Reference ranges for the intra‐amniotic umbilical cord vein diameter, peak velocity and blood flow in a regional NSW population

Abstract

Aim

To construct gestational age (GA)‐related reference ranges of the intra‐amniotic umbilical cord vein (UCV) diameter, peak velocity (PV) and blood flow (Qucv) using a Central West New South Wales population.

Materials and Methods

This was a prospective, quasi‐experimental study of low risk, singleton pregnancies (n = 321) between 16 and 42 weeks of gestation. Participation was voluntary following informed consent. The UCV diameter and PV were measured using B mode and duplex Doppler respectively, and Qucv calculated. Percentile values and reference range graphs were established using quantile regression modelling in R statistical software. Intraclass correlation coefficients (ICCs) were calculated to assess the intra and intersonographer reliability.

Results

Reference ranges for the UCV diameter, PV and Qucv were established and graphed. All three UCV measurements increased with advancing GA, with both diameter and Qucv exhibiting a decline in the late third trimester. The intrasonographer and intersonographer ICCs for the UCV diameter and PV showed almost perfect agreement within and between sonographers.

Conclusion

Gestational age‐related reference ranges for the UCV diameter, PV and Qucv were developed using quantile regression from a cohort of low risk, singleton pregnancies in Central West NSW. These reference ranges have the potential to assist in the diagnosis and monitoring of fetal growth restriction.

Introduction

The umbilical vein is the sole conduit transporting oxygen and nutrients to the fetus. Measurements of the intra‐amniotic umbilical cord vein (UCV) diameter, peak velocity (PV) and blood flow (Q_ucv) have potential clinical implications given these parameters have been shown to be reduced in fetal growth restriction (FGR). A growth restricted fetus does not reach its growth potential due to pathological reasons1 and FGR complicates 5–10% of all pregnancies2, 3 with more than half attributed to placental insufficiency.4

A reduction in the diameter5, 6 or cross‐sectional area7, 8, 9, 10 of the umbilical vein has been identified in the presence of FGR. A decreased umbilical vein diameter has been found to be more extreme in the presence of abnormal umbilical artery Doppler7 and poor pregnancy outcomes.6

The umbilical vein velocity has been reported to be persistently lower in FGR fetuses, with the trend evident in both longitudinal and cross‐sectional studies, in all sample sizes and across a range of gestational ages.5, 8, 11, 12, 13, 14

Umbilical vein blood flow has been investigated since the 1980s, with many researchers finding a reduction in blood flow in the presence of FGR5, 8, 11, 12, 15, 16, 17, 18, 19, 20 and some concluding that reduced blood flow was present prior to other ultrasound changes associated with FGR.12, 21

The aim of this research was to measure the diameter, PV and calculate the Q_ucv of the intra‐amniotic umbilical cord vein between the gestational ages of 16–42 weeks and establish reference ranges for each parameter using quantile regression.

Methods

Study sample

Between 5th April 2011 and 22nd March 2013, a total of 625 women consented to participate in this study and underwent ultrasound examinations at Orange or Bathurst Health Services. An introduction to the study and provision of an information sheet preceded an invitation to participate. Voluntary written consent was obtained from each client prior to entering the study.

In order to include only ‘normal’ pregnancies in the data set used to develop the UCV reference ranges, participants identified with risk factors for abnormal intrauterine growth or fetuses with structural anomalies were excluded. Criteria for inclusion were as follow: spontaneously conceived singleton pregnancy, maternal age greater than 16 years, accurate pregnancy dating from last menstrual period or first‐trimester ultrasound, gestational age (GA) greater than 16 weeks, absence of significant maternal risk factors for abnormal fetal growth including smoking and chronic diseases (diabetes, hypertension, renal diseases, etc.), absence of fetal structural anomalies including heart defects and two vessel cords, a live neonate delivered between 37 completed weeks (≥259 days) and 42 weeks (<294 days) GA, birthweight between the 10th and 90th percentiles by Australian standards,22 normal amniotic fluid index and cord artery Doppler indices, and successful measurement of the UCV diameter and PV. The selection criteria were met by 321 participants with 49.5% of participants having a single ultrasound examination, 29% two examinations, 10.9% three and 10.6% having four or more ultrasound examinations.

