A 50-year-old Caucasian male was admitted with a 2-day history of melena. His medical history was significant for liver cirrhosis secondary to autoimmune hepatitis, for which he underwent living donor right lobe liver transplantation in 2008. His posttransplant course was unremarkable, and he remained free of acute cellular rejection or recurrence of autoimmune hepatitis after liver transplantation. His immunosuppressant regimen included tacrolimus (1.5 g twice daily) and prednisone (5 mg daily).
On the day of admission, the patient was slightly tachycardic but hemodynamically stable. The physical examination revealed that the spleen was palpated 3 cm below the costal margin, whereas no hepatomegaly was appreciated. Laboratory data were notable for a hemoglobin level of 5.4 g/dL in the setting of normal liver function tests, a normal platelet count, and a normal coagulation panel.
After adequate resuscitation and blood transfusion, the patient underwent upper endoscopy (Fig. 1A). Because of suspected gastric varices on endoscopy, a transjugular hepatic venogram with portosystemic pressure measurements was performed. The hepatic venogram demonstrated no evidence of filling defects, focal stenosis, or web formations. The portal-atrial pressure gradient was 4 mm Hg. Additionally, a direct portal pressure measurement was obtained in the portal vein with a percutaneous approach. The direct portal pressure ranged from 7 to 8 mm Hg. Liver biopsy was performed and revealed minimal chronic inflammation with no evidence of acute cellular rejection. The trichrome stain highlighted no fibrosis. No Schistosoma eggs were identified.
Figure 1A reveals what appears to be gastric varices in the fundus of the stomach. A careful review of the endoscopic images revealed several features atypical for gastric varices: the submucosal vascular lesions were more scattered than would be expected with fundal varices secondary to portal hypertension, and there did not appear to be evidence of background portal hypertensive gastropathy.
Because of this as well as the absence of any liver disease since the patient's liver transplantation 4 years earlier, we were hesitant to proceed directly to a glue injection or transjugular intrahepatic portosystemic shunt without further information.
There was no evidence of portal hypertension from the wedge portal pressure measurements or the direct portal pressure measurements. The latter were obtained because of the concern that the patient might have intrahepatic presinusoidal portal hypertension, which might not be reflected by his wedge portal pressure measurements.
The triphasic computed tomography images in Fig. 1B,C reveal that the splenic artery was occluded approximately 2 to 3 cm distal to its ostium. The spleen measured 18.2 cm. Figure 1D shows several prominent gastrohepatic and perigastric collateral vessels that were identified. The celiac artery angiogram shown in Fig. 1E confirmed the occlusion of the splenic artery. Additionally, the angiogram revealed multiple enlarged vessels arching off the left gastric artery and the hepatic artery. These vessels collateralized around the medial gastric fundus (which corresponded to the location of the lesion seen on upper endoscopy) to supply blood to the reconstituted distal splenic artery (Fig. 1F).
The final diagnosis was submucosal fundic arterial lesions instead of gastric varices. These lesions represent prominent collateral arteries that arise as a result of splenic artery occlusion. Indeed, a careful review of the original living donor liver transplantation that was performed in 2008 revealed that the splenic artery was ligated during the surgery because of excessive portal inflow into a small-sized liver graft to prevent small-for-size syndrome.
Upper gastrointestinal bleeding secondary to prominent collateral arteries is rare, and there is no clear-cut recommendation for its management. Cyanoacrylate, the therapeutic modality of choice for gastric varices, is associated with a significant risk of systemic embolization and has the potential for devastating adverse events if it is directly injected into arterial lesions. Arterial angiography with embolization of the bleeding collaterals can be used to control the bleeding. The disadvantage of this therapeutic modality is that embolization of the visceral arteries can induce ischemia in the abdominal viscera (particularly the stomach and pancreas), which may necessitate an emergent surgical intervention. Therefore, it is advisable to superselect the bleeding arterial branches and to insert the embolization catheter as distally as possible before the embolization materials are injected. Splenectomy and devisualization of the greater curvature of the stomach constitute a definitive therapy for bleeding secondary to prominent collateral arteries. However, this carries a high risk of surgical morbidity and mortality.
Transcatheter embolization of the left gastric artery was performed via a coaxial microcatheter with both Gelfoam slurry and pledgets; this produced a marked decrease in the anterograde flow and a pruned appearance for the collaterals. Three days later, the patient experienced another episode of melena along with a significant drop in hemoglobin. Upper endoscopy confirmed that the fundic collaterals were the source of the bleeding. After resuscitation and blood transfusion, the patient was brought to the operating room for splenectomy. The patient tolerated the surgery well with no recurrence of gastrointestinal bleeding postoperatively.
Upper gastrointestinal bleeding secondary to prominent collateral arteries is rare, with only 4 reported cases in the English literature. Two cases involved gastric bleeding from a prominent submucosal short gastric artery in the context of a congenital absence of the splenic artery.[2, 3] Those were successfully managed by splenectomy, although partial gastrectomy in addition to splenectomy was necessary to control the bleeding in one case. Two other cases involved this phenomenon in the setting of an occluded splenic artery.[4, 5] In one case, the bleeding was successfully controlled after splenectomy. Radiographic arterial angiography with embolization was sufficient in the second case with no need for surgical intervention.
Splenic artery ligation is commonly performed today, especially during living donor liver transplantation, to prevent portal hyperperfusion and small-for-size syndrome. The significance of the current case is that it involved a life-threatening, long-term complication of splenic artery ligation after liver transplantation. A large study, possibly combining data from multiple centers, is needed to explore the outcomes of splenic artery ligation after liver transplantation and to determine whether gastrointestinal bleeding from arterial collaterals is common after this procedure.
Furthermore, the current case involved submucosal gastric arterial lesions with an endoscopic appearance similar to that of gastric varices. Cyanoacrylate, the therapeutic modality of choice for gastric varices, is associated with a significant risk of systemic embolization and has the potential for devastating adverse events if it is directly injected into arterial lesions. Therefore, a high level of diagnostic certainty of gastric varices is required before one proceeds with cyanoacrylate therapy.