GDF-15 is associated with sudden cardiac death due to incident myocardial infarction

http://www.diva-portal.org

This is the published version of a paper published in Resuscitation.

Citation for the original published paper (version of record):

Andersson, J., Fall, T., Delicano, R., Wennberg, P., Jansson, J-H. (2020)

GDF-15 is associated with sudden cardiac death due to incident myocardial infarction

Resuscitation, 152: 165-169

https://doi.org/10.1016/j.resuscitation.2020.05.001

Access to the published version may require subscription.

N.B. When citing this work, cite the original published paper.

Permanent link to this version:

Clinical

paper

GDF-15

is

associated

with

sudden

cardiac

death

due

to

incident

myocardial

infarction

Jonas

Andersson

*

,

Tove

Fall

,

Rachel

Delicano

,

Patrik

Wennberg

,

Jan-Ha˚kan

Jansson

a_{DepartmentofPublicHealthandClinicalMedicine,Skelleftea˚ ResearchUnit,Umea˚ University,Sweden} b

UppsalaUniversityandScienceforLifeLaboratory,UppsalaUniversity,Uppsala,Sweden

c

DepartmentofPublicHealthandClinicalMedicine,FamilyMedicine,Umea˚ University,Sweden

Abstract

Aims:Preventingsuddencardiacdeath(SCD)duetoacutemyocardialinfarction(MI)inpreviouslyhealthypatientsischallenging.Proteomicanalysis mayleadtoanunderstandingofbiologicalmechanismsandprovidepredictivebiomarkers.

Methods:Inthisprospective,nestedcase-controlstudyfromnorthernSweden,87candidatecardiovascularproteinbiomarkerswerestudiedin244 individualswholaterdiedwithin24hfromanincidentMIand244referentswithoutMIandindividuallymatchedforage,sexanddateofhealth examinationandaliveatthedateofeventintheindexperson.Associationanalysiswasconductedusingconditionallogisticregression.Bonferroni correctionwasappliedtoavoidfalsepositivefindings.

Results:TenproteinswereassociatedwithfutureSCDduetoacuteMIinthenon-adjustedanalysis.Thestrongestassociationwerefoundforgrowth differentiationfactor15(GDF-15)withanoddsratio(OR)of1.79(95%confidenceinterval[CI]1.41,2.25)perstandarddeviationincreaseinprotein,and urokinase-typeplasminogenactivatorreceptorwithanORof1.66(95%CI1.34,2.06).Inmodelsadjustedforlipidlevels,bodymassindex,education, smoking,hypertensionandC-reactiveprotein,onlyassociationwithGDF-15remained(OR1.47(95%1.11,1.95)).

Conclusion:ElevatedlevelsofGDF-15areassociatedwithincreasedriskofSCDwithin24hofincidentMI.Furtherresearchmayenabletheuseof GDF-15togetherwithotherclinicalandbiologicalmarkerstoguideprimarypreventiveinterventionsforindividualsathighriskforSCD.

Keywords:Suddencardiacdeath,Myocardialinfarction,Proteomics,GDF-15

Introduction

Cardiovasculardisease(CVD)accountsforone-thirdofalldeaths globallyandisthesingleleadingcauseofpremature mortality.1,2

About50%ofallheart-relateddeathsiscausedbysuddencardiac death(SCD)triggeredbycoronaryheartdiseaseevents.3Innorthern Sweden the yearlyincidence of SCD in individuals aged 35 64 withoutpreviousCVDis12per100,000forwomenand65formen,4

andtheincidencerateincreasesrapidlywithhigherage.5_Prevention

ischallengingasSCDmightbethefirstclinicalmanifestationofCVD andbecausetheunderlyingetiologyisnotfullyunderstood.“Among

individuals without known heart disease many suffers from a concealedischemicheartdisease.Whenthisisnotthecasegenetic andotherstillunknowncausespredisposeindividualstoSCD.”Risk factor profiling is proposed to be a feasible strategy to identify individualswithincreasedriskforSCD.

