Abstract
BACKGROUND AND AIMS: Approximately 10% of patients with acute decompensated (AD) cirrhosis develop acute-on-chronic liver failure (ACLF) within 28 days. Such cases have high mortality and are difficult to predict. Therefore, we aimed to establish and validate an algorithm to identify these patients on hospitalization. METHODS: Hospitalized patients with AD who developed ACLF within 28 days were considered pre-ACLF. Organ dysfunction was defined according to the chronic liver failure-sequential organ failure assessment (CLIF-SOFA) criteria, and proven bacterial infection was taken to indicate immune system dysfunction. A retrospective multicenter cohort and prospective one were used to derive and to validate the potential algorithm, respectively. A miss rate of <5% was acceptable for the calculating algorithm to rule out pre-ACLF. RESULTS: In the derivation cohort (n=673), 46 patients developed ACLF within 28 days. Serum total bilirubin, creatinine, international normalized ratio, and present proven bacterial infection at admission were associated with the development of ACLF. AD patients with ≥2 organ dysfunctions had a higher risk for pre-ACLF patients [odds ratio=16.581 95% confidence interval: (4.271-64.363), p<0.001]. In the derivation cohort, 67.5% of patients (454/673) had ≤1 organ dysfunction and two patients (0.4%) were pre-ACLF, with a miss rate of 4.3% (missed/total, 2/46). In the validation cohort, 65.9% of patients (914/1388) had ≤1 organ dysfunction, and four (0.3%) of them were pre-ACLF, with a miss rate of 3.4% (missed/total, 4/117). CONCLUSIONS: AD patients with ≤1 organ dysfunction had a significantly lower risk of developing ACLF within 28 days of admission and could be safely ruled out with a pre-ACLF miss rate of <5%.