Bleeding ectopic varices in cirrhosis: the role of transjugular intrahepatic portosystemic stent shunts


Dr N. Kochar, Department of Hepatology, Royal Infirmary of Edinburgh, 51, Little France Crescent, Edinburgh EH16 4SU, UK.


Aliment Pharmacol Ther28, 294–303


Background  Bleeding from ectopic varices is uncommon but can be difficult to manage.

Aim  To report our experience of the use of transjugular intrahepatic portosystemic stent shunts (TIPSS) in the management of uncontrolled bleeding from ectopic varices.

Methods  A retrospective study of patients who had TIPSS for bleeding ectopic varices. Patients were selected from a dedicated data base.

Results  Over 14 years, of 750 TIPSS insertions, 28 patients had TIPSS for bleeding ectopic varices (Child–Pugh score: 8.8 ± 1.8). Varices were rectal (12), stomal (8), duodenal (4) and at other sites (4). Concomitant variceal embolization was performed in five. Portal pressure gradient fell from 18.2 ± 6.4 to 7.2 ± 3.5 mmHg. TIPSS achieved haemostasis in six of nine patients who presented with active bleeding. Five patients rebled from ectopic varices. This was related to shunt dysfunction in two and responded to shunt interventions. Three patients rebled despite a functional shunt. Of these, thrombin controlled bleeding in one. Eight patients developed hepatic encephalopathy post-TIPSS.

Conclusions  Transjugular intrahepatic portosystemic stent shunt is a safe and effective treatment for bleeding ectopic varices. Rebleeding from ectopic varices related to shunt dysfunction responds to shunt intervention. A significant proportion of patients have rebleeding despite a patent shunt, when other adjunctive measures like thrombin injection may be tried.


Portal hypertension is a significant complication of cirrhosis and often leads to the development of abnormal collaterals between the portal and systemic circulation. These can occur anywhere in the gastrointestinal tract (GIT), but occur most commonly at the cardio-oesophageal junction. Collaterals developing at other sites have been broadly termed as ectopic varices. These can develop anywhere in the abdomen including the entire GIT, in the abdominal wall, sites of previous abdominal surgery, stomal sites, the falciform ligament and gall bladder. Figure 1 shows stomal varices supplied by the inferior mesenteric vein.

Figure 1.

 Stomal varices supplied by inferior mesenteric vein.

The prevalence of ectopic varices seems to depend on the cause of portal hypertension and the technique used to demonstrate the varices. Although these are a relatively common finding at endoscopy, they are an unusual cause of gastrointestinal haemorrhage and can account for up to 5% of variceal bleeding.1 In a review of 169 cases of bleeding ectopic varices, 17% occurred in the duodenum, 17% occurred in the jejunum or ileum, 14% in the colon, 8% in the rectum and 9% in the peritoneum. Twenty-six per cent bled from peristomal varices and a few from other infrequent sites.1

Management of bleeding from ectopic varices can be challenging both because of inaccessibility and initial difficulty in diagnosis and subsequent difficulty in treatment. It is difficult to determine the best treatment strategy for ectopic varices as most literature on this subject consists mainly of small series and case reports with no randomized trials of therapy.

Traditional conservative treatments include local ligation,2 cauterization, sclerotherapy3 or embolization of the feeding vessels. Described surgical options range from surgical excision, mucocutaneous detachment, stomal resiting or revision,4 to portosystemic shunting5, 6 and liver transplantation. High rebleeding rates are observed following local treatment of varices and both selective variceal embolization (VE) and portocaval surgery are associated with a high post-operative morbidity and mortality. More recently transjugular intrahepatic portosystemic stent shunts (TIPSS) have been used in management of bleeding ectopic varices.1

We report here experience from a single centre on the use of TIPSS to control bleeding from ectopic varices.



Clinical data on all patients who had a TIPSS inserted in our centre from 1991 until the start of this study have been prospectively collected and entered into a dedicated database. We retrospectively reviewed and collected data from the database and case notes of all patients who required TIPSS insertion for bleeding ectopic varices.

