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,ia
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.
©2021PublishedbyElsevierB.V.
*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
0301-2115/©2021PublishedbyElsevierB.V.
ContentslistsavailableatScienceDirect
European
Journal
of
Obstetrics
&
Gynecology
and
Reproductive
Biology
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 difficult 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),defined astheruptureofthemembranesbefore37weeksofgestation Cesareansection(CS)
Tocolytictherapy Steroidadministration Vaginaldischarge Vaginalinfections Urinarytractinfections Chorioamnionitis
Intra-uterinedeath(IUD),definedasdeathofeithertwinfrom22 weeksofgestations
Neonataldeath(NND),definedasthedeathofeithernewbornup to28daysoflife
Perinataldeath,definedasthesumofIUDandNND Apgarscore<7at3minutes
Birthweight(BW)<2500g BW<1500g
BWexpressedascontinuousvariable Respiratorydistresssyndrome(RDS)
F.D’Antonio,V.Berghella,D.DiMascioetal. EuropeanJournalofObstetrics&GynecologyandReproductiveBiology261(2021)166–177
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 quantifiedusingtheI2metric[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].
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 BiotequeTMcup CL30 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
Vaginal progesterone
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
Vaginal progesterone
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
Vaginal progesterone
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 progesterone
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
Vaginal progesterone
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
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.
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).
F.D’Antonio,V.Berghella,D.DiMascioetal. EuropeanJournalofObstetrics&GynecologyandReproductiveBiology261(2021)166–177
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.
(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).
F.D’Antonio,V.Berghella,D.DiMascioetal. EuropeanJournalofObstetrics&GynecologyandReproductiveBiology261(2021)166–177
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 spuriousfindingratherthananactualeffectofprogesteroneonthe 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
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.
F.D’Antonio,V.Berghella,D.DiMascioetal. EuropeanJournalofObstetrics&GynecologyandReproductiveBiology261(2021)166–177
[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.
[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.
F.D’Antonio,V.Berghella,D.DiMascioetal. EuropeanJournalofObstetrics&GynecologyandReproductiveBiology261(2021)166–177