Inflammation of left atrial epicardial adipose tissue is associated with paroxysmal atrial fibrillation

(1)

Inflammation of left atrial epicardial adipose tissue is associated with paroxysmal atrial fibrillation

著者草山隆志

著者別表示 Kusayama Takashi journal or

publication title

博士論文本文Full 学位授与番号 13301甲第4444号

学位名博士（医学）

学位授与年月日 2016‑06‑30

URL http://hdl.handle.net/2297/46454

doi: 10.1016/j.jjcc.2015.11.005

Creative Commons : 表示 ‑ 非営利 ‑ 改変禁止 http://creativecommons.org/licenses/by‑nc‑nd/3.0/deed.ja

(2)

Original article

Inﬂammation of left atrial epicardial adipose tissue is associated with paroxysmal atrial ﬁbrillation

Takashi Kusayama (MD), Hiroshi Furusho (MD, PhD)*, Haruka Kashiwagi (MD), Takeshi Kato (MD, PhD), Hisayoshi Murai (MD, PhD), Soichiro Usui (MD, PhD), Shuichi Kaneko (MD, PhD), Masayuki Takamura (MD, PhD)

DepartmentofDiseaseControlandHomeostasis,KanazawaUniversityGraduateSchoolofMedicalScience,Kanazawa,Ishikawa,Japan

Introduction

Atrialﬁbrillation(AF)isacommonlyencounteredarrhythmia thatisassociatedwithhighmorbidityandmortality[1].Obesityas deﬁnedbybody massindex(BMI)is oneoftheriskfactorsfor developingAF,asareadvancedage,diabetes,hypertension,and coronaryarterydisease(CAD)[2–5].Ectopicfatdepositsinmuscle, liver,andthepancreascorrelatewithinsulinresistanceandCAD [4]. Epicardial adipose tissue (EAT) is ectopic visceral fat that surroundstheheart,andincreasedEATvolumescorrelatewiththe prevalenceofAF,aswellaswiththerecurrenceofAFafterablation therapy[6,7].Increasedfatthicknessattheposteriorleftatrium (LA)maybeparticularlyrelatedtotheAFburden,independentof age,BMI,andtheLAarea[8,9].Nevertheless,themechanismsofAF pathogenesisandtheformationsofAFsubstrateduetoEATarenot wellunderstood.

Chronicinflammationisnotonlyassociatedwiththepresence ofAF,butalsopredictspatientsatincreasedriskofdevelopingAF [10].InCAD,theextentofEATaccumulationisassociatedwith diseaseseverity,andperivascularadipocytesareknowntosecrete inflammatorycytokines[4,11–14].Thepresenceofinflammation in pericardial fat results in a higher density in computed tomography(CT)imagesandiscorrelatedwithculpritlesionsin patients withCAD[15].However,theassociationbetween EAT- associatedinflammationandAFremainsunclear.

This study investigated the hypothesis that increased EAT densityaroundtheLAasdeterminedbyCTisassociatedwiththe presenceofparoxysmalAF(PAF).

Methods Studygroup

Thestudypopulationconsistedof64patientswhounderwent cardiacCT.Intotal,51consecutivepatientsunderwenttheirﬁrst catheterablationforAFwithcardiacCTbetweenJanuary2011and September 2013 at our hospital; 19 of these patients were ARTICLE INFO

Articlehistory:

Received1July2015

Receivedinrevisedform20October2015 Accepted4November2015

Availableonlinexxx

Keywords:

Inﬂammation

Epicardialadiposetissue Leftatrium

Paroxysmalatrialﬁbrillation Computedtomography

ABSTRACT

Background: Althoughanincreasedepicardialadiposetissue(EAT)volumearoundtheleftatrium(LA)is related to the atrial fibrillation (AF) burden, the role of EAT inflammation in AF is unclear. We investigatedtheassociationbetweenAFandinflammationoftheEATaroundtheLA.

Methods:WeretrospectivelyidentiﬁedregionsofEATaroundtheLAandmeasuredthedensityofthese areasusingcomputedtomography(CT).

