Role of progesterone, cerclage and pessary in preventing preterm birth in twin pregnancies: A systematic review and network meta-analysis

(1)

Review

article

Role

of

progesterone,

cerclage

and

pessary

in

preventing

preterm

birth

in

twin

pregnancies:

A

systematic

review

and

network

meta-analysis

Francesco

D

’Antonio

a,

*

,

Vincenzo

Berghella

b

,

Daniele

Di

Mascio

c

,

Gabriele

Saccone

d

,

Filomena

Sileo

e

,

Maria

Elena

Flacco

f

,

Anthony

O. Odibo

g

,

Marco

Liberati

a

,

Lamberto

Manzoli

f

,

Asma

Khalil

h,i

a

CenterforHigh-RiskPregnancyandFetalCare,DepartmentofObstetricsandGynecology,UniversityofChieti,Chieti,Italy

b

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

c

DepartmentofMaternalandChildHealthandUrologicalSciences,SapienzaUniversityofRome,Italy

d

DepartmentofNeuroscience,ReproductiveSciencesandDentistry,SchoolofMedicine,UniversityofNaplesFedericoII,Naples,Italy

e

PrenatalMedicineUnit,ObstetricsandGynecologyUnit,DepartmentofMedicalandSurgicalSciencesforMother,ChildandAdult,UniversityofModena andReggioEmilia,Modena,Italy

f

DepartmentofMedicalSciences,UniversityofFerrara,Italy

g

DivisionofMaternal-FetalMedicine,DepartmentofObstetricsandGynecology,UniversityofSouthFlorida,MorsaniCollegeofMedicine,UnitedStates

h

FetalMedicineUnit,SaintGeorge’sUniversityofLondon,London,UnitedKingdom

i

VascularBiologyResearchCentre,MolecularandClinicalSciencesResearchInstitute,StGeorge'sUniversityofLondon,UnitedKingdom

ARTICLE INFO

Articlehistory:

Received23December2020 Receivedinrevisedform10April2021 Accepted17April2021 Availableonlinexxx Keywords: Twins Twinpregnancy Multiplepregnancy Pretermbirth Pretermdelivery Pessary Progesterone Cerclage ABSTRACT

Objective:Todeterminetheroleofprogesterone,pessaryandcervicalcerclageinreducingtheriskof (pretermbirth)PTBintwinpregnanciesandcomparetheseinterventionsusingpairwiseandnetwork meta-analysis.

Studydesign:Medline,Embase,CINAHLandCochranedatabaseswereexplored.Theinclusioncriteria werestudiesinwhichtwinpregnancieswererandomizedtoaninterventionforthepreventionofPTB (anytypeofprogesterone,cervicalcerclage,cervicalpessary,oranycombinationofthese)ortoacontrol group(e.g.placeboortreatmentasusual).Interventionsofinterestwereeitherprogesterone[vaginalor oralnatural progesteroneor intramuscular17a-hydroxyprogesteronecaproate (17-OHPC)],cerclage (McDonaldorShirodkar),orcervicalpessary.

TheprimaryoutcomewasPTB<34weeksofgestation.Bothprimaryandsecondaryoutcomeswere exploredinanunselectedpopulationoftwinpregnanciesandinwomenathigherriskofPTB(deﬁnedas thosewithcervicallength<25mm).Random-effecthead-to-headandamultiple-treatment meta-analyseswereusedtoanalyzethedataandresultsexpressedasriskratios.

Results:26studieswereincludedinthemeta-analysis.Whenconsideringanunselectedpopulationof twinpregnancies,vaginalprogesterone,intra-muscular17-OHPCorpessarydidnotreducetheriskofPTB <34weeksofgestation(allp>0.05).WhenstratifyingtheanalysisforspontaneousPTB,neitherpessary, vaginalor intramuscular17-OHPCwere associated withasigniﬁcant reductioninthe risk ofPTB comparedtocontrols(allp>0.05),whiletherewasnostudyoncerclagewhichexploredthisoutcomein anunselectedpopulationoftwinpregnancies.Whenconsideringtwinpregnancieswithshortcervical length(25mm),therewasnocontributionofeitherpessary,vaginalprogesterone,intra-muscular 17-OHPCorcerclageinreducingtheriskofoverallPTB<34weeksofgestation.

Conclusions:Cervicalpessary,progesteroneandcerclagedonotshowasigniﬁcanteffectinreducingthe rateofPTBorperinatalmorbidityintwins,eitherwhentheseinterventionsareappliedtoanunselected populationoftwinsorinpregnancieswithashortcervix.

*Correspondingauthorat:CenterforHighRiskPregnancyandFetalCare-DepartmentofObstetricsandGynecology,UniversityofChieti,ViadeiVestini31,66100,Chieti, Italy.

E-mailaddress:francesco.dantonio@unich.it(F.D’Antonio).

https://doi.org/10.1016/j.ejogrb.2021.04.023

ContentslistsavailableatScienceDirect

European

Journal

of

Obstetrics

&

Gynecology

and

Reproductive

Biology

(2)

Contents

Introduction ... 167

Materialandmethods ... 167

Datasources ... 167

Eligibilitycriteria,mainoutcomesmeasures ... 167

Datacollectionandanalysis ... 168

Results ... 170

Generalcharacteristicsofthestudies ... 170

Synthesisoftheresults ... 170

Primaryoutcome(PTB<34weeks) ... 170

Secondaryoutcomes ... 173

Secondaryoutcomes:fetalandneonataloutcomes ... 173

Sub-groupanalyses:twinpregnancieswithshortcervicallength(25mm) ... 174

Discussion ... 174

Mainﬁndings... 174

Strengthsandlimitations ... 174

Comparisonwithothersystematicreviews ... 175

Clinicalandresearchimplications ... 175

Conclusions ... 175

Funding ... 175

References ... 175

Introduction

Twinpregnanciesareatincreasedrisk ofperinatalmortality andmorbiditycomparedtosingleton,mainlyasaresultoffetal anomalies,growthdisordersorpretermbirth(PTB)[1,2].PTBisthe majordeterminantofadverseperinataloutcomeintwin pregnan-cies,withanestimatedincidenceof59%and11%before37and32 weeksofgestation,respectively[1].ThelargemajorityofPTBin twinsisspontaneous,followingeitherpretermprematurerupture ofmembranes(PROM)orspontaneousonsetoflabor,althougha signiﬁcantproportionofthesepregnanciesaredeliveredpreterm afteriatrogenicobstetricalinterventionsformaternal-fetal indi-cations[1].

In singleton pregnancies, prophylactic administration of progesterone inwomen atriskhasbeenshowntodecreasethe riskofPTB,perinatalmortalityandmorbidity[3].

Conversely, extrapolating objectiveevidenceontheeffective prevention intwins is challenging. Smallsample size,different outcomemeasuresandtheinclusionofcasesatincreasedriskof PTB,suchaspregnancieswithmonochorionicity-related compli-cations hampers the ability to determine whether such inter-ventions mayreducetheincidenceof PTB intwinpregnancies. Moreimportantly,studiesonPTBpreventionintwinpregnancies do notcomparedifferenttypesofinterventions.Therefore,it is dif_ﬁcult to conclude which intervention is likely to be more effective.

Theaimofthissystematicreviewwastodeterminetheroleof progesterone,pessaryandcervicalcerclageinreducingtheriskof PTB intwinpregnanciesandcomparetheseinterventionsusing pairwiseandnetworkmeta-analyses.

Materialandmethods Datasources

Thisreviewwasperformedaccordingtoana-prioridesigned protocol and recommended for systematic reviews and meta-analysis [4]. MEDLINE, Embase and CINAHL were searched electronically on theMarch2021 utilizing combinationsof the relevantmedicalsubjectheading(MeSH)terms,keywords, and word variants for “twin pregnancies” and “preterm birth” (Supplementary Table1).Thesearchandselection criteriawere restricted to English language. The reference lists of relevant

articlesandreviewswerehandsearchedforadditionalreports.The PRISMAguidelineswerefollowed[5]. Thestudywas registered with the PROSPERO database (Registration number: CRD42019127901).

Eligibilitycriteria,mainoutcomesmeasures

Inclusioncriteriawerestudiesinwhichtwinpregnancieswere randomizedtoaninterventionforthepreventionofpretermbirth (anytypeofprogesterone,cervicalcerclage,cervicalpessary,orany combination of these) or to a control group (e.g. placebo or treatmentasusual)Interventionsofinterestwereeithertypeof progesterone (natural progesterone per vagina or oral, or intramuscular 17a-hydroxyprogesterone caproate [17-OHPC]), cerclage(McDonaldorShirodkar),orpessary.