TIPSS insertion.  Index TIPSS placement was performed by a standard procedure and has been described elsewhere.7, 8 All patients had pre-TIPSS right atrial pressure, inferior vena cava and direct portal pressure measurements as well as post-TIPSS portal and right atrial pressure measurements.

In patients with significantly large collaterals and where feeding vessel was easily accessible, concomitant embolization of varices with coils was also performed. All patients were administered 1 g of cefotaxime intravenously and this was continued for 24–72 h postprocedure.


Clinically significant bleeding was defined as a decrease in haemoglobin by 2 g/dL, or the requirement of more than 2 units of packed red cells within 24 h to stabilize haemoglobin concentration or signs of volume depletion (systolic blood pressure below 100 mmHg and/or heart rate above 100/min).

Rebleeding was defined as clinically significant bleeding occurring after 24 h or more after a 24-h point of stable haemoglobin/haematocrit following an episode of acute bleeding.9

Shunt dysfunction was defined as portocaval gradient higher than 12 mm of Hg or portographic evidence of an occluded shunt.


Doppler ultrasonography was performed within a week of TIPSS insertion with portography, if indicated. Portography for assessment of shunt function was carried out routinely at 6 monthly intervals thereafter or earlier if rebleeding occurred or shunt dysfunction was suspected. When the gradient was higher than 12 mm of Hg, dilatation of the shunt was performed and a new stent was placed, if needed. If the rebleeding occurred despite a patent shunt and low portal pressure gradient (PPG), the varices were injected with thrombin, where available or referred for VE.

Data were collected on the patients till they were last seen, had a liver transplant or died.

Statistical analysis

Descriptive statistics are provided as the mean ± s.d., median (range) or percentage. The actuarial probability of remaining free from rebleeding, remaining free from shunt dysfunction and survival was calculated via Kaplan–Meier analysis. To determine significant associations, univariate analysis and where applicable multivariate analysis were performed using a Cox-regression model. Statistical analysis was performed using ssps version 11.5 (SPSS Inc., Chicago, IL, USA).


Over a 14-year period, TIPSS was inserted for bleeding ectopic varices in 28 patients, from a total of 750 patients (3.7%).

Baseline characteristics

Table 1 shows baseline characteristics of these patients. Three patients had been on beta-blockers prior to TIPSS insertion.

Table 1.   Clinical characteristics of patients with ectopic varices (n = 28)
Age (years), median (range)61.5 (35.7–75.9)
Child’s grade, n (%)
 A2 (7)
 B17 (61)
 C9 (32)
Child–Pugh score, median (range)8.5 (5–13)
Bilirubin (μmol/dL), median (range)48 (8–873)
Albumin (g/dL), median (range)29 (18–44)
PT (s), median (range)14 (9–28)
Hepatic encephalopathy7
Aetiology of liver disease (n)
 Alcohol induced17
 Primary biliary cirrhosis 1
 Primary sclerosing cholangitis 2
 Viral 2
 Auto immune1
TIPSS unsuccessful1
Site of ectopic varices
 Stomal 8
 Duodenal 4
 Falciform ligament 1
 Caput medusa 1
 Intraperitoneal 1

Twelve patients had presented with rectal bleeding and were found to have rectal varices. Eight patients had visible bleeding from stomal varices. Four patients presented with signs of upper GIT bleed caused by duodenal varices. One patient had visible bleeding from a caput medusa. Three other patients had ectopic varices at other sites. In two of these patients, ectopic varices were found during investigation for obscure GIT bleeding. One was found to have peritoneal varices during laparoscopic assisted enteroscopy and the other was found to have mesenteric varices on angiography. One patient had presented as an acute surgical emergency with hypovolemic shock with signs of intraperitoneal bleeding and was found to have bleeding from a falciform ligament varix on exploratory laparotomy.

