04 versus before embolization). Eight patients of 37 were still in need of nonabsorbable antibiotics compared to 17 before embolization. One patient was successfully transplanted after embolization because of persisting bouts of encephalopathy. Univariate analysis of a wide spectrum of biochemical and clinical parameters identified sex, time interval between diagnosis of HE and SPSS, serum albumin, International Normalized Ratio (INR), the presence of ascites preembolization, hemoglobin level, Child and MELD score (all with P < 0.05) as predictors of HE recurrence CDK inhibitor post-SPSS-embolization. After weighing these different variables to exclude multicollinearity,

logistic regression ascertained the following parameters to be predictive of HE recurrence BI 6727 mw postembolization: sex (odds ratio [OR] 0.06, 95% confidence interval [CI] 0.005-0.971, P = 0.048) and MELD preembolization (OR 1.52, 95% CI 1.073-2.180, P = 0.019). We further evaluated the discrimination ability of the MELD score in predicting HE recurrence after SPSS embolization by using the area under ROC curve. The MELD score showed good accuracy

to discriminate between patients with recurrence or not (95% CI = 0.637- 0.914, P < 0.0001). Using the Youden index, the best cutoff point for the MELD score was 11 with a sensitivity and specificity of 68.4% and 77.6%, respectively (Fig. 5). Overall there were eight early procedure-related complications, of which seven were mild and symptomatically treated (one cutaneous infection at the puncture site, one contrast-induced nephropathy, three hematomas limited to the puncture site, and two self-limiting episodes of fever). One patient had a capsular bleeding due to a transhepatic approach complicated with hypovolemic shock for which surgical hemostasis was needed. All complications were nonlethal and without permanent

injury or morbidity. With regard to long-term complications, we observed no significant aggravation of portal hypertension during follow-up. More specifically, there was no increase postembolization in the number of patients with gastroesophageal varices (48.6 versus 52%, P = 0.94) or with portal hypertensive gastropathy (50 versus 66%, P = 0.18). Two patients developed de novo esophageal varices (grade 1 and grade 2, respectively). Overall, one patient with preexisting varices experienced a nonfatal variceal bleeding SB-3CT at 55 months postembolization which was managed by combined pharmacological and endoscopic intervention. There was no difference with respect to the number of patients with ascites (13/37 pre- and 15/37 postembolization, P = 0.92). In the postembolization group, 6 of 15 patients developed de novo ascites (of which five were patients with recurrent HE). From these 15 patients, seven were in need of large-volume paracentesis (of which six were also nonresponders to embolization) and two developed spontaneous bacterial peritonitis during follow-up.