Article Text

Technical and clinical outcomes following EUS-guided thrombin injection and coil implantation for parastomal varices
  1. Andrew Todd1,
  2. Chander Shekhar2,
  3. Joanne O’Rourke1,
  4. Colm Forde1,
  5. Arvind Pallan1,
  6. Sharan Suresh Wadhwani1,
  7. Dhiraj Tripathi1,3,
  8. Brinder Singh Mahon1
  1. 1Liver Unit, University Hospitals Birmingham NHS Foundation Trust, Birmingham, UK
  2. 2Department of Gastroenterology, Walsall Healthcare NHS Trust, Walsall, UK
  3. 3Institute of Immunulogy and Immunotherapy, University of Birmingham, Birmingham, UK
  1. Correspondence to Professor Dhiraj Tripathi; dhiraj.tripathi{at}


Background and aims Bleeding from parastomal varices causes significant morbidity and mortality. Treatment options are limited, particularly in high-risk patients with significant underlying liver disease and other comorbidities. The use of EUS-guided embolisation coils combined with thrombin injection in gastric varices has been shown to be safe and effective. Our institution has applied the same technique to the treatment of parastomal varices.

Methods A retrospective review was performed of 37 procedures on 24 patients to assess efficacy and safety of EUS-guided injection of thrombin, with or without embolisation coils for treatment of bleeding parastomal varices. All patients had been discussed in a multidisciplinary team meeting, and correction of portal hypertension was deemed to be contraindicated. Rebleeding was defined as stomal bleeding that required hospital admission or transfusion.

Results All patients had significant parastomal bleeding at the time of referral. 100% technical success rate was achieved. 70.8% of patients had no further significant bleeding in the follow-up period (median 26.2 months) following one procedure. 1-year rebleed-free survival was 80.8% following first procedure. 7 patients (29.1%) had repeat procedures. There was no significant difference in rebleed-free survival following repeat procedures. Higher age was associated with higher risk of rebleeding. No major procedure-related complications were identified.

Conclusions EUS-guided thrombin injection, with or without embolisation coils, is a safe and effective technique for the treatment of bleeding parastomal varices, particularly for patients for whom correction of portal venous hypertension is contraindicated.


Data availability statement

Data are available upon reasonable request.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Bleeding from parastomal varices causes significant morbidity for patients with a stoma and liver cirrhosis.

  • Definitive treatment options for portal hypertension are not available for all patients due to comorbidities.

  • Treatment of gastric varices using endoscopic ultrasound (EUS)-guided embolisation coils and human thrombin has been established as an alternative to cyanoacrylate glue.


  • We report a new technique for management of parastomal varices using EUS-guided embolisation.


  • Embolisation of parastomal varices may reduce bleeding and comorbidity for patients with portal hypertension.


Portal hypertension in cirrhosis results from increased intrahepatic resistance and increased portal inflow due to splanchnic vasodilatation. Clinically significant portal hypertension develops at a hepatic venous pressure gradient of ≥10 mm Hg and signals the development of varices and bleeding. Parastomal varices form when splanchnic veins from the portal venous system are surgically juxtaposed with the systemic veins of the abdominal wall. Bleeding from parastomal varices (also referred to as stomal or peristomal varices) has been reported in up to 5% of people with a stoma and 27% of the stoma patients with liver cirrhosis.1 2 Recurrent bleeding causes significant morbidity, and mortality from haemorrhage has been reported in 2.6%–4% of patients undergoing treatment.3 4

Liver transplantation to treat the underlying liver disease is only available for a minority of patients; however, portal pressure can be reduced or normalised by creating a shunt between the portal and central venous systems. These patients often have multiple comorbidities and surgical portosystemic shunt formation is associated with perioperative mortality of 10%–20%5 and has been largely superseded by minimally invasive transjugular intrahepatic portosystemic stent shunt (TIPSS) with or without embolisation.6

Where a portosystemic shunt is contra indicated, or where bleeding persists despite normalisation of the portosystemic gradient, treatment is directed towards the varices at the stoma site. In acute haemorrhage, first-line, non-operative interventions such as compression, epinephrine-soaked gauze, ligation and cautery may be effective, and can be used in combination with medical therapy such as non-selective beta-blockade.3 Surgical revision of a stoma, with mucocutaneous disconnection, is performed in some cases; however, these patients are often high-risk surgical candidates as described above. Fluoroscopic-guided embolisation of parastomal varices with coils has been described using a transjugular approach (either alone7 or in combination with TIPSS8), via direct puncture of the liver in percutaneous transhepatic obliteration9 10 or via accessing the efferent systemic vein in balloon-occluded retrograde transvenous obliteration.11

