Cerclage for Short Cervix on Ultrasound in Singleton Gestations without Prior Spontaneous Preterm Birth: a Systematic Re view and Meta-analysis of Trials using individual patient-level data 1 2 3 4 5 Vincenzo Berghella, Andrea Ciardulli, Orion A. Rust, Meekai To, Katsufumi Otsuki, Sietske 6 7 1 8 Althuisius, Kypros Nicolaides, Amanda Roman, Gabriele Saccone 1 Department of Obstetrics and Gynecology, Division of Maternal-Fetal Medicine, Sidney Kimmel Medical College of Thomas Jefferson University, Philadelphia, PA, USA 2 Department of Obstetrics and Gynecology, Catholic University of Sacred Heart, Rome, Italy 3 Department of Obstetrics and Gynecology, Lehigh Valley Health Network, Allentown, PA, USA 4 Kings College Hospital, London, UK 5 Department of Obstetrics and Gynecology, Showa University Koto Toyosu Hospital, Tokyo, Japan 6 Department of Obstetrics and Gynecology, Dr. Horacio E. Oduber Hospital, Oranjestad, Aruba 7 Harris Birthright Research Centre for Fetal Medicine, Kings College Hospital, London, UK 8 Department of Neuroscience, Reproductive Sciences and Dentistry, School of Medicine, University of Naples Federico II, Naples, Italy Correspondence : Vincenzo Berghella, MD, Department of Obstetrics and Gynecology, Division of Maternal-Fetal Medicine, Thomas Jefferson University, 833 Chestnut, Philadelphia, PA 19107, USA. E-mail: Disclosure : The authors report no conflict of interest Financial Support : No financial support was received for this study Key word : ultrasound-indicated cerclage, prematurity, transvaginal ultrasound, cervical length, intensive care, preterm birth Running title : Cerclage in singletons with no prior preterm birth and short cervical length ABSTRACT Objective: The aim of this systematic review and meta-analysis was to quantify the efficacy of cervical cerclage in preventing preterm birth (PTB) in asymptomatic singleton pregnancies with a mid-trimester short transvaginal ultrasound cervical length (TVU CL) and without prior spontaneous PTB. Methods: Electronic databases were searched from inception of each database until February 2017. No language restrictions were applied. We included all randomized controlled trials (RCTs) of asymptomatic singleton pregnancies without prior spontaneous PTB screened with TVU CL, found to have a midtrimester short CL <25mm, and then randomized to management with either cerclage (i.e. intervention group) or no cerclage (i.e. control group). We contacted corresponding authors of all the included trials to request access to the data and perform a meta- analysis of individual patient data. Data provided by the investigators were merged into a master database specifically constructed for the review. The primary outcome was PTB <35 weeks. The summary measures were reported as relative risk (RR) with 95% confidence interval (CI). The quality of the evidence was assessed using the GRADE approach. Results: Five RCTs, including 419 asymptomatic singleton gestations with TVU CL <25mm and without prior SPTB, were analyzed. No statistically significant differences were found in 2 PTB <35 (21.9% vs 27.7%; RR 0.88, 95% CI 0.63 to 1.23; I =0%; 5 studies, 419 participants), This article is protected by copyright. All rights reserved. Accepted Article <34, <32, <28, and <24 weeks, mean gestational age at delivery, preterm premature rupture of membranes, and neonatal outcomes, comparing women who were randomized in the cerclage group with those who were randomized in the control group, respectively. Planned subgroup analyses revealed a significant decrease in PTB <35 weeks in women with TVU CL <10mm 2 (39.5% vs 58.0%; RR 0.68, 95% CI 0.47 to 0.98; I =0%; 5 studies, 126 participants), in women 2 who received tocolytics (17.5% vs 25.7%; RR 0.61, 95% CI 0.38 to 0.98; I =0%; 5 studies, 154 participants), and in those who received antibiotics (18.3% vs 31.5%; RR 0.58, 95% CI 0.33 to 2 0.98; I =0%; 3 studies, 163 participants). The quality of evidence was downgraded two levels because of serious “imprecision” and serious “indirectness,” and therefore was judged as low. Conclusions: In women with singleton gestation, without prior spontaneous PTB but with TVU CL <25mm in the second trimester, cerclage does not prevent preterm delivery or improve neonatal outcome. Cerclage, in singletons without prior spontaneous PTB, seems to be possible efficacious at lower CLs, such as <10mm, and when tocolytics or antibiotics were used as additional therapy, requiring further studies in these subgroups. Given the low quality of evidence, further well-designed RCT is necessary to confirm the findings of this study. This article is protected by copyright. All rights reserved. Accepted Article INTRODUCTION 1 Preterm birth (PTB) is a major cause of perinatal morbidity and mortality. Worldwide, about 15 million babies are born too soon every year, causing 1.1 million deaths, as well as short- and 2,3 long-term disability in countless survivors. 