Interventions for women with mid‐trimester short cervix: which ones work?

Two of the most significant advances in obstetrics over the last 20 years are (1) the discovery that a short mid-trimester cervix on transvaginal ultrasound is the most powerful predictor of preterm birth (PTB), for both high- and low-risk women1-3, thus enabling universal cervical length screening4; and (2) the introduction of potentially promising treatments for the prevention of PTB in women with a short cervix. In women with a short mid-trimester cervix, randomized controlled trials (RCTs) have used progestogens, cervical cerclage and, lately, cervical pessary as PTB prevention strategies. The aims of this Editorial are to review Level-I evidence regarding the effectiveness of interventions in women identified as having a short cervix on mid-trimester ultrasound, discuss the challenges remaining and outline areas for future research. The evidence is clear with respect to the use of progestogens in asymptomatic women with short mid-trimester cervix: synthetic progestogens are not effective, whereas natural vaginal progesterone seems to be quite effective, in reducing preterm births. Two trials that randomized women to weekly intramuscular injections of 17α-hydroxyprogesterone caproate did not show any benefit, regardless of the participant's risk5, 6. In contrast to synthetic progestogens, there is evidence that natural progesterone, in the form of vaginal capsules or gel, for PTB prevention in women with a singleton pregnancy and short cervix appears to be associated with substantial reduction not only in PTB ≤ 34 weeks but also in neonatal morbidity. The first RCT to use vaginal progesterone (200 mg capsule nightly) in the general population of women with short cervical length (≤ 15 mm) was reported by Fonseca et al.7. They found that vaginal progesterone significantly reduced the rate of spontaneous PTB < 34 weeks (19.2% vs 34.4%; relative risk (RR), 0.56 (95% CI, 0.36–0.86)) but the reduction in neonatal morbidity was not significant (8.1% vs 13.8%; RR, 0.59 (95% CI, 0.26–1.25)). The historical significance of this RCT is that it paved the way for screening of the general population, among whom the majority of PTBs occur. The second trial to use vaginal progesterone (90 mg gel daily) in women with short cervical length (10–20 mm) was reported by Hassan et al.8. The results were even more promising than those of Fonseca's7 because, in addition to the reduction in PTB, it was the first trial to show significant improvement in neonatal outcomes in women with a short cervix treated with vaginal progesterone. Treated patients, as compared with placebo patients, had a 45% reduction in the primary outcome of PTB < 33 weeks (8.9% vs 16.1%; RR, 0.55 (95% CI, 0.33–0.92)), 50% reduction in PTB < 28 weeks (5.1% vs 10.3%; RR, 0.50 (95% CI, 0.25–0.97)) and 38% reduction in PTB < 35 weeks (14.5% vs 23.3%; RR, 0.62 (95% CI, 0.42–0.92))8. In the treated group, there was significant improvement in neonatal outcomes including reduced rates of respiratory distress syndrome (3.0% vs 7.6%; RR, 0.39 (95% CI, 0.17–0.92)), any neonatal morbidity or mortality event (7.7% vs 13.5%; RR, 0.57 (95% CI, 0.33–0.99)) and birth weight < 1500 g (6.4% vs 13.6%; RR, 0.47 (95% CI, 0.26–0.85))8. The RCTs by Fonseca et al.7 and Hassan et al.8 were unique because they focused on women with a short mid-trimester cervix. Three additional trials focused on high-risk patients, using varying concentrations of vaginal progesterone (90–100 mg daily) and cervical length cut-offs ranging from 25 to 32 mm9-11. These three trials focused primarily on high-risk patients, based on historical factors such as history of spontaneous PTB, uterine malformation, twin pregnancy, positive cervicovaginal fibronectin or a short cervix. The largest and most recent of the three trials (OPPTIMUM)11 involved a very heterogeneous population from 65 UK National Health Service hospitals and one Swedish hospital. Candidates for inclusion were women with singleton pregnancy who had previous spontaneous PTB ≤ 34 weeks, cervical length ≤ 25 mm, or positive fetal fibronectin combined with other risk factors (i.e. history of PTB, second-trimester pregnancy loss, preterm prelabor rupture of membranes or history of a cervical procedure)11. In this heterogeneous population, with a compliance rate of only 69%, the investigators could not demonstrate a benefit of vaginal progesterone