Utilisation of endoscopic ultrasound (EUS) for the treatment of gastric varices is increasingly established, although its use in parastomal varices is very limited.12 13 Worldwide, the most commonly used obliterative agent for varices is cyanoacrylate (N-butyl-2-cyanoacrylate; Histoacryl ‘glue’); however, embolisation coils are increasingly considered in gastric varices, with evidence that a combination of coil and cyanoacrylate is superior to either coils or cyanoacrylate alone.14 Human thrombin has been proposed as an alternative to cyanoacrylate, and used with and without coils in gastric varices.15–18

Our institution has successfully treated parastomal varices with human thrombin injection±embolisation coil(s), under EUS guidance to treat parastomal varices. A retrospective review was performed to assess the outcomes from this technique.


The aim of this study is to assess safety and efficacy of EUS-guided management of parastomal varices.



Procedural and follow-up data for all EUS-guided interventions for bleeding parastomal varices from January 2014 to January 2020 at a regional liver transplant centre in the UK were retrospectively analysed. All procedures were performed by one of two experienced endosonographers or a trainee fellow under their direct supervision. The project was registered and approved as an audit following local policy.

Patients were identified using the Radiology Information System. Procedure reports and electronic medical records such as clinic letters and discharge summaries were reviewed to obtain clinical information. Model for End-stage Liver Disease (MELD) scores were calculated using blood test results at the time of procedure.19 For rebleeding, where no date of admission was documented, the date of clinic letter was used as surrogate.

Preprocedure assessment

All cases were referred to the tertiary-care liver unit at our institution, mostly from other hospitals in the region, although a minority of patients were referred from within our own Trust. All had refractory bleeding; patients with acute severe bleeding were transferred under inpatient care and treated as emergent, with more stable cases admitted electively. All patients had portal venous CT scan to assist in management planning. All cases were discussed in liver multidisciplinary team meetings to review treatment options depending on history, comorbidities and portal vein anatomy. Where the risk of TIPSS or an alternative interventional radiology approach was deemed to be too high or not technically feasible, patients were referred for an EUS-guided treatment. Patients were consented for risk of bleeding, non-target embolisation (including intestinal ischaemia, stroke and pulmonary embolism), unsuccessful procedure, and repeat intervention including endovascular or surgical approach.

Standard procedure

Most procedures are performed without sedation or analgesia, but intravenous midazolam and fentanyl are available if required. The room was set up to allow the endoscopist to comfortably support and manipulate the lower end of the endoscope with their right hand throughout the procedure. Thrombin 500IU3/ml in prefilled syringes (Tisseel®, Baxter Healthcare, Newbury, UK) is prepared by defrosting over 10 min.

With the patient supine, the linear echoendoscope (GF-UCT240; Olympus, Southend, UK) in combination with a Hitachi Aloka ultrasound platform (ProSound F75 or Aloka Alpha 10; Hitachi Medical Systems, Wellingborough, UK) is inserted into the stoma. A deep vein or veins usually feeding a network of varices are identified, often within 2–5 cm of the stoma opening but no deeper than 10 cm (figure 1). The dominant feeding vessel diameter is measured, with Doppler imaging used to confirm flow direction and measure flow rate. Those feeding vessels deemed to have relatively high flow are considered for a two-step approach to thromboembolisation, with initial slowing of blood flow by coil placement, and subsequently completion of embolisation with thrombin.

Figure 1

Feeding vessel identified using colour Doppler imaging.

The needle is prepared by flushing with saline to avoid introduction of air and degradation of the ultrasound image. A 19-gauge vascular access needle (EchoTip Ultra, Cook Medical, Limerick, Ireland) is used for 0.036″ embolisation coils (Nester, Cook Medical, Limerick, Ireland), or a 22-gauge EUS needle is used for 0.018″ microcoils or thrombin injection alone. The needle is placed in the target vessel and test injection of 10 mL saline is performed to ensure in the needle tip is intravascular (figure 2).

Figure 2

Needle (arrows), with tip in feeding vessel.

The initial coil diameter selected is slightly oversized by approximately 20%–25% relative to the diameter of the target vessel to be obliterated, in order to minimise the risk of coil migration and non-target embolisation. Any subsequent coil(s) placed can be of smaller diameter. The coil holding device is attached to the needle channel, the coil is loaded into the needle using the needle stylet, and the holding device is removed. Then coil is then delivered into the vessel using the needle stylet, carefully monitored under EUS imaging to ensure the entire length of coil is deployed within the vessel (figure 3, online supplemental video 1). Luminal flow is reassessed and subsequent smaller coils may be used to further reduce blood flow. Some loss of visibility of the distant wall of the vein should be expected, caused by acoustic shadowing from the coil(s).