4,5 Few prognostic tests are available to predict PTB. A short transvaginal ultrasound cervical length (TVU CL) has been shown to be a good predictor of spontaneous PTB, in both singletons 4-8 and twins. 9-24 Different strategies have been adopted for prevention of PTB, including progesterone, cerclage, cervical pessary, as well as lifestyle modification, such as smoking cessation, diet, aerobic exercise, and nutritional supplements. The evidence supports the use of vaginal 9 progesterone in singleton pregnancies with short cervix, while cervical cerclage seems to be beneficial only in the subgroup of singleton gestations with both prior spontaneous PTB and 10 11 24 TVU CL ≤25mm, and not in singletons without prior PTB, nor in multiple gestations. Cervical pessary is relatively non-invasive, easy to use, does not require anesthesia, can be used in an outpatient clinic setting, and it is easily removed when necessary. However, data published are contradictory, and meta-analyses have shown no efficacy in prevention of PTB in both 13 23 singleton, and multiple pregnancies. Interestingly, only 235 women have been included in randomized controlled trials (RCTs) on 11 cerclage for TVU CL <25mm for singleton pregnancies without prior spontaneous PTB, while 10 504 for singleton pregnancies with prior spontaneous PTB. Indeed, Berghella et al. in an individual patient data (IPD) meta-analysis of four RCTs found a non-significant 16% reduction in PTB <35 weeks in singletons without prior spontaneous PTB but with a TVU CL <25mm who 11 were randomized to cerclage compared to no cerclage. This article is protected by copyright. All rights reserved. Accepted Article Recently, Otsuki et al. reported data from a new RCT on cerclage in women with short TVU CL, 24 including also singleton gestations without prior spontaneous PTB. They showed that for women with TVU CL <25 mm between 16 and 26 weeks of gestation, cerclage might be 25 considered to reduce the occurrence of threatened preterm labor. 11 Our objective was to update and expand the previous IPD meta-analysis, and to quantify the efficacy of cervical cerclage in preventing PTB and perinatal morbidity and mortality in asymptomatic singleton pregnancies with a mid-trimester sonographic short TVU CL and without prior spontaneous PTB. METHODS Search strategy The review protocol was established by two investigators (VB, GS) prior to commencement and was registered with the PROSPERO International Prospective Register of Systematic Reviews (registration No. CRD42016048269). MEDLINE, ClinicalTrials.gov, the PROSPERO International Prospective Register of Systematic Reviews, and the Cochrane Central Register of Controlled Trials were searched for the following terms: ‘cerclage,’ ‘cervical cerclage,’ ‘salvage cerclage,’ ‘rescue cerclage,’ ‘emergency cerclage,’ ‘ultrasound-indicated cerclage,’ ‘short cervix,’ ‘cervical length,’ ‘ultrasound,’ and ‘randomized trial,’ from inception of each database until February 2017. All manuscripts were reviewed for pertinent references. No language restrictions were applied. This article is protected by copyright. All rights reserved. Accepted Article Study Selection We included all RCTs of asymptomatic singleton pregnancies without prior spontaneous PTB screened with TVU CL, found to have a midtrimester CL <25mm, and then randomized to management with either cerclage (i.e. intervention group) or no cerclage (i.e. control group).. Quasi-randomized trials (i.e. trials in which allocation was done on the basis of a pseudo-random sequence, e.g. odd/even hospital number or date of birth, alternation), studies on multiple pregnancies and studies on symptomatic women were excluded. Trials