Change in model for end-stage liver disease score at two weeks, as an indicator of mortality or liver transplantation at 60 days in acute-on-chronic liver failure.

Background: Acute-on-chronic liver failure (ACLF) is characterised by a sudden deterioration of underlying chronic liver disease, resulting in increased rates of mortality and liver transplantation. Early prognostication can benefit optimal allocation of resources. Methods: ACLF was defined as per the disease criteria of the Asian Pacific Association for the Study of the Liver. Inpatient discharge summaries from between January 2001 and April 2013 were reviewed. The primary outcome was mortality or liver transplantation within 60 days from onset of ACLF. Absolute 'model for end-stage liver disease' (MELD) score and change in MELD at Weeks 1, 2 and 4 were reviewed in order to identify the earliest point for prediction of mortality or liver transplantation. Results: Clinical data were collected on 53 subjects who fulfilled the inclusion and exclusion criteria. At 60 days from presentation, 20 patients (37.7%) died and 4 (7.5%) underwent liver transplantation. Increased MELD of >= 2 after 2 weeks was 75.0% sensitive and 75.9% specific for predicting mortality or liver transplantation. If the MELD score did not increase at 2 weeks, predictive chance of survival was 93.8% over the next 60 days. MELD change at 1 week showed poor sensitivity and specificity. Change at 4 weeks was too late for intervention. Conclusion: Change in MELD score at 2 weeks provides an early opportunity for prognostication in ACLF. A MELD score that does not deteriorate by Week 2 would predict 93.8% chance of survival for the next 60 days. This finding warrants further validation in larger cohort studies.