[MELD score in prediction of early mortality in patients suffering refractory ascites treated by TIPS]

Transjugular Intrahepatic Portosystemic Shunt (TIPS) is now well established in the treatment of complications of symptomatic portal hypertension such as acute or recurrent variceal bleeding, refractory ascites and Budd-Chiari syndrome. In some patients with refractory ascites who belong to group C according to Child-Pugh classification (score around 12), the indication of the procedure could be very questionable and early mortality is quite high. However, in some cases, the subgroup of such risky patients can profit from TIPS. Child-Pugh classification is used for the stratification of the patients routinely. During the last decade other scoring systems occured to bring a better prognostic value. MELD (Model for End stage Liver Disease) score, based only on laboratory values is one of them. Comparison of these two scoring systems in patients treated by TIPS in previous trials brought certain discrepancy, but MELD score seems to be better in predicting early mortality. The aim of our study was to determine retrospectively the predictive accuracy of MELD score for the early mortality in comparison to Child-Pugh score in patients treated for refractory ascites by TIPS.

We evaluated 110 patients (mean age 55 years) with liver cirrhosis (61% of patients with alcoholic etiology), who underwent TIPS for refractory ascites in our center from September 1992 to December 2003. MELD and Child-Pugh score was calculated and then compared between groups with early (one month), three month and one year mortality, and those who survived over this period (one, three and twelve months), comparing MELD and Child-Pugh score (ROC analysis and Student's T test were used).

Mean follow up was 23 months. Average MELD score in the whole group was (16). In patients, who died within one month the score before TIPS was 21, three months 20 and 18 one year. Comparing MELD score between subgroups and then Child-Pugh score, only for MELD score there was a statistically significant difference (p < 0.05) in one month. Using ROC (AUC) analysis, discriminant power of MELD score was superior to Child-Pugh score for one (0.73 vs 0.63) and three month (0.73 vs 0.67) mortality. The discriminant power for one year mortality was low in both scores.

MELD scoring system is a better tool to predict the risk of early mortality in patients with refractory ascites treated by TIPS than Child-Pugh classification. The discriminant power was low in both scores in one year horizon.