Paulina Okuńska 1, Anna Małecka 2, Joanna Renke 2, Ninela Irga-Jaworska 2, Paweł Wiczling 1
1 Department of Biopharmaceutics and Pharmacodynamics, Medical University of Gdansk (Gdańsk, Poland), 2 Department of Paediatrics, Haemathology & Oncology, Medical University of Gdansk (Gdańsk, Poland)
Introduction
Clinicians are increasingly faced with a wide array of pharmacokinetic (PK) models developed for various patient populations. Selecting the most appropriate model to guide individualized therapy can be challenging, particularly when extrapolating data from small cohorts, which may not accurately reflect the pharmacokinetics of specific patient groups. This is especially critical for antibiotics such as vancomycin, where the narrow therapeutic window necessitates precise dosing throughout the entire antimicrobial course. The complexity of treatment regimens in oncology patients, combined with frequent co-medications and potential drug-drug interactions, further complicates dose optimization. Despite the abundance of PK models in the literature, few are directly applicable to pediatric hematology populations, leaving clinicians with limited guidance for evidence-based dosing.​ [1]​
Objectives
The study aimed to analyze historical clinical and laboratory data on vancomycin serum concentrations and serum creatinine levels in pediatric hematology patients. The goal was to develop an integrated pharmacokinetic model to inform the safest and most effective dosing strategy for vancomycin in this population and to support therapeutic drug monitoring (TDM), improving both efficacy and safety.
Methods
Data were obtained from 15 treatment cycles in 9 pediatric hematology patients (3 boys, 6 girls; 3.66–16.8 years) treated for leukemia or lymphoma at the University Clinical Centre in Gdansk. Vancomycin and serum creatinine concentrations were measured together with demographic and clinical covariates, including body weight, age, and sex.
Estimated glomerular filtration rate (eGFR), calculated using the Schwartz equation, exceeded physiological values (>120 mL/min/1.73 m²) for most of the subjects. As this cohort differs substantially from the population in which the equation was developed, these values were interpreted cautiously. Therefore, renal function was characterized within the joint pharmacokinetic model rather than relying solely on eGFR.
Vancomycin and creatinine pharmacokinetics were described using two one-compartment models with shared clearance. A hierarchical Bayesian framework with informative priors derived from the literature was applied to capture both interindividual and population-level variability​. [2] Model parameters (volume of distribution, clearance, and creatinine production rate) were allometrically scaled to body size. The Bayesian approach enabled integration of prior knowledge with observed pediatric data and provided parameter estimates together with quantification of predictive uncertainty, which is particularly valuable in small populations.
Results
The joint one-compartment model accurately described vancomycin and creatinine kinetics. Typical pharmacokinetic parameter estimates (posterior mean vs. prior mean) were: θCL = 7.56 L/h vs 7.21 L/h; θCRP = 7.07 vs. 7.22; θV1 =44.9 L vs. 44.8 L. Interindividual variability estimates (posterior mean vs. prior mean) were: ωCL = 0.375 vs. 0.305; ωCRP = 0.314 vs. 0.306; ωV1 = 0.179 vs. 0.205. The model reliably reproduced observed concentration-time profiles and allowed individualized dose prediction based on prior vancomycin exposure, measured drug levels, and serum creatinine. Serum creatinine emerged as a critical covariate explaining variability and supporting dose individualization.
Conclusions
This study presents a single, integrated Bayesian model capturing both vancomycin and creatinine kinetics in pediatric hematology patients. By incorporating prior knowledge and quantifying uncertainty, the approach provides a robust framework for individualizing vancomycin therapy in small, special populations. The methodology has potential to improve clinical outcomes, minimize toxicity, and support evidence-based therapeutic drug monitoring in pediatric oncology. Future work could expand this framework to include other renal biomarkers or co-administered drugs, further enhancing dosing precision in complex clinical scenarios.
Clinical Application
The clinical applicability of the model was illustrated through individualized dose predictions generated under varying levels of available prior information: no prior measurements, a single vancomycin concentration, a single creatinine value, or both measurements combined.
References:
[1] Guilhaumou, R. et al. Pediatric Patients With Solid or Hematological Tumor Disease: Vancomycin Population Pharmacokinetics and Dosage Optimization. www.r-project.org (2016).
[​2] Aljutayli, A., Marsot, A. & Nekka, F. An Update on Population Pharmacokinetic Analyses of Vancomycin, Part I: In Adults. Clinical Pharmacokinetics vol. 59 671–698 Preprint at https://doi.org/10.1007/s40262-020-00866-2 (2020).
Reference: PAGE 34 (2026) Abstr 11940 [www.page-meeting.org/?abstract=11940]
Poster: Clinical Applications