Country of birth modifies the association of fatty liver index with insulin action in Middle Eastern immigrants to Sweden

http://www.diva-portal.org

This is the published version of a paper published in Diabetes Research and Clinical Practice.

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

Bennet, L., Groop, L., Franks, P W. (2015)

Country of birth modifies the association of fatty liver index with insulin action in Middle

Eastern immigrants to Sweden.

Diabetes Research and Clinical Practice, 110(1): 66-74

http://dx.doi.org/10.1016/j.diabres.2015.07.011

Access to the published version may require subscription.

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

Permanent link to this version:

Country

of

birth

modifies

the

association

of

fatty

liver

index

with

insulin

action

in

Middle

Eastern

immigrants

to

Sweden

Louise

Bennet

*

,

Leif

Groop

,

Paul

W.

Franks

a_{DepartmentofClinicalSciences,LundUniversity,Ska˚neUniversityHospital,Malmo¨,Sweden} b_{FamilyMedicine,LundUniversity,Malmo¨,Sweden}

c_{GeneticandMolecularEpidemiologyUnit,LundUniversityDiabetesCentre,Malmo¨,Sweden} d

DepartmentofDiabetesandEndocrinology,LundUniversityDiabetesCentre,Malmo¨,Sweden

e_{DepartmentofNutrition,HarvardSchoolofPublicHealth,Boston,MA,USA} f_{DepartmentofPublicHealthandClinicalMedicine,Umea˚ University,Umea˚,Sweden}

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received29September2014 Receivedinrevisedform 2May2015

Accepted27July2015

Availableonline3August2015 Keywords:

Type2diabetes Insulinsensitivity

Non-alcoholfattyliverdisease MiddleEast

a

b

s

t

r

a

c

t

Aims:Non-alcoholfattyliverdisease(NAFLD)isastrongriskfactorforinsulinresistance andtype2diabetes.TheprevalenceofNAFLDvariesacrosspopulationsofdifferentethnic backgroundsbuttheprevalenceinMiddleEasternpopulations,whichareathighriskoftype 2diabetes,islargelyunknown.Usingfattyliverindex(FLI)asaproxyforNAFLDtheaimwas tocalculatetheoddsofNAFLD(FLI70)givencountryoforiginandfurthertoinvestigate theassociationsbetweenISIandFLI.

Methods: In2010–2012weconductedapopulation-basedstudyofindividualsaged30–75 yearsborninIraqorSweden,inwhomanthropometrics,fastingbloodsamplesandoral glucose tolerance tests were performed and sociodemography and lifestyle behaviors characterized.

Results: AhigherproportionofIraqis(N=1085)thanSwedes(N=605)hadahighprobability ofNAFLD(FLI70,32.5vs.22.6%,p<0.001,age-andsex-adjusteddata)andISIwasmore severelyimpaired(70.7vs.95.9%,p<0.001).Independentlyoftraditionalriskfactorsfor NAFLD, being born inIraqi increased therisk of FLI70 (OR 1.59: 95% CI 1.15, 2.20). Furthermore,countryofbirthpresentedastrongerassociationbetweenISIandFLI70 inIraqisthaninSwedes(Pinteraction=0.019).

Conclusions:OurdataindicatethatimmigrantsfromIraqareathigherriskofNAFLD.The findingthatcountryofbirthmodifiestherelationshipofFLIwithISI,suggeststhatliverfat maybeastronger determinantofimpairedinsulinaction andincreased riskoftype2 diabetesinIraqisthaninSwedes.

#_{2015TheAuthors.PublishedbyElsevierIrelandLtd.Thisisanopenaccessarticleunder} theCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

*Correspondingauthorat:CenterforPrimaryHealthCareResearch,ClinicalResearchCentre,Building28,Floor11,JanWaldenstro¨ms gata35,Ska˚neUniversityHospital,20502Malmo¨,Sweden.Tel.:+4640391367;fax:+4640391370.

E-mailaddress:Louise.Bennet@med.lu.se(L.Bennet).

Abbreviations: ALT,alanineaminotransferase; AST,aspartateaminotransferase;CIR,correctedinsulinresponse;DIo,disposition index;GGT,gamma-glutamyltransferase;HDL,high-densitylipoprotein;IQR,interquartilerange;ISI,insulinsensitivityindex;LDL, low-densitylipoprotein;p,plasma;SD,standarddeviation;TGs,triglycerides.

ContentsavailableatScienceDirect

Diabetes

Research

and

Clinical

Practice

j o u r n a lh o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / d i a b r e s

http://dx.doi.org/10.1016/j.diabres.2015.07.011

0168-8227/#2015TheAuthors.PublishedbyElsevierIrelandLtd.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://

1.

Introduction

ImmigrantsfromtheMiddleEastrepresentthelargest non-EuropeanimmigrantgrouptoSwedenandare,comparedtothe native Swedishpopulation, at highriskofimpaired insulin actionandtype2diabetes[1].Thehigherprevalenceispartly explainedbyanexcessprevalenceofabdominalobesityand familyhistoryofdiabetes;nonetheless,ourdatahasshownthat beingborninIraqiisanindependentriskfactorfordiabetes[2]. Furthermore,theMEDIM(impactofMigrationandEthnicityon DiabetesinMalmo¨)studyhasshownthatinsulinsensitivity index(ISI)islowerinIraqisthaninSwedes,andthatISImay haveastrongerimpactontype2diabetesriskinIraqisthanin Swedes[1].ThelowerISIvaluesinimmigrantsfromIraqarenot fullyexplainedbytraditionalriskfactorssuchasfamilyhistory of diabetes, abdominal obesity or physical inactivity, as illustratedbythefactthatinindividualswithanormalwaist circumference and/or body mass index (BMI), ISI is more impairedthaninIraqisthaninSwedeswithequivalentlevelsof adiposity [1]. This suggests that mechanisms other than abdominalobesityorothertraditionalriskfactorsforinsulin resistancemayinfluenceinsulinactioninIraqis.