Figure 3

Coil deployment under endoscopic ultrasound guidance, with annotated image showing location of the needle (parallel lines) and coil (single line) within the vessel.

Human thrombin 500 IU3/mL (Tisseel; Baxter Healthcare, Newbury, UK) is injected in 1 mL aliquots, until visible clot formation is seen obliterating the vessel lumen (figures 4 and 5).

Figure 4

Injection of thrombin into a vessel showing colour Doppler flow (no coil in vessel) with immediate formation of echogenic thrombus in the vessel lumen (arrowheads).

Figure 5

Complete occlusion of feeding vessel with thrombus (arrowheads) after coil embolisation and thrombin injection.

Example EUS images are shown in figures 1–5 and online supplemental video 1.

Post procedure

Patients are advised to stay in bed for at least 4 hours post procedure and observed for up to 12 hours. Paracetamol is prescribed as required for analgesia, and the patient is discharged the following day if well.

Routine reassessment with EUS is not usually required due to the relatively small size of parastomal varices and low risk of catastrophic haemorrhage (in contrast to follow-up of gastric varices).


Significant rebleeding (SR): postprocedure stomal bleeding episodes requiring hospital admission or blood transfusion.

Mild rebleeding: postprocedure stomal bleeding documented but not requiring hospital admission or blood transfusion.

Procedural (technical) success: the ability to obliterate the vessel on direct EUS view at the time of the procedure.

Treatment success: no significant rebleeding in the follow-up period.

Statistical analysis

Statistical analysis was performed using IBM SPSS Statistics for Windows, V.25 (IBM, Armonk, USA). Kaplan-Meier estimator was used to analyse the SR-free survival, with log-rank test to compare between groups and Cox’s proportional hazards model used to assess correlation with continuous variables. HRs with 95% CIs were calculated for each variable, and the result was considered statistically significant if p<0.05. Continuous data were reported as median (IQR) values. Mann-Whitney U test was used to compare continuous variables, as appropriate. Categorical data were reported as proportions, and comparisons were made using either Fisher’s exact test or χ2 test, as appropriate.


Baseline characteristics

In total, 37 procedures were identified on 24 patients. Only 3 patients (12.5%) were referred from within our own institution, with 21 (87.5%) referred from other hospitals.

Background demographics and medical history are shown in table 1. The most common cause for cirrhosis was primary sclerosing cholangitis and is associated with inflammatory bowel disease, which was the most common underlying cause for the stoma in this series.

Table 1

Baseline characteristics

Six patients had concurrent esophagogastric varices, with three of these patients having had previous banding of oesophageal varices as primary prevention. No recorded instances of previous upper gastrointestinal haemorrhage were identified. Most patients (70.8%) had been prescribed a non-selective beta-blocker (carvedilol or propranolol) to reduce the risk of variceal haemorrhage. One patient had a patent covered TIPSS in situ at the time of procedure, and another had an occluded TIPSS. All other patients were documented as being unsuitable for TIPSS—either due to risk of encephalopathy, or due to previously failed attempt at TIPSS insertion. Median MELD Score at the time of procedure was 12.

Previous blood transfusion for parastomal variceal bleeding was recorded in 19 of 24 patients (79.1%) prior to their first procedure EUS-guided intervention.

Procedural details

Of the 37 procedures, 18 (48.6%) were performed as emergent, with treatment of acute haemorrhage being the most common reason for admission (table 2). Patients undergoing elective procedures were admitted for routine overnight observation following the procedure. Based on preprocedural blood results, three patients required blood products (fresh frozen plasma and/or platelets) for correction of coagulopathy (international normalised ratio (INR)>1.4, platelets<50×109 /L) prior to the procedure.

Table 2

Summary of procedural details

Sedation was used as required in a minority of procedures, depending on patient anxiety and discomfort. Intravenous midazolam was used in five cases (maximum 2.5 mg) and fentanyl was used in two cases (maximum 100 µg).

The median diameter of the largest vessels treated in the group receiving combination therapy with coils and thrombin was larger than the group receiving thrombin alone, reflecting operator preference to use coils in larger, high-flow vessels. The median diameter of the largest coil used was 8 mm, slightly larger than the median diameter of the largest vessel in this group (6 mm), also reflecting operator preference to oversize the coils in order to prevent non-target embolisation.

