Antenatal Umbilical Coiling Index and Newborn Outcomes: Cohort Study

Introduction

Coiling of the umbilical cord has been the subject of anatomical and sonographic study. In modern literature, Edmonds was the first to consider the anatomy and described an “index of twist.”[1] Based on a concept that abnormal vasculogenesis and/or development of the Wharton's jelly ground substance might be a marker for adverse fetal growth and obstetric outcomes, several studies of cords examined postnatally have explored possible associations between coiling and outcomes. As a basis for such studies, the normal range for coiling at birth following uncomplicated singleton pregnancies was established in a series of 122 births.[2] Extending the concept into the antenatal period, a detailed review considered the development and structure of the umbilical cord and examined the rationale and potential to take advantage of sonography to identify abnormal coiling in utero as a marker of fetal compromise. The review concluded that although promising, further evidence was needed to substantiate this approach.[3] The same authors subsequently reported a strong correlation between sonographic and postnatal measurement of umbilical coiling. There was an indication that the risk of small for gestational age (SGA) at birth and a need for intervention due to nonreassuring fetal status were higher in the presence of an abnormal coiling index, but the confidence intervals (CIs) around the risk estimates were very wide, leading the authors to conclude that larger studies would be needed to confirm useful predictive potential.[4]

Despite technical advances in imaging and other methods of fetal assessment, the available screening and diagnostic tools have important limitations leading to adverse outcomes as a result of failure to detect fetal compromise at the same time as a high rate of obstetric intervention for suspected compromise that turns out to have been unnecessary. Similarly, the available strategies for identifying women at risk of preterm birth have their limitations. We aimed to test the predictive value of antenatal umbilical coiling index (aUCI) among a prospectively recruited cohort of antenatal women.

Methods

Study procedures

Participants were recruited on attendance for ultrasound scans as described above. They were provided with an information sheet about the study and informed consent was obtained. Transabdominal ultrasound scanning was carried out using one of the five machines that were maintained according to the manufacturer's recommendations. The models used were GE Voluson P8 Logiq E9, Philips Epiq 7G, iU22, and HD 15. Fetal biometry was undertaken according to a standardized departmental protocol based on the INTERGROWTH-21^st consortium procedures.[7] Six sonographers carried out study scans and were trained to assess the umbilical cord using a standard protocol, according to which images including a longitudinal view of the cord showing at least two complete segments (two arteries and one vein) were acquired excluding the two extremes of the cord (placental and fetal attachments). This middle portion of the cord has been shown to be the approximate arithmetic mean of the cord at the fetal and placental ends.[8] It should be noted that the practical inability to measure coiling over the entire length of the cord is an inherent limitation of the study. The images were captured and stored in the hospital's picture archiving and communication system and were available online for subsequent assessment of image quality, measurement, and calculation of the aUCI. After assessment of the acceptability of the images using the criteria described above, the distance between the coils was measured from the inner edge of an arterial or venous wall to the outer edge of the next coil along the ipsilateral side of the umbilical cord. The aUCI was calculated as a reciprocal of the distance between a pair of coils (aUCI = 1/distance in cm) as shown in Figure 1. MN performed all the calculations. If the umbilical coiling index (UCI) was too low to measure one complete coil in one view, the largest segment of cord without a complete coil was measured. The aUCI was then calculated as the reciprocal value of the mean of the two measurements of the pitch of one complete coil or as the reciprocal value of the largest length of umbilical cord without one complete coil [Figure 1].

Close

Figure 1

Color Doppler ultrasound showing measurement of the umbilical coiling index from the inner edge of an artery to the outer edge of the same artery at the adjacent umbilical twist along the ipsilateral cord side. The umbilical coiling index was 0.52.

Export to PPT

Only one image from each participant was included in the data set. Participants’ medical records were reviewed for maternal demographic characteristics, antepartum complications, and birth data. Variables retrieved include maternal age, gravidity, parity, gestational age at which the second-trimester ultrasound was performed, gestational age at delivery, mode of delivery, need for intervention at delivery, and neonatal birth weight.

Results

Data were collected prospectively between December 2013 and October 2014 where a total of 602 pregnant women who fulfilled the inclusion criteria gave informed consent. Complete data were obtained from 442 women who either delivered at the Aga Khan University Hospital or for whom complete delivery records were available from other facilities. One late miscarriage and two intrauterine fetal deaths occurred. Data from 12 participants were excluded due to poorly obtained coiling indices whose longitudinal ultrasound images were deemed inadequate. 430 participants were included in the final analysis [Figure 2].

Close

Figure 2

A flow chart demonstrating the patient inclusion process.

Export to PPT

The range of the aUCI was from uncoiled (0) to 1.49, with a median aUCI of 0.43. Normocoiling in this study was defined by an aUCI between 0.21 and 0.59.[12] Of the 430 coiling indices that were calculated, 324 (75%) were normocoiled, 45 (11%) were hypocoiled, and 61 (14%) were hypercoiled. Table 1 shows the participant characteristics and clinical outcomes grouped by aUCI category.

