EASL Clinical Practice Guidelines

Extrahepatic portal vein obstruction (non-cirrhotic, non-malignant)

Extrahepatic portal vein obstruction (EHPVO) occurs due to the three following mechanisms: malignant invasion (frequently but improperly referred to as malignant thrombosis), portal vein narrowing within a malignant tumor, and thrombosis. Malignant invasion and portal vein narrowing will not be discussed further in this section. Following acute thrombosis, in the absence of recanalisation, the portal venous lumen obliterates and porto-portal collaterals develop. This process is called cavernomatous transformation of the portal vein, the result of which is the portal cavernoma, which fully develops in a couple of months after acute thrombosis. Chronic PVT has been used to designate the latter condition although these terms are not as factual as cavernoma or cavernomatous transformation. There is a debate as to whether portal cavernoma may result from other mechanisms than thrombosis. In children, aetiological investigation has been negative or has shown only weak common prothrombotic conditions [98]. When a cavernoma is found in infancy or childhood in the absence of local or general factors for thrombosis, the hypothesis of a congenital malformation cannot be ruled out although evidence for this hypothesis is still poor [98].


Available data in patients with non-cirrhotic non-malignant EHPVO come from short-term prospective studies following acute thrombosis [3], or from retrospective cohort studies where patients have received various forms of treatment [[40], [42], [76], [99]]. Due to the improved sensitivity of non-invasive imaging, diagnosis of EHPVO is increasingly being made at an early stage of acute PVT [[40], [42], [76], [99]]. Among features of PH, gastrointestinal bleeding has become a rare mode of presentation, by contrast with frequent fortuitous findings of an enlarged spleen, reduced blood cell counts, gastroesophageal varices or portal hypertensive gastropathy, or portosystemic collaterals at abdominal imaging [[40], [42], [76], [99]]. The severity of portal hypertension typically contrasts with a mild or absent liver dysfunction and with normal levels of transaminases, alkaline phosphatase, and gamma-glutamyl transferase. Some patients may experience post-prandial abdominal pain, or features of incomplete bowel obstruction related to ischemic stenosis. Less frequently, initial manifestations are with biliary symptoms (biliary pain, pancreatitis, cholecystitis) related to portal cholangiopathy, a condition characterized by compression and deformation of intra- and extrahepatic bile ducts by the collateral veins constituting the cavernoma. Progressive cholestatic disease or recurrent bacterial cholangitis are rare in patients with portal cholangiopathy [[42], [76], [99]].


The most frequent complication is gastrointestinal bleeding related to PH [[40], [42], [76], [99]], followed by recurrent thrombosis (mostly in the splanchnic area) [[40], [42], [76], [99]] and more rarely, biliary complications [100]. Asymptomatic recurrent thrombosis in the splanchnic area is underestimated and its actual clinical significance requires further evaluation. Ascites, bacterial infections and overt encephalopathy are uncommon except following an episode of gastrointestinal bleeding [101]. Subclinical encephalopathy appears to be much more common than previously suspected [102]. In children, a specific additional consequence appears to be growth failure [103]. Regenerative macronodules may develop but HCC has not been reported yet [104]. Previous gastrointestinal bleeding and size of esophageal varices have been identified as independent predictors for gastrointestinal bleeding [[40], [42]]; the presence of an underlying prothrombotic condition as a predictor for recurrent thrombosis [[40], [42]]; dilated segments of the bile ducts, for clinical biliary complications [97]; and age, ascites, extension to the superior mesenteric vein and severity of underlying conditions as predictors for death [[40], [42], [76], [99]].


The diagnosis of EHPVO should be considered in patients with features of PH or hypersplenism; in patients affected with a condition associated with a risk for PVT (general: MPN antiphospholipid syndrome, inherited thrombophilic factors, or local: pancreatitis, diverticulitis, inflammatory bowel disease); in patients with abdominal pain; and in patients with biliary disease. Rarely, the diagnosis has to be considered in a context resembling decompensated cirrhosis (encephalopathy, and/or ascites, and/or bacterial infection).

A diagnosis of EHPVO is based on the findings of Doppler ultrasound, and axial CT or MR imaging using vascular contrast agents. The experience and awareness of the radiologist is crucial. Essential features are; (a) the absence of visible lumen corresponding to the portal vein; and (b) the presence of numerous, serpiginous vascular channels in porta hepatis [[105], [106]]. Other less specific features may provide indirect clues for an obstructed portal vein: a dysmorphic liver where segment 1 and segment 4 are enlarged but surface is smooth; a mosaic pattern of parenchymal enhancement in the arterial phase, with homogeneous enhancement at a later phase; an increased enhancement of the peripheral parts of the liver at the arterial phase; a dilated hepatic artery; and a mild irregular dilatation of the bile ducts [107]. A thickened gallbladder wall due to collateral veins should be differentiated from cholecystitis. A thickened heterogenous pancreas due to collateral veins should be differentiated from pancreatic cancer and chronic pancreatitis. In cases of pure portal vein obstruction, liver biopsy shows an essentially normal liver. However, a cavernomatous transformation of the portal vein can be superimposed on cirrhosis or obliterative portal venopathy where diagnosis requires a liver biopsy [[3], [99]]. Liver biopsy in EHPVO is indicated in patients with persistently abnormal liver tests or a dysmorphic liver whose aspect is not typical for extrahepatic venous obstruction as described above. Non-invasive tests like elastometry would be most useful in recognising underlying liver disease [108].

