Translational in vitro and in vivo PKPD modelling for apramycin against Gram-negative lung pathogens to facilitate prediction of human efficacious dose in pneumonia

Abstract

ObjectivesNew drugs and methods to efficiently fight carbapenem-resistant gram-negative pathogens are sorely needed. In this study, we characterized the preclinical pharmacokinetics (PK) and pharmacodynamics of the clinical stage drug candidate apramycin in time kill and mouse lung infection models. Based on in vitro and in vivo data, we developed a mathematical model to predict human efficacy. MethodsThree pneumonia-inducing gram-negative species Acinetobacter baumannii, Pseudomonas aeruginosa, and Klebsiella pneumoniae were studied. Bactericidal kinetics were evaluated with time-kill curves; in vivo PK were studied in healthy and infected mice, with sampling in plasma and epithelial lining fluid after subcutaneous administration; in vivo efficacy was measured in a neutropenic mouse pneumonia model. A pharmacokinetic-pharmacodynamic model, integrating all the data, was developed and simulations were performed. ResultsGood lung penetration of apramycin in epithelial lining fluid (ELF) was shown (area under the curve (AUC)ELF/AUCplasma = 88%). Plasma clearance was 48% lower in lung infected mice compared to healthy mice. For two out of five strains studied, a delay in growth (∼5 h) was observed in vivo but not in vitro. The mathematical model enabled integration of lung PK to drive mouse PK and pharmacodynamics. Simulations predicted that 30 mg/kg of apramycin once daily would result in bacteriostasis in patients. DiscussionApramycin is a candidate for treatment of carbapenem-resistant gram-negative pneumonia as demonstrated in an integrated modeling framework for three bacterial species. We show that mathematical modelling is a useful tool for simultaneous inclusion of multiple data sources, notably plasma and lung in vivo PK and simulation of expected scenarios in a clinical setting, notably lung infections.