Secondary Logo

Journal Logo

Congenital Vascular Malformations of the Liver

An Association With Trisomy 21

Burdall, Oliver C.*; Grammatikopoulos, Tassos; Sellars, Maria; Hadzic, Nedim; Davenport, Mark*

Journal of Pediatric Gastroenterology and Nutrition: December 2016 - Volume 63 - Issue 6 - p e141–e146
doi: 10.1097/MPG.0000000000001405
Original Articles: Hepatology

ABSTRACT A link between congenital vascular malformation (CVM) of the liver and trisomy 21 has been suggested. We reviewed all children with trisomy 21 referred for investigation to a specialist pediatric hepatobiliary unit (1985–2015). Forty-five children with trisomy 21 were identified; 7 (15%) had a defined CVMs (4 girls). All such infants were also diagnosed with a range of cardiac defects. CVMs were divided according to the nature of the vascular connection. Group (i) (n = 3): Abnormal venovenous anomaly. This included portocaval shunt and patent ductus venosus (n = 2). Group (ii) (n = 4): Involvement of all 3 vascular systems. Two infants had arterioportal hypertension caused by hepatic arteries feeding into a left portal vein aneurysm within the umbilical fissure. Two infants had more complex hepatic artery to hepatic vein shunts developing early cardiorespiratory failure with progressive jaundice. Our series shows a clear association between cardiac anomalies and CVM in children with trisomy 21.

*Department of Paediatric Surgery

Department of Paediatric Hepatology

Department of Radiology, Kings College Hospital, London, UK.

Address correspondence and reprint requests to Prof Mark Davenport, ChM, Department of Paediatric Surgery, Kings College Hospital, London, SE5 9RS, UK (e-mail:

Received 4 March, 2016

Accepted 30 August, 2016

The authors report no conflicts of interest.

What Is Known

  • Congenital cardiac anomalies are a common feature in children with trisomy 21 (Down syndrome).

What Is New

  • It strongly suggests a relation between trisomy 21 (Down syndrome), cardiac anomalies, and a range of congenital vascular malformations of the liver.
  • A novel and important association not previously described is the combination of left portal venous aneurysm and arterioportal hypertension.

Congenital vascular malformations (CVMs) of the liver can be defined functionally by the degree of interconnection between the hepatic arterial and portal venous network and systemic venous drainage (1,2). The clinical features and mode of presentations depend on the exact nature of the connection and its sequelae. Routine use of abdominal ultrasound (US) (including Doppler flow studies) since the 1990s has allowed detection of such anomalies at a much earlier stage (3).

There is an association between trisomy 21 and congenital cardiovascular anomalies (4–6), and intra-abdominal malformations such as duodenal atresia (7–9).

An association of CVM (left portal vein to inferior vena cava [IVC]) with trisomy 21 was first reported by Kieran et al (10). Since then there have been 6 case reports (10–15) and 3 small case series (3,16,17). The largest of these is limited to a description of the clinical features of 3 children (16). The aim of the present study was to review our experience and management of 7 children with trisomy 21 and various types of CVM of the liver, suggest a classification and review the literature in this context.

Back to Top | Article Outline


The Paediatric Liver Service at King's College Hospital is one of only 3 specialized units within the UK for the investigation and treatment of liver disorders in childhood. We retrospectively reviewed all children held on a departmental database with trisomy 21 who had been referred for further investigation of liver pathology between 1985 and 2015.

We also carried out a review of the English language literature previously published through searches of the PubMed, Medline, and the Excerpta Medica Database (Embase).

We classified the CVM into 2 basic types.

  • Group (i)—no evidence of hepatic arterial involvement and therefore simply a connection between the portal venous and systemic venous systems. These also included those with a patent ductus venosus (PDV).
  • Group (ii)—involvement of all 3 vascular systems—hepatic arterial and either/or/both portal venous and systemic venous systems. Both arterioportal hypertension and high arteriovenous flow with the potential for cardiac failure were possible.