Results:Atotalof32patientswhounderwenttheirfirstcatheterablationforparoxysmalAF(PAF)were enrolled(meanage62.511.1years).PatientswithoutahistoryofAF(n=32),butwhounderwentcardiac CT and were matched by age, sex, and metabolic risk factors, were enrolled in the control group (62.212.1years).ThemeanEATdensityaroundtheLAwassignificantlyhigherinthePAFgroupthaninthe controlgroup( 108.16.7vs. 111.65.5Hounsfieldunits;p=0.02),whilethedensitiesofsubcutaneous adiposetissue(SAT)intheabdomenandthoraxdidnotdifferbetweenthetwogroups.Inamultiplelogistic regressionanalysis,ahigherEATdensitywassignificantlyassociatedwiththepresenceofPAFafteradjusting forotherriskfactors(oddsratio:1.25;95%confidenceinterval:1.08–1.45,p=0.003).

Conclusions: ThisstudysupportsthehypothesisthatinﬂammationofEATaroundtheLA,butnotSAT,is relatedtothepresenceofPAF.

ß2015JapaneseCollegeofCardiology.PublishedbyElsevierLtd.Allrightsreserved.

* Correspondingauthorat:DepartmentofDiseaseControlandHomeostasis, KanazawaUniversity Graduate School of Medical Science, 13-1 Takaramachi, Kanazawa,Ishikawa920-8641,Japan.Tel.:+81762652233;fax:+81762344250.

E-mailaddress:[email protected](H.Furusho).

ContentslistsavailableatScienceDirect

Journal of Cardiology

j our na l ho me pa g e : w ww . e l se v i e r . com / l oca t e / j j cc

http://dx.doi.org/10.1016/j.jjcc.2015.11.005

0914-5087/ß2015JapaneseCollegeofCardiology.PublishedbyElsevierLtd.Allrightsreserved.

(3)

excluded,leaving 32 patients in the PAF group. The exclusion criteriawereasfollows:persistentorlong-lastingpersistentAF, moderate or severe valvular disease, left ventricular ejection fraction (LVEF)<55%, BMI>30kg/m², heart failure, B-type natriuretic peptide level>200pg/mL, history of open heart surgery, and evidence of active infection or inﬂammation.

PersistentAFwasdeﬁnedasAFlasting>7days,andlong-lasting persistentAFwasAFlasting>1year.

Anadditional347consecutivepatientsunderwentcardiacCTto examinetheir coronary arteryregions between April2011 and March2012.Ofthese,32patientswithoutadocumentedhistoryof AF and who were matched for age, sex, BMI, the presence of hypertension,diabetesmellitus,and dyslipidemia withthePAF groupwereenrolledinthecontrolgroup.

This study was approved by our local institutional review boardsandethicscommittees,andwasperformedinaccordance withinstitutionalpolicies, nationallegal requirements, and the revised Helsinki declaration. Written informed consent was obtainedfrom eachPAF patient. Thisstudyis registeredin the Universal Hospital Medical Information Network Clinical Trials Registry(UMIN000012606;http://www.umin.ac.jp/ctr).

CardiacCTprotocol

A 128-section DSCT SOMATOM Deﬁnition Flash scanner (SiemensHealthcare,Erlangen,Germany)wasusedforallpatients inthisstudy.Ifnecessary,thepatientsweregivenanintravenous

b

-blocker to reduce their heart rate. The raw data were reconstructed using algorithms optimized for electrocardio- gram-gated multi-slice spiral reconstruction. For contrast-en- hanced scans, 90mL of nonionic contrast agent (Optiray 310;

CovidienJapan,Tokyo,Japan)wereinjectedintravenouslyataﬂow rateof4.5mL/s.

CardiacCTimageanalysis

EATregionscontainingtheﬁvemajorLAganglionatedplexuses (GPs)wereidentiﬁed:superiorleft(SL),inferiorleft(IL),anterior right(AR),inferiorright(IR),andMarshalltract(MT).Theanatomic areasoftheGPswerepresumedbasedonpreviousreportsofthe autonomicinnervationoftheheart(Fig.1)[16,17].

Apredeﬁnedimagedisplaysettingwasused[windowwidth 195to 45Hounsﬁeldunits(HU),windowcenter 120HU]to identifypixelsthatcorrespondedtofattissue(Fig.1)[4].