Study design

This quantitative research project was quasi‐experimental in design using non‐random sample selection from a naturally formed group of potential participants. The study sample was non‐random due to automatic exclusion of part of the population based on maternal and gestational ages, and the self‐selection of potential participants. UCV continuous quantitative data were collected prospectively, using both cross‐sectional and longitudinal sampling. Participant characteristics were collected prospectively, and pregnancy outcome information was retrieved retrospectively.

The study protocol was approved by The Greater Western Human Research Ethics Committee (HREC/10/GWAHS/34) on the 25th January 2011 and Charles Sturt University Human Research Ethics Committee (2011/020) on the 15th February 2011.

Ultrasound examination

Toshiba Aplio XG (Version 3, model SSA‐790A, Toshiba Medical Systems Corporation, Otawara‐shi, Tochigi‐ken, Japan) ultrasound machines manufactured in 2008 and 2009 were used for all ultrasound examinations in this research project. All measurements were recorded using either a transabdominal curved PVT‐375BT or PVT‐674BT transducer with optimised pre‐sets including compounding and harmonics.

UCV diameter measurements were undertaken on a horizontal segment of intra‐amniotic umbilical cord with the vein walls perpendicular to the ultrasound beam and the scanning plane aligned to the central widest part of the vessel. The image was optimised and magnified to occupy approximately 25% of the screen. The diameter (mm) was measured to one decimal place according to nuchal measurement guidelines23, 24 with the inner border of the horizontal line of the calipers placed on the line that defined the inner vein walls (Figure 1).

Figure 1.

Measurement of the UCV Diameter. Image Used with Permission of the Individual.

UCV PV measurements were undertaken using a vertical portion of intra‐amniotic umbilical cord vein using the longest length of vertical cord to allow maximum development of the parabolic flow profile.25 Colour, power or ADF™ Doppler imaging was used to enhance cord localisation. The Doppler gate, with a default width of 3 mm, was placed in the centre of the vessel as blood flow in the UCV was assumed to be symmetrically parabolic15 and laminar26 with the peak velocity found in the centre of the vessel. The optimised spectral trace ran for at least 4‐s with the scale, pulse repetition frequency and baseline set so the venous waveform occupied approximately three quarters of the available spectral window. Using an angle of insonation of zero, or less than 15 degrees with angle correction, the velocity waveform was recorded as a continuous, monophasic flow. The PV was visually selected on the frozen image of the spectral trace and manually measured in units of cms⁻¹ to one decimal place (Figure 2). The UCV diameter and PV measurement were repeated thrice, on different segments of cord if possible. The mean of the three measurements, rounded to one decimal place, was used for statistical analyses and construction of the GA‐related reference ranges, while the unrounded mean value was used in blood flow calculations to eliminate truncation errors.

Figure 2.

Measurement of the UCV PV. Image Used with Permission of the Individual.

The segment of cord used for PV measurements was different to the portion used to measure the diameter as the ideal measurement in spectral Doppler and B‐mode necessitate vertical and horizontal orientation of the cord, respectively. All measurements were recorded during fetal quiescence.

UCV blood flow (ml/min) was calculated as Q_UCV = π (diameter/2)² × (0.5 × UCV PV) × 6027, 28, 29, 30 where 0.5 was the spatial velocity profile coefficient applicable to a parabolic profile and the UCV was assumed to be circular.8, 27, 30, 31, 32

The intrasonographer and intersonographer reliabilities were tested on images from ten randomly selected participants. Four sonographers undertook measurements of the diameter and PV three times for each stored image.

Data analysis

All data were de‐identified and analysed by a statistician using R statistical software version 3.1.1.33 Quantile regression using the Quantreg procedure was used to model the GA‐related reference ranges for UCV diameter, PV and Q_ucv. Polynomial linear, quadratic and cubic quantile regression models were applied with backward selection of power, and the most parsimonious model was selected for each variable. For each percentile the standard error of the coefficients and P values were produced and tabulated. A value of P < 0.05 was considered statistically significant.

