fetal celiac and splenic artery flow velocity and pulsatility index: longitudinal reference ranges...

Ultrasound Obstet Gynecol 2008; 32: 663–672Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/uog.6145

Fetal celiac and splenic artery flow velocity and pulsatilityindex: longitudinal reference ranges and evidence forvasodilation at a low portocaval pressure gradient

C. EBBING*†, S. RASMUSSEN*†‡, K. M. GODFREY§, M. A. HANSON§ and T. KISERUD*†*Department of Obstetrics and Gynecology, Haukeland University Hospital, †Department of Clinical Medicine, University of Bergen and‡Medical Birth Registry of Norway, Locus of Registry Based Epidemiology, University of Bergen and the Norwegian Institute of PublicHealth, Bergen, Norway and §Division of Developmental Origins of Health and Disease, University of Southampton, Southampton, UK

KEYWORDS: celiac artery; circulation; Doppler; fetus; hemodynamics; portocaval pressure; reference ranges; splenic artery

ABSTRACT

Objectives To establish longitudinal reference rangesfor the fetal celiac and splenic arteries flow velocityand pulsatility index (PI), and to determine theirhemodynamic relationship to venous liver perfusion anddistribution and to other essential arteries.

Methods This was a prospective longitudinal study of 161low-risk pregnancies. Doppler recordings of the celiac andsplenic arteries were made on three to five occasions at3–5-week intervals to establish reference ranges for bloodvelocity and PI measurements. Peak systolic velocity in theductus venosus, a shunt between the umbilical and inferiorcaval veins, was used to represent the umbilicocaval (i.e.portocaval) pressure gradient, and the left portal veinblood velocity represented the umbilical distribution tothe right liver lobe. The correlations between the celiac,splenic and hepatic arteries were determined, and theirassociation with the middle cerebral and umbilical arteryPIs (MCA-PI and UA-PI) was assessed.

Results Longitudinal reference ranges for the fetal celiacand splenic arteries were established based on 510and 521 observations, respectively, during gestationalweeks 21–39. Terms for calculating conditional referenceranges to be used for repeat observations are provided.Celiac and splenic artery PIs were low when portocavalpressure and umbilical supply to the right lobe were low(P < 0.0001). Their peak systolic velocity and PI werecorrelated (r = 0.7 (95% CI, 0.6–0.8) and r = 0.5 (95%CI, 0.3–0.6), respectively), while the PI of the hepaticartery correlated weakly with those of the celiac andsplenic arteries. They were positively associated with theMCA-PI and UA-PI (P < 0.0001).

Conclusion We provide longitudinal reference ranges forthe fetal celiac and splenic arteries Doppler measurementsand show that they are involved in maintaining portalliver perfusion independently from the hepatic artery.Copyright 2008 ISUOG. Published by John Wiley &Sons, Ltd.

INTRODUCTION

At mid-gestation in the human fetus, one third of thecombined ventricular output is directed to the placenta1,2,and this declines to one fifth near term1. Nutrient-richblood returning from the placenta perfuses primarilythe fetal liver (70% and 80% at 20 and 30–40 weeksof gestation, respectively) and a correspondingly smallfraction (30–20%) is shunted through the ductus venosus(DV)3–6 to supply the left heart with oxygenated bloodthrough the foramen ovale7,8. Blood velocity in theDV reflects directly the umbilicocaval (i.e. portocaval)pressure gradient (0.5–3.5 mmHg) that drives umbilicaland portal liver perfusion9–11 (Figure 1c).

The large proportion of umbilical venous bloodperfusing the liver is linked to liver tissue proliferation,production of insulin-like growth factors 1 and 2 andoverall fetal growth12,13 and can be modified by externalfactors such as maternal body composition and diet14.While the umbilical vein is the most important supplier ofvenous blood to the fetal liver, with advancing gestationthere is an increasing contribution of deoxygenatedsplanchnic blood from the portal stem15 (14–20%at 21–39 weeks of gestation4,6). This mixes with theumbilical blood in the right liver lobe. Differences inperfusion between the left and right lobes of the fetal liver

Correspondence to: Dr C. Ebbing, Department of Obstetrics and Gynecology, Haukeland University Hospital, N-5021 Bergen, Norway(e-mail: [email protected])

Accepted: 8 May 2008

Copyright 2008 ISUOG. Published by John Wiley & Sons, Ltd. ORIGINAL PAPER

664 Ebbing et al.

SAAO

UVS

CA

AO

UVSMA

s

Spl

AO

HA

CA

DV

LPV

UV

IVC

SA

(a)

DV

HA

IVC

PVCA

SMA

SA

UV

(c)

Figure 1 Fetal vasculature of the upper abdomen in axial and sagittal views (a and c) with corresponding color Doppler images (b and d).The celiac artery (CA) is the first anterior branch from the descending aorta (AO) below the diaphragm; the Doppler gate is placed over theproximal part of the vessel (d). The splenic artery (SA) arises from the CA and continues posterior to the stomach (S) to reach the hilum ofthe spleen (Spl); the Doppler gate is placed over the proximal part of the vessel (b). DV, ductus venosus; HA, hepatic artery; IVC, inferiorvena cava; LPV, left portal vein; PV, portal vein; SMA, superior mesenteric artery; UV, umbilical vein.

could be important as they may underlie the differencesin cellular architecture16, hematopoiesis17, distribution ofenzymes18 and gene expression19 between the two liverlobes. The circulatory distribution between the left andright lobes is reflected closely in the blood velocity ofthe junction between the portal and umbilical veins, i.e.the left portal vein (LPV), which has been suggested as

a marker of this watershed phenomenon20–22. Its flowvelocity is a direct reflection of the umbilical venoussupply to the right liver lobe22.