Thecardiologistsinterpretingtheresultswereblindedtothe clinical results of the patients. They trimmed along the LA, pulmonaryveins,andleftatrialappendagesusingaxial,coronal, andsagittalslices.EATareasweredefinedasanyadiposetissue locatedwithin thepericardial sac. TheEAT densities in the CT imagesweremeasuredaroundtheLA usingplainCTimages in both groups. The cardiologists interpreting the results only referredtoenhancedCTimageswhentheyhaddifficultydetecting theanatomicallocationsoftheEAT.Themeanvalueofthefive regionsofinterestwasrecordedateachEAT.Theregionsofinterest weresetasfollows:circularshape,size5mm²,andthewhole regionwasfullycontainedineachlocalEATarea(Fig.2).Whenthe EATareacouldnotcontainfiveregionsinasingleaxiallevel,some regionsof interestwereset atupperor loweraxial levels.The densitiesofsubcutaneousadiposetissue(SAT)intheabdomenand thoraxweremeasuredsimilarly.

Assessmentofriskfactorsandcovariates

The following patient data were collected: the presence of hypertension, the presence of dyslipidemia, the presence of diabetes mellitus, high-sensitivity C-reactive protein (hs-CRP),

BMI,andcurrentmedication.Hypertensionwasdefinedasblood pressure>140/90mmHgonrepeatedmeasurements,orcurrent treatment with antihypertensive agents. Blood samples were obtainedatthetimeofcardiacCTtomeasuretheserumlevelsof theindicatedmarkers.Dyslipidemiawasdefinedaslow-density lipoproteincholesterol (LDL-C) levels>140mg/dL, high-density lipoprotein cholesterol (HDL-C) levels<40mg/dL, triglyceride levels>150mg/dL,or theuseofa statin.LDL-Cwasexamined directlyorwascalculatedusingtheFriedewaldmethod[18].Dia- betesmellituswasdefinedasafastingplasmaglucose>126mg/

dL,hemoglobinA1c>6.5%,orcurrenttreatmentwithhypoglyce- micagents.Echocardiographywasperformedtomeasurethesize of each cardiac chamber and to assess cardiac function at approximatelythesametimeasthecardiacCT.TheLAdiameter wasmeasured attheendofsystoleintheparasternallong-axis view, and LAenlargementwasdeﬁned asa diameter>42mm.

LVEF was measured using the M-mode or modiﬁed Simpson’s method;anLVEFof<55%wasconsideredabnormal.Moderateor highvalvular regurgitationwasdiagnosed usingechocardiogra- phy.

Statisticalanalyses

Categorical variables are presented as numbers (%), and continuousvariablesarepresentedasmeansstandarddeviations (SDs). Student’s t-test or the Mann–Whitney U-test was used to analyze differences between the two study groups in baseline continuousvariables,andthechi-squaretestwasusedfordichoto- mousvariables. Thevaluesforhs-CRPandtriglycerideswere log- transformedtoimprovenormality.Theresultswereﬁnallyexpressed ontheoriginalscale,afterexponentiation,asgeometricmeansand thecorrespondingasymmetric95%conﬁdenceintervals(CIs).

Thenon-LAEATdensitiesinthePAFgroupwerecalculatedby twooperatorsandintra-andinter-observerreliabilityfortheEAT densitymeasurementswasexaminedusingintra-andinter-class correlationcoefﬁcients,respectively,withtheir95%CIs.Addition- ally,inter-observeragreementswerealsoshownusingtheBland andAltmanplot.Themeandifferencewaspresentedasthebias and95%limitsofagreementaroundthebiasexpressedasthemean difference1.96SDs.

Multiplelogisticregressionanalysiswasusedtocomparethe associationbetweenthedensityoftheEATandPAFbetweenthe control and PAF groups. A value of p<0.05 was considered statisticallysigniﬁcant.Allanalyseswereperformed usingSPSS version19.0forWindows(SPSS;Chicago,IL,USA).

Results

Baselinecharacteristics

Between January 2011 and September 2013, 32 patients underwenttheir ﬁrstcatheterablationfor PAFwithcardiacCT, andwereincludedinthePAFgroup(meanage62.511.1years;

19males).ThemeandiseasedurationofPAFwas4.65.2years.In addition,32patientswereenrolledinthecontrolgroup(meanage 62.212.1years;19males).Thebaselinecharacteristicsofthetwo groupsareshowninTable1.ThePAFgrouphadalargerLAdiameter, andmorepatientsinthisgroupused

b

-blockers.

EATdensityandPAF

TheEATdensitywascalculatedbytwoindependentoperators.