The mechanisms underlyingthe profoundinsulin resis-tanceintheIraqibornpopulationarepoorlyunderstood,which motivatedthecurrentstudy.Whilstinsulinactionisinfluenced by abdominal and visceral obesity, other organs such as peripheral muscle and the liver are also involved [3]. For instance,non-alcoholfattyliverdisease(NAFLD)representsa strongriskfactorforimpairedinsulinaction,pre-diabetesand type2diabetes[4].PatientsdiagnosedwithNAFLDalsohavea higherreleaseofinflammatorymarkersthanpatientswithout NAFLD[5]andtheyalsohaveahigherriskofcardiovascular disease(CVD)[6].Theseresultsareinconsistencywithprevious findingsfromtheMEDIMstudy,showingthattheassociation between lower ISI and inflammatory markers – such as cytokines – irrespective ofother risk factors [7]is stronger andthattheprevalenceofCVDindiabeticsishigherinIraqis than in Swedes [8]. Altogether thesefindings indicate that NAFLDmaybemoreprevalentandhaveagreaterimpacton insulinactionanddiabetesriskinIraqisthannativeSwedes.

Globally,NAFLDisthemostprevalenttypeofliverdisease

[9–11]anditsprevalencediffersaccordingtoethnicbackground

[12],with12to15%inAsians[13]and30%inAmericans[14]

affected.AlthoughthecharacteristicsoftheIraqiimmigrant population resemble those of patients with NAFLD, the prevalenceofNAFLDand associationwithinsulin actionin MiddleEasternpopulationshastothebestofourknowledgenot beeninvestigatedbefore.Thus,usingthefattyliverindex(FLI) asaproxyforNAFLD[15],intheMEDIMstudywedetermined whetherFLIlevelsdifferedbetweenimmigrantsfromIraqand nativeSwedesandsoughttocharacterizethefactors underly-ingthesedifferences.Further,wesoughttostudyassociations betweeninsulinsensitivityindex(ISI)andFLI.

2.

Subjects

CitizensofMalmo¨ borninIraqandaged30to75yearswere randomlyselected fromthecensus register and invited by

mailandphonetoparticipateinapopulation-basedsurvey. According to the census register, the population of Iraqi immigrants30to75yearsofageinMalmo¨ consistedin2010of 4397personswithameanageof44.8yearsandofwhom57.8% weremen.Swedishborncitizenslivinginthesame geograph-icalareainMalmo¨ wererandomlyselectedfromthecensus registertoreachasimilarageandgenderdistributionasthe Iraqipopulation(meanage45.2years,p=0.08;57.4%males p=0.74).IraqiandSwedishindividualswerethencontacted andinvitedbymailandphonetoparticipateinthestudy.We aimed to recruit a final sample of 2:1 Iraqi and Swedish participants with the goal to reach a similar age and sex distribution amongst the finalparticipants asamongst the originalbackgroundpopulation.Peoplewithtype1diabetes, severephysicalormentalillnessordisabilitieswereexcluded fromthestudy.Aprerequisiteforinclusioninthestudywas also that all values included in the FLI (gamma-glutamyl transferase(p-GGT),BMI,waistcircumferenceand/orplasma triglycerides(p-TG))wereassessed.

To minimize cohort effects and assessment biases, examinationswereconductedwithinarelativelyshort time-frame(February1,2010throughDecember31,2012).Aflow chart describing the recruitment and participation rate of MEDIMispresentedinasupplementaryfigure.Allparticipants conductinganoralglucosetolerancetest(OGTT)thatdidnot haveexcessivealcoholhabits(<9standardglasses/weekfor women and <14 standard glasses/week for men) were includedinthestudy.

3.

Materials

and

methods

3.1. Physicalexamination

Standard physicalexaminations wereperformedbytrained Swedish- and Arabic-speakingresearch nurses and clinical variables such as blood pressure, height, weight, waist circumferences, BMIand abdominalobesity were assessed anddefinedaspreviouslydescribed[16,17].

3.2. Bloodsamplesandoralglucosetolerancetests

Participantswereinstructedtobefastingandnottoeator drinkanythingbutwaterandnottousetobaccoafter10pm thedaybeforetesting.Inthefollowingmorning,fastingblood samples were taken and a 75-gOGTT was performed and bloodsamplesforplasmaglucoseandseruminsulinat30,60, 90 and 120min were collected thereafter. Blood glucose, serum insulin, HbA1c, totalcholesterol, p-TG, high-density lipoprotein cholesterol (p-HDL)and low-density-lipoprotein cholesterol (p-LDL) levels were determined as previously described [16,18]. Plasma alanine aminotransferase(p-ALT, IU/L),aspartateaminotransferase(p-AST,IU/L)and gamma-glutamyltransferase(GGT,IU/L)weremeasuredusingaCobas analyzer(RocheDiagnostics,Mannheim,Germany).

