Can third‐trimester assessment of uterine scar in women with prior Cesarean section predict uterine rupture?

The rate of Cesarean delivery has increased worldwide in recent decades1. Concomitantly, the rates of vaginal birth after Cesarean (VBAC) have decreased steadily1. These current trends are attributed partially to concerns about the safety of trial of labor after Cesarean (TOLAC), including concerns regarding the risk of intrapartum uterine rupture2. This complication, estimated to occur in about 0.4–1% of women undergoing TOLAC, is associated with severe neonatal morbidity, including asphyxia, neurological sequelae and even death, as well as with severe maternal complications, such as genitourinary tract damage, hemorrhage and hysterectomy2-8. Uterine rupture following Cesarean delivery can generally be avoided by performing elective repeat Cesarean (ERC), but this is associated with a risk both of short-term maternal and neonatal complications9-13 and of severe maternal complications in future pregnancies, for example abnormal placentation (placenta previa and placenta accreta) and difficulties during surgery, which become more likely with increasing numbers of Cesarean sections14-17. Therefore, pregnant women with previous Cesarean section face a difficult choice for their next delivery: an ERC, with the possibility of short- and long-term maternal complications, or a TOLAC, with the risk of uterine rupture. As highlighted by the NIH (National Institutes of Health) statement in 20101, TOLAC is a safe option for most women with a prior Cesarean section. However, it is important to improve the selection of women with high chances of successful VBAC and low risk of uterine rupture. Certain conditions are associated with high rates of uterine rupture and constitute contraindications to TOLAC, such as previous classical or inverted T incision and previous uterine rupture18. Other factors, such as multiple previous Cesarean sections19, 20, type of prior hysterotomy closure21, short interdelivery interval22, fetal macrosomia23 and postpartum endometritis at previous Cesarean24, may be associated with uterine rupture, but none of these, neither alone nor in combination, has been demonstrated useful clinically with respect to the prediction of outcome25, 26. Here, we review the literature regarding third-trimester ultrasonographic assessment of uterine scar in the evaluation of the risk of uterine rupture following previous Cesarean section. Sonographic measurement of lower uterine segment (LUS) thickness near term has been shown to be correlated inversely with the risk at delivery of uterine scar defect, including uterine scar dehiscence at ERC and/or uterine rupture27-33. In 1988, Fukuda et al. suggested an association between a thin LUS and the risk of uterine scar dehiscence at ERC30. Using transabdominal sonography near term, they observed uterine scar dehiscence in 35% (5/14) of women with LUS thickness < 2.0 mm, compared with 0% (0/70) in women with LUS thickness > 3.0 mm. From a prospective cohort of 642 women with previous Cesarean section, Rozenberg et al.29 observed an inverse correlation between LUS thickness measured at 36–38 weeks' gestation with transabdominal sonography and the risk of uterine scar dehiscence/rupture. There were two (0.4%) cases of scar dehiscence and one (0.2%) case of uterine rupture among 455 women with LUS thickness > 3.5 mm, five (3.7%) cases of scar dehiscence and nine (6.6%) uterine ruptures among 136 women with a measurement between 2.6 and 3.5 mm, and three (5.9%) cases of scar dehiscence and five (9.8%) uterine ruptures among 51 women with a measurement between 1.6 and 2.5 mm (P < 0.01). They concluded that 3.5 mm appears to be the optimal cut-off value, allowing excellent sensitivity (88%) and negative predictive value (99.3%)29. The same authors later observed that introduction of LUS thickness measurement into clinical practice was associated with a downwards trend in the prevalence of uterine rupture in their center34. These promising results were limited by the low specificity (73%) and positive predictive value (12%) of the 3.5-mm cut-off value. A few studies which used complete uterine rupture as the primary outcome suggested a lower cut-off value27, 35, 36. In these studies, transabdominal assessment of the LUS was supplemented by transvaginal ultrasound, which allowed increased precision of LUS measurement, especially in its lowest part. In 2009, Bujold et al.27 published a prospective cohort study combining transabdominal and transvaginal ultrasound in the assessment of LUS thickness. Using the thinnest of the two LUS measurements, one with each modality, they proposed a cut-off of 2.3 mm as optimal for the prediction of uterine rupture in women undergoing TOLAC. They observed three (9.1%) cases of uterine rupture among 33 women with LUS thickness < 2.3 mm compared with no (0%) case among 92 women with LUS ≥ 2.3 mm who underwent TOLAC (P = 0.02). This study was limited by the fact that most women with LUS measurement < 2.0 mm were informed about the potential risk of uterine rupture; it is possible that this information affected the number of these women who opted to undergo TOLAC and therefore the number of uterine ruptures in this group, resulting in an underestimation of the risk of uterine rupture in women with LUS measurement < 2.3 mm. In a study conducted by Gotoh et al.35, it was found that there may be incomplete uterine rupture at delivery when LUS thickness at transvaginal ultrasonography is < 2 mm within 1 week of delivery, with positive and negative predictive values of 73.9% and 100%, respectively. A systematic review of 12 studies, including 1834 