Review
First-trimester
ultrasound
determination
of
chorionicity
in
twin
gestations
using
the
lambda
sign:
a
systematic
review
and
meta-analysis
G.M.
Maruotti,
G.
Saccone,
M.
Morlando,
P.
Martinelli
*
DepartmentofNeuroscience,ReproductiveSciencesandDentistry,SchoolofMedicine,UniversityofNaplesFedericoII,Naples,Italy
Contents
Introduction... 67
Materialsandmethods... 67
Results... 67 Comment... 68 Funding ... 70 Conflictofinterest... 70 References... 70 A R T I C L E I N F O Articlehistory: Received22January2016 Accepted21April2016 Keywords: Multiplegestations
Twin–twintransfusionsyndrome
Ultrasound
Firsttrimester
A B S T R A C T
Objective:Toevaluatetheaccuracyoffirst-trimestersonographicdeterminationofchorionicityintwin gestationsusingthelambdasign.
Studydesign: Electronicdatabases(MEDLINE,PROSPERO,Scopus,ClinicalTrials.gov,EMBASE, Science-direct)weresearchedfromtheirinceptionuntilApril2016.We includedonlystudyassessingthe accuracylambdasigninpredictionofmonochorionicityinthefirsttrimester.Forestplotsforpooled sensitivityandspecificitywith95%confidenceintervals(CI)weregenerated.Inaddition,symmetric summaryreceiver-operatingcharacteristiccurveswereplotted.Theareaunderthecurve(AUC)wasalso computedtoevaluatetheoverallaccuracyofthediagnostictest.
Results:Ninestudies,including2292twins,wereanalysed.Inallofthesestudies,identificationofthe lambdasignwasusedtodiagnosechorionicityonreal-timeB-modeimaging.Twinswereclassifiedas monochorioniciftherewasasingleplacentalmassintheabsenceofthelambdasign,anddichorionicif therewasasingleplacentalmassbutthelambdasignwaspresentortheplacentaswerenotadjacentto each other. In all nine studies, placental histology or discordant fetal sex wereused to confirm chorionicity.Pooledresultsfromthemeta-analysisshowedthatsensitivityofthepresenceofthelambda signinthepredictionofdichorionicitywas99%(95%CI98–100%),andspecificitywas95%(95%CI92– 97%).Pooledsensitivityoftheabsenceofthelambdasigninthepredictionofmonochorionicitywas96% (95%CI92–98%)andpooledspecificitywas99%(95%CI98–99%).TheAUCfordiagnosticaccuracywas 0.99,andsuggestedveryhighdiagnosticaccuracy.
Conclusion:Thelambdasignpredictschorionicitywithahighdegreeofaccuracybefore14weeksof gestation.Presenceofthelambdasignindicatesdichorionicity,andabsenceofthelambdasignindicates monochorionicity. All hospitals should encourage departments providing ultrasound services to determine chorionicity when examining women with twin pregnancies in the first trimester.As determinationofchorionicityismostaccuratebefore14weekswhentheamnionandchorionhavenot yetfused,thefirst-trimesterscanintwinpregnancyisparamount.
ß2016ElsevierIrelandLtd.Allrightsreserved.
* Corresponding author.Tel.:+390817462966;fax:+390817462966.
E-mailaddress:martinel@unina.it(P.Martinelli).
ContentslistsavailableatScienceDirect
European
Journal
of
Obstetrics
&
Gynecology
and
Reproductive
Biology
j o urn a l hom e pa ge : ww w. e l s e v i e r. c om/ l o ca t e / e j ogr b
http://dx.doi.org/10.1016/j.ejogrb.2016.04.023
Introduction
TheincidenceoftwingestationsintheUSAhasincreasedover thepast decades[1].Compared withdichorionic twins, mono-chorionic twins have inherently different complication rates, includingfetal loss,fetal anomalies,intrauterinegrowth restric-tion, prematurity, twin anaemia polycythaemia sequence and twin–twintransfusionsyndrome[2].Therefore,accurate determi-nation of chorionicity is vital when managing women with multiplegestations.Thesonographicdeterminationofchorionicity isbasedonthenumberofplacentalsites,thevisualizationofan intertwinmembrane(ITM),andtheidentificationoftheso-called ‘lambda’or‘twin-peak’signontheonehandorthe‘T’signonthe other[3].