Normalglucosetolerance(NGT),impairedfastingglucose (IFG), impaired glucose tolerance (IGT), impaired glucose regulation (IGR, IFG in combination with IGT) and type 2 diabetesweredefinedaccordingtoWorldHealthOrganization criteria[19];NGT,fastingglucoselevelof<6.1mmol/Landa

2-hplasmaglucoselevelof<7.8mmol/L;IFG,fastingplasma glucose level of 6.1mmol/L and <7.0mmol/L and a 2-h plasma glucose level of <7.8mmol/L, IGT, fasting plasma glucoselevelof<6.1mmol/Landa2-hplasmaglucoselevelof 7.8mmol/Land<11.1mmol/Landtype2diabetesfasting plasma glucose level of 7.0mmol/L and/or a 2-h plasma glucoselevelof11.1mmol/L.Ifonlyoneglucosevaluewas pathologic,theOGTTwasrepeatedonanotherdaywithin2 weekswiththesamefastingprocedures.Twovalues exceed-ingthesethresholdswereneededfordiagnosis[19].IFG,IGT andIGRareinthispapercollectivelyreferredtoas ‘prediabe-tes’. Estimation of insulin sensitivity (ISI) and beta cell function(correctedinsulinresponse,CIRandoraldisposition index,DIo)wereassessedusingtheMatsudaindices calculat-edfromtheOGTTasdescribedpreviously[20–23].

FLIwascalculatedaspreviouslydescribedbyBalkauetal.

[15]:

FLI¼ e

L

ð1þeL_Þ100

where L=0.953 (loge p-TG)+0.139BMI+0.718 (loge

(GGT)+0.053waistcircumference15.745.

Liver fatscore was calculatedas previously described by Kotronenetal.[24]:

Liver fat score=2.89+1.18metabolic syndrome [17](yes=1/no=0)+0.45type 2 diabetes (yes=2/no=0)+ 0.15fasting S-insulin (mU/L)+0.04fasting S-AST (U/ L)0.94AST/ALT

3.3. Questionnaires

Information on previous diagnoses of diabetes, current medication,familyhistoryofdiabetes(inbiologicalparents and/or siblings), lifestyle habits and sociodemography was collected in interviews by Arabic- and Swedish-speaking nurses using structured questionnaires in Swedish and Arabic.Questionnairesweretranslatedand back-translated bytwoindependentprofessionaltranslatorswithArabicas theirnativelanguage[16].

Smokinghabits,alcoholconsumptionandhoursphysically active/weekwereassessedasdescribedpreviously[1].

3.4. Statisticalanalysis

AnalyseswereperformedusingSTATAIC/12.1.Leastsquares meanswerederivedafterageandsexadjustmentusinggeneral linearmodels;differencesinproportionswereadjustedforage andsexusinglogisticregression.AssociationswithFLI70% wereassessedusingstepwiselogisticregressionanalysisand associationswithISI(logtransformed)wereassessedbylinear regression analysis. Units were standardized in strata of ethnicityandsexper1standarddeviation(SD)unitvariance forthecontinuousindependentvariables.Dataareexpressed as odds ratios (ORs) (FLI as the dependent variable) and b coefficients(ISIasthedependentvariable)with95%confidence intervals(CIs)(Tables2aand2b).Alltestsweretwo-sidedanda p-valueof<0.05wasconsideredstatisticallysignificant.

Inordertominimizethemultipletestingburden, interac-tions were considered only when the included marginal effectsweresignificant.

Multicollinearitywasnotconsideredanissue,asvariance inflationfactor(VIF)valuesinthefinalmultivariateregression modelswere<3.1.

3.5. Ethicalconsiderations

All participantsprovidedwritteninformedconsent andthe Ethics Committee at Lund University approved the study (application nos. 2009/36 and 2010/561).The MEDIM study conforms to the principles outlined in the Declaration of Helsinki[25].

4.

Results

Intotal, 1085participantsborn inIraq and605participants borninSwedenwereincludedinthestudy.DespitetheIraqias comparedtotheSwedish populationwasyoungerthanthe Swedish population(45.2 vs. 49.1years, p<0.001) their FLI valueswerehigher(54.6vs.43.2,p<0.001)andtheirinsulin sensitivitywasmoreimpaired(ISI70.7vs.95.9%,p<0.001).A higherproportionofIraqisthanSwedeshadahighprobability offattyliver(FLI70,32.5vs.22.6%,p<0.001,age-and sex-adjusteddata,Fig.1).

Amongst participants with ahigh probability of NAFLD (FLI70), theIraqiswereparadoxicallyyounger, hadlower totalcholesterollevelsandlowerp-LDLlevelsascomparedto theSwedes.Furthermore,theirprevalenceofthemetabolic syndromewaslowerandinwomenwaistcircumferencewas smaller than in Swedes(Table 1). Inparticipants with low (FLI<20)andmediumFLIvalues(FLI20–69),Iraqishadworse glucoseandlipidmetabolism,asindicatedbylowerISIvalues, higherBMI,lowerp-HDLlevels,higherp-TGlevelsandhigher fastingglucoselevels(Table1).

In a stepwise logistic regression modelwe studiedrisk factorsassociatedwithFLI70(Table2a).Adjustingforthe confoundingeffectofotherknownriskfactorsforNAFLDsuch asage,sex,abdominalobesity,type2diabetesand/oralcohol consumption, beingborn inIraq remained anindependent risk factor for FLI70. Other risk factors independently contributingtohigheroddsofFLI70werefemalesex,type 2 diabetes and last but not least abdominal obesity that contributedwithover27timestheoddsofFLI70.Together, allriskfactorsexplained34%ofthevariance(R2_)inliverfat

scores (Table 2a). The variance did not differ between ethnicities(datanotshown).

