In-utero evaluation of the fetal umbilical–portal venous system: two- and three-dimensional ultrasonic study

Authors

  • Z. Kivilevitch,

    1. Maccabi Health Services, Ultrasound Unit, The Negev Medical Center, Beer Sheba, Israel
    Search for more papers by this author
  • L. Gindes,

    1. Department of Obstetrics and Gynecology, The Chaim Sheba Medical Center, Tel-Hashomer, Ramat Gan and Sackler School of Medicine, Tel-Aviv University, Tel Aviv, Israel
    Search for more papers by this author
  • H. Deutsch,

    1. Department of Obstetrics and Gynecology, Kaplan Medical Center, Rehovot, Israel
    Search for more papers by this author
  • R. Achiron

    Corresponding author
    1. Department of Obstetrics and Gynecology, The Chaim Sheba Medical Center, Tel-Hashomer, Ramat Gan and Sackler School of Medicine, Tel-Aviv University, Tel Aviv, Israel
    • Ultrasound Unit, Department of Obstetrics & Gynecology, Chaim Sheba Medical Center, Tel Hashomer, Ramat Gan 52621, Israel
    Search for more papers by this author

Abstract

Objectives

To describe the normal anatomy of the fetal umbilical–portal venous system (UPVS) and to assess possible anatomical variants of the main portal vein (MPV) insertion into the portal sinus (PS).

Methods

This was a prospective cross-sectional study of low-risk patients between 14 and 36 weeks of gestation. Two- (2D) and three-dimensional (3D) ultrasound techniques combined with color and high-definition flow Doppler were used to evaluate the fetal UPVS. The standard transverse plane of the fetal upper abdomen, used for measuring the abdominal circumference, was taken in all cases as the point of reference. A longitudinal section was taken to identify the normal course of the umbilical vein and ductus venosus (DV). We performed offline analysis of all gray-scale and color Doppler 2D and 3D volume datasets.

Results

Two hundred and eight fetuses were included in the study. The umbilical vein was observed to course in a cephalad direction from its entry point into the fetal abdomen, joining the L-shaped PS, a confluence of vessels that is the main segment of the left portal vein (LPV). Three branches emerge from the LPV: two to the left, the inferior and superior branches, and one to the right, the medial branch. The main LPV then courses abruptly to the right. Following the emergence of the DV, the communication of the MPV with the LPV marks the point at which the vessel becomes the right portal vein (RPV), giving rise to its anterior and posterior branches. We were able to define three main variants of connection between the MPV and the PS. In 140 (67.3%) fetuses the MPV was connected to the LPV in an end-to-side T-shaped anastomosis, in 26 (12.5%) fetuses the MPV connected with a side-to-side X-shaped anastomosis and in 30 (14.4%) fetuses the two vessels ran in parallel with a short communicating segment, in an H-shaped anastomosis. In the remaining 12 (5.7%) cases classification into one of these three groups was not possible due to intermediate morphology.

Conclusions

Knowing the normal anatomy of the UPVS and being aware of the possible variants of the connection between the MPV and the PS is a fundamental requirement for accurate prenatal diagnosis of the anomalies of the fetal UPVS. Copyright © 2009 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

Assessment of the normal fetal umbilical–portal venous system (UPVS) has become an important part of antenatal fetal surveillance. Anomalies of the UPVS are associated with fetal chromosomal and structural anomalies, and Doppler blood flow of the ductus venosus (DV) has become a screening tool for Down syndrome in the first trimester1–4. Furthermore, recent studies have emphasized the need to evaluate liver blood flow perfusion in fetuses with intrauterine growth restriction5.

In humans, the venous perfusion of the fetal liver is unique since it is supplied by two embryologically and functionally different blood systems: the umbilical and the portal/vitelline systems. Between 5 and 10 weeks of gestation, the growing liver induces a network of anastomoses between the umbilical and vitelline systems, and increasing amounts of placental blood flow are shunted to the heart via this hepatic network. The intra- and extrahepatic portal venous system develops from the right vitelline vein. In the umbilical system, the right umbilical vein regresses and the remaining left umbilical vein joins the portal system directly, forming the umbilical–portal system. The DV, which emerges from the umbilical–portal system, provides oxygenated blood directly to the heart6, 7.