Clinical outcomes

All procedures reported technical success in identifying and thrombosing a variceal vessel as imaged on EUS.

Length of follow-up was defined by the period between the date of procedure and the date of the most recent follow-up letter. Median follow-up was 26.2 months (range 0–46 months, IQR 10.2–33.2 months).

Following their first procedure, SR occurred in 7 of 24 patients (29.1%) during the follow-up period, at a median of 10.8 months following the procedure. Seven patients had a second procedure, and three of these patients (42.9%) had further SR, at a median of 13.5 months post procedure. Three patients had a third procedure, and one patient had five procedures in total (figure 6, table 3).

Figure 6

Flowchart of rebleeding and reintervention. Numbers denote number of patients at each step. EUS, endoscopic ultrasound.

Table 3

SR occurrences by procedure number

Kaplan-Meier analysis was performed to assess SR-free survival (figure 6). One-year SR-free survival was 80.8% following first procedure, 83.3% following second procedure, and 76.6% with all procedures included (figure 7). Log-rank test did not show significant difference when comparing first, second and third procedures (p=0.132). With only a single fourth and fifth procedure recorded (performed on the same patient), meaningful SR-free survival for these cases could not be determined.

Figure 7

Significant rebleeding (SR)-free survival following endoscopic ultrasound-guided embolisation of stomal varices, all procedures included. Censored points denote time of last available follow-up information or death.

SR-free survival for categorical data is shown in table 4. No statistically significant differences were identified between groups, except for longer estimated mean SR-free survival in patients below the median age of 59.0 (figure 8).

Figure 8

Significant rebleeding (SR)-free survival, patients grouped above or below median age (59 years) at time of procedure. Censored points denote time of last available follow-up information or death. Statistically significant difference between groups (p=0.023).

Table 4

Comparison of categorical variables using Kaplan-Meier estimation to determine SR-free survival and log-rank analysis to determine significance

HRs for continuous variables are shown in table 5. The only statistically significant correlation was that of age, which is associated with an increased risk of SR of 1.072 for every additional year of age.

Table 5

Comparison of continuous variables


Postprocedure pain was recorded in 3 of 37 discharge summaries (8.1%). Postprocedure fever was recorded on 2 (7.4%)—both were treated with antibiotics and were otherwise uneventful—all other elective procedures were discharged on day 1 post procedure (median stay 1 day). Median length of stay for emergency cases was 3 days, with no discharge delays relating to the endoscopic procedure identified. One patient had three procedures within 3 months for recurrent bleeding. After the third procedure, the patient had further bleeding but was treated with palliative intent only and died 1 month later. Three other patients died within the follow-up period due to underlying liver disease—one due to metastatic colorectal cancer, one with PSC, previous transplant and portal vein stenosis (MELD Score 28), and one patient with primary biliary cholangitis, awaiting liver transplant, (MELD Score 13) with unknown cause of death.


Bleeding from parastomal varices is a relatively rare complication, but it poses a unique challenge in management. TIPSS has been found to be effective in achieving haemostasis, with one study recording a rebleeding rate of only 5% in parastomal varices, although post-TIPSS hepatic encephalopathy manifested or worsened in 30% of patients20 and direct procedural mortality from TIPSS is 1%–2%.21 22 Guidelines from both the British Society of Gastroenterology6 15 and the American Association for the study of liver diseases23 recommend TIPSS for control of variceal haemorrhage, but recognise the need to tailor treatment for individual patients.


The use of cyanoacrylate glue in the treatment of gastric varices has been associated with serious thromboembolic events,24 with one series reporting pulmonary emboli being in 4% of cases.25 In addition, glue can damage the endoscope if not administered appropriately, which adds complexity to the procedure, particularly for clinicians who do not use such adhesives on a regular basis. Thrombin has been suggested as an alternative to cyanoacetate glue with the benefit of being technically easier to use and potentially having fewer adverse events.16 26 One recent randomised controlled trial in acute gastric variceal haemorrhage has shown similar effectiveness between cyanoacrylate and thrombin, with reduced complications in the thrombin group.27 Although there are no dedicated studies in parastomal varices, there are potential advantages to using thrombin over cyanoacrylate.