The arterial supply to the upper abdomen, celiac artery(CA) and superior mesenteric artery is also involved inmaintaining liver perfusion. One of the three CA branches(Figure 1), the hepatic artery (HA), is connected directly to

Copyright 2008 ISUOG. Published by John Wiley & Sons, Ltd. Ultrasound Obstet Gynecol 2008; 32: 663–672.

Fetal celiac and splenic artery 665

the liver, while the other two branches, the splenic artery(SA) and left gastric artery, feed the splanchnic circuitthat drains into the portal vein. During postnatal life,increases in HA blood flow buffer liver tissue perfusionby an adenosine-mediated vasodilation in the face ofa fall in portal flow23. Our recent study suggests thatthis mechanism also operates in the human fetus, eventhough at this time the portal system is connected tothe umbilical circuit24. The fetal HA is not an easyvessel to investigate using Doppler ultrasound24. Itsorigin, the CA, arising directly from the aorta, seemsmore accessible and could represent an alternative forDoppler assessment. To date, there is no standardizedinsonation technique and reference ranges for the CAhave not been published, and there is no information asto whether a fall in umbilical venous liver perfusion leadsto a compensatory vasodilation in the CA as it does inthe HA.

The SA, another of the CA branches (Figure 1) is amajor contributor to portal flow through the splenic veinin postnatal life25. During prenatal life, the spleen hasa higher density of adrenergic neurons than it does inadults26, is involved in immunological and hematopoieticactivities27, and operates as a blood reservoir28. However,little is known of its hemodynamic regulation in humanfetuses; for example, it is not known whether localvariation in umbilical and portal liver perfusion or changesin the systemic circulation reflected in the umbilical artery(UA) and middle cerebral artery (MCA) influence SAblood flow velocity. Changes in peak systolic velocity(PSV) and pulsatility index (PI) of the SA (SA-PI) in anemicand growth-restricted fetuses29–34 indicate a potentialclinical use, and cross-sectional reference ranges havebeen published29,31,32,34. However, serial measurementsand longitudinal reference ranges, which provide a moresecure basis for clinical and research purposes35, have yetto be published.

The aim of this study was firstly to establish longitudinalreference ranges for CA and SA flow velocity and PI andthe conditional terms needed for serial measurements (i.e.to calculate the expected mean and ranges based on aprevious observation in the actual fetus). Secondly, weaddressed the hemodynamic relationships: (1) of CA andSA to the venous liver perfusion, (2) of two other majorcirculatory circuits, the UA and MCA, and (3) betweenthe CA and two of its branches, the HA and SA.

METHODS

Subjects

This study was part of a larger prospective longitudinalstudy of the fetal cerebral, umbilical and splanchniccirculation, approved by the regional committee formedical research ethics (REK Vest no 203.03). Wehave previously published the results of the UA, MCAand HA24,36. Here we present the results of the CAand SA. The participants were 161 women with low-risk pregnancies recruited after a routine ultrasound

examination at 17–20 weeks, when gestational agewas assessed37. They gave written informed consent toparticipate. Exclusion criteria were: multiple pregnancy,fetal anomaly, history of pregnancy complications (e.g.pregnancy-induced hypertension, fetal growth restrictionor abruptio placentae), or any general chronic disease.Participants who developed complications after enrolmentwere not excluded. Each woman was scheduled for anexamination on three to five occasions, at 3–5-weekintervals. Pregnancy complications, birth weight, Apgarscore, gestational age at delivery, gender and congenitalabnormalities were noted.

Measurements

For Doppler ultrasound measurements, women wereplaced in a semirecumbent position with a pillowunderneath their knees. Measurements were recorded byone operator (C.E.), except for the interobserver study,in which two operators (C.E. and T.K.) participated.We used a 2–5-, 2–7- or 4–8-MHz transabdominaltransducer (Voluson 730 Expert ultrasound machine,GE Medical Systems, Zipf, Austria) with the high-passfilter set to 70 Hz. Mechanical and thermal indices in themajority of sessions were < 1.1 and < 0.9, respectively,and were kept < 1.9 and < 1.5, respectively. All insonationbeams for Doppler measurements were aligned in thedirection of the vessel, and the angle of interrogation wasalways kept at ≤ 30◦. When this angle was not zero, theDoppler shift was corrected accordingly. Recordings wereacquired during fetal quiescence over three to six uniformheart cycles to determine PSV, time-averaged maximumvelocity (TAMXV) and PI38. At each session, which lastedno more than 60 min, we measured blood flow velocityin the CA, superior mesenteric artery, SA, HA, DV, LPV,MCA, UA and umbilical vein.

The CA was identified as the most proximal of theunpaired branches of the abdominal aorta (Figure 1).The fetal abdomen was visualized in a sagittal (oraxial) plane using color Doppler, and flow velocitywaveforms were recorded close to the abdominal aortausing pulsed Doppler. The diaphragm, abdominal aortaand neighboring superior mesenteric artery were valuableanatomical landmarks when identifying the CA (Figure 1).The SA was visualized in an axial plane, identifying itsorigin at the CA in front of the aorta and its course behindthe stomach to the spleen (Figure 1). The sample volumewas placed over the proximal part of the vessel. The HAwas assessed close to the DV24 and results included in theanalysis to assess its hemodynamic correlations to the CAand SA.