TheprimaryoutcomewasPTB<34weeksofgestation. Thesecondaryoutcomeswere:

PTB<37weeks PTB<32weeks PTB<28weeks PTB<24weeks

Gestational age at delivery (weeks), expressed ascontinuous variable

Pretermprematureruptureofthemembranes(PPROM),de_ﬁned astheruptureofthemembranesbefore37weeksofgestation Cesareansection(CS)

Tocolytictherapy Steroidadministration Vaginaldischarge Vaginalinfections Urinarytractinfections Chorioamnionitis

Intra-uterinedeath(IUD),deﬁnedasdeathofeithertwinfrom22 weeksofgestations

Neonataldeath(NND),deﬁnedasthedeathofeithernewbornup to28daysoflife

Perinataldeath,de_ﬁnedasthesumofIUDandNND Apgarscore<7at3minutes

Birthweight(BW)<2500g BW<1500g

BWexpressedascontinuousvariable Respiratorydistresssyndrome(RDS)

F.D’Antonio,V.Berghella,D.DiMascioetal. EuropeanJournalofObstetrics&GynecologyandReproductiveBiology261(2021)166–177

(3)

Bronchopulmonarydysplasia(BPD) Needformechanicalventilation

Intraventricularhemorrhage(IVH)(anygrade) Periventricularleukomalacia(PVL)(anygrade) Necrotizingenterocolitis(NEC)

Retinopathyofprematurity Sepsis

Admissiontoneonatalintensivecareunit(NICU) Lengthofin-hospitalstay

Both primary and secondary outcomeswere explored in an unselectedpopulationoftwinpregnanciesandinthatathigher riskofPTB[includingonlytwinswithacervicallength(CL)<25 mm)].

Datacollectionandanalysis

Two reviewers (FDA, AK) independently extracted data. Inconsistencies were discussed among the reviewers until consensuswasreached. Forthosearticlesinwhich therelevant informationwasnotreportedbutthemethodologywassuchthat theinformation mighthavebeenrecordedinitially,theauthors werecontactedrequestingthedata.

RiskofbiaswasassessedusingtheRevisedCochrane risk-of-biastoolforrandomizedtrials(RoB2)[6].Accordingtothistool, theriskofbiasofeachincludedstudyisjudgedaccordingtoﬁve domains:biasarisingfromtherandomizationprocess,biasdueto deviations from intended interventions, bias due to missing outcomedata,biasinthemeasurementoftheoutcomeandbias inselectionofthereportedresult.AlthoughtheRoB2tooldoesnot provideanoverallriskofbiasassessment,theoverallriskofbias wasconsideredlowiffourormoredomainswereratedaslowrisk (not counting ‘other biases’), with at least one of them being sequencegenerationorallocationconcealment,accordingtowhat reportedinprevioussystematicreviewsofintervention.Finally, thequality ofevidenceand strength ofrecommendationswere assessed using the Grading of Recommendations Assessment, Development and Evaluation(GRADE)methodologyfor network meta-analyses(GRADEpro,Version20.McMasterUniversity,2014) [7].

Thepurposeofthepresentmeta-analysiswastointegratethe availableevidenceinPTBmanagementamongtwinpregnancies, and tohelp identifywhich, among fourinterventions (pessary, vaginal administration of progesterone, intramuscular adminis-tration of 17α-hydroxyprogesterone caproate – IM 17-OHPC; cerclage) plusa watchful waitingapproach (expectant manage-ment),isthemosteffectiveinpreventingPTB.Todothis,twodata synthesis approaches wereused. First, we performed random-effecthead-to-headmeta-analysescomparingtheriskofPTBand of other maternal and neonatal adverse outcomes in women treatedwith(a)pessary;(b)vaginalprogesterone;(c)IM17-OHPC; (d)cerclage.Onlydataextrapolatedfromprospectivelyregistered studies were includedin the pooled analysis. Allhead-to-head meta-analyses wereperformed twice:(a) consideringan unse-lected populationof twin pregnancies; (b) including onlytwin pregnancieswithaCL<25mm.Overall,atotalof25outcomes were analyzed. The main outcome was the incidence of PTB, deﬁnedasbirthoccurringbefore34completedweeksofgestation, and assessedtwice: (a)includingallPTBs;(b) consideringonly spontaneousPTB.Additionally,PTBwasassessedatotherfourtime points(before37,32,28and24weeksofgestation).Amongthe secondary outcomes,three werecontinuous (gestationalageat birth;birthweight;lengthofin-hospitalstay),theotherwere21 categorical. Of the latter, eight outcomes assessed maternal morbidity(PPROM;cesareansection;needfortocolysis;steroids administration;vaginaldischarge;urinarytractinfection;vaginal

infection;chorioamnionitis);theremaining13outcomes evaluat-edseveralneonataladverseevents:death(stratifiedas intrauter-ine,neonatalorperinatal);compositemorbidity;Apgarscore<7; lowbirthweight(separatelyassessedas<2500gand<1500g); RDS; BPD; need for mechanical ventilation; IVH; PVL; NEC; retinopathy; sepsis; admission to NICU. For each head-to-head meta-analysis,theresultswereexpressedasRiskRatiosand95% confidence intervals (CI) [8]. The statistical heterogeneity was quantifiedusingtheI2_metric_[₉_].

Second, for theprimary outcome, in theabsence of an RCT directly comparing all interventions to prevent PTB in twin pregnancies _<34 weeks, a multiple-treatment meta-analysis (MTM)providesthebestevidenceonthemosteffectiveapproach [10].Thismethodologyallowsindirecthead-to-headcomparisons of allinterventionsthrough a commoncomparator,and subse-quentrankingoftheinterventions[11].MTMhastheadvantage that it can incorporate the evidence from all comparisons of differenttreatmentswithinasingleanalysis.Thisallowsabetter appreciation of the relative merits of each treatment within a common analytical framework [11]. The model was ﬁtted in a Bayesianframework,usingthenetworkpackageinStata[12],and its estimates were relative effect sizes and their credibility intervals[11].IntheMTMframework,weconsideredthefollowing networknodes:(a)pessary;(b)vaginalprogesterone;(c)IM 17-OHPC;(d)cerclage.WerunseparateMTManalysesaccordingto thegestationalrisk,eitherconsidering(a)anunselected popula-tionoftwinpregnancies,and(b)onlywomenathighriskofPTB, foratotalof12separateMTManalyses.

In order to assess the different contribution of each direct comparisontotheestimationofthenetworksummaryeffect,a contribution plot using the Stata command netweight was performed [13]. In the plot, the weight of each comparison (expressedaspercentage),isacombinationofthevarianceofthe directtreatmenteffectandthenetworkstructure[14].

In the absence of closed loops in the network, a formal statisticalassessmentofinconsistencywasnotpossible[15],and inconsistencyisnotassumedbydeﬁnition[16].However,thisdoes notimplythatthetransitivityassumptionwillnecessarilyhold:to checkthevalidityofthetransitivityassumption,weevaluatedthe potentialdifferencesinthegestationalageandthecervicallength (CL)acrossallthestudiesincludedinthenetworkmeta-analysis. Whendatawereinsufﬁcientforthisassessment,weassumedthat thetransitivityassumptionwasmet[17].

Finally,fortheoutcomeswhereanetworkmeta-analysiswas performed the interventions were ranked according to their Surface Under the Cumulative RAnking curve values (SUCRA), whichreﬂectthelikelihoodofaninterventionofbeingamongthe best[18].HigherSUCRAvaluesindicatehigherprobabilitiesofan interventionconsistentlybeingamongthebest.

Thequalityratingofeachanalysiswasassessedfollowingthe GRADE Working Group approach for network meta-analyses which, amongotheraspectssuchas riskof biasof thestudies, takes into consideration the network assumptions to rate the quality of the evidence, downgrading its score if there is heterogeneity,intransitivityorincoherence[18].Formostofthe secondary outcomes, given the relative scarcity of the studies included in each comparison, and the missingness of data, inconsistencywouldbelikelytooccur,thuswedidnotperform anyMTMcalculation,andusedonlydirectcomparison head-to-headmeta-analyses.

Small study effects (potentially caused by publication bias) wereassessedusingfunnelplots,andformallytestedthroughthe Eggerregressionasymmetrytestforthosemeta-analyses includ-ing 10 studies. When less than ten studies are included, the availabletestsareatveryhighriskofbiasbecauseofthelackof statisticalpower[6].

(4)

Table 1

General characteristics of the included studies.

