Mechanism-based modelling and sequential dosing to elucidate subpopulation synergy for antibiotic combinations

Background: Rational and prospective approaches to optimize antibiotic combination therapy are scarce. Due to the emerging global healthcare crisis caused by multidrug-resistant bacteria, more efficient and quantitative approaches to optimize combination therapy are urgently required. Nisin, a peptide antibiotic affecting pore formation in bacterial membranes and preventing peptidoglycan synthesis, is rapidly bactericidal against multiple-resistant Staphylococcus aureus (MRSA). Quantitative modelling of antimicrobial combination therapy is a powerful tool to elucidate synergy mechanisms against multiple bacterial populations.

Objectives: To develop an efficient experimental strategy for mechanism-based modelling that evaluates different synergy mechanisms and to model the synergy of nisin and amikacin as well as nisin and linezolid against MRSA.

Methods: Time-kill experiments were performed over 48 h against an MRSA (USA300) strain. An initial inoculum of 10^8 CFU/mL was used and serial viable counts were determined for various concentrations of nisin and amikacin. All three antibiotics were studied in monotherapy. A sequential design (initial dosing of 8 or 32 mg/L of nisin, switched to amikacin or linezolid at 1.5h) was used to quantify the rate of growth and rate of killing of the nisin-intermediate and nisin-resistant populations by amikacin or linezolid. Simultaneous combinations were subsequently studied. All profiles were modelled simultaneously using NONMEM.

Results: A mechanism-based model with six subpopulations in total (three for nisin times two for amikacin) provided unbiased and precise (r=0.94, slope=1.00) population predictions. Pre-treatment with 8 mg/L nisin killed the nisin-susceptible subpopulation and pre-treatment with 32 mg/L nisin killed both the nisin-susceptible and nisin-intermediate subpopulation. The EC50 of amikacin for stimulation of killing was 16.5 mg/L, and maximum killing rate constants were 9.1 1/h for the susceptible, and 0.64 1/h for the subpopulation less susceptible to amikacin. The second-order killing rate constants for nisin were 3.58, 0.0605, and 0.00484 L/(mg•h) against the three subpopulations. After accounting for the effects of nisin and amikacin on each subpopulation a model with additive effects described all profiles well and no additional synergy function was necessary. Linezolid decreased the probability of successful bacterial replication but did not efficiently kill subpopulations that were less susceptible to nisin.

Conclusions: Subpopulation synergy was found for nisin plus amikacin against MRSA. It was essential to model these effects using multiple bacterial subpopulations. The proposed sequential and simultaneous dosing design combined with mechanism-based modelling offers an efficient approach to quantitatively characterize antibiotic synergy over time and prospectively evaluate antibiotic combination dosing strategies.