Theaimofthis systematic reviewand meta-analysiswasto evaluatetheaccuracyoffirst-trimesterultrasounddetermination ofchorionicityintwingestationsusingthelambdasign. Materialsandmethods
Thisreviewwasperformedaccordingtoaprotocoldesigneda priori and recommended for systematic review [4]. Electronic databases (MEDLINE, PROSPERO, Scopus, ClinicalTrials.gov, EMBASE,Sciencedirect,Cochrane Library, Scielo)were searched fromtheirinceptionuntilApril2016withnolanguagelimitations. Thefollowingsearchtermswereused:‘twin’,‘multiple’, ‘amnio-nicity’, ‘chorionicity’, ‘gestations’, ‘ultrasound’, ‘sonographic’, ‘pregnancy’, ‘transvaginal’, ‘2D’, ‘lambda’, ‘accuracy’, ‘cohort’, ‘case–control’, ‘prediction’, ‘first trimester’, ‘obstetric’, ‘triplet’ and‘sign’.Norestrictionswereappliedforlanguageor geographi-callocation.Inaddition,thereferencelistsofallidentifiedarticles wereexaminedtoidentifyanystudiesthatwerenotcapturedby theelectronicsearch.Theelectronicsearchandstudyeligibility wereassessedindependentlybytwoauthors(GS,PM).Differences werediscussedandconsensuswasreached.
This study considered randomized controlled trials, case– controlstudiesandcohortstudies.Studieswereincludedifthey reporteddataallowingconstructionofa22table.Onlystudies thatassessedtheaccuracyofthelambdasignforthepredictionof chorionicity in twin gestations during the first trimester were included.Studiesthatevaluatedtheaccuracyofothersigns(e.g. ITM, gestational sac number)for the prediction of chorionicity were excluded. Studies on triplet and high-order multifetal pregnancies were also excluded. The primary outcome of this meta-analysiswasplannedaprioriastheaccuracyofthelambda signforthepredictionofchorionicityintwingestations.
Dataabstractionand methodologicalquality of theincluded studieswerecompletedbytwoindependentinvestigators(GMM, PM).Eachinvestigatorabstracteddatafromeachstudy indepen-dently. Data from each eligible study were extracted without modificationoftheoriginaldatatocustom-madedatacollection forms. Disagreementswereresolved by consensuswitha third reviewer(GS).Allauthorsoftheoriginalstudieswerecontactedfor missingdataifpossible.
ThequalityofeachincludedstudywasassessedusingQuality AssessmentofDiagnosticAccuracyStudies(QUADAS)criteria[5]. Eachitemisscoredas‘yes’,‘no’or‘unclear’(ifthereisinsufficient informationtomakeanaccuratejudgement)[5].Aqualityscore wascalculated,definedasthetotalnumberofitemscategorizedas ‘yes’amongthesevenitemsoftheQUADAScriteria[5].Astudy withascore5outof7wasjudgedtohavelowoveralllowriskof bias[5].
The meta-analysis was reported in accordance with the Preferred Reporting Item for Systematic Reviews and Meta-analyses (PRISMA) statement [6]. Before data extraction, the protocol for this review was registered with the PROSPERO
InternationalProspectiveRegisterofSystematicReviews (Regis-trationNo.CRD42015027286)followingthePRISMAguidelinesfor protocols(PRISMA-P)[7].
Forallincludedstudies,a22tablewasconstructedto cross-classifytheultrasoundmeasurementofthelambdasignandthe predictionofmonochorionicity.Forestplotsforpooledsensitivity (i.e.detectionrate)andspecificityoffirst-trimesterlambdasignfor thepredictionofchorionicityintwinsweregenerated,with95% confidence intervals (CI). Additionally, symmetric summary receiver-operating characteristic (SROC) curves were plotted. The area under the curve (AUC) and the Q* index were also computedtoevaluatetheoverallaccuracyofthediagnostictest. TheAUCofanSROCcurveisameasureoftheoverallperformance ofadiagnostictesttodifferentiateaccuratelybetweencaseswith andwithouttheconditionofinterest.TheQ*indexisdefinedby thepointatwhichsensitivityandspecificityareequal,whichis closesttotheidealtop-leftcorneroftheSROCspace.Bothvalues rangebetween0and1,withhighervaluesindicatingbettertest performance[8].Thefollowingguidelineshavebeensuggestedfor interpretation of AUC values: 0.5AUC<0.7, low accuracy; 0.7AUC<0.9, moderate accuracy; and 0.9AUC1, high accuracy[8].TheauthorsplannedtoassessAUCforthesensitivity analysisaccordingtothestudydesignoftheincludedstudies.