Author Year Country Period

considered

Inclusion criteriaa _Intervention _{Description of intervention} _{Indication for}

intervention Primary outcome Intervention sample size Comparison sample size Rehal [19] 2021 United Kingdom, Italy,

Spain, Bulgaria, France, Belgium

2017 2019 All twin pregnancies at 11 14 weeks Progesterone Vaginal progesterone (300 mg/2 daily from 11 14 to 34 weeks

Prophylactic sPTB< 34 weeks 582 587

Merced [20]

2019 Spain 2010 2014 Twin pregnancies with CL less than 20 mm

after an episode of threatened preterm labor

Arabin Pessary Arabin Pessary CL20 mm sPTB< 34 weeks 67 65

Dang [21] 2019 Vietnam 2016 2017 Asymptomatic twin pregnancies with CL

less than 38 mm

Pessary vs Progesterone

Arabin Pessary; Vaginal progesterone (400 mg/daily) from 16 22 weeks

CL< 38 mm PTB< 34 weeks 150 150

Berghella [22]

2017 United States 2004 2016 Asymptomatic twin pregnancies with

short cervix between 18 + 0 27 + 6 weeks

Pessary BiotequeTM_cup _CL_{30 mm} _{PTB (overall)}

<34 weeks

23 23

Goya [23] 2016 Spain 2011 2014 Asymptomatic twin pregnancies with a

short cervix at between 18 22 weeks

Pessary Arabin Pessary CL 25 mm sPTB < 34 weeks 68 66

Nicolaides [24]

2016 United Kingdom 2008 2011 All twin pregnancies at 20 24 + 6 weeks Pessary Arabin Pessary Prophylactic sPTB< 34 weeks 590 590

Liem [25] 2013 The Netherlands 2009 2012 All twin pregnancies at 12 20 weeks Pessary Arabin Pessary Prophylactic Composite poor

perinatal outcome

403 410

El-Refaie [26]

2016 Egypt 2012 2014 DC twin pregnancies with CL between

20 25 mm at 20 24 weeks

Vaginal progesterone

Progesterone suppositories (400 mg/daily from 20 24 to 37 weeks

Cl between 20 25 mm

sPTB< 34 weeks 116 108

Awwad [27]

2015 Lebanon 2006 2011 All twin pregnancies at 16 20 weeks Intramuscular

17OHPC

Intramuscular injection of 250 mg 17 OHPC /weekly from 16 20 to 36 weeks

None (unselected population)

PTB< 37 weeks 194 94

Brizot [28] 2015 Brazil 2007 2013 All twin pregnancies between 18 and 21 + 6

weeks of gestation

Progesterone ovules 100 mg/daily None (unselected population)

Difference in mean GA at birth

189 191

Senat [29] 2013 France 2006 2010 Asymptomatic twin pregnancies with CL<

25 mm at 24 + 0 31 + 6 weeks

Intramuscular 17OHPC

Intramuscular injection of 500 mg 17 OHPC /weekly from 16 20 to 36 weeks

CL< 25 mm Time from randomization to delivery

82 83

Serra [30] 2012 Spain 2005 2008 All DC twin pregnancies from 11 13 to 20

weeks of gestation

Progesterone pessaries 200 400 mg/daily from 20 to 34 weeks

None (unselected population)

PTB< 37 weeks 194 96

Wood [31] 2102 Canada 2006 2010 All twin pregnancies at 16 + 0–20 + 6 weeks

of gestation

Progesterone gel 90 mg/daily None

(unselected population)

GA at birth 42 42

Aboulghar [32]

2012 Egypt 2008 2010 All DC twin pregnancies at 15 19 weeks of

gestation

Vaginal natural progesterone suppositories 200 mg twice daily

None (unselected population) PTB< 37 and 34 weeks 49 42

Combs [33] 2011 United States 2004 2009 All DC twin pregnancies at 16 24 weeks of

gestation Intramuscular 17OHPC Intramuscular injection of 250 mg 17 P/weekly from 16 24 to 34 weeks None (unselected population) Composite neonatal morbidity 160 78

Lim [34] 2011 The Netherlands 2006 2009 All twin pregnancies at 15 19 weeks of

gestation

Intramuscular 17OHPC

Intramuscular injection of 250 mg 17 OHPC /weekly from 15 19 to 34 weeks None (unselected population) Adverse neonatal outcome 336 335 Rode [35] 2011 Denmark/Austria/United Kingdom

2006 2008 All diamniotic pregnancies at 18 24 weeks Vaginal progesterone

Progesterone pessaries 200 mg/ daily from 20 23 + 6–33 + 6 weeks

None (unselected population) PTB< 34 weeks 334 343 Klein [36] 2011 Denmark/Austria/United Kingdom

2006 2009 Twin pregnancies with CL< 30 mm at 20 24 weeks

Progesterone pessaries 200 mg/ daily from 20 23 + 6–33 + 6 weeks

CL< 30 mm PTB< 34 weeks 17 30

Cetingoz [37]

2011 Turkey 2004 2007 All twin pregnancies at 24 weeks Vaginal

progesterone

Progesterone suppositories (100 mg/daily from 24 to 34 weeks

None (unselected population) PTB< 37 weeks 28 39 Durnwald [38]

2010 United States 2004 2005 Twin pregnancies with a short C at 16 20

weeks of gestation Intramuscular 17OHPC Intramuscular injection of 250 mg 17 OHPC/weekly from 16 20 to 35 weeks CL<10th or 25th percentile Delivery or fetal death<35 weeks 20 201 F. D ’Antonio, V. Berghella, D. Di Mascio et al. European Journal of Obstetrics & Gynecology and Repr oductiv e Biology 26 1 (202 1) 16 6– 17 7 16 9

(5)

RevMan 5.3 (Copenhagen: The Nordic Cochrane Centre,The CochraneCollaboration,2014)andStata,version13.1(StataCorp., College Station, TX, 2013) were used to perform head-to-head meta-analysesMTMs,respectively.

Results

Generalcharacteristicsofthestudies

383articleswereidentiﬁed,41wereassessedwithrespectto theireligibilityforinclusionand28studieswereincludedinthe systematicreview(Table1andFig.1)[19_–46],while26 studies were included in the meta-analysis [19,21–25,27,20–46]. The excludedstudiesandthereasonsfortheirexclusionareoutlinedin SupplementaryTable2.

Fourstudies(2473 pregnancies) exploredtheuseofcervical pessary vs nointervention in reducing the risk of PTB in twin pregnancies; out of these, two studies included an unselected populationof twins while two includedthose with a short CL (Table1).Regardingthetypeofpessaryadopted,threestudiesused ArabinwhileoneusedBiotequecup.

Eighteen studies (5821 pregnancies) explored the role of progesterone compared tonointervention in reducing therisk of PTB intwins (Table1); outof these,14included unselected populations of twin pregnancies, while 4 exclusively included those with a short cervical length (CL). Regarding the type of progesterone,7studiesusedintra-muscularinjectionof17-OHCP, while11vaginalprogesterone.

Fourstudies(224 pregnancies)exploredthe roleof cerclage comparedtonointerventioninreducingtheriskofPTB(Table1); alltheincludedstudiesusedMcDonaldcerclageandallexceptone includedtwinswithashortCLonmid-trimesterultrasound.There was only one study comparing two interventions (pessary vs vaginalprogesterone).

Theresultsofthequalityassessmentoftheincludedstudies usingRoB2toolarepresentedinSupplementaryTable3.Resultsof thedirect,indirectand networkestimatesofallnetwork meta-analyses comparingtherisk of PTB (<34weeks),including the qualityofevidenceasassessedbytheGRADEscoreandthereasons fordowngrading itare reportedinSupplementaryTable 4,and

Figs.2and3.Forthemainoutcome,thequalityofevidencefor mostoftheperformedcomparisonswasmoderate.

Synthesisoftheresults

Primaryoutcome(PTB<34weeks)

Tenstudiesexploredtheroleofvaginalprogesteronecompared to no intervention in reducing the risk of PTB in unselected populationsoftwinpregnancies.Overall,vaginalprogesteronedid notreducetheriskofPTB<34weeksofgestation(6studies,2672 pregnancies;RR:1.04;95%CI:0.84 1.30-p=0.7).Likewise,there wasnocontributionofeitherintra-muscular17-OHPC(3studies, 923pregnancies;RR:1.09;95%CI:0.74 1.60-p=0.7)orpessary (1study,1177pregnancies;RR:1.07;95%CI:0.82 1.38-p=0.6)in affectingtheriskofPTB,whiletherewasnostudyexploringthe roleofcerclageinreducingPTBinunselectedpopulation(Table2,

Fig.4,SupplementaryFigs.1–4).

Therewas onlyonestudycomparingtwo different interven-tions (pessary vs vaginal progesterone) in asymptomatic twin pregnancieswithCL<38mmat16 22weeksofgestation).The studyreported nodifferencein the risk of PTB <34 weeks of gestationbetweenthetwogroups[19].