women and with a 6.6% rate of uterine scar defects, confirmed the strong association between the degree of LUS thinning measured in the third trimester of pregnancy and the risk of uterine scar rupture/dehiscence at delivery28. The cut-off value proposed for predicting these complications varied between 2.0 and 3.5 mm28. Due to the heterogeneity of the studies, no precise cut-off value could be recommended. These findings were confirmed by another recent meta-analysis of 21 studies37. The measurement technique for full LUS and myometrial thicknesses in the third trimester have been described by several authors27, 29, 35. Bujold et al.27 suggested performing the measurement using transabdominal and transvaginal ultrasound between 35 + 0 and 38 + 6 weeks of gestation for a fetus in cephalic presentation. Sonographically, LUS is typically a two-layer structure that consists of a hyperechogenic layer (including the bladder wall) and a hypoechogenic layer that is considered to represent the myometrium. The borders of the LUS are visualized more clearly if the patient's bladder is full. The LUS is examined longitudinally and transversely in order to find the thinnest area. To measure the full LUS thickness, one caliper is placed at the interface between urine and the bladder wall and another is placed at the interface between amniotic fluid (or fetal scalp) and the decidua. The myometrial thickness is measured with the first caliper at the interface of the bladder wall and the myometrium, so that it includes only the hypoechogenic layer. Figure 1 shows a thick and a thin LUS measured by transabdominal and transvaginal approaches. Most authors suggest that at least three measurements should be taken, with the lowest value being retained27, 35. Ultrasonographic measurement of LUS thickness can be performed using a transabdominal or a transvaginal probe27, 28, 33, 35, 38. Using a combination of transvaginal and transabdominal ultrasound, the 10th, 25th and 50th percentiles of LUS thickness are around 2.0 mm, 2.3 mm and 3.2 mm, respectively27, 28. One study, with the patients' bladders full at examination, observed a strong correlation between the two approaches, but another study, with patients' bladders empty at examination, observed no correlation33, 39. However, studies using a transabdominal approach reported greater cut-off values than did those using a transvaginal approach for the prediction of uterine scar dehiscence/rupture27, 28. Many experts suggest that a combination of both approaches is probably the best way to measure LUS thickness: transabdominal ultrasound can detect scar defects located high on the LUS, which are seen typically in women who had a previous Cesarean section performed in early gestation and/or before labor, while transvaginal ultrasound is probably the best approach to visualize scar defects located in the lowest part of the LUS, seen typically in women whose prior Cesarean section was performed late in the first or in the second stage of labor40. Both full LUS and myometrial thickness measurements have been related to the risk of uterine scar dehiscence/rupture27, 31, 32, 35, 37, 41 and opinions vary regarding the potential benefits of measuring myometrial thickness instead of full LUS thickness. However, it is sometimes difficult to distinguish the limits of the myometrial layer in the LUS, which probably reduces the accuracy of this measurement28, 37. The systematic review of Jastrow et al.28 suggests that full LUS thickness is more predictive of uterine scar dehiscence/rupture than is myometrial thickness, but this was not confirmed by the recent meta-analysis of Kok et al.37, which found no significant differences between these measurements for prediction of uterine scar dehiscence/rupture. Several studies have evaluated the reproducibility of this measurement27, 39, 42-45 and all found intra- and interobserver agreement for measurement of full LUS thickness to be good (defined as difference between observers < 1 mm). The transvaginal approach was shown to have better interobserver agreement than had the transabdominal approach27, 39, 42, 44, 45. It was also demonstrated that reproducibility between observers improves with experience27. We are not aware of any studies that evaluated the reproducibility of measurement of the myometrial layer. Two- (2D) and three-dimensional (3D) sonography and video sequences have also been used to save LUS images for external evaluation44, 45. These studies found measurements on 3D sonography to have excellent intra- and interobserver reproducibility and those on 2D sonography to have good reproducibility. Comparisons between measurements made from 2D video sequences and those at standard live 2D ultrasound examination showed high reproducibility between the techniques44. According to Martins et al., using 3D sonography could significantly improve the interobserver reliability over that possible using 2D sonography, but Cheung et al. did not confirm this observation42, 43. A major advantage of 3D sonography and video sequences of the LUS is the possibility to postpone the evaluation, allowing measurement of LUS thickness to be performed by experts located elsewhere. Few studies have evaluated factors associated with third-trimester LUS thickness31, 46-48. Cheung et al.31 observed that the LUS was thinner in women who had delivered previously by Cesarean section, compared with women who had delivered vaginally, findings confirmed recently by Ginsberg et al.48, but only in post-term pregnant women. Some studies have shown that full LUS thickness, when measured in the late second and third trimesters, decreases with progression of pregnancy, both in women