in reducing the primary obstetric outcome (fetal death or PTB < 34 weeks), neonatal composite outcome (death, brain injury or bronchopulmonary dysplasia) or childhood outcome at 22–26 months of age (Bayley-III cognitive composite score). However, among patients with a short cervix (≤ 25 mm) there was a non-significant reduction in PTB ≤ 34 weeks (24.8% vs 32.2%; RR, 0.77 (95% CI, 0.52–1.14))11. The number of women with a short cervix in the OPPTIMUM trial11 was comparable to that of Fonseca et al.7, but was much less than that of Hassan et al.8. To clarify further the evidence for the benefit of vaginal progesterone, Romero et al. performed two meta-analyses, one before12 and one after13 the OPPTIMUM trial, thus including its data, with similar positive results. Using the same primary outcome as in the OPPTIMUM trial, they found that vaginal progesterone significantly decreased the risk of PTB ≤ 34 weeks or fetal death compared with placebo (18.1% vs 27.5%; RR, 0.66 (95% CI, 0.52–0.83))13. The study also found that, in singleton pregnancies with a short cervix (≤ 25 mm), vaginal progesterone reduced the risks of PTB occurring at < 28 to < 36 weeks, respiratory distress syndrome, composite neonatal morbidity and mortality, birth weight < 1500 g and admission to the neonatal intensive care unit13. Importantly, no differences in neurodevelopmental outcomes at 2 years of age were found between the progesterone and placebo groups. Romero et al. concluded that clinicians should continue to offer mid-trimester cervical length screening at 18–24 weeks in women with a singleton pregnancy and offer vaginal progesterone to those with cervical length ≤ 25 mm13. The efficacy of vaginal progesterone in women with twins and a short mid-trimester cervix is reported in this issue by Romero et al. who performed an individual patient data meta-analysis using data from six randomized controlled trials; it was found that administration of vaginal progesterone to this group of women significantly decreased the risk of PTB < 34 weeks by approximately 30% and, in addition, it decreased neonatal mortality and composite neonatal morbidity/mortality by approximately 50%14. In our opinion, considering the data altogether, it appears that there is strong evidence for the benefit of vaginal progesterone in women with a short mid-trimester cervix. However, tantalizing questions remain regarding the best group of candidates, as defined by the degree of cervical shortening, the optimal dose of progesterone and possible short- and long-term side effects. The two trials that focused on treating only women with short cervix used different cervical length cut-offs of < 15 mm7 and 10–20 mm8. The other three trials that included women with a short cervix used cervical length cut-offs ranging from 25 to 32 mm9-11, whereas traditionally, a short cervix is defined as < 25 mm15. The individual and aggregate meta-analyses demonstrated that vaginal progesterone is effective in women with a cervical length ≤ 25 mm13, 14. However, it should be noted that all trials used a single static cervical length cut-off for intervention. From serial observations in clinical practice, we have noticed that some patients with a short cervix will develop progressive cervical shortening and others will remain stable. Unfortunately, both of these groups of patients have been grouped together in the aforementioned trials so that we have no information regarding the presence or absence of progressive cervical shortening prior to randomization. It is possible that the beneficial effects of progesterone could be attributed to women with progressive cervical shortening. Hopefully, future studies will determine the optimal group of candidates on which to focus treatment. Although the trials used different doses of progesterone, ranging from 90 to 200 mg7-11, an individual patient data meta-analysis12 showed that there was no difference in the efficacy of vaginal progesterone according to the dose administered. In the meta-analysis of twin gestations with a short cervix, a dose of 400 mg daily appeared to be more effective, as compared to 100 mg or 200 mg, but this should be interpreted with caution given the small number of patients in the subgroup analyses14. Also, a careful prospective evaluation for not only short- but also long-term side effects beyond 2 years of age is necessary. So far, trials have shown that there is better short-term