075: Precision Dosing in Critically Ill Patients: A Voriconazole Case Study

Statement of Purpose: Voriconazole is the drug of choice for treating invasive aspergillosis (IA), a severe fungal infection associated with high mortality in immunocompromised patients. It has a relatively narrow therapeutic window and shows high between-subject variability as well as non-linear pharmacokinetic, resulting in the need for therapeutic drug level monitoring (TDM). TDM of voriconazole in isolation is challenging because it relies on steady-state concentrations in blood or plasma. Due to voriconazole’s long half-life and its non-linear kinetics it is not feasible to wait for several days in this critically-ill patient population until steady-state is reached to adjust dosing. Thus, the use of non-steady-state concentrations in conjunction with a modeling and simulation approach offers a viable alternative to optimize dosing earlier in these patients. Therefore, the objective of this study is to develop a bedside-ready decision support tool for optimizing voriconazole dosing in IA patients.

Description of Methods & Materials: Plasma peak and/or trough samples were collected from 25 critically ill patients with hematological malignancies and invasive fungal infections, including IA. The patients received voriconazole (70 mg to 430 mg every 8 or 12 hours) via intravenous short-term (1h) infusion or per os. A non-linear mixed effects approach implemented in NONMEM software was used to develop a population PK model. Statistics criteria, the goodness of fit plots, and simulation-based diagnostics were used for model evaluation and qualification. Then, the model was used to perform simulations of PTA driven by the PK/PD index of efficacy: AUC_0-24/MIC > 25.

Data & Results: Voriconazole PK was well described with a 1-compartment model (V/F, 179 L) with first-order absorption (k_a fixed to 0.654 h-1), non-linear clearance calculated using Michaelis-Menten model (V_max, 33.2 mg/h; K_m fixed to 3.35 mg/L). Polymorphism of CYP2C19 affecting V_max has been included in the model to account for increased clearance in ultra-rapid metabolizers (V_max, UM, 74.5 mg/h). Simulations obtained using the developed population PK model showed that the probability of achieving AUC/MIC > 25 falls below 90% for MIC >= 1 mg/L (epidemiological cut-off reported by EUCAST) at the actual dosing regimen. Moreover, PTA falls below 90% for MIC 2- to 4-fold smaller in ultra-rapid metabolizers in comparison with normal metabolizers, while the use of a loading dose (6 mg/kg in adults; 9 mg/kg in pediatrics) and a maintenance dose of 400 mg ensures a PTA above 90 % for MIC equal or smaller than 1 mg/L. Oral and IV routes of administration have demonstrated similar results since the bioavailability of the drug is very high.

Interpretation, Conclusion or Significance: The findings of our study suggest: 1) adding the loading dose in the regimen to reach efficient concentrations immediately in each subgroup of patients independently of the magnitude of the clearance; 2) collection of samples during the first day of treatment to measure the drug concentrations; 3) using the developed model to inform about individual voriconazole's metabolizing rate and to adjust the maintenance dose, even when the CYP2C19 phenotyping information is missing. In order to optimally treat these critically ill patients, the generation of MIC distribution data for the local hospital population will be further important.

Disclosures: Ahmed A. Abulfathi is funded through UF/GSK Post-Doctoral Fellowship; All authors have no conflict of interest to declare. This project was approved by the Ethics Committee of Research of the Federal Univ of Santa Maria