Variceal bleeding (VB) refers to bleeding from abnormally dilated, fragile varices in the esophagus or stomach. Varices usually develop due to liver cirrhosis and the resultant portal hypertension (PHT). The varices are fragile and prone to rupture, often leading to life-threatening bleeding.

Atypical or ectopic VB refers to an unusual VB outside the esophagus or stomach, including the duodenum, jejunum, and colon. Atypical VBs, which are less common, are also typically caused by PHT. However, atypical varices may not be readily detected by conventional endoscopic techniques, potentially resulting in diagnostic and therapeutic delays. Treatment of atypical VB may include endoscopic therapy to locate and ligate bleeding varices, or surgery to remove the affected segment of the intestine. In some cases, radiological interventions such as embolization or shunt procedures may be used to redirect blood flow around the damaged veins.

Although liver cirrhosis is adults’ most common cause of PHT, extrahepatic portal vein obstruction (EHPVO) can also cause PHT and a life-threatening VB [1]. EHPVO is a condition in which the portal vein (PV) is occluded outside the liver, leading to PHT. The etiology of EHPVO is diverse; however, most pediatric cases are idiopathic in nature [1,2]. Risk factors include perinatal events, such as umbilical vein catheterization, omphalitis, perinatal sepsis, and PV anomalies, as well as thrombophilia resulting from prothrombin mutation or deficiency of protein C, protein S, and antithrombin III. Although primary thrombosis of the PV can be a causal factor in adults, EHPVO develops more commonly after biliary surgery or pancreaticoduodenectomy (PD). Causative factors include lymph node dissection, inflammation, complicated fluid collection due to anastomotic leakage or infection, and tumor recurrence. The gold standard diagnostic modalities for EHPVO are imaging studies, such as computed tomography (CT) or magnetic resonance imaging. Both can visualize the PV occlusion and the resultant cavernous transformation of the porta hepatis.

Afferent loop varices and bleeding are serious complications of hepaticojejunostomy (HJ), which is a common surgical procedure performed to restore bile flow after choledochal cyst excision, bile duct stone removal, or bile duct or pancreatic cancer surgery. If EHPVO develops after the HJ, the portal flow is reconstructed through venous drainage in the afferent loop jejunum around the HJ anastomosis site, serving as the only hepatopetal flow. Due to the severe adhesion around the HJ site and the engorgement of multiple variceal vessels, a direct surgical approach to the HJ is often impossible and dangerous.

After pancreatic surgery, EHPVO is not uncommon, often attributable to inflammation associated with complicated fluid collection or an increased incidence of oncovascular surgery with vessel reconstruction during tumor resection [3]. Kang et al. [4] showed that PV stenosis/occlusion is common after PD, particularly after PV resection or chemoradiotherapy. While tumor recurrence was the most common cause, PV stenosis/occlusion was detected in 17.3% of the patients without disease recurrence.

Improvements in surgical techniques and adjuvant therapies have inadvertently led to an increased incidence of VB due to EHPVO, as patient survival rates after pancreatic cancer surgery continue to improve. Although gastroesophageal VB is often effectively managed with endoscopic treatment and radiological embolization, HJ bleeding presents a significant therapeutic challenge. Herein, the various treatment results are summarized in a literature review.

The diagnosis of atypical VB is difficult because of its intermittent nature and a small amount of herald bleeding. The symptoms typically include abdominal pain, distention, and gastrointestinal bleeding with melena. Patients typically have a history of HJ as part of surgery for choledochal cysts or periampullary cancer. Contrast-enhanced CT is the standard diagnostic imaging modality, showing occlusion of the extrahepatic PV and cavernous transformation around the HJ site in post-PD patients (Fig. 1, 2) [5]. The most common CT imaging pattern of EHPVO in children was occlusion of the PV bifurcation (48.4%), followed by occlusion of the main PV (29%) [6].

Figure 1. Preoperative computed tomography scan demonstrating a cavernous formation around the porta hepatis (A) and portal vein thrombosis (B). Adapted from the article by Oh et al. (Vasc Specialist Int 2020;36:186-192) [5].

Figure 2. Preoperative computed tomography scan revealing the obliteration of the portal vein (thin arrows in coronal view [A] and axial view [B]). Gastric and hepaticojejunostomy varices (thick arrows in [A]) due to portal hypertension were observed. Adapted from the article by Cho and Min (Vasc Specialist Int 2022;38:27) [36].

Routine colonoscopy or esophagogastroduodenoscopy is frequently ineffective in identifying the offending VB. Diagnosing bleeding from the afferent loop presents a notable challenge because of technical difficulties caused by the angulation of the jejunojejunostomy, long limb length, and the presence of adhesions. Some studies have reported anecdotal success in approaching the HJ with adult variable stiffness colonoscopy [7], single balloon endoscopy (SBE) [8,9], or double balloon endoscopy (DBE) [10,11] for the Roux-en-Y afferent loop.

