Extrahepatic portal vein obstruction (EHPVO) with cavernous transformation is an important cause of portal hypertension in children. It is associated with gastrointestinal bleeding mainly from oesophageal and/or gastric varices (1). The precise aetiology of the development of EHPVO in the majority of these children is unknown. The predisposing factors are believed to be the following: conditions that directly injure the vessel, rare portal vein congenital anomalies, and a group of systemic causes such as neonatal sepsis, abdominal sepsis, dehydration, multiple exchange transfusions, and hypercoagulable states (2). Prothrombotic disorders include deficiencies in the naturally occurring anticoagulants (protein C, protein S, and antithrombin) and factor V Leiden and prothrombin gene mutations (PT20210 G/A). Acquired prothrombotic disorders include the antiphospholipid syndrome, paroxysmal nocturnal haemoglobinuria, and myeloproliferative disorders (MPDs): polycythaemia vera (PV), essential thrombocythaemia (ET), and idiopathic myelofibrosis (3). Recent reports confirm that the JAK2 V617F mutation is strongly implicated in the pathogenesis of myeloproliferative disorder that may be associated with increased rates of thrombosis (4,5). The battery of tests for hypercoagulable state allowed us to study 30 children with EHPVO for these conditions.
MATERIALS AND METHODS
Case records of 108 patients with EHPVO presenting between August 1979 and September 2005 were analysed retrospectively. Sixty-seven (62%) were male and the median age at diagnosis was 4.75 years (range 1 day–16.3 years). The patient who was diagnosed at 1 day of life was a white male, diagnosed antenatally with an abnormality of the hepatic venous and arterial system and portal vein atresia. Of the patients, 91 (84%) were white, 15 (14%) were of Asian descent, and 2 (2%) were of African descent. The diagnosis of EHPVO was made on the basis of signs and symptoms of portal hypertension, including gastrointestinal bleeding and splenomegaly, as well as ultrasonographic finding of portal vein cavernoma with the absence of any evidence of chronic liver disease. The length of hospitalisation at first admission was a median 8 (1–65) days. Since 1992, thrombophilia testing was performed including protein C, protein S, antithrombin, lupus anticoagulant, factor V Leiden, and prothrombin gene mutation. JAK2 V617F mutation screening was performed using frozen blood specimens from 30 patients retrospectively. A group of patients with nonthrombotic liver disease (n = 27) was used as a negative control group for JAK2 V617F. Management of EHPVO included haemodynamic stabilisation in those with bleeding varices. From 1992 onward, octreotide was administered as a continuous infusion (25 μg/hour). This reduced the need for Sengstaken tube placement during periods of acute bleeding. Oesophageal gastroduodenoscopy was performed with sclerotherapy and/or banding until variceal obliteration was achieved. Oesophageal varices were graded on a scale of 1 to 4 based on the estimated size of the largest varix using Conn criteria (6). Patients were followed up for a period of 6 (0.4–19.5) years. Data were entered into SPSS version 13.0 (SPSS, Chicago, IL) and descriptive analysis was made. Results are presented as median (range).
The median age at first presentation at our centre was 7.5 years (range 1 day–24.2 years). Forty-five (42%) patients were 5 years old or younger, with a median age of 2.8 years (1 day–4.7 years) and 63 (58%) patients were older than 5, with median age 9.6 (5.1–24.2) years. Fourteen (12%) patients were found to have a history of umbilical vein catheterisation, 9 (8%) had abdominal sepsis, and 8 (7%) had neonatal sepsis. Associated congenital anomalies were found in 26 patients (24%). The majority of patients, 67 (62%), had no described predisposing factors. The characteristics at presentation are shown in Table 1 and laboratory indices in Table 2. There were no differences in laboratory indices according to age at presentation. Detailed prothrombotic profiles were available in 30 patients. The median values of protein C and protein S were 67 (7–124) U/dL and 77 (46–126) U/dL, respectively. Protein C activity was low in 6 patients and protein S antigen was low in 2 patients; however, only 1 patient had a significantly low protein C value (7 U/dL). Lupus anticoagulant was positive in 1 patient. Antithrombin was normal in all and factor V Leiden mutations were absent in all cases. JAK2 V617F mutation screening was found to be negative in all 30 patients tested.
