Midtrimester transvaginal ultrasound cervical length screening for spontaneous preterm birth in diamniotic twin pregnancies according to chorionicity

(1)

Full

length

article

Midtrimester

transvaginal

ultrasound

cervical

length

screening

for

spontaneous

preterm

birth

in

diamniotic

twin

pregnancies

according

to

chorionicity

Amanda

Roman

a

,

Gabriele

Saccone

b,

*

,

Carolynn

M. Dude

c

,

Andrew

Ward

a

,

Hannah

Anastasio

a

,

Lorraine

Dugoff

c

,

Fulvio

Zullo

b

,

Vincenzo

Berghella

a

DivisionofMaternal-FetalMedicine,DepartmentofObstetricsandGynecology,SidneyKimmelMedicalCollegeofThomasJeffersonUniversity,Philadelphia, PA,USA

b_Department_of_{Neuroscience,}_Reproductive_Sciences_and_Dentistry,_School_of_Medicine,_University_of_Naples_Federico_II,_Naples,_Italy c

DivisionofMaternal-FetalMedicine,DepartmentofObstetricsandGynecology,UniversityofPennsylvania,Philadelphia,PA,USA

ARTICLE INFO

Articlehistory: Received8July2018

Receivedinrevisedform2August2018 Accepted3August2018 Availableonlinexxx Keywords: Multiplegestations Prematurity Pretermbirth Pessary Cerclage NICU Progesterone ABSTRACT

Objective:Tocomparethemeantransvaginalultrasound(TVU) cervicallength(CL)atmidtrimester

screeningforspontaneouspretermbirthinasymptomaticmonochorionicdiamnioticversusdichorionic

diamniotictwinpregnancies

Studydesign:Thiswasamulticenterretrospectivecohortstudy.Studysubjectswereidentiﬁedatthetime

ofaroutinesecondtrimesterfetalultrasoundexamat180/7–236/7weeksgestation.Weexcluded

women that received progesterone, pessary, or cerclage. Distribution of CL was determined and

normalitywasexamined.MeanofTVUCL werecomparedbetweenmonochorionicdiamnioticand

dichorionicdiamnioticpregnancies.TherelationshipofTVUCLwithgestationalage(GA)atdeliveryand

incidenceofspontaneouspretermbirth(SPTB)atdifferentTVUCLcutoffswereassessed.Incidenceof

shortTVUCL,deﬁnedasTVUCL30mm,wasalsocalculatedinthetwogroups.

Results:580womenwithdiamniotictwinpregnanciesunderwentTVUCLscreeningbetween180/6and

236/7weeks. 175(30.2%)weremonochorionicdiamnioticpregnancies,and405(69.8%)weredichorionic

pregnancies.Thedemographiccharacteristicsweresimilaronbothgroups.ThemeanGAatTVUCLwas

about20weekinbothgroups.ThemeanTVUCLwassigniﬁcantlylowerinthemonochorionicdiamniotic

(32.810.1)comparedtothedichorionic(34.98.6)group(MD 2.10mm,95%CI 3.91to 0.29).TVU

CL30mmwas16.6%(29/175)inthemonochorionicgroup,and11.9%(48/405)inthedichorionicgroup

(aOR1.48,95%CI1.03–2.43).Twinswithamonochorionicdiamnioticpregnancyhadasigniﬁcantly

higherincidenceofSPTB(53.1%vs44.9%;aOR1.22,95%CI1.22–1.79).ForanygivenCLmeasuredbetween

180–7and236/7weeks,gestationalageatdeliveryformonochorionicdiamnioticpregnancieswas

about2weeksearliercomparedtodichorionicpregnancies(MD-2.1weeks;ANCOVAP<0.001).

Conclusion:Monochorionicdiamniotictwinpregnancieshadahigherrateofspontaneouspretermbirth

than dichorionic diamniotic pregnancies. The higher rate of spontaneous preterm delivery in

monochorionic pregnancies is associated with lower midtrimester TVU CL when compared to

dichorionicpregnancies.

Introduction

Pretermbirth(PTB)isaleadingcauseofperinatalmorbidityand mortality[1]. Over the last few years, cervical assessment has moved from digital examination toultrasound evaluation, and ultrasoundofthecervixhasbeenthefocusofmuchresearch[2–6].

Transvaginal ultrasound (TVU) cervical length (CL) has been assessed in several populations (e.g.women with and without symptomsofpretermlabororprematureruptureofmembranes) to evaluate the risk of spontaneousPTB (SPTB) [7], in women beforeinductionoflabortopredictinductionoutcome[8],andat term topredictthe onset of spontaneouslabor withmoderate degreeofaccuracy[9].AshortTVUCLhasbeenshowntobeagood predictorofSPTBinbothsingletonsandtwins[4];andhasbeen shown to be more accurate than digital examination and fetal ﬁbronectininthepredictionofSPTB[4,10].

* Correspondingauthor.

E-mailaddress:gabriele.saccone.1990@gmail.com(G.Saccone).

https://doi.org/10.1016/j.ejogrb.2018.08.006

ContentslistsavailableatScienceDirect

European

Journal

of

Obstetrics

&

Gynecology

and

Reproductive

Biology

(2)

TwingestationsareatincreasedriskofbothSPTBandindicated PTB (eithermaternal orfetal indications) [1]. For example,the overallrateofPTBat<37,<34and<32weeksintwinpregnancies hasbeenshowntobe56%,17%and9%,comparedwithabout9.8%, 2%and1%insingletonpregnancies,respectively[1].