perinatal outcome, and no difference in long-term neurodevelopment at 2 years of age, in pregnancies with a short cervix treated with vaginal progesterone, as compared with controls. As intra-amniotic inflammation is observed in a significant proportion of women with a short cervix (approximately 10%)16, it is conceivable that some pregnancies with short cervix and intra-amniotic inflammation may be prolonged by the use of progesterone and that the long-term childhood morbidity in such cases may be increased. We hope that future studies will be able to address this important question. The first trial that focused on high-risk pregnant women with a short cervix (< 25 mm) used cervical cerclage as the mode of treatment17. Despite the initial encouraging results of this RCT, which had a very small number of subjects17, three subsequent RCTs involving many more patients produced negative results18-20. A meta-analysis of the aforementioned four trials performed 40 subgroup analyses, taking into consideration not only gestational age at delivery but also factors such as singletons, twins, history of prior PTB, second-trimester fetal loss, history of cone biopsy, number of prior dilations and evacuations, cervical funneling, gestational age at cervical shortening and degree of shortening21. Of the 40 subgroup comparisons, six were statistically significant with five of the six comparisons having upper limit estimates of the 95% confidence intervals of 0.92–0.99. The authors concluded that in the subgroup analyses, cerclage was associated with a significant reduction in PTB among women, who in addition to a short cervix, had a prior PTB (RR, 0.61 (95% CI, 0.40–0.92)) or second-trimester pregnancy loss (RR, 0.57 (95% CI, 0.33–0.99)), and recommended a well-powered RCT in this group of patients21. Four years later, Owen et al.22 reported the results of a multicenter RCT using cerclage for PTB prevention in women with short mid-trimester cervix (< 25 mm) and a history of spontaneous PTB between 17 + 0 and 33 + 6 weeks' gestation. The authors found that cerclage did not prevent PTB < 35 weeks (primary outcome) but was associated with its reduction if cervical length was < 15 mm (secondary outcome)22. In other words, the results of the trial were negative. At face value, one can conclude that cerclage does not work for women with prior spontaneous PTB who have a cervical length of 15–24 mm. If one were to accept the results of the planned secondary analysis in this trial22 then cerclage remains effective in reducing PTB among women with a cervical length < 15 mm. However, as women who had visible membranes on speculum examination were excluded, it is highly likely that many patients with very short cervix (i.e. < 5 mm) may have been deemed ineligible, and hence excluded from the trial. Thus, if cerclage works, its benefits are confined to a very small subset of women with a history of spontaneous PTB or second-trimester pregnancy loss who have a mid-trimester cervical length of 5–14 mm. Therefore, from the point of view of the general population, the role of cerclage as a prevention strategy for PTB remains very limited, at best. It is worth commenting on the sequence of events that may have led to the trial by Owen et al.22. This trial was carried out as a result of the findings of the multiple subgroup analyses of a prior meta-analysis, rather than the traditional ‘other way around’ of pooling small RCTs to draw a conclusion from the evidence. In our view, from a statistical perspective, when undertaking a meta-analysis, one should ensure that the meta-analysis preserves (or reduces) both Type I and Type II errors. An accumulation of studies in a meta-analysis will almost always decrease the Type II error, and hence increase the power to detect treatment efficacy, but attention to Type I error is often pushed to the wayside. Performing a meta-analysis with several underpowered subgroup analyses will inflate the Type I error. While such subgroup analyses may compromise the statistical power, they will just as likely compromise Type I error too. The third modality that has been attempted recently for preventing PTB in women with a short cervix is cervical pessary. In our view, the evidence regarding the use of pessaries in women with short mid-trimester cervix appears inconclusive for both singleton and twin gestations. For singleton gestations with a short mid-trimester cervix, we evaluated