evaluating history- 27 indicated cerclage (placed for the sole indication of prior spontaneous PTB), or ultrasound- 10,11 indicated (placed for a short TVU CL) in women with also a prior spontaneous PTB, or physical-exam indicated cerclage (placed for second trimester cervical dilatation detected on 27 27 physical exam), as well as studies on technical aspects of cerclage, were also excluded. Therefore, eligible RCTs had to include women with singleton gestations, without prior spontaneous PTB, found to have upon TVU screening a short CL in the second trimester, who were randomized to cerclage versus no cerclage, and were followed for the primary outcome of PTB. Data extraction and risk of bias assessment The risk of bias in each included study was assessed by using the criteria outlined in the 29 Cochrane Handbook for Systematic Reviews of Interventions. Seven domains related to risk of bias were assessed in each included trial since there is evidence that these issues are associated with biased estimates of treatment effect: 1) random sequence generation; 2) allocation concealment; 3) blinding of participants and personnel; 4) blinding of outcome assessment; 5) This article is protected by copyright. All rights reserved. Accepted Article incomplete outcome data; 6) selective reporting; and 7) other bias. Review authors’ judgments 29 were categorized as “low risk,” “high risk” or “unclear risk” of bias. We contacted corresponding authors of all the included RCTs to request access to the data and perform a meta-analysis of IPD. Authors were asked to supply anonymized data (without identifiers) about patient baseline characteristics, experimental intervention, control intervention, co-interventions, and pre-specified outcome measures for every randomly assigned subject and were invited to become part of the collaborative group with joint authorship of the final publication. Data provided by the investigators were merged into a master database specifically constructed for the review. Data were checked for missing information, errors, and inconsistencies by cross-referencing the publications of the original trials. Quality and integrity of the randomization processes were assessed by reviewing the chronological randomization sequence and pattern of assignment, as well as the balance of baseline characteristics across treatment groups. Inconsistencies or missing data were discussed with the authors and corrections were made when deemed necessary. Quality of evidence For this review, the quality of the evidence was assessed using the GRADE approach in order to assess the quality of the body of evidence relating to the primary and secondary outcomes. GRADEpro Guideline Development Tool was used to import data from Review Manager 5.3 (Copenhagen: The Nordic Cochrane Centre, Cochrane Collaboration, 2014) in order to create ’Summary of findings’ tables. A summary of the intervention effect and a measure of quality for each of the above outcomes was produced using the GRADE approach. The evidence can be downgraded from 'high quality' by one level for serious (or by two levels for very serious) This article is protected by copyright. All rights reserved. Accepted Article limitations, depending on assessments for risk of bias, indirectness of evidence, serious 29 inconsistency, imprecision of effect estimates or potential publication bias. Outcomes Primary and secondary outcomes were established a priori. The primary outcome was PTB <35 weeks. Secondary outcomes were: PTB <37, <34, <32, <28 and <24 weeks, mean gestational age at delivery in weeks, mean of latency in days (i.e. time from randomization to delivery), incidence of preterm premature rupture of membranes (PPROM), and neonatal outcomes including birth weight, low birth weight (LBW) (i.e. birth weight <2500 grams), very LBW (VLBW) (i.e. birth weight <1500 grams), respiratory distress syndrome (RDS), intraventricular hemorrhage (IVH) grade 3 or 4, sepsis, necrotizing enterocolitis (NEC), admission to neonatal intensive care unit (NICU), mean of length of stay (LOS) in NICU in days, and neonatal death (i.e. death of a live-born baby within the first 28 days of life). We planned to assess the primary outcome (i.e. PTB <35 weeks) according to different TVU CL cutoffs (i.e. ≤20, ≤15, <10, <5 mm), according to race, according