When stratifying the analysis on spontaneous PTB, neither pessary(1study,1177pregnancies;RR:1.05;95%CI:0.79 1.41-p =0.7), vaginal (3studies,2134 pregnancies; RR:1.08;95 % CI: 0.80 1.46- p=0.6)or intra-muscular17-OHPC(2studies,893

T able 1 (Continued ) A uthor Y ear Country Period considere d Inclusion crite ria a Inte rvention Descrip tion of int erven tion Indication for int erve ntion Primary outcome Int erve ntion sample size Comparison sample size Briery [ 39 ] 2 0 0 9 U nite d S tate s N S All twin pre gnancies at 20 30 w eeks Intra muscular 1 7OHPC Intr amuscular injection of 250 mg 17 OHPC w eekl y fro m 2 0 30 to 34 w eeks N one (unselect ed population) PTB < 3 5 w eeks 1 6 1 4 N orman [40 ] 20 09 U nite d Kingdom 20 04 20 08 All twin pre gnancies at 22 weeks of ges tation Vaginal pro g ester one Pr oge ste ro ne g e l 1 0 0 mg/dail y N one (unselect ed population) PTB or IUD < 34 w eeks 250 250 R ouse [ 41 ] 2 0 0 7 U nite d S tate s 2 0 0 4 20 06 All twin pre gnancies at 1 6 20 weeks of ges tation Intra muscular 1 7OHPC Intr amuscular injection of 250 mg 17 OHPC /w eekl y fr o m 1 6 20 to 35 w eeks N one (unselect ed population) Deli v ery or fetal death < 3 5 weeks 325 330 F onseca [42 ] 20 07 U nite d Kingdom 20 03 20 06 Twin pr egnancies with CL < 1 5 mm at 20 25 w eeks Vaginal pro g ester one Vaginal micr onized pr oge ste ron e 20 0 mg/dail y fro m 2 4 to 3 3 + 6 w eeks CL < 1 5 mm sPTB < 34 weeks 11 1 3 Berghella [ 43 ] 20 04 U nite d S tate s 1 998 20 03 Asympt omatic twin pr egnancies with CL < 25 mm Cerclag e M cDonald cer clage CL < 25 mm PTB < 3 5 w eeks 3 1 Eskndar [ 44 ] 20 07 Saudi Ar abia 20 04 20 06 All twin pre gnancies at 1 2 1 4 w eeks of ges tation Cerclag e M cDonald cer clage N one (unselect ed population) PTB < 3 7 weeks 7 6 1 0 0 Althusius [ 45 ] 20 0 1 The N ether lands 1 995 20 0 0 Twin pr egnancies with CL < 25 mm Cerclag e M cDonald cer clage CL < 25 mm PTB < 34 w eeks 8 9 R ust [ 46 ] 2 0 0 1 U nite d S tate s 1 998 20 0 0 Twin pr egnancies with US dilatation or shortening of the cervix at 1 6 2 4 weeks of ges tation Cerclag e M cDonald cer clage CL < 25 mm G A at birth 1 3 1 4 CL, cervical length; PTB, pret erm birth; sPTB, spontaneous pret erm birth; 1 7 P, 1 7-pr og ester one; 17-OHPC, 1 7a-h ydr o xypro g est erone caproa te ; G A, g estational ag e; IUD, intra uter ine death; NS, no t speci ﬁ ed. a inclusion crit eria refer to the main population anal yzed in each indiv idual study , n o t considering the sub-grou p anal y ses.

(6)

pregnancies;RR:1.16;95%CI:0.92 1.46-p=0.2)wereassociated withasigniﬁcantreductionintheriskofPTBcomparedtocontrols, whiletherewasnostudyoncerclagewhichexploredthisoutcome inanunselectedpopulationoftwinpregnancies(Table2).

When a network meta-analysis was used to compare the differentinterventions, nodifferenceemergedbetweenpessary, vaginalorintra-muscular17-OHPCinreducingtherateofPTB<34 weeks when such interventions were applied to both an unselectedpopulationoftwins(Table3).

The cumulative probabilities of being the most efﬁcacious treatments for pessary, vaginal progesterone, intramuscular progesterone or expectant management were: 31.6 %, 20.4 %, 43.8%and 4.2 %,respectively,when consideringan unselected populationoftwinpregnancies.ForwomenwithaCL25mm,the probabilitiesofbeingthebestapproachwere:4.0%(pessary),10.7 %(vaginalprogesterone),24.4%(intramuscularprogesterone)and 0.6 % (expectant management). For this subgroup of women, cerclageshowedthehighestprobabilityofbeingthebestapproach Fig.1.Systematicreviewﬂowchart.

Fig.2.Allwomen–Studylimitationsforeachnetworkestimateforpairwisecomparisonsofdifferenttreatments,inwomenatriskofpretermbirth(<34weeks).Calculations arebasedonthecontributionsofdirectevidence.Thecoloursarebasedontheriskofbias(green:low;yellow:moderate;red:high).Theinitialjudgementsabouttheriskof biasinthedirectestimatesareshownontherightsideoftheﬁgure(thereisnodirectevidenceforAC,CDandAD).Thenamesofthetreatmentsarereportedbelow. A=Pessary;B=Expectantmanagement;C=VaginalProgesterone;D=intramuscular17α-hydroxyprogesteronecaproate(IM17-OHPC).

(7)

Fig.3. Womenwithcervicallength<25mm-Studylimitationsforeachnetworkestimateforpairwisecomparisonsofdifferenttreatments,inwomenatriskofpretermbirth (<34weeks).Calculationsarebasedonthecontributionsofdirectevidence.Thecoloursarebasedontheriskofbias(green:low;yellow:moderate;red:high).Theinitial judgementsabouttheriskofbiasinthedirectestimatesareshownontherightsideoftheﬁgure(thereisnodirectevidenceforBDandBE).Thenamesofthetreatmentsare reportedbelow.

A=Pessary;B=Expectantmanagement;C=VaginalProgesterone;D=intramuscular17α-hydroxyprogesteronecaproate(IM17-OHPC);E=Cerclage.

Table2

Resultsofthehead-to-headmeta-analysiscomparingthelikelihoodofpretermbirthintwinpregnanciestreatedwith(1)pessary;(2)vaginalprogesterone;(3)intramuscular 17α-hydroxyprogesteronecaproate;(4)cerclageversusuntreatedtwinpregnancies(expectantmanagement).Foreachoutcome,pooledriskestimatesarereportedforall interventionscombinedandseparatelyforeachintervention.Allanalyseswereperformedtwice,eitherconsidering(a)anunselectedpopulationoftwinpregnanciesand(b) twinpregnancieswithcervicallength<25mm.

A.Unselectedpopulationoftwinpregnancies B.Womenwithcervicallength<25mm N.studies (sample) PooledRR (95%CI) p I2 ,% N.studies (sample) PooledRR (95%CI) p I2 ,%

1.Pretermbirth<34weeksintreatedvsuntreatedtwinpregnancies

Pessaryvsexpectantmanagement 1(1177)[22] 1.07(0.82 1.38) 0.6 – 3(371)[20–22] 0.91(0.42 1.96) 0.8 79 Vaginalprogesteronevs

expectantmanagement

6(2672)[19,28,30,32,35,37] 1.04(0.84 1.30) 0.7 38 1(35)[26] 0.95(0.64 1.40) 0.8 – IM17-OHPCvsexpectantmanagement 3(923)[31,37,39] 1.09(0.74 1.60) 0.7 39 1(165)[27] 1.45(0.94 2.25) 0.09 – Cerclagevsexpectantmanagement 0 – – – 3(48)[41,43,44] 2.13(0.74 6.15) 0.16 31 2.Spontaneouspretermbirth<34weeksintreatedvsuntreatedtwinpregnancies

Pessaryvsexpectantmanagement 1(1177)[22] 1.05(0.79 1.41) 0.7 – 2(348)[21,22] 0.72(0.25 2.06) 0.5 67 Vaginalprogesteronev

expectantmanagement

3(2134)[19,28,35] 1.08(0.80 1.46) 0.6 41 0 – – – IM17-OHPCvsexpectantmanagement 2(893)[31,39] 1.16(0.92 1.46) 0.2 0 0 – – – Cerclagevsexpectantmanagement 0 – – – 2(21)[41,43] 1.20(0.29 4.90) 0.8 12 3.Pretermbirth<37weeksintreatedvsuntreatedtwinpregnancies

Pessaryvsexpectantmanagement 2(1985)[22,23] 0.96(0.89 1.02) 0.2 0 2(157)[21,22] 0.95(0.77 1.16) 0.6 0 Vaginalprogesteronevs

expectantmanagement

8(3250)[19,28,30–32,35,37,39] 1.03(0.94 1.13) 0.5 33 1(35)[26] 0.98(0.80 1.21) 0.9 – IM17-OHPCvsexpectantmanagement 5(1879)[25,31,32,37,39] 1.03(0.95 1.12) 0.4 19 2(178)[27,32] 0.95(0.70 1.30) 0.8 39 Cerclagevsexpectantmanagement 1(176)[42] 1.15(0.70 1.91) 0.6 – 3(48)[41,43,44] 1.20(0.92 1.57) 0.2 0 4.Pretermbirth<32weeksintreatedvsuntreatedtwinpregnancies