three trials that used cervical pessary. The first reported trial by Goya et al., which used the Arabin pessary in asymptomatic singleton gestations with a short cervix (≤ 25 mm), reported a significant reduction in PTB < 34 weeks in those with pessary compared with those with expectant management (6% vs 27%; OR, 0.18 (95% CI, 0.08–0.37))23. The authors hypothesized that the mechanism by which the pessary works is by supporting ‘the immunologic barrier between the chorioamnion-extraovular space and the microbiological flora as cerclage has been postulated to do’23. This hypothesis, if true, would argue against the potential benefit of the pessary given the poor record of cerclage in women with a short cervix. A subsequent smaller trial by Hui et al.24, which used cervical length cut-off < 25 mm, and a considerably larger trial by Nicolaides et al.25, which used cervical length cut-off ≤ 15 mm, reported negative results (PTB < 34 weeks: 9.4% vs 5.5%; OR, 1.80 (95% CI, 0.40–7.96) by Hui et al.24 and 12.0% vs 10.8%; OR, 1.12 (95% CI, 0.75–1.69) by Nicolaides et al.25). However, it is worth noting that the rates of PTB < 34 weeks in the control groups of Goya et al.23 and Nicolaides et al.25 were significantly different (27.0% in Goya et al.23 vs 10.8% in Nicolaides et al.25; RR, 2.49 (95% CI, 1.75–3.54)). Given the differences in cervical length used in the two trials (≤ 25 vs ≤ 15 mm), one would expect the control group in Goya et al. to have lower PTB rate; however, the opposite occurred. This observation suggests that there was considerable heterogeneity or dissimilarity between these two populations. Interestingly, the same pattern of differences in the control groups was also noted in the cervical pessary trials of these two investigators in women with twin gestations and a short cervix (discussed below). The first trial using cervical pessary for prevention of PTB in unselected asymptomatic women with multiple pregnancy was reported by Liem et al.26. In this trial, the primary outcome was composite poor perinatal outcome (stillbirth, periventricular leukomalacia, severe respiratory distress syndrome, bronchopulmonary dysplasia, intraventricular hemorrhage, necrotizing enterocolitis, neonatal sepsis and neonatal death); PTB was one of the secondary outcomes. Overall, the pessary did not improve composite perinatal outcome or gestational age at birth; however, a secondary analysis showed that, among women with a mid-trimester cervical length < 38 mm, who received the pessary, poor perinatal outcome was less frequent and PTB rates < 28 and < 32 weeks were lower when compared with the control group. These findings were an impetus for two subsequent trials in twin gestation27, 28. The largest trial was reported by Nicolaides et al. in unselected twin gestations (n = 1180; 590 in each group) and had negative results as there was no difference between the pessary and control groups in terms of PTB or any other adverse perinatal outcome27. In addition, a post-hoc subgroup analysis of 214 women with a short cervix (≤ 25 mm) did not show any benefit among those who received the pessary (n = 106) and those who did not (n = 108); the rates of spontaneous PTB < 34 weeks were 31.1% and 25.9%, respectively (RR, 1.20 (95% CI, 0.78–1.83))27. Goya et al.28 reported a smaller trial in twin gestations focusing specifically on those with a short cervix (≤ 25 mm) and found significant benefit of the pessary in terms of less frequent spontaneous PTB < 34 weeks (16.2% vs 39.4%; RR, 0.41 (95% CI, 0.22–0.76)) for those who received the pessary (n = 68) compared with those who did not (n = 66); however, there was no difference in neonatal morbidity or mortality between the two groups. The discrepant results between the latter two trials were attributed by Goya28 to a high rate of pessary removal < 34 weeks and the lack of a training program for pessary insertion and follow-up in the Nicolaides trial. However, in our view, it appears that the populations were, once again, highly heterogeneous given the discrepant rates of PTB < 34 weeks between the control groups of these two trials (39.4% in Goya et al.28 vs 25.9% in Nicolaides et al.27). For all the aforementioned reasons, it is our belief that the evidence regarding the use of pessary in women with a short cervix is inconclusive, in both singleton and twin gestations, due to differences in the studied populations and perhaps in training in pessary