Background
Studies revealing a discrepancy in umbilical artery Dopplers between the two umbilical arteries in normally-grown fetuses necessitates further evaluation of the paired umbilical arteries in the setting of fetal growth restriction as this is a critical component in the surveillance of this population.
Objective
Umbilical artery Doppler sampling in fetal growth restriction is typically assessed in 1 umbilical artery in a free loop of cord. Although discrepancies of >20% between the 2 umbilical arteries occur in 1 of 3 normal pregnancies, this has not been assessed in fetal growth restriction. Our objectives were to determine the frequency of discordant Doppler pulsatility indices between paired umbilical arteries in a fetal growth restriction cohort and to determine if sampling of 1 or both arteries alters surveillance or timing of delivery.
Study Design
A cohort of 425 growth-restricted fetuses between 25 and 39 weeks of gestation had umbilical artery Doppler pulsatility indices determined from both umbilical arteries in a midsegment of the cord to determine: (1) the discrepancy percentage between paired umbilical artery pulsatility indices and (2) the frequency of both arteries being normal, abnormal, or discordant (pulsatility index < and >95th percentile). To determine what sampling method increased the detection of an abnormal Doppler index, 3 sampling methods were compared: (1) average pulsatility index from both umbilical arteries, (2) pulsatility index from 1 umbilical artery chosen randomly, and (3) highest pulsatility index of the 2 umbilical arteries.
Results
The mean percentage difference between umbilical artery pulsatility indices was 11.7%, and in 15.8% of cases, it exceeded 20%. Both umbilical artery pulsatility indices were normal in 71.1% (302/425), abnormal in 12.2% (52/425), and discordant in 16.7% (71/425) of cases ( P <.0001). Of the 3 sampling methods, the pulsatility index was abnormal in: (1) 19.2% (82/425) of cases when averaged from both umbilical arteries, (2) 22.1% (94/425) of cases when choosing 1 umbilical artery at random, and (3) 28.9% (123/425) of cases when the highest umbilical artery pulsatility index was used ( P =.003).
Conclusion
In this large fetal growth restriction cohort, the overall discrepancy between the 2 umbilical artery pulsatility indices was 11.7%. Among fetuses with at least 1 abnormal umbilical artery pulsatility index, 71 of 123 (57.7%) had 1 normal pulsatility index and 1 abnormal. Thus, the number of arteries sampled and the sampling method used may alter clinical decision-making, including frequency of surveillance and timing of delivery.
Introduction
Fetal growth restriction (FGR) complicates up to 10% of all pregnancies, carries multiple short- and long-term risks to the fetus, is associated with hypertensive disorders of pregnancy, and places the mother at high risk for cesarean delivery. , An important component of managing FGR pregnancies is the use of umbilical artery (UA) Doppler velocimetry, which has been shown to reduce perinatal death by 29% and the need for antenatal admission, induction of labor, and cesarean delivery. Although there is no debate on the benefit of UA Doppler velocimetry in FGR management in general, the recommended sampling method has not been questioned since its first description for use in FGR pregnancies 35 years ago. Although there are 2 UAs in 99% of fetuses, current recommendations are to sample a single UA in the midsegment of the umbilical cord. National and international guidelines show general agreement regarding the importance of Doppler and frequency of surveillance based on Doppler severity, but none specifically address the sampling of either 1 or both of paired UA vessels, perhaps because data are lacking.
Why was this study conducted?
This study was performed to determine whether Doppler sampling of 1 or 2 umbilical arteries (UAs) shows discrepancies between each vessel pair in growth-restricted fetuses and if management might be altered.
Key findings
In our cohort of fetuses with fetal growth restriction (FGR), the mean difference in UA pulsatility index (PI) between each UA in a pair was 11.7%; however, 16.7% (71/425) of fetuses had 1 abnormal (UA PI <95th percentile) and 1 normal UA PI. Gestational age and birthweight at delivery in this group were intermediate between fetuses with 2 abnormal or 2 normal UA Dopplers. Sampling each paired UA vessel and using the highest of the 2 UA PI values results in clinical management changes in a small percentage of patients.
What does this add to what is known?
Sampling just 1 UA vessel or averaging the UA PI values of an UA pair underestimates the number of FGR fetuses with an abnormal UA PI, which could otherwise have altered surveillance or timing of delivery. Without sampling both UAs in each fetus, it cannot be known for certain whether there is improvement, worsening, or stable Doppler values from 1 week to the next given the frequent discrepancy between UAs in a pair.
Previous studies of normal pregnancies have revealed discrepancies in UA Doppler values between the 2 paired UA in up to 38% of normally grown fetuses. The frequency of discrepant Doppler pulsatility and resistance indices including laterality (left vs right UA) in growth-restricted pregnancies is not well elucidated, with 1 study containing only 19 FGR fetuses. , These previous studies were conducted in unselected populations. Discrepancies in the frequency of absent or reversed end-diastolic flow in severe FGR have been observed, but such cases represent a small percentage of FGR seen in clinical practice. , The objectives of this study include determining the frequency of discordant Doppler pulsatility indices (PIs) between paired UA vessels and establishing how varying sampling methods of obtaining the UA PI affect the detection of abnormal UA PI in an outpatient FGR cohort. As UA Dopplers in pregnancies complicated by FGR inform the frequency of outpatient visits, timing of admission to the hospital, and timing of delivery, understanding how the evaluation of 1 or both UAs may impact management is important.