One child (case 3) in group (i) has previously been reported with an unusual presentation of hyperinsulinemic hypoglycemic episodes (17). Data are presented as median with ranges, unless stated otherwise.

Back to Top | Article Outline


Forty-five children with trisomy 21 were identified. CVMs of the liver were identified in 7 (16%) of these children (4 girls). The diagnosis of trisomy 21 (Down syndrome) was confirmed by karyotype in all.

Median gestational age was 35 years (32–42) weeks and median birth weight 2.32 (1.04–3.68) kg. All of the infants identified were also diagnosed with a range of midline cardiac defects (Tables 1 and 2). Three of these required surgical correction, including an atrioventricular septal defect repair (case 1), pulmonary artery banding (case 4), and patent ductus arteriosus ligations for patients cases 6 and 7. Only 1 infant was diagnosed antenatally (case 5).





We then divided the group into portal venous-systemic venous shunt (n = 3) (Table 1) and (ii) arteriovenous or arterioportal shunt (n = 4) (Table 2).

Clinical features at presentation in group (i) included signs of heart failure, jaundice, and in 1, previously described case, hyperinsulinemic hypoglycemia, whereas the clinical features of group (ii) included heart failure, recurrent gastrointestinal bleeding, poor feeding/failure to thrive, thrombocytopenia, coagulopathy, and jaundice (Tables 1 and 2).

Median serum bilirubin was the same in both groups (70 [12–107] μmol/L vs 70 [3–129] μmol/L). Random serum ammonia was measured in 5 of the 7 children, with a median of 111 (47–121) μmol/L (normal range 15–50 μmol/L). There were no clinical signs of encephalopathy in any of the children.

Back to Top | Article Outline

Clinical Course

Group (i) (n = 3)

Two children had shunts from left portal vein to the hepatic vein confluence in keeping with a PDV.

The PDV in case 1 was confirmed to have resolved by follow-up US scans at 1 year. Case 2 died at the age of 3 months due to overwhelming sepsis and multiorgan failure, neither of which was thought to be related to their PDV.

The third child (case 3), who had undergone uncomplicated surgical repair of an esophageal atresia shortly after birth, was readmitted at the age of 21 months after 2 hypoglycemic seizures. An US scan at this point identified a venous shunt between the confluence of the portal vein and the IVC (Abernethy malformation) (1,18,19). In the absence of other defined cause for her hypoglycemia, and the persisting need for parenteral nutrition she underwent surgical closure of the portosystemic shunt at 2 years. This normalized blood glucose postoperatively and parenteral nutrition was discontinued over a period of 2 months. She remains well on a normal diet at 5 years of age.

Back to Top | Article Outline

Group (ii) (n = 4)

The CVMs in this group were much more variable and included arterioportal fistula (n = 2) and complex hepatic artery to hepatic vein shunts (n = 2).

Infants 4 and 5 had a similar arterioportal connection with large hepatic arteries feeding a portal vein aneurysm located within the umbilical fissure. This caused arterialization of their portal venous systems with at times reversal of flow in the portal vein synchronous with systole. Neither of these cases appeared to have a significant shunt into the hepatic veins from the portal venous system. Both presented early, 1 case 5) with duodenal atresia requiring reconstructive surgery and the other (case 4) with failure to establish enteral nutrition and later recurrent gastrointestinal bleeding. She also developed heart failure due to significant cardiac anomalies (ASD, ventricular septal defect, patent ductus arteriosus) and following failure of medical therapy required pulmonary artery banding. Liver investigations included hepatic angiography, which delineated the vascular malformation. In case 4, we attempted embolization of the feeding arteries during 2 sessions at 12 months. Neither was clinically or radiologically successful and she then underwent laparotomy at 2 years. There was hypertrophy of both left and right hepatic arteries with a thick-walled aneurysmal dilatation measuring 2 cm in diameter within the umbilical fissure. This was dissected from within the hepatic parenchyma, disconnected from its arterial inflow and excised. US scans at 3 months postoperatively showed tortuous collateral veins in keeping with thrombosis of the portal vein and she subsequently developed splenomegaly, with evidence of ongoing gastrointestinal bleeding. Recurrent episodes of upper gastrointestinal bleeding requiring transfusions continued until the age of 6 years. Monthly depot octreotide injections were started at this point with complete resolution of bleeding and need for transfusion (20).