Correlation between FLI and ISI (log-transformed) in participants without abdominal obesity who were born in IraqorSwedenarepresentedinFig.2.Thisfigureshowsthat withdecreasingISI,FLIincreasedevenmoreinIraqisthanin Swedes,indicatingastrongerinfluenceofNAFLDoninsulin resistanceintheIraqipopulation.Wethenanalysed associa-tions between ISI and traditional risk factors for insulin resistanceafteradjustingforFLI(Table2b).BeingborninIraq remainedanindependentriskfactorforimpairedISIinthefull multivariatemodel.Furthermore,weobservedthatcountryof birth modified the effect of FLI70 (Pinteraction=0.019),

indicatingastrongerassociationbetweenISIand FLI70% inIraqisthaninSwedes.Together,allriskfactorsexplained 35%ofthevariance(R2_)inISI.

4.1. Representativenessofthestudysample

InimmigrantsfromIraqtheageandsexdistributiondidnot differ in studyparticipants compared tothe eligible back-groundpopulation.ParticipantsborninSweden wereolder (49.1vs.44.8years,p<0.001),butthesexdistributiondidnot differ compared to the eligible native Swedish population (datanotshown).

5.

Discussion

Akeynovelfindingfromthisstudyisthatirrespectiveofother traditional risk factors for fatty liver disease, Iraqis had a higherriskofNAFLD(FLI70)thannativeSwedes. Further-more,highFLIandIraqioriginwereindependentlyassociated withimpairedinsulin actionirrespective oftraditional risk factorssuchasphysicalinactivity,abdominalobesityandtype 2diabetes.

Thefindingthat countryofbirth modifiesthe relation-shipofFLIwithISI,suggestsastrongereffectoffattyliver on insulin action in Iraqis than in Swedes, and opens a possible effect of fatty liver as a strong contributor to impairedinsulinactionandincreasedriskoftype2diabetes in this population that to our knowledge has not been reportedbefore.

5.1. NAFLDandinsulinresistance

Diminished insulin secretion and action are important pathological determinants of type 2 diabetes [26]. Insulin actionisnotonlyregulatedbyabdominalorvisceralobesity butthatotherorgansandtissues,suchasperipheralmuscle andtheliverarealsoinvolved[3].NAFLDisassociatedwith impairedinsulinactionand waspreviouslyidentifiedasan independentriskfactorforpre-diabetesandtype2diabetes

[3,4]. PreviousdatafromtheMEDIMstudyshowedstronger

associations betweentype2 diabetesand insulinaction in Iraqis than in Swedes, indicating that in Iraqis impaired insulin action appears tohave a greater impact on type2 diabetes[1].

ThemetaboliccharacteristicsoftheIraqiparticipantsin theMEDIMstudyresembleaNAFLDphenotype[19–21],with pronouncedinsulinresistance,highplasmaTGlevelsanda highprevalenceofdiabetesdespitelessabdominalobesity

[1].Further,patientsdiagnosedwithNAFLDarereportedto haveahigherreleaseofinflammatorymarkersthanpatients without NAFLD [5]. Consistent with those findings, we previouslyshowedthatthe associationsbetweenlowerISI andinflammatorymarkerssuchascytokinesarestrongerin IraqisthaninSwedes,irrespectiveofotherriskfactors[7].We havealsoshownthattheassociationbetweencardiovascular disease (CVD) in Iraqis depends on diabetes status, with three-fold higher odds of CVD in Iraqis with diabetes as comparedSwedeswithdiabetes;amongstindividuals with-out diabetes the odds of CVD amongst Iraqis half that observedinSwedes[8]. Thisisconsistentwitha previous study of diabetes patients which reported that patients diagnosedwithNAFLD havingahigherprevalenceofCVD thanthosewithoutNAFLD[6].Sincethese dataaltogether indicatethatahighproportionoftheIraqipopulationmay haveNAFLD inthisstudy weinvestigated whether differ-encesinNAFLDcouldexplaintheprofoundinsulinresistance intheIraqiborngroup.AsaproxyforNAFLDweincluded FLI70inouranalysis.Bydoingthatthevarianceincreased morethanthree-fold(from0.13to0.34),reflectingthestrong influenceofNAFLDoninsulinresistance.Stillbeingbornin Iraqremainedanindependentriskfactorforethnic differ-ences ininsulin resistanceindicating there areadditional genetic-, behavioral or metabolic risk factors influencing insulinresistanceinthisimmigrantpopulationthatremain tobedetected.

Fig.1–Distributionoflow(<20),medium(20–69)andhigh(I70)fattyliverindex(FLI)valuesinparticipantsborninIraqor Sweden.HighFLIvaluesweremoreprevalentinIraqisthaninSwedes(32.5vs.22.6%,p<0.001),whereaslowFLIvalues werelessprevalent(13.9vs.28.7%,p<0.001).

Table 1 – Characteristics of the study population born in Iraq or Sweden that underwent an oral glucose tolerance test in relation to low (<20), medium (20–69) or high (I70) fatty liver index values [15].