Treatment of HJ bleeding depends on its severity and location. Mild bleeding may resolve spontaneously. However, severe or persistent bleeding requires prompt medical attention. Although the existing evidence on the efficacy of medical treatment for EHPVO-related bleeding remains inconclusive, consensus guidelines suggest that recommendations for acute VB associated with cirrhosis should be applied [12,13]. Initial management should prioritize stabilizing the patient’s condition and addressing any accompanying symptoms, such as pain. Blood or blood products should be transfused as needed to restore the lost volume; however, this is somewhat conservative because volume overload increases portal pressure. Antibiotics or anti-inflammatory medications may be prescribed to reduce inflammation and prevent further bleeding if bleeding is associated with an infection or inflammation.

In the context of acute VB, administering vasoactive drugs to stimulate splanchnic vasoconstriction, thereby reducing portal pressure, is essential. Therapeutic agents such as terlipressin, somatostatin, or octreotide may be used and maintained for up to five days [12,14]. If terlipressin is not available, vasopressin and transdermal nitroglycerin can be used as an alternative. Nonselective beta-blockers, including propranolol, nadolol, and carvedilol, serve as the cornerstones of secondary prophylaxis. Beta-blockers reduce portal pressure by concurrently blocking β1-adrenoreceptors to reduce cardiac output and β2-adrenoreceptors in the splanchnic vessels to induce vasoconstriction.

In some cases, endoscopic therapies, such as electrocautery, argon plasma coagulation, or hemostatic clips, can be used to control bleeding. This approach involves inserting a flexible tube with a camera and a small instrument into the patient’s body to locate the source of the bleeding and administer the treatment. Due to the technical difficulties in reaching the HJ stomy, only a few reports are available in the literature. Prachayakul et al. [15] reported successful treatment of HJ bleeding with Histoacryl injection using SBE. Marinoni et al. [16] reported successful hemostasis using DBE with a dual-emission laser after argon plasma coagulation and endoclipping.

Due to its minimally invasive nature, PV stenting is the preferred treatment modality for PV recanalization. Depending on the anatomy, steno-occlusive PV lesions can be accessed percutaneously via a trans-splenic or transhepatic route or surgically through the ileocolic vein. The insertion of a stent or stent graft across the steno-occlusion site can cure this disease. Lee et al. [17] reported the successful treatment of EHPVO with portal stenting, with no rebleeding during a 32-month follow-up period. However, stenting failed in another case and required a meso-caval shunt for treatment. Ota et al. [18] reported a case of successful treatment of HJ bleeding using stenting. PV stenting is less invasive and can restore adequate portal flow with a reduction in blood flow to the collateral vessels.

However, PV stenting is not always possible and has several limitations.

(1) Technical failure due to chronic occlusion is common. PV stenosis or early thrombosis is a good indication for PV stenting; however, passing a guidewire through a chronic thrombotic lesion is extremely challenging because of tight fibrotic occlusion.

(2) The long-term patency of portal stents remains undefined [19,20]. Moreover, the results of PV stenting are controversial depending on the clinical setting: after liver transplantation (LT) vs. after periampullary cancer surgery. Some studies have reported good patency of stenting after LT [19,21], while others have reported poor stenting outcomes after other hepato-biliary-pancreatic surgeries [20,22,23]. If the stent is occluded, extensive portomesenteric venous occlusion often develops, leading to VB.

(3) An optimal stent design is yet to be established. Covered stents can be used in some malignant PV steno-occlusions to prevent tumor ingrowth. However, it carries the risk of occluding the major portal branches or splanchnic veins. Therefore, bare stents are preferred. Recently, dedicated venous stents, typically used for iliofemoral deep vein-dedicated stents, have become available; however, their effectiveness in PV thrombosis remains uncertain.

(4) Optimal antithrombotic medication after stenting has not yet been established. Anticoagulation with warfarin is generally recommended to prevent early stent thrombosis, despite the lack of robust evidence [19,24-26]. However, early initiation of anticoagulation therapy may expose the patient to risks such as puncture site bleeding, gastrointestinal bleeding, and heparin-induced thrombocytopenia.

(5) The percutaneous transhepatic or transsplenic procedure carries risks of various complications, such as puncture site pain, mild fever, liver abscess, hepatic artery pseudoaneurysm, subcapsular hematoma, bile leakage from an intrahepatic duct, PV thrombus during venoplasty, and intraperitoneal hemorrhage. Given its technical demands, PV interventions should be undertaken in high-volume centers with sufficient expertise and experience.