Overall, 83 (77%) patients bled. The median age at the first episode of bleeding was 4.6 (0.02–16.4) years. Thirty-nine (47%) patients presented at King's College Hospital at the first bleed, at a median age of 5.2 (0.02–16.3) years. Thirty-nine (47%) patients presented with bleeding at the local hospital, at a median age of 3.7 (0.6–14.8) years. The remaining 5 (6%) patients who bled presented without bleeding but bled during follow-up, at an age of 10.2 (5.3–15.9) years. Thirty-eight (46%) patients had a second bleed during the follow-up period. The median interval between the first and second bleeding was 2.9 (0.02–11.1) years. The median number of bleeds in those who bled more than once was 2 (range 2–10). Since 1992, all of the patients who presented with gastrointestinal bleeding received intravenous octreotide. Before 1992, 4 patients received pitressin. The median duration of octreotide treatment was 4 (range 2–9) days. Fifty-one (47%) patients received blood transfusions, 49 as part of emergency resuscitation. Of the 108 patients, 22 (20%) received propranolol, which was still being administered in 14 (64%) at last follow-up. The median duration of therapy was 2.9 (0.01–7.4) years. One patient had a further bleeding episode after starting propranolol.
Thirteen (12%) patients required Sengstaken tube placement (11 before 1992 and 2 after 1992). Repeated Sengstaken tube placement was required in 4 patients: 2 had 2, and 2 patients had 3, at a median of 1.26 days (range 20 minutes–5 days) apart. All received sclerotherapy after Sengstaken tube placement and no complications from Sengstaken tube placement were recorded.
On first endoscopy, oesophageal varices were present in 92 patients. Of these patients, 70 (76%) received sclerotherapy, 5 (5%) were banded, and 16 (17%) had both. A median of 6 (1–20) sclerotherapy sessions or 2 (1–5) banding sessions was required to eradicate the varices. During acute bleeding, the median interval between bleed and sclerotherapy session was 0 days (0–14 days). The oesophageal varices were grade I in 8 (9%), grade II in 23 (25%), grade III in 53 (58%), grade IV in 4 (4%), and the grade was not specified in 4 (4%). Of all 108 subjects, 69 (64%) had gastric varices, 28 (26%) had gastropathy, 12 (11%) had duodenal lesions, and 44 (41%) had anorectal varices. Complications of sclerotherapy were seen in 34 (32%) patients in the form of ulcers in 16, strictures in 10, and both in 7, while 1 had erosive gastritis and duodenitis. Fifteen patients needed a median of 2 (range 1–16) sessions of endoscopic dilatations to treat strictures. Ulcers and gastritis were treated with omeprazole and sucralfate.
Shunt operation was performed in 17 (16%) patients, at a median age of 9.7 (5.2–23.7) years. The indications for surgery were recurrent gastrointestinal bleeding despite regular sclerotherapy in 13 and emergency shunt for failure to control bleeding in 4. The types of shunts were lienorenal in 9, mesocaval in 5, portomesenteric in 2, and portoazygos in 1. The shunt was functioning in 9 of the 17 patients (53%) at last follow-up after 12.8 (2.3–19.5) years. Those with nonfunctioning shunts had gastrointestinal bleeding, controlled with sclerotherapy in 7, and 1 had a second shunt operation, which was successful.
Patients were followed up for a median of 6 (0.4–19.5) years. The age at last follow-up was 14.6 (2.2–29.2) years. The last endoscopy showed 58 (54%) patients were free of varices, 27 (25%) had grade I varices, 7 (7%) had grade II varices, and 1 (1%) had grade III varices, but this patient was still receiving treatment at the time of follow-up. The median time free from bleeding (period of time between last bleeding and last follow-up) was 5.1 (0.04–19.9) years. Four (4%) died because of associated cardiac anomalies (2 with unknown congenital heart defects and 2 with pulmonary stenosis/atresia); none died because of gastrointestinal bleeding.
The results of the present study support earlier publications in that the cause of EHPVO in the majority of patients remains unknown (7). In neonates, EHPVO most commonly occurs following umbilical vein catheterisation, with or without concurrent infection (8). This may not be recognised at the time and may present later in childhood with the complication of portal hypertension. In older children, EHPVO has been associated with liver transplantation, splenectomy, sickle cell anaemia, intraabdominal sepsis, and hypercoagulable states (9). Umbilical vein catheterisation, abdominal sepsis, and neonatal sepsis were noted as possible causes of EHPVO in a small number of children in the present series.