The DV-PSV was recorded7 and included in the analysisto determine the effect of portocaval pressure9–11 on CA-PI and SA-PI. We used the LPV-TAMXV20–22 to assesswhether low umbilical perfusion of the right lobe wasassociated with low impedance in the CA and SA, i.e. lowCA-PI and SA-PI. The velocity was recorded in an axial(oblique) view of the fetal abdomen using a standardizedtechnique22. We measured the UA-PI and MCA-PI using


666 Ebbing et al.

standardized techniques39,40 and calculated the umbilicalvenous flow (UV-Q) from repeated measurements ofthe inner diameter and velocity in the intra-abdominalumbilical vein24, to determine a possible association withthe CA-PI and SA-PI. This was in order to test thehypothesis that these vessels have a common dependencyon central fetal hemodynamics.

Statistics

The statistical analysis was performed using SPSS (Sta-tistical Package for the Social Sciences, SPSS Inc,Chicago, IL, USA) and the MlWin program (MlWin,Centre for Multilevel Modelling, University of Bris-tol, UK).

Multilevel modeling was used in order to calculatemean and centiles for the PSV, TAMXV and PI ofthe CA and SA according to gestational age. Toachieve normal distribution of the outcome variables,we used power transformation. The outcome variableswere regressed against gestational age using fractionalpolynomials. The 2.5th, 5th, 10th and 25th centiles werecalculated by subtracting 1.96 SD, 1.645 SD, 1.282 SD,and 0.674 SD, respectively, from the mean. The 97.5th,95th, 90th and 75th centiles were calculated by addingthe respective multiples of the SD to the mean. The95% CI of the mean, 5th and 95th centiles werederived.

To assess the effect of DV-PSV, LPV-TAMXV, UV-Q,MCA-PI, and UA-PI on PI and PSV in the CAand SA, we included these measurements (in threecategories: < 10th centile, 10–90th centile and >90th

centile, because we expected responses to occur morecommonly in extreme conditions) as indicator variables inthe multilevel regression models, which describe the meanof CA-PI, SA-PI, CA-PSV and SA-PSV. The gestationalage-dependent centiles of these independent variableswere also calculated by multilevel regression. Indicatorvariables with significant improvement in the goodness offit to the models, as assessed by the deviance statistics(χ2 with P < 0.05), were considered to significantlyinfluence the outcome variables. Associations betweenPSV and PI in the CA and SA and those of the HA and SAwere assessed by simple linear regression of transformeddata.

Reproducibility for the CA and SA measurements wasassessed by intra- and interobserver variations calculatedwith a paired sample t-test and intra- and interclasscorrelations41.

RESULTS

The median maternal age at inclusion was 29 (range,20–40) years and the median gestational age at deliverywas 40 + 3 (range, 35 + 3 to 42 + 4) weeks. TheCesarean section rate (10.6%) and birth weights (median,3700 g; range, 2260–4980) were similar to values inthe local population (the Cesarean section rate being11.8%)42. Five (3.1%) women developed pre-eclampsia.

90(a)

(b)

(c)

80

70

60

50

40

CA

-PSV

(cm

/s)

30

20

10

018 20 22 24 26 28 30 32 34 36 38 40

35

30

25

20

15C

A-T

AM

XV

(cm

/s)

10

5

018 20 22 24 26 28 30 32 34 36 38 40

3.5

3.0

2.5

2.0

1.5CA

-PI

1.0

0.5

018 20 22 24 26 28 30 32 34 36 38 40

Gestational age (weeks)



Figure 2 Longitudinal reference ranges for the celiac artery peaksystolic velocity (CA-PSV) (a), time-averaged maximum velocity(CA-TAMXV) (b), and pulsatility index (CA-PI) (c), showing fitted5th, 50th and 95th centiles with 95% CIs based on 510 observations.

Population characteristics have been further detailedpreviously36. The intra- and interobserver study showedgood reproducibility, with intra- and interclass correlationcoefficients ≥ 85% for all parameters tested (Table S1available online).

In a total of 633 examinations, we obtained 510 mea-surements from the CA (80.6% success rate) (Figure 2).Unfavorable fetal position and fetal movements werereasons for failure to obtain measurements. There wasno angle correction required in 179 recordings. The



Table 1 Reference ranges for the celiac artery peak systolic velocity(cm/s) based on 510 observations in 161 low-risk pregnancies

CentileGA(weeks) 50th 2.5th 5th 10th 25th 75th 90th 95th 97.5th

21 26 16 17 19 22 30 35 38 4022 28 17 19 21 24 33 37 40 4323 30 19 20 22 26 35 40 43 4624 32 20 22 24 28 37 43 46 4925 34 22 24 26 30 40 45 48 5226 36 23 25 27 31 42 47 51 5427 38 25 27 29 33 44 50 53 5728 40 26 28 30 35 46 52 56 5929 42 27 29 32 36 48 54 58 6130 43 28 30 33 38 50 56 60 6331 45 29 32 34 39 51 58 62 6532 46 30 33 35 40 53 59 63 6733 47 31 33 36 41 54 60 65 6834 48 32 34 37 42 55 62 66 7035 49 32 35 38 43 56 62 67 7136 49 33 35 38 43 56 63 67 7137 50 33 35 38 43 57 64 68 7239 50 33 36 38 44 57 64 68 72

GA, gestational age.

