Establishment and validation of a predictive nomogram for voriconazole-associated liver injury in lung transplant patients.

Abstract

Introduction: Voriconazole serves as a cornerstone antifungal therapy for lung transplant (LTx) patients; however, its use is associated with a notable risk of hepatotoxicity.

Aim: This study aimed to identify potential prognostic factors and develop a robust prediction nomogram for voriconazole-associated liver injury (VALI) in this population.

Method: This retrospective observational study included 97 LTx patients treated with voriconazole between 2017 and 2024. Patients were randomly divided into a training cohort and a validation cohort, with a ratio of 75:25. Voriconazole blood concentrations were measured, and genetic polymorphisms related to voriconazole metabolism were analyzed. Potential prognostic factors associated with VALI were identified by using univariate and multivariate Cox proportional hazard models. The nomogram was evaluated by C-index, calibration curve and clinical utility.

Results: The overall incidence of VALI was 30.9%. In the multivariate analysis, voriconazole concentration (HR 1.318, 95% CI 1.052-1.650, p = 0.016), voriconazole dose (HR 1.005, 95% CI 1.001-1.008, p = 0.006), concomitant mycophenolate mofetil (MMF) use (HR 0.320, 95% CI 0.116-0.885, p = 0.028), and rs4244285 polymorphism (HR 2.466, 95% CI 0.936-6.498, p = 0.068) were identified as independent predictive factors for VALI. The C-indexes of the nomogram were 0.807 (95% CI 0.725-0.889) for the training cohort and 0.82 (95% CI 0.681-0.959) for the validation cohort. Furthermore, the nomogram demonstrated excellent calibration and clinical applicability. Receiver operating characteristic (ROC) curve analysis determined the voriconazole cut-off trough concentration for hepatotoxicity stratified by MMF use. Stratified analysis indicated that the optimal voriconazole trough thresholds for minimizing VALI risk were 2.8 ng/mL in MMF users (AUC 0.769, p < 0.001) and 2.6 ng/mL in non-users (AUC 0.795, p = 0.011).

Conclusion: This study highlights the significance of therapeutic drug monitoring for voriconazole in stratifying the risk probability of VALI among LTx patients. Additionally, we have established a predictive nomogram to aid clinicians and pharmacists in assessing the likelihood of VALI following voriconazole administration in this patient population. However, before clinical implementation, external validation in independent cohorts and prospective utility studies are imperative to confirm the generalizability and clinical efficacy of this model.