External validation of the SALT-M score in living donor liver transplantation for acute-on-chronic liver failure.

BACKGROUND & AIMS: The Sundaram Acute-on-Chronic Liver Failure-Liver Transplant Mortality (SALT-M) score predicts 1-year post-transplant mortality in patients with severe ACLF undergoing deceased donor liver transplantation (DDLT). Its performance in living donor liver transplantation (LDLT) is unknown.

METHODS

We externally validated SALT-M in two independent ACLF-LDLT cohorts: King Faisal Specialist Hospital and Research Centre (KFSHRC), Riyadh (n=121), and a global International LDLT Registry cohort (n=57). Discrimination (AUROC with 95% CI), calibration (observed-to-expected [O:E] ratio, calibration slope and intercept), and association with observed outcomes were assessed overall and, in the KFSHRC cohort, separately in ACLF grade 1 and severe ACLF (grades 2-3).

RESULTS

In the KFSHRC cohort, ACLF grade was 1 in 47.9%, 2 in 43.0%, and 3 in 9.1%; 1-year mortality was 14.9% (global cohort 21.1%). SALT-M predicted ≤30% mortality in 95.9% of the KFSHRC and 98.2% of the global cohort. Discrimination was limited and statistically inconclusive: AUROC 0.56 (95% CI 0.40-0.72) overall, 0.62 (0.42-0.81) in severe ACLF, and 0.51 (0.25-0.76) in ACLF-1. Calibration-in-the-large was acceptable (O:E 1.18, 95% CI 0.74-1.87) but the calibration slope was shallow (0.16). None of the five SALT-M components was associated with 1-year mortality; white blood cell count (OR 1.14, p=0.03) and female sex (OR 3.49, p=0.02) were. MELD-Na and MELD 3.0 were higher in non-survivors in the global cohort (p=0.03 and p=0.02).

CONCLUSIONS

In LDLT-dominant ACLF cohorts - evaluated largely outside the severe-ACLF population for which SALT-M was derived - discrimination could not be confirmed, reflecting profound spectrum restriction and few outcome events rather than demonstrated model failure. SALT-M should be applied cautiously as an adjunctive risk-stratification tool, not as a universal threshold for transplant candidacy.