Wepreviouslyreportedthattype2diabetesmellitusaswellasa highbodymassindex(BMI)predictaneightfoldriskforSCDdueto acutemyocardialinfarction(MI).6However,cardiovascularriskfactor screeningmethodsstillneedstobeimproved,7,8_{particularytoidentify}

high-riskindividuals.Recenttechnicaladvancementshaveenabled proteomicplasmaprofilinginlargestudiesandpavingthepathforthe discoveryofnovelbiomarkersandpathwaysleadingtoSCD.

* Correspondingauthorat:Skellefteålasarett,93186Skellefteå,Sweden. E-mailaddress:jonas.so.andersson@regionvasterbotten.se(J.Andersson). https://doi.org/10.1016/j.resuscitation.2020.05.001

Received16January2020;Receivedinrevisedform9April2020;Accepted1May2020

0300-9572/©2020TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/ by/4.0/).ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).

Available

online

at

www.sciencedirect.com

Resuscitation

Theaimofthisstudywastoevaluatetheassociationofalargeset ofcardiovascularcandidatebiomarkerswithSCDduetoacuteMI usinganestedcase-referencedesigninapopulation-basedsettingin northernSweden.

Methods

Thestudysamplewas derivedfromtheVästerbottenIntervention Program(VIP),9_{ahealthinterventionprogramforpreventionofCVD}

and the World Health Organization's Multinational Monitoring of TrendsandDeterminantsinCardiovascularDisease(MONICA),a population survey innorthern Sweden.10 Participationrates were about59%and77%,respectively.Thedatacollectionandregistration inVIPandMONICAaresimilarandaredescribedindetailinRefs.9,10

AllacuteMIeventsinnorthernSwedenareevaluatedandregistered intheMONICAregistry,accordingtostandardizedWHOcriteriaand based on reports from general practitioners as well as hospital dischargerecordsanddeathcertificates.10

Inthisprospectivenestedcase-controlstudy,weidentifiedcases offirsteveracuteMIinVIPandMONICAoccurringbetween1986and 2006throughtheMONICAregistry.Individualsincludedare25 64 yearsandfrom2000,individuals_74yearshavealsobeenincluded. MIcasesthatdiedwithin24hfromonsetofsymptomswereclassified as SCD due to MI. Subjects with ahistory of MI before health examinationwereexcluded,asexperiencingaMImayinfluencethe studieddeterminants.Plasmawasavailablefrom244caseswithSCD duetoMI.OnereferentwithoutMIforeachcase,aliveatthetimeof SCDintheindexperson,matchedbysex,birthyear(2years),year ofhealthexaminationandtypeofhealthexamination,wereselected fromthebiobank.ThestudywasapprovedbytheResearchEthics Committee of Umeå University. All participants gave informed consent.

Baselinevariables

Smoking habits were classified into “ever smoking” (including previous smokers and occasional smokers) or _{“never-smoking”.} BMIwascalculatedasweight(kg)/squareheight(m2).Hypertension wasdefinedasasystolicbloodpressure140mmHgoradiastolic blood pressure 90mmHg or reported use of anti-hypertensive medicationduringthelast14days.Educationallevelwas dichotom-isedinto9yearscompulsoryeducationorhigher.

Bloodsamplesweretakenafteraminimum4-hourfastandwere analyzedfortotalcholesterolandplasmaglucoseusingabench-top analyser (Reflotron1; Boehringer Mannheim GmbH Diagnostica, Mannheim,Germany).Since2005,aHemoCuebench-topanalyser (Quest Diagnostics) has been used for glucosevalues. Anoral glucosetolerancetestwasperformedwitha75goralglucoseload accordingtoWHOstandards.Diabeteswasdefinedasself-reported diseasefromthequestionnaireorafastingglucose7mmol/Land/ or2-hpost-loadplasmaglucose11.0mmol/L(12.2mmol/Linthe VIP, ascapillary plasma was drawn). Biochemical analysis was performedontheCobas8000modularmultianalyserusing Tina-quantkitsforA1(APOAT)andB(APOBT),aswellasc-reactive protein (CRP)-kit (CRPL3) from the same manufacturer (Roche DiagnosticsGmbH, Mannheim,Germany).Plasmasampleswere obtainedafteraminimumof4hfast(extendedto8h1992),and

storedinadeep-freezerat 80Cuntilanalyses.Allmeasurements were made by laboratory staff, blinded to participants’ disease status.

