The LiverTox database compiles cases of idiosyncratic drug-induced liver injury (iDILI) with the promised aims to help identify hepatotoxicants and provide evidence-based information on iDILI. Weaknesses of this approach include case selection merely based on published case number and not on a strong causality assessment method such as the Roussel Uclaf Causality Assessment Method (RUCAM). The aim of this analysis was to find out whether the promised aims have been achieved by comparison of current iDILI case data with those promised in 2012 in LiverTox. First, the LiverTox criteria of likelihood categories applied to iDILI cases were analyzed regarding robustness. Second, the quality was analyzed in LiverTox cases caused by 46 selected drugs implicated in iDILI. LiverTox included iDILI cases of insufficient quality because most promised details were not fulfilled: (1) Standard liver injury definition; (2) incomplete narratives or inaccurate for alternative causes; and (3) not a single case was assessed for causality with RUCAM, as promised. Instead, causality was arbitrarily judged on the iDILI case number presented in published reports with the same drug. All of these issues characterize the paradox of LiverTox, requiring changes in the method to improve data quality and database reliability. In conclusion, establishing LiverTox is recognized as a valuable effort, but the paradox due to weaknesses between promised data quality and actual data must be settled by substantial improvements, including, for instance, clear definition and identification of iDILI cases after evaluation with RUCAM to establish a robust causality grading.
Keywords: DILI database case quality; LiverTox; Roussel Uclaf Causality Assessment Method (RUCAM); iDILI; idiosyncratic drug-induced liver injury.