Pessaryvsexpectantmanagement 2(1985)[22,23] 0.92(0.70 1.20) 0.5 0 1(23)[41] 4.58(0.66 33.4) 0.13 – Vaginalprogesterone

vsexpectantmanagement

4(2514)[19,28,30,35] 1.02(0.82 1.26) 0.9 0 1(35)[26] 0.72(0.42 1.26) 0.3 – IM17-OHPCvsexpectantmanagement 3(1194)[25,31,32] 1.09(0.56 2.11) 0.8 66 2(178)[27,32] 1.24(0.29 5.36) 0.8 85 Cerclagevsexpectantmanagement 0 – – – 3(48)[41,43,44] 1.62(0.52 4.99) 0.4 46 5.Pretermbirth<28weeksintreatedvsuntreatedtwinpregnancies

Pessaryvsexpectantmanagement 2(1985)22,23 0.98(0.60 1.59) 0.9 10 2(157)[20,21] 0.87(0.15 1.08) 0.9 57 Vaginalprogesterone

vsexpectantmanagement

4(2514)[19,28,30,35] 0.94(0.69 1.27) 0.7 0 1(35)[26] 0.71(0.27 1.87) 0.5 – IM17-OHPCvsexpectantmanagement 3(1194)[25,31,32] 0.85(0.51 1.42) 0.5 0 1(15)[32] 0.33(0.09 1.28) 0.1 – Cerclagevsexpectantmanagement 0 – – – 3(48)[41,43,44] 1.40(0.48 4.08) 0.5 24 6.Pretermbirth<24weeksintreatedvsuntreatedtwinpregnancies

Pessaryvsexpectantmanagement 0 – – – 1(23)[20] 1.83(0.19 17.5) 0.6 – Vaginalprogesteronevs

expectantmanagement

1(1169)19

0.58(0.31 1.09) 0.09 – 0 – – – Cerclagevsexpectantmanagement 0 – – – 3(48)[41,43,44] 1.15(0.51 2.56) 0.7 0 RR,RiskRatio;CI,ConﬁdenceInterval;IM17-OHPC=intramuscular17α-hydroxyprogesteronecaproate.

(8)

(60.3 %). However, theseﬁgures werebases upon avery small numberofincludedcases,whichaffectstherobustnessofthese ﬁndings.

Secondaryoutcomes

Preventionof PTBand maternaloutcomes. Fifteenstudies(6121 pregnancies)explored the roleof thedifferent interventionsin preventing PTB < 37 weeks of gestation. None of these interventions, including pessary (2 studies, 1985 pregnancies; RR:0.96;95 %CI:0.89 1.02-p =0.2),vaginalprogesterone(8 studies,3250pregnancies;RR:1.03;95%CI:0.94 1.13-p=0.5), intra-muscular17-OHPC(5studies,1879pregnancies;RR:1.03;95 %CI:0.95 1.12-p=0.4)orcerclage(1study,176pregnancies;RR: 1.15;95%CI:0.10–1.91-p=0.6),signiﬁcantlyreducedtheriskof PTB, whencompared towomenwithnointervention.Likewise, noneoftheexploredinterventionsreducedtheriskofPTB<32or PTB<28weekscomparedtocontrols(allp>0.05;Table2).

Fivestudies(2084pregnancies)exploredtheriskofPPROMin women receiving theseinterventions.Neither pessary(1 study, 808pregnancies;RR:1.04;95%CI:0.67–1.64-p=0.9),vaginal progesterone(1study,290pregnancies;RR:0.82;95%CI:0.20– 3.38 - p = 0.8) nor intra-muscular 17-HPC (3 studies; 986 pregnancies; RR:1.21;95%CI:0.77–1.89-p=0.4)reducedthe risk of PPROM compared to women not receiving these inter-ventions(SupplementaryTable5).

Prophylacticadministrationofvaginalprogesterone(5studies, 1924 pregnancies; RR 0.93, 95 % CI 0.88 0.99, p = 0.02) was associatedwithasigniﬁcantlylowerrateofCS,whiletherewasno differenceintheneedforCSinwomenreceivingintra-muscular 17-OHPCcomparedtocontrols(3studies,1558pregnancies;RR: 1.01; 95%CI:0.93–1.10,p =0.8).Conversely,theriskofCSwas signiﬁcantlyincreased(1study,808pregnancies;RR1.18,95%CI 1.02–1.36,p=0.03)inwomenreceivingpessary,althoughonlyone studywasincludedinthisoutcome(SupplementaryTable5).

Whenassessing theassociationbetweentheexplored inter-ventionsandtheneedforadministrationofsteroidsortocolytics, the use of either pessary, vaginal and intramuscular 17-OHPC progesterone was not associated with a reduction in steroids

administrationortocolytictherapycomparedtocontrols(allp> 0.05)(SupplementaryTable5).

Therewas nodifferenceintherate ofvaginaldischargeand infections,UTIorchorioamnionitisinwomentreatedwithpessary, vaginal progesterone or intra-muscular 17-OHPC compared to controls(allp>0.05)(SupplementaryTable5).

Furthermore,therewasnosigniﬁcantdifferenceinthemean gestational age at birth between women receiving pessary (2 studies,1985pregnancies;MD:0.09;95%CI:-0.30;0.48-p=0.6), vaginalprogesterone(5studies,2023pregnancies;MD:-0.09;95% CI:-0.32;0.14-p=0.5)orintra-muscular17-OHPC(5studies,1879 pregnancies:MD:0.33;95%CI:-0.71;1.38-p=0.6)comparedto controls when such interventions were used in an unselected populationoftwinpregnancies(SupplementaryTable6).

Nodifferencewasalsofoundwhencomparingpessary(1study, 2354pregnancies; MD:-22.0; 95 % CI:-64.0; -20.0 - p = 0.3), vaginalprogesterone(4studies,3152pregnancies;MD:-24.6;95% CI:-113.0;63.5-p=0.6)orintra-muscular17-OHPC(3studies, 1980pregnancies:MD:36.5;95%CI:-119;192-p=0.6)intermsof birthweight(SupplementaryTable6).

Finally,pessarywasnotassociatedwithareductionoflengthof in-hospitalstay(1study,136pregnancies;MD:-3.0;95%CI:-9.21; 3.21-p=0.3),whilewomenreceivingvaginalprogesteronehada meanlongerhospitalizationcomparedtocontrols(2studies,801 pregnancies;MD:3.30;95%CI:1.25;5.35–p<0.01).Conversely, women receiving intra-muscular 17-OHPC progesterone had a meanlowerhospitalizationcomparedtocontrols(2studies,636 pregnancies; MD: -3.93; 95 % CI: -6.87; -0.99 – p = 0.009) (SupplementaryTable6).

Secondaryoutcomes:fetalandneonataloutcomes

Twelve studies (2 pessary, 5 vaginal progesterone, 5 on intramuscular 17-OHPC and nonecerclage; 11612 pregnancies) explored the role of each intervention in reducing the risk of perinatalmortality.Overall, neitherpessary (RR:0.69; 95 %CI: 0.40–1.19-p=0.2),norvaginalprogesterone(RR:1.13;95%CI: 0.61–2.10-p=0.7)orintra-muscular17-OHPC(RR:0.82;95%CI: 0.46–1.49-p=0.5)wereassociatedwithasigniﬁcantreductionin theriskofIUD.Likewise,noneoftheseinterventionsreducedthe

Fig.4. Allwomen-Contributionplotforthemultiple-treatmentmeta-analysisofpretermbirth(<34weeks).Thesizeofeachsquareisproportionaltotheweightattachedto eachdirectsummaryeffect(horizontalaxis)fortheestimationofeachnetworksummaryeffect(verticalaxis).Thenumbersexpresstheweightsaspercentages. a=Pessary;b=Expectantmanagement;c=VaginalProgesterone;d=intramuscular17α-hydroxyprogesteronecaproate(IM17-OHPC).

(9)

riskofNNDorPNDwhenappliedtoanunselectedpopulationof twinpregnancieswithaCL25mmonultrasound(allp>0.05; SupplementaryTable7).

Neonataland perinatalmortalitywas notsigniﬁcantlyin ﬂu-encedbypessary,vaginalprogesteroneorintra-muscular17-OHPC whencomparedtopregnanciesnotreceivingtheseinterventions. Finally,cervical pessary,vaginalprogesterone orintra-muscular 17-OHPCdidnotmodifytheriskofcompositemorbidity,Apgar score<7,BW<2500and1500g,RDS,BPD,needformechanical ventilation, IVH, PVL, NEC,PVL, NEC,retinopathy, and neonatal sepsisoradmissiontoNICU(SupplementaryTable7).