In case 5 we opted for early surgery without an attempt at embolization. Initially hepatic artery ligation was performed. Because of minimal therapeutic effect, resection of the left lateral liver segments II and III together with the portal vein aneurysm was carried out. He developed multiple jejunal perforations requiring further laparotomy and a period of stomal drainage before intestinal continuity was re-established. Currently, aged 5 years, he is asymptomatic and fully enterally fed. He has a normal portal venous system on US scan.

The remaining 2 infants (cases 6 and 7) developed early cardiorespiratory failure requiring ventilation from birth and day 5 of life, respectively. They also showed progressive signs of conjugated hyperbilirubinemia, coagulopathy, and thrombocytopenia requiring blood product correction. Infant 6 had an antenatally diagnosed arteriovenous malformation (AVM) or CVM in the left lobe of the liver at 36 weeks. The nature of the CVM was delineated using contrast computed tomography scan and Doppler US. It consisted of a left-sided arteriovenous malformation with high flow. The lesion measured 35 mm in diameter with a dilated left hepatic vein. The CVM in case 7 was more complicated with additional evidence of shunting into the portal venous system through a complex fistula between the left hepatic artery and portal vein with an enlarged ductus venosus.

There was no response to maximal medical therapy including digoxin, inotropic support, and diuretics; and open surgery was performed in both at 5 and 3 months, respectively. In case 6, this consisted of an initial attempt at reducing arterial input by ligation of the common hepatic artery. This was then followed 5 weeks later with resection of the CVM by left lateral segmentectomy, due to further clinical deterioration including the development of sepsis, high output cardiac failure, and pulmonary hypertension requiring nitric oxide, sildenafil, and high pressure ventilation. His ventilation deteriorated further and he died at 24 hours postoperatively. A left lateral segmentectomy was performed ab initio in case 7. She remained ventilated with no evident improvement in her clinical condition and died 1 month later.

Back to Top | Article Outline

Literature Search

We identified 14 cases from the literature, 1 of which (18) is included in our case series and so excluded from this analysis, which is detailed in Table 3. From their available clinical descriptions we divided them using our classification into group (i) (n = 7) and (ii) (n = 6) (Figs. 1 and 2).







Of the venovenous malformations in group (i); 4 were a direct portocaval shunt with the remaining 3 having communications between the left hepatic and left portal vein (10,12,13,15–17). Of the portocaval shunts; 1 was diagnosed antenatally with a dilated umbilical vein communicating directly with an IVC which was confirmed as a connection between the portal vein and IVC on postnatal US. The management was not described. The other was diagnosed following referral at 1 month of age due to suspected heart defect with tachypnea, mild conjugated hyperbilirubinemia, poor feeding, and dusky appearance. This patient died at the age of 52 days due to cardiac failure and Staphylococcus aureus sepsis without intervention (17). Of the portovenous shunts, 1 was successfully treated with coil embolization due to increasing “cardiac failure” (15) and the others resolved with conservative management (n = 2) (11,13,16). The decisions for conservative management were based on very low birth weight as a contraindication for invasive intervention and parental refusal of treatment (11,13,16).

The cases within group (ii) of our classification included 2 children with what were described as “portal vein sinuses” and communication of this with hepatic arteries. Doppler US identified both communication between left hepatic artery and intrahepatic IVC, via a “tangle” of small arterial vessels, and a superior mesenteric vein that emptied into blind-ending portal segment, which communicated with IVC. This case was treated conservatively with dietary restriction (not specified) only (14). The other child, who was found to have 2 separate fistulae between the 2 vessels, underwent successful coil embolization (11).