Overthepastdecadestheincidenceoftwinpregnanciesinthe USAhasincreased,reachinganewhighforthenationof33.7per 1000 total births in the 2015 [1]. Monochorionic pregnancies comprises 20–33% of all twin gestations they have inherently different complication rates when compared with dichorionic pregnancy, including higher risk of fetal demise, congenital anomalies,intrauterinegrowth restriction,twinanemia polycy-themiasequence,twin-twintransfusionsyndrome,andhigherrisk ofspontaneouspretermbirth[11,12].However,dataregardingTVU CLintwinpregnanciesstratiﬁedbychorionicity,andspeciﬁcally studies assessing whether the risk of spontaneous PTB in monochorioniccompared todichorionictwinscanbepredicted byadifferenceinTVUCL,arelimited[13].

Objective

The aim of this study was to compare the TVU CL at midtrimester in screening for SPTB in asymptomatic twins in monochorionic diamniotic compared to dichorionic diamniotic pregnancy.

Methods Studypopulation

Thiswasamulticenterretrospectivecohortstudy.Dataonall consecutiveasymptomatictwinpregnancieswhounderwentTVU CLscreeningatUniversityofNaplesFedericoII(Naples,Italy),at DivisionofMaternalFetalMedicineThomasJeffersonUniversity Hospital (Philadelphia, PA), and at Division of Maternal Fetal MedicineUniversityofPennsylvania(Philadelphia,PA)atthetime ofroutinesecondtrimesterfetalultrasoundexamat180/7–236/7 weeks from January 2014 to January 2017 were included in a dedicateddatabase

Monoamniotictwins,twinpregnancieswithtwin-twin trans-fusionsyndrome,useofvaginalprogesterone,pessaryorcerclage inplace,aswellasmajorfetalmalformationsorgeneticanomalies atthetimeoftheTVUCLwereexcluded.Fetaldemiseorselective reductionofanyofthetwinsbeforedeliverywerealsoexcluded from the analysis. Therefore the analyzed cohort included all consecutiveasymptomatictwinpregnancieswithnormal,viable twinsatthetimeofdeliverywhounderwentmidtrimesterTVUCL screening. Women were divided in two groups according to chorionicity:monochorionicand dichorionic.To avoidselection bias all consecutive twin pregnancies who received TVU CL screeningwereincludedandanalyzed.

Potentialstudysubjectswereidentifiedatthetimeofaroutine secondtrimester fetalultrasoundexamat18 0/7–236/7weeks gestation. Only TVU screening was employed for cervical screening;and onlyone TVU CLmeasurement was performed. Physiciansand sonographerswho performed TVUCLscreening werecertifiedthroughthe FetalMedicineFoundation (FMF) or throughCLEAR.Briefly,themeasurementofCLwasperformedin thesagittalplane,visualizingthefulllengthofthecervicalcanal from the internal os to the external cervical os. At least 3 measurements were obtained and the shortest was recorded [14,15].Chorionicitywasassessedusingthelambdasigninthefirst trimesterand confirmed withplacental analysis atthe time of delivery[11].Womenthatreceivedpessary,cerclage,or proges-teronewereexcluded[16].Accordingtoourprotocol,deliverywas plannedat370/7–386/7weeksfordichorionictwins,andat360/

7–376/7 weeksfor monochorionictwins [17,18]. Indicationfor deliverywasrecorded.

Outcomes

Theprimary outcomewas themeanof TVUCLstratiﬁedby chorionicity.DistributionofCL,normality,andincidenceofshort cervixinseveralcutoffs(30,25,15,10,and5mm)inboth group werecalculated. Receiver-operating characteristics (ROC) curvewas used toassess the CLvaluepredictive for SPTB<32 weeks. Sensitivity, speciﬁcity, positive and negative likelihood ratio(LR+andLR-,respectively)werecalculatedfortheeachcutoff point(30,25,15,10,and5mm).Wealsoassessedtherelationship ofTVUCLwithgestationalageatdeliverybetweenmonochorionic diamnioticanddichorionicpregnancy.

Secondaryoutcomesweregestationalageatdelivery,incidence ofPTBandofSPTB<37,<34,<32and<28weeks,andindication fordelivery.

Data on pregnancy outcomes were obtained from hospital maternity records. In case of PTB, records were examined to determine whether the delivery was medically indicated or spontaneous PTB. SPTB included either spontaneous onset of pretermlabororPPROM.

Dataanalysis

Data are shown as meansstandard deviation (SD), or as number (percentage). Univariate comparisons of dichotomous datawereperformedwiththeuseofthechi-squareorFisherexact test.Comparisonsbetweengroupswereperformedwiththeuseof theMann—WhitneyUtest,totestgroupmedians;andwiththeuse of the T-test to test group means with SD. Outcomes were estimatedwithmultivariateanalyses.

Logisticregression,presentedasunadjustedoddsratio(crude OR)oradjustedoddsratio(aOR)orasmeandifference(MD)with the 95% of conﬁdence interval (CI), was performed. Adjusted analysis was performed to correct data for relevant baseline characteristics. All potentially relevant baseline characteristics were added to the model as covariates. Relevant baseline characteristics included: age, body mass index (BMI), parity, historyofSPTB,andsmoking.Thisanalysiswasperformedtoshow robustnessofourresults[19].

Distributionofcervicallengthwasdeterminedandnormality wasexaminedbytheKolmogorov-Smirnovtest.Wealsoplanned toassesstheincidenceofTVUCL30mminsubgroupanalysisin womenwithandwithoutpriorSPTB.