Sub-groupanalyses:twinpregnancieswithshortcervicallength(25 mm)

WhentheanalysiswasrestrictedtothepregnancieswithCL 25mmonultrasound,therewasnocontributionofeitherpessary (3studies,371pregnancies;RR:0.91;95%CI:0.42 1.96-p=0.8), vaginalprogesterone(1study,35pregnancies;RR:0.95;95%CI: 0.64 1.40 - p = 0.8), intra-muscular 17-OHPC (1 study, 165 pregnancies;RR:1.45;95%CI:0.94 2.25-p=0.09)orcerclage(3 studies,48pregnancies;RR:2.13;95%CI:0.74 6.15-p=0.16)in reducingtheriskofoverallPTB_<34weeksofgestation(Table2, SupplementaryFigs.5–9).

When taking into account studies reporting the risk of spontaneousPTB,both pessary(2studies,348pregnancies;RR: 0.72;95 %CI: 0.25 2.06–p =0.5) and cerclage(2studies,21 pregnancies;RR:1.20;95%CI:0.29 4.90–p=0.8)didnotreduce the risk of spontaneous PTB in twin pregnancies with a short cervix.Thecervicalpessary,vaginalprogesterone,intra-muscular 17-OHPCorcerclagedidnotreducetheriskofPTB<37,<32,<28or <24weeksofgestationsinwomenwithashortcervix(Table2). Cervical pessary, vaginal progesterone, intra-muscular 17-OHPCorcerclagedidnotreducetheriskofPPROM,CS,needfor tocolytics,steroidsadministration,chorioamnionitisorinfection inwomenwithashortcervix(SupplementaryTable5).

When assessing perinatal outcomes, none of the explored intervention was associated with a reduced risk of IUD, NND, composite morbidity, Apgar score _<7, IVH, PVL, retinopathy, neonatal sepsis or admission toNICU. Conversely, IM 17-OHPC wasassociatedwithahigherriskofperinataldeath(RR:9.11,95% CI1.17–71.1,p=0.04),althoughthisoutcomewasassessedbyonly onestudy.Vaginalprogesteronesigniﬁcantlyreducedtheriskof

verylowBW(RR:0.46,95%CI0.2 0.7,p=0.002)althoughthis outcomewasassessedonlybyonestudy,whilecervicalcerclage wasassociatedwithasigniﬁcantlyhigherriskofbothBW<2500 (RR:1.36,95%CI1.0–1.8,p=0.02)andBW<1500(RR:3.14,95%CI 1.5–6.6,p=0.002),botharelikelytobeastheconsequenceofPTB (SupplementaryTable7).

Asperunselectedtwinpregnancies,nodifferencewasfound betweenpessary,vaginalorintra-muscular17-OHPCinreducing therateofPTB<34weekswhenanetworkmeta-analysiswasused to compare the different interventions also in pregnancies presentingwithaCL25mmonultrasound(Table3).

Discussion Mainﬁndings

The ﬁndings fromthis systematicreview show that neither cervicalpessarynorprogesteroneareassociatedwithasigniﬁcant reduction in the risk of PTB when applied to an unselected population of twin pregnancies. Likewise,none of these inter-ventions are associated with a reduced risk of maternal and perinataloutcomes,includingmortalityandmorbidity.

Strengthsandlimitations

Thorough literature search, assessment of a multitude of maternal and perinatal clinical outcome, stratification of the analysisaccordingtocervicallengthandcomputationofmultiple treatment meta-analyses represent the main strengths of the present systematic review. The relatively small sample size of someoftheexploredtreatmentswasoneofthemajorlimitations of the present systematic review. Furthermore, the primary outcome of the included studies was mainly the incidence of PTB,andthelargemajorityofthesestudieswerenotpoweredfor theneonataloutcomes,includingmorbidityandmortality.Inthis scenario, the beneficial effect of progesterone in reducing the incidenceofsomeaspectofneonatalmorbiditymayrepresenta spurious_findingratherthananactualeffectofprogesteroneonthe perinatal outcome, highlighting the need of larger studies adequately powered on neonatal outcome in order to confirm such association. Likewise, we could not stratify the analysis according to maternal and pregnancy characteristics. More Table3

Resultsofthemultiple-treatmentmeta-analysiscomparingtheriskofpre-termbirth(<34weeks)intwinpregnanciestreatedwith(1)pessary;(2)vaginalprogesterone;(3) intramuscular17α-hydroxyprogesteronecaproate;(4)cerclage.Allanalyseswereperformedtwice,eitherconsidering(a)anunselectedpopulationoftwinpregnanciesand (b)twinpregnancieswithcervicallength<25mm.

Unselectedpopulationoftwinpregnancies Twinpregnancieswithcervicallength<25mm

RR(95%CI) p RR(95%CI) p 1.Pessaryvs VaginalProgesterone 0.98 (0.58 1.67) 0.9 0.89 (0.08 16.7) 0.9 2.Pessary vs IM17-OHPC 1.12 (0.60 2.07) 0.7 2.0 (0.24 16.6) 0.5 3.VaginalProgesterone vs IM17-OHPC 1.14 (0.70 1.85) 0.6 2.19 (0.07 65.2) 0.7 4.Pessary vs cerclage – – 4.73 (0.62 36.1) 0.13 5.VaginalProgesterone vs cerclage – – 5.24 (0.21 129) 0.3 6.IM17-OHPC vs cerclage – – 2.37 (0.20 27.7) 0.5

(10)

importantly,therewasalargeheterogeneityinthedefinitionof shortcervixamongtheincludedstudiesandthesubgroupanalysis includingthepregnancieswithaCL<25mmonlywasaffectedby small number of includedstudies and even smallernumber of events. Moreover, computation of the effect of cerclage was affected by the lack of data on unselected population of twin pregnanciesandverysmallsamplesizeoftheincludedcasesinthe sub-group analyses includingwomen with a short cervix only. Finally,anynetworkmeta-analysisisbasedonthreeassumptions: absence of heterogeneity, transitivity and coherence. Although significant incoherence was not found, heterogeneity and/or intransitivitywasdetectedinseveralofthenetworks.Thisisan important limitationof theresults,althoughassessedusingthe GRADE approach [13]. Although this resulted in several “low” quality of evidence ratings - which should be interpreted accordingly - italso givesconfidencein resultsevaluatedas of “moderate” and “high” quality. Despite these limitations, the present study represents the most comprehensive up to date systematicreviewontheroleofcervicalpessary,progesteroneand cerclageinreducingtheriskofPTBintwinpregnancies. Comparisonwithothersystematicreviews

ArecentsystematicreviewbyJardeetal.exploredtheroleof theseinterventionsinreducingtheriskofPTBintwins.Theprimary outcomeswerePTB<37and34weeksofgestationandneonatal death.Theauthorsreportedthatnoneoftheexploredinterventions reduced the risk of the primary outcomes. However, women receivingvaginalprogesterone,hadasignificantreductioninsome secondaryoutcomes,includingverylowbirthweightandneedfor mechanical ventilation[47].Inanindividualpatientdata(IPD) meta-analysisincludingonlyRCTscomparingvaginalprogesteronewith placebo/notreatmentinwomenwithatwingestationanda mid-trimester sonographic cervical length 25 mm, Romero et al. reported that vaginal progesterone, compared with placebo/no treatment,wasassociatedwithastatisticallysignificantreductionin the risk of PTB < 35, 34, 33, 32 and 30 weeks of gestation. Furthermore,progesteronewasalsoassociatedwithareductionin NND, RDS, need for mechanical ventilation, BW < 1500 g, and compositeneonatal morbidityandmortality [48].There wasno significantdifferenceintheneurodevelopmentaloutcomeat4–5 yearsofagebetweenthevaginalprogesteroneandplacebogroups. However,thestudywasaffectedbythesmallnumberofincluded women(n=303)resultinginlowstatisticalpowerforsomeofthe exploredoutcomes.However,itisimportanttoacknowledgethe advantagesoftheIPDapproachofthismeta-analysis[48].

Likewise,asystematicreviewbySacconeetal.,includingthree RCTonpessary vs no interventionin twins didnot report any signiﬁcantcontributionofsuchinterventioninreducingtheriskof PTBorimprovingtheneonataloutcomealthoughthereviewwas hamperedbythesmallnumberofincludedstudiesandthelarge heterogeneityoftheinclusioncriteria[49].

Finally,anotherIPDmeta-analysisbySacconeetal.exploredthe roleofcerclageintwinpregnancieswithashortcervicallength reporting no difference in the rate of PTB between women undergoingcerclageandcontrols.Conversely,theratesofverylow BWandofRDSweresigniﬁcantlyhigherinthecerclagegroupthan inthecontrolgroup[50].