Variable Fatty liver index <20 Fatty liver index 20–69 Fatty liver index 70

Born in Iraq N = 168 (13.9%) Born in Sweden N = 188 (28.7%) p Born in Iraq N = 524 (43.3%) Born in Sweden N = 269 (41.1%) p Born in Iraq N = 393 (32.5%) Born in Sweden N = 148 (22.6%) p Age (years) 41.9 (9.0) 46.8 (10.7) <0.001 45.7 (9.4) 49.4 (11.1) <0.001 46.2 (8.5) 50.6 (11.3) <0.001 Male gender, n (%) 50 (29.8) 63 (33.5) 0.448 299 (57.1) 162 (60.2) 0.393 296 (75.3) 104 (70.3) 0.234 BMI (kg/m2_{) 24.6 (2.2) 23.2 (2.2) <0.001 27.8 (2.6) 27.0 (2.7) <0.001 32.4 (4.1) 32.3 (4.3) 0.419} Waist men (cm) 85.3 (4.9) 84.1 (7.4) 0.329 93.9 (6.2) 94.9 (6.4) 0.301 105.9 (9.1) 107.7 (8.9) 0.231 Waist women (cm) 81.4 (6.6) 78.6 (7.1) 0.005 92.1 (5.9) 93.0 (6.6) 0.219 103.9 (7.6) 109.9 (9.9) <0.001 ISIa_{(mmol/L  mU/L}1_{) 121.8}

(97.8–161.0) 157.9 (112.2–211.8) <0.001 83.2 (59.8–18.3) 94.1 (72.1–133.9) <0.001 53.2 (37.0–74.8) 56.2 (43.0–76.7) 0.177 CIRa (mU/L  mmol/L1 mmol/L1₎ 150.8 (95.0–252.8) 130.8 (76.5–204.9) 0.135 164.4 (92.7–267.8) 146.0 (86.9–214.5) 0.973 199.8 (116.8–347.3) 180.9 (111.1–289.4) 0.399 DIoa _19021.9 10605.3–33537.2) 20263.8 (12719.3–30.958.4) 0.221 12879.6 (7302.2–23703.2) 13689.1 (7934.4–24145.3) 0.010 10454.7 (5522–18235) 9801.3 (5186.2–17701.5) 0.834 Total cholesterol (mmol/L) 4.4 (0.8) 4.9 (1.0) <0.001 4.8 (0.9) 5.3 (1.0) <0.001 5.2 (1.0) 5.5 (1.1) 0.001

p-LDL (mmol/L) 2.8 (0.7) 3.0 (0.8) 0.110 3.2 (0.8) 3.4 (0.9) 0.023 3.4 (0.9) 3.7 (1.0) 0.007 p-HDL (mmol/L) 1.4 (0.3) 1.7 (0.4) <0.001 1.2 (0.3) 1.4 (0.4) <0.001 1.1 (0.3) 1.1 (0.3) 0.728 p-TGs (mmol/L) 0.8 (0.3) 0.7 (0.2) 0.008 1.3 (0.6) 1.2 (0.5) <0.001 2.2 (1.2) 2.0 (1.1) 0.551 p-ALT (IU/L) 9.3 (3.3) 8.9 (3.2) 0.573 11.3 (4.9) 12.5 (8.5) 0.011 14.5 (9.1) 13.7 (7.9) 0.879 p-AST (IU/L) 22.9 (6.0) 23.7 (6.3) 0.491 25.9 (6.6) 27.4 (12.1) 0.027 29.8 (10.3) 29.0 (9.4) 0.386 p-GGT (IU/L) 13.9 (7.3) 16.4 (9.4) 0.125 27.7 (16.9) 32.3 (32.0) 0.019 54.0 (52.6) 59.8 (63.3) 0.210 Fasting glucose (mmol/L) 5.5 (0.6) 5.4 (0.5) 0.024 5.7 (0.6) 5.6 (0.6) 0.051 5.6 (0.8) 5.8 (1.0) 0.382 2-h glucose (mmol/L) 5.5 (1.6) 5.4 (1.6) 0.274 5.8 (1.9) 5.6 (1.9) 0.052 6.3 (2.5) 6.6 (2.6) 0.742 Liver fat score[24] 3.5 (1.1) 3.8 (1.1) 0.051 2.2 (1.5) 2.4 (1.3) 0.134 0.6 (1.9) 0.6 (2.1) 0.598 New cases of type 2

diabetes, n (%)

0 0 0.314 10 (1.9) 2 (0.7) 0.220 16 (4.1) 6 (4.1) 0.947

Metabolic syndrome[17], n (%) 19 (11.3) 12 (6.4) 0.066 213 (40.6) 101 (37.5) 0.022 294 (74.8) 125 (84.5) 0.036 Participants with excessive alcohol consumption (9 standard glasses/week in women and 14 standard glasses/week in men) were excluded from the study. Data are presented as the mean (standard deviation, SD), number (percentage) or median (interquartile range, IQR). Differences in means between groups were adjusted for age and sex using general linear models (for continuous variables) while differences in proportions between groups were studied using logistic regression adjusted for age and gender. All tests were two-sided and a p-value of <0.05 was considered statistically significant.

a _{Data presented as IQR CIR and DIo only included cases where the glucose level at 30 min was >4.44 mmol/L and was greater than the fasting glucose level[22].}

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 1 1 0 ( 2 0 1 5 ) 6 6 – 7 4

5.2. NAFLDandethnicdifferences

Studies of the Pima Indians, a population that shares similaritieswithIraqisinthattheyarehighlyinsulinresistant andhaveoneofthehighestprevalenceratesoftype2diabetes intheworld,haveshownthatincreasedlipidcontentinthe

liver independently links hypoadiponectinaemia, hypertro-phicobesityandincreasedvisceraladipositywithperipheral and hepatic insulin resistance [27]. These findings are consistentwithourfindingthattype2diabetesandabdominal obesityincreasedtheoddsofNAFLD(FLI70).Ourfindings that beingborn inIraq was anindependent risk factorfor NAFLDtogetherwiththefindingthatcountryofbirthmodified theeffectofFLI70oninsulinaction,suggeststhatNAFLD has apotentially crucialrolein thedevelopmentoftype2 diabetes in Iraqis, representing the largest non-European immigrantgroupinSweden.Ourdatafurthersuggeststhat theIraqipopulationwithNAFLDisyoungerthantheSwedish bornpopulation.Thismayreflectanearlieronsetofdisturbed fat metabolism contributing to insulin resistance and an earlier diabetes onset in Iraqi immigrants that we have previouslyreported[28].

