meta分析范文展示Word文档下载推荐.docx
《meta分析范文展示Word文档下载推荐.docx》由会员分享,可在线阅读,更多相关《meta分析范文展示Word文档下载推荐.docx(25页珍藏版)》请在冰豆网上搜索。
//jn.nutrition.org
RUNNINGTITLE:
Fructoseandcholesterol
WORDCOUNT:
5618;
NUMBEROFFIGUREA:
3;
NUMBEROFTABLES:
2
SUPPLEMENTARYMATERIAL:
OnlineSupportingMaterials:
5
AUTHORLISTFORINDEXING:
An,Zhang,Zhang,Zhou,Huang,Zhang
1ThestudywassupportedbytheMinistryofScienceandTechnologyofChinawithgrantoftheNationalHigh-techResearchandDevelopmentProgramofChinatoDrJianZhang.
2Authordisclosures:
T.An,R.C.Zhang,Y.H.Zhang,Q.Zhou,Y.Hung,J.Zhanghavenoconflictsofinterest.
*Towhomcorrespondenceshouldbeaddressed.Mailingaddress:
HeartFailureCenter,CardiovascularInstituteandFuwaiHospital,ChineseAcademyofMedicalSciencesandPekingUnionMedicalCollege,167Beilishilu,Beijing,China;
Zipcode:
100000;
Telephonenumber:
86-10-88396180;
Faxnumber:
E-mail:
Fwzhangjian62@
PROSPEROREGISTRATIONNUMBERS:
CRD42012003351
ABSTRACT
Fructoseiswidelyusedasasweetenerinproductionofmanyfoods,yettherelationbetweenfructoseintakeandcholesterolremainsuncertain.Weperformedasystematicreviewandmeta-analysisofhumancontrolledfeedingtrialsofisocaloricfructoseexchangeforothercarbohydratestoquantifytheeffectsoffructoseontotalcholesterol(TC),LDLcholesterol(LDL-C),andHDLcholesterol(HDL-C)inadulthumans.Weightedmeandifferenceswerecalculatedforchangesfrombaselinecholesterolconcentrationsbyusinggenericinversevariancerandom-effectsmodels.TheHeylandMethodologicalQualitywasusedtoassessstudyquality.Subgroupanalysesandmeta-regressionwereconductedtoexplorepossibleinfluenceofstudycharacteristics.Twenty-fourtrials(withatotalof474subjects)wereincludedinourmeta-analysis.Inanoverallpooledestimate,fructoseexertednoeffectonTC,LDL-CandHDL-C.Meta-regressionanalysisindicatedthatfructosedosewaspositivelycorrelatedwiththeeffectsizesofTCandLDL-C.SubgroupanalysesshowedthatisocaloricfructoseexchangeforcarbohydratescouldsignificantlyincreaseTCby12.97mg/dL(95%CI:
4.66,21.29;
P=0.002)andLDL-Cby11.59mg/dL(95%CI:
4.39,18.78;
P=0.002)at>
100gfructose/dbuthadnoeffectonTCandLDL-Cwhenfructoseintakewas≤100g/d.Inconclusion,veryhighfructoseintake(>
100g/d)couldleadtosignificantlyincreaseinserumLDL-CandTC.Larger,longerandhigher-qualityhumancontrolledfeedingtrialsareneededtoconfirmtheseresults.
Keywords:
fructose,cholesterol,meta-analysis
INTRODUCTION
Hyperlipidemiaisacommonriskfactorforcoronaryheartdisease(CHD),with44.4%ofadultsintheUnitedStateshavingabnormalTCvaluesand32%havingelevatedLDL-Clevels
(1).Comparedtosubjectswithnormalbloodlipid,thosewithhyperlipidemiahavea3-foldriskofheartattacks
(2).LifestylemodificationshouldbeinitiatedinconjunctionbothprimaryandsecondarypreventionofCHD.Moreconsiderationexistsastowhatconstituteshealthyeating.
Fructoseisthemostnaturallyoccurringmonosaccharide,andhasbecomeamajorconstituentofourmoderndiet.Fruit,vegetables,andothernaturalsourcesprovidenearlyone-thirdofdietaryfructose,andtwo-thirdscomefrombeveragesandfoodsinthediets(eg,candies,jam,syrups,etc)(3).Fructoseispreferredbymanypeople,especiallythosewithdiabetesmellitusbecauseofitslowglycemicindex(23%versusglucose100%)(4).Afterintestinaluptake,fructoseismainlyremovedfromthebloodstreambytheliverinaninsulin-independentmanner,andisusedforintrahepaticproductionofglucose,fattyacidsorlactate.Cross-sectionalstudiesinhumansuggestthatexcessivefructoseconsumptioncanleadtoadversemetaboliceffects,suchasdyslipidemiaandincreasedvisceraladiposity(5-7).TheDietaryGuidelinesforAmericans,2010,pointoutthatitislackofsufficientevidencetosetatolerableupperintakeofcarbohydratesforadults(8).AlthoughTheCandianDiabetesAssociationsuggestsconsumptionofnomorethan60gofaddedfructoseperdaybypeoplewithdiabetesforitstriglyceride-raisingeffect(9),thethresholddoseoffructoseatwhichtheadverseinfluenceoncholesteroliscontroversial.