The cubic polynomial quantile regression equation expressing percentiles (Q_p) of the UCV variables (y) (diameter, PV and Q_ucv) as a function of GA was written as follows:

$${\mathrm{Q}}_{\mathrm{p}}(\mathrm{y})={\mathrm{b}}_{0_{\mathrm{p}}}+{\mathrm{b}}_{1_{\mathrm{p}}}\text{GA}+{\mathrm{b}}_{2_{\mathrm{p}}}{\text{GA}}^2+{\mathrm{b}}_{3_{\mathrm{p}}}{\text{GA}}^3$$

Where ${\mathrm{b}}_{0_{\mathrm{p}}}$ represents the intercept, ${\mathrm{b}}_{1_{\mathrm{p}}},{\mathrm{b}}_{2_{\mathrm{p}}}\text{and}{\mathrm{b}}_{3_{\mathrm{p}}}$ the regression coefficients, p the 5th, 10th, 25th, 50th, 75th, 90th and 95th percentiles and GA the gestational age in days.

The intrasonographer and intersonographer reliabilities were assessed using a two‐way random model including participants and sonographers. The intraclass correlation coefficient (ICC) was used to assess the interaction of these terms, and the ICC was calculated according to Gwet34 using the ASReml‐R package.35

Results

The general characteristics of the study sample are reported in Table 1, and a description of the whole study population has been published previously.36 Overall the study sample was similar to the broader 2012 Australian population in respect to the proportion of Indigenous mothers, vaginal deliveries and low Apgar scores at 5 min.37 The mothers in the study sample were significantly younger than the 2012 average Australian mother,37 with non‐Indigenous mothers dominating this trend. Significantly fewer mothers in the study sample were expecting their first child, and less male babies were born compared to 2012 national data.37 There was no significant difference in the average birthweight between the study sample and the 2012 Australian average birthweight.37

Table 1.

Maternal characteristics and pregnancy outcomes of the study sample

Parameter	Mean (±SD) or %
Number of participants	321
Number of examinations	603
Maternal age (years)	27.8 (±5.4)
Parity
0	36.2%
>1	63.8%
Ethnicity
Indigenous	2.8%
Non‐Indigenous	97.2%
Gestational age at delivery (days)	276.2 (±7.5)
Birthweight (g)	3398 (±317)
Mode of delivery
Caesarean	27.1%
Vaginal	72.9%
Sex
Males	45.5%
Females	54.5%
Apgar score <7 at 5 min	0.9%

The intrasonographer and intersonographer ICCs for the UCV diameter and PV are shown in Table 2.

Table 2.

Intrasonographer and intersonographer ICC results

Ultrasound parameter	Intraclass correlation coefficient
	Intrasonographer	Intersonographer
UCVa diameter	0.99 (95% CIb 0.36–0.99)	0.98 (95% CI 0.41–0.99)
UCV peak velocity	0.99 (95% CI 0.42–0.99)	0.98 (95% CI 0.43–0.99)

^aUmbilical cord vein.

^bConfidence interval.

All UCV variables increased with advancing GA. The relationship between UCV diameter and GA was best described by a quadratic (order 2) quantile regression model. A linear (order 1) model best described the relation of UCV PV with GA; while a cubic (order 3) model best described the relation between Qucv and GA. The regression coefficients, standard errors of the coefficients and P values for all percentiles are shown in Table 3. The P values indicate the significance of each order of the polynomial in the quantile regression with P < 0.05 considered statistically significant. The standard error of the coefficients relates to quantile regression analysis and provides an estimate of the range of the errors for coefficient estimation. The reference ranges are presented as graphs in Figures 3, 4 and 5 showing the 5th, 10th, 50th, 90th and 95th percentiles and data.

Table 3.

The fitted regression coefficients based on the polynomial quantile regression formula for the UCV diameter, PV and Q_ucv. Standard error and P values documented. A P value < 0.05 was considered statistically significant