with and in those without previous Cesarean section35, 49. Gotoh et al.35 observed that some women with previous Cesarean section had a persistently thin uterine wall from the second trimester until delivery, while in other women with previous Cesarean section, LUS thickness decreased progressively, in a manner similar to women without previous Cesarean section. Jastrow et al.46 found, in a prospective study of 235 women, that labor before Cesarean section in a previous pregnancy was the only factor associated with LUS thickness46; full LUS thickness was approximately 0.6 mm greater in women whose previous Cesarean section was performed during labor. They also observed, in women who had had a Cesarean section after labor commenced, that full LUS was thicker if the Cesarean section had been performed in the active phase of labor, compared with in the latent phase (P = 0.02). These results were confirmed in a retrospective cohort study on 377 women47. The same authors showed that myometrial LUS was thicker in women following hysterotomy when it had been closed with synthetic sutures versus closure with catgut (P = 0.04). Several factors could potentially explain the positive association between labor during previous Cesarean section and LUS thickness in subsequent pregnancies. When Cesarean section is performed during labor, the hysterotomy is probably carried out low on the LUS, sometimes near the cervix, and may thus involve the less contractile part of the uterus. Moreover, in labor, the LUS is more developed and may contribute to better apposition of the suture margins, allowing better scar healing40, 46. This finding is in agreement with the observation that women with previous Cesarean section performed during labor have a lower risk of uterine rupture compared with those with Cesarean section prior to labor50. However, this association could also be explained by a shorter duration of labor and a lower risk of dystocia in women who experienced labor at previous Cesarean section46. In contrast to these findings, studies performed in non-pregnant women have shown that Cesarean delivery carried out in advanced labor increased the likelihood of a large scar defect, as evaluated transvaginally 6–9 months after delivery, and that scars with large defects were located lower in the uterus than were intact scars51, 52. These discrepancies may be explained by the fact that uterine scars located very low on the LUS, sometimes into the cervical tissue, could be missed or evaluated less easily by ultrasound in future pregnancies. One of the disadvantages of third-trimester measurement of LUS thickness is that the obstetrician receives the information late in the pregnancy, when the delivery plan has already been made. Moreover, some women may commence preterm labor before their appointment for LUS thickness measurement. Women with a previous preterm Cesarean section are more likely to deliver before term and could also be at higher risk of uterine rupture53. Additionally, the procedure is often uncomfortable, because it is carried out with a full bladder at the end of pregnancy. For these reasons, an evaluation before the third trimester may be beneficial. Some authors have proposed evaluation of the scar in non-pregnant women51, 52, 54, 55, suggesting an association between large scar defects and the risk of uterine rupture/dehiscence at subsequent delivery. Vikhareva Osser and Valentin found that 5.3% (1/19) of women with an intact scar or a small scar defect had uterine dehiscence or rupture compared with 42.9% (3/7) of those with a large defect (P = 0.047)54. This association should be validated by further studies before being used in clinical practice. First-trimester Cesarean scar assessment is a promising tool that could help to categorize women at the beginning of pregnancy into those with low and those with high risk of Cesarean scar defects and could also help detect impaired placentation56, 57. As yet, however, no study has demonstrated an association between first-trimester uterine scar assessment and such complications. Furthermore, using LUS thickness measurement as a single predictor of uterine rupture in routine clinical practice could have deleterious consequences. Obstetricians who rely only on this measurement for intrapartum management may be prone to take excessive risks, which could lead to uterine rupture even in the presence of a thick sonographic LUS58. LUS thickness measured by ultrasound during the third trimester of pregnancy is inversely correlated with uterine scar rupture/dehiscence at delivery. To better visualize the thinnest LUS, transvaginal ultrasound should be added to the transabdominal approach. Ultrasound examination in the third trimester may be used as a supplementary method to predict uterine rupture in women contemplating VBAC. Taking into account clinical factors in combination with sonographic findings and careful intrapartum management could lead to increased numbers of successful VBACs with low rates of uterine rupture. Further studies are required to determine the optimal way in which to implement LUS measurement in clinical practice. Two randomized trials (LUSTrial, ClinicalTrials.gov: NCT01916044 and PRISMA trial, ISRCTN registry: 15346559) on LUS thickness measurement are currently underway and should help to clarify the potential role of the sonographic assessment of LUS in the management of women with uterine scar. There is also a need to understand better the correlation between previous uterine scar location, appearance of the scar in non-pregnant women and in pregnant women in the first and second trimesters, and LUS thickness in the late second and third trimesters.