Table 2 Reference ranges for the celiac artery pulsatility indexbased on 510 observations in 161 low-risk pregnancies


21 1.73 1.23 1.30 1.38 1.53 1.96 2.19 2.35 2.4922 1.81 1.28 1.36 1.44 1.60 2.05 2.30 2.46 2.6223 1.88 1.33 1.41 1.50 1.67 2.13 2.39 2.57 2.7324 1.94 1.37 1.45 1.54 1.72 2.20 2.48 2.66 2.8325 2.00 1.41 1.48 1.58 1.76 2.26 2.54 2.73 2.9126 2.04 1.43 1.51 1.61 1.80 2.31 2.60 2.79 2.9727 2.07 1.45 1.53 1.64 1.83 2.35 2.64 2.84 3.0228 2.09 1.47 1.55 1.65 1.84 2.37 2.67 2.87 3.0529 2.10 1.47 1.56 1.66 1.85 2.38 2.68 2.88 3.0730 2.10 1.47 1.56 1.66 1.85 2.39 2.68 2.88 3.0731 2.09 1.47 1.55 1.66 1.85 2.38 2.67 2.87 3.0632 2.08 1.46 1.54 1.64 1.83 2.36 2.65 2.85 3.0433 2.06 1.44 1.53 1.63 1.82 2.33 2.62 2.82 3.0034 2.03 1.43 1.51 1.61 1.79 2.30 2.59 2.78 2.9635 1.99 1.40 1.48 1.58 1.76 2.26 2.54 2.73 2.9036 1.95 1.38 1.45 1.55 1.73 2.21 2.49 2.67 2.8437 1.91 1.35 1.42 1.52 1.69 2.16 2.43 2.61 2.7739 1.81 1.28 1.35 1.44 1.60 2.05 2.30 2.46 2.62


median angle of insonation was 12◦ (range, 0–30◦),and the median sample volume was 3 (range, 1–8) mm.CA-PSV, CA-TAMXV and CA-PI with fitted meanand reference intervals are presented in Figure 2, andthe corresponding gestational age-specific centiles arepresented in Tables 1, 2 and S2 (available online).The end-diastolic velocity in the CA was alwayspositive.

0

10

20

30

40

50

60

70(a)

18 20 22 24 26 28 30 32 34 36 38 40

SA-P

SV (

cm/s

)

0

5

10

15

20

25

30

35

40

45(b)

18 20 22 24 26 28 30 32 34 36 38 40

SA-T

AM

XV

(cm

/s)

0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0(c)

18 20 22 24 26 28 30 32 34 36 38 40




SA-P

I

Figure 3 Longitudinal reference ranges for the splenic artery peaksystolic velocity (SA-PSV) (a), time-averaged maximum velocity(SA-TAMXV) (b), and pulsatility index (SA-PI) (c), showing fitted5th, 50th and 95th centiles with 95% CIs based on 521observations.

We obtained 521 measurements from the SA (82.3%success rate) (Figure 3). The median angle of insonationwas 0◦ (range, 0–30◦), and the median sample volumewas 4 (range, 2–8) mm. Similar to the CA, the SA velocityincreased throughout the second half of pregnancy, andthe curve describing the PI had an inverted U-shape(Figure 3 and Tables 3, 4 and S3, available online).Terms for calculating conditional ranges for repeatmeasurements of the CA and SA are presented inAppendix S1 (available online).


668 Ebbing et al.

Table 3 Reference ranges for the splenic artery peak systolicvelocity (cm/s) based on 521 observations in 161 low-riskpregnancies


21 19 8 9 11 15 24 28 31 3322 22 10 11 13 17 26 31 34 3623 24 12 13 16 19 29 34 36 3924 26 14 15 18 22 31 36 39 4225 29 15 17 20 24 34 39 42 4426 31 17 19 22 26 36 41 44 4727 33 19 21 24 28 39 44 47 4928 35 21 23 26 30 41 46 49 5229 37 23 25 27 32 43 48 51 5430 39 24 26 29 34 45 50 53 5631 41 26 28 31 35 46 52 55 5832 42 27 29 32 37 48 53 57 6033 44 28 31 33 38 49 55 58 6134 45 29 32 34 39 51 56 60 6335 46 30 33 35 40 52 57 61 6436 47 31 34 36 41 53 58 62 6537 48 32 34 37 42 54 59 63 6639 49 33 35 38 43 55 61 64 67


Table 4 Reference ranges for the splenic artery pulsatility indexbased on 521 observations in 161 low-risk pregnancies


21 1.39 0.90 0.96 1.04 1.19 1.64 1.91 2.09 2.2822 1.50 0.96 1.03 1.11 1.28 1.77 2.07 2.28 2.4923 1.60 1.01 1.09 1.18 1.36 1.89 2.22 2.45 2.6724 1.68 1.06 1.14 1.23 1.42 1.99 2.34 2.58 2.8225 1.74 1.10 1.18 1.28 1.48 2.07 2.43 2.69 2.9426 1.79 1.12 1.20 1.31 1.51 2.13 2.51 2.77 3.0327 1.82 1.14 1.22 1.33 1.54 2.17 2.56 2.83 3.1028 1.84 1.15 1.24 1.35 1.56 2.19 2.59 2.87 3.1429 1.85 1.16 1.24 1.35 1.56 2.20 2.60 2.88 3.1530 1.85 1.15 1.24 1.35 1.56 2.20 2.60 2.88 3.1531 1.84 1.15 1.23 1.34 1.55 2.19 2.58 2.86 3.1332 1.82 1.14 1.22 1.33 1.54 2.16 2.55 2.82 3.0933 1.79 1.12 1.21 1.31 1.52 2.13 2.51 2.78 3.0434 1.76 1.11 1.19 1.29 1.49 2.10 2.47 2.73 2.9935 1.73 1.09 1.17 1.27 1.47 2.06 2.42 2.67 2.9236 1.69 1.07 1.15 1.25 1.44 2.01 2.36 2.61 2.8537 1.66 1.05 1.12 1.22 1.41 1.96 2.31 2.55 2.7839 1.58 1.00 1.07 1.17 1.34 1.86 2.18 2.41 2.63


For the LPV Doppler measurements (Figure S2,online) the median angle of insonation was 1◦ (range,0–30◦) and the median sample volume was 2 (range,0.7–4.0) mm.

