FIPS AUC 0.821 outperforms MELD scores for 90-day mortality prediction in TIPS for ascites

This is a meta-analysis of studies evaluating prognostic scores for patients undergoing transjugular intrahepatic portosystemic shunt (TIPS) procedures for ascites or gastroesophageal bleeding. The analysis pooled data from a total sample size of 5,180 patients. The setting was not reported in the abstract. The primary intervention was the comparison of Model for End-Stage Liver Disease (MELD)-based scores, including MELD, MELD-Na, and MELD 3.0, against the Freiburg index of post-TIPS survival (FIPS). The comparator was the performance of MELD-based scores versus the FIPS index. The primary outcome was the area under the curve (AUC) for 90-day mortality prediction. The follow-up period for all outcomes was 90 days.

For patients with ascites, the pooled AUC for 90-day mortality prediction with the standard MELD score was 0.703 (95% CI, 0.606-0.800). The MELD-Na score had a pooled AUC of 0.699 (95% CI, 0.570-0.828). The MELD 3.0 score showed an AUC of 0.790 (95% CI, 0.689-0.873). The FIPS index demonstrated the highest AUC for ascites at 0.821 (95% CI, 0.656-0.985). For patients with gastroesophageal bleeding (GEB), the MELD score AUC was 0.827 (95% CI, 0.740-0.914). The MELD-Na score AUC was 0.781 (95% CI, 0.689-0.873). The MELD 3.0 score AUC was 0.797 (95% CI, 0.641-0.953). In contrast, the FIPS index had a lower AUC of 0.689 (95% CI, 0.580-0.797) for GEB.

The meta-analysis did not report key secondary outcomes. Safety and tolerability data, including adverse event rates, serious adverse events, and discontinuations, were not reported in the abstract. The analysis noted that subgroup differences were not significant (P = .412 for ascites; P = .274 for GEB), indicating that the observed variations in score performance between indications may not be statistically robust.

These results can be compared to prior landmark studies and guidelines on TIPS prognostication, which have often relied on MELD-based scores. This meta-analysis suggests that the FIPS index may offer improved discrimination for ascites, while MELD scores remain strong predictors for gastroesophageal bleeding. The performance of MELD 3.0 showed modest improvement over earlier versions in both indications.

Key methodological limitations include sources of heterogeneity, though specific sources were not detailed in the abstract. The analysis is a meta-review and does not report individual study designs, settings, or potential biases from the primary studies. The lack of reported safety data is a significant limitation for clinical application.

Clinically, these findings suggest that score selection for risk stratification before TIPS should be indication-specific. For ascites, FIPS may provide better discrimination, while MELD-based scores are effective for gastroesophageal bleeding. However, the non-significant subgroup differences advise cautious interpretation.

Unanswered questions include the generalizability of these findings to broader populations, the impact of score selection on clinical decision-making and outcomes, and the validation of FIPS in prospective cohorts. Future research should address these gaps and incorporate safety data.

In summary, this meta-analysis synthesizes evidence on prognostic scores for TIPS patients, highlighting indication-specific performance differences. The results support tailored risk assessment but underscore the need for further validation and comprehensive safety reporting.