Clinicalandresearchimplications

Many would argue that the prevention of PTB is the most important challengein maternal-fetalmedicine.PTB represents the maindeterminant of theperinatalmortalityand morbidity worldwidewithestimatedﬁnancialcostofaround$26billionper yearonlyintheUnitedStates[1].Themainproblemwhentryingto

applypreventativestrategiesisthatPTBisnotauniquediseasebut ratherasyndromecharacterizedbymultipleetiologieswhichacts byactivating what it was deﬁned as“the commonpathwayof parturition”,resultingintheanatomical,biochemical, endocrino-logic,andclinicaleventswhichleadtoPTBmainlythroughthree events: increased uterine contractility, cervical ripening and decidualmembraneactivation.

In singleton pregnancies, cervical pessary, cerclage and progesterone have been similarly shown to reduce the risk of PTB, especially when these interventions are applied to a populationatriskforthiscondition,suchaswomenwithapast historyofprior PTBora shortcervicallengthonmid-trimester ultrasound[3].

Inthepresentsystematicreview,wedidnotﬁndanybeneﬁcial effectofanyoftheexploredinterventioninreducingtheprimary outcomeintwinpregnancies.

However, PTB is themajor cause of perinatalmorbidity and mortalityalsointwingestations[1,2,51].Alikelyexplanationofthe lackof associationbetween the exploredinterventions and the observedoutcomesmightbetheheterogeneouspathophysiologyof PTBintwinpregnancies.Contrarytosingletonpregnancies,PTBin twins is mostly the result of uterine overdistension leading to increasedcontractility.Inthis scenario,interventionsdirectedto maintaincervicalcompetencearelesslikelytobeeffectivethanin singletons.Moreover,theincreasedriskofPTBintwinsisalsorelated tocomplicationsrelatedtomonochorionicityandmonoamnionicity [52–55].ThereforeTherefore, drawingobjective evidenceonthe actualroleofthese strategiesintwinpregnanciesischallenging becausemostoftheincludedstudiesconsideredmainlyunselected populations oftwins, while those focusing on the twinpregnancies at risk were limited by thevery small sample size and thelarge heterogeneityinthedeﬁnitionofpregnanciesatrisk.

Conclusions

Cervicalpessary,progesteroneandcerclagedonotsigniﬁcantly reducetheriskofPTBintwins,eitherwhentheseinterventionsare appliedtoallpregnanciesandtothosewithashortcervix.Large multicenterstudieswhichareadequatelypoweredforperinatal outcomesareneededinordertoconﬁrmtheseobservationsand elucidatewhethertheadministrationofprogesteroneinwomen withtwinpregnancyandshortcervixcouldimprovetheperinatal outcome.

Funding

Nofundingwasobtainedforthepresentsystematicreview. DeclarationofCompetingInterest

The authors declare that they have no known competing ﬁnancial interests or personal relationships that could have appearedtoinﬂuencetheworkreportedinthispaper.

AppendixA.Supplementarydata

Supplementarymaterialrelatedtothisarticlecanbefound,inthe onlineversion,atdoi:https://doi.org/10.1016/j.ejogrb.2021.04.023. References

[1]ChauhanSP,ScardoJA,HayesE,AbuhamadAZ,BerghellaV.Twins:prevalence, problems,andpretermbirths.AmJObstetGynecol2010;203:305–15. [2]DiMascioD, AcharyaG,KhalilA,OdiboA,PrefumoF,LiberatiM,etal.

Birthweightdiscordance and neonatalmorbidity in twin pregnancies:a systematic review and meta-analysis. Acta Obstet Gynecol Scand 2019;98:1245–57.

(11)

[3]Jarde A, Lutsiv O, Beyene J, McDonald SD. Vaginal progesterone, oral progesterone,17-OHPC,cerclage,andpessaryforpreventingpretermbirth inat-risksingletonpregnancies:anupdatedsystematicreviewandnetwork meta-analysis.BJOG2019;126:556–67.

[4]Higgins JPT, Green S. Cochrane handbook for systematic reviews of interventions. Availablefrom:.The CochraneCollaboration; 2011.www.

cochrane-handbook.org.

[5]MoherD, ShamseerL,ClarkeM,GhersiD,LiberatiA,PetticrewM,etal. Preferredreportingitemsforsystematicreviewandmeta-analysisprotocols (PRISMA-P)2015statement.SystRev2015;1(4):1.

[6]HigginsJPT,SterneJAC,SavovicJ,PageMJ,HróbjartssonA,BoutronI,etal.A revisedtoolforassessingriskofbiasinrandomizedtrials.In:ChandlerJ, McKenzieJ,BoutronI,WelchV,editors.Cochranemethods.CochraneDatabase ofSystematicReviews;2016Issue10(Suppl1).

[7]PuhanMA,SchunemannHJ,MuradMH,LiT,Brignardello-PetersenR,SinghJA. AGRADEWorkingGroupapproachforratingthequalityoftreatmenteffect estimatesfromnetworkmeta-analysis.BMJ2014;349:g5630.

[8]DerSimonianR,LairdN.Meta-analysisinclinicaltrials.ControlClinTrials 1986;7(3):177–88.

[9]HigginsJP,ThompsonSG,DeeksJJ,AltmanDG.Measuringinconsistencyin meta-analyses.BMJ2003;327:557–60.

[10]Salanti G, Higgins JP, Ades A, Ioannidis JP. Evaluation of networks of randomizedtrials.StatMethodsMedRes2008;17:279–301.

[11]FlaccoME,ManzoliL,RossoA,MarzuilloC,BergaminiM,StefanatiA,etal. ImmunogenicityandsafetyofthemulticomponentmeningococcalBvaccine (4CMenB): a systematic review and meta-analysis. Lancet Infect Dis 2018;4:461–72.

[12]WhiteIR.Networkmeta-analysis.StataJ2015;15:951–85.

[13]ChaimaniA,HigginsJPT,MavridisD,SpyridonosP,SalantiG.Graphicaltoolsfor networkmeta-analysisinSTATA.PLoSOne2013;8:e76654.

[14]LuG,WeltonNJ,HigginsJPT,WhiteIR,AedesAE.Linearinferenceformixed treatmentcomparisonmeta-analysis:atwo-stageapproach.ResSynthMeth 2011;2:43–60.

[15]VeronikiAA,VasiliadisHS,HigginsJP,SalantiG.Evaluationofinconsistencyin networksofinterventions[publishedcorrectionappearsinIntJEpidemiol. 2013Jun;42(3):919].IntJEpidemiol2013;42:332–45.

[16]EfthimiouO,DebrayTP,vanValkenhoefG,TrelleS,PanayidouK,MoonsKGM, etal.GetRealinnetworkmeta-analysis:areviewofthemethodology.Res SynthMethods2016;7:236–63.

[17]ShiC,DumvilleJoC,CullumN.Supportsurfacesforpressureulcerprevention: anetworkmeta-analysis.PLoSOne2018;13:e0192707.

[18]SalantiG,DelGiovaneC,ChaimaniA,CaldwellDM,HigginsJP.Evaluatingthe qualityofevidencefromanetworkmeta-analysis.PLoSOne2014;9:e99682. [19]RehalA,BenkÅZ,DePacoMatallanaC,SyngelakiA,JangaD,CiceroS,etal.Early vaginalprogesteroneversusplacebointwinpregnanciesforthepreventionof spontaneouspretermbirth:arandomized,double-blindtrial.AmJObstet Gynecol2021;224:86.e1-86.e19.

[20]MercedC,GoyaM,PratcoronaL,RodóC,LlurbaE,HiguerasT,etal.Cervical pessaryforpreventingpretermbirthintwinpregnancieswithmaternalshort cervixafteranepisodeofthreatenedpretermlabor:randomisedcontrolled trial.AmJObstetGynecol2019;221(55)e1-55.e14.

[21]DangVQ,NguyenLK,PhamTD,PhaMTN,VuKN,HeYTN,etal.Pessary comparedwithvaginalprogesteroneforthepreventionofpretermbirthin womenwith twin pregnancies andcervical length lessthan 38mm: a randomizedcontrolledtrial.ObstetGynecol2019;133:459–67.

[22]BerghellaV,DugoffL,LudmirJ.Preventionofpretermbirthwithpessaryin twins(PoPPT):a randomizedcontrolledtrial. UltrasoundObstetGynecol 2017;49:567–72.

[23]GoyaM,delaCalleM,PratcoronaL,MercedM,RodoR,MunozB,etal.Cervical pessary to prevent preterm birth in women with twin gestation and sonographic short cervix: a multicenter randomized controlled trial (PECEP-Twins).AmJObstetGynecol2016;214:145–52.

[24]NicolaidesKH,SyngelakiA,PoonLC,dePacoMatallanaC,PlasenciaW,Molina FS,et al. Cervicalpessaryplacement forprevention ofpreterm birthin unselectedtwinpregnancies:arandomizedcontrolledtrial. AmJObstet Gynecol2016;214:3.e1-9.

