Ultrasound in Med. & Biol., Vol. 35, No. 8, pp. 1278–1283, 2009Copyright � 2009 World Federation for Ultrasound in Medicine & Biology

Printed in the USA. All rights reserved0301-5629/09/$–see front matter

asmedbio.2009.03.014

doi:10.1016/j.ultr

d Original Contribution

MATERNAL HEPATIC VEIN DOPPLER VELOCIMETRY DURINGUNCOMPLICATED PREGNANCY AND PRE-ECLAMPSIA

WILFRIED GYSELAERS,*y GEERT MOLENBERGHS,y TINNE MESENS,* and LOUIS PEETERSz

*Department of Obstetrics and Gynecology, East Limburg Hospital, Genk, Belgium and Hasselt University, Diepenbeek,Belgium; yCenter for Statistics, Hasselt University, Diepenbeek, Belgium; and zDepartment of Obstetrics and Gynecology,

Maastricht University Medical Center, Maastricht, The Netherlands

(Received 3 December 2008, revised 4 March 2009, in final form 19 March 2009)

AObstetr6, B-36

Abstract—Changes of Doppler velocity measurements of distinct hepatic vein (HV) Doppler wave components wereevaluated during uncomplicated pregnancy (UP) as a reference to measurements in pre-eclampsia (PE). Women withUP (n 5 13) were submitted to standardised duplex scanning of HVat 11 stages of pregnancy between 10 and 38 weeks.For each stage, mean ± SD was calculated for HVA-, X-, V- and Y-peaks. Women with PE (n 5 30) were evaluated once,and mean ± SD was calculated for pregnancies ,32 weeks, 32–3416 weeks and $35 weeks. PE and UP values at cor-responding gestational age were compared statistically using t-test. HVA-velocity measurements changed markedlyfrom negative values in early uncomplicated pregnancy, converting around 22–24 weeks to positive values until term.Changes throughout gestation were less prominent for HV X-, V- and Y-velocities. HVA-velocity measurements weresignificantly lower in PE than in UP, the difference being more pronounced at 30 weeks (–3.59 ± 3.41 vs. 6.12 ± 3.43, p 50.0001) than at 37 weeks (2.35 ± 4.54 vs. 5.32 ± 1.92, p 5 0.04). From our results, we conclude that HV Doppler veloc-imetry shows a gradual shift from central venous reversed flow during atrial contraction in uncomplicated early preg-nancy to constantly forward moving flow until term. HV A-velocities are significantly lower in PE than in UP, thedifferences being more pronounced in late second trimester than near term. (E-mail: wilfried.gyselaers@zol.be)� 2009 World Federation for Ultrasound in Medicine & Biology.

Key Words: Maternal venous hemodynamics, Hepatic veins, Gestational physiology, Pre-eclampsia, Doppler.

INTRODUCTION

During pregnancy, the maternal circulation adapts to an

initial fall of total peripheral vascular resistance by

initiating a high-flow and low-resistance circulation,

combined with plasma volume expansion (Duvekot and

Peeters 1994). Also, in the maternal venous compartment,

gestational changes have been reported. Using Doppler

ultrasound at the level of the renal interlobar veins

(RIV), different evolutions of maximum and minimum

blood flow velocities have been observed during uncom-

plicated pregnancy (Gyselaers et al. 2009), resulting in

lower third trimester values of RIV impedance index

compared with the first trimester of pregnancy (Gyselaers

et al. 2008; Karabulut et al. 2003). Next to this, the RIV

impedance index was higher in pre-eclampsia than in

uncomplicated pregnancy (Bateman et al. 2004; Gyselaers

et al. 2009).

ddress correspondence to: Wilfried Gyselaers, Department ofics and Gynecology, Ziekenhuis Oost Limburg, Schiepse Bos00 Genk, Belgium. E-mail: wilfried.gyselaers@zol.be

1278

Hepatic vein (HV) Doppler waveforms have been re-

ported to change markedly during pregnancy (Roobottom

et al. 1995). A spectrum of venous HV Doppler-wave

patterns has been observed in the course of both unevent-

ful pregnancies and those complicated with pre-eclampsia

(Gyselaers 2008). Characteristics of normal HV Doppler

waveforms (A, X, V and Y) resemble those observed at

the level of the inferior vena cava (Appleton et al. 1987)

and relate to the cardiac-cycle dependent changes of the

right atrium (Fig. 1) (Downey 2005; Ommen et al. 2000).

