Look to the Left: A Rare Case of Recurrent Upper GI Bleeding

We present the case of a 63 year old male who developed recurrent massive gastric varicose vein bleeding episodes years after necrotising pancreatitis. Endoscopy revealed extensive gastric but no esophageal varices. Ultrasonographic findings included an enlarged spleen, normal portal vein diameter and flow and a regular liver texture without signs of cirrhosis. CT scan demonstrated massive perigastric and splenic collateral veins due to chronic splenic vein occlusion. Left-sided portal hypertension was diagnosed and the patient treated with splenectomy. In an endoscopic 2-year follow-up, gastric varicose veins have completely regressed and the patient has remained free of recurrent GI bleeding complications.


Case Report
We report on a rare case of recurrent gastric variceal bleeding in a 63 year old male with a history of necrotising pancreatitis in 2005 who underwent pancreatico-jejunostomy for large pancreatic pseudocysts as well as jejuno-jejunostomy for gastrointestinal stroma tumor in 2006. In December 2010, the patient presented twice with massive upper gastrointestinal (GI) bleeding due to extensive gastric varices, 14 days apart. Both episodes could be stabilized endoscopically using histoacryl sclerotherapy. Amongst others, paraclinic abnormalities included anaemia (hemoglobin 5.2 mmol/l), thrombocytopenia (58 GPt/l) and leucopenia (1.87 GPt/l).

Clinical course and results of imaging procedures
Endoscopy revealed a portal-hypertensive gastropathy as well as extensive gastric varices at the great gastric curvature but no esophageal varices. Ultrasonography demonstrated a normal portal vein diameter and flow as well as a regular liver texture without signs of cirrhosis. On the other hand, ultrasound revealed an enlarged spleen (160 × 70 mm) and a CT scan demonstrated massive perigastric and splenic collateral veins due to chronic thrombotic splenic vein occlusion. Of note, splenomegaly was not suspected during clinical examination and was only found in imaging. Furthermore, re-evaluation of the CT scan from 2005 demonstrated splenic vein compression by a large pancreatic pseudocyst (Figure 1a-1c). Despite successful histoacryl treatment of both upper GI bleedings the patient remained unstable with recurrent minor bleeding episodes. Within 6 days of the last sclerotherapy, the patient developed a third, massive gastric variceal bleeding (Figure 2a and 2b), which led to hemorrhagic shock and required repeated histoacryl injections, terlipressin infusions and transfusions of 16 units red blood cell concentrates, 2 units of platelets, 5 units of fresh frozen plasma, 2.000 units of prothrombin complex concentrate and 2500 units of factor XIII concentrate in ICU, following our institutional transfusion protocol.

Diagnosis
We diagnosed recurrent gastric variceal bleeding and hypersplenism  or splenic vein thrombosis and anticoagulation should be considered in the early phase to achieve recanalization of the affected vein to prevent development of collateral veins [14,15]. In our case, anticoagulation was not provided in the acute phase of splenic vein thrombosis, since the thrombotic occlusion remained undiagnosed. During our treatment, anticoagulation was considered to be ineffective due to chronic vein occlusion and dangerous because of recurrent acute variceal bleeding. Transjugular portosystemic shunt (TIPS) as an option for treating recurrent variceal bleeding16 was not indicated, since neither cirrhosis nor portal vein hypertension was present [16].

Conclusion
Patients with isolated gastric varicose veins should be assessed for signs of splenomegaly or hypersplenism and splenic vein obstruction. LSPH should be suspected, if splenomegaly or hypersplenism are present in the absence of portal vein occlusion or liver cirrhosis. Presence of collateral veins in splenic vein occlusion is an indicator of chronic disease and correlates to an increased bleeding risk. As a consequence, anticoagulant treatment will have little benefit but could potentially increase bleeding risk and intensity. If acute gastric variceal bleeding occurs in LSPH patients, splenectomy can drastically reduce the risk of bleeding recurrence and, due to the reduced splenic consumption seen in hypersplenism, normalize concomitant thrombocytopenia.
due to left-sided portal hypertension (LSPH), based on the findings of an occluded splenic vein with pronounced collateral veins as well as splenic enlargement with thrombocytopenia (58 GPt/l) and leukopenia (1.87 GPt/l) without portal hypertension or liver cirrhosis.

Treatment
Due to three episodes of massive upper GI bleeding within 3 weeks, interdisciplinary consensus led to a splenectomy after interventional coil occlusion of the splenic artery. During surgery the chronic thrombotic occlusion of V. lienalis was confirmed and histology revealed massive vascular dilatation of intrasplenic vessels ( Figure  2c). Following an uncomplicated post-operative period the patient was discharged on day 7, at which time both platelet count and leucocyte count had returned to normal.

Clinical follow up 2 years later
Recently, the patient presented for an endoscopic 2-year followup. After splenectomy, the massive gastric varicose veins have completely regressed (Figure 2d). Hemoglobin (10.4 mmol/l), platelet and leucocyte count (256 and 9.6 Gpt/l, respectively) were still within normal range and since the splenectomy, the patient has remained free of recurrent GI bleeding complications.

Discussion
Left-sided portal hypertension (LSPH), also referred to as sinistral portal hypertension, is a segmental form of portal hypertension usually due to an isolated splenic vein obstruction [1][2][3]. Since splenic vein occlusion often develops without specific symptoms, the incidence remains unclear but it is suggested to be present in less than 5% of all patients with portal hypertension [1] and to be mainly related to splenic vein thrombosis in acute or chronic pancreatitis or malignant diseases [4][5][6][7].
Compensation of splenic vein occlusion and increased venous pressure consists of splenic vein drainage through short gastric veins with the consequence of varicose vein formation and risk of rupture [8]. Diagnosis is usually made in acute upper GI bleeding situations when massive and often recurrent gastric variceal bleeding manifest [9]. On the other hand, incidences of gastric variceal bleeding in LSPH vary between 4 and 25%, probably due to the low number of patients included in these analyses [3,[8][9][10].
Another diagnostic criterion is the finding of sonographic and paraclinic signs of hypersplenism (spleen enlargement, thrombocytopenia, leucopenia) [8,11] in the absence of portal vein occlusion or liver cirrhosis. Of note, our case did not have clinically apparent splenomegaly, which was only confirmed by imaging. Consequently, the imaging procedures for patients with esophageal or gastric varices should not only focus on liver cirrhosis but should include a screening for splenomegaly ("look to the left"). Acute therapy of gastric variceal bleeding consists of histoacryl sclerotherapy [12], but endoscopic treatment may only be effective in the acute setting and not over long. Recurrent bleeding episodes, as in our case, are frequent with rates higher than 50% [12], due to the persistent increase of venous pressure. Definitive treatment may therefore only be achieved by splenectomy, which can reduce the rates of recurrent bleeding complications below 1% [6,8,11]. In contrast, the benefit of prophylactic splenectomy for incidental finding of gastric varicose veins without bleeding events is unclear [3,5,13].
Several treatment options exist for patients with variceal bleeding. The underlying pathomechanism of portal hypertension often is portal