Value of post-cerclage transvaginal ultrasound in predicting preterm birth at <28 weeks in twin pregnancy with ultrasound-indicated cerclage

Abstract Objective To evaluate the value of transvaginal ultrasound parameters before and after cerclage in twins in predicting spontaneous preterm birth (sPTB) before 28+0 weeks. Methods We retrospectively studied the medical records of twin-pregnant women who underwent ultrasound-indicated cerclage between January 2016 and February 2022 at our hospital. Recorded transvaginal ultrasound images before and after cerclage were reassessed for cervical length (CL), uterocervical angle (UCA), funneling, and sludge. Multivariate logistic and Cox regression analyses were performed to identify the independent risk factors associated with sPTB before 28 weeks. Results A total of 69 women were included. Among them, 17 women (24.64%) delivered before 28 weeks of age. Regression analysis revealed a significant association of post-cerclage CL, UCA, white blood cell (WBC) count, and gestational age (GA) at cerclage with sPTB before 28 weeks. The area under the curve of these predictors was 0.938 (95% confidence interval, 0.882–0.994; p < .001), with a sensitivity of 88.2%, specificity of 92.3%, positive predictive value of 78.9%, and negative predictive value of 96.0%. Cox analysis showed that post-cerclage UCA was an independent risk factor affecting the cerclage-to-delivery interval (hazard ratios, 1.026; 95% confidence interval (CI), 1.004–1.048; p = .019). Conclusions The combination of post-cerclage CL, UCA, WBC count, and GA at cerclage showed good performance in predicting sPTB at <28 weeks in twin pregnancy. Post-cerclage UCA is also associated with pregnancy latency. We found that post-cerclage cervical ultrasound may be useful to predict preterm birth before 28 weeks in twins who undergo ultrasound-indicated cerclage.


Introduction
Preterm birth (PTB) remains the leading cause of neonatal morbidity and mortality despite continuous progress in obstetrics and neonatology [1].Multiple pregnancies are associated with a higher risk of spontaneous preterm birth (sPTB), maternal and perinatal morbidity and mortality, and worse neonatal outcomes than single pregnancies [2].According to the existing literature, twin pregnancies represent 3% of all live births, while they account for 17-20% of all PTBs, with up to half of them delivering before 37 weeks of gestation [3,4].
However, the accurate identification of women who are at a higher risk of early sPTB remains a great challenge [5].This is mostly because of the complex causes and phenotypes of sPTB.In multiple pregnancies, it seems that the pathophysiological processes of overdistention of the uterus and cervical insufficiency are the most common etiology of sPTB [6].Unlike singleton gestations, cervical cerclage may not be shown to benefit twins, especially in twins with ultrasound-detected short cervix (<25 mm) [7].Some studies have shown a relative risk of 2.15 for PTB in twin pregnancies with ultrasound-indicated (cervical length (CL) <25 mm) cerclage [8].Therefore, early prediction can help clinicians in risk stratification and antenatal assessment in twins with an elevated risk of sPTB after ultrasound-indicated cerclage.A transvaginal CL less than 25 mm is the most commonly used predictive tool for sPTB in multiple pregnancies and singletons [9].However, it alone may only detect one-third of sPTB cases [10].Recently, novel screening strategies for predicting sPTB have been proposed.Adjunctive sonographic measurements, such as the uterocervical angle (UCA), funneling, and presence of amniotic fluid sludge, have been considered promising approaches to estimate the risk of sPTB in singleton pregnancies [11][12][13].These sonographic parameters have also been evaluated in twins, but with limited success [14].Thus, we aimed to evaluate the value of transvaginal ultrasound parameters before and after cerclage in twins to predict sPTB before 28 þ0 weeks.

