatedsignificantlywith thememory domainscores,whereastheS100bscoreat48hdid.

InFig.3,weshowROC-curvesforS100BandNSEwithregardto cognitiveimpairment(yes/no)aspresentedinthegroupsinTable1.

TheresultsoftheROC-analysesareshownindetailinTable3.

None of the S100b variables were significant predictors of cognitive impairment. However, NSE at 48h was a predictor of Table1–Descriptivestatistics.

TTM24 TTM48 p Total

N(Male/Female) 36(34/2) 43(37/6) =.280 79(71/8)

Cognitivelyimpaired(%) 12(33%) 5(12%) =.028 17(22%)

Median(IQR) Median(IQR) Median(IQR)

Age 59(10) 60(19) =.672 60.00(15.00)

ROSCtime 22(15) 17(12) =.969 18.50(13.00)

Glasgowcomascalearrival* 3(0) 3(0) =.113 3(0)

Glasgowcomasvare64h** 14(6) 11(11) =.108 14(9)

GlasgowcomascaleICU*** 14.5(1) 14.0(1) =.636 14.0(1)

S100barrival(N=54) .75(1.43) .90(1.89) =.332 .88(1.43)

S100b24h(N=59) .89(.78) .72(.67) =.813 .83(.06)

S100b48h(N=58) .88(.13) .73(.47) =.777 .82(.06)

S100b72h(N=56) .45(.65) .08(.06) =.012 .75(.08)

NSEarrival(N=54) 17.67(16.96) 22.22(23.26) =.164 20.23(18.26)

NSE24h(N=59) 8.33(6.41) 10.51(6.47) =.192 9.07(7.35)

NSE48h(N=57) 9.56(7.61) 7.96(8.11) =.872 9.31(7.89)

NSE72h(N=55) 7.27(7.84) 7.59(5.85) =.301 7.43(6.56)

IQR:Interquartilerange;ROSCtime:Timefromcardiacarresttoreturnofspontaneouscirculation.

S100bandNSEunits:m^g/l.

*Threepatientsscored>3(Scores:4/6/7).

**Measured64hafterarrival,thus8hafterendofhypothermiaintheTTM48group.

***MeasuredatICUdischarge.

(4)

cognitiveimpairmentatsixmonths.Theoptimalcut-offwasanNSE valueof8.41,whichresultedinasensitivityof100%andaspecificityof 56%.NSEvaluesof13.68,18.87and29.31resultedinsensitivity/

specificityof50/83,30/97and20/100respectively.

To investigate whether the TTM intervention at 48 vs 24h moderated thepredictive value of NSEat 48h, we performed a sequentiallogisticregressionanalysiswithNSEat48haspredictor andcognitiveimpairmentasdependentvariableinblock1(Oddsratio:

1.122,p=.019).Inblock2,weenteredallocationtoTTM24orTTM48 treatmentandbothNSE48(Oddsratio:1.154,p=.014)andtreatment condition(Oddsratio:1.113,p=.012)predictedcognitiveimpairment.

Inblock3,theinteractionbetweenNSE48andtreatmentwasnon- significant (p=.485). Thus, NSE based prediction at 48h was independentoflengthofTTM.

Discussion

NSEat48hwasapredictorofcognitiveimpairmentsixmonthsafter out-of-hospitalcardiacarrestamongpatientswithseeminglygood outcome (CPC2) and thepredictive valuewas independent of lengthofTTM.NSEat48correlatedwithverbalfluencymeasuresand visuo-motormeasuresandNSEat24hcorrelatedwithverbalfluency measures.Further,S100bat48hcorrelatedwithmemoryandverbal fluencyandS-100batarrivalalsocorrelatedwithverbalfluencysix monthslater.Thus,S-100bwasalsorelatedtocognitiveoutcome althoughitwasnotapredictorofglobalcognitiveimpairment.

ThecognitivedomainmostaffectedbyOHCAwasmemory,ashas beenfoundinmostotherstudies,²andthepresentstudyisconsistent

Fig.1–Boxplotsofs100bvaluesforimpairedvsnon-impairedsubjectsat4differenttimes.

(5)

Table2–CorrelationsbetweentheneuropsychologicaldomainscoresandS100b/NSE.

Neuropsychologicaldomainzscore Arrival 24h 48h 72h

Memorydomain S100b .229 .158 .276* .151

NSE .164 .002 .239 .126

Verbalfluencydomain S100b .341* .179 .309* .054

NSE .215 .265* .381** .107

Visuo-motordomain S100b .139 .003 .140 .249

NSE .236 .133 .349** .190

Workingmemorydomain S100b .030 .056 .053 .194

NSE .233 .185 .242 .101

*p<.05.

