Improved Discrimination and Predictive Ability of Novel Prognostic Scores for Long-term Mortality in Hospitalized Patients with Cirrhosis.

Abstract

Background and aims: Since the adoption of novel prognostic scores, such as the iterative model for end-stage liver disease (MELD 3.0) and the gender-equity model for liver allocation (GEMA), their utility has markedly expanded to diverse clinical scenarios. However, data concerning their prognostic value in more generalized cirrhotic populations are scarce. In this study, we aimed to elucidate the MELD 3.0/GEMA-Na for long-term mortality risk stratification and refine their usage scope.

Methods: This study retrospectively reviewed 310 hospitalized patients with decompensated cirrhosis. Discrimination and stratification were compared between MELD 3.0/GEMA-Na and other scores. Validation was performed in another 120 subjects.

Results: In the investigated cohort, the median MELD-Na, MELD 3.0, and GEMA-Na were 9 (7, 12), 12 (10, 17), and 12 (9, 17), respectively. Compared to their predecessors, both MELD 3.0 and GEMA-Na models exhibited consistently better discriminative ability, especially in relation to long-term mortality. This effect was more pronounced for GEMA-Na, which was the only score to present an area under the receiver operating characteristic curve greater than 0.8 up to two years (0.807). Statistical analysis indicated that a MELD 3.0 score of 18 and a GEMA-Na score of 20 were the most optimal cutoffs to rank the risk of death, both of which were independently associated with two-year all-cause transplant-free mortality (MELD 3.0: hazard ratio: 1.13, 95% confidence interval: 1.10, 1.17; GEMA-Na: hazard ratio: 1.12, 95% confidence interval: 1.10, 1.17, both P < 0.001). Similar findings were affirmed in the validation cohort.

Conclusions: MELD 3.0 is superior to other MELD-based scores for long-term prognostication in hospitalized patients with cirrhosis, while GEMA-Na demonstrated even better accuracy and performance.