Theintra-andinter-observercorrelationswere0.890and0.910, respectively(p<0.001).TheBlandandAltmanplotshowedagood agreementbetweentheobservers,withabiasof 0.757.25HU andthe95%limitsofagreementwas 14.97to13.47HU(Fig.3).

T.Kusayamaetal./JournalofCardiologyxxx(2015)xxx–xxx 2

(4)

ThemeandensityoftheEATscontainingallﬁveEATareas(LA- EAT)in thePAFgroupwassigniﬁcantlyhigherthanthatin the control group ( 108.16.7HU vs. 111.55.5HU, p=0.02;

Table2).In theregionalanalysis ofEAT,the densitiesoftheAR- EATandMT-EATweresigniﬁcantlyhigher,andthatoftheSL-EATand

IR-EATtendedtobehigher,inthePAFgroupthaninthecontrols.In comparison,thedensitiesofSATintheabdomenandthoraxdidnot differbetweenthetwogroups.Severalparameterswereincludedin themultiplelogisticregressionanalysis:age,sex,BMI,LAdiameter, and LA-EAT density. The density of LA-EAT was significantly Fig.1.Computedtomography(CT)imagesettingsandlocationoftheEATs.(A,C,andE)EnhancedCTimagesfromaparoxysmalatrialfibrillationpatient.(B,D,andF) PredefinedplainCTimagedisplaysettings(windowwidth 195to 45HU,windowcenter 120HU)wereusedtoidentifypixelscorrespondingtofattissue.Eachimagein B,D,andFshowsthesameaxiallevelwiththeimageofA,C,andErespectively.YellowarrowsindicatetypicalEATandthoracicSATregions.Ao,aorta;PA,pulmonaryartery;

SVC,superiorvenacava;LA,leftatrium;LAA,leftatrialappendage;RS,rightsuperior;LS,leftsuperior;RI,rightinferior;PV,pulmonaryvein;RA,rightatrium;RV,right ventricle;LV,leftventricle;EAT,epicardialadiposetissue;SAT,subcutaneousadiposetissue;MT,Marshalltract;AR,anteriorright;SL,superiorleft;IL,inferiorleft.

(5)

associatedwiththepresenceofPAF(oddsratio:1.25;95%CI:1.08–

1.45,p=0.003),aswellastheLAdiameterafteradjustingforAFrisk factors,includingage,sex,andBMI(Table3).

Discussion

Our resultsshowthat theEAT density aroundthe LA in CT imageswasassociatedwiththepresenceofPAF.ThedensityofLA- EATwassigniﬁcantlygreaterinthePAFgroupthaninthecontrols.

Furthermore,thisdifferencewassigniﬁcantlycorrelatedwiththe presenceof PAF, and was independent of other metabolic risk factors.

EATaroundtheLA

EAT is contiguous with cardiac structures; it overlies the right ventricle, coronary arteries, left ventricular apex, and atria, without forming an intervening fascia between these

Fig.2.RepresentativemeasurementofEATdensity.(A)Predeﬁnedplaincomputedtomographyimagefromaparoxysmalatrialﬁbrillationpatient,and(B)acontrolpatient.

FigureAshowsthemagniﬁedimageoftheareaencircledbydashedsquareinFig.1B.Amongthecolorscorrespondingtoadiposetissue,blueareaindicatesrelativelyhigher HU,andredareaindicateslowerHU.Whitecirclesindicateregionsofinterest(ROI)containedineachAR-EATarea.ThemeanvalueoftheﬁveROIswasrecordedateachEAT area.AR,anteriorright;EAT,epicardialadiposetissue.

Table1

Baselinepatientcharacteristics.

Control(n=32) PAF(n=32) p-Value

Age,years 62.212.1 62.511.1 0.91

Sex,male 19(59.4) 19(59.4) 1.00

BMI,kg/m² 23.12.8 23.32.7 0.79

LVEF,% 69.06.2 66.25.6 0.06

LAdiameter,mm 33.74.2 39.77.1 <0.01

Hypertension 17(53.1) 17(53.1) 1.00

Dyslipidemia 21(65.6) 21(65.6) 1.00

Diabetesmellitus 4(12.5) 4(12.5) 1.00

hs-CRP,mg/dL^a 0.07(0.04–0.10) 0.04(0.03–0.06) 0.07 TG,mg/dL^a 113.1(94.6–144.2) 117.9(99.0–141.8) 0.75

HDL-C,mg/dL 53.413.9 52.911.5 0.87

LDL-C,mg/dL 106.736.5 101.620.1 0.50

HbA1c,% 5.750.50 5.780.69 0.86

b-Blockers,% 3(9.4) 16(50.0) <0.01

ACEI/ARB,% 11(34.4) 13(40.6) 0.61

Statins,% 10(31.3) 13(40.6) 0.43

AllvaluesarepresentedasmeansSDsorn(%).

aThevalues arethegeometric meanandcorrespondingasymmetric95%

conﬁdenceintervals.