Underlying prothrombotic disorders and local factors are common in adults. These disorders constitute major determinants of outcome, and may require specific therapy (Fig. 3).

Fig. 3
Proposed algorithm for making a decision of permanent anticoagulation in patients with chronic extrahepatic portal vein obstruction. *Assessment based on personal and familial history of unprovoked deep vein thrombosis, and on findings of isolated or combined prothrombotic conditions.

View Large Image | View Hi-Res Image | Download PowerPoint Slide


Prevention of thrombotic extension or recurrence

The effect of specific treatments for underlying conditions has not been evaluated. Evidence for a favourable benefit/risk ratio of anticoagulation is low as no prospective study has ever been performed. In three retrospective cohort studies on non-cirrhotic PVT patients, long-term anticoagulation has been associated with a reduced risk of recurrent thrombosis. In a multivariate analysis it was found to be an independent factor in one study (risk ratio 0.39, p = 0.02) [42] and borderline in the other (hazard ratio 0.2, p = 0.1) [41]. Prevention of rethrombosis was also observed at univariate analysis in a large cohort of patients whose initial presentation was with abdominal pain or intestinal ischemia [40]. When evaluated in patients with EHPVO receiving anticoagulation, the risk of recurrent bleeding has not been shown to be increased in the context where prophylaxis for bleeding has been routinely performed [[40], [42]]. In another study where the strategy for bleeding prophylaxis has not been evaluated, anticoagulation therapy was significantly associated with an increased risk of bleeding [41]. The severity of bleeding on anticoagulation has been found to be similar in patients with and without anticoagulation at the time of bleeding [42]. Multivariate analysis indicated a favourable impact of anticoagulation therapy on survival with a statistically significant decrease in mortality in one study [99], and a non-significant decrease in the other [41]. Extrapolation of these data collected between 1983–1998 [83], 1973–2005 [84] and 1985–2009 [41] requires caution.


Prevention of the complications of EHPVO

In most available surveys, patients have been treated for portal hypertension according to the recommendations for patients with cirrhosis. Hemodynamic data in animals with pre-hepatic PH [109] and in patients with non-cirrhotic portal hypertension [110] indicate beneficial effects of non-selective beta adrenergic blockade on splanchnic haemodynamics. Theoretical deleterious effects of non-selective beta blockers on patients with extended thrombosis promoting abdominal pain or intestinal ischemia have never been proved.

According to multivariate analysis, beta adrenergic blockade decreases the risk of bleeding in patients with large varices [42], and improves survival in patients with chronic portomesenteric venous obstruction [99]. Sclerotherapy reduces the incidence of bleeding in previously untreated patients. Endoscopic variceal band ligation is superior to sclerotherapy according to a short-term randomized controlled trial in children [111]. In children, combination of ligation and sclerotherapy provides marginal advantage to either band ligation alone or sclerotherapy alone. In adults, by two years of follow-up, there was no difference in the rate of recurrent bleeding between treatment with propranolol or with band ligation for non-cirrhotic portal hypertension (including a majority of patients with EHPVO) [112]. In the latter study, none of the patients were receiving anticoagulation. Rebleeding rate was about 20% at two years.

In selected patients, low mortality and rebleeding rates have been observed with surgical portosystemic shunting using superior mesenteric or splenic veins [113]. However, the proportion of patients where these shunts are feasible remains unclear. The data with TIPS are still extremely limited in patients without cirrhosis or malignancy. While covered TIPS insertion appears to be feasible when intrahepatic portal veins are visible, results are available only on a short-term follow-up (average 18 months) [114]. Encephalopathy appears to occur at a similar rate as in patients with cirrhosis.

In children with patent superior mesenteric and left portal veins, a bypass can be constructed between these two veins (so-called mesenterico-Rex shunt). The feasibility and long-term patency appears to be high. Gastrointestinal bleeding is effectively prevented. An improvement in mental status and in coagulation factor levels as been observed [[115], [116]]. There is no report of adult patients treated with mesenterico-Rex shunt.

Only patients with clinical manifestations of portal cholangiopathy should be considered for a specific treatment [100]. Bile stones should be treated endoscopically. Risk of endobiliary maneuvers is haemobilia from ruptured intrabiliary varices, which can be massive. Biliary stricture associated with jaundice or bile stones can also be treated endoscopically with repeated stenting. When superior mesenteric vein or splenic veins are evident a surgical shunt can be considered. Because of anecdotal reports of successful TIPS placement, such a procedure can also be considered although results beyond a few months of follow-up have not been reported [[114], [117]].

Overall outcome

Overall outcome is relatively good in patients with extrahepatic PVT in the absence of cirrhosis or malignancy. Five-year survival rates above 70% have been reported in large cohorts spanning over the last 20 years [[40], [41], [42], [76], [99]]. No comparison with the general population is available.