With respect to the hemodynamic relationship betweenthe CA, SA and HA, there was no correlation betweenSA-PI and HA-PI (r = 0.1; 95% CI, −0.05 to 0.30)

and a weak correlation between HA-PI and CA-PI(r = 0.3; 95% CI, 0.1–0.5), while there was a strongercorrelation between SA-PI and CA-PI (r = 0.5; 95%CI, 0.3–0.6) (Figure 4). The CA-PI/SA-PI ratio showeda stable relationship throughout the second half ofpregnancy, with a mean of 1.019 (95% CI, 1.016–1.023)(Figure S3 online).

We hypothesized that low DV-PSV, representing lowportocaval pressure10, would be associated with lowimpedance in the CA and SA, and the regression analysisconfirmed that fetuses with DV-PSV < 10th centile hada lower CA-PI and SA-PI (P < 0.0001) (Figure 5). Wehypothesized that low velocity in the LPV (indicatingless oxygenated umbilical blood to the right lobe) wouldbe associated with corresponding arterial vasodilation,and the regression analysis confirmed that LPV-TAMXV< 10th centile was associated with lower CA-PI andSA-PI (P < 0.0001) (Figure 5). UV-Q < 10th centile wasassociated with low CA-PI and SA-PI (P < 0.0001), in linewith our hypothesis. However, we also found a reduced

4.0(a)

3.0

2.0

SA-P

I

1.0

00 1.0 2.0

CA-PI

3.0 4.0

2.0

(b)

HA

-PI

1.0

00 1.0 2.0

SA-PI3.0

Figure 4 Simple linear regression analysis of transformed data of:(a) celiac and splenic artery pulsatility indices (CA-PI and SA-PI),showing their hemodynamic relationship (r = 0.5); (b) hepatic andsplenic artery pulsatility indices (HA-PI and SA-PI), showing theirindependence (r = 0.1). Regression lines with 95% CI are indicated.



2.0(a)

1.5

1.0

SA-P

I

0.5

0

2.0(b)

1.5

1.0

SA-P

I

0.5

0

2.0(c)

1.5

1.0

SA-P

I

0.5

020 24 28

Gestational age (weeks)32 36 40

20 24 28Gestational age (weeks)

32 36 40

20 24 28Gestational age (weeks)

32 36 40

Figure 5 Influence of portocaval (i.e. umbilicocaval) pressuregradient, expressed by the ductus venosus peak systolic velocity (a),umbilical venous distribution to the right liver lobe expressed bythe left portal vein time-averaged maximum velocity (b) andumbilical venous flow (c), on local regulation, expressed by thesplenic artery pulsatility index (SA-PI). Lines indicate < 10th centile( ), 10th –90th centile ( ) and > 90th centile ( ) forthe variables and the difference between the lines represents theeffect, all being significant.

CA-PI and SA-PI when the umbilical blood flow was high(UV-Q > 90th centile) (Figure 5c).

The MCA and UA are essential parts of the fetalcirculation. We tested whether a low MCA-PI and UA-PIwere associated with a low CA-PI and SA-PI. Again,regression analysis showed that MCA-PI and UA-PI

< 10th centile were associated with low CA-PI and SA-PI(P < 0.0001) (Figure S4 online).

Absolute blood flow velocity is related positively tovolume flow43,44. Regression analysis showed that highvenous velocity and flow (DV-PSV, LPV-TAMXV andUV-Q > 90th centile) were associated with high velocityin the CA and SA (P < 0.005 for the DV-PSV vs. CA-PSV relationship; P < 0.0001 for all others). There wasa strong correlation between the absolute velocity inthe arteries: HA-PSV vs. CA-PSV (r = 0.6; 95% CI,0.5–0.8) and SA-PSV vs. CA-PSV (r = 0.7; 95% CI,0.6–0.8).

DISCUSSION

The spleen has important hematopoietic and immuno-logical functions in the developing fetus26,28,45. We haveshown that the fetal spleen, rich in adrenergic neurons26,is also involved in maintaining portal perfusion (Figure 5),but independently from the HA (Figure 4), which buffersliver perfusion directly by increasing arterial supplyto the sinusoids23,24. There is also low impedance inthe SA when the right liver lobe has a low umbili-cal supply (Figure 5). We expect that the pattern seenin these physiological pregnancies represents a regula-tory mechanism of vasodilation that is augmented inextreme cases, explaining the previously reported changesin SA-PI and SA-PSV in fetal growth restriction andanemia29–34.

The CA feeds both the HA and the SA, and ourresults suggest that the SA and HA have independent localregulatory mechanisms, and that the CA hemodynamicsare more closely related to those of the SA than they areto those of the HA (Figure 4). Thus, the CA-PI cannot beused as a substitute for the less accessible HA-PI (as wehad hoped for), nor can their absolute velocities be usedinterchangeably. Rather, we found the CA-PI/SA-PI ratioto be constant throughout the second half of gestation(Figure S3, online). This finding is surprising becausegrowth velocity and most hemodynamic relationships aregestational age-dependent.