Table 1 Participant characteristics and clinical outcomes by category of antenatal umbilical coiling index

Using the intergrowth newborn weight for gestational age standards, 45 (10.5%) of the neonates were classified as SGA, 334 (77.7%) as normal for gestational age (NGA), and 51 (11.9%) as large for gestational age (LGA). The distribution of the SGA, NGA, and LGA in the normocoiled, hypocoiled, and hypercoiled groups is shown in Table 2. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of abnormal coiling for predicting SGA were 20%, 75%, 9.8%, and 87.4%, respectively. The odds ratio (OR) for the association between abnormal coiling index versus SGA at birth was 0.7 (95% CI: 0.3–1.6, P = 0.445). After adjusting for potential confounders using multivariable logistic regression models, there was no significant change in the above estimate with an OR of 0.6 (95% CI: 0.32–1.4, P = 0.26). The distribution of potential confounders and univariate probabilities and their distribution in the normal versus abnormal coiling are outlined in Table 3.

Table 2 The distribution of small, normal, and large for gestational age neonates by antenatal umbilical cord coiling

Table 3 The distribution of variables of interest and univariate probabilities

With regard to aUCI for the prediction of preterm birth, the sensitivity, specificity, PPV, and NPV were 47.6%, 76.9%, 9.6%, and 96.6%, respectively. The corresponding OR was 3.3 (95% CI: 1.4–7.7, P = 0.003).

The degree of agreement between the two observers independently computing the coiling index using 82 observations selected as every fifth umbilical cord image included was high, with a Kappa statistic of 0.95 and an AC1 statistic of 0.91.

Discussion

We have demonstrated a lack of association between abnormal antenatal umbilical coiling and fetal growth as reflected in SGA status at birth, but our findings indicate that there is a statistical association between abnormal coiling seen at the second-trimester scan and subsequent preterm birth.

Our results with regard to the range of observed indices of coiling are consistent with those noted in previous studies. Normocoiling was between 0.22 and 0.67 with a median of 0.43, which is comparable to previous studies where the cutoff values for hypo- and hypercoiled were between 0.2 and 0.6,[13] with a mean aUCI ranging from 0.3 to 0.42.[13141516] We observed a number of uncoiled umbilical cords (aUCI = 0). These were often associated with normal outcomes; similar findings were noted in previous studies.[17] A plausible explanation that has been put forward to explain this is that the second-trimester aUCI is a poor predictor of the degree of coiling in the third trimester or at birth, with an increased degree of coiling at delivery. This was demonstrated in an earlier study which found a poor level of agreement (kappa = 0.005) between the second trimester UCI and the values at birth with ultrasound having a poor sensitivity in detecting hypocoiling (17.3%) and hypercoiling (9.1%).[8] Naturally, this may affect the usefulness of aUCI as a predictive test in clinical practice.

In common with other reports with smaller participant numbers,[18] we found no association between SGA at birth and aUCI while some smaller studies have indicated a potentially useful association.[19] We undertook the study intending to achieve a number sufficient to confirm or refute such an association: our findings constitute “evidence of no association,” and we consider it unlikely that a further increase in the sample size would generate findings strong enough to point toward inclusion of this test to identify fetuses at risk of growth restriction. We did not undertake detailed clinical characterization of our participants at the time of scanning for aUCI to identify very high-risk clinical subgroups such as those with hypertensive disease or other medical conditions: our findings and interpretation are therefore relevant to a general unselected obstetric population.

It should be noted that the large study of Sharma et al.,[14] with 600 participants, did find statistical associations between abnormal coiling and low birth weight (rather than gestation-specific weight centiles as used in the present study). Their findings are likely to indicate an association with preterm birth rather than growth restriction. Taking data from their table of term and preterm births delivered vaginally, the PPV and NPV values would appear to be 52.8% and 97.6%, respectively. Our PPV of 9.6% was much lower, and at least in our African population, aUCI determination would not have clinical utility as a part of preterm birth risk screening. However, further exploration of this examination in combination with established approaches such as cervical length measurement might be worthwhile.

Almost perfect interobserver variability was observed in the present study with a Kappa statistic of 95.1% and an AC1 statistic of 0.91. This, like in other studies,[8] demonstrates that the measurement of aUCI is highly reproducible. However, while it has been suggested that abnormal umbilical coiling (hyper- or hypo-coiling) can be recognized during the fetal anatomic ultrasound survey in the second trimester without significantly increasing examination time,[13] our experience was that imaging for aUCI was not straightforward, especially in earlier gestations. Although we did not quantify the additional imaging time, the acquisition of at least two segments of umbilical cord coiling, with two arteries and a vein, away from the placental and fetal insertion so as to achieve the degree of reliability required was challenging and time-consuming. This can be explained by the significant variation that exists in the relationship of the umbilical vessels, especially depending on the ultrasound plane,[8] and is thus not simply a matter of sonographer training and experience.

Conclusion

Previous studies have reported contrasting findings with regard to potential associations between antenatal umbilical coiling and relevant perinatal outcomes, especially SGA and preterm birth. Our investigation had sufficient power to identify clinically important differences; while some potentially useful and statistically significant association is noted with regard to preterm birth, the practical limitations and low PPV do not indicate that priority should be given to further development of this approach.