Risk stratification of voriconazole-induced hepatotoxicity through therapeutic drug monitoring combined with the albumin–bilirubin score: a multicenter retrospective cohort study in Japan

Background Patients receiving voriconazole often have insufficient hepatic reserve, making it difficult to stratify the risk of severe hepatotoxicity using therapeutic drug monitoring (TDM) alone. This multicenter study aimed to evaluate whether the albumin–bilirubin (ALBI) score can stratify the risk of hepatotoxicity in patients undergoing TDM. Methods This multicenter, retrospective cohort study included 608 patients. The primary outcome was voriconazole-induced hepatotoxicity. The cumulative risk of hepatotoxicity was evaluated using a Gray’s test between ALBI score ≥−2.0 or <−2.0, and a landmark approach was applied as a sensitivity analysis. The Fine–Gray test was performed, and independent risk factors were used for decision tree analysis. Results The cumulative risk of hepatotoxicity was significantly higher in the ALBI score ≥-2.0 group than that in the ALBI score <-2.0 group ( p = 0.006), and this trend was confirmed in a sensitivity analysis. The Fine–Gray model revealed that the ALBI score ≥−2.0 (subdistribution hazard ratio, 1.951; 95% confidence interval, 1.135–3.355; p = 0.016) was an independent risk factor. In the decision tree model, the development of hepatotoxicity was lowest (6.5%, 12/186) in patients with ALBI score <-2.0 and voriconazole trough concentration (C min ) <4.0 μg/mL, whereas that in patients with ALBI score ≥-2.0 and C min ≥4.0 μg/mL was highest at 18% (27/153), with an accuracy of 88.3%. Conclusions The present study demonstrated that combining an ALBI score ≥−2.0 with the C min of voriconazole allows risk stratification for voriconazole-induced hepatotoxicity with high accuracy, and this model may provide a practical risk stratification tool.