PAF,paroxysmalatrialﬁbrillation;BMI,bodymassindex;LVEF,leftventricular ejectionfraction;LA,leftatrium;hs-CRP,high-sensitivityC-reactiveprotein;

TG,triglycerides;HDL-C,high-density lipoproteincholesterol;LDL-C, low- densitylipoproteincholesterol;HbA1c,hemoglobinA1c;ACEI/ARB,angiotensin- convertingenzymeinhibitor/angiotensinreceptorblocker.

Table2

DensityofEATaroundtheLAandSAT.

Control PAF p-Value

LA-EAT,HU 111.65.5 108.16.7 0.02

(n=32) (n=32)

SL-EAT,HU 110.78.5 107.57.7 0.15

(n=26) (n=30)

IL-EAT,HU 114.48.4 112.48.7 0.41

(n=22) (n=29)

AR-EAT,HU 111.85.8 107.77.4 0.045

(n=19) (n=31)

IR-EAT,HU 111.77.6 108.710.8 0.25

(n=23) (n=31)

MT-EAT,HU 112.27.1 105.010.4 <0.01

(n=28) (n=26)

AbdominalSAT,HU 121.53.7 123.05.8 0.21

(n=32) (n=32)

ThoracicSAT,HU 122.76.3 124.95.5 0.40

(n=32) (n=32) AlldataarepresentedasmeansSDs.

PAF,paroxysmalatrialﬁbrillation;LA,leftatrium;EAT,epicardialadiposetissue;

SAT,subcutaneousadiposetissue;SL,superiorleft;IL,inferiorleft;AR,anterior right;IR,inferiorright;MT,Marshalltract;HU,Hounsﬁeldunits.

LA-EATisshownasthemeandensityofEATscontainingallﬁveEATareas.

Fig.3.BlandandAltmanplotsofinter-observersagreements.BlandandAltman plotsshowsinter-observeragreementsfortheepicardialadiposetissuedensity measurementswithmeandifference(thickline)and95%limits ofagreement (1.96standarddeviations,dashedlines).

(6)

structures [19]. Our study targeted EAT surrounding the LA becausetheLAplaysanimportantroleinAF,particularlyaround the pulmonary veins. We identified sites of EAT anatomically, corresponding to five major LA GPs based on reports of the autonomicinnervationoftheheart[8,16,17].Infact,EATaround theLAwasfoundinthesefivemajorGPregions.Clinically,GPs havealargeeffecton theinitiationandperpetuationofAF,and theirexistencehasbeenconfirmedbytheinductionofhypoten- sion and bradycardia with high-frequency pacing at that site.

AlthoughweassessedthedensityofEATineachanatomicalGP region,wedidnotconﬁrmthateachEATareaactuallycontaineda GPinthisstudy.

Inﬂammationanddensity

Inﬂamed tissue has higher attenuation values than non- inﬂammatorytissue onCT [20].Furthermore,Konishietal.[15]

reportedthatahigherdensityinCTimagesreflectsthepresenceof pathological inflammation in pericardial fat and is significantly correlatedwithCAD.WemeasuredthedensityoftheEATtoassess EATinflammationinPAFpatients,andwefoundthatthedensityof LA-EATwassignificantlyhigherinthePAFgroupthaninthecontrol group.

InﬂammationofEATandAF

Several prospective epidemiological studies confirmed that inflammationmightconferanincreasedriskofAF[10].Aprevious study assessed inflammation of EAT that was harvested from patients undergoingcardiac surgery,and showedthat it repre- sentedanimportantlocalsourceoftheinflammatorymediators tumornecrosisfactor-

a

and interleukin-6[21,22],whichmight have direct arrhythmogenic effects on atrial tissue and be associated with AF pathogenesis [23]. Moreover, lone AF and rapidacutepacingwererecentlyreportedtoprovokesigniﬁcant changesinatrialadipogenesisandtopromoteadipocytedifferen- tiation and adipose tissue expansion [24]. These mechanisms mightplayanimportantroleintheformationoftheAFsubstrate duringthepathogenesisofAF.Therefore,thesequalitativechanges inEATmayberelatedtohigherEATdensity.