This study was designed to explore gestation-depen-

dent change of distinct HV Doppler wave characteristics

in a group of women with uneventful pregnancy for

comparison with measurements of pre-eclampsia (PE)

pregnancies.

METHODS

Before study onset, approval of the local ethical

committee was obtained and the reproducibility of our

methodology was evaluated in a set of 24 women. For

this group, measurements, as explained further, were

Fig. 1. This figure shows the three different types of hepatic vein (HV) Doppler waveforms, as seen during normalpregnancy: triphasic (a), biphasic (b) and monophasic (c). Negative velocities (below zero line) are consistent with forwardvenous flow towards the heart, whereas those above the zero line indicate reversed venous flow, away from the heart. Asillustrated by the simultaneously recorded maternal ECG signal, the components of the pulsatile waveform follow differentstages of the cardiac cycle in the right atrium: (1) A indicates backward venous flow caused by atrial contraction; (2) Xindicates forward venous flow after atrial relaxation; (3) V indicates venous flow deceleration after atrial filling, shortlybefore tricuspid-valve opening; and (4) Y indicates forward venous flow after opening of the tricuspid valve and followingventricular relaxation. These components can be identified in the tri - and biphasic but not in the monophasic waveforms.

Maternal hepatic vein Doppler velocimetry d W. GYSELAERS et al. 1279

performed twice in the same individual by the principal

investigator (W.G.) and the intraclass correlation coeffi-

cient was calculated using maximum likelihood estima-

tion for the linear mixed model (Laenen et al. 2006;

Verbeke and Molenberghs 2001).

Only women without previous history or symptoms

of hepatic disease were included and only singleton preg-

nancies were considered. Two populations were studied:

one group represented 13 women with uncomplicated

pregnancies (UP), the other group consisted of 30 women,

admitted for PE to the Maternal-Fetal Medicine Unit of

East Limburg Hospital, Genk, Belgium. UP pregnancies

were evaluated prospectively at 11 occasions: at 12, 16,

19, 22, 24, 26, 28, 30, 33, 36 and 38 weeks. PE pregnan-

cies were only evaluated once, shortly after admission to

the hospital.

After informed consent, all women had a conven-

tional ultrasound scan as well as a pulsed Doppler flow

examination of hepatic veins. All examinations were

performed by the same ultrasonographer (W.G.), using a

3.5–7 MHz probe (Hitachi EUB 6500; Hitachi Medical

Systems N.V./S.A. Interleuvenlaan, Heverlee, Belgium).

All women were examined in supine position at random

occasions throughout the day, irrespective of food intake

(Teichgraber et al. 1997). Right, left and middle branch

of the hepatovenous tree were identified using colour

Doppler flow mapping and differentiated from hepatic

arteries and the portal system. The impact of breathing

movements on the ultrasound image was demonstrated

to every patient and the relevance of holding breath during

Doppler measurements was explained and demonstrated.

Once the patient was familiar with the instructions of

the ultrasonographer, the examination was performed

according to a standard protocol. (1) Doppler signals

were sampled at three different locations from the cranio-

caudal midportion in the liver, preferably one sample of

each of the main branches. (2) The real-time ultrasonic

B-image and Doppler signal were visualised simulta-

neously and the scanning image was frozen after visualisa-

tion of at least two to three similar Doppler waveforms

during interrupted breathing. (3) As the direction of the

Doppler beam was mostly parallel with the examined

vessels, Doppler angle correction was rarely needed. If

so, the axis of adjustment was always within 30�. (4)

Velocities were measured of the HV Doppler wave charac-

teristics A, X, V and Y (Fig. 1a and b). For the monophasic

Doppler waveform, where Doppler wave characteristics

could not be identified, venous maximum velocity

(MxV) and minimum velocity (MnV) were considered to

represent the equivalents of X and Y and of A and V,

respectively (Fig. 1c). Throughout the course of the ultra-

sound examination, the ultrasonographer was blinded

to the depicted results on the screen by covering the upper

left corner of the screen or by not being made aware of the

depicted values. (5) For every woman, each of three

consecutive measurements was recorded. After the scan,

mean values of the three measured values of A-, X-,

V- and Y-velocities were calculated and these results

were registered in the database.