Pre-TIPSS therapy

Vasoactive medications (octreotide/terlipressin), where administered (n = 4), were unsuccessful in controlling bleeding. Local treatments including suture ligation and application of gauze soaked in adrenaline were ineffective in controlling bleeding from a caput medusa. Eight patients had endoscopic interventions prior to TIPSS insertion. These were: banding (rectal varices – five; duodenal varices – one) and thrombin injection (n = 2) (rectal varices – one; duodenal varices – one). Bleeding recurred despite thrombin injection in both patients. Endoscopic band ligation proved to be initially successful in three patients (rectal varices – two; duodenal varices – one) of the six patients. However, bleeding recurred in two of these patients (rectal varices – one; duodenal varices – one) necessitating TIPSS insertion. The third patient was deemed to be at a high risk of rebleeding and a TIPSS insertion was attempted. However, this was unsuccessful in this patient.

TIPSS procedure

Table 2 shows TIPSS-related characteristics of these patients.

Table 2.   Data regarding TIPSS (n = 27) (mean ± s.d.)
  1. TIPPS, transjugular intrahepatic portosystemic shunts; PPG, portal pressure gradient.

Pre-TIPSS PPG (mmHg)18.2 ± 6.4
 Stomal16.2 ± 3.5
 Rectal18.8 ± 8.1
 Duodenal20.5 ± 5.2
 Other18.0 ± 6.5
Post-TIPSS PPG (mmHg)7.2 ± 3.5
Mean reduction PPG (mmHg)11.0 ± 5.2
Percent reduction in PPG (%)60
Covered stents8
Mean stent diameter (mm)10.4 ± 0.7
 Covered stents10 ± 0
 Uncovered stents10.6 ± 0.9
Embolization of varices5
Other treatments pre-TIPSS8
Thrombin injection2

Transjugular intrahepatic portosystemic stent shunt was successfully inserted in 27 of the 28 patients. In one patient who had bled from rectal varices, TIPSS was unsuccessful because of abnormal anatomy of suprahepatic veins. This patient underwent repeat banding of rectal varices, was initiated on propranolol and had no further bleeding.

Transjugular intrahepatic portosystemic stent shunt was inserted as an emergency procedure in nine patients. Covered stents were used in eight patients (Viator Gore; WL Gore & Associates, Inc., Flagstaff, AZ, USA). There was no significant difference in pre TIPSS PPG in ectopic varices at different sites. VE was performed at the initial procedure in five patients.

Control of bleeding

Of the nine patients who underwent TIPSS insertion as an emergency procedure to control acute bleeding, haemostasis was achieved in six of the nine (66%) patients. Of these, two patients had been on vasoactive medications prior to TIPSS. In the other three Child’s C patients, bleeding could not be controlled (bleeding falciform ligament varix (n = 1), duodenal varix (n = 1) and rectal varices (n = 1). In these patients, bleeding was exacerbated by the development of significant coagulopathy secondary to development of progressive liver failure. These patients were not suitable for any further intervention and died between days 1 and 5 post-TIPSS.


Five (21%) rebled from ectopic varices at a median duration of 101 (2–202) days (Table 3). TIPSS was inserted as an elective procedure in all these patients. Four of these patients rebled within 40 days of TIPSS insertion. Concomitant VE was not performed at initial procedure in any of these patients. Two patients had covered stents in situ.

Table 3.   Characteristics of patients who had rebleeding from ectopic varices
Site of varices C–P grade and scorePre-TIPSS PPGPost-TIPSS PPGVE at TIPSSTIPSS to rebleed (days)Shunt at rebleedPPG at rebleedInterventionOutcome
  1. C–P, Child–Pugh; TIPPS, transjugular intrahepatic portosystemic shunts; PPG, portal pressure gradient; VE, variceal embolization.

StomalB, 9165No101Intimal hyperplasia24Parallel shunt; VEBleeding controlled
StomalA, 5140No11Occluded16Parallel shuntBleeding controlled
StomalB, 8235No202Patent6ThrombinBleeding controlled
RectalB, 986No2Patent3Thrombin VE not possibleBleeding not controlled
StomalA, 5103No195Patent0NoneBleeding not controlled

Rebleeding in two of these patients was related to shunt dysfunction, and responded to shunt interventions. One patient had additional VE at the time of shunt intervention. Cumulative risk of rebleeding at 1, 6 and 12 months was 13%, 21% and 29% respectively. Figure 2 shows proportion of patients free of rebleeding.

Figure 2.