The degree of between-study heterogeneity was evaluated usingtheI2statistic,whichrepresentsthepercentageof between-studyvariationthat isduetoheterogeneityratherthanchance. Avalue30%indicatesasubstantiallevelofheterogeneity[4,9]. PotentialpublicationbiaswasassessedstatisticallyusingBegg’s andEgger’stests[9].
Dataanalysis wascompletedindependently by authors(GS, PM) usingMeta-DiSc1.4 (HospitalUniversitarioRamonyCajal, Madrid,Spain).Thecompletedanalyseswerecompared,andany differenceswereresolvedbyreviewofthefulldata.
Results
ThestudyflowchartisshowninFig.1.Thirteenstudieswere assessed for eligibility [10–22].Four studiesthat evaluatedthe accuracyofITMforthepredictionofamnionicitywereexcluded
[10–12,17].Assuch, ninestudies,including2292 pairof twins, wereanalysed[13–16,18–22].
Begg’sandEgger’stestsshowednosignificantpublicationbias (p=0.62andp=0.71,respectively).Thestatisticalheterogeneity betweentheincludedstudieswaslow(I2=0%).Fig.2showsthe resultsofthequalityassessment.Noneoftheninestudieshada highriskofbiasinpatientselectionandreferencestandard.Most studies(5/9)hadalowoverallriskofbias(i.e.QUADASscore5)
[13,16,18,19,21].
Table1showsthecharacteristicsofthenineincludedstudies. Of the 2292 women with twin gestations, 428 (19%) were monochorionic twins and 1864 (81%) were dichorionic twins. FourstudieswerefromtheUSA[14,15,18,21].Fourstudieswere retrospective cohort studies [13–16] and five were prospective cohortstudies[18–22].Onestudywasasecondaryanalysisofa randomizedtrialofpretermbirth preventionintwins[18].The methodofultrasoundascertainmentwasdescribedclearlyinall ninestudies.Ultrasoundscanswereperformedat<14weeksof gestation. In sevenstudies, transvaginalultrasound scans were performed using a 5-MHz transducerforB-mode imaging bya doctorwithappropriateexperienceinearly-pregnancyultrasound assessment [14,15,18–22]. In theothertwo studies, ultrasound scans were performed transabdominally [13,16]. In one study, where views were suboptimal with the transabdominal ultra-sound, the examination was performed transvaginally with an 8-MHztransducer[16].
Inallninestudies,identificationofthelambdasignwasusedto diagnose chorionicity on real-time B-mode imaging. If the placentaswerefusedandawedge-shapedjunctionwasobserved (i.e. lambda sign), this was interpretated as extension of the chorionic tissue into the base of the ITM, and therefore a dichorionictwinpregnancy.Twinswereclassifiedas monochor-ioniciftherewasa singleplacental massintheabsenceofthe lambdasign,anddichorioniciftherewasasingleplacentalmass butthelambdasignwaspresentortheplacentaswerenotadjacent to each other. In the absence of an ITM, the pregnancy was considered to be monoamniotic. In all nine studies, placental histologyordiscordantfetalsexwereusedtoconfirmchorionicity. A 22 table was constructed for all nine studies for the prediction of chorionicity using the lambda sign in the first-trimesterultrasoundscan.
Pooledresultsfromthemeta-analysisshowedthatsensitivity of the presence of the lambda sign for the prediction of dichorionicityranged from98% to100%,and specificity ranged from82%to100%.Pooledsensitivity(i.e.detectionrate)was99% (95%CI98–100%)andpooledspecificitywas95%(95%CI92–97%). Sensitivityoftheabsenceofthelambdasignforthepredictionof monochorionicityrangedfrom79%to100%,andspecificityranged from97%to100%.Pooledsensitivitywas96%(95%CI92–98%)and pooledspecificitywas99%(95%CI98–99%).
TheAUCfordiagnosticaccuracywas0.99,whichsuggestedvery highdiagnosticaccuracy (Fig.3). Accuracywasalsohighinthe sensitivityanalysisofretrospectivestudiesalone(AUC=0.98)and prospectivestudiesalone(AUC=0.98).