Golewale et al (16) described 2 cases including portal vein aneurysm fed by hepatic arteries in a child with multiple cardiac anomalies who ultimately died despite both surgical ligation and embolization of the feeder vessels, with the latter triggering initial symptomatic improvement and a more complicated connection between phrenic artery, vein, and portal vein who survived after embolization of the left phrenic vein. Hartung et al (3) described 2 still births resulting from arteriovenous shunts via abnormal vessels. In the first case a US at 14 weeks, performed due to fetal edema, showed a large vessel arising directly from the aorta, coursing through the liver, then spiraling around and eventually entering the intra-abdominal umbilical vein. In the other case the pregnancy was electively terminated (3).

Back to Top | Article Outline


We describe the largest series of a multiplicity of CVM in infants with trisomy 21. Actual data on prevalence in the normal population are not available. One estimate of 1 in 30,000 births for one of these (congenital portocaval shunts) appears in a recent review (19) and even with referral bias this strongly suggests that children with trisomy 21 are indeed susceptible to CVM of the liver. Although no particular malformation predominated and we were able to describe examples of most CVM; some appeared to be typical of this association (eg, arterioportal hypertension secondary to a left portal vein aneurysm [aka “sinus”] fed by hepatic arteries).

To date, there is no uniformly accepted classification of CVM of the liver and most reports are pragmatic in describing connections and their functional consequences. For example, the simplest situation of an abnormal congenital portosystemic venous connection has several classifications (1,2,20,21) with the latest review by Guerin et al (2) dividing them on the basis of side (right or left) and nature of connecting veins from either portal or systemic origin. We describe 1 child with a congenital portosystemic venous shunt and that was a relatively small side-to-side shunt (ie, type II). She, however, was clearly unusual in developing persisting symptomatic hypoglycemic episodes, which resolved after ligation suggesting that her underlying Down syndrome also contributed to her metabolic symptoms (18). So, even children with absolutely no intrahepatic portal venous flow due to end-to-side shunts manage to maintain their blood glucose without problems (2). The other, much commoner example of portosystemic venous connection is that of a PDV. Mean time to anatomical duct closure is usually described as between 4 and 7 days dependent upon gestational age though exceptions are not uncommon (22,23). Most are entirely asymptomatic and were incidental findings upon investigation of persisting conjugated jaundice. In contradistinction to the other types of portosystemic venous connection most will close spontaneously and almost never require therapeutic closure. Indeed this occurred in 2 of our cases demonstrably by their first birthday and 1 in the literature group (13).

The classification and pathology of congenital AVMs are also contentious, overlapping with that of “congenital” or “infantile” hemangiomas of the liver (24). Discrimination is not always straightforward in those with an isolated tumor-like malformation (eg, case 6). There was no evidence of any diminution of A-V shunting over time and all these came to surgical intervention because of failure of medical intervention. For those centers with access to and experience in arterial embolization in infants then this appears to be a reasonable intervention and has been reported previously (15). Our previously reported experience with hepatic artery ligation in infants with multifocal hemangiomas (25) showed rapid control of cardiac failure without any concomitant worsening of liver function. This was not replicated in the current series and surgical resection of the affected segments appeared to be a better strategy if the child is fit enough to withstand the intervention.

Congenital connection between the arterial and portal venous system is the least common CVM reported in the literature (26–29). Congenital arterioportal hypertension has been reported, usually as an isolated anomaly, with often multiple symptoms making diagnosis difficult. Most infants seem to have failure to thrive, recurrent gastrointestinal bleeding and usually splenomegaly (26,27). There is not usually a structural anomaly of the liver evident on computed tomography or magnetic resonance imaging and most are detected on US with Doppler flow studies. Treatment can be difficult and has to be individualized. Our 2 cases did have an isolated structural defect at the level of the left portal vein but the resulting aneurysm obtained its arterial input from both left and right sides of the liver. This probably accounted for failure of embolization or specifically left-sided arterial ligation. One residual artery appeared to be enough to maintain the high portal venous pressures and ongoing symptoms. These infants were much more stable than those with predominantly arteriovenous malformations and cardiorespiratory failure but still both infants developed complications possibly attributable to an acute reduction in portal venous pressure. In 1 this turned out to be portal vein thrombosis following a straightforward dissection and initially uncomplicated postoperative period. In the other there were multiple intestinal perforations and although the dissection was more complicated here because of the previous duodenal atresia reconstruction it was still unexpected and we speculate that it may have been due to some acute change in portal venous hemodynamics. Indeed we found that all our arterioportal malformations (n = 4) needed to go on to either segmentectomy or lobectomy despite prior attempts at either embolization or ligation of the feeding vessels. Successful embolization of arterioportal lesions (14,16) has been reported (16).