insertion. Altogether, the evidence suggests that there is progress in preventing PTB and that the most effective intervention is in women with singleton pregnancy and a short mid-trimester cervix, regardless of their risk status, by vaginal progesterone. However, we still face some serious challenges relating to (1) the accuracy of cervical length measurements in the setting of universal screening, (2) the appropriateness of universal screening in with PTB and the very small in reducing PTB rate by on the between a short cervix and PTB are in that approximately 30% of women with a singleton pregnancy and a cervical length ≤ 25 mm in some in the mid-trimester will preterm and that these patients for of women with a spontaneous However, most studies on cervical length screening from It remains to be if the aforementioned rates can be by universal screening of all populations. In a study in which were to their ultrasound of the cervix for it was found that of did not The of in the clinical setting may be even considering that of this study most likely their may work but only be for use by of which there are not and are Given the it seems likely that the effectiveness of universal cervical length screening may be in issue to universal screening is that the of a short cervix (≤ 25 mm) seems to considerably between populations. For in a mid-trimester cervical length ≤ 25 mm is only in of the is in The a mid-trimester cervical length ≤ mm, to the 20 patients had to be in to patients with a cervix (≤ The randomized trial in this population of low-risk singleton using this cervical length as a and vaginal progesterone (200 mg capsule daily) vs placebo had negative results reduction in PTB < 32 weeks and < 34 In with a very of a short cervix or PTB, for which universal screening is not consideration may be given to use cervical length only in patients with preterm to the for or For asymptomatic patients, may be more In mid-trimester cervical length evaluation is to high-risk patients such as women with a previous PTB and screening of women based on previous cervical and using such a the number of cervical may be reduced by however, such a may the of women preterm by which may be in with PTB In contrast to the aforementioned in the there is a high rate of PTB (approximately in universal cervical length screening has been found to be as compared with screening or no screening and vaginal progesterone administration to women with a short mid-trimester cervix in the regardless of risk or prior has been reported to prevent one of PTB < 34 weeks for women and one of neonatal morbidity or mortality for women which into an reduction of approximately PTBs < 34 weeks and 17 cases of neonatal morbidity or Although vaginal progesterone seems to the rate of PTB < 34 weeks in women with a singleton pregnancy and a short cervix, the on the PTB rate is An analysis of and potential with PTB interventions in with very high concluded that five combined interventions may decrease the rate of PTB by no more than The possible of our interventions in reducing PTB, based on from is in We did not cerclage as one of the prevention because women with a short cervix and prior history of PTB or second-trimester pregnancy loss may be treated with vaginal progesterone. Unfortunately, interventions can prevent only approximately of the number of PTBs Therefore, considerable progress in the prevention of PTB in women with a short cervix by vaginal progesterone, serious challenges As shown in in to have a in reducing PTB, future should focus on the four groups of (1) multiple only a small proportion will to vaginal (2) those with those with short mid-trimester cervical length who not to vaginal and those with mid-trimester cervical length who develop spontaneous We have a of because, if we are to have a on reducing PTB, future studies should focus on all four groups. will to in that our or of should into consideration the of PTB, we have data to that the may the or of our In our opinion, taking into consideration the of each individual preterm we will only and In the it to offer universal cervical length screening, with or the use of transvaginal ultrasound, and offer treatment with vaginal progesterone in women with singleton pregnancy and short mid-trimester cervix in with high PTB such as the For with PTB such as a more of screening and vaginal progesterone to those with a singleton pregnancy and a short cervix seems