Proteomicprofiling

PlasmasampleswereanalyzedusingtheProximityExtensionAssay technique11_{ontheOlinkMultiplexCVDIIIpanel,ahigh-specificity}

assaythatsimultaneouslymeasuresconcentrationsof92 cardiovas-cularcandidateproteins.Inbrief,theassayusesastandard96-well microplate format including four quality control standards. Each sampleismixedwith92pairsofoligonucleotide-labeledantibodies.If bothhigh-specificityantibodiesbindthetargetprotein,theattached oligonucleotides form a unique DNA reporter sequence that is subsequently amplifiedandquantifiedbystandardPCR.Samples wereanalyzedinindividualwellsoneightplates,keepingeachsetof case and referent together on the same plate. Fluorescence detection-thresholdPCRvalueswerelog2-transformedandcorrected fortechnicalvariationbynegativeandinter-platecontrols.11_Lower

limitsofdetection(LOD)wasdeterminedthroughnegativecontrol samples. Quality control included the removal of five proteins (SPON1,NTproBNP,EPHB4,PCSK9andPSPD)with>15%missing values.Oneindividualwith<95%ofmeasurementsbelowLODinthe remaining87proteinswereexcluded.Intheremainingdata,values belowtheLODwereimputedtoLOD/2(intotal107replacementsout of 42,891 data points). Because principal component analysis indicated associations of protein measurements with plate and storagetime,weusedstandardizedresidualsfromlinearregression modelsadjustedforplateandstoragetimeasindependentvariables inthisstudy.

Statisticalanalysis

WeusedSTATA14.1forallstatisticalanalyses.Asafirststep,a seriesofnon-adjustedconditionallogisticregressionswereusedto estimatetheassociationofeachproteinwithcasestatususingthe clogitcommandinSTATA.WeusedaBonferronicorrectedpvalueof 0.05/87tests=5.7e 4todefinestatisticalsignificance.

For the proteinsassociated with case status ata Bonferroni-adjustedp-valuethreshold,weperformedasetofadjustedconditional logistic regression models. For many variables, especially CRP, apolipoprotein A (ApoA), and ApoB1, there were many missing observations.Therefore,weappliedthefollowingapproach.First,we used thecomplete-case approach that excluded individuals with missingcovariatedataandadjustedforhypertensionstatus,smoking habits,diabetesmellitus,educationlevel,fastingstatus,BMIandtotal cholesterol.Wefurtherusedmultipleimputationbychainedequation with 20 iterationsto impute missingvalues in diabetes mellitus, ApoA1,ApoB,education,BMIandCRPbasedoninformationinthese variablesandhypertension,smokinghabits,ageandsex.Thus,the imputationincludedsomeofthematchingvariablesbutexcludedthe identifier of matched pairs, consistent with the method “Multiple imputation using matching variables” described by Seaman and Keogh.12ThenumberofimputedvaluesisshowninSupplementary Table1.Conditionallogisticregressionmodelsadjustingcoefficients andstandarderrorsforthevariabilitybetweenimputationswasdone accordingtothecombinationrulesbyRubin.13_{Thesemodelswere}

runforeachoftheproteinspassingstep1adjustingforhypertension status, smokinghabits, diabetes mellitus, educationlevel, fasting status,BMI,totalcholesterol,ApoA1,ApoBandCRP.