Although the normal anatomy of the UPVS in the human fetus has been described on pathological examination8 and in utero by three-dimensional (3D)-4D ultrasound9, the precise relationship between the umbilical vein and the portal system has not been described rigorously in-vivo. A recent study by Kessler et al.10 investigated the flow in the main portal vein (MPV) during the second half of human gestation, but anatomical variability of its junction with the portal sinus (PS) was not evaluated systematically. A literature search produced only one pathological study which described three variants of communication between the MPV and the PS, depending on the angle of the junction11. Precise knowledge of normal human fetal anatomy and its possible variants is a prerequisite for accurate prenatal diagnosis of vascular anomalies. We therefore aimed in this study to focus on the intrahepatic connections of the UPVS, and to describe for the first time the normal in-vivo anatomical variants.

Methods

A prospective, cross-sectional study of anatomically normal fetuses was conducted as part of routine antenatal care in a low-risk population. In most cases, examinations were performed during fetal anatomical sonography at 14–16 weeks and 19–24 weeks, or in the third trimester as part of fetal growth estimation.

We excluded fetuses with any abnormal sonographic findings, including ‘soft markers’ for aneuploidy in which normal karyotype was not ascertained and pregnancies complicated by maternal diseases that could affect fetal growth or with abnormal amniotic fluid volume whether or not this was associated with abnormal intrauterine growth.

Ultrasound examinations were performed with the Voluson 730 Expert or E8 machines (GE Medical Systems, Zipf, Austria) equipped with a 4–8-MHz transabdominal or a 5–9-MHz transvaginal probe, with a high-pass filter of 70 Hz. To avoid any bioeffect risk, the ALARA principle was adopted and, as recently recommended12, care was taken to keep mechanical and thermal indices below 1.1 and 0.9, respectively, when color Doppler was used. All women provided written and oral informed consent prior to the ultrasound examination and the study protocol was approved by the institutional review board.

Only cases with optimal visualization of the UPVS were included in the study. All evaluations were made in the standard transverse plane of the fetal upper abdomen, which is usually used for measurement of the abdominal circumference. The approach we adopted was from the right side of the fetus, with the stomach distal to the transducer, as a standard technique in all examinations, as this was determined to be the best approach with which to visualize the UPVS. This plane includes the stomach and the L-shaped PS, a confluence of vessels extending from the end of the umbilical vein which has been defined by Mavrides et al.8 as the vascular space extending from the point of origin of the inferior branch of the left portal vein (LPVi) to that of the right portal vein (RPV) (Figure 1a). From this point, a right downward sweep was performed in order to visualize the junction of the PS and the MPV, which runs from the left side between the stomach and the descending aorta. We tried to visualize the hepatic artery running close to the MPV, as a referral point, as shown in Figure 2b. The junction of the MPV with the PS was first identified by two-dimensional (2D) ultrasound. Thereafter, color Doppler with high-definition flow (HDF) was applied to achieve the best mode of visualization, and to verify the direction of flow (Figure 2a and b). The 3D technique was applied only in cases in which the PS and the MPV could not be visualized in the same plane with the other imaging modes. As the portal branches are situated on different anatomical levels, and convey a relatively low velocity blood flow, they are depicted poorly with 2D gray-scale imaging and even with color Doppler. 2D and 3D color HDF are essential techniques to optimize visualization of the LPV and RPV branching systems13. For 3D HDF we used a 30–35° angle sample volume, adjusting the balance between surface color mode and maximum transparent gray-scale mode, and with the lowest filter threshold needed to achieve optimal visualization of the vessels (Figure 3). According to the angle of the junction between the MPV and the PS, variants were defined. In order to evaluate the intrahepatic branches of the portal system, we adopted the Couinaud liver segmentation system14. A longitudinal section was additionally taken to identify the normal course of the umbilical vein and the DV.

Figure 1.

Ultrasound images in a 23-week fetus showing the normal intrahepatic umbilical vein (UV) connection to the left (LPV) and right (RPV) portal veins at the level at which the abdominal circumference is usually measured. (a) Transverse plane. (b) Corresponding section in the sagittal plane with the transverse plane of the section indicated by a dashed line. Note that neither the ductus venosus (DV) nor the main portal vein (MPV) can be visualized in the transverse plane, which passes between them and crosses the inferior vena cava (IVC) and the descending aorta (DA) posteriorly. LPVi, left portal vein inferior branch; LPVm, left portal vein medial branch; LPVs, left portal vein superior branch; PS, portal sinus; SP, spine; ST, stomach.

Figure 2.

