Mechanism-based Modelling of the Antagonism between Polymyxin B and Levofloxacin against Pseudomonas aeruginosa using Mono- and Combination Therapy
H. Xu(1), A. Forrest(1,2), B. T. Tsuji(1), J. B. Bulitta(1,2)
(1) School of Pharmacy and Pharmaceutical Sciences, SUNY at Buffalo, Buffalo, NY, USA (2) Ordway Research Institute, Albany, NY, USA
Objectives: Polymyxins have become the last line of defence against multidrug resistant Pseudomonas aeruginosa and other gram-negative bacteria. Polymyxin B (PB) first displaces Ca2+ and Mg2+ from their binding sites on the outer membrane and causes leakage of intracellular components and bacterial killing. Levofloxacin (LEV), a fluoroquinolone targeting DNA gyrase, has different mechanisms of action and of resistance compared to PB. Our objective was to develop a mechanism-based population pharmacodynamic (PD) model for the time course of killing of P. aeruginosa by PB and LEV alone or in combination via an Importance Sampling algorithm.
Methods: In vitro time kill experiments were performed using eight PB concentrations up to 64 mg/L and LEV from 1 to 16 mg/L against P. aeruginosa (strain PAO1) at an initial inoculum of 108 colony forming units per mL (CFU/mL). In addition to monotherapy by PB or LEV, bacterial killing was studied by simultaneous and sequential dosing. For sequential dosing, cells were exposed to 4 mg/L PB for 1.5 h, PB was then removed by centrifugation and resuspension in fresh broth containing LEV (or PB). Serial viable counts were determined over 24 to 48 h. All log10 CFU/mL were simultaneously modelled using the importance sampling Monte Carlo Parametric Expectation Maximization (MC-PEM) algorithm in parallelized S-ADAPT (v. 1.57) and SADAPT-TRAN.
Results: The rapid, concentration-dependent killing of P. aeruginosa by PB (4 log10 killing at 2 h for 4 mg/L and 6 log10 for 8 mg/L PB) was modelled as a second order killing process subject to a competitive inhibition with divalent cations . Levofloxacin was assumed to stimulate the natural death rate constant up to 17-fold with an SC50 of 2.8 mg/L. The model contained 5 subpopulations with different susceptibility to each antibiotic. A joint resistant population was not required. There was a slight, but noticeable antagonism when PB and LEV were given simultaneously with 1-2 log10 less killing compared to LEV monotherapy. PB was assumed to increase the SC50 of LEV.
Conclusions: The proposed mechanism-based PD model described the rate and extent of killing by PB and LEV in monotherapy as well as in sequential and simultaneous combination therapy excellently. We suspect that polymyxin B may have up-regulated the efflux of levofloxacin.
 Bulitta JB, Yang JC, Yohonn L, et al. Antimicrob Agents Chemother 2010; 54:2051-62.