244 - Prognostic Value of Acute vs Baseline MELD Scores in Predicting Mortality After Emergent TIPS

Purpose: The aim of this study was to assess whether acute MELD scores (Original, MELD-NA, and MELD 3.0) obtained at the time of emergent TIPS provide superior prognostic accuracy for 3-month mortality compared with baseline MELD values, with the goal of improving risk stratification in urgent clinical decision-making.

Materials and Methods: We conducted a single-center retrospective cohort study of adults undergoing transjugular intrahepatic portosystemic shunt (TIPS) from 2004–2024. The patient data was obtained from the EMR and PACS. There were 545 TIPS encounters identified, of which 109 met our operational definition of emergent TIPS (acute hemorrhagic presentation with day-of-procedure labs). Our primary cohort comprised 54 emergent cases with both a “baseline” laboratory panel (−6 months to −2 weeks pre-procedure) and an “acute” panel (day of TIPS). The primary outcome was 3-month all-cause mortality and exposures were MELD (original), MELD-Na, and MELD-3.0. We fit separate univariable logistic regression models for each score at the baseline and acute timepoints to estimate the odds of 3-month mortality. We reported odds ratios and ROC/AUC with 95% CIs. Given only eight events, we limited models to single predictors to reduce the risk of overfitting. We summarized model fit with Tjur and Cox–Snell pseudo-R² and analysis was completed in GraphPad Prism.

Results: Among 54 TIPS patients there were 8 deaths at 3 months. Simple logistic regressions using acute MELD values found a significant association with 3-month mortality for MELD-Original (OR per point 1.133, 95% CI 1.037–1.257; AUC 0.781, 95% CI 0.564–0.999), MELD-Na (OR 1.131, 95% CI 1.038–1.25; AUC 0.774, 95% CI 0.556–0.993), and MELD-3.0 (OR 1.134, 95% CI 1.040–1.259; AUC 0.772, 95% CI 0.559–0.985); all p< 0.02 for AUC and p≤0.006 for slope. In contrast, baseline MELD values were not significantly associated with mortality (all slope p≥0.18; AUCs 0.677–0.696 with p≥0.08).

Conclusion: Our findings indicate that acute MELD scores provide better discrimination for 3-month mortality when compared to baseline measures across MELD-Original, MELD-Na, and MELD-3.0 in this cohort. This will help guide clinical decision making, risk stratification, and informed consent.