Comment
This systematic review and meta-analysis showed that the lambdasignhasveryhighdiagnosticaccuracyforthepredictionof chorionicity.Thepooleddetectionratewas99%.
Thisstudyhasseveralstrengths.Totheauthors’knowledge, this is the first meta-analysis to evaluate the accuracy of the lambdasignforthepredictionofchorionicity;no similar meta-analyseswerefoundduringthesystematicreview.Thenumberof womenincludedinthisstudywashigh,andtheoverallriskofbias oftheincludedstudieswaslow.Alloftheincludedstudieshadthe
Fig.1.Studyflowdiagram.
Fig.2.Reviewauthors’judgementofriskofbiasandapplicabilityconcernsbasedonQualityAssessmentofDiagnosticAccuracyStudiestool.(A)Summaryofriskofbiasfor
eachstudy:+,lowriskofbias; ,highriskofbias;?,unclearriskofbias.(B)Riskofbiasgraphabouteachriskofbiasitempresentedaspercentagesacrossincludedstudies.
same primary outcome (i.e. prediction of chorionicity). The protocol of this review was registered a priori on PROSPERO. Statisticaltestsshowednosignificantpotentialpublicationbias. Finally,statisticalheterogeneitybetweentheincludedstudieswas low,withnoinconsistencyinthepooledresults(I2=0%).
The limitations of this study are mostly inherent to the limitationsoftheincludedstudies.Thequalityofthefindingsis dependentonthequalityoftheprimarystudiesincluded.Allofthe included studies were cohort studies. The generalizability and externalvalidityofthesefindingsmaybelimitedduetothequality of ultrasound equipment at these institutions and the patient populationevaluated.Noadjustmentsforpotentialconfounders weremadebytheoriginalstudies.Subgroupanalysisaccordingto gestationalagewasnotfeasible;onlyBlumenfeldetal.included second-trimestermeasurements,butdidtheynotstratifythedata by gestational age. However, they showedthat for each week
increaseingestationalage,theodds ofmisclassificationroseby 10%[18].HierarchicalSROCcurveswerenotassessed.
Twins accountforapproximately1%ofall pregnancies,with two-thirdsoftwinsbeingdizygotic(i.e.non-identical)and one-thirdbeingmonozygotic(i.e.identical)[1].Overthelast20years, the rate of twinning has increased, and this increase is most markedfordizygotictwins[1].Two-thirdsoftheincreaseintwins isduetotheuseofassistedreproductiontechniques,andone-third oftheincreaseisduetoincreasingmaternalage[1].
Chorionicity refersto thetype of placentationand does not reflectzygosity,whichdenotesthetypeofconception. Chorioni-city, rather than zygosity, is the main factor that determines pregnancyoutcome[23,24].Indizygotictwins,eachfetushasits own placenta and amniotic sac, and therefore dizygotic twins alwayshave a dichorionicplacenta[23]. Inmonozygotic twins, there may be sharing of the same placenta (monochorionic),
Table1
Characteristicsoftheincludedstudies.
Monteagudo 1994[14] Copperman 1995[15] Sepulveda 1996[13] Carroll 2002[16] Menon 2005[22] Lee 2006[21] Bora 2008[19] Dias 2011[20] Blumenfeld 2014[18]
Location USA USA UK Ireland Malaysia USA UK UK USA
Studydesign PC PC PC PC RC RC RC RC RC
Numberoftwinsincluded 64 47 279 150 463 247 67 613 362
Numberofmonochorionictwinsincluded 9 3 63 34 50 49 14 146 60
Numberofdichorionictwinsincluded 55 44 216 116 413 198 53 467 302
Gestationalageatscan(weeks) 5–14 8 10–14 10–14 10–14 <14 7–9 11–14 <14
Placentalpathologicalexamination Yes Yes Yes Yes Yes Yes Yes Yes Yes
PC,prospectivecohort;RC,retrospectivecohort.