Association between venous thrombosis and thrombophilia has been well defined (7,10). Factor V Leiden mutation, which is the most prevalent of the heritable thrombophilias, increased the relative risk of venous thrombosis in adults by 4- to 6-fold (11,12). Although its association with hepatic vein thrombosis has been described, its association with EHPVO is questionable (13,14). Gurakan et al (12) and Seixas et al (15) failed to demonstrate an association between factor V Leiden mutation and EHPVO. In another recent study, factor V Leiden mutation was infrequent in Indian children with EHPVO. Inherited deficiencies of proteins C and S and antithrombin III increase the risk of venous thrombosis, but they occur at a lower incidence than factor V Leiden (3). In adults, isolated protein C and protein S deficiency have been detected among approximately 30% of patients with EHPVO. However, antithrombin deficiency was found to be less prevalent (16). Seixas et al (17) found that hereditary protein C or protein S deficiency was not an etiological factor for EHPVO in children. The anti-phospholipid antibody syndrome may manifest as either arterial or venous thromboembolism, and is characterised by the presence of anti-phospholipid antibodies (lupus anti-coagulant and/or anti-cardiolipids antibodies) (18).
In our study, protein C was significantly low in 1 patient only and lupus anticoagulant was present in another. Neither patient was treated with anticoagulant therapy because there is a lack of evidence of the benefit of anticoagulant or thrombolytic therapy in this setting (14). Mildly low values of protein C and S could be a reflection of affected liver synthetic function rather than genetic deficiency. We extended our procoagulant screen to include JAK2 V617F mutation analysis. MPDs are the most frequent cause of both Budd-Chiari syndrome and EHPVO in adults (19,20). The detection of MPDs in patients with abdominal vein thrombosis has been facilitated by the discovery of a point mutation in the JAK2 gene. JAK2 is a cytosolic tyrosine kinase that transduses signals induced by haemopoietic growth factors, including erythropoeitin (21).
In 2005, an acquired JAK2 mutation (JAK2V617F) was reported in more than 90% of patients with PV and half of the patients with ET (21). This mutation leads to constitutive activation of JAK2 and stem cell proliferation. JAK2 V617F has not been reported in normal healthy controls (5). Primigani et al (22) and Kiladjian et al (23) have examined JAK2 V617F in adult patients with idiopathic EHPVO and Budd-Chiari syndrome. They found that the JAK2 mutation occurred in 36% and 34% of patients with EHPVO and 40% and 45% of patients with Budd-Chiari syndrome, respectively. Other studies of EHPVO have demonstrated the JAK2 mutation in 17.2% and 18.2% of cases, respectively (24,25). Although positive screening for JAK2 V617F in the blood of patients with abdominal vein thrombosis confirms the diagnosis of underlying MPD, a negative result does not exclude MPD because up to 10% of patients with PV and half of those with ET will be JAK2 negative (21). This is nevertheless a valuable diagnostic tool and marker of clonality, particularly in acute abdominal vein thrombosis, in which a diagnosis of MPD is often difficult to make. This is also of value in patients with “occult MPD,” who do not fulfill the traditional diagnostic criteria for MPD at presentation, but develop the classic MPD months or years following the initial thrombotic event (5).
Outcome of patients with EHPVO depends on the control of gastrointestinal bleeding from varices. Recent advances in the nonsurgical treatment of gastroesophageal varices have resulted in remarkable improvement in the clinical course of patients. Sclerotherapy has emerged as an effective treatment for bleeding oesophageal varices both in adults and children (26). The results of our study show that sclerotherapy is an effective and safe treatment for oesophageal varices in children. Although the majority of patients had varices on admission, more than 50% were free of varices at the time of follow-up. The significant reduction in bleeding episodes after variceal eradication is a result of effective sclerotherapy and it is unlikely that this would have occurred naturally (26). It must be emphasised that recurrent bleeding episodes after variceal obliteration were successfully managed by a further course of sclerotherapy.
Major complications associated with sclerotherapy before the attainment of complete variceal eradication have been reported previously (27). Similarly to the present study, the most common complications were reported as oesophageal strictures and oesophageal ulcers (28). Other rare complications have been reported including oesophageal perforation (15), spinal artery occlusion with spinal cord necrosis, and paraplegia (29) and perforation leading to development of an oesophagopleural fistula (30). Other potential long-term effects of sclerotherapy on oesophageal motility include reduced lower oesophageal pressure, poor motor contraction, reduced acid clearance, and development of gastroesophageal reflux (31). None of these were noted in the present study. Banding of varices has been shown to be superior to sclerotherapy because varices are eradicated rapidly with fewer sessions, and banding has fewer associated complications compared with sclerotherapy (26,27). In our study, only one fifth of patients had banding; however, this is our favoured mode of treatment now when technically possible.
