用QUADAS-2作为文献质量评价范例

MAGNETICRESONANCEMRdiffusionimagingforpreoperativestagingofmyometrialinvasioninpatientswithendometrialcancer:asystematicreviewandmeta-analysisAnitaAndreano&GildaRechichi&PaolaRebora&SandroSironi&MariaGraziaValsecchi&StefaniaGalimbertiReceived:31October2013/Revised:27January2014/Accepted:19February2014#EuropeanSocietyofRadiology2014AbstractObjectivesTocomparethediagnosticaccuracyofdynamiccontrast-enhanced(DCE)anddiffusion-weighted(DW)MRimagingindetectingdeepmyometrialinvasioninendometrialcancer,usingsurgical-pathologicalstagingasreferencestandard.MethodsAftersearchingawiderangeofelectronicdatabasesandscreeningtitles/abstracts,weobtainedfullpapersforpotentiallyeligiblestudiesandevaluatedaccordingtopredefinedinclusioncriteria.Qualityassessmentwasconduct-edbyadaptingtheQualityAssessmentofDiagnosticAccu-racyStudies-2(QUADAS-2)checklist.Fromeachstudy,weextractedinformationondiagnosticperformanceofDWandDCEsequences.Afterexploringheterogeneity,weadoptedabivariategeneralizedlinearmixedmodeltocomparetheeffectofthetwoMRsequencesjointlyonsensitivityandspecificity.ResultsNinestudies(442patients)wereconsidered.Signifi-cantevidenceofheterogeneitywasfoundonlyforspecificity,bothinDWandDCEimaging(I2=70.8%and70.6%).PooledsensitivityofDWandDCEwas0.86andspecificitydidnotsignificantlydiffer(p=0.16)betweenthetwosequences(DW=0.86andDCE=0.82).Nodifferencewasfoundbetween3-Tand1.5-TMR.Therewasnoevidenceofpublicationbias.ConclusionsMRdiagnosticaccuracyinpresurgicaldetectionofdeepmyometrialinfiltrationinendometrialcancerishigh.DCEandDWimagingdonotdifferinsensitivityandspecificity.KeyPoints�Myometrialinvasionisthemostimportantmorphologicalprognosticfeatureofendometrialcancer�MRdiagnosticaccuracyinpresurgicaldetectionofdeepmyometrialinfiltrationishigh�MRexaminationincludingT2andDCEimagingisconsid-eredthereferencestandard�DWimaginghasbeenincreasinglyemployedwithheteroge-neousresults�Thismeta-analysisshowsthatDCEandDWdonotdifferindiagnosticaccuracyKeywordsEndometrialneoplasms.Magneticresonanceimaging.Diffusionmagneticresonanceimaging.Meta-analysis.ReviewAbbreviationsandacronymsADCapparentdiffusioncoefficientAICAkaikeinformationcriterionCIconfidenceintervalCTcomputedtomographyDCEdynamiccontrast-enhancedDORdiagnosticoddsratioDWdiffusion-weightedESSeffectivesamplesizeFIGOInternationalFederationofGynecologyandObstetricsGLMMgeneralizedlinearmixedmodelLR+andLR−positiveandnegativelikelihoodratioElectronicsupplementarymaterialTheonlineversionofthisarticle(doi:10.1007/s00330-014-3139-4)containssupplementarymaterial,whichisavailabletoauthorizedusers.A.Andreano:P.Rebora:S.Sironi:M.G.Valsecchi(*):S.GalimbertiCenterofBiostatisticsforClinicalEpidemiology,DepartmentofHealthSciences,UniversityofMilano-Bicocca,ViaCadore48,Monza,MB20900,Italye-mail:[e-mail]:S.SironiDepartmentofRadiology,S.GerardoHospital,ViaPergolesi33,Monza,MB20900,ItalyEurRadiolDOI10.1007/s00330-014-3139-4MRmagneticresonanceQUADAS-2QualityAssessmentofDiagnosticAccuracyStudies-2ROCreceiveroperatingcharacteristicSNRsignaltonoiseratioUSultrasoundIntroductionEndometrialcanceristhesixthmostfrequentcancerinwom-enworldwide,withabout290,000newcasesand74,000deathsin2008[1].Itisthemostcommongynaecologicalmalignancyinwesterncountrieswith49,560estimatednewcasesin2013intheUSAand98,900in2012inEurope[2,3].Well-knownnegativeprognosticfactorsarehightumourgrade,deepmyometrialinvasion(≥50%myometrialthick-ness),lymphovascularspaceinvasion,non-endometrioidhis-tology,andcervicalstromainvolvement[4].Amongthose,themostimportantsinglemorphologicalprognosticfeatureisdepthofmyometrialinvasionwithaone-halfcut-off,whichdividesthecurrentstageIoftheInternationalFederationofGynecologyandObstetrics(FIGO)stagingsystemintoIaandIb[5,6].Deepmyometrialinvasionisassociatedwithbothpelviclymphnodeinvolvementandextensionintotheparametrium[7].Toassessthedepthofmyometrialinvasion,anumberofimagingprocedureshavebeenapplied,includingultrasound(US),computedtomography(CT)andmagneticresonance(MR).MRisconsideredsuperiortoUSandCTfortheevaluationofmyometrialinvasion[8].EssentialsequencestodeterminetheextentofmyometrialinvasionarehighspatialresolutionT2imagesperformedonmultiplesplanes[9].However,thedetectionofdeepmyometrialinvasionatnon-contrastMRhasbeenassociatedwithsomepitfalls,includingthepresenceoflargeleiomyomas,adenomyosis,andexten-sionofthetumourintothecornua[10].Dynamiccontrast-enhanced(DCE)MRimaginghelpsindeterminingmyometrialinvasioninnon-easilyinterpretablecasesonT2imaging[9].Atpresent,aMRexaminationincludingT2andDCEimagingisconsideredthereferencestandardforpreoperativedetectionofdeepmyometrialinva-sioninwomenwithendometrialcancer[9,11,12].Neverthe-less,somestudiesreportedsensitivitiesandspecificitiesaslowas72%and44%andamodestinterobserveragreementforDCEMR[13,14].Diffusion-weighted(DW)imagingisincreasinglyusedasanadd-ontoT2andDCEMRimaging[15,16],evenifDWimagingisnotyetincludedincurrentimagingguidelinesupdatedby2011[9,12].DWsequencesareabletomeasuretherandommotionofwaterintissues,thushelpingtocharacteriz