Recently,therelationshipbetweenPAFandpersistentAFand inflammatory activity of the EAT measured using fluorodeox- yglucose-positronemissiontomographywasreported[25].Inour study,thedensityofLA-EATinCTimageswassignificantlyhigher inthePAFgroup,whilethatofSATwasnot.Inaddition,hs-CRPasa markerofsystemicinflammationdidnotdifferbetweenthetwo groups;however,theEATdensitywashigherin thePAF group.

Theseresultssupporttheexistenceoflocalinﬂammationinthefat padssurroundingtheatria,butnotinsubcutaneousfatinthePAF patients.Itiswellknownthaths-CRPlevelsareoftenincreasedin AFpatients[26].However,inourstudy,itdidnotdifferbetween thetwogroups.Inthisstudy,moreCADpatientswereincludedin thecontrolgroup(control:n=6vs.PAF:n=1),andtheusageratio of

b

-blockers,whichhasbeenreportedtoassociatewithlowerCRP levels[27],washigherinthePAFgroup(Table2).Thesemightbe

thereasonswhy hs-CRPlevels inthis study werenotdifferent betweenthegroups.Furtherstudyisneededtodeterminewhether atrialEATcausesinﬂammationandsubsequentAF.

Inthisstudy,theLAdiameterwassigniﬁcantlylargerinthePAF group.AlthoughincreasedLAsizeisastrongriskfactorforAF[28], theEATdensityonCTwasasigniﬁcantpredictorofthepresenceof PAF in the multiple logistic regression analysis including LA diameterinourstudy.

Studylimitations

There were some limitations to this study that must be considered. For example, it was a retrospective study with a relatively smallnumber ofPAF patients. The subjectswerenot randomly selected from the general population, and our data containedriskfactors,includingbloodpressuresandlipidprofiles, that were likely to be modified by therapeutic agents. Future studiesshouldinvestigatetherelationshipbetweenelectrophysi- ological and pathological changes in the atrial myocardium in patientswithAFandinflammationoftheEATaroundtheLA.

Conclusions

In this study, the mean EAT density around the LA as determined by CT was higher in PAF patients, and this result wassigniﬁcantlycorrelatedwiththepresenceofPAF,whichwas independent of other metabolic risk factors. This supports the hypothesisthatinﬂammationofEATaroundtheLA(butnotofSAT) isrelatedtothepresenceofPAF.

Fundingsources

Thisresearchreceived no grant fromanyfundingagency in public,commercial,ornot-for-proﬁtsectors.

Conﬂictofinterest

Theauthorsdeclarethatthereisnoconﬂictofinterest.

Acknowledgments

Theauthorsthankallofthepatients whoparticipatedinthe clinicaltrialsandwhocontributeddatatothisanalysis.Wewould alsoliketoexpressourdeepestgratitudetoDrMasaruSakuraiof KanazawaMedicalUniversityforhisvaluablecommentsaboutthe statisticalanalysis.

References

[1]BenjaminEJ,WolfPA,D’AgostinoRB,SilbershatzH,KannelWB,LevyD.Impact of atrialﬁbrillationon therisk ofdeath:TheFramingham HeartStudy.

Circulation1998;98:946–52.

[2]BenjaminEJ,LevyD,VaziriSM,D’AgostinoRB,BelangerAJ,WolfPA.Indepen- dent riskfactors foratrial ﬁbrillationin apopulation-based cohort. The FraminghamHeartStudy.JAMA1994;271:840–4.

[3]RosengrenA,HauptmanPJ,LappasG,OlssonL,WilhelmsenL,SwedbergK.Big menandatrialﬁbrillation:effectsofbodysizeandweightgainonriskofatrial ﬁbrillationinmen.EurHeartJ2009;30:1113–20.