[25]LiemS,SchuitE,HegemanM,BaisJ,deBoerK,BloemenkamK,etal.Cervical pessariesforpreventionofpretermbirthinwomenwithamultiplepregnancy (ProTWIN): amulticentre, open-labelrandomisedcontrolled trial.Lancet 2013;382:1341–9.

[26]El-RefaieW,AbdelhafezMS,BadawyA.Vaginalprogesteroneforpreventionof preterm laborinasymptomatictwin pregnancieswithsonographicshort cervix:arandomizedclinicaltrialofefﬁcacyandsafety.ArchGynecolObstet 2016;293:61–7.

[27]AwwadJ,UstaIM,GhazeeriG,YacoubN,SuccarJ,HayekS,etal.Arandomised controlleddouble-blindclinicaltrialof17-hydroxyprogesteronecaproatefor thepreventionofpretermbirthintwingestation(PROGESTWIN):evidencefor reducedneonatalmorbidity.BJOG2015;122:71–9.

[28]RobertoE2,RossanaPV2,VeraLJ.BrizotML,HernandezW,LiaoAW,BittarRE, FranciscoRPV,KrebsVLJ,etal.Vaginalprogesteroneforthepreventionof pretermbirthintwingestations:arandomizedplacebo-controlled double-blindstudy.AmJObstetGynecol2015;213:82.e1-82.e9.

[29]SenatMV,Porcher R,WinerN,Vayssière C,DeruelleP,CapelleM,etal. Preventionofpretermdeliveryby17alpha-hydroxyprogesteronecaproatein asymptomatictwinpregnancieswithashortcervix:arandomizedcontrolled trial.AmJObstetGynecol2013;208:194.e1-8.

[30]SerraV,PeralesA,MeseguerJ,ParrillaJJ,LaraC,BellverJ,etal.Increaseddoses of vaginal progesterone for the prevention of preterm birth in twin pregnancies:arandomisedcontrolleddouble-blindmulticentretrial.BJOG 2013;120:50–7.

[31]WoodS,RossS,TangS,MillerL,SauveR,BrantR.Vaginalprogesteroneto preventpretermbirthinmultiplepregnancy:arandomizedcontrolledtrial.J PerinatMed2012;40:593–9.

[32]AboulgharMM,AboulgharMA,AminYM,Al-InanyHG,MansourRT,SerourGI. TheuseofvaginalnaturalprogesteroneforpreventionofpretermbirthinIVF/ ICSIpregnancies.ReprodBiomedOnline2012;25:133–8.

[33]CombsCA,GariteT,MaurelK,DasA,PortoM,ObstetrixCollaborativeResearch Network.17-hydroxyprogesteronecaproatefortwinpregnancy:a double-blind,randomizedclinicaltrial.AmJObstetGynecol2011;204:221.e1-8. [34]LimAC,SchuitE,BloemenkampK,BernardusRE,DuvekotJJ,ErwichJJHM,etal.

17α-hydroxyprogesteronecaproateforthepreventionofadverseneonatal outcome in multiple pregnancies:a randomized controlledtrial. Obstet Gynecol2011;118:513–20.

[35]RodeL,KleinK,NicolaidesKH,Krampl-BettelheimE,TaborA,PREDICTGroup. Preventionofpretermdeliveryintwingestations(PREDICT):amulticenter, randomized, placebo-controlled trialonthe effectofvaginal micronized progesterone.UltrasoundObstetGynecol2011;38:272–80.

[36]KleinK,RodeL,NicolaidesKH,Krampl-BettelheimE,TaborA,PREDICTGroup. Vaginalmicronizedprogesteroneandriskofpretermdeliveryinhigh-risk twinpregnancies:secondaryanalysisofaplacebo-controlledrandomizedtrial andmeta-analysis.UltrasoundObstetGynecol2011;38:281–7.

[37] CetingozE,CamC,SakallıM,KaratekeA,CelikC,SancakA.Progesterone effectsonpretermbirthinhigh-riskpregnancies:arandomized placebo-controlledtrial.ArchGynecolObstet2011;283:423–9.

[38]DurnwaldCP,MomirovaV,RouseDJ,CaritisSN,PeacemanAM,SciscioneA, etal.Secondtrimestercervicallengthandriskofpretermbirthinwomenwith twingestationstreatedwith17alphahydroxyprogesteronecaproate.JMatern FetalNeonatalMed2010;23:1360–4.

[39]BrieryCM,VeillonEW,KlauserCK,MartinRW,ChauhanSC,MagannEF,etal. Progesteronedoesnotpreventpretermbirthsinwomenwithtwins.South MedJ2009;102:900–4.

[40]NormanJE,MackenzieF,OwenP,MactierH,HanrettyK,CooperS,etal. Progesteroneforthepreventionofpretermbirthintwinpregnancy(STOPPIT): a randomised,double-blind,placebo-controlled studyand meta-analysis. Lancet2009;373:2034–40.

[41]RouseDJ,CaritisSN,PeacemanAM,SciscioneA,ThomEA,SpongCY,etal.A trialof17alpha-hydroxyprogesteronecaproatetopreventprematurity in twins.NEnglJMed2007;357:454–61.

[42]FonsecaEB,CelikE,ParraM,SinghM,NicolaidesKH,FetalMedicineFoundation SecondTrimesterScreeningGroup.Progesteroneandtheriskofpretermbirth amongwomenwithashortcervix.NEnglJMed2007;357:462–9.

[43]BerghellaV,OdiboAO,TolosaJE.Cerclageforpreventionofpretermbirthin womenwithashortcervixfoundontransvaginalultrasoundexamination:a randomizedtrial.AmJObstetGynecol2004;191:1311–7.

[44]EskandarM,ShaﬁqH,AlmushaitMA,SobandeA,BaharAM.Cervicalcerclage forpreventionofpretermbirthinwomenwithtwinpregnancy.IntJGynaecol Obstet2007;99:110–2.

[45]AlthuisiusSM,DekkerGA,HummelP,BekedamDJ,vanGeijnHP.Finalresults of the Cervical Incompetence Prevention Randomized Cerclage Trial (CIPRACT):therapeuticcerclagewithbedrestversusbedrestalone.AmJ ObstetGynecol2001;185:1106–12.

[46]RustOA,AtlasRO,ReedJ,vanGaalenJ,BalducciJ.Revisitingtheshortcervix detectedbytransvaginalultrasoundinthesecondtrimester:whycerclage therapymaynothelp.AmJObstetGynecol2001;185:1098–105.

[47]JardeA,LutsivO,ParkCK,BarrettJ,BeyeneJ,SaitoS,etal.Pretermbirth preventionintwinpregnancieswithprogesterone,pessary,orcerclage:a systematicreviewandmeta-analysis.BJOG2017;124:1163–73.

[48]RomeroR,Conde-AgudeloA,El-RefaieW,RodeL,BrizotML,CetingozE,etal. Vaginalprogesteronedecreasespretermbirthandneonatalmorbidityand mortalityinwomenwithatwingestationandashortcervix:anupdated meta-analysis of individual patient data. Ultrasound Obstet Gynecol 2017;49:303–14.

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

[50]SacconeG,RustO,AlthuisiusS,RomanA,BerghellaV.Cerclageforshortcervix intwinpregnancies:systematicreviewandmeta-analysisofrandomized trials using individual patient-level data. Acta Obstet Gynecol Scand 2015;94:352–8.

[51]DiMascioD,KhalilA,RizzoG,BucaD,LiberatiM,MartellucciCA,etal.Riskof fetal loss following amniocentesis orchorionic villus sampling in twin pregnancy:systematicreviewandmeta-analysis.UltrasoundObstetGynecol 2020;56:647–55.

(12)

[52]MONOMONO Working Group. Inpatient vsoutpatient management and timingof deliveryof uncomplicatedmonochorionic monoamniotic twin pregnancy: the MONOMONO study. Ultrasound Obstet Gynecol 2019;53:175–83.

[53]SacconeG,KhalilA,ThilaganathanB,GlinianaiaSV,BerghellaV,D’AntonioF, etal. Weightdiscordance andperinatalmortalityin monoamniotictwin pregnancy: analysis of MONOMONO, NorSTAMP and STORK multiple-pregnancycohorts.UltrasoundObstetGynecol2020;55:332–8.

[54]DiMascioD,KhalilA,D’AmicoA,BucaD,BenedettiPaniciP,FlaccoME,etal. Outcomeoftwin-twintransfusionsyndromeaccordingtoQuinterostageof disease:systematicreviewand meta-analysis.UltrasoundObstetGynecol 2020;56:811–20.

[55]BucaD,DiMascioD,KhalilA,AcharyaG,VanMieghemT,HackK,etal. Neonatalmorbidityofmonoamniotictwinpregnancies:asystematicreview andmeta-analysis.AmJPerinatol2020(July),doi:

http://dx.doi.org/10.1055/s-0040-1714420Epubaheadofprint.PMID:32722824.