Abstract
The emergence of plasmid-mediated resistance threatens the efficacy of polymyxins as the last line of defense against pan-drug-resistant infections. However, we have found that using Mueller-Hinton II (MHII), the standard minimum inhibitory concentration (MIC) medium, results in MIC data that are disconnected from in vivo treatment outcomes. We found that culturing putative colistin-resistant Acinetobacter baumannii clinical isolates, as defined by MICs of >2 mg/L in standard MHII testing conditions, in bicarbonate-containing media reduced MICs to the susceptible range by preventing colistin resistance-conferring lipopolysaccharide modifications from occurring. Furthermore, the lower MICs in bicarbonate-containing media accurately predicted in vivo efficacy of a human-simulated dosing strategy of colistin and polymyxin B in a lethal murine infection model for some polymyxin-resistant A. baumannii strains. Thus, current polymyxin susceptibility testing methods overestimate the contribution of polymyxin resistance-conferring mutations and incorrectly predict antibiotic activity in vivo. Polymyxins may remain a viable therapeutic option against Acinetobacter baumannii strains heretofore determined to be "pan-resistant."