Methods
We conducted a retrospective cohort study involving twin pregnancies that underwent ultrasound-indicated cerclage at 14 þ0 -26 þ6 weeks of gestation in Women's Hospital, Zhejiang University, School of Medicine (a tertiary hospital) between January 2016 and February 2022.Exclusion criteria were as follows: patients who underwent history indicated cerclage or exam indicated cerclage, genetic or major fetal anomalies, elective termination of pregnancy, and iatrogenic PTB (twin-twin transfusion syndrome, placental previa, severe preeclampsia, or fetal distress).Women were excluded if they did not undergo cervical ultrasound before and after cerclage or if they did not deliver at our hospital.This study was approved by the ethical committee of Zhejiang University (IRB-20200160-R).This study was exempt from informed consent as it included no personal information.
At this tertiary hospital, women with multiple pregnancies would receive routinely US cervical check every two weeks from 14 to 16 weeks of gestation to 28 weeks.Short cervix was defined as CL less than 25 mm.Ultrasound-indicated cerclage was offered to women with CL <25 mm at 14-26 weeks of gestation.The patients opted to have a cerclage after counseling about risks, benefits, and limitations of their respective options.Other options, such as expectant management, medical therapy like progesterone were also offered to patients.Cerclage was offered when cervical dilation was present, but these patients were not included in the present study.Cerclage was performed in a McDonald manner with a single stitch (four treads of no. 10 silk tread) in the operating room under epidural anesthesia.Additional antepartum therapies such as indomethacin for 48 h, and antibiotic for 48 h such as cephalosporins if there was no allergy were prescribed to patients as standard perioperative management.Ritodrine hydrochloride was administered for uterine contractions.Nifedipine and progesterone were prescribed on discharge.Post-cerclage ultrasound was routinely performed after cerclage within one week transvaginally.
Ultrasound images stored in our hospital database system (ViewPoint) before and after cerclage were reviewed separately by two investigators.For women with multiple post-cerclage ultrasounds, measurements on the first one after cerclage were considered.Cervical length was defined as a linear measurement between the internal and external cervical os [15] (Figure 1(A)).The UCA was defined as the angle formed by one line that corresponds to the measurement of the CL and a second line placed delineating the lower anterior uterine segment and passing through the internal cervical os [16] (Figure 1(B)).Amniotic fluid sludge was defined as the presence of dense aggregates of particulate matter in the proximity of the internal cervical os [15,17] (Figure 1(C)).Cervical funneling was defined as protrusion of the amniotic membranes into the cervical canal, and measured along the lateral border of the funnel [15] (Figure 1(D)).
Other data were extracted from the electronic medical records as follows: age, pre-gestational body mass index (BMI), use of in vitro fertilization (IVF), gravidity, parity, history of hysteroscopy, history of polycystic ovary syndrome (PCOS), chorionicity, intervals from prior pregnancy, gestational age (GA) at the time of admission, transvaginal ultrasound parameters, results of microbiological and blood tests such as white blood cells (WBCs) and C-reactive proteins (CRPs), pregnancy complications such as gestational diabetes mellitus (GDM), preterm premature rupture of membranes (PPROMs), placental abruption, and hypertensive disorders in pregnancy.The primary outcome was the prediction of sPTB at <28 weeks of gestation by using transvaginal ultrasound parameters before and after cerclage in twins.

Statistical analysis
The Kolmogorov-Smirnov test was used to test the normal distribution of continuous variables.Continuous variables are presented as means (±standard deviation) or medians (range) if non-normal.
Categorical variables are presented as numbers and percentages.Fisher's exact test or Chi-square test for categorical variables and Wilcoxon-Mann-Whitney's test or Student's t-test for continuous variables were applied when appropriate.A p value of <.05 was considered statistically significant, and all tests were two-tailed.
Multivariate logistic regression analysis was performed to identify risk factors associated with sPTB at <28 weeks.Collinearity analysis among variables was performed to eliminate variables with a variance inflation factor (VIF) of >5.Results are presented as odds ratios (ORs) with 95% confidence intervals (CIs).Receiver operating characteristic (ROC) curves were developed to determine the area under the curve (AUC) for the prediction of sPTB at <28 weeks' gestation.Cox regression analyses were performed to evaluate the associations between the variables and pregnancy latency.Predictive strength was estimated using hazard ratios (HRs), 95% CIs, and p values.