ROC curve to assess the CL value predictive for SPTB<32 weeks was plotted for both monochorionic diamniotic and dichorionic pregnancies. The area under the curve (AUC) was computedtoevaluatetheoverallperformanceofthediagnostic testaccuracyinpredictionofSPTB_<32weeks[9,20].TheAUCofa ROCcurveisameasureoftheoverallperformanceofadiagnostic test in accurately differentiating those cases with and those withouttheconditionofinterest[9,20].Differencebetweenthe AUCoftheROCcurveforthemonochorionicdiamniotictwinsand the AUC of the ROC curve for the dichorionic twins were calculated by using the DeLong nonparametric test [20]. The DeLong test assessed the standard error of the AUC and the differencebetweenthetwoAUCs[20].

ThecorrelationbetweenCLandgestationalageatdeliveryin monochorionicdiamnioticpregnanciesanddichorionic pregnan-cies was assessed with the use of the Spearman’s correlation coefﬁcient. Comparison of coefﬁcients was done by using the Fisher’s Z-transformation. ANCOVA analysis of covariance was usedtoplotthegenerallinearmodelfortherelationshipbetween CLandgestationalageatdeliveryandtheMDinweeksbetween

(3)

monochorionicanddichorionicwascalculated.Wecalculatedtwo sided p-values. A p-value <0.05 was considered to indicated statistical signiﬁcance. Statistical analysiswas performedusing Statistical Package for Social Sciences (SPSS) v. 19.0 (IBM Inc., Armonk,NY,USA).

Results

Characteristicsofthestudypopulation

Overall, 580 women with twin pregnancies, who met the inclusion criteria and underwent TVU CL screening at our institutions were analyzed. 175 (30.2%) were monochorionic diamniotic pregnancies, and 405 (69.8%) were dichorionic pregnancies.The maternaldemographics werenotsigniﬁcantly differentbetweenmonochorionicvs.dichorionic(Table1).Fifteen women in the monochorionic diamniotic group (8.6%) and 33 womeninthedichorionicgroup(8.1%)hadhistoryofSPTB.Noneof theincludedwomenreceivedprogesterone,pessary,orcerclage. Cervicallengthmeasurement

ThemeangestationalageatTVUCLwasperformedatabout20 weekinbothgroups.ThemeanTVUCLwassigniﬁcantlylowerin the monochorionic diamniotic (32.810.1) compared to the dichorionic (34.98.6) group(MD 2.10mm, 95% CI 3.91 to -0.29). Fig.1 shows the distribution of CL in both groups. In monochorionicdiamnioticandindichorionicpregnancies,the5th percentilewere10.5mmand20.7mm,respectively;andthe95th

percentilewere45.0mmand50.0mm,respectively(Fig.2). Theincidenceofshortcervix,deﬁnedasTVUCL30mm,was 16.6%(29/175)inthemonochorionicgroup,and11.9%(48/405)in thedichorionicgroup(aOR1.48,95%CI1.03–2.43)(Table2).

InwomenwithoutpriorSPTB,theincidenceofTVUCL30mm was14.4% (26/160)inthemonochorionicgroup,and 10.8%(40/ 372)in thedichorionic group (aOR1.39, 95% CI 0.80–2.42). In womenwithprior SPTB,theincidenceof TVUCL30mmwas 40.0%(6/15)inthemonochorionicgroup,and24.2%(8/33)inthe dichorionicgroup(aOR2.08,95%CI0.57–7.68).

OnROCcurveanalysis,TVUCLforthepredictionofPTB<32 weeksshowedanareaunderthecurveof0.71(95%CI0.59–0.85)in themonochorionicdiamnioticpregnancy,andof0.67(95%0.55to 0.79)inthedichorionicdiamnioticpregnancy(Fig.3).TVUCLin monochorionicdiamniotic twinpregnancieshad a better value comparedtoTVUCLindichorionicpregnanciesinpredictionof PTB<32 weeks (DeLong test: difference between areas 0.049, standarderror0.0577,95%CI0.031to0.063,p=0.003).Sensitivity, speciﬁcity, LR+and LR-,for each TVU CL cutoffs are shown in

Table3 for monochorionicdiamniotictwins,and inTable4 for dichorionictwins.

Relationshipbetweencervicallengthandgestationalageatdelivery

Fig.3showstherelationshipbetweenTVUCLandgestational ageatdelivery.ForanygivenCLmeasuredbetween180–7and23 6/7 weeks, gestational age at delivery for monochorionic was earlierbyabout2weeksonaveragecomparedwithdichorionic pregnancies (MD -2.1 weeks; ANCOVA P<0.001). For TVU CL measurement assessed between18 0/7and 23 6/7 weeks, the correlationwithgestationalageatdeliveryinmonochorionictwin gestationswasrelativelyconstant[Spearman’scorrelationcoef ﬁ-cient0.57(0.42–0.66)]andwasstatisticallystronger(Fisher’s Z-transformation P=0.02)thanin dichorionicpregnancies [Spear-man’scorrelationcoefﬁcient0.33(0.27–0.52)].

