Understanding variability with voriconazole using a population pharmacokinetic approach: implications for optimal dosing

Voriconazole exhibits highly variable, non-linear pharmacokinetics and is associated with a narrow therapeutic range. This study aimed to investigate the population pharmacokinetics of voriconazole in adults, including the effect of CYP2C19 genotype and drug-drug interactions. Non-linear mixed effects modelling (NONMEM) was undertaken of six voriconazole studies in healthy volunteers and patients. Dosing simulations to examine influential covariate effects and voriconazole target attainment (2-5 mg/L) stratified by CYP2C19 phenotype were performed. We analysed 3352 voriconazole concentration measurements from 240 participants. A two-compartment pharmacokinetic model with first-order oral absorption with lag time and Michaelis-Menten elimination best described voriconazole pharmacokinetics. Participants with one or more CYP2C19 loss-of-function (LoF) alleles had a 41.2% lower V-max for voriconazole. Co-administration of phenytoin or rifampicin, St John's wort or glucocorticoids significantly increased voriconazole elimination. Among patients receiving 200 mg of voriconazole twice daily, predicted trough concentrations on day 7 were < 2 mg/L for oral and intravenous regimens for 72% and 63% of patients without CYP2C19 LoF alleles, respectively, with 49% and 35% below this threshold with 300 mg twice daily dosing. Conversely, these regimens resulted in 29%, 39%, 57% and 77% of patients with CYP2C19 LoF alleles with voriconazole trough concentrations a parts per thousand yen5 mg/L. Current dosing regimens for voriconazole result in subtherapeutic exposure in many patients without CYP2C19 LoF alleles, suggesting the need for higher doses, whereas these regimens result in supratherapeutic exposure in a high proportion of patients with reduced CYP2C19 activity. These findings support the essential role of therapeutic drug monitoring in ensuring efficacious and safe voriconazole exposure.