Since Abernethy first discovered the congenital portosystemic shunt (CPSS) in 1793 [7], an increasing number of studies have been conducted on this disease [8,9,10,11]. According to the location of the shunt between the portal and systemic veins, the CPSS can be divided into intrahepatic and extrahepatic types and various subtypes. The previous description of foetal CPSS also used the adult classification until Achiron proposed the foetal umbilical–portal–systemic shunt (UPSVS) classification [12], including type I umbilical–systemic shunt, type II ductus venosus–systemic shunt, and type III portal–systemic shunt. Type III is further divided into two subtypes: type IIIa, intrahepatic portal–systemic shunt, and type IIIb, extrahepatic portal–systemic shunt. This classification is currently the only foetal type of CPSS that can better reflect the unique blood circulation characteristics of the foetal period and is more suitable for the diagnosis of foetal CPSS. However, prenatal ultrasound manifestations of type IIIa are diverse, and no further elaboration has been made regarding the characteristics of this type. It must be emphasised that prenatal classification is not the same as postnatal classification. Furthermore, for optimal treatment and management after birth, various imaging modalities, including CT angiography and Doppler ultrasound, should be considered to precisely define the involved vessels, type of shunt, and presence of aneurysm/dilation [13].

The direct manifestation of the congenital IHPSS is an abnormal connection between the hepatic and intrahepatic portal veins. Due to the limitations of imaging conditions, prenatal classification of IHPSS based on the Park classification is unreliable [10]. However, it is not difficult to distinguish single or multiple shunts by prenatal ultrasound. The abnormal connection was often close to the liver capsule (78.9%, 15/19), indicating that the shunts were mostly located in the peripheral branches of the hepatic and intrahepatic portal veins. The morphology of the abnormal connection was diverse: 33.3% (4/12) of the single-shunt group and 28.6% (2/7) of the multiple-shunt group showed cystic dilatation at the shunt. In the multiple-shunt group, 57.1% had irregular shunts and 42.9% had regular shunts.

In our experience, most IHPSS cases diagnosed in the foetal period, whether single or multiple shunts, mostly originate in the left portal branch on the portal side. In the single-shunt group, 100% (12/12) originated from the left portal branch, of which 75% (9/12) were in the LPVi, 16.7% (2/12) in the LPVm, and 8.3% (1/12) in the LPVs (Fig. 2). In the multiple-shunt group, excluding one patient with both intrahepatic and extrahepatic shunts, this patient had abnormal communications between the left portal vein and the hepatic vein and between the right portal vein and the hepatic vein, the remaining shunts all originated from the left portal vein. On the hepatic vein side, the most common location of IHPSS is the left hepatic vein (LHV). In the single-shunt group, 75% of the shunts were located in the LHV, 16.7% in the middle hepatic vein (MHV), and 8.3% in the right hepatic vein (RHV). In the multiple-shunt group, 85.7% (6/7) of the cases included the LHV shunts, 71.4% (5/7) included MHV shunts, and 14.3% (1/7) included RHV shunts.

We found that the left lobe of the IHPSS diagnosed in the foetal period was more common than the right lobe, whether on the portal or hepatic venous side. This conclusion is consistent with that reported by Kivilevitch et al. [4]. In their cases, except for one foetus in which the left portal vein could not be identified, the starting position of the shunt in all other cases was located in the left portal vein. Cytter-Kuint et al. studied IHPSS in 15 children, of which 14 shunts were located between the LPV and hepatic vein, and only one case was located between the RPV and RHV [14]. However, none of these studies have discussed this feature further.

The hepatic primordium emerges in the fourth week of gestation and is in contact with the vitelline and umbilical venous systems. Eventually, liver blood vessels are formed via angiogenesis, vasculogenesis, and vascular differentiation [15]. The segments of the vitelline veins located between the subdiaphragmatic and subhepatic anastomoses are interrupted by the developing hepatic trabeculae and resolve into a plexus of small vessels connected to the hepatic sinusoids. The hepatic afferent veins (venæ advehentes) convey blood from the subhepatic anastomosis to the sinusoidal plexus, while the hepatic efferent veins (venæ revehentes) drain blood from the sinusoidal plexus to the subdiaphragmatic anastomosis [16,17,18]. The foetal umbilical vein completely supplies the left portal vein of the foetal liver, whereas the right portal vein of the foetus is supplied by both the umbilical vein and the main portal vein. The blood flow in the left portal vein of the foetal liver was greater than that in the right portal vein [19, 20]. IHPSS is a vascular malformation caused by incomplete vascular remodelling of the vitelline vein. We speculate that in the process of gradual remodelling of the small vascular plexus connected to the sinus plexus into the peripheral branches of the portal vein and hepatic vein, the peripheral branch of the LPV has greater blood flow because the LPV connects to the umbilical vein, making it easier to “maintain” the abnormal connection with the peripheral branch of the hepatic vein. That is, increased blood flow facilitates the formation of an IHPSS. This may be the reason the IHPSS we observed during the foetal period was more common in the left branch of the portal vein. In the single-shunt group, the closer the portal vein branch was to the umbilical vein, the more shunts were observed (LPVi > LPVm > LPVs). The spontaneous closure of IHPSS diagnosed prenatally after birth may be related to the closure of the umbilical vein and the decrease in portal venous blood flow after birth, which, on the other hand, indicates that low blood flow is not conducive to the “maintenance” of the abnormal shunt.