Results

ThemeanageatSCDwas55yearsand78%weremen(Table1).In thefirststep,10proteinswereassociatedwithfutureSCDinthe non-adjustedanalysis.TheseassociationsarereportedinSupplementary Table2. Nineproteinswereassociated with increasedriskwhile concentrationsofparaoxonase3(PON3)wasassociatedwithlower risk.ThestrongestassociationswerefoundforGDF-15,withanodds ratio(OR)perstandarddeviation(SD)increaseinproteinof1.79(95% CI1.41,2.25)andurokinaseplasminogenactivatorsurfacereceptor (U-PAR)1.66(95%CI1.34,2.06).Infullyadjustedmodelsusingthe completecaseapproach,estimateswereattenuated,indicatingthat the association was confounded or mediated by the included covariates(hypertensionstatus,smokinghabits,diabetesmellitus, educationlevel,fastingstatus,BMI,andtotalcholesterol)althoughthe associationofGDF-15andU-PARwithSCDwasstillsignificant.In

modelsadditionallyadjustedforApoA1,ApoB andCRP,onlythe associationofGDF-15wasstillsignificant(OR1.47,95%CI1.11, 1.95).TheresultsareshowninTable2.

Discussion

Ourmainfindingwasidentificationofthestrongassociationof GDF-15 independently with SCD. We further identified nine other associated proteins, but their association was dependent on cardiovascularriskfactors.

GDF-15 was identified in 1997 as a macrophage derived inflammatoryresponsecytokine.14_{Itwaslaterfoundtobeinduced}

in themyocardium in response to ischemia.15 _{Since then it has}

becomerecognizedasariskmarkerforcardiovascularandall-cause mortality,16 20andin acutecoronarysyndromes asapredictorof prognosis.21 27

Table1_–Descriptionofthe244caseswithmyocardialinfarctionandsuddencardiacdeathand244referents.

Caseswithdata Mean(SD)/n(%) Referentswithdata Mean(SD)/n(%)

Ageatsampling 244 54.7(7.0) 244 54.9(7.1)

Ageatevent 244 63.2(7.3) NA NA

Female 244 54(22.1%) 244 54(22.1%)

Previousoroccasionalsmokers 244 174(71.3%) 244 135(55.3%)

Bodymassindex 242 28.1(4.6) 239 26.0(3.9)

Triglycerides 201 2.0(1.2) 187 1.6(0.9) Totalcholesterol 235 6.5(1.3) 234 6.1(1.1) Hdlcholesterol 105 1.5(3.1) 86 1.3(0.4) Apoa1 173 1.4(0.2) 175 1.4(0.3) Apob 174 1.3(0.3) 175 1.2(0.3) Diabetesmellitus 244 85(34.8%) 241 44(18.3%) Fastingglucose 207 6.1(2.5) 214 5.5(1.0)

2-h-postOGTTglucose 186 7.0(2.7) 213 6.6(1.8)

C-reactiveprotein 172 3.1(6.0) 171 2.1(4.7)

Hypertension 244 158(64.8%) 244 108(44.3%)

Systolicbloodpressure 238 142.3(18.4) 240 133.9(16.9)

Diastolicbloodpressure 238 88.4(9.1) 240 84.9(8.9)

Secondaryeducation 227 100(44.1%) 230 138(60.0%)

Table2_–Multivariable-adjustedmodelsforthoseproteinsthatwereassociatedwithsuddencardiacdeathinthe discoveryanalysis.

Protein Protein,fullname Non-adjusted,n=488 Adjusteda_{,n=426 Imputed,adjusted}b_,n=488

