Useful biomarkers for predicting poor prognosis of patients with drug-induced liver injury: A retrospective cohort study.

Background: Drug-induced liver injury (DILI) plays an important role in liver failure and causes mortality. Patients with DILI compatible with Hy's law are associated with poorer outcomes. However, the predictive accuracy of Hy's law is not good enough in clinical practice. This study aimed to investigate the optimal values of biomarkers associated with the prognosis of DILI.

Methods: From June 1, 2014-May 30, 2022, patients with reported DILI were included. Patients' characteristics, drugs, DILI type, liver enzymes, and comorbidities were assessed. The associations with DILI-related comorbidities and survival were analyzed.

Results: Ninety-five DILI patients were enrolled, 5 patients died of liver failure, and 23 patients died within 56 weeks after DILI. This study found that 15 mg/dL of total bilirubin, 1000 U/L of ALT, and 2 of PT-INR were optimal cut-off values in predicting DILI-related mortality. For the overall survival, patients with sepsis (HR:5.053, 95% CI:1.594-16.018, p = 0.006), malignancy (HR:4.371, 95% CI:1.573-12.147, p = 0.005), or end-stage renal disease (HR:7.409, 95% CI:1.404-39.103, p = 0.018) were independent poor prognostic factors in multivariate Cox regression analysis.

Conclusions: Total bilirubin >15 mg/dL, ALT >1000 U/L, and PT-INR >2 are useful biomarkers in predicting DILI-related mortality. DILI patients with sepsis, malignancy, or end-stage renal disease are associated with worse overall survival.