Abstract
BACKGROUND: An extracorporeal membrane oxygenation (ECMO) device represents an additional functional body compartment, potentially impacting vancomycin pharmacokinetics. OBJECTIVES: This systematic review aimed to: (1) provide a comprehensive summary of vancomycin population pharmacokinetics (Pop-PK) in critically ill patients receiving ECMO and; (2) associate the findings with clinical practices and dosing recommendations. METHODS: PubMed, Embase and Google Scholar databases were searched for vancomycin non-linear mixed-effects modelling Pop-PK studies in ECMO patients (inception-May 2024). Standardized, pre-tested and pilot-tested tools were used for quality assessment and data extraction utilizing triangulation. Summary measures included typical values for vancomycin pharmacokinetic parameters, influential covariates, and associated interindividual and residual variabilities. RESULTS: Seven studies reporting an approximate total of 1600 vancomycin blood concentrations (range: 33-433 concentrations per study) collected from 215 patients (range: 11-93 patients per study) were included. Pop-PK models fitted three-compartment (n = 1), two-compartment (n = 5), or one-compartment (n = 1) models. Various modalities [venovenous-ECMO (n = 6); venoarterial-ECMO (n = 5), centrifugal-pump (n = 6), roller pump (n = 1)] and age groups were reported [pediatrics (n = 2); adults (n = 6)]. Vancomycin clearance and central volume of distribution (V(1)) ranged between 0.0579-4.32 L/h and 0.13-3.32L/kg, respectively. Relatively high between-subject variability (BSV) exists (clearance: 0.1%-77%; V(1): 0.33%-94.8%). Models failed to explain maximally 19.4% of the overall variations, successfully explaining at least 80.6% of the variabilities in vancomycin disposition in ECMO. Age and body weight were significant covariates on clearance and V(1). Additionally, glomerular filtration rate, serum creatinine, creatinine clearance, and use of dialysis were significant covariates on clearance. Clinical status-related covariates (i.e., acute physiology and chronic health evaluation II (APACHE-2) score, sequential organ failure assessment (SOFA) score, presence of shock) and ECMO modality-related covariates (pump type/flow rate, ECMO mode) were not significant covariates in the final clearance and V(1) models. CONCLUSION: This work comprehensively compiles up-to-date population-approach-based dosing simulations and pharmacokinetic models of vancomycin in ECMO, with special focus on influential predictors of vancomycin disposition to guide clinical dosing decisions. Relatively small sample sizes of patients were reported, a limitation that needs to be addressed in future large-scale studies. Observed BSV mandates therapeutic drug monitoring using Pop-PK models and parameters summarized in this work. Large-scale studies are needed to explore unexplored covariates, resulting in more informative dosing nomograms.