Evaluation of available scoring systems in liver transplantation for severe acute-on-chronic liver failure (ACLF).

Abstract

The Transplantation for ACLF-3 Model (TAM) and Sundaram ACLF-LT-Mortality score (SALT-M) have been proposed to help select liver transplantation (LT) candidates with acute-on-chronic liver failure (ACLF) to optimize post-LT outcomes. We assessed the performance and clinical utility of these scores and their components on a large retrospective international multicenter study. All patients transplanted in the context of ACLF grade 3 (n=336) from 9 international cohorts (2008-2024) were included in the study. The primary endpoint was 1-year post-LT survival. Score discrimination, calibration, and clinical utility using decision-curve (net benefit) analysis were assessed. The 1-year post-LT survival rate was 75.3%. Both TAM (OR 1.91, p =0.0001) and SALT-M (OR 1.09, p <0.0001) were associated with 1-year mortality but showed only modest discriminative performance (AUROC 0.61 and 0.64, respectively). Clinically relevant candidate high-risk thresholds were identified: TAM ≥3 (1-year survival 42.9% vs. 79.0%) and SALT-M >25 (46.9% vs. 78.7%). Respiratory failure (PaO 2 /FiO 2 ≤200) was associated with 1-year post-LT mortality, whereas other variables included in the original scores, diabetes, BMI, and leukocyte count, did not demonstrate significant prognostic value in this cohort. In this large multicenter study, TAM and SALT-M were associated with post-LT mortality but demonstrated limited discriminative accuracy. Candidate high-risk thresholds (TAM ≥3, SALT-M >25) may help identify a small subset of patients with markedly poor post-LT outcomes. However, given their modest performance, these scores should support, not replace, multidisciplinary decision-making, and their applicability should be evaluated in a prospective fashion.