PKPD-modeling of time-kill curves from E. coli mutants exposed to ciprofloxacin
David Khan (1), Pernilla Lagerbäck (2), Cao Sha (3), Diarmaid Hughes (3), Otto Cars (2), Dan I. Andersson (4), Lena E. Friberg (1)
(1) Department of Pharmaceutical Biosciences, (2) Department of Medical Sciences, (3) Department of Cell and Molecular Biology, and (4) Department of Medical Biochemistry and Microbiology, Uppsala University, Sweden
Background: There is a need to better understand how antibiotics should be dosed to overcome and minimize development of resistance and much information can be gained from modeling and simulations based on in vitro experiments of bacteria kill. A semi-mechanistic model describing in vitro antibiotic effects has been proposed based on a strain of Streptococcus pyogenes [1]. The model includes two subpopulations of bacteria, one drug-susceptible population and one resting insusceptible population. The semi-mechanistic model structure allows for application on other types of bacteria with parameter estimates dependent on the degree of bacterial resistance and fitness. The aim of this work was to develop a PKPD-model describing the time-kill curves of wild-type and three well-characterized mutants of E. coli exposed to different concentrations of ciprofloxacin.
Methods: Time-kill curve data from 24h static in vitro experiments with E. coli MG1655 and three mutants thereof were used for model development. Ciprofloxacin concentrations were constant during the experiment and ranged from 0.06 to 8 x MIC for each bacterial strain. Modeling was performed in NONMEM 7. The drug effect on the susceptible bacteria was described by an Emax-model. Differences in the parameters for the wild-type and resistant bacteria were searched for. Regrowth in the wild-type bacteria experiments were described as pre-existing resistance for some bacteria in the starting inocula.
Results: Time-kill curves for all investigated strains and concentrations were well predicted by the model. Growth rates were 10% lower for the three mutants compared to wild type. EC50 was higher for the mutants and different for all strains. Emax was the same for the wild type and two of the mutants while one mutant had a 30-fold higher Emax. Allowing for pre-existing resistant bacteria in the wild-type starting inocula resulted in a decrease in OFV of 75 units with an estimate of 8 bacteria per 106 bacteria.
Conclusions: The model successfully described the time-kill curves following ciprofloxacin exposure for all investigated mutants and explained the regrowth occurring in the wild type bacteria experiments. The model can be used to predict the time-course of bacterial kill following different dosing regimens and in the presence of mixtures of wild-type and mutant bacteria, and may thereby be a valuable tool in the search for dosing regimens that minimize the growth of resistant mutants.
References:
[1] Nielsen, E.I., et al., Semimechanistic pharmacokinetic/pharmacodynamic model for assessment of activity of antibacterial agents from time-kill curve experiments. Antimicrob Agents Chemother, 2007. 51(1): p. 128-36.
