Comparative evaluation of eight in vitro pharmacodynamic models of infection: Activity of moxifloxacin against Escherichia coli and Streptococcus pneumoniae as an exemplary example

Abstract

Use of pharmacodynamic in vitro models provides more clinically relevant information about the activities of antibiotics than static endpoints. Several models are used to simulate pharmacokinetics by dilution of the medium. It is discussed whether this procedure would result in a washout of bacteria, particularly if profiles with a short half-life are simulated. Methods have been developed to minimise the washout of bacteria. Bacteria are retained in the system either by centrifugation and resuspension, use of filters, a capillary unit, dialysis tubing or mathematical correction, versus systems with an unprotected outflow allowing a continuous washout of bacteria. None of these eight models has been directly compared with another. Therefore, an interlaboratory study was performed to address the question of whether or not washout matters. All laboratories used identical batches of media, bacteria, antibiotics and simulated pharmacokinetic profiles with a short or long half-life. Values of area under the bacterial kill-time curve (AUBKC), single-point kill rate and time to 3-log10 reduction of inoculum were calculated. These parameters did not differ significantly between the models. Differences were noted if the inoculum was prepared from the early logarithmic growth phase compared with the late logarithmic or stationary growth phase, resulting either in a pronounced or reduced antibacterial activity. Thus, preparation of inocula affects the results generated, whereas washout of bacteria has apparently a negligible impact on antibacterial activities.