The second interventional treatment for HJ bleeding involves the transcatheter embolization of jejunal veins supplying the afferent loop. Theoretically, this method is unrealistic and challenging because of the abundant collaterals at the porta hepatis, potential liver dysfunction caused by the blockage of the hepatopetal flow, and inevitable recurrence. However, if combined with PV stenting or shunt surgery, this approach can be particularly effective in directly halting HJ bleeding, thereby preventing early recurrent bleeding. Furthermore, Kato et al. [23] reported that embolization of collateral flow during PV stenting can improve stent patency.

Shim et al. [20] reported a case series of PV stenting with and without varix embolization. Stenting was technically successful in 19 of 22 (86.4%) patients. Bleeding cessation without recurrence within 1 month was achieved in 18 of 19 patients (94.7%). One patient with recurrent bleeding after 4 days required additional variceal embolization. During the 258 days of follow-up, stent occlusion developed in 6 of the 19 (31.6%) patients, and recurrent bleeding developed in 4 of the 19 patients (21%).

1) Direct approach to HJ anastomosis

Although resection of the HJ anastomosis and re-anastomosis have been reported [27], this is not the preferred surgical treatment because of the associated risks, including massive intraoperative bleeding, difficulties in dissecting the fibrotic adhesion, difficulty in re-anastomosing the HJ, and recurrence of jejunal varices.

We previously attempted to access the anterior wall of the HJ and control the VB by using direct suture ligation of the varices. However, the procedure was unsuccessful. HJ bleeding caused by hepatopetal collateral growth in the submucosa of the jejunum was diffuse and multiple. Moreover, the swollen mucosa frequently lacerates, making secure anastomosis difficult. In addition, because PHT is not controlled with this method, recurrence is inevitable.

2) Nonphysiologic porto-systemic shunt

Shunt operations create a shortcut to decompress the PHT. These surgeries can be classified as either “nonphysiological”—diverting blood flow away from the liver, or “physiological”—redirecting blood flow back to the liver.

Portosystemic shunts (PSSs) are non-physiological routes created to divert portal and mesenteric blood flow into the systemic circulation. The net result of this diversion is a reduction in pressure and the subsequent decompression of the portal venous system [9]. Distal spleno-renal shunt (DSRS), the preferred procedure for children with PHT at our center, is a selective PSS that has proven effective and reliable [28]. However, PSSs carry the potential risk of reduced liver perfusion, which may worsen liver function and increase the risk of hepatic encephalopathy.

There have been reports of atypical VB treated with shunt surgeries. Paquet et al. [29] reported successful meso-caval shunt treatment of the afferent loop VB with no recurrence for 24 months. Lee et al. [17] reported a case of a patient treated with a meso-caval shunt after failed portal stenting. However, the patient experienced rebleeding a month later despite a patent shunt. The patient was treated conservatively and has had no further bleeding for 8 months.

Despite these successful cases, shunt operations carry the risks of shunt occlusion, hepatic encephalopathy, and operative morbidities.

3) Meso-Rex shunt

The meso-Rex shunt (MRS) is an attractive treatment for EHPVO that involves shunting the mesenteric vein to the intrahepatic left portal vein (LPV) within the Rex recess using an autologous vein graft. MRS is the best option for children with EHPVO because it can perfuse the liver to allow growth [30].

MRS, the most physiological shunt for EHPVO, was first described in 1992 by de Ville de Goyet et al. [31] and was originally designed to treat post-LT patients with PVT, but has also been successfully applied to non-transplant cases. The operation consists of bypassing the EHPVO area by creating a conduit between the mesenteric venous system and the LPV within the Rex recess, which is a part of the left portal system that runs sagittally within the umbilical fissure between Couinaud segments II, III, and IV (Fig. 3). This shunt reduces portal pressure, restores physiological portal flow toward the liver (hepatopetal), and can ultimately alleviate the side effects of PHT or portosystemic connections [2]. Studies have shown that it provides additional metabolic benefits over PSSs by restoring normal portal venous circulation in the liver. Furthermore, it can reverse coagulopathy, hyperammonemia, hepatopulmonary syndrome, and encephalopathy and improve neurocognitive ability, nutrition, and somatic growth [32,33].

Figure 3. Intraoperative findings depicting the procedure of the modified coronary vein-left portal vein (LPV) shunt. (A) Coronary vein isolation. (B) LPV isolation. (C) Distal end-to-end anastomosis of the great saphenous vein (GSV) graft to the LPV. (D) Proximal end-to-end anastomosis of the coronary vein to the GSV. Adapted from the article by Oh et al. (Vasc Specialist Int 2020;36:186-192) [5].

MRS is superior to DSRS for the treatment of childhood EHPVO by improving liver function [34]. Khamag et al. [35] mentioned that DSRS controls VB, but should only be considered when MRS is not technically feasible or as a salvage procedure when MRS fails.

