Treatment of serious bacterial infections with antimicrobial agents, such as antibiotics, is a major clinical challenge, because of growing bacterial resistance to multiple agents. Combination therapy (i.e. combined dosing of more than one agent) is often used for the purpose, but its systematic design remains a challenge. To address this, we recently reported a method to mathematically model bacterial response to antimicrobial agents, and to use this model for systematic design of clinically relevant combination therapy. The method relies on (a) longitudinal data of bacterial load, estimated from optical density measurements during time-kill experiments in an automated instrument, and (b) use of these data to fit a mathematical model for combination therapy design. In this work, we studied an application of the proposed method to (a) an important bacterial pathogen (Acinetobacter baumannii) and (b) two cases of antibiotic combinations (ceftazidime / amikacin and ceftazidime / avibactam) in synchronous and asynchronous dosing, not otherwise studied to date. Following the proposed method, optical density based data of bacterial load under antibiotic exposure for 20 h were used to calibrate the mathematical model and subsequently predict outcomes of various dosing regimens with clinically relevant pharmacokinetics. Representative predictions by the mathematical model were tested in vitro in a hollow fiber infection model over 120 h. Test outcomes validated these predictions. Collectively, this study both provides guidance for design of A. baumannii infection treatments with the agents studied and underscores the broader applicability of the proposed method for design of clinically relevant combination therapy.
Read full abstract