Increased knowledge of the mechanisms triggering insulin action in different populations is important for betterunderstandinginthishighriskpopulationfortype2 diabetes.

Toourknowledge,thereisadearthofdataonNAFLDfrom Iraq.However,ourdataofanestimatedprevalenceofNAFLD of over 30% in this population correspond with previous studiesconductedintheMiddleEastregion. Forinstancea comparisonbetweenpopulationsofArabsandSouthAsians inKuwaitrevealedaprevalenceofNAFLDin33.3and29.0%, respectively[29]whereastheprevalenceofNAFLDinIsraelis wasreportedto30%[30]inbothstudiesNAFLDwasverifiedby abdominalultrasonography.Furtherethnicaswellasgender differencesintheprevalenceofNAFLDisreportedinstudiesof Blacks,HispanicsandWhitesconductedintheUSA,witha higherprevalenceinHispanicsthaninBlacksandahigher prevalence in menthan in women [14,29,30]. Thelatter is

Fig.2–Correlationbetweenfattyliverindex(FLI)andinsulinsensitivityindex(ISI,log-transformed)inparticipantswithout abdominalobesitywhowereborninIraqorSweden.TheassociationbetweenISIandFLII70wasmodifiedbycountryof birth(Pinteraction=0.019),indicatingthatfattyliverdiseasemayhaveastrongerimpactoninsulinactioninIraqisthanin

Swedes.

Table2a–Riskfactorsassociatedwithfattyliverindex I70%inthetotalstudypopulation.

Variable Model1 (R2_=0.09) OR 95%CI Model2 (R2_=0.10) OR 95%CI Model3 (R2_=0.34) OR 95%CI BorninSweden BorninIraq Reference 1.73 1.38to2.18 Reference 1.71 1.36to2.16 Reference 1.59 1.15to2.20 Age(years),per1SD 1.16

1.04to1.30 1.14 1.02to1.28 1.02 0.90to1.15 Femalesex Malesex Reference 2.84 2.26to3.56 Reference 2.85 2.27to3.58 Reference 4.78 3.71to6.15 Nodiabetes Type2diabetes Reference 3.72 1.77to7.82 Reference 3.30 1.48to7.34 Noalcohol consumption Alcoholconsumption Reference 1.05 0.77to1.44 Noabdominalobesity Abdominalobesity Reference 27.04 16.43to44.49 Associationswerestudiedusingstepwiselogisticregressionand expressedasoddsratios(ORs)with95%confidenceintervals(CIs). Thevarianceis expressedasR2_{.Oddsratioswerestandardized}

(SD)inthestrataofethnicityandsexper1SDunitvarianceforthe continuousindependentvariables.

consistentwithourdatashowinghigheroddsofFLI70in menthan in women, irrespective of abdominalobesity or otherriskfactorsforNAFLD.AhigherprevalenceofNAFLDin differentpopulationscould alsobecausedby genetics.For instance, NAFLD shares genetic risk factors with other intermediatemetabolicriskfactorssuchasdiminishedinsulin sensitivityandhypertriglyceridemia[20]andgeneticfactorsis shown to contribute to disturbed fat metabolism and increased fat storage in the liver, causing steatohepatosis andsubsequentlyfibrosis[3].

5.3. Strengthsandlimitations

Ourstudyispopulation-basedandrepresentsalargefraction oftheIraqipopulationinthestudiedregion.Itisalsodistinct fromother studiesofthistopic inthatit includes detailed metabolic phenotyping and assessments of lifestyle expo-sures.Thestudyislimitedbytheinabilitytoinfercausality duetothe cross-sectionaldesign.TheparticipatingSwedes weresomewhatolderthanthecorrespondingnativeSwedish backgroundpopulation,whichmayhaveintroducedselection bias. However, we consider our data reliable since age is adjustedforinthemultivariateregressionmodels.FLIorother estimationsofNAFLDhavebeen developedinaCaucasian populationandhasnotbeenstudiedor validatedbeforein Middle Eastern populations. Another limitation is that the study lacks quantitative information on liver fat content assessed by abdominal ultrasound or magnetic resonance imaging(MRI).

6.

Conclusions

Thisstudyreportsthatirrespectiveoftraditionalriskfactors, immigrantsfromIraqhaveahigherprobabilityofNAFLDand worse insulinactionthan nativeSwedes.Althoughliverfat content was not investigated by ultrasound or MRI in the presentstudy,ourdataindicatethatIraqisandinparticular IraqimaleshaveaveryhighprobabilityofNAFLD,andthatthis probabilityincreasesfurtherinthosewithabdominalobesity. Ourdatasuggeststhatliverfatcontentmayplayastrongerrole in insulin action in the Iraqi compared to the Swedish population.Population-basedfocusedonobjectivelyverifying liverfatcontentandstudyingtheinfluenceofNAFLDontype2 diabetesriskinMiddleEasternpopulationsmaybewarranted.

Conflict

of

interest

statement

LouiseBennet,LeifGroopandPaulW.Franksdeclarethatthey havenoconflictofinterest.

Statement

of

human

and

animal

rights

Allproceduresfollowedwereinaccordancewiththeethical standards of the responsible committee on human experi-mentation(institutionalandnational)andwiththeHelsinki Declarationof1975,asrevisedin2008(5).

Table2b–Riskfactorsassociatedwithinsulinsensitivityindex(log-transformed)inthetotalstudypopulation.

