Development of vancomycin population pharmacokinetic models for pediatric patients using a real-world web application database.

Background: The Japanese Society of Chemotherapy developed Practical AUC-guided Therapeutic Drug Monitoring (PAT), a freely available web-based application widely used in Japan to support area under the concentration-time curve (AUC)-guided dosing of vancomycin. Its broad adoption has generated a substantial real-world pharmacokinetic database. Although PAT includes a preliminary pediatric model based on data from American children aged three months or older, that model requires further optimization. This study aimed to develop a population pharmacokinetic (popPK) model specifically for Japanese pediatric patients and to compare its performance with existing models.

Methods: We utilized a real-world database collected through PAT between December 2022 and October 2024, comprising 1,673 pediatric patients aged three months or older. PopPK analysis was performed using nonlinear mixed-effects modeling, with comparisons made against five existing models.

Results: The developed model demonstrated strong a priori predictive performance for empirical vancomycin dosing, with a mean prediction error of 0.2 μg/mL and a mean absolute prediction error of 5.6 μg/mL. These results showed no apparent bias. Graphical diagnostics confirmed optimal a priori prediction in the developed model. A posteriori predictive performance-used for Bayesian posterior dosing-was similarly favorable across all models. Notably, two-point sampling produced significantly different clearance estimates compared to one-point sampling in 21 of 116 patients (18.1%).

Conclusions: For Japanese pediatric patients aged three months or older, the developed popPK model is suitable for both empirical and Bayesian-guided vancomycin dosing. Two-point sampling improves the accuracy of clearance estimation and is recommended when feasible.