Fig.3.Symmetricsummaryreceiveroperatingcharacteristicscurvewith95%confidenceintervalfortheaccuracyoffirsttrimestersonographicdeterminationofchorionicity
amnioticsac(monoamniotic)orevenfetalorgans(conjoinedor Siamese). In monozygotic twins, the type of placentation will dependonwhenthezygotedivides.Ifthedivisionoccurswithin thefirst3daysoffertilization,dichorionic–diamniotictwinsoccur; if the split occurs 4–7 days after fertilization, the result is monochorionic–diamniotictwins;andifthedivisionoccurs8–12 daysafterfertilization,monochorionic–monoamniotictwinsoccur
[23].Approximately25% ofallmonozygotictwingestationsare dichorionic[1].
While all twin pregnancies need increased surveillance comparedwithsingletonpregnancies,monochorionictwinsneed particularfollow-up[2,24].Twinpregnancieshavea monochor-ionicplacentain20%ofcases,and perinatalmortalityistwo to threetimeshigherthanindichorionictwins[25,26].
Indichorionictwins,theITMiscomposedofacentrallayerof chorionic tissue sandwiched between two layers of amnion, whereas in monochorionic twins, no chorionic layer is present
[3,24].Determinationofchorionicityhelpstoidentifythosetwins athigherrisk,andenablesmonitoringtobetailoredfortheearly diagnosis of complications [1,25–27]. Being able to predict chorionicityintwingestationshasseveralpotentialbenefits,as failuretodetectchorionicitymaybeassociatedwithhigherratesof neonatalmorbidityandmortality[2,24,25,28].Thisinformation may assist birth providers when planning staff coverage [29]. These data may also be helpfulfor women choosing between plannedcaesareandeliveryandwaitingforspontaneouslabourto attempt vaginal delivery [29,30]. For pregnant women, this informationmayhelpthemtoarrangetheirsocialactivitiesand dealwithanxiety.
Insummary,thelambdasignpredictschorionicitywithahigh degreeofaccuracybefore14weeksofgestation.Theresultsofthis study support the observation that the lambda sign indicates dichorionicity, and the absence of the lambda sign indicates monochorionicity.Asthereisregressionofthechorionlaeveand thelambdasignbecomesmoredifficulttoidentifywithadvancing gestation[3,13,18,23,30],allhospitalsshouldencourage depart-ments providing ultrasound services to determine chorionicity when examining women with twin pregnancies in the first trimester.
As determination of chorionicityis mostaccurate before 14 weekswhentheamnionandchorionhavenotyetfused,the first-trimesterscanintwinpregnancyisparamount.
Funding None.
Conflictofinterest Nonedeclared. References
[1]MartinJA,HamiltonBE,OstermanMJ,CurtinSC,MatthewsTJ.Births:finaldata for2013.NatlVitalStatRep2015;64:1–68.
[2]DanonD,SekarR,HackKE,FiskNM.Increasedstillbirthinuncomplicated monochorionictwinpregnancies:asystematicreview andmeta-analysis. ObstetGynecol2013;121:1318–26.
[3]SalomonLJ,AlfirevicZ,BilardoCM,etal.ISUOGpracticeguidelines: perfor-manceoffirst-trimesterfetalultrasoundscan.UltrasoundObstetGynecol 2013;41:102–13.
[4]HigginsJPT,GreenS.Cochranehandbookforsystematicreviewsof
interven-tions,Version5.1.0.Oxford:TheCochraneCollaboration;2011,March,
Avail-ableat:www.cochrane-handbook.org[accessed15.10.15].
[5]WhitingPE,RutijesAW,WestwoodME,etal.QUADAS-2:arevisedtoolforthe qualityassessmentofdiagnosticaccuracystudies.AnnInternMed2011;155: 529–36.
[6]MoherD,LiberatiA,TetzlaffJ,AltmanDG,PRISMAGroup.Preferredreporting itemsforsystematicreviewsandmeta-analyses:thePRISMAstatement.JClin Epidemiol2009;62:1006–12.
[7]ShamseerL,MoherD,ClarkeM,etal.Preferredreportingitemsforsystematic reviewandmeta-analysisprotocols(PRISMA-P)2015:elaborationand expla-nation.BMJ2015;349:g7647.
[8]Walter SD. Propertiesof the summeryreceiver operatingcharacteristics (SROC)curvefordiagnostictestdata.StatMed2002;21:1237–56.
[9]ZhouY,DendukuriN.Statisticsforquantifyingheterogeneityinunivariateand bivariatemeta-analysesofbinarydata:thecaseofmeta-analysesofdiagnostic accuracy.StatMed2014;33:2701–17.
