抗真菌薬ボリコナゾールを安全に使用するための薬物動態パラメータの検討

Abstract

Voriconazole (VRCZ) is a triazole antifungal agent that exhibits large inter-individual variability in plasma concentration. Reasons for this include non-linear pharmacokinetics due to its saturated metabolism and genetic polymorphism in CYP2C19. Because VRCZ has the potential to produce adverse reactions such as hepatotoxicity and visual disturbances, therapeutic drug monitoring is required to optimize treatment. Therefore, this study aimed to evaluate the predictive performance of published population pharmacokinetic (PPK) models of VRCZ. Demographic and laboratory data were retrospectively collected from the medical records of 39 patients who underwent VRCZ treatment. Patients were classified into lower and higher concentration groups by visual inspection of observed weight- and dose-normalized trough concentration. We assumed that the patients in the lower concentration group were more extensive metabolizers of CYP2C19 versus patients in the higher concentration group. We calculated PPK-predicted concentrations of VRCZ for each patient after taking the estimated genotype into account. Next, we examined their predictive performance by calculating the mean prediction error (ME), mean absolute prediction error (MAE), and root mean squared error (RMSE). Thirty and 9 patients were included in the lower and higher concentration groups, respectively. The predictive performance of the PPK model constructed from a Japanese phase Ⅲ study was significantly smaller than that from the other PPK studies (r = 0.868 vs 0.694, P = 0.001; ME －0.27 µg/mL, MAE ＋0.90 µg/mL, and RMSE ＋1.15 µg/mL for the Japanese PPK model). Finally, we constructed a suggested VRCZ safe dose schedule. Our findings provide useful information for adjusting the VRCZ dosage to improve clinical safety and effectiveness.