Saliva for Precision Dosing of Antifungal Drugs: Saliva Population PK Model for Voriconazole Based on a Systematic Review.

Precision dosing for many antifungal drugs is now recommended. Saliva sampling is considered as a non-invasive alternative to plasma sampling for therapeutic drug monitoring (TDM). However, there are currently no clinically validated saliva models available. The aim of this study is firstly, to conduct a systematic review to evaluate the evidence supporting saliva-based TDM for azoles, echinocandins, amphotericin B, and flucytosine. The second aim is to develop a saliva population pharmacokinetic (PK) model for eligible drugs, based on the evidence. Databases were searched up to July 2019 on PubMed(R)and Embase(R), and 14 studies were included in the systematic review for fluconazole, voriconazole, itraconazole, and ketoconazole. No studies were identified for isavuconazole, posaconazole, flucytosine, amphotericin B, caspofungin, micafungin, or anidulafungin. Fluconazole and voriconazole demonstrated a good saliva penetration with an average S/P ratio of 1.21 (+/- 0.31) for fluconazole and 0.56 (+/- 0.18) for voriconazole, both with strong correlation (r = 0.89-0.98). Based on the evidence for TDM and available data, population PK analysis was performed on voriconazole using Nonlinear Mixed Effects Modeling (NONMEM 7.4). 137 voriconazole plasma and saliva concentrations from 11 patients (10 adults, 1 child) were obtained from the authors of the included study. Voriconazole pharmacokinetics was best described by one-compartment PK model with first-order absorption, parameterized by clearance of 4.56 L/h (36.9% CV), volume of distribution of 60.7 L, absorption rate constant of 0.858 (fixed), and bioavailability of 0.849. Kinetics of the voriconazole distribution from plasma to saliva was identical to the plasma kinetics, but the extent of distribution was lower, modeled by a scale factor of 0.5 (4% CV). A proportional error model best accounted for the residual variability. The visual and simulation-based model diagnostics confirmed a good predictive performance of the saliva model. The developed saliva model provides a promising framework to facilitate saliva-based precision dosing of voriconazole.