OR(95%CI) p OR(95%CI) p OR(95%CI) p

GDF15 Growth/differentiationfactor15 1.79(1.41,2.25) 1.1E 06 1.39(1.05,1.84) 0.02 1.47(1.11,1.95) 0.01 UPAR Urokinaseplasminogenactivatorsurfacereceptor 1.66(1.34,2.06) 4.0E 06 1.35(1.02,1.78) 0.04 1.32(0.99,1.75) 0.05 PON3 Paraoxonase 0.60(0.48,0.75) 7.2E 06 0.76(0.56,1.02) 0.07 0.79(0.58,1.07) 0.12 LDLreceptor Low-densitylipoproteinreceptor 1.59(1.30,1.95) 8.2E 06 1.25(0.95,1.63) 0.11 1.16(0.88,1.51) 0.29 OPG Osteoprotegerin 1.54(1.24,1.90) 6.8E 05 1.16(0.90,1.50) 0.26 1.22(0.93,1.60) 0.14 FABP4 Fattyacid-bindingprotein,adipocyte 1.56(1.25,1.95) 6.9E 05 1.14(0.84,1.55) 0.41 1.08(0.80,1.46) 0.61 RARRES2 Retinoicacidreceptorresponderprotein2 1.50(1.22,1.84) 1.3E 04 1.13(0.88,1.45) 0.35 1.05(0.81,1.35) 0.73 tPA Tissue-typeplasminogenactivator 1.47(1.21,1.80) 1.4E 04 1.09(0.83,1.42) 0.55 1.03(0.79,1.34) 0.84 CSTB Cystatin-B 1.48(1.20,1.81) 1.8E 04 1.17(0.91,1.51) 0.22 1.13(0.89,1.45) 0.31 CTSD CathepsinD 1.43(1.18,1.75) 3.6E 04 1.03(0.81,1.32) 0.81 1.01(0.79,1.30) 0.92 a

Adjustedforhypertensionstatus,smokinghabits,diabetesmellitus,educationlevel,fastingstatus,bodymassindexandtotalcholesterol.

b_{Imputedcovariates,adjustedforhypertensionstatus,smokinghabits,diabetesmellitus,educationlevel,fastingstatus,bodymassindex,totalcholesterol,} ApoA1,ApoBandC-reactiveprotein.

GDF-15isinvolvedintheinflammatorypathwaybutincontrast totherapidriseandfallofnatriureticpeptides,cardiactroponinand CRPduringCVDeventsitseemsmorestableandmayreflecta chronic CVD burden.28 It isreasonabletoassumethat GDF-15 identifiesindividualswithsubclinicalcardiovasculardiseaseatrisk forSCDfromanacutemyocardialinfarction,explainingourresults. Our study adds to the potential use of GDF-15 for SCD risk prediction,asthisisthefirststudytoshowGDF-15asariskmarker for SCD among previously healthy individuals. However, the discriminative abilityand clinicalutility ofGDF-15 in thatsetting needstobedeterminedinfuturestudies.Intheabsenceofprevious cardiovascular events, primary prevention is the only possible measurementtoreduceSCD.Thus,improvedmethodstoidentify individuals with high risk are essential and GDF-15 may be a suitablecandidateriskmarker.Furtherresearchmayleadtothe useofGDF-15,togetherwithotherclinicalandbiologicalmarkers, toguideprimarypreventiveinterventionsforindividualswithhigh riskforSCD.Currently,implantablecardioverterdefibrillatorsare usedtoreduceSCDamongpatientswithheartfailure.Theuseof GDF-15 may better identify patientsthat canbenefit from such treatment and beyond present indications. GDF-15 may also becomeatooltoevaluateprimarypreventivemeasuresthataimto reducetheriskofSCD.

Usingasinglemarkerstrategyinindividualriskpredictioninthe generalpopulationisnotreasonablebasedontheheterogenicityof SCD.29,30GDF-15itselfcanonlyexplainalimitedriskincreasefor SCD,anditisunlikelythatanyothersinglebiomarkerwillhaveenough powertoidentifyasatisfactorynumberofhigh-riskindividualsina generalpopulation.Furtherstudiesareneeded,andaconstellationof biomarkersandotherriskfactorscouldbethebasisforariskscore. U-PARhasalsobeenidentifiedasaninflammatorymarkerfrom studiesmainlyincancer31andkidneydisease.32 35IntheMalmöDiet andCancerStudyU-PARwasassociatedwithincreasedincidenceof CVDinelderly.36_{Wecouldnotreproducetheirresultsinourfully}

adjustedmodel.Thediscrepancycanbeexplainedinseveralways including the differently adjusted models, thefact that U-PAR is associatedwith CVD butnotSCD,andthat theMalmöDiet and CancerStudypopulationwasabout10yearsolder.