[10]WatsonWJ,ValeaFA,SeedsJW.Sonographicevaluationofgrowthdiscordance andchorionicityintwingestation.AmJPerinatol1991;8:342–4.
[11]KurtzAB,WapnerRJ,MataJ,JohnsonA,MorganP.Twinpregnancies:accuracy offirst-trimesterabdominalUSinpredictionchorionicityandamnionicity. Radiology1992;185:759–62.
[12]TannirandornY,PhaosavasdiS.Accuracyofultrasonographiccriteriaforthe prenataldiagnosisofplacentalamnionicityandchorionicityintwin gesta-tions.JMedAssocThail1993;76:190–5.
[13]SepulvedaW,SebireNJ,HughesK,OdiboA,NicolaidesKH.Thelambdasignat 10–14weeksofgestationasapredictorofchorionicityintwinpregnancies. UltrasoundObstetGynecol1996;7:421–3.
[14]MonteagudoA,Timor-TritschIE,SharmaS.Earlyandsimpledeterminationof chorionicandamniotictypeinmultifetalgestationsinthefirstfourteenweeks by high-frequency transvaginal ultrasonography. Am J Obstet Gynecol 1994;170:824–9.
[15]CoppermanAB,KaltenbacherL,WalkerB,etal.Earlyfirst-trimester ultra-soundprovidesawindowthroughwhichthechorionicityoftwinscanbe diagnosedinaninvitrofertilization(IVF)population.JAssistReprodGenet 1995;12:693–7.
[16]CarrollSG,SoothillPW,Abdel-FattahSA,etal.Predictionofchorionicityin twinpregnanciesat10–14weeksofgestations.BJOG2002;109:182–6.
[17]WanJJ,SchrimmerD,TacheV,etal.Currentpracticeindetermining amnio-nicityandchorionicityinmultiplegestations.PrenatDiagn2011;31:125–30.
[18]BlumenfeldYJ,MomirovaV,RouseDJ,etal.Accuracyofsonographic chor-ionicityclassificationintwingestations.JUltrasoundMed2014;33:2187–92.
[19]BoraSA,PapageorghiouAT, BottomleyC, KirkE,BourneT.Reliability of transvaginalultrasonographyat7–9weeks’gestationinthedetermination of chorionicity and amnionicity in twin pregnancies. Ultrasound Obstet Gynecol2008;32:618–21.
[20]DiasT,ArcangeliT,BhideA,etal.First-trimesterultrasounddetermination of chorionicity in twin pregnancy. Ultrasound Obstet Gynecol 2011;38: 530–2.
[21]LeeYM,Cleary-GoldmanJ,ThakerHM,SimpsonLL.Antenatalsonographic predictionoftwinchorionicity.AmJObstetGynecol2006;195:863–7.
[22]MenonDK.Aretrospectivestudyoftheaccuracyofsonographicchorionicity determinationintwinpregnancies.TwinResHumGenet2005;8:259–61.
[23]ShettyA,SmithAP.Thesonographicdiagnosisofchorionicity.PrenatDiagn 2005;25:735–9.
[24]AmericanCollegeofObstetriciansandGynecologists,Societyfor Maternal-FetalMedicine.ACOGPracticeBulletinNo.144:Multifetalgestations:twin, triplet,and high-ordermultifetal pregnancies. Obstet Gynecol2014;123: 1118–32.
[25]BernirschkeK,KimCK.Multiplepregnancy.NEnglJMed1973;288:1276–84.
[26]BernirschkeK,KimCK.Multiplepregnancy.NEnglJMed1973;288:1329–36.
[27]CarrollSG,Tyfield L,ReeveL, etal.Iszygosityorchorionicitythemain determinantoffetaloutcomeintwinpregnancies? AmJ ObstetGynecol 2005;193:757–61.
[28]SacconeG,BerghellaV.Planneddeliveryat37weeksintwins:asystematic reviewandmeta-analysisofrandomizedcontrolledtrials.JMaternFetal NeoantalMed2015;26:15.
[29]BibboC,RobinsonJN.Managementoftwins:vaginalorcesareandelivery?Clin ObstetGynecol2015;58:294–308.
[30]HillLM,CheneveyP,HeckerJ,MartinJG.Sonographicdeterminationoffirst trimestertwin chorionicity andamnionicity. J Clin Ultrasound 1996;24: 305–8.