Predicting mutant selection in competition experiments with ciprofloxacin-exposed Escherichia coli

Highlights

• A previously developed PK/PD model was challenged to predict time–kill competition experiments.

• Selection of wild-type versus mutant strains was reasonably well forecast.

• PK/PD models can be of value in exploring the impact of exposure on resistance selection.

Abstract

Predicting competition between antibiotic-susceptible wild-type (WT) and less susceptible mutant (MT) bacteria is valuable for understanding how drug concentrations influence the emergence of resistance. Pharmacokinetic/pharmacodynamic (PK/PD) models predicting the rate and extent of takeover of resistant bacteria during different antibiotic pressures can thus be a valuable tool in improving treatment regimens. The aim of this study was to evaluate a previously developed mechanism-based PK/PD model for its ability to predict in vitro mixed-population experiments with competition between Escherichia coli (E. coli) WT and three well-defined E. coli resistant MTs when exposed to ciprofloxacin. Model predictions for each bacterial strain and ciprofloxacin concentration were made for in vitro static and dynamic time–kill experiments measuring CFU (colony forming units)/mL up to 24 h with concentrations close to or below the minimum inhibitory concentration (MIC), as well as for serial passage experiments with concentrations well below the MIC measuring ratios between the two strains with flow cytometry. The model was found to reasonably well predict the initial bacterial growth and killing of most static and dynamic time–kill competition experiments without need for parameter re-estimation. With parameter re-estimation of growth rates, an adequate fit was also obtained for the 6-day serial passage competition experiments. No bacterial interaction in growth was observed. This study demonstrates the predictive capacity of a PK/PD model and further supports the application of PK/PD modelling for prediction of bacterial kill in different settings, including resistance selection.

Introduction

The increasing selection of antibiotic-resistant bacteria is an important global healthcare issue, especially for Gram-negative bacteria. To meet the need for efficient treatments, the development of new antibacterial agents has regained interest, with 40 antibiotics in clinical programmes for the US market as of September 2016 [1], [2], [3]. Both for new antibacterial agents and for currently existing antibiotics it is critical to identify dosing regimens that minimise the selection of resistance, and consequently it is of importance to understand the dynamics of resistance selection [4].

Pharmacokinetic/pharmacodynamic (PK/PD) modelling has been suggested as a tool for improved use of existing antibiotics as well as to increase the efficiency of the development pipeline of new antibiotics [5]. Models that can predict drug effects and resistance development can be highly valuable tools, with the potential to promote optimised dosing regimens and a more streamlined drug development process. We have previously demonstrated the potential of PK/PD models developed based on in vitro data to predict PK/PD indices in vivo [5], [6], [7].

Mechanism-based PK/PD models incorporate knowledge of the studied system. During the last decades, in silico PK/PD models have been developed for a variety of different bacterial strains and antibiotics [5], [8], [9]. PK/PD models predicting outcomes outside the traditional experimental setting are valuable for making drug development more efficient and attractive [5] and may allow for more reliable adjusted dosing regimens in special patient populations. Mechanism-based models are typically believed to have increased predictive capacity compared with empirical models [10], [11]. A mechanism-based PK/PD model previously developed by Nielsen et al. [12] that considers the co-existence of growing drug-susceptible and non-growing non-susceptible bacteria has been shown to be applicable to a variety of different antibiotics, and the basic structure may be shared across drugs and bacterial strains [12], [13], [14]. The original model structure has been further developed to describe adaptive resistance of Pseudomonas aeruginosa to colistin [14] and to describe the growth and killing of Escherichia coli (E. coli) wild-type (WT) and mutant (MT) bacteria exposed to ciprofloxacin [13]. The latter model successfully predicted the time course of bacterial growth and killing kinetics in experiments with high bacterial densities in the starting inoculum as well as for several new E. coli MTs [15].

To understand the relevance of a MT in the presence of susceptible bacteria, two bacterial strains can be mixed in the starting inoculum to study the competition dynamics under different antibacterial drug pressures [16]. Since such experiments can result in a wide range of conditions to study that would be quite resource demanding, it would be valuable if PK/PD models can be applied to predict the selection of resistance under different antibiotic pressure.

The aim of this work was to evaluate the predictive performance of a mechanism-based PK/PD model [13] for in vitro competition experiments between E. coli WT and three well-defined E. coli resistant MTs exposed to a range of ciprofloxacin concentrations close to or below the minimum inhibitory concentration (MIC).