Performance of the cervical shortening for prediction of spontaneous preterm birth in uncomplicated twins

Abstract

Objective There is a lack of consensus about the management of twins with significant cervical length (CL) shortening, especially if CL is above 25 mm. Therefore, it is important to define “abnormal” CL change over time, and to compare the performance of different strategies. The aim of this study was twofold, to describe the performance of the cervical shortening and that of an integrated strategy that includes both the cervical shortening and a fixed CL cutoff <25 mm in any measurement as predictor of spontaneous PTB (sPTB) < 34 weeks in uncomplicated twin pregnancies. Material and methods Retrospective cohort study of twins followed in our Twins Clinic at Hospital Italiano de Buenos Aires from 2013 to 2017. Inclusion criteria were dichorionic or monochorionic diamniotic twins with CL measurement between 18 and 33 + 6 weeks with available data of the delivery. Exclusion criteria included any of the following complications: iatrogenic preterm delivery <34 weeks, cerclage, fetal growth restriction, fetal death, structural anomalies, polyhydramnios, twin–twin transfusion syndrome, selective fetal growth restriction, twin anemia–polycythemia sequence, and twin reversed arterial perfusion sequence. Spontaneous preterm birth was defined as spontaneous delivery <34 weeks. Cervical shortening was analyzed in the following periods: 20–24 weeks, 20–28 weeks, 24–28 weeks, 24–32 weeks and 28–32 weeks. Cervical changes were analyzed as velocity of shortening over time (mm/week) and as the ratio of shortening over time (%/week). ROC curves for each period were constructed and two different cutoffs were used to classify changes of the CL as positive or negative screening: a) the shortening of CL associated to the highest value of the Youden Index and b) fixing a 10% false positive rate (FPR). For the second objective, we analyzed an integrated strategy considering a fixed cutoff of 25 mm at any GA and/or a significant shortening. The screening was considered positive if any CL measurement was <25 mm at any GA or there was a shortening of the CL ≥ the cutoff obtained for each period. We report sensitivity, specificity, positive and negative predictive value, positive and negative likelihood ratio and area under the ROC curve. Results We included 378 patients and 1417 measurements, 284 (75%) dichorionic and 94 (25%) monochorionic. Between 20 and 28 weeks, with a change in CL cutoff = 1.6 mm/week or 4.1%/week the detection rate was 54.2% (32.8–74.4%) and the specificity 80.5% (75.1–85.1%) and 83.5% (78.5–87.8%) respectively. In the integrated strategy, the detection rate was 65.7% (47.8–80.9%) and the specificity 69 (63.7–74). All the ROC curves of the periods studied showed an AUC < 0.7. In the group of patients that delivered preterm the initial mean CL was shorter than in the term group, 39 (±12) mm vs. 43 (± 7.7) mm (p = .02) and the most important change in CL was at 20–24 weeks both in the velocity and in the ratio of shortening over time. Conversely, patients that delivered at term showed a higher change in CL in the third trimester. Conclusion The performance of all the strategies analyzed as a predictor of sPTB <34 weeks was moderate. The period 20–28 weeks detected half of the patients at risk with a FPR around 10–20% and the integrated strategy increased the sensitivity up to a detection of two thirds of the patients at risk but with a FPR of ∼30%. Future analyses need to explore other strategies to improve the performance and to really identify the patients at higher risk.