The junction point between the main portal vein and the bifurcation of the right and left portal branches at the portal sinus in a 23-week fetus, shown without (a) and with (b,c) high definition flow. Images (a) and (b) show the transverse plane. The arrow indicates the hepatic artery. Image (c) is the corresponding section in the sagittal plane with the transverse oblique main portal vein (MPV) plane indicated by the dashed line. Note that in the sagittal plane, the intrahepatic umbilical vein segment is not included. ARPV, anterior branch of the right portal vein; DA, descending aorta; DV, ductus venosus; HA, hepatic artery; IVC, inferior vena cava; LPV, left portal vein; LPVi, left portal vein inferior branch; LPVm, left portal vein medial branch; PRPV, posterior branch of the right portal vein; RHV, right hepatic vein; RPV, right portal vein; ST, stomach; UV, umbilical vein.

Figure 3.

The portal vein in a 24-week fetus: normal intrahepatic vascular anatomy, shown in the transverse abdominal plane (a). The high definition three-dimensional flow technique made it possible to visualize the main portal vein (MPV) and the branches simultaneously, while this was not possible on two-dimensional imaging (b–d). ARPV, anterior branch of the right portal vein; LPV, left portal vein; LPVi, left portal vein inferior branch; LPVm, left portal vein medial branch; LPVs, left portal vein superior branch; PRPV, posterior branch of the right portal vein; RPV, right portal vein; SP, spine; UV, umbilical vein.

All samples were performed by a single operator (K.Z.). The kappa index of agreement between two operators (K.Z. and A.R.) was assessed for the first 50 cases, and calculated according to the formula: K = ∑a − ∑ef/N − ∑ef, where Σa is the sum of agreement, Σef is the sum of expected frequency and N is the total number. The other statistical analyses (mean gestational age at examination, 95% CI, range and normality of distribution (Kolmogorov–Smirnoff)) were assessed by MedCalc (Version 9.2.1.0, MedCalc Software bvba, Mariakerke, Belgium) and Excel 2007 (Microsoft Corp., Redmond, WA, USA) statistical software.

Results

During the study we examined 208 fetuses. The mean gestational age at examination was 25.1 ( ± 5.3 SD; 95% CI, 24.4–25.8; range, 14–39) weeks. The age distribution was not normal (Kolmogorov–Smirnoff, P < 0.001), as 70.8% of the study group was recruited at 20–26 weeks of gestation, during second-trimester routine anatomy screening. We observed that in the longitudinal plane, the umbilical vein courses in a cephalad direction and enters the liver, where it connects with the portal system. In the left intersegmental fissure of the liver it joins the LPV, which then courses abruptly to the right, creating the L-shaped segment known as the PS. The MPV enters obliquely the main lobar fissure from the left. The site of connection of the MPV to the PS represents the anatomical point of division between its right and left branches, and is situated inferiorly (Figure 2c) and to the right with respect to the DV origin. The RPV bifurcates into two major ramifications: the anterior and posterior branches, at a distance from the MPV–PS junction that varies between fetuses. Three branches emerge from the LPV: two to the left, the inferior (LPVi) and superior (LPVs) branches, and one to the right, the medial branch (LPVm), at approximately the level of origin of the LPVi (Figure 3). During the study period, in only one (0.4%) case were we unable to detect the L-shaped segment of the LPV, indicating absence of the horizontal part of the LPV. In this case, the DV emerged from the RPV rather than from the PS (Figure 4).

Figure 4.

Umbilicoportal vein variant in a 23-week fetus (a,b). The typical L shape of the left portal vein cannot be identified (a) and the ductus venosus has a different emergence (b, arrow), in comparison to the normal course (c, arrow). Ao, aorta; mpv, main portal vein; RPV, right portal vein; St, stomach; UV, umbilical vein.

