MELD score < 18 rule out 28-day ACLF development among inpatients with hepatitis B-related previous compensated liver disease.

The acute-on-chronic liver failure (ACLF) development is highly dynamic. Currently, no satisfactory algorithm identifies patients with HBV at risk of this complication. The aim of the study was to characterize ACLF development in hospitalized HBV-related patients without previous decompensation and to test the performance of traditional prognostic models in ruling out ACLF development within 28 days on admission we conducted a cohort study. Two multi-center cohorts with hospitalized HBV-related previous compensated patients were analyzed. Performances of MELD, MELD-Na, CLIF-C AD, and CLIF-C ACLF-D in ruling out ACLF development within 28 days were compared and further validated by ROC analyses. In the derivation cohort (n = 892), there were 102 patients developed ACLF within 28 days, with profound systemic inflammatory levels and higher 28-day mortality rate (31.4% vs. 1.0%) than those without ACLF development. The MELD score (cut-off = 18) achieved acceptable missing rate (missed/total ACLF development) at 2.9%. In the validation cohort (n = 1656), the MELD score (<18) was able to rule out ACLF development within 28 days with missing rate at 3.0%. ACLF development within 28 days were both lower than 1% (0.6%, derivation cohort; 0.5%, validation cohort) in patients with MELD < 18. While in patients with MELD ≥ 18, 26.6% (99/372, derivation cohort) and 17.8% (130/732, validation cohort) developed into ACLF within 28 days, respectively. While MELD-Na score cut-off at 20 and CLIF-AD score cut-off at 42 did not have consistent performance in our two cohorts. MELD < 18 was able to safely rule out patients with ACLF development within 28 days in HBV-related patients without previous decompensation, which had a high 28-day mortality.