Data of UP were aligned according to gestational

age: 12, 16, 19, 22, 24, 26, 28, 30, 33, 36, and 38 weeks.

At each stage, mean 6 SD was calculated for HV A-, X-,

V- and Y-velocities and gestational evolutions were

plotted graphically.

For UP, evolutions of HV A-velocities of individual

women were categorised in three groups, according to

1280 Ultrasound in Medicine and Biology Volume 35, Number 8, 2009

the presence or absence of mono-, bi- or triphasic Doppler

waveforms at gestations later than 30 weeks. Group 1

showed monophasic waveforms only; group 2 had both

mono- and biphasic waveforms and group 3 showed

mono-, bi- and tri-phasic waveforms. Standard deviations

of HV A-velocity measurements of these three groups

were calculated and compared statistically using F-test.

Evolutions of HV A-velocities before and after 30 weeks

were also compared between the three groups using

a repeated-measures model of the so-called mixed model

type, taking the repeated measures nature into account

(SAS procedure MIXED) (Verbeke and Molenberghs

2001). For each of these groups, a fitted group-specific

profile was calculated and presented graphically.

Mild and severe PE were defined according to the

criteria reported by Walfish and Hallak (2006). Data of

PE were also categorized in three groups: measurements

taken (1) at ,32 weeks, (2) between 32 and 3416 weeks

and (3) at $35 weeks of gestation. This classification

was used for comparison with data from groups of uncom-

plicated pregnancies at corresponding gestation as ex-

plained above. For each category, mean 6 SD was

calculated for HV A-, X-, V- and Y-velocities. T-Test

Fig. 2. Pregnancy-induced change in hepatic vein (HV) Dopwith uncomplicated pregnancy. The initially negative HV Aflow) at around 20–24 weeks. This indicates that the Doppler wagestation to predominantly monophasic near term (see Fig. 1). Th

V and Y with advancing gest

was used for comparison between normal pregnancies

and pre-eclampsia.

RESULTS

Intraobserver correlation between two consecutive

measurements of A, X, V and Y in a set of 24 women

was 0.78, 0.42, 0.88 and 0.62, respectively.

Figure 2 shows the gestational evolutions of HV A-,

X-, V- and Y-velocities in normal pregnancy (n 5 13).

HV-A velocity shows a remarkable evolution from nega-

tive velocity (reversed venous flow during atrial contrac-

tion) in early pregnancy, converting around 22 to 24

weeks to positive velocities (forward venous flow during

atrial contraction) throughout the further course of preg-

nancy (Fig. 2a). The values of the other HV velocity

components changed much less throughout pregnancy:

Figure 2b, c and d show a modest rise, a modest fall and

no appreciable change with pregnancy in the HV V-,

X- and Y-velocities, respectively.

Figure 3 shows the gestation-induced changes of

HV-A velocity for each of the 13 women with an uncom-

plicated pregnancy. A total of four women showed quite

pler wave characteristics A, X, V and Y in 13 women-velocity (reversed flow) becomes positive (5 forwardveform pattern shifts from predominantly triphasic in earlye changes of the HV Doppler waveform characteristics X,ation are more modest.

Fig. 3. Three different types of evolution of HV A-velocity during uncomplicated pregnancies categorised according to theevolution $30 weeks. Fig 3a shows four examples of stable HV A-velocities $30 w, with presentation of monophasicwaveforms only (group 1). Figure 3b represents six examples of instable, undulating HV A-velocities $30 weeks,indicating an intra-individual variation between mono- and biphasic waveforms (group 2). As shown in Figure 3c, threewomen also presented triphasic waveforms $30 weeks, next to bi- and monophasic types, indicating temporary reoccur-rence of HV backflow during atrial contraction in the third trimester of pregnancy (group 3). Figure 3d shows the fitted

group-specific profiles for these three groups.