Pregnancyoutcome

Thegestationalageatdeliverywassignificantlyearlierinthe monochorionic diamniotic group compared to the dichorionic groupbyabout2week(MD 2.20weeks,95%CI 2.75to 1.65). Monochorionicpregnancieshadasignificantlyhigherincidenceof PTB_<37,_<34and _<32weeks.Theresultsdidnotchangewhen medicallyindicatedbirthswereexcluded.Indeed,theincidenceof SPTB<37 and <34 weeks were significantly higher in the monochorionicgroupcomparedtothedichorionicgroup,while the rate of SPTB<32 weeks was not statistically significant differentafteradjustingforconfounders.Notably,the monochor-ionicgrouphadahigherincidenceofdeliveryduetospontaneous onsetoflabor(61.7%vs48.1%;aOR1.74,95%CI1.21–2.49)(Table5). Discussion

Mainﬁndings

Thisstudyshowedthatmonochorionicdiamnioticpregnancies had a higher rate of SPTB compared todichorionic diamniotic pregnancies. Thishigher rateof SPTB was reﬂected bya lower meanmidtrimesterTVUCL,byhigherincidenceofshortTVUCL, and by earlier gestational age at birth per any given CL in monochorionicdiamnioticcomparedtodichorionicpregnancies. Table1

Characteristicsoftheincludedwomen. Monochorionic N=175(30.2%) Dichorionic N=405(69.8%) pvalue Maternalage MeanSD >35yn(%) 31.24.79 46(26.3%) 32.05.1 111(27.4%) 0.10 0.78 BMI MeanSD 26.55.5 26.17.0 0.46 Nulliparity n(%) 111(63.4%) 254(62.7%) 0.87 PriorSPTB n(%) 15(8.6%) 33(8.1%) 0.87 Smoking n(%) 22(12.6%) 51(12.6%) 0.98

SD,standarddeviation;SPTB,spontaneouspretermbirth;BMI,Bodymassindex.

Fig. 1.Distributionoftransvaginalcervicallengthbetween180–7and236/7weeks indiamniotictwinspregnancies.Yellowboxesshowmonochorionicdiamniotic pregnanciesandblueboxesshowdichorionicpregnancies.TVUCL,transvaginal ultrasoundcervical.

(4)

Forany givenCLmeasured between180–7and 23 6/7weeks, gestational age at delivery for monochorionic diamniotic was earlierbyabout2weekscomparedwithdichorionicpregnancies. Twinswithamonochorionicdiamnioticpregnancyhadahigher rate of delivery at any GA due to spontaneous onset of labor comparedtodichorionicpregnancies.

Strengthsandlimitations

This study has several strengths. CL measurements were performedtransvaginallybycertiﬁed operators.Thisisa retro-spective study of a prospectively collecteddata in a dedicated database.Toourknowledge,therearenosimilarstudiesinthe Fig.2.Receiver-operatingcharacteristicscurveshowingpredictiveabilityofcervicallengthforspontaneouspretermbirth<32weeksinmonochorionicdiamniotic pregnancy(A),andindichorionicdiamnioticpregnancy(B).

Table2

Cervicallengthmeasurementofthemonochorionicdiamnioticanddichorionictwingroups. Monochorionic

N=175(30.2%)

Dichorionic N=405(69.8%)

aORorMD(95%CI)a

GAatTVUCLassessment(weeks) MeanSD 20.97.4 20.68.8 0.30week( 1.09to1.69) TVUCL(mm) MeanSD 32.510.1 34.97.7 2.40mm( 4.07to 0.73) >30mm 146(83.4%) 357(88.1%) 0.68(0.41–1.12) 30mm 29(16.6%) 48(11.9%) 1.48(1.03–2.43) 25mm 26(14.9%) 35(8.6%) 1.84(1.07–3.17) 20mm 18(10.3%) 19(4.7%) 2.33(1.19–4.56) 15mm 11(6.3%) 12(3.0%) 2.31(1.03–5.67) 10mm 8(4.6%) 8(2.0%) 2.38(0.88–6.44) 5mm 7(4.0%) 4(1.0%) 4.18(1.21–14.46)

G9A:Gestationalage;TVUCL:Transvaginalultrasoundcervicallength;SD,standarddeviation;CI,conﬁdenceinterval;MD,meandifference;aOR,adjustedoddsratio. Boldfacedata,statisticallysigniﬁcant.

a

AdjustedforallvariablesreportedinTable1.

Fig.3. Relationshipbetweencervicallengthbetween180–7and236/7weeksandgestationalageatdeliveryinmonochorionicanddichorionictwingroups.Therelationship ispresentedforasymptomatictwinswithmonochorionicpregnancies(redline)anddichorionicpregnancies(blueline).X-axis,cervicallength(inmm);Y-axis,gestational ageatdelivery(inweeks).Meandifference-2.1weeks;ANCOVAanalysisofcovarianceP<0.001.

(5)

literaturecomparingTVUCLinmonochorionicversusdichorionic twins. We excluded women who received cervical cerclage, pessaryorvaginalprogesteronetoavoidfurtherconfoundersin theincidenceofSPTB.

The most important shortcoming of this study was the retrospectivenonrandomizedapproach.TheCLwasassessedonly oncebetween180–7and236/7weeks.Therefore,comparingthe rateofcervicalshorteningduringgestationwasnotfeasible. Discussion

Different strategies have been evaluated for prediction and preventionofSPTB[21–40],Mid-trimesterTVUCLhasbeenshown

tobeagoodpredictorofSPTBinasymptomaticwomenwithtwin pregnancies[21,22].Ameta-analysisof21studies,including3523 women, showed that among asymptomatic women with twin pregnanciesaCL20mmat20–24weekswasthemostaccuratein predictingSPTB<32 and<34weekswitha pooledsensitivities, speciﬁcities,andpositiveandnegativelikelihoodratiosof39%and 29%,96%and97%,10.1and9.0,and0.64and0.74,respectively[22]. Sperling et al. in a prospective multicenter study of 383 twin pregnanciesshowedthatCLmeasurementat23weekswasagood screeningtestforpredictingtwinsatlowriskofpretermandvery pretermdelivery[23].TheyalsofoundthattheincidenceofSPTB was higher in the monochorionic compared tothe dichorionic twins. Our study showed that the higher rate of SPTB among Table3

Sensitivityandspeciﬁcityforeachcervicallengthmeasurementwith95%conﬁdenceintervalinpredictionofspontaneouspretermbirth<32weeksinmonochorionic diamniotictwinpregnancies.