At the junction point of the MPV and the PS (Figure 2), we noticed that their angle of communication ranged continuously from perpendicular to almost completely parallel. Accordingly, three main types of connection between the MPV and PS could be characterized. The most common type, observed in 140 (67.3%) fetuses, showed a T shape on imaging, with end-to-side anastomosis between the MPV and the PS (Figure 5). This type of connection showed a large range of angle of insertion and distance from the origin of the posterior branch of the RPV (PRPV). It ranged from a vertical T-shaped insertion into the PS, distant from the origin of the PRPV (Figure 5a), to a more acute angle of insertion and shorter distance from the origin (Figure 5b and 5c), to finally a cross-shaped structure formed by four vessels: the MPV, the LPV and the two branches, anterior and posterior, of the RPV (Figure 5d). This was the point at which the morphology became borderline, being intermediate between the T-shaped connection and the second type of connection between the MPV and PS: the X-shaped communication (Figure 6), which presented in 26 fetuses (12.5%). This junction was characterized by a side-to-side anastomosis between the MPV and PS, which ran almost parallel. The posterior branch of the RPV was continuous with the MPV, and was connected by an opening of variable width (Figure 6a and b). In some cases a detachment between MPV and LPV was observed, representing an intermediate stage between this second type and the third type of connection between the MPV and PS, which showed an H-shape on imaging and was observed in 30 (14.4%) fetuses. In this type of connection, the MPV and the PRPV were separated from the RPV by a thin vessel (Figure 7) and we observed a varying degree of distance between the LPV/ARPV and the MPV/PRPV complexes. In the most extreme case they could not be visualized together on the same plane using 2D gray-scale imaging and only the use of a 3D color HDF multiplanar technique could demonstrate the thin vascular connection between them (Figure 7c). In our series, classification of the type of connection between the MPV and PS was not possible in 12 (5.6%) cases due mainly to intermediate morphology. Eight were between Types T and X, and four were between Types X and H. The kappa index of agreement between the two operators was 0.89.

Figure 5.

Variants of the main portal vein (MPV)–portal sinus (PS) end-to-side connection at 24 weeks. (a) T-shaped anastomosis. (b) Variable distance from the origin of the posterior branch of the right portal vein (PRPV) was noted; in some cases the left portal vein (LPV) and the right portal vein (RPV) branches originated directly from the MPV in a trifurcated form (c). (d) The more acute angle of insertion is an intermediate form between end-to-side and side-to-side connection types. ARPV, anterior branch of the right portal vein; DA, descending aorta; IVC, inferior vena cava; SP, spine; ST, stomach; UV, umbilical vein.

Figure 6.

Variants of the main portal vein (MPV)–portal sinus (PS) side-to-side connection at 24 weeks: X-shaped anastomosis. In (a) and (b) the variable connection width between the MPV/posterior branch of the right portal vein (PRPV) complex and the left portal vein (LPV)/anterior branch of the right portal vein (ARPV) complex is shown, with the almost complete detachment representing an intermediate stage between the X and H shapes. (c) Three-dimensional high-definition flow reconstruction of the image in part (a). DA, descending aorta; IVC, inferior vena cava; SP, spine; ST, stomach; UV, umbilical vein.

Figure 7.

Variants of the main portal vein (MPV)–portal sinus (PS) H-shaped anastomosis at 24 weeks. The MPV and posterior branch of the right portal vein (PRPV) are separated from the left portal vein (LPV) and anterior branch of the right portal vein (ARPV), with small bridging vessels communicating between them (a and b). In (c) is a case in which the LPV/ARPV and MPV/PRPV complexes were so distant from one another that they could be visualized only by using the three-dimensional multiplanar color high-definition flow mode. RPV, right portal vein; SP, spine; ST, stomach; UV, umbilical vein.

Discussion

In this study, we examined in-vivo the precise communication between the MPV and the PS. The UPVS is a complex vascular network that supplies the liver as well as the heart of the developing fetus. Although this network has been evaluated and named from the early years of fetal sonographic imaging15, there is still some confusing, if not contradictory, nomenclature regarding certain segments, such as the PS (or ‘pars transversa’), and even such as regarding the exact point of contact between the portal and umbilical systems16.

We chose to adopt the anatomical nomenclature proposed by Mavrides et al.8, and to use the term ‘portal sinus’ for the L-shaped umbilical portion of the LPV, rather than the term ‘pars transvera’ (transverse portion), as suggested by Chinn et al.17, 27 years ago. The main reason for this was our ability, using 2D and 3D HDF, to visualize with ease the LPVi branch as a landmark for the origin of the PS. Moreover, this technique made it possible for us to visualize simultaneously the MPV and its branches, which was not possible on the initial 2D examination (Figure 3b–d).