to type of cerclage, and according to additional therapy used. Data analysis The data analysis was completed independently by two authors (VB, GS) using Review Manager 5.3 (Copenhagen: The Nordic Cochrane Centre, Cochrane Collaboration, 2014). The completed analyses were then compared, and any difference was resolved with review of the entire data and independent analysis. IPD were analyzed using the so-called two-stage approach. In this approach, the IPD are first analyzed separately in each study to produce study-specific estimates of relative treatment effect. A combined estimate is then obtained in the second step by This article is protected by copyright. All rights reserved. Accepted Article calculating a weighted average (inverse error-variance-based) of the individual estimates using methods analogous to meta-analyses of aggregate data. Between-study heterogeneity was explored using the I-squared, which represents the percentage of between-study variation that is due to heterogeneity rather than chance. Meta-analysis was performed using the random effects model of DerSimonian and Laird, to produce summary treatment effects in terms of either a RR or a mean difference (MD) with 95% confidence interval (CI). 30 Potential publication biases were assessed statistically by using Begg’s and Egger’s tests. Two- tailed p-value<0.05 was considered statistically significant. Characteristics of the included women obtained in the merged database were analyzed using Statistical Package for Social Sciences (SPSS) v. 19.0 (IBM Inc., Armonk, NY, USA). Data are shown as means ± standard deviation (SD), or as number (percentage). Univariate comparisons of dichotomous data were performed with the use of the chi-square or Fisher exact test. Comparisons between groups were performed with the use of the T-test to test group means with SD. Two sided p-values <0.05 were considered statistically significant. All review stages were conducted independently by two reviewers (VB, GS). The two authors independently assessed electronic search, eligibility of the studies, inclusion criteria, risk of bias, data extraction and data analysis. Disagreements were resolved by discussion with a third reviewer (AC). The meta-analysis was reported following the Preferred Reporting Item for Systematic Reviews 31 and Meta-analyses (PRISMA) statement. This article is protected by copyright. All rights reserved. Accepted Article RESULTS Study selection and population characteristics Figure 1 shows the flow diagram (PRISMA template) of information derived from reviewing of 25,32-35 potentially relevant articles. Five RCTs, including 419 asymptomatic singleton gestations with short mid-trimester TVU CL and without prior spontaneous PTB, were included in the meta-analysis. The overall risk of bias of the included trials was low (Figure 2). All studies had a low risk of bias in “random sequence generation”, “incomplete outcome data”, and “selective reporting.” Adequate methods for allocation of women were used. All randomized women were included in an intention-to-treat analysis. Given the intervention, double-blinding was not feasible and all trials were judged as high risk of bias in performance bias. Publication bias, assessed using Begg’s and Egger’s tests, showed no significant bias (P=0.39 and P=0.51, respectively). The statistical heterogeneity between the studies was low with no 2 inconsistency (I =0%) in the primary and most of the secondary outcomes. Table 1 shows the characteristics of the included trials. All the included trials enrolled also women with prior spontaneous PTB which were excluded from the IPD. Multiple gestations were also excluded. Therefore, the IPD was used in order to include only singleton gestations without prior spontaneous PTB. Out of the 419 women analyzed, 224 (53.5%) were included in the cerclage group (i.e. study group), and 195 (46.5%) in the control group. Only singleton gestations without prior spontaneous PTB and with short cervix <25mm were analyzed. Most of the included studies (4 25,32,33,35 34 out of the 5), defined short cervix as TVU CL <25 mm; while To et al. defined as TVU 32,33,35 34 CL ≤15 mm. Three trials used only McDonald cerclage, To et al. only Shirodkar, while This article is protected by copyright. All rights reserved. Accepted Article