Percentile	Regression Coefficient	Diameter			Peak velocity			Blood flow
		Coefficient Value	Standard Error of the Coefficient	P value	Coefficient Value	Standard Error of the Coefficient	P value	Coefficient Value	Standard Error of the Coefficient	P value
5th	b0	−8.75780405	1.3993055	<0.001	7.826576	0.736248	<0.001	525.2819658	236.148465	0.027
	b1	0.115346739	0.0158825	<0.001	0.019795	0.003745	<0.001	−9.76528029	4.0623865	0.017
	b2	−0.00021954	0.000042	<0.001	–	–	–	0.058617162	0.0226974	0.01
	b3	–	–	–	–	–	–	−0.00010142	0.0000411	0.014
10th	b0	−8.76521908	0.9897232	<0.001	7.704062	0.769246	<0.001	588.1080878	207.2729902	0.005
	b1	0.11585926	0.0108758	<0.001	0.025563	0.004174	<0.001	−10.8913919	3.5141597	0.002
	b2	−0.000214	0.0000277	<0.001	–	–		0.065122681	0.0192343	0.001
	b3	–	–	–	–	–	–	−0.00011246	0.000034	0.001
25th	b0	−10.3431763	0.7401659	<0.001	7.698352	0.459082	<0.001	567.3820467	169.7434624	0.001
	b1	0.133454751	0.008373	<0.001	0.033991	0.002683	<0.001	−10.8436533	2.7921243	<0.001
	b2	−0.00025093	0.000022	<0.001	–	–	–	0.066108233	0.0148089	<0.001
	b3	–	–	–	–	–	–	−0.00011292	0.0000254	<0.001
50th	b0	−10.8489549	0.9280011	<0.001	8.246923	0.591807	<0.001	795.9752009	210.3135293	<0.001
	b1	0.140566158	0.0106144	<0.001	0.039231	0.002962	<0.001	−14.7420598	3.4956642	<0.001
	b2	−0.00026341	0.0000284	<0.001	–	–	–	0.087580236	0.0188188	<0.001
	b3	–	–	–	–	–	–	−0.00014847	0.0000329	<0.001
75th	b0	−10.5470332	0.7361624	<0.001	8.652272	0.658489	<0.001	1142.772577	332.9427026	0.001
	b1	0.138452825	0.0080516	<0.001	0.047727	0.003502	<0.001	−20.4581868	5.4721738	<0.001
	b2	−0.00025159	0.0000204	<0.001	–	–	–	0.117951434	0.0291401	<0.001
	b3	–	–	–	–	–	–	−0.00019759	0.0000504	<0.001
90th	b0	−11.184482	0.8164241	<0.001	8.335517	0.644677	<0.001	840.3546836	312.1913629	0.007
	b1	0.147571362	0.0092837	<0.001	0.058507	0.003646	<0.001	−15.9904935	5.1199942	0.002
	b2	−0.00027081	0.0000247	<0.001	–	–	–	0.096783998	0.0272318	<0.001
	b3	–	–	–	–	–	–	−0.00016167	0.0000471	0.001
95th	b0	−10.8417373	1.4601604	<0.001	7.952513	1.037601	<0.001	1085.348224	452.2103257	0.017
	b1	0.146221486	0.0172645	<0.001	0.066528	0.005258	<0.001	−20.0188935	7.4751065	0.008
	b2	−0.00026603	0.0000474	<0.001	–	–	–	0.117912257	0.0401709	0.003
	b3	–	–	–	–	–	–	−0.00019569	0.0000705	0.006

Figure 3.

Reference Range of the UCV Diameter with Advancing GA Derived from Quadratic Quantile Regression with the Data Points Plotted in Black (n = 598). The Percentile Curves were Colour Coded: 5th and 95th (Blue), 10th and 90th (Gold) and 50th (Black).

Figure 4.

Reference Range of the UCV PV with Advancing GA Derived from Linear Quantile Regression with the Data Points Plotted in Black (n = 593). The Percentile Curves were Colour Coded as Follows: 5th and 95th (Blue), 10th and 90th (Gold) and 50th (Black).

Figure 5.

Reference Range of Q_ucv with Advancing GA Derived from Cubic Quantile Regression with the Data Points Plotted in Black (n = 588). The Percentile Curves were Colour Coded as Follows: 5th and 95th (Blue), 10th and 90th (Gold) and 50th (Black).

Discussion

This research presents reference ranges for the intra‐amniotic umbilical cord diameter, peak velocity and blood flow based on a regional New South Wales population.

