Can a pharmacokinetic/pharmacodynamic (PKPD) model be predictive across bacterial densities and strains? External evaluation of a PKPD model describing longitudinal in vitro data

Abstract

Background: Pharmacokinetic/pharmacodynamic (PKPD) models developed based on data from in vitro time- kill experiments have been suggested to contribute to more efficient drug development programmes and better dosing strategies for antibiotics. However, for satisfactory predictions such models would have to show good extrapolation properties. Objectives: To evaluate if a previously described mechanism-based PKPD model was able also to predict drug efficacy for higher bacterial densities and across bacterial strains. Methods: A PKPD model describing the efficacy of ciprofloxacin on Escherichia coli was evaluated. The predictive performance of the model was evaluated across several experimental conditions with respect to: (i) bacterial start inoculum ranging from the standard of ~106 cfu/mL up to late stationary-phase cultures; and (ii) efficacy for seven additional strains (three laboratory and four clinical strains), not included during the model development process, based only on information regarding their MIC. Model predictions were performed according to the intended experimental protocol and later compared with observed bacterial counts. Results: The mechanism-based PKPD model structure developed based on data from standard start inoculum experiments was able to accurately describe the inoculum effect. The model successfully predicted the time course of drug efficacy for additional laboratory and clinical strains based on only the MIC values. The model structure was further developed to better describe the stationary phase data. Conclusions: This study supports the use of mechanism-based PKPD models based on preclinical data for predictions of untested scenarios.