Abstract
BACKGROUND: The model for end-stage liver disease (MELD) score is widely used for the prognostication in end-stage liver disease but has limited performance in acute-on-chronic liver failure (ACLF). In this study, we identified additional predictive parameters and reformed the MELD score to predict ACLF more accurately. METHODS: A meta-analysis was performed on relevant studies to identify the predictive factors of 28-day/90-day outcomes of ACLF, which were validated in two large prospective cohorts. A prognostic score was developed by incorporating predictive parameters into the MELD score. The model was evaluated with a focus on discrimination and calibration. RESULTS: The meta-analysis incorporated 32 cohort studies with a total of 13 939 patients, of which 13 risk factors were identified, and 3 risk factors (age, neutrophil count and hepatic encephalopathy (HE) grade) besides MELD score were validated in 751 patients with ACLF derived from two prospective cohorts. A new model (Chinese Acute-on-Chronic Liver Failure Consortium (CATCH-LIFE)-MELD score) was developed as follows: 0.028×age+0.3×HE grade+0.039×neutrophil count+0.079×MELD score. CATCH-LIFE-MELD score achieved a concordance index of 0.791/0.788 for 28-day/90-day outcomes, which is superior to other traditional scores. Other discrimination indices, including net reclassification improvement, integrated discrimination improvement and probability density function, and calibration including Nagelkerke's R(2) and Brier scores confirmed its superiority. Moreover, the accuracy of CATCH-LIFE-MELD score remained stable. It was highest in patients with or without hepatitis B virus infection, cirrhosis, liver failure or under the Chinese Group on the Study of Severe Hepatitis B (COSSH) criteria or European Association for the Study of the Liver (EASL) criteria. All results were substantiated by an evaluation using an external cohort. CONCLUSIONS: CATCH-LIFE-MELD score, a modified MELD score exhibited improved accuracy in predicting the short-term prognosis of ACLF than other traditional scores.