Performance of scoring systems to predict mortality of patients with acute-on-chronic liver failure: A systematic review and meta-analysis

Background and Aim: Acute-on-chronic liver failure (ACLF) has characteristic feature of multisystem organ failure, rapid progression, and low early transplant-free survival. We performed a meta-analysis to determine the accuracy of five scoring systems in predicting mortality of ACLF patients. Methods: A systematic database search was performed, and retrieved articles were graded according to methodological quality. Collated data was meta-analyzed by hierarchical summarized receiver operating characteristic model and bivariate model to evaluate the diagnostic accuracy of scoring systems. Results: Of 4223 studies identified, 26 studies involving 4732 ACLF patients were included. The model of end-stage liver disease (MELD) score was found to have largest the area under summarized receiver operating characteristic (AUROC) (0.82) compared with other estimated scoring systems, especially for 3-month mortality. MELD serum sodium (MELD-Na) score showed homologous high accuracy, with the AUROC was 0.81. However, meta-analyses of 16 studies showed that Child-Pugh-Turcotte score had least AUROC (0.71). Sequential organ failure assessment (SOFA) score presented moderately lower diagnostic accuracy, with AUROC being 0.73. Moreover, chronic liver failure-SOFA score presented excellent accuracy of prognostication with highest diagnostic odds ratios. Conclusion: This review demonstrated that MELD had moderate diagnostic accuracy to predict mortality of ACLF patients. Considering the expectative diagnostic value, chronic liver failure-SOFA could be regarded as a promising replacement of MELD. To improve the predictive power of scoring systems, multicenter prospective studies of large sample sizes with long-term follow-up are needed.