**p<.01.

Fig.2–BoxplotsofNSEvaluesforimpairedvsnon-impairedsubjectsat4differenttimes.

(6)

withearlierresearchwhichhasshownmemorydeficitsinpatientswith CPC<=2.⁵

Theoptimalcut-offinthepresentstudywas8.41

m

^g/L^for^NSE^at

48h,withasensitivityof100%andaspecificityof56%.Thisisalower cut-offthanthecut-offfoundbyChoietal.⁷andDuezetal.⁸These previouslyidentifiedcut-offsarequitedifferentinthattheyarebased onalladmittedpatientsofwhichforinstanceonly2patientswith“poor” outcomesurvivedintheDuezetal.study.Thus,thehighcut-offis primarilyapredictorofsurvival,notaprognosticfactoramongthose whosurvived.Thelowcut-offonthepresentstudyshouldhoweverbe interpretedasawithin-survivorpredictorofcognitiveoutcomeand futurestudiesshouldexploreandvalidatethecut-offs.

Themainlimitationsinthepresentstudyarerelatedtothefactthat thiswasapost-hocstudybasedontheTTH48trialwhichfocusedon theeffects ofprolongedTTM.Thus,thepresentsampleishighly selectedwithagoodneurologicaloutcome,definedasCPC2and willingnesstocometothetest.Thisleadstorestrictionofrangeand weakensthepredictivepropertiesofthebiomarkerswithregardto cognitiveoutcome.Further,asmentionedabove,wedidnothave informationconcerningpremorbidcognitivefunctioningandthismay haveattenuatedthepredictionpropertiesofthebiomarkers.Finally, oursamplewassmall,alsolimitingthegeneralizabilityofthestudy.

Thus,theseresultsshouldbeviewedastentativeandexploratoryand futurestudiesshouldbedonetovalidatethefindings.

Thisis thefirst studywhichhas demonstrated theprognostic propertiesofS-100bandNSEinOHCAsurvivorswithCPC2using acomprehensiveneuropsychologicalbatteryandwhichalsohave showedthatlengthofTTMdidnotaffecttheseprognosticproperties.

Thus, the study is an important first step, both clinically and theoretically,inestablishingprognosticationofcognitiveimpairments amongOHCAsurvivors.

Futurestudiesshouldreplicatetheseresultsandestablishtwo differentcut-offscoresforNSEandS100Bforclinicallymeaningful prognostication.Ahighcut-offscoreforprognosticationofsurvival andalowerforprognosticationofintactcognitiveoutcome.Patients withbiomarkerlevelsbetweenthosetwocutofflevelswouldmost likelysurvivewith cognitiveimpairment andhencehaveneedfor specialattentionandsupport.

Conclusions

Biomarker prognostication of cognitive impairment may to some extentbe feasibleamongOHCAsurvivorswith goodoutcomeas definedbyCPC2.Manyofthesepatientswillsufferfromcognitive impairment.NSEat48hpredictedoverallcognitiveimpairmentwith anoptimalcut-offat8.41

m

^g/L^resultingⁱⁿsensitivityof100%and specificityof56%.

Fig.3–ROC-curvesforS100BandNSEwithcognitiveimpairmentaspredictedvariable.

Table3–AreaUndertheCurve:S100b/NSE.

TestResultVariable(s) Area Std.Error^a p^b 95%ConfidenceInterval

LowerBound UpperBound

Arrival S100b .342 .096 .128 .154 .530

NSE .494 .099 .958 .300 .689

24h S100b .478 .105 .829 .273 .683

NSE .700 .090 .055 .524 .876

48h S100b .513 .116 .899 .285 .741

NSE .789 .070 .006 .652 .926

72h S100b .318 .108 .080 .108 .529

NSE .619 .089 .252 .445 .794

Thetestresultvariable(s):S100bat24h,S100bat48h,S100bat72hhasatleastonetiebetweenthepositiveactualstategroupandthenegativeactualstate group.Statisticsmaybebiased.

aUnderthenonparametricassumption.

bNullhypothesis:truearea=.5.

(7)

Conflicts of interest

None.

CRediT authorship contribution statement

KolbjørnBrønnick:Conceptualization,Methodology,Datacuration, Writing - original draft, Writing - review & editing. Lars Evald:

Methodology,Datacuration,Writing-review&editing.Christophe HenriValdemarDuez:Methodology,Datacuration,Writing-review

&editing.AndersMortenGrejs:Methodology,Writing-review&

editing.AnniNørgaardJeppesen:Methodology,Writing-review&

editing.HansKirkegaard:Conceptualization,Methodology,Writing- review & editing. Jørgen Feldbæk Nielsen: Conceptualization, Methodology,Writing-review&editing.EldarSøreide:Conceptuali- zation,Methodology,Writing-review&editing.