Ourfindingsaremostlikelyexplainedbyatherosclerotic mecha-nismsinvolvingGDP-15andU-PAR.BothGDF-15andU-PARare involved in inflammatory processes which in turn are linked to atherosclerosis.Ahigherdegreeofinflammationcouldsuggestmore aggressivediseaseprogressionexplainingtheassociationofGDF-15 withSCDeventsamongMIcasesinourstudy.

Ourfindingsneedconfirmationinanotherpopulationtoverifythe results.Also,itremainsunknownifGDF-15isonlyariskmarkerorifit isacausalfactorandiftherapiestargetingGDF-15canlowertherisk ofSCD.Additional studiessuchasMendelianrandomization and randomizedcontrolledtrialsloweringGDF-15areneededtoanswer suchquestions.

Ourstudyhadsomeuniquestrengths.AmajorityofSCDoccurs outofthehospital.Wewereabletoincludepatientswhodiedoutside ofthehospitalduetocardiacarrestevenwhennoresuscitationwas attempted,andweonlyincludedincidentMIs.Secondaryprevention in cases with previous CVD may otherwise have altered the concentrationsofthestudiedproteins.Themethodalsoallowedus toanalyzeproteinssampledbeforetheevent.

Thestudyalsohadsomelimitations.Almost80%ofthecaseswere male,ascanbeexpected,whichlimitedthepossibilityforsex-specific analysis.The participants in this studywere mainly middle-aged

CaucasiansfromnorthernSwedenandtheresultscouldbedifferentin otherpopulations.Moreover,theproteinpanelonlyincludedalimited setofproteins,andotherproteinsmaybeoflargerimportance.The findingsshouldbeconfirmedinfurtherstudies,preferablyinanother population.

Conclusion

Elevated levels ofGDF-15areassociated with increased riskof SCDwithin24hofincidentMI.Furtherresearchmayenabletheuse of GDF-15 togetherwithotherclinical andbiological markersto guideprimarypreventiveinterventionsforindividualsathighriskfor SCD.

Authors’

contributions

JonasAnderssoncontributedtoacquisition,analysis,or interpreta-tion,draftedmanuscript,criticallyrevisedthemanuscript,gavefinal approval,agreestobeaccountableforallaspectsofworkensuring itegrityandaccuracy.

ToveFallcontributed toacquisition,analysis,orinterpretation, drafted manuscript, critically revised the manuscript, gave final approval,agreestobeaccountableforallaspectsofworkensuring itegrityandaccuracy.

RachelDelicanocontributedtoacquisition,analysis,or interpre-tation,criticallyrevisedthemanuscript,gavefinalapproval,agrees to be accountable for all aspects of work ensuring itegrity and accuracy.

PatrikWennbergcontributedtoacquisition,analysis,or interpre-tation,criticallyrevisedthemanuscript,gavefinalapproval,agreesto beaccountableforallaspectsofworkensuringitegrityandaccuracy. Jan-HåkanJanssoncontributedtoconceptionordesign, contrib-utedtoacquisition,analysis,orinterpretation,criticallyrevisedthe manuscript, gavefinal approval,agreesto be accountable forall aspectsofworkensuringitegrityandaccuracy.

Funding

ThisworkwassupportedbyERCStartingGrant(801965),aproject grantfromSwedishHeartandLungfoundation(20150429)andthe Swedish Research Council (2015-03477). The project was also supportedbytheSwedishHeartandLungfoundation(20160275).

Conflict

of

intererst

Nonedeclared.

Acknowledgements

Dr.FallreportsgrantsfromERCStartingGrant(801965),aproject grantfromSwedishHeartandLungfoundation(20150429)andthe SwedishResearchCouncil(2015 03477),duringtheconductofthe study.

Dr. Jansson reports grants from Swedish Heart and Lung foundation(20160275),duringtheconductofthestudy.

Appendix

A.

Supplementary

data

Supplementarydataassociatedwiththisarticlecanbefound,inthe onlineversion,athttps://doi.org/10.1016/j.resuscitation.2020.05.001.