The longitudinal reference ranges established forthe SA and CA in this study, together with thoserecently established for the HA24, provide the meansfor a differential hemodynamic assessment of the fetus,particularly in cases such as those with fetal growthrestriction or anemia. In addition to being suitablefor single measurements, these longitudinal ranges areappropriate for serial observations35 when used incombination with the terms for calculating conditionalreference ranges (Appendix S1, online). The narrow 95%CI for the centiles indicates that the reference ranges arealso reliable for extreme values (Figures 2 and 3).

The variation seen between these results and those ofprevious cross-sectional studies of the SA (Figure 6)31,32

may be due to differences in design, populationsample, insonation technique and analysis. Bahado-Singh et al. showed that their reference values forSA-PSV were less suitable for predicting anemia


670 Ebbing et al.

when the gestational age was < 28 weeks31. Com-paring our new reference ranges with their resultsfrom anemic fetuses, it seems justified to reassessthe reference ranges, particularly for pregnancies of< 28 weeks’ gestation. Our results for the mean LPV-TAMXV were similar to those of another longitudinalstudy22.

The inverted U-shaped curve of our reference rangesfor the CA-PI and SA-PI (Figures 2 and 3) is in line withthe results of previous cross-sectional studies of the SA-PI and SA resistance index29,34, but in contrast to thoseof another study32 (Figure 6). The inverted U-shape issimilar to that for the MCA-PI36,46, probably reflectingorgan growth and vascularization. The spleen exhibitslinear growth during the second half of pregnancy47,48,while there is a curvilinear increase in normalized portalflow (to which SA is an important contributor) towardsterm15, in agreement with increasing vascularization ofthe spleen.

Arterial blood velocity is not only related to the volumeof blood flowing through an organ43,44 but also tothe kinetic condition of the central hemodynamics and

0

10

20

30

40

50

60

70(a)

(b)

18 20 22 24 26 28 30 32 34 36 38 40Gestational age (weeks)

SA-P

SV (

cm/s

)

0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

18 20 22 24 26 28 30 32 34 36 38 40


SA-P

I

Figure 6 The 5th, 50th and 95th centiles with 95% CIs for thesplenic artery peak systolic velocity (SA-PSV) (a) and pulsatilityindex (SA-PI) (b) established in the present longitudinal study (solidlines), compared with ranges based on cross-sectional studies(dashed lines); (a) shows median and 95th centile fromBahado-Singh et al.31 and (b) shows 5th, 50th and 95th centiles fromCapponi et al.32.

cardiac output. This concept is further supported by ourfinding of a positive association of the absolute velocityin the CA and SA with venous velocity and flow. Ourfindings of growing blood flow velocity in the superiorsplanchnic arteries during late pregnancy are in line withan increasing perfusion of the splanchnic organs towardsthe end of pregnancy, and a correspondingly relativeincrease in portal flow to the fetal liver6. We have alsoshown low PI in the CA and SA to be associated with lowPI in the UA and MCA, indicating a common dependenceon general hemodynamics (Figure S4, online).

In conclusion, we have shown that the CA feedstwo independently regulated arteries, the HA and SA,of which both are involved in maintaining fetal venousliver perfusion. The longitudinal reference ranges that wepresent provide new means for the differential assessmentof the splanchnic circulation in the fetus.

ACKNOWLEDGMENT

The Western Norway Regional Health Authority (Grantno 911160) supported the study. M.A.H is supported bythe British Heart Foundation.

REFERENCES

1. Kiserud T, Ebbing C, Kessler J, Rasmussen S. Fetal cardiacoutput, distribution to the placenta and impact of pla-cental compromise. Ultrasound Obstet Gynecol 2006; 28:126–136.

2. Sutton MSJ, Plappert T, Doubilet P. Relationship betweenplacental blood flow and combined ventricular output withgestational age in normal fetuses. Cardiovasc Res 1991; 25:603–608.

3. Kiserud T, Rasmussen S, Skulstad S. Blood flow and the degreeof shunting through the ductus venosus in the human fetus.Am J Obstet Gynecol 2000; 182: 147–153.

4. Bellotti M, Pennati G, De Gasperi C, Battaglia FC, Ferrazzi E.Role of ductus venosus in distribution of umbilical blood flowin human fetuses during second half of pregnancy. Am J PhysiolHeart Circ Physiol 2000; 279: H1256–H1263.

5. Haugen G, Kiserud T, Godfrey K, Crozier S, Hanson M. Portaland umbilical venous blood supply to the liver in the human fetusnear term. Ultrasound Obstet Gynecol 2004; 24: 599–605.

6. Kessler J, Rasmussen S, Godfrey K, Hanson M, Kiserud T.Longitudinal study of umbilical and portal venous blood flowto the fetal liver: low pregnancy weight gain is associated withpreferential supply to the left fetal liver lobe. Ped Res 2008; 63:315–320.

7. Kiserud T, Eik-Nes SH, Blaas HGK, Hellevik LR. Ultrasono-graphic velocimetry of the fetal ductus venosus. Lancet 1991;338: 1412–1414.

8. Kiserud T, Eik-Nes SH, Blaas HG, Hellevik LR. Foramenovale – an ultrasonographic study of its relation to the inferiorvena-cava, ductus venosus and hepatic veins. Ultrasound ObstetGynecol 1992; 2: 389–396.