Acknowledgements

Wewishto thanktheresearchassistants attheHammel Neuro- rehabilitationandResearchCentreandtheDepartmentofPsychiatry attheStavangerUniversityHospital.

Appendix A. Supplementary data

Supplementarymaterialrelatedtothisarticlecanbefound,intheonline version,atdoi:https://doi.org/10.1016/j.resuscitation.2021.02.025.

REFERENCES

1.KirkegaardH,TacconeFS,SkrifvarsM,SoreideE.Postresuscitation careafterout-of-hospitalcardiacarrest:clinicalupdateandfocuson targetedtemperaturemanagement.Anesthesiology2019;131:186

208.

2.MoulaertVR,VerbuntJA,vanHeugtenCM,WadeDT.Cognitive impairmentsinsurvivorsofout-of-hospitalcardiacarrest:asystematic review.Resuscitation2009;80:297 305.

3.HofgrenC,Lundgren-NilssonA,EsbjornssonE,SunnerhagenKS.

Twoyearsaftercardiacarrest;cognitivestatus,ADLfunctionandliving situation.BrainInj2008;22:972 8.

4.JennettB,BondM.Assessmentofoutcomeafterseverebrain damage.Lancet1975;1:480 4.

5.SulzgruberP,KliegelA,WandallerC,etal.Survivorsofcardiacarrest withgoodneurologicaloutcomeshowconsiderableimpairmentsof memoryfunctioning.Resuscitation2015;88:120 5.

6.GulSS,HuesgenKW,WangKK,MarkK,TyndallJA.Prognosticutility ofneuroinjurybiomarkersinpostout-of-hospitalcardiacarrest (OHCA)patientmanagement.MedHypotheses2017;105:34 47.

7.ChoiS,ParkK,RyuS,KangT,KimH,ChoS,etal.UseofS-100B, NSE,CRPandESRtopredictneurologicaloutcomesinpatientswith returnofspontaneouscirculationandtreatedwithhypothermia.Emerg MedJ2016;33:690 5.

8.DuezCHV,GrejsAM,JeppesenAN,SchroderAD,SoreideE,Nielsen JF,etal.Neuron-specificenolaseandS-100binprolongedtargeted temperaturemanagementaftercardiacarrest:arandomisedstudy.

Resuscitation2018;122:79 86.

9.EvaldL,BronnickK,DuezCHV,etal.Prolongedtargetedtemperature managementreducesmemoryretrievaldeficitssixmonthspost- cardiacarrest:arandomisedcontrolledtrial.Resuscitation2019;134:1

9.

10.KirkegaardH,RasmussenBS,deHaasI,etal.Time-differentiated targettemperaturemanagementafterout-of-hospitalcardiacarrest:a multicentre,randomised,parallel-group,assessor-blindedclinicaltrial (theTTH48trial):studyprotocolforarandomisedcontrolledtrial.Trials 2016;17:228.

11.KirkegaardH,SoreideE,deHaasI,etal.Targetedtemperature managementfor48vs24hoursandneurologicoutcomeafterout-of- hospitalcardiacarrest:arandomizedclinicaltrial.JAMA2017;318:341

50.

12.ReyA.Mémorisationd’unesériede15motsen5répetitions.Paris, France:PressesUniversitairesdesFrance;1958.

13.OsterreithPA.Letestdecopied’unefigurecomplex:contributiona l’étudedelaperceptionetdelamémoire.ArchPsychol1944;30:286

356.

14.ReyA.L’examenpsychologiquedanslescasd’encéphalopathie traumatique.(Lesproblems.).ArchPsychol1941;28:286 340.

15.WechslerD.WechslerAdultIntelligenceScaleI.V.SanAntonio,TX, USA:Pearson,ThePsychologicalCorporation;2008.

16.ReitanRM.Therelationofthetrailmakingtesttoorganicbrain damage.JConsultingPsychol1955;19:393 4.

17.ReitanRM.ValidityoftheTrailmakingtestasanindicatoroforganic braindamage.PerceptualMotorSkills1958;8:271 6.

18.DelisDC,KaplanE,KramerJH.Delis-Kaplanexecutivefunction system(D-KEFS)technicalmanual.SanAntonio,TX,USA:Pearson, ThePsychologicalCorporation;2001.

19.MitrushinaMN.Handbookofnormativedataforneuropsychological assessment.2.ed.NewYork:OxfordUniversityPress;2005.

20.BewickV,CheekL,BallJ.Statisticsreview13:receiveroperating characteristiccurves.CritCare2004;8:508 12.