 Kaplan–Meir curve showing the proportion of patients free from rebleeding.

In three patients, rebleeding occurred despite patent shunts and low PPGs (Table 3). Two of these patients (rectal varices in one and stomal varices in one) underwent injection of thrombin, which effectively controlled bleeding in one and reduced the amount of bleeding to clinically more manageable levels in the other. In one patient who had recurrent bleeding from stomal varices, no further intervention was undertaken in the referring hospital, but the bleeding gradually reduced to clinically more manageable levels in this patient.

Figure 4 shows an overview of the patients who were treated with TIPSS for ectopic varices.

Figure 4.

 Summary of patients who had TIPSS for bleeding ectopic varices.

Two other patients had portal hypertension-related bleeding from sites other than ectopic varices. One patient had an episode of oesophageal variceal bleed related to shunt dysfunction and this responded to shunt extension. Another patient had recurrent ooze from portal hypertensive gastropathy, despite a patent shunt 9.6 months post-TIPSS. The patient developed progressive liver failure and died 28 days after this episode. Four patients had bleeding from other GI sources including haemorrhoids in three patients and a duodenal ulcer in one patient.

Shunt patency

Of the 40 portograms performed over the follow-up period, 10 episodes of shunt dysfunction were identified in four patients (all uncovered stents). Cumulative primary shunt patency rates at 1, 6 and 9 months were 95%, 89% and 81% respectively (Figure 3). Shunt insufficiency was mainly because of intimal hyperplasia8 and shunt occlusion.2 This led to rebleeding from ectopic varices in two patients and bleeding from oesophageal varices in one patient. The shunt interventions undertaken were parallel shunt insertion,2 balloon angioplasty,5 shunt extension2 and balloon angioplasty with restent.1

Figure 3.

 Kaplan–Meir curve showing primary shunt patency rates.

Hepatic encephalopathy

Post-TIPSS hepatic encephalopathy (HE) occurred in eight (29.6%) (six uncovered and two covered stents) patients. Two died of progressive liver failure 5 and 21 days post-TIPSS and two have undergone liver transplantation at 52 and 245 days post-TIPSS. Of the other four patients, HE is well controlled with medications in two patients, but two other patients require repeated admissions for refractory HE. One of these patients has recently undergone a shunt modification procedure for refractory HE. This patient has not rebled post shunt modification.


Median follow-up duration was 203 (1–1869) days. During the follow-up period, 16 patients died and six patients had a transplant at a median of 84 (1–1869) days post-TIPSS insertion. Eight patients died of liver-related causes (liver failure – seven; hepatocellular carcinoma – one), three patients died of uncontrolled bleeding and five patients died of medical causes (subdural haematoma – one; heart failure – two; sepsis – one; bronchial carcinoma – one). Cumulative survival at 1, 3 and 6 months were 81%, 72% and 61% respectively (Figure 5). Of the six nontransplanted patients who are still alive, four were well when last seen and two have had troublesome HE.

Figure 5.

 Kaplan–Meir curve showing survival following TIPSS for ectopic values.

On univariate analysis none of the patient-related factors such as components of Child–Pugh (C–P) score, Child’s grade, pre-TIPSS PPG, post-TIPSS PPG, type of shunt used at TIPSS (covered vs. uncovered) or the shunt diameter were found to be significantly associated with either improved survival or recurrence of bleeding from ectopic varices.

Procedure-related complications

The hepatic artery was inadvertently punctured in one patient and this was managed successfully by coil embolization.


Ectopic varices can occur at various sites in the abdomen including small bowel, rectum, stomas, falciform ligament, umbilicus, retroperitoneum, biliary tract, vagina and bladder. They are an unusual source of bleeding and account for up to 5% of all variceal bleeding.1 The bleeding from these sources can be difficult to manage because of an initial difficulty in determining the source of bleeding and subsequent difficulty in application of therapy because of inaccessibility of varices in some locations.