Maternal hepatic vein Doppler velocimetry d W. GYSELAERS et al. 1281

stable values of third trimester HV A-velocities, due to

presentation of monophasic Doppler waves only (group

1, Fig. 3a). This was represented by a standard deviation

of ,0.8 cm/s in each individual’s measurements .30

weeks. In six other women, the third trimester Doppler

wave patterns varied between mono- and biphasic types

and this variation was represented by a standard deviation

of each individual’s A-velocity measurements between 1

and 5 cm/s (group 2, Fig. 3b). In the three remaining

women with UP, triphasic Doppler wave types were also

present next to mono- and biphasic types and this variation

was represented by a third trimester HV A-SD of .5 cm/s

for each woman (group 3, Fig. 3c). F-test showed statisti-

cally significant differences between groups 1 and 2 (p 5

0.002) and also between group 1 and 3 (p , 0.002) but not

between groups 2 and 3 (p 5 0.79).

The evolutions of HV A-velocity measurements after

30 weeks of gestation were significantly different between

the three groups (F-statistic 5 19.07, p , 0.001) but not at

gestations before 30 weeks (F-statistic 5 1.93, p 5 0.09).

The individual evolutions of HV A-velocity measure-

ments in the three groups and the fitted group-specific

profiles are presented graphically in Figure 3.

Table 1 shows the comparison of HV Doppler measure-

ments between uncomplicated pregnancies and pre-

eclampsia at three different stages of gestation. As is shown,

HV A-velocity was lower in UP than in PE for the three

categories. At gestation ,32 weeks, V- velocity was also

significantly lower in PE than in UP. HV Y- and V- velocities

were borderline significantly lower in PE than in UP at

gestations ,32 weeks and 32–3416 weeks, respectively.

DISCUSSION

Results of Doppler flow studies are subject to high

intra- and interobserver variation (Lui et al. 2005). The

reproducibility of our methodology, as reported in this

article, seems to be acceptable, as is indicated by our intra-

class correlation coefficient $ 0.66 for A, V and Y-veloc-

ities. Only for X-velocity measurement, our intraclass

correlation was ,0.5. Although there is room for bias in

our method of blinding the ultrasonographer for the de-

picted measurements at the screen of the scanner, it is

unlikely that this invalidates the results.

Three basic patterns of HV Doppler waveforms have

been reported: triphasic, biphasic and flat (Roobottom

Table 1. Comparison of HV Doppler velocity measurements between uncomplicated pregnancies (UP) and pre-eclampsia (PE) atthree different stages of gestation (This classification was used for comparison with data from the groups of uncomplicated

pregnancies at corresponding gestation.)

28–3116 weeks UP (n 5 13) PE (n 5 5) p

Gestat (w) 30.41 6 0.49 29.45 6 2.28 0.15A (cm/s) 6.12 6 3.43 -3.59 6 3.41 0.0001X (cm/s) 11.16 6 2.13 10.66 6 2.15 0.66V (cm/s) 8.58 6 2.07 3.80 6 3.15 0.002Y (cm/s) 10.17 6 1.82 8.26 6 1.26 0.05

32–3416 weeks UP (n 5 13) PE (n 5 13) p

Gestat (w) 32.94 6 0.43 33.04 6 0.78 0.69A (cm/s) 6.16 6 4.21 0.93 6 6.15 0.018X (cm/s) 10.49 6 1.79 11.69 6 3.48 0.28V (cm/s) 8.70 6 1.43 5.48 6 5.43 0.05Y (cm/s) 10.31 6 1.62 10.36 6 2.35 0.95

$ 35 weeks UP (n 5 13) PE (n 5 12) p

Gestat (w) 37.07 6 1.19 36.73 6 0.82 0.42A (cm/s) 5.32 6 1.92 2.35 6 4.54 0.04X (cm/s) 9.53 6 2.17 9.09 6 1.65 0.58V (cm/s) 6.76 6 1.02 5.86 6 2.40 0.23Y (cm/s) 8.63 6 1.47 7.99 6 1.57 0.30

The differences are more pronounced at gestation 28–3116 weeks than at $35 weeks.UP 5 uncomplicated pregnancy; PE 5 pre-eclampsia; Gestat 5 gestation; HV 5hepatic vein; w 5 weeks.