9. Hellevik LR, Kiserud T, Irgens F, Ytrehus T, Eik-Nes SH.Simulation of pressure drop and energy dissipation for bloodflow in a human fetal bifurcation. J Biomech Eng 1998; 120:455–462.

10. Kiserud T, Hellevik LR, Eik-Nes SH, Angelsen BAJ, Blaas HG.Estimation of the pressure-gradient across the fetal ductusvenosus based on Doppler velocimetry. Ultrasound Med Biol1994; 20: 225–232.



11. Schroder HJ, Tchirikov M, Rybakowski C. Pressure pulses andflow velocities in central veins of the anesthetized sheep fetus.Am J Physiol Heart Circ Physiol 2003; 284: H1205–H1211.

12. Tchirikov M, Kertschanska S, Schroder HJ. Obstruction ofductus venosus stimulates cell proliferation in organs of fetalsheep. Placenta 2001; 22: 24–31.

13. Tchirikov M, Kertschanska S, Sturenberg HJ, Schroder HJ.Liver blood perfusion as a possible instrument for fetal growthregulation. Placenta 2002; 23: S153–S158.

14. Haugen G, Hanson M, Kiserud T, Crozier S, Inskip H,Godfrey KM. Fetal liver-sparing cardiovascular adaptationslinked to mother’s slimness and diet. Circ Res 2005; 96:12–14.

15. Kessler J, Rasmussen S, Kiserud T. The fetal portal vein: normalblood flow development during the second half of humanpregnancy. Ultrasound Obstet Gynecol 2007; 30: 52–60.

16. Emery JL. Functional asymmetry of liver. Ann N Y Acad Sci1963; 111: 37–44.

17. Emery JL. The distribution of haemopoietic foci in the infantilehuman liver. J Anat 1956; 90: 293–297.

18. Ring JA, Ghabrial H, Ching MS, Shulkes A, Smallwood RA,Morgan DJ. Propranolol elimination by right and left fetalliver: Studies in the intact isolated perfused fetal sheep liver.J Pharmacol Exp Ther 1998; 284: 535–541.

19. Cox LA, Schlabritz-Loutsevitch N, Hubbard GB, Nijland MJ,McDonald TJ, Nathanielsz PW. Gene expression profile dif-ferences in left and right liver lobes from mid-gestation fetalbaboons: a cautionary tale. J Physiol 2006; 572: 59–66.

20. Kiserud T, Kilavuz O, Hellevik LR. Venous pulsation in thefetal left portal branch: the effect of pulse and flow direction.Ultrasound Obstet Gynecol 2003; 21: 359–364.

21. Kilavuz O, Vetter K, Kiserud T, Vetter P. The left portal vein isthe watershed of the fetal venous system. J Perinat Med 2003;31: 184–187.

22. Kessler J, Rasmussen S, Kiserud T. The left portal vein as anindicator of watershed in the fetal circulation: developmentduring the second half of pregnancy and a suggestedmethod of evaluation. Ultrasound Obstet Gynecol 2007; 30:757–764.

23. Lautt WW. The 1995 Ciba-Geigy Award Lecture. Intrinsicregulation of hepatic blood flow. Can J Physiol Pharmacol1996; 74: 223–233.

24. Ebbing C, Rasmussen S, Godfrey KM, Hanson MA, Kiserud T.Hepatic artery hemodynamics suggest operation of a bufferresponse in the human fetus. Reprod Sci 2008; 15:166–178.

25. Tsukuda T, Ito K, Koike S, Sasaki K, Shimizu A, Fujita T,Miyazaki M, Kanazawa H, Jo C, Matsunaga N. Pre- andpostprandial alterations of portal venous flow: Evaluation withsingle breath-hold three dimensional half-Fourier fast spin-echoMR imaging and a selective inversion recovery tagging pulse.J Magn Reson Imaging 2005; 22: 527–533.

26. Anagnostou VK, Doussis-Anagnostopoulou I, Tiniakos DG,Karandrea D, Agapitos E, Karakitsos P, Kittas C. Ontogeny ofintrinsic innervation in the human thymus and spleen. J His-tochem Cytochem 2007; 55: 813–820.

27. Vatner SF. Effects of hemorrhage on regional blood-flowdistribution in dogs and primates. J Clin Invest 1974; 54:225–235.

28. Cumano A, Godin I. Ontogeny of the hematopoietic system.Ann Rev Immunol 2007; 25: 745–785.

29. Abuhamad AZ, Mari G, Bogdan D, Evans AT. Doppler flowvelocimetry of the splenic artery in the human fetus – Is it amarker of chronic hypoxia? Am J Obstet Gynecol 1995; 172:820–825.

30. Bahado-Singh R, Oz U, Deren O, Pirhonen J, Kovanci E,Copel J, Onderoglu L. A new splenic artery Doppler veloci-metric index for prediction of severe fetal anemia associated

with Rh alloimmunization. Am J Obstet Gynecol 1999; 180:49–54.

31. Bahado-Singh R, Oz U, Deren O, Kovanchi E, Hsu CD,Copel MJ, Mari G. Splenic artery Doppler peak systolic veloc-ity predicts severe fetal anemia in rhesus disease. Am J ObstetGynecol 2000; 182: 1222–1226.

32. Capponi A, Rizzo G, Arduini D, Romanini C. Splenic arteryvelocity waveforms in small-for-gestational-age fetuses: Rela-tionship with pH and blood gases measured in umbilical bloodat cordocentesis. Am J Obstet Gynecol 1997; 176: 300–307.