Results
During the study period, 90 twin pregnancies with ultrasound-indicated cerclages were screened.Among them, six women did not undergo post-cerclage ultrasound assessment and seven women delivered in other hospitals.Six women with iatrogenic deliveries due to medical complications were excluded from analysis.Two patients were excluded because of missing ultrasound images.Ultimately, 69 twin pregnancies were included in the study.
The mean post-cerclage UCA was 115.84 ± 29.87 � .Hence, we divided the women into two groups according to the mean post-cerclage UCA (115 � ).Women with a post-cerclage UCA of �115 � had a significantly longer cerclage-to-delivery interval than those with a UCA of >115 � (log-rank test, p ¼ .001).Kaplan-Meier's curves are shown in Figure 3.

Discussion
In the present study, we focused on the value of transvaginal ultrasound parameters before and after ultrasound-indicated cerclage, along with other clinical variables, to predict sPTB before 28 weeks in twin pregnancies.We identified the following four risk factors for delivery before 28 weeks: post-cerclage CL, UCA, WBC count, and GA at cerclage.Among all ultrasound markers, only post-cerclage CL and UCA were significantly associated with sPTB before 28 weeks, whereas post-cerclage UCA itself was an independent risk factor affecting the cerclage-to-delivery interval.
Management for the prevention of sPTB in twin pregnancies remains controversial, especially the use of cervical cerclage [7, 18,19].Ultrasound-indicated cerclage is performed to pregnant women with a prior sPTB and CL �25 mm by transvaginal ultrasound scans [20].Major guidelines [21] recommend against cerclage in the setting of multiple gestations for CL less than 25 mm, since it may increase the adverse outcomes.However, many recent studies have shown benefits in the use of cerclage after identification of a short cervix, especially in CL less than 15 mm [7,8,22].Therefore, antenatal assessment for risk stratification may identify those that could benefit from interventions such as antenatal corticosteroids or transfer to a facility with appropriate neonatal care.In our study,  we identified post-cerclage CL, UCA, WBC count, and GA at cerclage as independent predictors of sPTB before 28 weeks.Although many studies have recognized the importance of UCA as a screening tool for sPTB [11], few studies have involved UCA in a predictive model.Benito Vielba et al. first developed a  predictive model including UCA for prematurity in twin pregnancies [23], but they did not collect laboratory data.Our model had excellent predictive performance with an AUC of 0.938, sensitivity of 88.2%, specificity of 92.3%, positive predictive value of 78.9%, and negative predictive value of 96.0%.Hence, our model could provide comprehensive and accurate information about the prognosis after cerclage.In recent years, the UCA has been proposed as a promising ultrasonographic marker to provide more accurate sPTB prediction in both singleton and twin pregnancies.Many retrospective studies have suggested that UCA performed better than CL in predicting sPTB at different GAs in singleton and twin pregnancies [11,23,24].Khamees et al. [25] conducted a prospective study to evaluate the predictive role of UCA and CL in PTB and demonstrated that UCA provided a higher diagnostic performance in high-risk patients than CL.This result is consistent with our findings.In our study, we compared pre-and postcerclage ultrasound parameters; only post-cerclage UCA and CL showed significance in predicting sPTB in twins in the multivariable regression analysis, while the AUC of UCA was far better than that of CL.Since our findings suggested an association between GA at cerclage and the risk of sPTB, we performed Cox regression analysis to determine the risk factors associated with pregnancy latency and found that post-cerclage UCA remained an independent predictor affecting cerclage-to-delivery intervals (HR, 1.026; 95%CI, 1.004-1.048;p ¼ .019).To our knowledge, this is the first study to demonstrate that UCA are related to pregnancy latency.The hypothesis is that the more obtuse the angle, the easier it is for the cervix to open or be funneled by the force of the uterus, while a narrower angle may support an anatomical geometry to exert less direct force on the cervix [26].
A CL less than 25 mm in twin pregnancies is a widely accepted predictor of sPTB [6,27].However, few studies have investigated the association between postcerclage CL and pregnancy outcomes.Song et al. [28] conducted a retrospective study including 52 patients with singleton pregnancies to evaluate the value of CL before and after cerclage and found that patients with a post-cerclage CL less than 25 mm might have a higher risk of sPTB before 32 weeks.Our findings also suggest that post-cerclage CL was associated with sPTB before 28 weeks in twin pregnancies.Battarbee et al. [29] assessed the value of the first transvaginal ultrasound parameters after cerclage, including the UCA, CL, width of the anterior and posterior cervix at the level of cerclage, stitch distance from the cervical canal, and cervical funneling and sludge, to estimate the risk of sPTB in singleton pregnancies.They indicated that besides CL, the location of the cerclage, curvature of the cervical canal, and presence of funneling and sludge might help identify women at high risk of sPTB.We also involved funneling and sludge in this study, but we did not observe this finding on pre-or post-cerclage ultrasound.There were only five patients who had sludge before cerclage and three patients after cerclage.For the matter of infection, we involved the blood marker -WBC as an infection marker.Fuchs et al. developed a scoring system for emergency cerclage based on obstetric history, cervical dilation, membranes bulging into the vagina and blood tests (WBC �13,600/mm 3 or CRP >15 mg/L) to predict preterm delivery before 32 weeks [30].In terms of funneling in twins, it may not appear as important as in singletons.Our study is in agreement to prior studies [31] that have shown that CL is a better predictor than funneling for the prediction of sPTB.
Our study has several strengths.First, in the present study, we considered pre-and post-cerclage ultrasound markers to predict sPTB in twin pregnancies who underwent ultrasound-indicated cerclage.Although the American College of Obstetricians and Gynecologists [20] and the Society of Obstetricians and Gynaecologists of Canada's Maternal Fetal Medicine committee [7] do not recommend transvaginal cervical surveillance after cerclage, our findings suggest potential benefits of transvaginal ultrasound after cerclage in predicting the risk of sPTB.Second, we included multiple variables previously associated with PTB in both our pre-and post-cerclage models for the prediction of birth before 28 weeks.Therefore, the variables included in the model were reliable.Third, this study utilized Cox regression analysis to assess the relationship between ultrasound markers and pregnancy latency from cerclage to delivery.
This study had some limitations.First, it was a retrospective study.Data were based on electronic medical records, and some information might be inexact and missing; for example, membranes bulging into the vagina were not available in the medical records.Second, the sample size was relatively small.Therefore, large randomized controlled trials are needed to confirm these findings.Third, we did not consider the effect of different medical therapies on pregnancy outcomes after cerclage.However, the study was conducted at a single center; therefore, our therapeutic strategies were based on the same clinical guidelines and protocols.

