Introduction: Daptomycin is a key antibiotic for multidrug-resistant Gram-positive infections, but its substantial inter-individual pharmacokinetic variability and exposure-related efficacy and toxicity make optimal dosing challenging, particularly in elderly patients. Age-related changes in renal function may further alter daptomycin exposure, increasing the risk of underexposure or toxicity. Population pharmacokinetic (PopPK) models provide the foundation for model-informed precision dosing, enabling individualized exposure estimation and Bayesian dose adjustment from sparse therapeutic drug monitoring data. However, the external predictive performance and clinical transferability of published daptomycin models in geriatric patients remain unclear.
Aims: To externally evaluate published daptomycin PopPK models in an independent elderly cohort, identify model(s) suitable for local MIPD-guided dosing, and develop a maximum a posteriori Bayesian AUC estimator to support individualized therapy.
Methods: A systematic literature search was conducted in PubMed, Web of Science, and Embase to identify published parametric PopPK models of daptomycin up to 6 May 2025, following PRISMA recommendations. Eligible models were screened based on predefined criteria, and key information including study population, dosing regimen, sampling strategy, structural model, covariates, variability terms, and model application was extracted. Models describing total plasma daptomycin concentrations were selected for external evaluation and re-implemented in NONMEM to ensure consistency across simulations. Real-world PK data were retrospectively collected from elderly patients aged ≥65 years who received intravenous daptomycin and underwent therapeutic drug monitoring at Hospital del Mar, Barcelona, between 2015 and 2025. External evaluation was performed under three clinically relevant prediction scenarios: a priori prediction using only dosing history and covariates, Bayesian prediction after incorporating one paired Cmax/Cmin sample, and Bayesian prediction after incorporating two paired samples. Model performance was assessed using prediction-corrected visual predictive checks, normalized prediction distribution errors, prediction error, median prediction error, and median absolute prediction error. Finally, the best-performing model was implemented in an interactive R Shiny application integrating maximum a posteriori Bayesian estimation to calculate individualized AUC and support daptomycin precision dosing in elderly patients.
Results:
A total of 592 records were identified through the systematic search, including 274 from PubMed, 93 from Embase, and 225 from Web of Science. After screening and eligibility assessment, 33 full-text articles were reviewed, of which 15 studies met the inclusion criteria. Among the 15 published daptomycin PopPK models identified, 10 models describing total plasma concentrations were selected for external evaluation. The external evaluation cohort included 119 elderly patients, contributing 308 plasma daptomycin concentrations. A priori predictive performance varied substantially across models, with models developed in populations more comparable to the external cohort demonstrating better transferability. Bayesian updating using TDM data consistently improved model accuracy and precision, with prediction errors decreasing as additional concentration data were incorporated. The maximum median absolute prediction error decreased from 61.94% in the a priori scenario to 52.29% after Bayesian updating with one paired Cmax/Cmin sample, and further to 46.86% after two paired samples. Overall, the Takahashi[1] model demonstrated the best predictive performance and was selected for the development of a freely accessible Bayesian AUC calculator, implemented as an R Shiny web application.
Conclusion: This study comprehensively evaluated published daptomycin PopPK models in an elderly real-world cohort and demonstrated that Bayesian updating with TDM data substantially improves predictive performance. By identifying a model suitable for local implementation and translating it into a Bayesian AUC calculator, this work provides a practical and scientifically grounded tool to support model-informed precision dosing of daptomycin in geriatric patients.
Reference: 1. Takahashi, Tsuji, et al. (2023). Eur J Drug Metab Pharmacokinet, 48(2), 201–211.