CL Sensitivity(95%CI) Speciﬁcity(95%CI) LR+

LR-30mm 70%(56–79) 79%(71–80) 3.2(2.4–3.7) 0.4(0.3–0.6) 25mm 59%(48–66) 89%(87–91) 6.0(4.7–7.7) 0.5(0.4–0.6) 15mm 42%(39–61) 97%(95–97) 7.0(1.8–8.7) 0.5(0.3–0.7) 10mm 29%(19–39) 98%(97–99) 8.1(3.2–9.5) 0.6(0.5–0.8) 5mm 13%(7–29) 99%(98–100) 12.0(10.5–17.4) 0.9(0.6–0.9) CL,cervicallength;CI,conﬁdenceinterval;LR+,positivelikelihoodratio;LR-,negativelikelihoodratio.

Table4

Sensitivityandspeciﬁcityforeachcervicallengthmeasurementwith95%conﬁdenceintervalinpredictionofspontaneouspretermbirth<32weeksindichorionictwin pregnancies.

CL Sensitivity(95%CI) Speciﬁcity(95%CI) LR+

LR-30mm 67%(62–76) 77%(73–80) 3.1(2.5–3.5) 0.4(0.3–0.5) 25mm 57%(51–65) 88%(87–90) 5.7(0.49–7.1) 0.5(0.4–0.6) 15mm 40%(39–47) 97%(95–97) 3.0(2.3–4.7) 0.6(0.3–0.7) 10mm 29%(23–33) 98%(97–99) 7.5(4.2–8.3) 0.6(0.5–0.8)

5mm 12%(10–24) 99%(98–100) 12.0(10.5– 0.9(0.6–0.9)

CL,cervicallength;CI,conﬁdenceinterval;LR+,positivelikelihoodratio;LR-,negativelikelihoodratio.

Table5

Pregnancyoutcomesofthemonochorionicanddichorionictwingroups. Monochorionic

N=175(30.2%)

Dichorionic N=405(69.8%)

CrudeOR(95%CI) aORorMD(95%CI)a

GAatdelivery(weeks) MeanSD 34.03.2 36.22.9 – 2.20weeks( 2.75to 1.65) PTB<37weeks n(%) 100(57.1%) 202(49.9%) 1.34(1.20–1.88) 1.21(1.05–1.97) PTB<34weeks n(%) 61(34.9%) 93(23.0%) 1.85(1.20–2.77) 1.80(1.22–2.65) PTB<32weeks n(%) 23(13.1%) 40(9.9%) 1.37(1.19–2.03) 1.38(1.08–2.93) PTB<28weeks n(%) 9(5.1%) 16(4.0%) 1.32(0.57–3.04) 1.55(0.91–4.12) SPTB<37weeks n(%) 93(53.1%) 182(44.9%) 1.22(1.22–1.79) 1.39(1.13–1.90) SPTB<34weeks n(%) 52(29.7%) 81(20.0%) 1.90(1.12–2.99) 1.69(1.13–2.54) SPTB<32weeks n(%) 20(11.4%) 32(7.9%) 1.40(1.03–3.11) 1.50(0.83–2.71) SPTB<28weeks n(%) 8(4.6%) 12(3.0%) 1.57(0.63–3.91 1.67(0.70–5.03) Indicationfordeliveryn(%)

-Maternalindication 30(17.1%) 65(16.1%) 1.08(0.67–1.74) 1.05(0.55–1.70) -Fetalindication 22(12.6%) 20(4.9%) 2.77(1.47–5.22) 2.49(1.34–5.63) -Combinedmaternalandfetal 4(2.3%) 4(1.0%) 3.24(0.66–8.41) 2.35(0.58–9.49) -Plannedatterm 11(6.3%) 121(29.9%) 0.16(0.08–0.30) 0.30(0.11–0.73) -Spontaneousonsetoflabor 108(61.7%) 195(48.1%) 1.35(1.21–1.98) 1.74(1.21–2.49)

GA:Gestationalage;SD,standarddeviation;PTB,pretermbirth;SPTB,spontaneouspretermbirth;OR,oddsratio;aOR,adjustedoddsratio;CI,conﬁdenceinterval;MD,mean difference.Boldfacedata,statisticallysigniﬁcant.

a

(6)

monochorionic twins could be reﬂected and predicted by the higherrateofshortcervix.Notably,severaltreatmentsforSPTB have been shown to have different effects in monochorionic comparedtodichorionicpregnancies[24]. Theseﬁndings could supportthetheorythatthepathogenesisforSPTBwasdifferentin monochorionic and in dichorionic twins. Cervical pessary, for example,in one largetrialhave beenshown tobe effectivein monochorionicbutnotindichorionictwins[24].Ourstudyalso showedthat women with twinpregnancies can havedifferent baselineriskofSPTBbasedonthemidtrimesterTVUCL.Aprior large retrospective study, showed that IVF-conceived twin pregnancies had an increased risk of SPTB compared to those whoconceivedspontaneouslyandthatthisriskwaspredictedby lowermidtrimesterTVUCL.