We previously reported successful cases of MRS in children and adults [5,36]. The success of MRS relies on several key factors, including the absence of intrinsic liver parenchymal disease, patent LPV, suitable mesenteric venous inflow, and the availability of an adequate autologous conduit [4]. Therefore, thorough preoperative imaging and meticulous operative techniques are imperative for successful MRS surgery.

The internal jugular vein is the most frequently used conduit for MRS. Alternative conduits such as the great saphenous vein (GSV), prosthetic grafts, splenic, coronary, and umbilical veins have also been used [37-39]. The traditional MRS procedure involves shunting the SMV to the LPV using the internal jugular vein, whereas a modified Rex shunt employs an alternative conduit or inflow, including the coronary vein, which also demonstrates good patency (Fig. 4); however, the debate about the best conduit remains [39].

Figure 4. Follow-up computed tomography after two years displaying complete patency of the great saphenous vein graft (thick arrows) and coronary vein (thin arrows). Adapted from the article by Cho and Min (Vasc Specialist Int 2022;38:27) [36].

The proposed requisites for MRS include (1) EHPVO with patent SMV or other suitable inflow; (2) patent LPV in the Rex recess and patent intrahepatic portal tree; (3) absence of intrinsic liver disease; and (4) patent both internal jugular veins, as one of them is utilized as a venous graft [27,32].

Wu et al. [40] reported that preoperative CT evaluation before MRB is vital. The authors classified the LPV in the Rex recess according to its diameter on CT scan (type 1, diameter ≥5 mm; type 2, diameter 2-5 mm; type 3, <2 mm; and type 4, with PV within the Rex recess undistinguishable). They suggested that MRS may be successful as long as the PV in the Rex recess is displayed on CT. In type 4, poor visualization was related to true aplasia of the LPV in some patients, but in more than half of the cases, it was due to insufficient filling of the PV as a consequence of PHT, and they might be amenable to MRS. Thus, those with type 4 LPV in the Rex recess need further evaluation of the patency of the PV system with other imaging modalities, such as wedged hepatic vein portography. In scenarios where the Rex recess is deemed unsuitable, the DSRS shunt may be considered the next viable option. A simplified treatment algorithm for VB caused by EHPVO is shown in Fig. 5.

Figure 5. Treatment algorithm for atypical variceal bleeding due to extrahepatic portal vein obstruction. EHPVO, extrahepatic portal vein obstruction; PV, portal vein; CT, computed tomography; USG, ultrasonography; LPV, left portal vein.

Oncovascular surgery and neoadjuvant chemoradiotherapy can achieve curative resection of locally advanced pancreatic cancers that invade the portomesenteric vein/hepatic artery [3]. This finding suggests the possibility of long-term survival in these patients. However, some patients develop EHPVO after undergoing PD, which ironically results in more patients with VB from the HJ site returning to the hospital.

Once EHPVO develops, VB can be caused by esophagogastric or jejunal varices. This underscores the importance of PV reconstruction during initial surgery. Many of the tumor-involving vein segments were primarily repaired during operation. However, when tension-free anastomosis is not possible, spiral grafts with GSV or bovine patch grafts are good options for reducing stenosis and overcoming the size mismatch between the autologous and portomesenteric veins [41].

Kang et al. [4] reported a post-PD PV stenosis/occlusion rate of 19.6% during a median follow-up of 22.9 months, with a 5-year patency rate of 69.9%. PV steno-occlusion was particularly common if the PV had been resected or if postoperative chemoradiotherapy had been administered. Therefore, PV reconstruction during the initial operation should be carefully performed using a multidisciplinary team approach, including vascular surgeons familiar with the natural history of the disease and the techniques required for safe PV reconstruction.

Atypical variceal bleeding due to EHPVO is a life-threatening complication after PD, and very difficult to treat. Vascular surgeons should know the pathophysiology, diagnosis, and management strategies of this rare disease.

The treatment modalities range from endoscopic, radiologic, and surgical interventions, including shunt surgery. MRS is particularly promising for mitigating the hemodynamic and metabolic impacts of EHPVO.

Because the treatment is complex, prevention of PV occlusion is important. Therefore, PV reconstruction during the initial operation should be carefully performed using a multidisciplinary team approach, led by an oncovascular surgeon.

Ahram Han and Seung-Kee Min have been the editorial board members of the VSI since 2019. They were not involved in the review process. Otherwise, no potential conflict of interest relevant to this article was reported.

Ahram Han and Seung-Kee Min have been the editorial board members of the VSI since 2019. They were not involved in the review process. Otherwise, no potential conflict of interest relevant to this article was reported.

Concept and design: SKM. Writing the article: all authors. Critical revision of the article: all authors. Final approval of the article: all authors. Statistical analysis: none. Obtained funding: none. Overall responsibility: all authors.