Our cases of CVM of the liver show a fairly consistent pattern of presentation that is also reflected in the current literature. All 4 of the group (ii) arteriovenous malformations developed significant jaundice, cardiac failure requiring diuretics early in their clinical course, and 2 developed significant pulmonary hypertension requiring nitric oxide or sildenafil support.

A variety of antenatal presentation may also be seen, although this does not appear common and whether it is due to the accompanying cardiac anomalies is not clear. So, Hartung et al (3) reported 2 fetal cases with “edema” evident at 14 weeks gestation and ascites and polyhydramnios evident at 31 weeks. In both there was some visible umbilical vein anomalies present on the maternal US scan.

Congenital heart defects were present in all our cases and in 8 (80%) of the other published cases (3,10–17). Population studies from the USA and Norway suggest that approximately 45% of all children with trisomy 21 have congenital heart defects and of these almost half will have an atrioventricular septal defect and a third a ventricular septal defect (6). In view of our findings we suggest that infants with trisomy 21 and a defined cardiac malformation should also be considered for abdominal US to enable early detection of CVM in the liver.

In conclusion, our series shows an association between congenital cardiac and hepatic vascular malformations in those children with trisomy 21. Clinical presentation appears stereotypical according to the underlying nature of the CVM and those with AVMs are more likely to require intervention with potentially worse outcomes in terms of morbidity and mortality, when compared to patients with venovenous malformations.

