Omnia S. Heikal 1, Fouzia Lghoul-Oulad Saïd 1, Laura B. Zwep 1, J.G.C van Hasselt
1 Leiden Academic Centre for Drug Research, Leiden University (Leiden, Netherlands), 2 InsightRx (, USA)
Introduction: Vancomycin is an antibiotic commonly used in a hospital setting to treat Methicillin Resistant Staphylococcus Aureus (MRSA). The occurrence of vancomycin-associated acute kidney injury (AKI) remains a significant clinical concern1, with an incidence of 5% to 35% of patients2. This large range is influenced by factors such as comedication, comorbidities, and other patient-related characteristics. Yet there is currently limited quantitative understanding into the exposure-toxicodynamic relationship between vancomycin exposure and the occurrence of AKI, wherein AKI is usually described according to the clinical categorical AKI severity criteria, which are based on serum creatinine (SCr). The direct use of SCr as a continuous physiological marker for kidney function may enhance insight into PK/PD relationships for vancomycin-associated AKI, including understanding of onset and recovery of kidney function.
Objectives: The aim of this study was to (i) develop a population PK/PD model for the relationship between vancomycin pharmacokinetics (PK) and the dynamic response of SCr using a large scale real-world dataset (RWD), (ii) compare predictive value of such a full SCr dynamics-based approach against conventional categorical AKI outcome prediction.
Methods: We used large scale retrospective dataset collected for routine therapeutic drug monitoring and model-informed precision dosing (MIPD) of vancomycin from multiple hospitals in the United States, as present in the InsightRx database. We first performed rigorous curation of the dataset, using the following inclusion criteria: adult individuals who received at least four doses of vancomycin between 2000–2023, a minimum of four SCr measurements, and at least two vancomycin TDM measurements, leading to inclusion of a total of 85,226 patients. To infer individual vancomycin PK profiles, we compared multiple previously developed population PK models for their predictive performance, comparing goodness of fit diagnostics, prediction errors, stratified by renal function. Finally, we evaluated multiple pharmacodynamic model structures to characterize the exposure-response relationships between vancomycin PK and SCr dynamics.
Results: We evaluated the performance of five published PK models. We found that the population PK model developed by Colin et al.3 (2019) had optimal predictive performance, with a median bias in individual predictions of 0.515%. Model performance remained consistent across clinical subgroups, with global CWRES centered at -0.26 and subgroup median CWRES (stratified by eGFR and BMI) ranging from -0.41 to -0.08. Based on this, the model was selected as the foundation for the mechanism-based model. Subsequentially, we compared observed SCr trajectories and evaluated multiple pharmacodynamic model structures to capture the dynamics of SCr and associated exposure-response relationships.
Conclusion: The model-based approach being developed in this project can help provide insights into the exposure-response relationships for vancomycin-associated AKI, which may ultimately be of relevance to enable model-based precision dosing of vancomycin, optimizing model-informed therapeutic drug monitoring and dosing strategies to reduce the incidence of vancomycin-associated AKI.
References:
1. Filippone E, Kraft W, Farber J. The Nephrotoxicity of Vancomycin. Clin Pharmacol Ther. 2017;102(3):459-469. doi:10.1002/cpt.726
2. Bellos I, Daskalakis G, Pergialiotis V. Relationship of vancomycin trough levels with acute kidney injury risk: an exposure–toxicity meta-analysis. Journal of Antimicrobial Chemotherapy. 2020;75(10):2725-2734. doi:10.1093/jac/dkaa184
3. Colin PJ, Allegaert K, Thomson AH, et al. Vancomycin Pharmacokinetics Throughout Life: Results from a Pooled Population Analysis and Evaluation of Current Dosing Recommendations. Clin Pharmacokinet. 2019;58(6):767-780. doi:10.1007/s40262-018-0727-5
Reference: PAGE 34 (2026) Abstr 12037 [www.page-meeting.org/?abstract=12037]
Poster: Clinical Applications