Portal hypertension (PHT) that leads to esophageal varices is clinically present, at the time of diagnosis, in 60% of cirrhotic patients (Groszmann NEJM 2005). Variceal bleeding is a life-threatening complication. The yearly incidence of the first variceal bleeding in cirrhotic patients is estimated to 4%. This risk increases to 15% when medium or large varices are initially present, (D'amico Baillieres Clin Gastroenterol 1997); Therefore, for patient with medium or large varices a primary prophylaxis treatment with nonselective beta-blockers must be instituted decreasing risk of bleeding approximately by half (d'amico Hepatology 2005).

The method to evaluate the efficacy of beta blockers treatment is the hepatic venous pressure gradient (HVPG). HVPG < 12 mmHg or a decrease of HVPG of 20% after beta blocker treatment reduce dramatically the risk of bleeding (Feu F. Lancet 1995). But a good response (HVPG < 12 mmHg or reduction > 20%) to beta blockers was observed only in almost 30 % of patients. In non responder patients, the variceal bleeding incidence is estimated to 24 % at 2 years.

In this situation, alternative treatments to beta-blockers have been suggested: endoscopic ligation, which was shown to be as effective as beta blockers in primary prevention, and more recently carvedilol (Reiberger, Gut 2012).

So, HVPG monitoring may provide critical information for patients. However, because of its technical requirement and its invasivity, HVPG measurement is not feasible in clinical practice . So, there is a general consensus to continue indefinitely beta blockers treatment without control of hemodynamic efficiency (merkel J. Hepatology 2009).

Previous studies have suggested the interest of azygos blood flow measurements for evaluating hemodynamic changes in the esophageal collateral vessels of patients with portal hypertension (Bosch J. Hepatology 1985 ). More recently Magnetic Resonance Imaging (MRI) has been reported to be an efficient technique to assess azygos blood flow (Gouya Radiology 2011) in cirrhotic patients. Its feasibility is obviously higher than HVPG measurement.

However, in the same way that HVPG, a large-scale implementation in clinical practice of azygos blood flow measurement by MRI requires to specify minimal absolute threshold, or relative post-therapeutic decrease, related with no variceal bleeding.

Inclusion Criteria:

Cirrhotic patients requiring β blocker in primary prophylaxis of bleeding esophageal varices

Exclusion Criteria:

• Patients under 18 year's old
• life expectancy of less than one month
• Patient already has a vasoactive treatment
• patient with CHC or portal vein thrombosis
• patient without social security or deprived of freedom
• contraindication to MRI
• contraindication to beta-blockers
• mental state does not allow the signing of an informed consent