The surgical option in the management of EHPVO is the formation of a portosystemic shunt. Of our patients, 16% underwent shunt operation with a success rate of 53%. Although portosystemic shunt surgery provides good control of bleeding—possibly improving growth retardation and hypersplenism while protecting against future development of portal biliopathy—it is associated with surgical mortality and is sometimes not feasible due to nonavailability of a satisfactory vessel (32). Mathur et al (33) reported that surgical mortality after emergency shunts is approximately 10%. Shunt occlusion, overt chronic portosystemic encephalopathy, and rebleeding after elective shunt surgery are seen in approximately 10% of patients. In those patients who live in remote areas or who desire a 1-time treatment, elective shunt surgery is a reasonable option. Among our patients, portosystemic shunting was performed as an emergency control of gastrointestinal bleeding in 4 patients. Tomoaki et al (34) reported that it was a safe and effective method for the management of portal hypertension in children. Patients showed significant improvements in growth parameters after the procedure. The superior mesenteric vein to intrahepatic left portal vein (Rex) shunt has been shown to be an effective method of resolving portal hypertension caused by EHPVO including thrombosis after living donor transplantation (35). We have not yet evaluated this technique in this group, however.
EHPVO can be managed successfully with endoscopic sclerotherapy and lately with banding without bleeding-related mortality. Hypercoagulable state, albeit tested in a smaller but selected cohort over the last 10 years, did not reveal these children to be prothrombotic.
1. Gauthier F. Recent concepts regarding extra-hepatic portal hypertension. Semin Pediatr Surg 2005; 14:216–225.
2. Bellomo-Brandao MA, Morcillo AM, Hessel G, et al
. Growth assessment in children with extra-hepatic portal vein obstruction and portal hypertension. Arq Gastroenterol 2005; 40:247.
3. Sharma S, Kumar SI, Poddar U, et al
. Leiden and prothrombin gene G20210A mutations are uncommon in portal vein thrombosis in India. Indian J Gastroenterol 2006; 25:236–239.
4. Kralovics R, Passamonti F, Buser AS, et al
. A gain-of-function mutation of JAK2 in myeloproliferative disorders. N Engl J Med 2005; 352:1779–1790.
5. Patel RK, Lea NC, Henegan MA, et al
. Prevalence of the activating JAK2 tyrosine kinase mutation V617F in the Budd-Chiari syndrome. Gastroenterol 2006; 130:2031–2038.
6. Conn HO. Ammonia tolerance in the diagnosis of esophageal varices. A comparison of endoscopic radiologic and biochemical techniques. J Lab Clin Med 1967; 70:442–451.
7. Yadav S, Dutta AK, Sarin SK. Do umbilical vein catheterization and sepsis lead to portal vein thrombosis? A prospective, clinical, and sonographic evaluation. J Pediatr Gastroenterol Nutr 1993; 17:392–396.
8. Sarin SK, Sollano JD, Chawla YK, et al
. Consensus on extrahepatic portal vein obstruction
. Liver Int 2006; 26:512–519.
9. Perel Y, Dhermy D, Carrere A, et al
. Portal vein thrombosis after splenectomy for hereditary stomatocytosis in childhood. Eur J Pediatr 1999; 158:628–630.
10. Gurgey A, Mesci L, Renda Y, et al
. Factor V Q 506 mutation in children with thrombosis. Am J Hematol 1996; 53:37–39.
11. Janssen HL, Meinardi JR, Vleggaar FP, et al
. Factor V Leiden mutation, prothrombin gene mutation, and deficiencies in coagulation inhibitors associated with Budd-Chiari syndrome and portal vein thrombosis: results of a case-control study. Blood 2000; 96:2364–2368.
12. Gurakan F, Gurgey A, Bakkaloglu A, et al
. Homozygous factor V Leiden mutation in a child with Budd-Chiari syndrome. J Pediatr Gastroenterol Nutr 1999; 28:516–517.
13. Rangari M, Gupta R, Jain M, et al
. Hepatic dysfunction in patients with extrahepatic portal venous obstruction. Liver Int 2003; 23:434–439.