The importance of our study is the fact that we were able to describe precisely the various anatomical connections between the MPV and the PS in a large number of fetuses during gestation. Our study confirms previous pioneer fetal pathological observations published by Czubalski and Aleksandrowicz11 regarding the three types of communication. However, these three types of variant are only landmarks within a continuum of forms of communication, ranging from a T-shaped angle of insertion to almost separate pairs (LPV/ARPV and MPV/PRPV) of vessels. Knowledge of these anatomical variants is paramount in the diagnosis of anomalies of the portal venous system, such as total and partial portal vein agenesis. Although data on the diagnosis of complete fetal portal vein agenesis is scanty in the literature, sporadic cases have been reported regarding the in-utero diagnosis of such congenital anomalies2, 9, 18–20.

In order to understand the intrauterine anatomy of the fetal portal veins we adopted the concept of the functional anatomy of the liver as described by Couinaud14, which is based on the distribution of the portal vein branches and, as such, is useful in studying the anatomy of the fetal portal veins. According to this, the liver is formed by two main lobes: right and left, each containing four segments. The right lobe is divided into anterior and posterior (paramedian and lateral, respectively, according to Couinaud14) and the left into medial and lateral lobes, each of which is further subdivided into superior and inferior lobes. The segments are numbered counter-clockwise: 1–4 in the left lobe and 5–8 in the right. The left lobe contains the lateral (Segment 2), the median, divided into two segments, one to the right and one to the left of the umbilical fissure (Segments 3 and 4, respectively), and the posterior caudal lobe (Segment 1). The right lobe contains the anterior-inferior, posterior-inferior, posterior-superior, and anterior-superior segments (Segments 5, 6, 7 and 8, respectively) (Figure 8a). However, functional segmentation remains controversial among hepatopancreatobiliary surgeons21.

Figure 8.

Schematic diagram of the normal portal venous distribution with respect to the functional liver segmentation of Couinaud14 (Segments 1–8). (a) Type 1: T-shaped communication between main portal vein (MPV) and the portal sinus (PS). (b) Type 3: H-shaped communication between MPV and PS. ARPV, anterior branch of the right portal vein; LPV, left portal vein; LPVc, left portal vein caudate branch; LPVi, left portal vein inferior branch; LPVm, left portal vein medial branch; LPVs, left portal vein superior branch; PRPV, posterior branch of the right portal vein; RPV, right portal vein; UV, umbilical vein.

Lafortune et al.22 studied the sonographic appearance of the portal vein hepatic segmentation in adults. They described the ‘two H’ configuration, one for each lobe, which was concordant with Couinaud's classification. We focused on anatomical variants with respect to the communication between the MPV and the PS, which is the crossing point of three vessels: the UV and the left and right portal branches. Let us now attempt to extrapolate adult sonographic and computed tomographic anatomical studies of the portal intrahepatic architecture23–26 to the fetus. The first-degree bifurcation of the MPV is into the RPV and LPV, supplying the right and left lobes, respectively. On the right side, the RPV is further divided into the posterior branch that supplies, according to Lafortune et al.22, the posterior-inferior and posterior-superior segments (Couinaud's Segments 6 and 7), and the anterior branch that supplies the anterior-inferior and anterior-superior segments (Couinaud's Segments 5 and 8). On the left lobe the lateral segment (Couinaud's Segment 2) is supplied by the LPVs, and the LPVi supplies the left (para)median (Couinaud's Segment 3). On the right side of the umbilicoportal tract, the LPVm supplies the right (para)median segment (Couinaud's Segment 4) (Figures 3a and 8). Regarding the MPV we depicted a wide range of intermediate forms with respect to its angle of insertion and distance from the PRPV. A detachment between the MPV and the RPV was also observed, being intermediate in form between the X-shaped communication and the H-shaped one. This variant created a new physiological division, dividing the right lobe into the posterior part (Couinaud's Segments 6 and 7), supplied by a poorly oxygenated portal vein, and the anterior part (Couinaud's Segments 5 and 8), supplied almost exclusively, via the LPV, with the rich nutrients and oxygenated blood from the placenta (Figure 8b). These variants may have implications on the intrahepatic hemodynamics and fetal growth pattern, which may warrant further investigation27, 28.

In conclusion, understanding the normal appearance of the fetal portal system is a fundamental requirement for the accurate prenatal diagnosis of anomalies of the fetal venous system in general, and of the portal system in particular. Our study focused on the normal morphological variants of the fetal intrahepatic portal system, and their relationship to the liver functional segmentation classification used clinically in adults. This information may serve as a basic anatomical platform when evaluating the fetal portal system, especially when one must bear in mind the relatively high incidence of variants that can be seen in this important vascular system.

Ancillary