[4]Rosito GA,Massaro JM,HoffmannU,RubergFL,MahabadiAA,VasanRS, O’DonnellCJ,FoxCS.Pericardialfat,visceralabdominalfat,cardiovascular diseaseriskfactors,andvascularcalciﬁcationinacommunity-basedsample:

TheFraminghamHeartStudy.Circulation2008;117:605–13.

[5]GoudisCA,KorantzopoulosP,NtalasIV,KallergisEM,KetikoglouDG.Obesity and atrialﬁbrillation: acomprehensive reviewof thepathophysiological mechanismsandlinks.JCardiol2015;66:361–9.

[6]AlChekakieMO,WellesCC,MetoyerR,IbrahimA,ShapiraA,CytronJ,Santucci P,WilberDJ,AkarJG.Pericardialfatisindependentlyassociatedwithhuman atrialﬁbrillation.JAmCollCardiol2010;56:784–8.

[7]TsaoHM,HuWC,WuMH,TaiCT,LinYJ,ChangSL,LoLW,HuYF,TuanTC,Wu TJ, Sheu MH, ChangCY, Chen SA.Quantitative analysis ofquantity and distributionofepicardialadiposetissuesurroundingtheleftatriuminpatients withatrialﬁbrillationandeffectofrecurrenceafterablation.AmJCardiol

2011;107: 1498–503.

Table3

MultiplelogisticregressionanalysiscomparingthePAFandcontrolgroups.

Waldstatistics Oddsratio(95%CI) p-Value

LA-EAT,HU 9.008 1.25(1.08–1.45) 0.003

LAdiameter,mm 14.087 1.36(1.16–1.59) <0.001

Age,years 0.927 0.97(0.91–1.03) 0.34

Sex,male 1.18 2.21(0.53–9.34) 0.28

BMI,kg/m² 0.001 1.00(0.75–1.35) 0.98

CI,conﬁdenceinterval;PAF,paroxysmalatrialﬁbrillation;LA,leftatrium;EAT, epicardialadiposetissue;BMI,bodymassindex.

(7)

[8]BatalO,SchoenhagenP,ShaoM,AyyadAE,VanwagonerDR,HallburtonSS, TchouPJ,ChungMK.Leftatrialepicardialadiposityandatrialﬁbrillation.Circ ArrhythmElectrophysiol2010;3:230–6.

[9]OpolskiMP,StaruchAD,KusmierczykM,WitkowskiA,KwiecinskaS,KosekM, JastrzebskiJ,PregowskiJ,KrukM,RozanskiJ,DemkowM,RuzylloW,KepkaC.

Computedtomographyangiographyforpredictionofatrialﬁbrillationafter coronaryarterybypassgrafting:proofofconcept.JCardiol2015;65:285–92.

[10]MazurekT,ZhangL,ZalewskiA,MannionJD,DiehlJT,ArafatH,Sarov-BlatL, O’BrienS, Keiper EA, JohnsonAG,Martin J, Goldstein BJ,Shi Y. Human epicardialadiposetissueisasourceofinﬂammatorymediators.Circulation 2003;108:2460–6.

[11]KonishiM,SugiyamaS,SugamuraK,NozakiT,OhbaK,MatsubaraJ,Matsuzawa Y,SumidaH,NagayoshiY,NakauraT,AwaiK,YamashitaY,JinnouchiH,MatsuiK, KimuraK,etal.Associationofpericardialfataccumulationratherthanabdomi- nalobesitywithcoronaryatheroscleroticplaqueformationinpatientswith suspectedcoronaryarterydisease.Atherosclerosis2010;209:573–8.

[12]MahabadiAA,MassaroJM,RositoGA,LevyD,MurabitoJM,WolfPA,O’Donnell CJ,FoxCS,HoffmannU.Associationofpericardialfat,intrathoracicfat,and visceralabdominalfatwithcardiovasculardiseaseburden:theFramingham HeartStudy.EurHeartJ2009;30:850–6.

[13]JeongJW,JeongMH,YunKH,OhSK,ParkEM,KimYK,RheeSJ,LeeEM,LeeJ, Yoo NJ,KimNH,Park JC. Echocardiographicepicardialfat thicknessand coronaryarterydisease.CircJ2007;71:536–9.

[14]IwayamaT,NitobeJ,WatanabeT,IshinoM,TamuraH,NishiyamaS,Takahashi H,ArimotoT,ShishidoT,MiyashitaT,MiyamotoT,ToyamaS,SadahiroM, KubotaI.Roleofepicardialadiposetissueincoronaryarterydiseaseinnon- obesepatients.JCardiol2014;63:344–9.

