Evaluation of polymyxin B in combination with 13 other antibiotics against carbapenemase-producing Klebsiella pneumoniae in time-lapse microscopy and time-kill experiments.

This study aimed to explore the interactions of polymyxin B in combination with 13 other antibiotics against carbapenemase-producing Klebsiella pneumoniae. Five clinical isolates of multidrug-resistant K. pneumoniae producing KPC-2, KPC-3, NDM-1, OXA-48 and VIM-1 carbapenemases were used. Polymyxin B was tested alone and in combination with amikacin, aztreonam, cefepime, chloramphenicol, ciprofloxacin, fosfomycin, linezolid, meropenem, minocycline, rifampicin, temocillin, thiamphenicol and trimethoprim. Inhibition of growth during antibiotic exposure was evaluated in 24-hr automated time-lapse microscopy experiments. Combinations that showed positive interactions were subsequently evaluated in static time-kill experiments. All strains carried multiple (≥9) resistance genes as determined by whole-genome sequencing. In the initial screening the combination of polymyxin B and minocycline was most active with enhanced activity compared with the single antibiotics detected against all strains. Positive interactions were also observed with polymyxin B in combination with rifampicin and fosfomycin against four of five strains and less frequently with other antibiotics. Time-kill experiments demonstrated an additive or synergistic activity (1-2 log10 or ≥2 log10 CFU/mL reduction, respectively, compared with the most potent single antibiotic) with 21 of 23 tested combinations. However, because of regrowth, only 13 combinations were synergistic at 24 hr. Combinations with minocycline or rifampicin were most active, each showing synergy and bacteriostatic or bactericidal effects resulting in 1.93-3.97 and 2.55-5.91 log10 CFU/mL reductions, respectively, after 24 hr against four strains. Polymyxin B in combination with minocycline, rifampicin or fosfomycin could be of potential clinical interest. Time-lapse microscopy showed some discrepancy in results compared with the time-kill data but was useful for screening purposes.