Todeterminetheeffectoffructoseoncholesterol,asubstantialnumberofclinicaltrialshavebeenperformedonadulthumanswithdifferenthealthstatus(diabetic,obese,overweight,hyperinsulinemic,impairedglucose-tolerantandhealthy).Thesetrialsusedvariousintakelevelsoffructoseanddifferentprotocols.Thus,itisdifficulttoreachaconsistentconclusionacrossthesestudies.Therefore,weconductedasystematicreviewofthescientificliteratureandmeta-analysisofcontrolledfeedingtrialstoevaluatetheeffectofisocaloricoralfructoseexchangeforcarbohydratesoncholesterolandtoclarifytheactivefactorsoffructose.
MaterialsandMethods
Thismeta-analysisfollowedthePreferredReportingItemsforSystematicReviewsandMeta-analyses(PRISMA)criteria(10).
Searchstrategy.WesearchedPubMed(http:
//www.ncbi.nlm.nih.gov/pubmed;
from1966toDecember2012),Embase(;
from1966toDecember2012)andtheCochraneLibrarydatabase(http:
//www.cochrane.org)byusingthefollowingsearchterms:
fructoseand(lipemiaorlipaemiaorlipidsorcholesterolor“totalcholesterol”or“LDLcholesterol”or“HDLcholesterol”)inEnglish.WealsosearchedChinaNationalKnowledgeInfrastructure()andWangfangdatabase()inChineseaccordingtothesearchstrategy.Thesearchwasrestrictedtoreportsoftrialsonhumans.
Studyselection.Allclinicaltrialsusingfructoseandindexedwithintheabovedatabaseswerecollected.Twoindependentreviewers(T.A.,R.C.Z)screenedtheabstractsandtitlesforinitialinclusion.Ifthiswasnotsufficient,fulltextsarticleswereobtainedandreviewedbyatleasttwoindependentreviewers(T.A.,R.C.Z,Q.Z.,Y.H.).Thereferencelistsofretrievedarticlesalsousedtosupplementthedatabase.Anydisagreementswereresolvedthroughdiscussion.Weincludedcontrolledfeedingtrialsinvestigatingthechroniceffectoffructoseonbloodcholesterol,frombothrandomizedandnonrandomizedstudies,iftheymetthefollowingcriteria:
subjectsmusthavebeenadministeredfructoseforatleast2weeks;
studiesinvestigatedtheeffectoforalfree(unbound,monosaccharide)fructosewhencomparedwithisocaloriccontroldietwithanothercarbohydrateinplaceoffructose;
studieswereperformedinhumanadultswitheitheraparallelorcrossoverdesign;
subjectsinbothexperimentalgroupsandcontrolgroupswereinstructedtoconsumeisocaloricdiets.Ifthestudyreportedanycomparisons,weincludedallsuchcomparisonsinthemeta-analysis.
Dataextractionandqualityassessment.Tworeviewers(T.A.,R.C.Z)independentlyextractedrelevantdatafromeligiblestudies.Disagreementswereresolvedbyoneofthetwoauthors(Y.H.Z.,J.Z.).Thesedataincludedinformationonstudyfeatures(author,yearofpublication,studydesign,randomization,blinding,samplesize,comparator,fructoseform,dose,follow-upandmacronutrientprofileofthebackgrounddiet),participantcharacteristics(gender,ageandhealthystatus)andbaselineandfinalconcentrationsornetchangesoftotalcholesterol,LDL-CandHDL-C.Datainitiallyextractedwereconvertedtosysteminternationalunit(eg,TC:
1mmol/Lconvertedto38.6mg/dL).Formulti-armstudies,onlyinterventiongroupsthatmetinclusioncriteriawereusedinthisanalysis.Ifbloodlipidconcentrationsweremeasuredseveraltimesatdifferentstagesoftrials,onlyfinalrecordsoflipidconcentrationsattheendofthetrialswereextractedforthismeta-analysis.
ThequalityofeachstudywasassessedwiththeHeylandMethodologicalQualityScore(MQS)(11),generalizedasfollows:
randomization;
analysis;
blinding;
patientselection;
comparabilityofgroupsatbaseline;
extentoffollowup;
treatmentprotocol;
co-intervention;
outcomes.Thehighestscoreforeachareawastwopoints.Highernumbersrepresentedabetterquality(MQS≥8).
Datasynthesis.StatisticalanalyseswereperformedwithStatasoftware(version11.0;
StataCorporation,TX,USA)andREVMANsoftware(version5.2;
CochraneCollaboration,Oxford,UnitedKingdom).Separatepooledanalyseswereconductedbyusingthegenericinversevariancerandom-effectsmodelsevenwheretherewasnoevidenceofbetween-studyheterogeneitybecausethesemodelsgivemoreconservativesummaryeffectestimatesinthepresenceofundetectedresidualheterogeneitythanfixed-effectsmodels.Thedifferentchangesfrombaselinebetweenfructoseandcarbohydratecomparatorsfortotalcholesterol,LDLcholesterolandHDLcholesterolwereusedtoestimatetheprincipleeffect.WeappliedpairedanalysestoallcrossovertrialsaccordingtothemethodsofElbourneandcolleagues(12).Weightedmeandifferencesoffructoseconsumptiononcholesterolconcentrationsandcorresponding95%CIswerecalculated.A2-sidedPvalue<
0.05wassetasthelevelofsignificanceforaneffect.Thevariancesfornetchangesinserumcholesterolwereonlyreporteddirectlyintwotrials(29,31).Wecalculatenetchangesfor