The mean gestational age at diagnosis in our cases was 33.8 ± 4.5 weeks (25–40 weeks), which is consistent with the conclusions of other researchers [4, 21] that IHPSS is mostly diagnosed in the late second and third trimesters (Fig. 4). The mean gestational age at diagnosis was 35.6 ± 4.9 weeks in the multiple-shunt group and 32.8 ± 4.1 weeks in the single-shunt group. Kivilevitch Z pointed out that the reasons leading to late diagnosis include the relatively small amount and low velocity of the blood flowing in these venous systems in the first trimester, lack of disease awareness by examiners, and acquired late formation of the shunts [4]. They believe that the two cavernous shunts in their cases were caused by vascular thrombotic events; thus, the shunts were discovered in the third trimester. In our case, no similar evidence of acquired shunts (e.g. intrahepatic calcifications that appeared until mid-pregnancy) was observed.

The gestational age at diagnosis of nineteen cases. Note that all cases were diagnosed in the late second and third trimesters. GA: gestational age at diagnosis; SS: single shunt; MS: multiple shunts

The most common intrauterine concomitant abnormality in IHPSS was FGR, with an incidence of 47.4%. No significant difference was observed in the incidence of FGR between the single and multiple groups (Table 2). In foetal sheep, experimental reduction of umbilical vein hepatic perfusion results in the decreased foetal liver synthesis of insulin-like growth factors, impairing foetal growth [22]. Therefore, abnormal shunting of the left portal vein and the hepatic vein leads to a decrease in blood perfusion in the liver, which may cause FGR in IHPSS.

Another common intrauterine abnormality is cardiac enlargement, with an incidence rate of 21.1%. Abnormal shunts lead to increased cardiac preload and, thus, enlargement of the heart, which suggests that attention should be given to evaluating cardiac function in patients with IHPSS. Some researchers have proposed the potential benefit of f-TAPSE (tricuspid annular plane contraction drift) for the longitudinal monitoring of growth-restricted foetuses diagnosed concomitantly with cardiomegaly and CPSS [23].

In our case, there was no significant difference in the incidence between boys and girls, and in cases of known sex, the male-to-female ratio was 1:1 for both the single and multiple groups. There were 14 live births (73.7%, 14/19), with 75% in the single-shunt group and 71.4% in the multiple-shunt group. Thirteen cases were followed up for the mode of delivery; caesarean section accounted for the vast majority (76.9%, 10/13), and the average gestational week of delivery was 38 weeks (35–40 weeks). Higher caesarean section rates may be associated with foetal FGR.

The postnatal manifestations of IHPSS vary [24]; moreover, the most common biochemical abnormality in our case was an increase in GGT and bilirubin, which was more common in the multiple-shunt group of live births (80%, 4/5), probably resulting from liver ischaemia due to vascular deprivation. In addition, it should be noted that there were four cases of neonatal hypoglycaemia, all of which were in the single-shunt group. All were prenatally complicated with FGR, including two cases of low birth weight infants and one case of very low birth weight infants. IHPSS can lead to neonatal hypergalactosaemia. Unfortunately, serum galactose was not detected in these four cases; therefore, it was unclear whether hypoglycaemia was caused by transient hypergalactosaemia. There have also been reports of neonatal hypoglycaemia; in a study of the clinical features of congenital portosystemic shunts in the neonatal period, 40% (4/10) of patients experienced severe neonatal hypoglycaemia, and all 4 patients were diagnosed with intrahepatic portosystemic shunts [24]. In our study, hyperammonaemia occurred in 6 neonates, and the blood ammonia level decreased during follow-up observation without causing hepatic encephalopathy. Abdominal ultrasonography in another 2 neonates revealed transient hyperechoic nodules in the liver, which spontaneously subsided during the follow-up process. One of them developed multiple cutaneous haemangiomas, which also subsided spontaneously. Interestingly, both cases were in the multiple-shunt group.

Benign and malignant liver tumours associated with congenital portosystemic shunts have been reported, including focal nodular hyperplasia, hepatocellular adenoma, and hepatoblastoma; however, malignant tumours are only observed in extrahepatic CPSS [25].

Among the cases we followed up, two children underwent laparoscopy and open surgery for shunt ligation, with a good prognosis and no severe postoperative complications. The 10 children with the remaining follow-up adopted conservative treatment, and all of them had spontaneous closure. The average closure time was 8.1 months (1–28 months), which is similar to the conclusion of Achiron [12]. In their cases, the mean interval between delivery and spontaneous closure in the IHPSS was 8.7 months. A management strategy for CPSS proposed that asymptomatic patients with portal–hepatic shunts should be followed up to 2 years old without immediate surgical treatment [2]. Endovascular closure is surgical and the treatment of choice for children who are symptomatic or whose shunt has not closed spontaneously after 2 years of age [26].

Two cases of labour induction (single-shunt group) were found to have copy number variations in the specimens of induction, which were located on chromosomes 7 and X, respectively. The microdeletion on chromosome 7 completely covers the region reported in “Williams-Beuren syndrome” (OMIM 194050). Microduplication located on the short arm of the X chromosome is a variant of unknown clinical significance and does not contain a pathogenic gene included in OMIM. This region is related to the reported region of the “Chromosome Xp11.23-p11.22 duplication syndrome” (OMIM 300,801). Karyotype analysis of the peripheral blood chromosome of one live infant (multiple-shunt group) showed inter-arm inversion of chromosome 9. The inter-arm inversion of chromosome 9 is generally considered a polymorphism; however, some studies have shown that it can have adverse genetic effects. No reports of copy number variations or inter-arm inversions have been found to be associated with IHPSS. Since not all of our cases were examined by whole-genome sequencing, whether these copy number variations or inversions between chromosomal arms were detected incidentally or were associated with the pathogenesis of IHPSS requires further investigation.

Our main limitation was the small number of cases, which will require further large-scale multicentre studies in the future.