Mechanism-based modelling to assess suppression of bacterial resistance by high intensity, short duration tobramycin exposure


Aims: Infections involving hypermutable Pseudomonas aeruginosa (Pa) are highly problematic due to their enhanced resistance emergence. For aminoglycosides such as tobramycin (TOB), the area under the unbound concentration-time curve divided by the MIC (fAUC/MIC) is used to predict bacterial killing and clinical success. We hypothesized that delivering the same fAUC over short vs. long durations of exposure would provide better killing and minimize TOB resistance in both hypermutable and non-hypermutable Pa. To test this hypothesis, we aimed to quantitatively characterise bacterial killing and emergence of resistance of Pa for different concentration-time profiles with the same overall TOB exposure using in vitro time-kill experiments and mechanism-based modelling (MBM).


 Methods: P. aeruginosa PAO1 (wild-type) and PAOΔmutS (hypermutable) were studied in 24 h in vitro static concentration time-kill experiments with TOB (MIC=0.5 mg/L for both strains). This was carried out at fAUC/MIC of 36, 72 and 168 with initial inocula of 104 and 106 colony forming units (cfu)/mL, in duplicate. Antibiotic concentrations were calculated by the exposure durations and targeted fAUC/MIC. TOB was added at 0 h and removed at 1, 4, 10 or 24 h via multiple centrifugation and re-suspension of bacteria in antibiotic-free broth.  The time-courses of bacterial killing and resistance were determined through viable counts of the total and resistant populations and then characterized by novel MBM (80 profiles). The importance sampling algorithm (S-ADAPT pmethod=4) was used in parallelised S-ADAPT (version 1.57), facilitated by SADAPT-TRAN. The model included 3 bacterial populations to represent the different susceptibilities (TOB susceptible, TOB intermediate and TOB resistant). The bacterial killing by TOB was described by direct killing functions with a post antibiotic effect (PAE). The emergence of TOB resistance was included in the model as amplification of pre-existing resistant mutants and adaptive resistance.


 Results: For both PAO1 and PAOΔmutS at the same fAUC/MIC more rapid and extensive killing was observed with high concentrations over 1 and 4 h durations of exposure (4 to 6 log10) compared to 10 and 24 h exposures (< 4 log10). Regrowth at the 24 h time point was extensive (up to 9 log10) after 1 and 4 h duration of exposures, but bacteria which regrew remained TOB susceptible with unchanged mutation frequencies compared to the untreated control. While there was less pronounced regrowth (0 to 6 log10) for the 24 and 10 h exposure durations, susceptible bacteria had been completely replaced by TOB resistant bacteria at 24 h for the 106 cfu/mL inoculum. The hypermutable PAO∆mutS revealed higher MF at 0 and 24 h in comparison to PAO1. The killing rate constant (Kmax) was estimated to be over 3 times higher for the susceptible compared to resistant populations, with the KC50 being over 28 times higher in the resistant populations. Incorporation of functions for both pre-existing and adaptive resistance in the model was required. The MBM well described the time-courses of the total and resistant bacterial populations simultaneously (r=0.97).


 Conclusion: The concentration-time profile shape was important to prevent resistance emergence. Short durations of TOB exposure yielded extensive killing with resistance prevention, whereas emergence of resistance was substantial for 24 and 10 h durations. Therefore, high intensity, short durations of TOB exposure are highly promising for use in innovative combination dosage regimens where the second antibiotic would combat the TOB-susceptible regrowth. MBM allowed us to quantitatively characterise the time-course of TOB resistance emergence. Simulations using this model will aid the optimisation of future combination dosage regimens that allow resistance prevention.