14. Figen G, Makbule E, Nurten KA, et al
. Extrahepatic portal vein thrombosis in children: etiology and long-term follow-up. J Clin Gastroenterol 2004; 38:368–372.
15. Seixas CA, Hessel G, Ribeiro CC, et al
. Factor V Leiden is not common in children with portal vein thrombosis. Thromb Haemost 1997; 77:258–261.
16. Hackeng TM, Sere KM, Tans G, et al
. Protein S stimulates inhibition of the tissue factor pathway by tissue factor pathway inhibitor. Proc Natl Acad Sci USA 2006; 103:3106–3111.
17. Seixas CA, Hessel G, Siqueira LH, et al
. Study of hemostasis in pediatric patients with portal vein thrombosis. Haematologica 1998; 83:955.
18. Korenkov M, Depta A, Kentner R, et al
. Paraneoplastic lupus anticoagulans syndrome. Zentralbl Chir 2005; 130:372–374.
19. Primignani M, Martinelli I, Bucciarelli P, et al
. Risk factors for thrombophilia in extrahepatic portal vein obstruction
. Hepatology 2005; 41:603–608.
20. Valla D, Casadevall N, Huisse MG, et al
. Aetiology of portal vein thrombosis in adults. A prospective evaluation of primary proliferative disorders. Gastroenterology 1988; 94:1063–1069.
21. Baxter EJ, Scott LM, Campbell PJ. Acquired mutation of the tyrosine kinase JAK2 in human myeloproliferative disorders. Lancet 2005; 365:1054–1061.
22. Primigani M, Barosi G, Bergamaschi G, et al
. Role of the JAK2 mutation in the diagnosis of chronic myeloproliferative disorders in splanchnic vein thrombosis. Hepatology 2006; 44:1528–1534.
23. Kiladjian J, Cervantes F, Leebeek F, et al
. Role of JAK2 mutation detection in Budd Chiari syndrome (BCS) and portal vein thrombosis (PVT) associated MPD [abstract]. Blood 2006; 108:377.
24. Colaizzo D, Amitrano L, Tiscia GL, et al
. The JAK2 V617F mutation frequently occurs in patients with portal and mesenteric venous thrombosis. J Thromb Haemost 2006; 5:55–61.
25. Regina S, Herault O, D'Iteroche L, et al
. The JAK2 mutation V617F is specifically associated with idiopathic splanchnic vein thrombosis [abstract]. Blood 2006; 108:380.
26. McKierman PJ, Beth SV, Davison SM. A prospective study of endoscopic esophageal variceal ligation using multiband ligator. J Pediatr Gastroenterol Nutr 2002; 34:207–211.
27. Baroncini D, Milandri GL, Borioni D, et al
. A prospective randomized trial of sclerotherapy versus ligation in the elective treatment of bleeding esophageal varices. Endoscopy 1997; 27:235–240.
28. Hill ID, Bowie MD. Endoscopic sclerotherapy
for control of bleeding varices in children. Am J Gastroenterol 1991; 86:472–476.
29. Seidman E, Weber AM, Morin CL, et al
. Spinal cord paralysis following sclerotherapy for esophageal varices. Hepatology 1984; 4:950–954.
30. Prasad H, Poddar U, Thapa BR, et al
. Esophago-pleural fistula following endoscopic sclerotherapy
in a child. Gastrointest Endosc 2000; 52:804–806.
31. Greenholtz SK, Hall RJ, Sondheimer JM, et al
. Manometric and pH consequences of esophageal endosclerosis in children. J Pediatr Surg 1988; 23:38–41.
32. Sarin SK, Agarwal SR. Extrahepatic portal vein obstruction
. Semin Liv Dis 2002; 22:43–58.
33. Mathur SK, Shah SR, Nagral SS, et al
. Transabdominal extensive esophagogastric devascularisation gastroesophageal stapling for management of noncirrhotic portal hypertension: long-term results. World J Surg 1999; 23:1168–1174.
34. Tomoaki K, Rene R, Raghad K, et al
. Portosystemic shunting in children during the era of endoscopic therapy: improved postoperative growth parameters. J Pediatr Gastroenterol Nutr 2000; 30:419–424.
35. Bambini DA, Superina R, Almond PS, et al
. Experience with the Rex shunt (mesenterico-left portal bypass) in children with extrahepatic portal hypertension. J Pediatr Surg 2000; 35:13–18.