Back to Top | Article Outline


1. Stringer M. The clinical anatomy of congenital portosystemic venous shunts. Clin Anat 2008; 21:147–157.
2. Guerin F, Blanc T, Gauthier F. Congenital portosystemic vascular malformations. Semin Pediatr Surg 2012; 21:233–244.
3. Hartung J, Chaoui R, Kalache K, et al Prenatal diagnosis of intrahepatic communications of the umbilical vein with atypical arteries (A-V fistulae) in two cases of trisomy 21 using color Doppler ultrasound. Ultrasound Obstet Gynecol 2000; 16:271–274.
4. Weijerman ME, Van Furth AM, Van der Mooren MD, et al Prevalence of congenital heart defects and persistent pulmonary hypertension of the neonate with Down syndrome. Eur J Pediatr 2010; 169:1195–1199.
5. Freeman SB, Taft LF, Dooley KJ, et al Population-based study of congenital heart defects in Down syndrome. Am J Med Genet 2998; 80:213–217.
6. Freeman SB, Bean LH, Allen EG, et al Ethnicity, sex and the incidence of congenital heart defects: a report from the National Down Syndrome Project. Genet Med 2008; 10:173–180.
7. Moore L, Toi A, Chitayat D. Abnormalities of the intra-abdominal fetal umbilical vein: reports of four cases and a review of the literature. Ultrasound Obstet Gynecol 1996; 7:21–25.
8. Basu S, Pooniya V, Kumar A, et al Down syndrome with cerebral arteriovenous malformation. Ann Trop Paediatr 2009; 29:61–65.
9. Kantarci M, Doganay S, Bozkurt M, et al Infra-renal abdominal aorta hypoplasia in a case of Down syndrome depicted by multidetector computed tomography angiography. J Cardiovasc Med 2009; 10:276–278.
10. Kieran MW, Vekemans M, Robb LJ, et al Portohepatic shunt in a Down syndrome patient with an interchange trisomy 47,XY,-2,+der (2),+der (21)t (2;21)(p13;q22.1)mat. Am J Med Genet 1992; 44:288–292.
11. Kitagawa S, Gleason WA Jr, Nrothrup H, et al Symptomatic hyperammonemia caused by congenital portosystemic shunt. J Pediatr 1992; 121:917–919.
12. Courtens W, Segers V, Johansson A, et al Association between Down syndrome and portohepatic shunt. Am J Med Genet 2000; 93:166–168.
13. Saxena AK, Sodhi KS, Arora J, et al Congenital intrahepatic portosystemic venous shunt in an infant with Down syndrome. Am J Roentgenol 2004; 183:1783–1784.
14. Bogert JN, Potter DD, Crow S, et al An unusual presentation of a congenital intrahepatic arterioportal fistula in an infant with Down syndrome. J Pediatr Surg 2011; 46:252–255.
15. Chun PT, Chun T, Files M, et al Percutaneous embolization of congenital portosystemic venous fistula in an infant with Down's syndrome. Case Rep Vasc Med 2013; 2013:127023.
16. Golewale N, Paltiel HJ, Fishman SJ, et al Portal vascular anomalies in Down syndrome: spectrum of clinical presentation and management approach. J Pediatr Surg 2010; 45:1676–1681.
17. Pipitone S, Garofalo C, Corsello G, et al Abnormalities of the umbilico-portal venous system in Down syndrome: a report of two new patients. Am J Med Genet 2003; 120A:528–532.
18. Senniappan S, Pitt K, Shah P, et al Postprandial hyperinsulinaemic hypoglycaemia secondary to a congenital portosystemic shunt. Horm Res Paediatr 2015; 83:217–220.
19. Bernard O, Franchi-Abella S, Branchereau S, et al Congenital portosystemic shunts in children: recognition, evaluation and management. Semin Liver Dis 2012; 32:273–287.
20. Howard ER, Davenport M. Congenital extrahepatic portocaval shunts—the Abernethy malformation. J Pediatr Surg 1997; 32:494–497.
21. Morgan G, Superina R. Congenital absence of the portal vein: two cases and a proposed classification system for portasystemic vascular anomalies. J Pediatric Surg 1994; 29:1239–1241.
22. Kondo M, Itoh S, Kunikata T, et al Time of closure of ductus venosus in term and preterm neonates. Arch Dis Child 2001; 85:F57–F59.
23. Loberant N, Herskovits M, Barak M, et al Closure of the ductus venosus in premature infants: findings on real-time gray-scale, color-flow Doppler, and duplex Doppler sonography. Am J Roentgenol 1999; 172:227–229.
24. Christison-Lagay ER, Burrows PE, Alomari A, et al Hepatic hemangiomas: subtype classification and development of a clinical practice algorithm and registry. J Pediatr Surg 2007; 42:62–67.
25. Davenport M, Hansen L, Heaton ND, et al Hemangioendothelioma of the liver in infants. J Pediatr Surg 1995; 30:44–48.
26. Heaton ND, Davenport M, Karani J, et al Congenital hepatoportal arteriovenous fistula. Surgery 1995; 117:170–174.
27. Inon AE, D’Agostino D. Portal hypertension secondary to congenital arterioportal fistula. J Pediatr Gastroenterol Nutr 1987; 6:471–473.
28. Norton SP, Jacobson K, Moroz SP, et al The congenital intrahepatic arterioportal fistula syndrome: elucidation and proposed classification. J Pediatr Gastroenterol Nutr 2006; 43:248–255.
29. Davenport M, Redkar R, Howard ER, et al Arterioportal hypertension: a rare complication of partial hepatectomy. Pediatr Surg Int 1999; 15:543–545.

Down syndrome; hepatic arteriovenous malformation; patent ductus venosus; portocaval shunt; trisomy 21

© 2016 by European Society for Pediatric Gastroenterology, Hepatology, and Nutrition and North American Society for Pediatric Gastroenterology,