Conclusions
We reported a series of predictors combining postcerclage UCA, CL, WBC, and GA at cerclage with high predictive accuracy to predict sPTB before 28 þ0 weeks in twin gestations.We demonstrated that a post-cerclage ultrasound in twins with short cervix may identify those pregnancies with an increased risk for sPTB before 28 weeks.
Statistical analysis was performed using Statistical Packages of Social Sciences for Windows (version 22.0; SPSS, Chicago, IL).

Figure 2 .
Figure 2. Receiver operating characteristic curves of joint predictors (green line), uterocervical angle (blue line) and cervical length (yellow line) in prediction of spontaneous preterm birth at <28 weeks in twins.UCA: uterocervical angle; CL: cervical length; WBC: white blood cell; GA: gestational age.

Figure 3 .
Figure3.Kaplan-Meier's survival analysis of pregnancy latency according to UCA.Kaplan-Meier's curves were generated for gestational age at delivery by UCA �115 � .Comparison of whether UCA was larger than 115 � using log-rank test showed significant difference (p ¼ .001).UCA: uterocervical angle.

Table 1 .
Demographic and clinical characteristics and ultrasound parameters compared between sPTB before and after 28 weeks.

Table 2 .
Multivariate logistic regression analysis of independent predictors for delivery at <28 weeks in twin pregnancies with ultrasound-indicated cerclage.

Table 3 .
Univariate and multivariable COX regression analyses identified risk factors associated with pregnancy latency.