Various options have been tried for the treatment of ectopic varices, including medical therapies such as use of vasopressin analogues and octreotide,1 local therapies such as sclerotherapy,3, 10 thrombin injection,11 injection of tissue adhesives (histoacryl),10 ligation2 and application of argon plasma coagulation.12 Other treatments utilized include embolization,13, 14 surgical ligation, resection of a bowel segment with the varix and various local surgical measures for stomal varices.4, 15 Balloon-occluded retrograde transvenous obliteration16 and transileocaecal or percutaneous transhepatic obliteration17 of ectopic varices have also been tried and reported to be successful.

The evidence for efficacy of these procedures is mainly in the form of case reports and systematic evidence for the efficacy of these procedures is lacking. It is generally felt that these procedures have limited success in achieving initial haemostasis, and rebleeding rates following these procedures is high.1, 15, 18, 19 With all these treatments, the underlying portal hypertension remains uncorrected and thus the patients remain at a risk of rebleeding from ectopic varices because of continued formation of new feeding collateral vessels.

To treat the underlying portal hypertension, the options are either surgical creation of a portosystemic shunt5, 6, 20 or creation of TIPSS. Although surgically created portosystemic shunts may have a lower incidence of both rebleeding and need for additional procedures,20 these procedures are associated with high post-operative morbidity and mortality, especially in C–P class B or C patients and if performed in the acute setting. For reasons of the problems associated with portosystemic surgery, TIPSS is now the first line treatment for refractory variceal bleeding. In the literature, there have been several case reports and small series on the use of TIPSS in the management of bleeding ectopic varices at various sites. More recently, larger series of patients with ectopic varices treated with TIPSS have been reported.14, 21–23

In this study, we report the largest series of patients with bleeding ectopic varices treated with TIPSS at a single centre. TIPSS was successfully inserted in all but one patient and haemostasis in 67% (six of nine) patients was effectively achieved. This was achieved solely with TIPSS without concomitant VE in 21 of the 22 (95%) patients and in three of the five (60%) patients who had TIPSS and concomitant VE. Bleeding could not be controlled in three Child’s C category patients despite concomitant VE in two. Rebleeding from ectopic varices occurred in five (21%) patients. This was related to shunt dysfunction in two (40%) patients and responded to shunt-related interventions. In three patients rebleeding occurred despite a functioning shunt with low PPGs. The rate of procedure-related complications was low (3.7%).

Our observations regarding efficacy of TIPSS in achieving initial haemostasis are similar to the observations by Vangeli and Vidal (100%), Haskal (100%) and Shibata et al. (100%).

Rates of rebleeding from ectopic varices (21%) in our series are similar to those reported by Vidal et al. (25%) and Shibata et al. (16.5%), but lower than those reported by Vangeli et al. (37%), although it is not clear from the latter’s paper whether TIPSS effectively achieved initial haemostasis in all the patients. Like Vidal and Shibata et al., we also found that rebleeding can be related to shunt dysfunction and responds to shunt interventions. This was also observed to occur in one of the patients in Vangeli’s series. None of the factors such as components of C–P score, Child’s grade, pre-TIPSS PPG or the type of stent used at TIPSS (covered vs. uncovered) were significantly associated with either risk of recurrence of bleeding from ectopic varices or death in our series. This may, however, be due to the small number of patients.

An interesting observation in our and previous series is that patients with ectopic varices may rebleed after TIPSS despite initial achievement of haemodynamic targets and presence of a patent functioning shunt in situ. In the series by Vangeli et al., this occurred more commonly in those who had TIPSS alone without concomitant VE cf. those who had concomitant VE (48% vs. 28%). Furthermore, they also found that the rebleeding in the majority responded to subsequent VE. On the basis of this, Vangeli et al. recommended TIPSS with concomitant VE for control of ectopic variceal bleeding. However, in the series by Vidal et al.,23 the frequency of rebleeding with a patent TIPSS was found to be low and occurred in only two (8.3%) patients. These patients did not have concomitant VE with TIPSS to achieve initial haemostasis and in both the cases, the rebleeding responded to subsequent VE. In the series by Shibata et al. where again TIPSS alone was used to achieve initial haemostasis, none of the patients had rebleeding with a functional shunt in situ. In our series, we found rebleeding from ectopic varices despite a patent functional shunt occurred in three patients. This responded to thrombin injection in one, but did not respond to thrombin injection in the other patient. In the latter patient, unfortunately, VE was not technically possible because of inaccessibility of the vessel feeding the varices.