OurstudyalsoshowedthatthecorrelationbetweenshortTVUCL and earlier gestationalageat delivery was stronger in monochorionic comparedtothedichorionictwins,andthatanygivenTVUCLwas associated with lower gestational age of about 2 week in the monochorionic diamniotic compared to the dichorionic twins. FuturestudiesshoulddifferentiateTVUCLandperinataloutcomes betweenmonochorionicanddichorionicpregnancies.This infor-mationwouldbeusefulincounselingwomenregardingtheirriskof SPTBaswellasimprovingthedesignoffuturestudiesevaluating therapiesinthepreventionofpretermdeliveryintwinpregnancies. Thebiologicalplausibilitytoexplainourfindingsisnotcompletely clear.However,someunknownfactorsmaybemorelikelytotrigger subclinicalorovertuterine contractionsin monochorionic com-paredtodichorionictwins,whichwouldincreasetheriskofPTBin thepresenceofagivenCL.Inapriorstudy,ourgroupalsoshowed thatIVF-conceivedtwinpregnancieshadasignificantlylowermean TVUCL comparedto spontaneously-conceived twinpregnancies [25].Thesefindingsmayleadtothehypothesisthatdifferentsubset ofwomenhavedifferentbaselineriskofSPTBaccordingtothemean midtrimesterTVUCL.

Inourcohort,weexcludedwomenwhoreceivedprogesterone, pessary,orcerclage.Indeed,preventionofSPTBinwomenwith twin pregnancies is still a subject of debate. Progesterone is routinelyusedtopreventSPTBinsingletons[32,41,42],butisnot currentlyrecommendedfortwins[43].Arecentmeta-analysisof randomized trials showed that use of Arabin pessary in twin pregnancieswithshortTVUCLat16–24weeksdoesnotprevent SPTBorimproveperinataloutcome[31].Finally,dataoncervical cerclagearecontroversial.Whileitseemstobebeneﬁcialbasedon small retrospective studies [44,45], level-1 data showed an increased risk of perinatal complications in women with twin gestationsreceivingcerclage[16].

Conclusion

Insummary,ourstudyshowedthatthehigherrateofSPTBin monochorionicdiamnioticcompared withdichorionic pregnan-ciesispredictedbyalowermidtrimesterTVUCL,aswellasbythe earliergestationalageatbirthperanygivencervicallength. Disclosure

Theauthorsreportnoconﬂictofinterest Financialsupport

Noﬁnancialsupportwasreceivedforthisstudy References

[1]HamiltonB.E.,MartinJA,OstermanMJ.Births:preliminarydatafor2015.Natl VitalStatRep2016;(65):1–15.

[2]OrzechowskiKM,BoelingRC,BaxterJK,BerghellaV.Auniversaltransvaginal cervical length screening program for preterm birth prevention. Obstet Gynecol2014;124:520–5.

[3]NavatheR,SacconeG,VillaniM,KnappJ,CruzY,BoeligR,etal.Decreaseinthe incidenceofthreatenedpretermlaborafterimplementationoftransvaginal ultrasoundcervicallengthuniversalscreening.JMaternFetalNeonatalMed 2018;5(January)1–6, doi:http://dx.doi.org/10.1080/14767058.2017.1421166 [Epubaheadofprint].

[4]BerghellaV,PalacioM,NessA,AlﬁrevicZ,NicolaidesKH,SacconeG.Cervical lengthscreeningforpreventionofpretermbirthinsingletonpregnancywith threatenedpretermlabor:systematicreviewandmeta-analysisof random-izedcontrolledtrialsusingindividualpatient-leveldata.UltrasoundObstet Gynecol2017;49(March(3))322–9, doi:http://dx.doi.org/10.1002/uog.17388 Epub2017Feb8.Review.

[5]PereiraS,FrickAP,PoonLC,ZamprakouA,NicolaidesKH.Sucessfulinduction oflabor:predictionbypreinductioncervicallength,angleofprogressionand cervicalelastography.UltrasoundObstetGynecol2014;44:468–75. [6]SuhagA,ReinaJ,SanapoL,MartinelliP,SacconeG,SimonazziG,etal.Prior

ultrasound-indicatedcerclage:comparisonofcervicallengthscreeningor history-indicated cerclage in the next pregnancy. Obstet Gynecol 2015;126:962–8.

[7]OwenJ,YostN,BerghellaV,ThomE,SwainM,Dildy[157_TD$DIFF][148_TD $DIFF]3rd GA, et al. National Institute of Child Health and Human Development,maternal-fetalmedicineunitsnetwork.Mid-trimester endo-vaginalsonographyinwomenathighriskforspontaneouspretermbirth. JAMA2001;286:1340–8.

[8]Verhoeven CJ, OpmeerBC, Oei SG, LatourV,van der Post JA, Mol BW. Transvaginalsonographic assessmentof cervicallength and wedging for predictingoutcomeoflaborinductionatterm:asystematicreviewand meta-analysis.UltrasoundObstetGynecol2013;42:500–8.

[9]SacconeG,SimonettiB,BerghellaV.Transvaginalultrasoundcervicallength forpredictionofspontaneouslabouratterm:asystematicreviewand meta-analysis.BJOG2016;123:16–22.

[10]BerghellaV,SacconeG.Fetalﬁbronectintestingforpreventionofpreterm birthinsingletonpregnancieswiththreatenedpretermlabor:asystematic reviewandmetaanalysisofrandomizedcontrolledtrials.AmJObstetGynecol 2016;215(October(4)):431–8.

