Generating and Modeling Virtual Patient Data from Published Population Pharmacokinetic Analyses: A Vancomycin Case...

Background/Objectives: In recent years, clinical pharmacometrics has become vital for drug development and clinical practice, particularly for predicting drug efficacy and safety. Population pharmacokinetic models are used for drugs for which therapeutic drug monitoring is recommended in clinical practice. Numerous population pharmacokinetic models have been developed for patients with similar clinical and demographic characteristics, resulting in reduced inter-individual variability. Despite the existence of diverse-population pharmacokinetic models, selecting an appropriate model for bedside use remains challenging. This study proposes a model-simulated model-based meta-analysis (M-cubed) to construct a unified model capable of accommodating a wide range of patient backgrounds. Methods: Vancomycin (VCM), a drug used for therapeutic drug monitoring, was used as an example. Using information from published VCM models, the M-cubed method was employed to generate virtual patient data for each publication through simulation, followed by modeling the integrated dataset. Results: Population pharmacokinetic analysis was performed on data from 19 virtual patient models, resulting in a total of 2303 cases. Covariates in the final model included creatinine clearance and body weight. The predictive ability of the model was robust. Conclusions: A model that integrates several population studies using the M-cubed method is required to address the need in clinical practice.