1282 Ultrasound in Medicine and Biology Volume 35, Number 8, 2009

et al. 1995) (Fig. 1). The triphasic Doppler wave resem-

bles the pattern usually observed at the level of inferior

vena cava (Appleton et al. 1987). As is explained in

Figure 1, each of the Doppler wave components reflects

a physiologic stage of the cardiac cycle in the right atrium

(Downey 2005; Ommen et al. 2000). Patterns of HV

Doppler waveforms vary widely among healthy individ-

uals. This may reflect normal human variability and/or

depend on congenital anatomical variations, liver and/or

cardiac disease, fluid overload, respiration, orthostasis,

venous (sub)obstruction and the distance between the

point of measurement and the heart (Downey 2005).

Despite this interindividual variation, we found

a consistent pattern of gestational evolution in a set of

13 women with uncomplicated pregnancies ,30 weeks.

HV A-velocity measurements showed a shift from nega-

tive values in early pregnancy to positive values in the

third trimester, with the mean point of conversion around

23 weeks. It is well known that negative A-velocities at

the level of HV reflect retrograde flow of blood from the

right atrium into the venous system, following atrial

contraction (Downey 2005; Ommen et al. 2000). The

conversion from negative to positive A-velocity during

pregnancy, as observed in our study, indicates that gesta-

tional hemodynamic mechanisms counteract this back-

flow mechanism when pregnancy advances. From our

data, it cannot be concluded what mechanism is respon-

sible for this. One generally accepted explanation for flat-

tening of HV waveforms is that this is caused either by

a reduction of liver compliance or by a rise in intra-

abdominal pressure, which occurs in many pathologies

such as ascites or large intra-abdominal tumours, as well

as in pregnancy (Roobottom et al. 1995). The pattern of

gestational evolution of HV A-velocity, as presented in

Figure 2a, resembles very much the well known evolution

of plasma volume expansion during pregnancy (Duvekot

and Peeters 1994). It has been reported that the splanchnic

bed (liver, stomach and intestines) is the most important

capacitance bed for the mobilisation of blood, as it

contains one third of the total blood volume (Pang

2000). It is likely that the gestation induced expansion

of the plasma volume interferes with vascular hemody-

namic mechanisms at the level of the splanchnic organs.

Whether the Doppler changes of HV A-velocities during

pregnancy, as observed in our study, indeed are a reflection

of the gestation induced expansion of maternal blood

volume, is to be evaluated in further research.

Despite a consistent evolution of HV A-velocity

measurements at gestations ,30 weeks, again we

observed a high interindividual variation of HV A-velocity

measurements after 30 weeks of gestation. As shown in

Figure 3, in our small group of 13 uncomplicated pregnan-

cies, we identified three different types of gestational

evolution of HV A-velocity, depending on the observed

Doppler wave types $30 weeks in each individual: mono-

phasic types only (Fig. 1c) , mono- and biphasic types

(Fig. 1b and c) and mono-, bi- and triphasic types

(Fig. 1). A total number of 13 uncomplicated pregnancies

is too low to draw any conclusions on the clinical relevance

of this subgroup classification, however, it clearly

Maternal hepatic vein Doppler velocimetry d W. GYSELAERS et al. 1283

demonstrates the diversity of HV Doppler velocimetry

measurements throughout the third trimester of uncompli-

cated pregnancy.

Our results also show that HV A-velocity measure-

ments are significantly different between UP and PE and

that this difference is more prominent in the late second

trimester than near term (Table 1). At 28–32 weeks, A-,

V- and Y-velocities in PE are significantly lower than in

UP, whereas at gestation 32–34 weeks, this is only true

for A and V, and at $35 weeks, only A remains borderline

significantly different from UP. Fractions of severe PE at

these gestational stages are 60% (3/5) at 28–32 weeks,

38.5% (5/13) at 32–35 weeks and 8.3% (1/12) at $ 35

weeks, respectively. These findings correlate well with

our former observation of different venous flow patterns

between PE and UP at the level of the renal interlobar

veins: the difference between early-onset PE and UP

was significantly stronger than between late-onset PE

and UP (Gyselaers et al. 2009). Again, our current obser-

vation confirms that the maternal venous compartment is

involved in the pathophysiology of pre-eclampsia and

that the hemodynamic background mechanisms behind

early- and late-onset pre-eclampsia are different. We are

currently expanding our observations in pregnancies,

complicated with pre-eclampsia, in order to link renal

interlobar and hepatic vein Doppler parameters to clinical

outcome and to other parameters of vascular hyperac-

tivity.

