The primary objective of the study is to demonstrate the superiority of an "early tips" strategy over standard treatment by glue obliteration (G0) in preventing bleeding recurrence or death at one year after a non GOV1 gastric variceal bleeding in cirrhotic patients initially treated by GO.

Inclusion Criteria:

• Cirrhotic patients: the diagnosis of liver cirrhosis will be based on previous needle liver biopsy or on the combination of clinical, biochemical, and radiological findings. If biopsy findings are unavailable and in case of non-complicated cirrhosis, non-invasive markers will be used.
• Variceal bleeding at endoscopy from gastroesophageal gastric varices type 2 or isolated gastric varices type 1 or 2 (Sarin classification) according to the following criteria: endoscopic signs of an active spurting or oozing from gastric varices (GV); adherent blood clots, white nipple signs, or erosions on the GV and absence of other bleeding sources.
• Hemodynamically stable patient (Mean arterial pressure above 65 mmHg) without clinical significant rebleeding (Baveno criteria) within 12 hours after the initial endoscopy with glue obliteration.
• Written informed consent obtained.

Exclusion Criteria:

• Pregnant woman or breastfeeding.
• Minor and patients older than 75 years.
• Non cirrhotic portal hypertension.
• Hepatocellular carcinoma outside the Milan criteria or other cancer at a palliative stage.
• Child Pugh score > 13.
• History of severe or refractory hepatic encephalopathy unrelated to gastrointestinal bleeding.
• Congestive heart failure.
• History or presence of pulmonary hypertension.
• Patients with other indication for TIPS.
• Uncontrolled gastric variceal bleeding.
• Portal vein cavernoma.
• Patient who have previously received a TIPS procedure.
• Failure to receive clear information in patients without an identified trusted person.
• Refusal of the participation agreement by signing the information form and consent as defined.
• Exclusion period from another biomedical study.