Chinese Hepatolgy ›› 2020, Vol. 25 ›› Issue (7): 671-675.

Previous Articles     Next Articles

A comparative study of predictive models for acute liver failure in drug-induced liver injury

YANG Rui-yuan, LIU Li-wei, LUO Juan, LI Ke-xin, TIAN Qiu-ju, WANG Yan, ZHAO Xin-yan, JIA Ji-dong   

  1. Liver Research Center, Beijing Friendship Hospital, Capital Medical University, National Clinical Research Center for Digestive Diseases, Beijing 100050, China
  • Received:2020-03-07 Published:2020-08-06
  • Contact: ZHAO Xin-yan, Email: zhao_xinyan@ccmu.edu.cn

Abstract: Objective To compare the efficacy of different models in the prediction of acute liver failure in drug-induced liver injury (DILI) patients. Methods We conducted a retrospective cohort study of patients diagnosed with DILI from 2014 to 2018. Data of liver biochemical tests of these DILI patients were collected and acute liver failure (ALF) cases were identified among those fulfilled the criterion of Hy's law, new Hy's law and Robles model, respectively. The sensitivity, specificity, positive predictive and negative predictive value were calculated, respectively. Results A total of 298 patients were enrolled, with an average age of 54.7 ± 14.7 years, including 211 female patients (70.8%). Most cases of DILI were hepatocellular injury (77.2%), followed by mixed injury (16.4%) and cholestatic injury (6.4%). Among all the patients, 13 (4.4%) cases progressed to ALF, 5 (1.7%) underwent liver transplantation or died of liver-related events. In this study, 161 DILI patients met the criterion of Hy’s law, 5.8% of them progressed to ALF, 208 cases met the criterion of new Hy’s law, 5.6% progressed to ALF, while 73 cases met the criterion of Robles model and 15.1% developed ALF. DILI patients fulfilled the Robles model had the highest risk of developing ALF, with the sensitivity of 84.6%, the specificity of 78.2%, and the positive predictive value of 99.1%. DILI patients who met the criterion of Robles model had significantly lower rate of liver biochemical normalization at 1, 3 and 12 months compared with DILI patients who did not (P<0.05). Conclusion When applied at DILI recognition, the Robles model provides the best specificity and positive predictive value in the prediction of ALF development, and most accurately predicts the normalization rate of liver biochemical indices within 1 year.

Key words: Drug-induced liver injury, Acute liver failure, Predictive model, Prognosis