CONCLUSIONS

Our study on Doppler velocimetry in hepatic veins of

pregnant women confirms the reported observation by

Roobottom et al. (1995) and identifies a mean gestation

of 22–24 weeks as the point of conversion, where the

physiologic backflow of blood into the central venous

compartment during atrial contraction, changes to

a constant forward flow into the direction of the heart.

Our results also show that central venous backflow is

significantly higher in pre-eclampsia than in uncompli-

cated pregnancies, an effect which is more pronounced

in the late second trimester than near term. These results

illustrate once more that the maternal venous compart-

ment is involved in the pathophysiology of pre-eclampsia

and that Duplex ultrasound scanning may be a useful,

noninvasive method to obtain more information on

gestational hemodynamic background mechanisms in

both normal and complicated pregnancies.

REFERENCES

Appleton CP, Hatle LK, Popp RL. Superior vena cava and hepatic veinDoppler echocardiography in healthy adults. J Am Coll Cardiol

1987;10:1032–1039.Bateman GA, Giles W, England SL. Renal venous Doppler sonography

in pre-eclampsia. J Ultrasound Med 2004;23:1607–1611.Downey DB. The retroperitoneum and the great vessels. In:

Rumack CM, Wilson RD, Charboneau JW, Johnson JM, (eds). Diag-nostic ultrasound. Philadelphia, PA: Mosby Inc; 2005. p. 443–488.

Duvekot JJ, Peeters LL. Maternal cardiovascular hemodynamic adapta-tion to pregnancy. Obstet Gynecol Surv 1994;49:S1–S14.

Gyselaers W. Hemodynamics of the maternal venous compartment:A new area to explore in obstetric ultrasound imaging. Ultrasound

Obstet Gynecol 2008;32:716–717.Gyselaers W, Molenberghs G, Van Mieghem W, Ombelet W. Doppler

measurement of Renal Interlobar Vein Impedance Index in uncom-plicated and pre-eclamptic pregnancies. Hypertens Pregnancy2009;28:23–33.

Gyselaers W, Verswijvel G, Molenberghs G, Ombelet W. Interlobarvenous flow is different between left and right kidney in uncompli-cated third trimester pregnancy. Gynecol Obstet Invest 2008;65:6–11.

Karabulut N, Baki YA, Karabulut A. Renal vein Doppler ultrasound ofmaternal kidneys in normal second and third trimester pregnancy.

Br J Radiol 2003;76:444–447.Laenen A, Vangeneugden T, Geys H, Molenberghs G. Generalized

reliability estimation using repeated measurements. Br J Math StatPsychol 2006;59:113–131.

Lui EY, Steinman AH, Cobbold RS, Johnston KW. Human factors asa source of error in peak Doppler velocity measurement. J VascSurg 2005;42:972–979.

Ommen SR, Nishimura RA, Hurrell DG, Klarich KW. Assessment ofright atrial pressure with two-dimensional and Doppler echocardiog-raphy: A simultaneous catheterization and echocardiographic study.Mayo Clin Proc 2000;75:24–29.

Pang CC. Measurement of body venous tone. J Pharmacol ToxicolMethods 2000;44:341–360.

Roobottom CA, Hunter JD, Weston MJ, Dubbins PA. Hepatic venousDoppler waveforms: Changes in pregnancy. J Clin Ultrasound1995;23:477–482.

Teichgraber UK, Gebel M, Benter T, Manns MP. Effect of respiration,exercise, and food intake on hepatic vein circulation. J UltrasoundMed 1997;16:549–554.

Verbeke G, Molenberghs G. Linear mixed models for longitudinal data.New York: Springer; 2001.

Walfish A, Hallak M. Hypertension. In: James D, Steer P, Weiner C,Gonik B, (eds). High risk pregnancy: Management options. Philadel-phia, PA: Elsevier Inc; 2006. p. 772–789.