Abstract
This meta-analysis compares the prognostic accuracy of Model for End-Stage Liver Disease-based scores (MELD) and the Freiburg Index of Post-TIPS Survival (FIPS) for predicting mortality in patients undergoing transjugular intrahepatic portosystemic shunt (TIPS) for ascites or gastroesophageal bleeding (GEB) and examines the sources of heterogeneity.
The study followed PRISMA guidelines and searched major biomedical databases for retrospective studies reporting area under the curve (AUC) based mortality prediction.
This meta-analysis included 11 studies (5,180 patients). In ascites, pooled AUCs for 90-day mortality were 0.703 (95% CI: 0.606-0.800) for MELD (6 studies) and 0.699 (95% CI: 0.570-0.828) for MELD-Na (5 studies). FIPS (2 studies) and MELD 3.0 (1 study) showed good performance: AUCs of 0.821 (95% CI: 0.656-0.985), and 0.790 (95% CI: 0.689-0.873), respectively. Subgroup differences were not significant (p=0.412). In GEB, MELD (6 studies), MELD-Na (5 studies), and MELD 3.0 (2 studies) yielded AUCs of 0.827 (95% CI: 0.740-0.914), 0.781 (95% CI: 0.689-0.873), and 0.797 (95% CI: 0.641-0.953), respectively. FIPS (4 studies) showed lower performance (AUC = 0.689; 95% CI: 0.580-0.797), with no statistically significant subgroup differences (p = 0.274).
The discriminatory performance of MELD-based scores varies according to the underlying indication for TIPS. In patients treated for GEB, MELD, MELD-Na, and MELD 3.0 showed good discriminatory capacity for early mortality (30-90 days). In contrast, in patients undergoing TIPS for ascites, MELD and MELD-Na showed only fair accuracy, with some improvement observed for MELD 3.0.