Abstract
OBJECTIVE: The objective of this study was to develop a physiologically based pharmacokinetic model for meropenem using a retrograde approach, which could serve as a basis for prediction of the systemic and infection-site drug exposures in different populations and indications. We intended this model to be a useful tool to inform (local) pharmacokinetic-based optimal dosing of meropenem in different settings. METHODS: We developed a reduced physiologically based pharmacokinetic model with NONMEM software using a top-down approach. We used historical (previously published) data for model development and qualification. We used steady-state systemic and infection-site concentrations from 60 adult patients diagnosed with severe lung infection for model development and internal evaluation. The data included rich plasma and sparse epithelial lining fluid samples. We based the internal validation of the model on successful numerical convergence, adequate precision in parameter estimation, acceptable goodness-of-fit plot with no indication of bias, and acceptable performance of visual predictive checks. We performed external validation by fitting the model to independent data from five previously published studies: four studies in patients with pneumonia, with different grades of renal impairment, and one study in morbidly obese patients. RESULTS: We successfully fitted a reduced physiologically based pharmacokinetic model with six compartments (arterial and venous pools, infection site [lungs], liver, kidneys and rest of the body) to the data and adequately estimated model parameters. We successfully qualified the model (internally and externally) using established methods. Estimated values for tissue-to-plasma partition coefficients were 0.2629 and 0.1946 for lungs and non-fat tissues (kidneys and liver), respectively. Estimated total clearance was 8.174 L/h for a typical patient with a glomerular filtration rate of 65 mL/min. Consistent with the known mechanism of meropenem elimination and previously published models, renal clearance accounted for 70% of total clearance. The model had good predictive performances on data from five different sources including populations with different characteristics with regard to body size, renal function and morbidity. CONCLUSIONS: We successfully developed a physiologically based pharmacokinetic model for meropenem in adult patients to be used as a basis for prediction of concentrations in different groups of patients, and eventually for effective dose individualisation in different subgroups of the population.