Materials and Methods
A retrospective study was undertaken, evaluating patients in university-based on- and off-campus perinatal offices in Denver, Colorado. The database for this study included 425 patients between 25 and 39 weeks of gestation whose fetuses had no obvious anatomic or cord abnormalities, with estimated fetal weights (EFWs) <10th percentile (percentile) for gestational age. Dating was established per American College of Obstetricians and Gynecologists (ACOG) criteria. All initial dating was performed by the University of Colorado team unless referred from outside private clinics, in which case dating was validated by our investigators at enrollment. EFWs were calculated from measurements of the head circumference, biparietal diameter, abdominal circumference, and femur length using Hadlock nomograms. Percentiles for EFW were assigned according to the population-based fetal growth curve by Hadlock, used consistently in our perinatal practice and found to be applicable to our study population. GE Voluson 10 ultrasound machines (GE Healthcare, Chicago, IL) were used. The current study was approved by the Colorado Multiple Institutional Review Board (IRB) under 2 IRB protocols to maximize the number of study subjects, with the first being a retrospective analysis in which consent and Health Insurance Portability and Accountability Act of 1996 authorization was waived (IRB number 14-2326, date of approval February 23, 2015), and the second being a longitudinal study with prospectively collected data that required informed consent to be obtained from all participants (IRB number 14-1360, date of approval March 16, 2018).
The final ultrasound of record before delivery performed in our center was used to obtain UA PI from both UAs in a midsegment of the cord of all fetuses, optimizing the image for an ideal angle of insonation (<30°). The normal anatomy of the umbilical cord with juxtaposed paired arteries allows unequivocal sampling of each paired UA ( Figure 1 ). UA PI values were obtained from an average of multiple flow velocity waveform with a minimum of 3 consistent appearing cardiac cycles. All fetuses had UA Dopplers revealing forward diastolic flow. Fetuses with a single UA or with absent or reversed end-diastolic flow were excluded. The percentage difference in the UA PI between each UA pair was first calculated to determine the mean discrepancy in general for the study population. Next, the frequency of normal (UA PI <95th percentile) vs abnormal UA PI (UA PI >95th percentile) values in paired UAs was determined using a validated calculator. The paired arteries were labeled “discordant” if one UA PI was >95th percentile and the other UA PI was <95th percentile.
In a separate analysis, we compared 3 groups that represent different methods of sampling the UA and of determining the PI values. Each fetus in the study cohort had the 3 methods applied to their UA PI acquisition. First, to pragmatically replicate routine practice in the United States where only 1 UA at random is routinely evaluated, we calculated the mean UA PI from 1 UA chosen by a random number generator ( www.randomnumbergenerator.com ). Second, we calculated the mean UA PI averaged from both UAs, as would occur in some ultrasound software packages. Lastly, we collected UA PI from both UAs, but only selected the mean UA PI from the higher of 2 UAs for comparison. The 3 groups representing the 3 sampling methods are henceforth abbreviated as “random,” “average,” and “highest.”
Data are presented as percentages, percentage differences, or as mean ± standard error of the mean. Following a D’Agostino-Pearson normality test, repeated-measures Friedman test (nonparametric analysis of variance) and Dunn’s multiple comparison hypothesis testing were used to compare the raw UA PIs from the 3 sampling methods. We then converted these values to UA PI percentiles and compared how the 3 sampling methods altered the detection rate of abnormal PI percentile within the cohort via chi-square analysis.
Results
In this 425-patient cohort, the mean percentage difference in the PI between each paired UA was 11.7%. Mean percentage discrepancy in fetuses with only 1 abnormal UA PI was 19.5%. Almost half (198/425) of FGR fetuses had UA PIs that were >10% discrepant, and 15.8% (67/425) of study subjects had UA PIs that were >20% discrepant ( Figure 2 ).
The number of patients with 2 normal, 2 abnormal, and 1 normal paired with 1 abnormal UA PI was significantly different in this study population ( P <.0001) ( Table 1 ). More specifically, 71.1% (302/425) of FGR fetuses had 2 normal UA PIs, and 12.2% (52/425) had 2 abnormal UA PIs. Thus, although the 2 UAs were in agreement 83.3% of the time, the 2 UA PIs were “discordant” in 16.7% of cases (71/425) ( Table 1 ). Of those 71 fetuses with discordant UAs, 80.3% (57/71) were <37 weeks at the time of ultrasound, and 19.7% (14/71) were >37 weeks. Table 2 depicts gestational age at delivery and birthweights for the 3 sampling groups for 164 patients for which there was delivery information available.
| Normality of paired UAs | Number of patients (%) | P value |
|---|---|---|
| Both UA PIs were normal | 302/425 (71.1) | <.0001 |
| Both UA PIs were abnormal (>95th %ile) | 52/425 (12.2) | |
| One UA PI <95th %ile and 1 UA PI >95th %ile | 71/425 (16.7) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