REFERENCES

1.CollaboratorsGMaCoD.Global,regional,andnationallife

expectancy,all-causemortality,andcause-specificmortalityfor249 causesofdeath,1980 2015:asystematicanalysisfortheglobal burdenofdiseasestudy2015.Lancet2016;388:1459 544. 2.JosephP,LeongD,McKeeM,etal.Reducingtheglobalburdenof

cardiovasculardisease.Part1:theepidemiologyandriskfactors.Circ Res2017;121:677 94.

3.MyerburgRJ.Suddencardiacdeath:exploringthelimitsofour knowledge.JCardiovascElectrophysiol2001;12:369 81. 4.MessnerT,LundbergV.Trendsinsuddencardiacdeathinthe

northernSwedenMONICAarea1985 99.JInternMed2003;253:320 8.

5.ZipesDP,CammAJ,BorggrefeM.ACC/AHA/ESC2006guidelinesfor themanagementofpatientswithventriculararrhythmiasandthe preventionofsuddencardiacdeath:areportoftheAmericanCollege OfCardiology/AmericanHeartAssociationtaskforceandthe EuropeanSocietyOfCardiologyCommitteeforpracticeguidelines (writingcommitteetodevelopguidelinesformanagementofpatients withventriculararrhythmiasandthepreventionofsuddencardiac death).JAmCollCardiol2006;48:e247.

6.AnderssonJ,WennbergP,LundbladD,EscherSA,JanssonJH. Diabetesmellitus,highBMIandloweducationlevelpredictsudden cardiacdeathwithin24hoursofincidentmyocardialinfarction.EurJ PrevCardiol2016;17:1814 20.

7.MyerburgRJ,MitraniRD,InterianA,CastellanosA.Interpretationof outcomesofantiarrhythmicclinicaltrials:designfeaturesand populationimpact.Circulation1998;97:1514 21.

8.MyerburgRJ,CastellanosA.Emergingparadigmsoftheepidemiology anddemographicsofsuddencardiacarrest.HeartRhythm 2006;3:235 9.

9.NorbergM,WallS,BomanK,WeinehallL.TheVästerbotten InterventionProgramme:background,designandimplications.Glob HealthAction2010;3.

10.StegmayrB,LundbergV,AsplundK.Theeventsregistrationand surveyproceduresintheNorthernSwedenMONICAProject.ScandJ PublicHealthSuppl2003;9 17.

11.LundbergM,ErikssonA,TranB,AssarssonE,FredrikssonS. Homogeneousantibody-basedproximityextensionassaysprovide sensitiveandspecificdetectionoflow-abundantproteinsinhuman blood.NucleicAcidsRes2011;39:e102.

12.SeamanSR,KeoghRH.Handlingmissingdatainmatched case-controlstudiesusingmultipleimputation.Biometrics2015;71:1150

9.

13.RubinDB.Multipleimputationfornonresponseinsurveys.NewYork: Wiley;1987.

14.BootcovMR,BauskinAR,ValenzuelaSM,etal.MIC-1,anovel macrophageinhibitorycytokine,isadivergentmemberofthe TGF-betasuperfamily.ProcNatlAcadSciUSA1997;94:11514 9. 15.KempfT,EdenM,StrelauJ,etal.Thetransforminggrowthfactor-beta

superfamilymembergrowth-differentiationfactor-15protectsthe heartfromischemia/reperfusioninjury.CircRes2006;98:351 60. 16.RohatgiA,PatelP,DasSR,etal.Associationofgrowthdifferentiation

factor-15withcoronaryatherosclerosisandmortalityinayoung, multiethnicpopulation:observationsfromtheDallasHeartStudy.Clin Chem2012;58:172 82.

17.WallentinL,ZetheliusB,BerglundL,etal.GDF-15forprognostication ofcardiovascularandcancermorbidityandmortalityinmen.PLOS ONE2013;8:e78797.