The technique of EUS embolisation described has advantages over fluoroscopically guided interventional radiology procedures to embolise parastomal varices. The risks associated with vascular access (transjugular or transhepatic) are avoided, and procedure time and cost are likely to be lower with EUS. Concurrent fluoroscopy could be performed with EUS technique to confirm there is no migration of the coil from the target vessel; however, in our experience, B-mode and Doppler ultrasound are sufficient to visualise coil placement within the varix safely and efficaciously. Obtaining simultaneous fluoroscopic views would not justify the additional time, expense, complexity and radiation exposure required. If coil migration is deemed to be high risk, fluoroscopic imaging might be deemed necessary.


There was a high burden of liver disease among the patients in this study, with median MELD Score of 12 associated with 3-month mortality of 6.0%28 and a quarter of patients having MELD Score >20. All had significant stomal bleeding prior to the procedure and recurrence is not unexpected when the underlying portal hypertension has not been corrected. Overall, 100% of patients had required recent admission to hospital for bleeding or documented blood transfusions prior to their first procedure.

Our results showed 100% technical success over 37 procedures, which demonstrates that the EUS method is able to reliably identify a target vessel, insert coil(s) where required and inject thrombin to thrombose the vessel.

SR-free survival of 80.8% at 1 year following first procedure suggests a significant improvement in the degree of bleeding from parastomal varices. In patients who did suffer from recurrent bleeding, median time to documentation was 10.8 months. These results demonstrate a reduction in severe bleeding at least the short-to-medium term following embolisation. There was no significant difference in SR-free survival following first, second or third intervention, which suggests that repeat procedures are also effective.

Oey et al20 reported an estimated SR-free survival rate of 77% at 1 year for all ectopic variceal bleeding treated with TIPSS, using similar end-points to this study. However, only 1 of 21 patients in their series with parastomal varices had significant rebleeding. The numbers of patients included in both studies are limited, and the results are not directly comparable, but they suggest that where TIPSS is feasible it may be superior to EUS-guided embolisation.

There was a higher level of rebleeding among older patients. There are multiple age-related physiological factors which could influence recurrence rates (eg, changes in the vascular wall) but the underlying mechanism for this relationship is unknown.

In our experience, we have not identified any vessels previously treated with coils which showed flow on subsequent procedures (recanalisation). It is more difficult to identify in retrospect vessels previously treated with thrombin alone; however, our experience suggests that recurrence is predominantly due to the development of new collaterals rather than recanalisation of treated varices, although this could be investigated in future studies.


In our series of 37 procedures, there were two instances of postprocedural fever, which were treated with antibiotics. The aetiology of this is unclear; however, both patients were treated with antibiotics. Three patients suffered postprocedural pain, which was self-limiting. It is possible that a mild inflammatory response should be expected following venous thrombosis, as recognised in deep venous thrombosis and arterial embolisation procedures.29 30 In patients who were in-patient at the time of procedure, median time to discharge was 3 days and for elective cases the median stay was 1 day, as per our protocol.

No serious complications were identified in our cohort of patients, and no stomal complications such as necrosis, ulceration or stricturing were reported. Such stomal complications have been reported with the use of sclerosants such as sodium morrhuate or sodium tetradecyl sulphate in parastomal varices.3 31


This is a retrospective study with inherent limitations in the completeness and accuracy of data collection. There is no control group of patients who did not undergo the procedure, and benefit can only be inferred by comparing the severity of bleeding before and after the procedure.

This study does not compare EUS-guided embolisation with other embolisation techniques or occlusive agents; however, the potential advantages of this particular technique are outlined. As with any novel technique, there is some difficulty in sharing knowledge, which may limit the uptake of this technique in other centres. In particular, the decisions of whether or not to use coils, and what size of coils to use are based on endosonographer’s experience, depending on the size of the vessel and flow rate.


To our knowledge, EUS-guided embolisation of parastomal varices using thrombin has not previously been described. In the context of a difficult to treat problem in complex patients who are unable to undergo more definitive procedures, our results suggest this technique can provide a significant benefit at least in the short-term to mid-term. Importantly, the procedure is safe, minimally invasive and well-tolerated, allowing for multiple treatments if required. Further prospective studies shall help to establish its role in standard of management of bleeding parastomal varices.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This study was approved as a service evaluation project by University Hospitals Birmingham NHS Foundation Trust Clinical Audit Registration & Management System.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • DT and BSM are joint senior authors.

  • Contributors BSM designed and oversaw the study. AWMT performed the data collection with contribution from CS and JO. AWMT, BSM and DT analysed the data. DT is the guarantor. All authors wrote the manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests DT has received income from speaker fees and consultancy work with W,L. Gore and Associates. Other authors declared no competing interests.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.