[15]KonishiM,SugiyamaS,SatoY,OshimaS,SugamuraK,NozakiT,OhbaK, MatsubaraJ,SumidaH,NagayoshiY,SakamotoK,UtsunomiyaD,AwaiK, JinnouchiH,MatsuzawaY,etal.Pericardialfatinﬂammationcorrelateswith coronaryarterydisease.Atherosclerosis2010;213:649–55.

[16]NakagawaH,ScherlagBJ,PattersonE,IkedaA,LockwoodD,JackmanWM.

Pathophysiologicbasisofautonomicganglionatedplexusablationinpatients withatrialﬁbrillation.HeartRhythm2009;6:S26–34.

[17]KawashimaT.Theautonomicnervoussystemofthehumanheartwithspecial referencetoitsorigin,course,andperipheraldistribution.AnatEmbryol(Berl) 2005;209:425–38.

[18]FriedewaldWT,LevyRI,FredricksonDS.Estimationoftheconcentrationof low-densitylipoproteincholesterolinplasma,withoutuseofthepreparative ultracentrifuge.ClinChem1972;18:499–502.

[19]IacobellisG, CorradiD, SharmaAM. Epicardialadipose tissue: anatomic, biomolecularandclinicalrelationshipswiththeheart.NatClinPractCardi- ovascMed2005;2:536–43.

[20]ZhangM,KonoM.Solitary pulmonarynodules:evaluationofbloodﬂow patternswithdynamicCT.Radiology1997;205:471–8.

[21]SawayaSE,RajawatYS,RamiTG,SzalaiG,PriceRL,SivasubramanianN,Mann DL,KhouryDS.Downregulationofconnexin40andincreasedprevalenceof atrialarrhythmiasintransgenicmicewithcardiac-restrictedoverexpression oftumornecrosisfactor.AmJPhysiolHeartCircPhysiol2007;292:H1561–67.

[22]MarcusGM,WhooleyMA,GliddenDV,PawlikowskaL,ZaroffJG,OlginJE.

Interleukin-6andatrialﬁbrillationinpatientswithcoronaryarterydisease:

datafromtheHeartandSoulStudy.AmHeartJ2008;155:303–9.

[23]TselentakisEV,WoodfordE,ChandyJ,GaudetteGR,SaltmanAE.Inﬂammation effectsontheelectricalpropertiesofatrialtissueandinducibilityofpostop- erativeatrialﬁbrillation.JSurgRes2006;135:68–75.

[24]ChilukotiRK,GieseA,MalenkeW,HomuthG,BukowskaA,FelixSB,KanaanJ, WollertHG,EvertK,VerheuleS,JaisP,HatemSN,LendeckelU,WolkeC.Atrial ﬁbrillationandrapidacutepacingregulateadipocyte/adipositas-relatedgene expressionintheatria.IntJCardiol2015;187:604–13.

[25]Mazurek T,KiliszekM,Kobylecka M,Skubisz-Gluchowska J,Kochman J, FilipiakK,Kro´licki L, OpolskiG. Relation ofproinﬂammatory activity of epicardialadiposetissuetotheoccurrenceofatrialﬁbrillation.AmJCardiol 2014;113:1505–8.

[26]AvilesRJ,Martin DO,Apperson-HansenC, HoughtalingPL,Rautaharju P, KronmalRA,TracyRP,VanWagonerDR,PsatyBM,LauerMS,ChungMK.

Inﬂammationasarisk factorforatrial ﬁbrillation.Circulation2003;108:

3006–10.

[27]PalmasW,MaS,PsatyB,GoffJrDC,DarwinC,BarrRG.Antihypertensive medicationsandC-reactiveproteininthemulti-ethnicstudyofatherosclero- sis.AmJHypertens2007;20:233–41.

[28]KizerJR,BellaJN,PalmieriV,LiuJE,BestLG,LeeET,RomanMJ,DevereuxRB.

Leftatrialdiameterasanindependentpredictorofﬁrstclinicalcardiovascular eventsinmiddle-agedandelderlyadults:TheStrongHeartStudy(SHS).Am HeartJ2006;151:412–8.

Inflammation of left atrial epicardial adipose tissue is associated with paroxysmal atrial fibrillation