18.DanielsLB,CloptonP,LaughlinGA,MaiselAS,Barrett-ConnorE. Growth-differentiationfactor-15isarobust,independentpredictorof 11-yearmortalityriskincommunity-dwellingolderadults:theRancho BernardoStudy.Circulation2011;123:2102 10.

19.WangTJ,WollertKC,LarsonMG,etal.Prognosticutilityofnovel biomarkersofcardiovascularstress:theFraminghamHeartStudy. Circulation2012;126:1596 604.

20.JenniferEH,AsyaL,PaulC,etal.Proteinbiomarkersof

cardiovasculardiseaseandmortalityinthecommunity.JAmHeart Assoc2018;7:e008108.

21.SkauE,HenriksenE,WagnerP,HedbergP,SiegbahnA,LeppertJ. GDF-15andTRAIL-R2arepowerfulpredictorsoflong-termmortality inpatientswithacutemyocardialinfarction.EurJPrevCardiol 2017;24:1576 83.

22.BonacaMP,MorrowDA,BraunwaldE,etal.Growthdifferentiation factor-15andriskofrecurrenteventsinpatientsstabilizedafteracute coronarysyndrome:observationsfromPROVEITTIMI22.Arterioscler ThrombVascBiol2011;31:203 10.

23.KhanSQ,NgK,DhillonO,etal.Growthdifferentiationfactor-15asa prognosticmarkerinpatientswithacutemyocardialinfarction.Eur HeartJ2009;30:1057 65.

24.WollertKC,KempfT,LagerqvistB,etal.Growthdifferentiationfactor 15forriskstratificationandselectionofaninvasivetreatmentstrategy innonST-elevationacutecoronarysyndrome.Circulation

2007;116:1540 8.

25.KempfT,SinningJM,QuintA,etal.Growth-differentiationfactor-15for riskstratificationinpatientswithstableandunstablecoronaryheart disease:resultsfromtheAtheroGenestudy.CircCardiovascGenet 2009;2:286 92.

26.KempfT,BjörklundE,OlofssonS,etal.Growth-differentiation factor-15improvesriskstratificationinST-segmentelevationmyocardial infarction.EurHeartJ2007;28:2858 65.

27.WollertKC,KempfT,PeterT,etal.Prognosticvalueof growth-differentiationfactor-15inpatientswithnon-ST-elevationacute coronarysyndrome.Circulation2007;115:962 71.

28.WollertKC,KempfT,WallentinL.Growthdifferentiationfactor15asa biomarkerincardiovasculardisease.ClinChem2017;63:140 51. 29.HertzCL,Ferrero-MilianiL,Frank-HansenR,MorlingN,BundgaardH.

Acomparisonofgeneticfindingsinsuddencardiacdeathvictimsand cardiacpatients:theimportanceofphenotypicclassification. Europace2015;17:350 7.

30.StattinEL,WestinIM,CederquistK,etal.Geneticscreeninginsudden cardiacdeathintheyoungcansavefuturelives.IntJLegalMed 2016;130:59 66.

31.LiuKL,FanJH,WuJ.PrognosticroleofcirculatingsolubleuPARin variouscancers:asystematicreviewandmeta-analysis.ClinLab 2017;63:871 80.

32.TummalapalliL,NadkarniGN,CocaSG.Biomarkersforpredicting outcomesinchronickidneydisease.CurrOpinNephrolHypertens 2016;25:480 6.

33.HahmE,PeevV,ReiserJ.Extrarenaldeterminantsofkidneyfilter function.CellTissueRes2017;369:211 6.

34.DandeRR,PeevV,AltintasMM,ReiserJ.Solubleurokinasereceptor andthekidneyresponseindiabetesmellitus.JDiabetesRes 2017;2017:3232848.

35.ZeierM,ReiserJ.suPARandchronickidneydisease-apodocyte story.PflugersArch2017;469:1017 20.

36.PerssonM,EngströmG,BjörkbackaH,HedbladB.Solubleurokinase plasminogenactivatorreceptorinplasmaisassociatedwithincidence ofCVD.ResultsfromtheMalmöDietandCancerStudy.