Rebleeding from ectopic varices in the presence of a patent functional shunt raises some important issues. First, it is important to recognize that it can occur. It is difficult to say why this occurs. A historical comparison with a previous series from our unit revealed that far fewer patients rebled, despite a patent shunt where the indication for TIPSS was bleeding oesophageal varices (15% cf. 60% for our series).24 This difference in rebleeding rates at the two sites may relate to anatomical differences between ectopic varices and varices at other sites. Variceal bleeding occurs when an expanding force within the varix exceeds the maximum wall tension. The wall tension, calculated by the Laplace equation, is directly proportional to the transmural pressure difference and the radius of the varix but is inversely proportional to the wall thickness of the varix.25 Ectopic varices, unlike oesophageal varices are true veins and are likely to have larger diameters resulting in greater wall tension25, 26 and this may be responsible for higher rates of rebleeding.

The second important issue is regarding the prevention of rebleeding in the presence of a patent TIPSS. To minimize its occurrence, Vangeli et al. have recommended concomitant VE at the time of index TIPSS. Although their data are persuasive, observations from our study and others’ series21, 23 suggest that the overall likelihood of rebleeding with a functional shunt is low (three of 24 patients who had a patent stent rebled in our series). Also, VE at the initial procedure may not be effective in achieving initial haemostasis as two of the three patients in our series where bleeding could not be controlled had concomitant VE at index TIPSS. Technically, it is not always possible to achieve VE as ectopic varices are frequently numerous, tortuous and complex and access to them may be challenging. Even when accessible, complete obliteration of ectopic varices may not be possible because of the presence of other communications with the systemic or mesenteric venous system. These anatomical differences may underlie differences in rebleeding rates seen in the various series, which have or have not used concomitant VE with TIPSS. Concomitant VE, in addition to prolonging the procedure, is invasive, has its own inherent set of complications, such as propagative thrombus or paradoxical systemic embolization.27 We therefore feel that concomitant VE should not be routinely performed at the initial procedure. In the event of rebleeding, if the varices are accessible, local treatments may be tried first. If these measures are ineffective or not possible and if varices are accessible radiologically, VE may be tried.

In the event of uncontrolled bleeding despite TIPSS with or without concomitant VE, options are very limited especially if the patients have advanced liver disease and hence are extremely poor operative candidates. Bleeding could not be controlled in three patients (all Child’s C) and sadly all these three patients developed progressive liver failure and died within 5 days of the initial procedure indicating the dismal prognosis of these patients. Further treatment options in such patients should be explored.

Cumulative primary shunt patency rates are higher in this series compared with previous series.28 This may be related to a greater proportion of patients who had covered stents at time of TIPSS. It is notable that no episodes of shunt insufficiency occurred in patients who had covered stents in situ. The incidence of post TIPSS HE in our series is comparable to the other series in the literature.29

As in previously reported series, the incidence of procedure-related complications in our series was low (3.6%) and these are usually manageable by further radiological interventions.

In conclusion, we feel, TIPSS alone is safe and effective both in achieving initial haemostasis and in preventing rebleeding from ectopic varices with a low risk of complications and an acceptable rate of post TIPSS HE. It should therefore be the first line of treatment in the management of bleeding from ectopic varices. Rebleeding may be related to shunt dysfunction, which responds to shunt interventions, and therefore shunt should be investigated first in the event of rebleeding. It is important to be aware that a significant proportion of patients rebleed, despite functioning shunts. Adjunctive measures such as thrombin injection, or VE should be reserved for patients for such patients.


Declaration of personal interests: We would like to thank Dr NDC Finlayson, Dr AJ MacGilchrist, Dr KJ Simpson, Dr AJ Bathgate, Dr JN Plevris and Dr C Blair whose patients were used in this study. We also wish to thank research nurses Sister Kim Macbeth, Sister Jane Holmes and Sister Gwenyth Wilkie for their administrative assistance and help with data collection. Declaration of funding interests: None.