Variable Model1 R2_=0.09 b 95%CI Model2 R2_=0.10 b 95%CI Model3 R2_=0.13 b 95%CI Model4 R2_=0.34 b 95%CI Model5 R2=0.35 b 95%CI BorninSweden BorninIraq Reference .12 .15to.10 Reference .13 .14to.09 Reference .12 .15to.10 Reference .09 .11to.07 Reference .11 .13to.08 Age(years),per1SD .02

.04to.01 .02 .03to.01 .03 .04to.01 .01 .02to.01 .01 .02to.01 Femalesex Malesex Reference .10 .12to.07 Reference .10 .13to.08 Reference .10 .13to.08 Reference .08 .11to.06 Reference .08 .11to.06 Nofamilyhistoryofdiabetes

Familyhistoryofdiabetes

Reference .04 .07to.02 Reference .04 .07to.02 Reference .02 .04to.01 Reference .02 .04to.01 Physicalactivity(h/week),per1SD .04

.03to.06 .03 .02to.04 .03 .02to.04 Noabdominalobesity Abdominalobesity Reference .14 .17to.12 Reference .14 .17to.11 Notype2diabetes Type2diabetes Reference .14 .22to.06 Reference .14 .22to.06 Fattyliverindex<70

Fattyliverindex70

Reference .19 .22to.16 Reference .24 .28to.19

BorninIraqFattyliverindex70 .06

.01to.12 Associationswerestudiedusingstepwiselinearregression andareexpressedasbcoefficientswith95%confidenceintervals(CIs).The varianceisexpressedasR2_{.b-coefficientswerestandardized(SD)inthestrataofethnicityandsexper1SDunitvarianceforthecontinuous}

Statement

of

informed

consent

Informed consent wasobtained fromall patientsfor being includedinthestudy.

Funding

ThisstudywasfundedbygrantsfromLundUniversity(ALF grants20101641,20101837and162641),RegionSka˚ne(226661 and121811),theSwedishSocietyofMedicine(SLS97081and 176831),theCrafoordFoundation(20110719)andtheSwedish ResearchCouncil(Linne´ grant349-2008-6589awardedtothe Lund University Diabetes Centre, Exodiab 2009-1039 and ANDIS825-2010-5983).

Acknowledgements

L.B.designedthestudy,wrotethemanuscript,andobtained, analyzed and interpreted the data. L.G. contributed tothe designofthestudy,interpretationofthedataanddiscussions ofthestudy’sfindings.P.W.F.contributedtointerpretationof thedata,discussionsofthestudy’sfindingsandwritingthe manuscript. All authors revised/edited the article critically andapprovedthefinalversionofthemanuscript.

We areindebtedtoStephen Gilliver,Center forPrimary HealthCareResearchatLundUniversityforlanguageediting, MaritaOlsson,Katarina BalckerLundgren,EnasBasheer El-Soussi,AsmaSalehandJanniHigafortheirexcellentworkin examiningtheparticipantsandcollectingdata.

Appendix

A.

Supplementary

data

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.

diabres.2015.07.011.

r

e

f

e

r

e

n

c

e

s

[1] BennetL,GroopL,FranksPW.Ethnicdifferencesinthe

contributionofinsulinactionandsecretiontotype2

diabetesinimmigrantsfromtheMiddleEastcomparedto

nativeSwedes.DiabetesResClinPract2014;105:79–87.

[2] BennetL,GroopL,LindbladU,AgardhCD,FranksPW.

Ethnicityisanindependentriskindicatorwhenestimating

diabetesriskwithFINDRISCscores:acrosssectionalstudy

comparingimmigrantsfromtheMiddleEastandnative

Swedes.PrimCareDiabetes2014;8(3):231–8.

[3] deAlwisNM,DayCP.Non-alcoholicfattyliverdisease:the

mistgraduallyclears.JHepatol2008;48(Suppl.1):S104–12.

[4] Zelber-SagiS,LotanR,ShiboletO,WebbM,BuchA,

Nitzan-KaluskiD,etal.Non-alcoholicfattyliverdisease

independentlypredictsprediabetesduringa7-year

prospectivefollow-up.LiverInt2013;33:1406–12.

[5] TargherG,ArcaroG.Non-alcoholicfattyliverdiseaseand

increasedriskofcardiovasculardisease.Atherosclerosis

2007;191:235–40.

[6] TargherG,BertoliniL,PadovaniR,RodellaS,TessariR,

ZenariL,etal.Prevalenceofnonalcoholicfattyliverdisease

anditsassociationwithcardiovasculardiseaseamongtype

2diabeticpatients.DiabetesCare2007;30:1212–8.

[7] MemonAA,SundquistJ,WangX,Palme´rK,SundquistK,

BennetL.Theassociationbetweencytokinesandinsulin

sensitivityinIraqiimmigrantsandnativeSwedes.BMJ

Open2013;3:e003473.

[8] BennetL,AgardhCD,LindbladU.Cardiovasculardiseasein

relationtodiabetesstatusinimmigrantsfromtheMiddle

EastcomparedtonativeSwedes:across-sectionalstudy.

BMCPublicHealth2013;13:1133.

[9] WongVW,ChuWC,WongGL,ChanRS,ChimAM,OngA,

etal.Prevalenceofnon-alcoholicfattyliverdiseaseand

advancedfibrosisinHongKongChinese:apopulation

studyusingproton-magneticresonancespectroscopyand

transientelastography.Gut2012;61:409–15.