Quantile regression has been used as an alternative to standard regression models for statistical modelling of ultrasound data in recent years30, 38, 39 and has several features that overcome assumptions of standard regression analysis and provide a better analysis of data. Quantile regression is more robust to outliers, does not assume a normal distribution of the data or that the residuals have a mean of zero or constant variance (homoscedasticity).30, 39, 40 Quantile regression considers the distribution of data rather than the mean of data thereby allowing all portions of the distribution to be assessed, such that for each percentile a different fitted function is produced in response to advancing GA.30, 38 This last point allows a better portrayal of the relationship between the variables as there may be a different rate of response with increasing GA depending on the percentile; however, a limitation of percentiles is that extreme values are grouped into the lowest and highest percentiles.41

The ICC is an assessment of the uniformity of measurements made by individuals measuring the same item.42 The ICC compares the variance of different measurements with the total variance of all measurements; with the variance derived from ANOVA using a linear mixed model in this analysis. The ICC value can range from 0 to 1, with a value greater than 0.8 indicating almost perfect agreement.43, 44 The ICC is considered superior to Pearson's and Spearman's analyses as it takes into account the differences between observers 42 and the variance of all measurements.45 Intrasonographer and intersonographer ICC were excellent for measurement of the UCV diameter and PV, with similar or superior values to other published results.25, 44, 45

Published data indicated that the largest umbilical vein diameter occurs during the third trimester, somewhere between 32 and 42 weeks GA. The majority of research found the rate of umbilical vein diameter growth slowed in the third trimester.27, 29, 46 In this research, the diameter reached a peak value at 37 weeks GA and remained relatively stable until 40 weeks GA, before declining. Similar declines after a third‐trimester peak have been noted in other studies using measurements of the intra‐amniotic UCV diameter,47 measurement of the UCV cross‐sectional area,48 measurement of the diameter and area of the whole umbilical cord 49 and assessment of clinical specimens.50

The majority of published work over the last 17 years has modelled the relationship between the intra‐abdominal or intra‐amniotic umbilical vein velocity and GA as a positive linear trend,12, 29, 30, 31, 51 similar to that found in this research.

The consensus of many studies is that umbilical vein blood flow increases exponentially throughout pregnancy. However, there is some variation in opinion as to whether this substantial increase continues up and past 40 weeks or whether there is a decline in the later weeks of pregnancy. The reference range produced from this research showed an exponential increase up to a peak at 39 weeks and then a slight decline until 42 weeks GA. The reference range trend was very similar to that produced in pioneering work undertaken by Gill et al. (1984).21 The rate of increase reduced towards term in several studies, but did not actually decline as it did in this research.27, 29, 30 The decline after 40 weeks GA documented in this research for both the UCV diameter and Qucv may be the result of a small number of observations during the 40–42 week period. This trend may also reflect the upper limit of gestational ages over which data were collected, as this research collected data up to 42 weeks GA as did both Acharya et al.27 and Sutton et al.;26 while Boito et al. collected to 36 weeks,8 Barbera et al. to 38 weeks31 and others to 40 weeks GA.21, 29, 30

There are several features and assumptions that may underlie variations between published data including measurement techniques, averaging of measurements, umbilical vein measurement sites, effects of umbilical cord coiling, fetal behavioural states, umbilical vein shape, flow profile within the UCV and spatial velocity profile coefficients. The limitations of this research include a sparsity of data between 23 to 28 weeks and after 39 weeks GA which reflects the referral patterns for obstetric ultrasounds in a regional hospital and the use of completed whole weeks of pregnancy to define the newborn as appropriately grown.22

Conclusion

This research presents the first sonographic UCV reference ranges developed from an Australian based ‘normal’ obstetrics population using quantile regression. This research demonstrated a curvilinear increase in the UCV diameter with increasing GA up until 37 weeks, followed by a plateau lasting 4 weeks and then a slight decline to 42 weeks GA, a positive linear increase in UCV PV with increasing GA; and an exponential increase in Q_ucv up to a peak at 39 weeks and then a slight decline until 42 weeks GA. The measurement techniques employed in this research demonstrated excellent reliability ensuring the data used to construct these reference ranges were reproducible at Orange and Bathurst Health Services.

These reference ranges provide normative data from an Australian population sample and add another biometric measurement in the assessment of fetal growth. These reference ranges may be employed in further research to assess their efficacy in identifying and monitoring growth restricted fetuses.

Conflict of interest

None of the authors have any financial support or relationships that may pose a potential conflict of interest. The principal author, Jacqueline Spurway received a financial assistance grant from the NSW Department of Health and a fee waiver scholarship from Charles Sturt University.

Authorship declaration

All authors have contributed to intellectual planning of the project or intellectual analysis of the data and writing of the manuscript. In addition, Jacqueline Spurway completed all data collection. All four authors are in agreement with the content of the submitted manuscript.