Aspartate aminotransferase and model for end-stage liver disease reliably predict mortality in drug-induced liver injury

Abstract Drug-induced liver injury (DILI) is associated with high mortality risk. However, no biomarker can reliably predict outcome. In order to identify baseline parameters that are associated with mortality the data of 268 prospectively collected DILI patients were analyzed. Multivariate logistic regression and receiver operating characteristic curves were used to identify the parameters being most predictive for a fatal outcome, as defined by orthotopic liver transplantation (OLT) or death. 10.4% of patients had a fatal outcome, which was associated with higher levels of transaminases, total bilirubin, INR and model for end-stage liver disease (MELD) scores (25 vs. 12, p < 0.001). Multivariate analysis revealed that only MELD and aspartate aminotransferase (AST) were independently associated with a poor outcome, the MELD score in particular had an extraordinarily high c-statistic of 0.93 (95% CI: 0.87–0.97). At a cut-off of ≥ 20, MELD could predict a fatal outcome with a sensitivity and specificity of 88% and 81%. The predictive performance of the MELD score could even be enhanced by combing it with AST elevation: At a cut-off of ≥ 20 and ≥ 29.6xULN, respectively, positive and negative predictive values of 44% and 96% were observed, which were higher than for any other baseline parameter.