[10] BellentaniS,SaccoccioG,MasuttiF,Croce` LS,BrandiG,

SassoF,etal.Prevalenceofandriskfactorsforhepatic

steatosisinNorthernItaly.AnnInternMed2000;132:112–7.

[11] AdamsLA,LympJF,StSauverJ,SandersonSO,LindorKD,

FeldsteinA,etal.Thenaturalhistoryofnonalcoholicfatty

liverdisease:apopulation-basedcohortstudy.

Gastroenterology2005;129:113–21.

[12]WilliamsCD,StengelJ,AsikeMI,TorresDM,ShawJ,

ContrerasM,etal.Prevalenceofnonalcoholicfattyliver

diseaseandnonalcoholicsteatohepatitisamongalargely

middle-agedpopulationutilizingultrasoundandliverbiopsy:

aprospectivestudy.Gastroenterology2011;140:124–31.

[13] FanJG,ZhuJ,LiXJ,ChenL,LiL,DaiF,etal.Prevalenceof

andriskfactorsforfattyliverinageneralpopulationof

ShanghaiChina.JHepatol2005;43:508–14.

[14] BrowningJD,SzczepaniakLS,DobbinsR,NurembergP,

HortonJD,CohenJC,etal.Prevalenceofhepaticsteatosisin

anurbanpopulationintheUnitedStates:impactof

ethnicity.Hepatology2004;40:1387–95.

[15] BalkauB,LangeC,VolS,FumeronF,BonnetF,DESIRGS.

Nine-yearincidentdiabetesispredictedbyfattyliver

indices:theFrenchD.E.S.I.R.study.BMCGastroenterol

2010;10:56.

[16] BennetL,JohanssonSE,AgardhCD,GroopL,SundquistJ,

RastamL,etal.Highprevalenceoftype2diabetesinIraqi

andSwedishresidentsinadeprivedSwedish

neighbourhood—apopulationbasedstudy.BMCpublic

health2011;11:303.

[17] AlbertiKG,EckelRH,GrundySM,ZimmetPZ,CleemanJI,

DonatoKA,etal.Harmonizingthemetabolicsyndrome:a

jointinterimstatementoftheInternationalDiabetes

FederationTaskForceonEpidemiologyandPrevention;

NationalHeartLung,andBloodInstitute;AmericanHeart

Association;WorldHeartFederation;International

AtherosclerosisSociety;andInternationalAssociationfor

theStudyofObesity.Circulation2009;120:1640–5.

[18] ThorellJ,LarsonSM.Radioimmunoassayandrelated

techniques.STLouis:TheCVMosbyCompany;1978:

205–11.

[19] WorldHealthOrganization.Definition,diagnosisand

classificationofdiabetesmellitusanditscomplications:

Reportofa,WHO.,consultation.,Part,1.,Diagnosisand

classificationofdiabetes,mellitus.Geneva:WorldHealth

Organization;1999.

[20] MatsudaM,DeFronzoRA.Insulinsensitivityindices

obtainedfromoralglucosetolerancetesting:comparison

withtheeuglycemicinsulinclamp.DiabetesCare

1999;22:1462–70.

[21] HansonRL,PratleyRE,BogardusC,NarayanKM,Roumain

JM,ImperatoreG,etal.Evaluationofsimpleindicesof

insulinsensitivityandinsulinsecretionforusein

epidemiologicstudies.AmJEpidemiol2000;151:190–8.

[22] SluiterWJ,ErkelensDW,ReitsmaWD,DoorenbosH.

Glucosetoleranceandinsulinrelease,amathematical

approachIAssayofthebeta-cellresponseafteroral

glucoseloading.Diabetes1976;25:241–4.

[23] BergmanRN,AderM,HueckingK,VanCittersG.Accurate

assessmentofbeta-cellfunction:thehyperboliccorrection.

Diabetes2002;51(Suppl.1):S212–20.

[24] KotronenA,PeltonenM,HakkarainenA,SevastianovaK,

BergholmR,JohanssonLM,etal.Predictionof

non-alcoholicfattyliverdiseaseandliverfatusingmetabolic

andgeneticfactors.Gastroenterology2009;137:865–72.

[25] WMA. DeclarationofHelsinki—Ethicalprinciplesof

medicalresearchinvolvinghumansubjects2008.h**ttp://

w**ww.wma.net/en/30publications/10policies/b33/ index.html.

[26] DeFronzoRA.Lillylecture1987.Thetriumvirate:beta-cell,

muscle,liver.AcollusionresponsibleforNIDDM.Diabetes

1988;37:667–87.

[27]KoskaJ,StefanN,PermanaPA,WeyerC,SonodaM,Bogardus

C,etal.Increasedfataccumulationinlivermaylinkinsulin

resistancewithsubcutaneousabdominaladipocyte

enlargement,visceraladiposity,andhypoadiponectinemia

inobeseindividuals.AmJClinNutr2008;87:295–302.

[28] BennetL,LindbladU,FranksPW.Afamilyhistoryof

diabetesdeterminespoorerglycaemiccontrolandyounger

ageofdiabetesonsetinimmigrantsfromtheMiddleEast

comparedwithnativeSwedes.DiabetesMetab

2015;41(1):45–54.

[29] BabusikP,BilalM,DurisI.Nonalcoholicfattyliverdisease

oftwoethnicgroupsinKuwait:comparisonofprevalence

andriskfactors.MedPrincPract2012;21:56–62.

[30] Zelber-SagiS,Nitzan-KaluskiD,HalpernZ,OrenR.

Prevalenceofprimarynon-alcoholicfattyliverdiseaseina

population-basedstudyanditsassociationwith

biochemicalandanthropometricmeasures.LiverInt