[11]MaruottiGM,SacconeG,MorlandoM,MartinelliP.First-trimesterultrasound determinationofchorionicityintwingestationsusingthelambdasign:a systematic review and meta-analysis. EurJ ObstetGynecol Reprod Biol 2016;202:66–70.

[12]TheMONOMONOstudy.UltrasoundObstetGynecol2018(inpress). [13]PaganiG,StagnatiV,FicheraA,PrefumoF.Cervicallengthatmid-gestationin

screeningforpretermbirthintwinpregnancy.UltrasoundObstetGynecol 2016;48:56–60.

[14]SocietyforMaternal-FetalMedicine(SMFM).Theroleofroutinecervical lengthscreeninginselectedhigh-andlow-riskwomenforpretermbirth prevention.AmJObstetGynecol2016;215:B2–7.

[15]BerghellaV,BerghellaM.Cervicallengthassessmentbyultrasound. Acta ObstetGynecolScand2005;84(June(6)):543–4.

[16]SacconeG,RustO,AlthuisiusS,RomanA,BerghellaV.Cerclageforshortcervix intwinpregnancies:systematicreviewandmeta-analysisofrandomized trialsusingindividualpatient-leveldata.ActaObstetGynecolScand2015;94 (April(4))352–8,doi:http://dx.doi.org/10.1111/aogs.12600Epub2015Mar1.. [17]Cheong-SeeF,SchuitE,Arroyoo-ManzanoD,etal.Prospectiveriskofstillbirth andneonatalcomplicationsintwinpregnancies:systematicreviewand meta-analysis.BMJ2016;354:i4353.

[18]SacconeG,BerghellaV.Planneddeliveryat37weeksintwins:asystematic reviewandmeta-analysisofrandomizedcontrolledtrials.JMatern Fetal Neonatal Med 2016;29(March (5)):685–9, doi:http://dx.doi.org/10.3109/ 14767058.2015.1016423.

[19]Smith AH, Bates MN. Conﬁdence limit analyses should replace power calculation in the interpretation of epidemiologicstudies. Epidemiology 1992;3:449–52.

[20]McNameeR.Regressionmodellingandothermethodstocontrolconfounding. OccupEnvironMed2005;62:500–6.

[21]SacconeG,CiardulliA,XodoS,DugoffL,LudmirJ,D’AntonioF,etal.Cervical pessaryforpreventingpretermbirthintwinpregnancieswithshortcervical length:asystematicreviewandmeta-analysis.JMaternFetalNeonatalMed 2017;30(December(24)):2918–25.

[22]Conde-AgudeloA,RomeroR, HassanSS,YeoL. Transvaginalsonographic cervical length forthe prediction ofspontaneous preterm birthintwin pregnancies:asystematicreviewandmetaanalysis.AmJObstetGynecol 2010;203(August(2))128.e1-12.

[23]SperlingL,Kiil C,LarsenLU,etal. Howtoidentifytwinsatlow riskof spontaneouspretermdelivery.UltrasoundObstetGynecol2005;26:138–44. [24]TajikP,MonfranceM,VanHooftJ,etal.Amultivariablemodeltoguidethe

decisionforpessaryplacementtopreventpretermbirthinwomenwitha multiplepregnancy:asecondaryanalysisoftheProTWINtrial.Ultrasound ObstetGynecol2016;48:48–55.

[25]SacconeG,ZulloF,RomanA,WardA,MaruottiG,MartinelliP,etal.Riskof spontaneouspretermbirthinIVF-conceivedtwinpregnancies.JMaternFetal Neonatal Med 2017;21(September)1–8, doi:http://dx.doi.org/10.1080/ 14767058.2017.1378339[Epubaheadofprint].

[26]SacconeG,MaruottiGM,GiudicepietroA,MartinelliP,ItalianPretermBirth Prevention(IPP)WorkingGroup.Effectofcervicalpessaryonspontaneous

(7)

pretermbirthinwomenwithsingletonpregnanciesandshortcervicallength: arandomizedclinicaltrial.JAMA2017;9318(December(23)):2317–24,doi: http://dx.doi.org/10.1001/jama.2017.18956.

[27] SentilhesL,SénatMV,AncelPY,AzriaE,BenoistG,BlancJ,etal.Preventionof spontaneouspretermbirth:guidelinesforclinicalpracticefromtheFrench CollegeofGynaecologistsandObstetricians(CNGOF).EurJObstetGynecol ReprodBiol2017;210(March):217–22.

[28]SacconeG,BerghellaV.Omega-3longchainpolyunsaturatedfattyacidsto preventpretermbirth:asystematicreviewandmeta-analysis.ObstetGynecol 2015;125(March (3)):663–72, doi:http://dx.doi.org/10.1097/ AOG.0000000000000668.

[29]SacconeG,BerghellaV.Folicacidsupplementationinpregnancytoprevent preterm birth: a systematic review and meta-analysis of randomized controlledtrials.EurJObstetGynecolReprodBiol2016;199(April)76–81, doi:http://dx.doi.org/10.1016/j.ejogrb.2016.01.042Epub2016Feb8.Review. [30]MagroMalossoER,SacconeG,SimonettiB,SquillanteM,BerghellaV.US

trendsinabortionandpretermbirth.JMaternFetalNeonatalMed2017;6(July) 1–5, doi:http://dx.doi.org/10.1080/14767058.2017.1344963 [Epub ahead of print].