33. Oz U, Kovanci E, Jeffress A, Mendilicioglu I, Mari G, Bahado-Singh RO. Splenic artery Doppler in the prediction of thesmall-for-gestational age infant. Ultrasound Obstet Gynecol2002; 20: 346–350.

34. Dubiel M, Gudmundsson S, Kozber H, Debniak B, Breborow-icz GH, Marsal K. Fetal splenic color Doppler energy imagingand velocimetry in normal and high-risk pregnancies. PrenatNeonat Med 1997; 2: 205–209.

35. Royston P. Calculation of unconditional and conditionalreference intervals for fetal size and growth from longitudinalmeasurements. Stat Med 1995; 14: 1417–1436.

36. Ebbing C, Rasmussen S, Kiserud T. Middle cerebral arteryblood flow velocities and pulsatility index and the cerebro-placental ratio: longitudinal reference ranges and terms forserial measurements. Ultrasound Obstet Gynecol 2007; 30:287–296.

37. Johnsen SL, Rasmussen S, Sollien R, Kiserud T. Fetal ageassessment based on ultrasound head biometry and the effect ofmaternal and fetal factors. Acta Obstet Gynecol Scand 2004;83: 716–723.

38. Gosling R, King D. Ultrasound angiology. In Arteries and Veins,Macus A, Adamson J (eds). Churchill-Livingstone: Edinburgh,1975; 61–98.

39. Acharya G, Wilsgaard T, Berntsen GKR, Maltau JM, Kiserud T.Reference ranges for serial measurements of umbilical arteryDoppler indices in the second half of pregnancy. Am J ObstetGynecol 2005; 192: 937–944.

40. Mari G, Abuhamad AZ, Cosmi E, Segata M, Altaye M,Akiyama M. Middle cerebral artery peak systolic veloc-ity – technique and variability. J Ultrasound Med 2005; 24:425–430.

41. Bland JM, Altman DG. Statistical notes: Measurement error andcorrelation coeffecients. BMJ 1996; 313: 41–42. [With correc-tion in http://www.bmj.com/cgi/eletters/313/7048/41#21854].

42. Skjaerven R, Gjessing HK, Bakketeig LS. Birthweight by ges-tational age in Norway. Acta Obst Gynecol Scand 2000; 79:440–449.

43. Acharya G, Erkinaro T, Makikallio K, Lappalainen T,Rasanen J. Relationships among Doppler-derived umbilicalartery absolute velocities, cardiac function, and placentalvolume blood flow and resistance in fetal sheep. Am J PhysiolHeart Circ Physiol 2004; 286: H1266–H1272.

44. Acharya G, Wilsgaard T, Berntsen GKR, Maltau JM, Kiserud T.Doppler-derived umbilical artery absolute velocities and theirrelationship to fetoplacental volume blood flow: a longitudinalstudy. Ultrasound Obstet Gynecol 2005; 25: 444–453.

45. Christensen JL, Wright DE, Wagers AJ, Weissman IL. Circula-tion and chemotaxis of fetal hematopoietic stem cells. PLoSBiol 2004; 2: 368–377.

46. Mari G, Deter RL. Middle cerebral artery flow velocity wave-forms in normal and small-for-gestational-age fetuses. Am JObstet Gynecol 1992; 166: 1262–1270.

47. Aoki S, Hata T, Kitao M. Ultrasonographic assessment of fetaland neonatal spleen. Am J Perinat 1992; 9: 361–367.

48. Oepkes D, Meerman RH, Vandenbussche FPHA, Vankamp IL,Kok FG, Kanhai HHH. Ultrasonographic fetal spleen measure-ments in red-blood-cell alloimmunized pregnancies. Am J ObstetGynecol 1993; 169: 121–128.


672 Ebbing et al.

SUPPORTING INFORMATION ON THE INTERNET

The following supporting information may be found in the online version of this article:

Figure S1 Doppler waveforms from the celiac artery (A) and the splenic artery (B). Note the non-pulsatile echoesbelow the zeroline from the portal and splenic veins, respectively.

Figure S2 The left portal vein time-averaged maximum velocity (LPV-TAMXV) showing fitted 5th, 50th and 95th

centiles with 95% CIs based on 282 observations.

Figure S3 Ratio of the celiac and splenic artery pulsatility indices (CA-PI/SA-PI) showing fitted 5th, 50th and 95th

centiles with 95% CIs based on 427 observations.

Figure S4 Influence of general circulatory status, expressed by the umbilical artery (A) and middle cerebralartery (B) pulsatility indices, on local regulation, expressed by the splenic artery pulsatility index (SA-PI). Linesindicate < 10th centile (–ž–), 10–90th centile ( ) and > 90th centile (–°–) for the variables and the differencebetween the lines represents the effect, all being significant.

Table S1 Intra- and interobserver measurement variation of celiac and splenic artery flow velocities and pulsatilityindices based on 16 and 14 pairs of celiac artery observations, and 16 and 15 pairs of splenic artery observations,respectively.

Table S2 Reference ranges for the celiac artery time-averaged maximum velocity (in cm/s) based on 510observations in 161 low-risk pregnancies.

Table S3 Reference ranges for the splenic artery time-averaged maximum velocity (in cm/s) based on 521observations in 161 low-risk pregnancies.

Appendix S1 Terms for calculating conditional ranges for repeat measurements of the celiac (CA) and splenic (SA)artery pulsatility index (PI), peak systolic velocity (PSV) and time-averaged maximum velocity (TAMXV).


fetal celiac and splenic artery flow velocity and pulsatility index: longitudinal reference ranges...

Documents