Abstract
BACKGROUND AND AIMS: Acute liver injury (ALI) due to autoimmune hepatitis (AIH) can be treated by immunosuppression. In contrast, idiosyncratic drug-induced liver injury (DILI) had a poor prognosis. DILI thus needs to be distinguished from non-DILI. METHODS: Twenty-nine patients with DILI and 77 with non-DILI (42 of AIH and 35 with undetermined cause) diagnosed during 2005-2017 comprised the derivation cohort. 110 patients with ALI due to either AIH, DILI, or obscure causes at 6 liver centers during 2010-2015 were the validation cohort. Revised international AIH group scores (IAIHGs) and the Roussel-Uclaf Causality Assessment Method (RUCAM) were modified to calculate results using medical interviews and laboratory data without chronological changes. Diagnostic accuracy for the distinction of DILI and non-DILI was evaluated by receiver operating characteristic analysis and results were expressed as the area under the curve (AUC). This study received institutional institutional review board approval (MH2020-205). RESULTS: The AUCs of modified IAIHGs and RUCAM scores for the diagnosis of DILI were 0.96 and 1.00 when cut-off values were set at 3 for the modified RUCAM and 5 for the modified IAIHGs in the derivation cohort. In the validation cohort, the AUCs of modified IAIHGs and RUCAM scores for the diagnosis of DILI were 0.95 and 0.97, respectively. The accuracy of the combination of the modified scores was 81% (89/110). CONCLUSION: Modified diagnostic scores based on detailed medical interviews and routine laboratory data can distinguish DILI from non-DILI in patients with ALI.