Quantifying effects of antimicrobial drug combinations by modelling and simulation

Antibiotic resistance is becoming an accelerating issue, both in the treatment of nosocomial infections and for public health in general. There is a pressing need for innovative methods to understand how available antibiotics could be combined to overcome and limit the emergence of resistance. Pharmacokinetic-pharmacodynamic (PKPD) modelling and simulation (M&S) is a promising tool in quantitatively characterising longitudinal antibiotic drug effects, especially in exploring antibiotic combinations where the concentration ratios of the component drugs, their interactions and relative contributions to the combined drug effects change over time. This thesis aimed to develop approaches to quantify the effects of antibiotic combinations on gram-negative bacteria by PKPD M&S. Pre-clinical data for antibiotics alone and in combination, including in vitro static and dynamic time-kill data and in vivo murine thigh infection data, were quantitatively described by semi-mechanistic PKPD models, as exemplified in the thesis by the combination of colistin and ciprofloxacin against four Escherichia coli strains and by the combination of polymyxin B and minocycline against two Klebsiella pneumoniae strains. The developed PKPD models showed good translational ability, as demonstrated by bridging across strains, predicting in vitro drug effects under dynamic settings, integrating all available in vitro and in vivo data, and quantifying the discrepancy between the in vitro and in vivo settings. These developed models were used to explore the therapeutic potential of the combinations. The value of adding low dosages of colistin (thus low toxicity) to ciprofloxacin treatment for E. coli strains with decreased susceptibility, and of combining polymyxin B and minocycline against K. pneumoniae when either drug alone has a limited antibacterial effect, was shown. In addition, a methodological framework was developed to quantify antibiotic effects from time-lapse microscopic images, exemplified by meropenem-exposed Pseudomonas aeruginosa and ertapenem-exposed E. coli. The framework could be extended to the antibiotic combinations in the future. To conclude, this thesis illustrated how semi-mechanistic PKPD M&S could be leveraged to evaluate the effects of antibiotic combinations quantitatively and to understand the antibiotic-bacteria interaction. As a future perspective, in a world where the rational use of antibiotic combinations is important to overcome and minimise resistance, semi-mechanistic PKPD M&S approaches could be applied to streamline the development and use of antibiotic drug combinations