보리코나졸의 Therapeutic Drug Monitoring을 위한 최적의 채혈시각 탐색을 위한 시뮬레이션 연구

Abstract

Objective: Voriconazole is a broad-spectrum triazole antifungal agent with activity against a wide range of yeasts and filamentous fungi which has been approved worldwide for invasive fungal infections. It has a narrow therapeutic range, nonlinear pharmacokinetic profile, and high interindividual variability. Area under the time-concentration curve during 12-hour dosing interval (AUC0-12) and pre-dose concentration (Ctrough) are clinically important variables based on which dose adjustment is made. The purpose of this study is to explore optimal sampling time for therapeutic drug monitoring of voriconazole. Methods: Plasma voriconazole concentrations following three dosing regimens (Scenario 1, loading dose of 6 mg/kg IV q12hr on day 1, followed by maintenance dose of 4 mg/kg IV q12hr; Scenario 2, loading dose of 6 mg/kg IV q12hr on day 1, followed by 3 mg/kg IV q12hr; Scenario 3, loading dose of 6 mg/kg IV q12h on day 1, followed by 200 mg PO q12hr) were simulated to generate 1,000 sets of data for each scenario using NONMEM® (version 7.4.4). Using one or two concentration data early after the initiation of therapy, plasma concentration over time, AUC0-12, and Ctrough at steady state of each individual were predicted by maximum a posteriori (MAP) method. By comparing the accuracy and precision of these values, the optimal pharmacokinetic (PK) sampling time was explored. During the MAP prediction, deviation in sampling time from the planned time was also taken into account. Results: Plasma voriconazole concentration over time was well predicted by MAP method for all of the sampling times explored in this study with minimal bias (less than 10%). However, precision of predicted concentration was different by time points used in MAP, with low precision especially for the concentration around mid-point of the dosing interval. AUC0-12 was best predicted with high accuracy and precision using concentration at 2- or 3- hour sampling time in scenario 1 and 2. In scenario 3, AUC0-12 was well predicted with similar accuracies across all the sampling times of 1 through 12 hours, but the precision was predicted to be high when using later time points near 12 hour. Conclusions: We successfully reconstructed a voriconazole PK model and conducted a simulation study. The simulation suggested optimal sampling time points that can be implemented in clinical setting for the therapeutic drug monitoring of voriconazole. The current study provides useful information for individualized, optimal therapy of voriconazole.