Predictive Accuracy Comparison of Prognostic Scoring Systems for Survival in Patients Undergoing TIPS Placement: A Systematic Review and Meta-analysis.

RATIONALE AND OBJECTIVES: This meta-analysis aimed to evaluate the performance of different risk assessment models (RAMs) for survival after Transjugular Intrahepatic Portosystemic Shunt (TIPS) in patients with cirrhotic portal hypertension.

MATERIALS AND METHODS: A systematic search of PubMed, WOS, Embase, Cochrane, and CNKI from inception to February 2023 was conducted. We comprehensively reviewed and aggregated data from numerous studies covering prevalent RAMs such as Child-Turcotte-Pugh, the Model for End-Stage Liver Disease (MELD), MELD-Sodium (MELD-Na), the Freiburg Index of Post-TIPS Survival (FIPS), Bilirubin-platelet, Chronic Liver Failure Consortium Acute Decompensation score, and Albumin-Bilirubin grade across different timeframes. For this study, short-term is defined as outcomes within a year while long-term refers to outcomes beyond one year. The area under the receiver operating characteristic (AUC) curve or Concordance Statistics was chosen as the metric to assess predictive capacity for mortality outcomes across six predetermined time intervals. Mean effect sizes at various time points were determined using robust variance estimation.

RESULTS: MELD consistently stood out as a primary short-term survival predictor, particularly for 1 month (± 2 weeks) (AUC: 0.72) and 3 months of (± 1 month) survival (AUC: 0.72). MELD-Na showed the best long-term predictive ability, with an AUC of 0.70 at 3.5 years (± 1.5 years). FIPS performed well for 6 months of (± 2 months) survival (AUC: 0.68) and overall transplant-free survival (AUC: 0.75). Efficacy nuances were observed in RAMs when applied to particular subgroups. Meta-regression emphasized the potential predictor overlaps in models like MELD and FIPS.

CONCLUSION: This meta-analysis underscores the MELD score as the premier predictor for short-term survival following TIPS. Meanwhile, the FIPS score and MELD-Na model exhibit potential in forecasting long-term outcomes. The study accentuates the significance of RAM selection for enhancing patient outcomes and advocates for additional research to corroborate these findings and fine-tune risk assessment in TIPS.