[31] SacconeG,CiardulliA,XodoS,DugoffL,LudmirJ,D’AntonioF,etal.Cervical pessaryforpreventingpretermbirthintwinpregnancieswithshortcervical length:asystematicreviewandmeta-analysis.JMaternFetalNeonatalMed 2017;30(December (24))2918–25, doi:http://dx.doi.org/10.1080/ 14767058.2016.1268595Epub2017Jan12..

[32]SuhagA, SacconeG, Berghella V.Vaginal progesterone for maintenance tocolysis:asystematicreviewandmetaanalysisofrandomizedtrials.AmJ ObstetGynecol2015;213(October(4)):479–87,doi:http://dx.doi.org/10.1016/j. ajog.2015.03.0.

[33]EhsanipoorRM,SeligmanNS,SacconeG,SzymanskiLM,WissingerC,Werner EF,etal.Physicalexamination-indicatedcerclage:asystematicreviewand meta-analysis.ObstetGynecol2015;126(July(1))125–35,doi:http://dx.doi. org/10.1097/AOG.0000000000000850Review.

[34]SacconeG, Berghella V. Omega-3 supplementation to prevent recurrent pretermbirth:asystematicreviewandmetaanalysisofrandomizedcontrolled trials.AmJObstetGynecol2015;213(August(2))135–40,doi:http://dx.doi.org/ 10.1016/j.ajog.2015.03.013Epub2015Mar7.Review.

[35]Saccone G, Berghella V, Maruotti GM, Sarno L, Martinelli P. Omega-3 supplementationduringpregnancytopreventrecurrentintrauterinegrowth restriction:systematicreviewandmeta-analysisofrandomizedcontrolled trials.UltrasoundObstetGynecol2015;46(December(2))659–64,doi:http:// dx.doi.org/10.1002/uog.14910Epub2015Nov4.Review.

[36]SacconeG,SacconeI,BerghellaV.Omega-3long-chainpolyunsaturatedfatty acidsandﬁshoilsupplementationduring pregnancy:which evidence?J

Matern Fetal Neonatal Med 2016;29(15)2389–97, doi:http://dx.doi.org/ 10.3109/14767058.2015.1086742Epub2015Sep18.Review..

[37]SacconeG,BerghellaV,SarnoL,MaruottiGM,CetinI,GrecoL,etal.Celiac diseaseandobstetriccomplications:asystematicreviewandmetaanalysis. AmJObstetGynecol2016;214(February(2))225–34,doi:http://dx.doi.org/ 10.1016/j.ajog.2015.09.080Epub2015Oct9.Review..

[38]EkeAC,SacconeG,BerghellaV.Selectiveserotoninreuptakeinhibitor(SSRI) useduringpregnancyandriskofpretermbirth:asystematicreviewand meta-analysis. BJOG 2016;123(November (12))1900–7, doi:http://dx.doi.org/ 10.1111/1471-0528.14144Epub2016May30.Review.

[39]SacconeG,BerghellaV,MaruottiGM,GhiT,RizzoG, SimonazziG, etal. Antiphospholipid antibody proﬁle based obstetric outcomes of primary antiphospholipidsyndrome:thePREGNANTSstudy. AmJObstetGynecol 2017;216(May (5))525.e1–525.e12, doi:http://dx.doi.org/10.1016/j. ajog.2017.01.026Epub2017Jan30.

[40]SacconeG,KhalifehA,ElimianA,BahramiE,Chaman-AraK,BahramiMA,etal. Vaginalprogesteronevsintramuscular17α-hydroxyprogesteronecaproatefor preventionofrecurrentspontaneouspretermbirthinsingletongestations: systematic review and meta-analysis of randomized controlled trials. UltrasoundObstetGynecol2017;49(March(3))315–21,doi:http://dx.doi.org/ 10.1002/uog.17245Epub2017Feb6.Review.

[41]SacconeG,SuhagA,BerghellaV.17-alpha-hydroxyprogesteronecaproatefor maintenancetocolysis:asystematicreviewandmetaanalysisofrandomized trials.AmJObstetGynecol2015;213(July(1)):16–22,doi:http://dx.doi.org/ 10.1016/j.ajog.2015.01.054.

[42]Quist-NelsonJ,ParkerP,MokhtariN,DiSarnoR,SacconeG,BerghellaV. Progestogens in singleton gestations with preterm prelabor rupture of membranes:asystematicreviewandmetaanalysisofrandomizedcontrolled trials. Am J Obstet Gynecol 2018(March), doi:http://dx.doi.org/10.1016/j. ajog.2018.03.027pii:S0002-9378(18)30242-4.[Epubaheadofprint]. [43]SocietyforMaternal-FetalMedicinePublicationsCommittee,withassistance

ofVincenzoBerghella.Progesteroneandpretermbirthprevention:translating clinicaltrialsdataintoclinicalpractice.AmJObstetGynecol2012;206(May (5)):376–86,doi:http://dx.doi.org/10.1016/j.ajog.2012.03.010.

[44]RomanA,RochelsonB,FoxNS,HoffmanM,BerghellaV,PatelV,etal.Efﬁcacyof ultrasound-indicated cerclage in twin pregnancies. AmJ Obstet Gynecol 2015;212(June(6)),doi:http://dx.doi.org/10.1016/j.ajog.2015.01.031788.e1-6. [45]RomanA,RochelsonB,MartinelliP,SacconeG,HarrisK,ZorkN,etal.Cerclage intwinpregnancywithdilatedcervixbetween16to24weeksofgestation: retrospectivecohortstudy.AmJObstetGynecol2016;215(July(1)),doi:http:// dx.doi.org/10.1016/j.ajog.2016.01.17298.e1-98.e11.