Performance of different methods for testing polymyxin B: comparison of broth microdilution, agar dilution and MIC test strip in mcr-1 positive and negative Escherichia coli.

Antimicrobial susceptibility testing with the last-resort antibiotics polymyxins (polymyxin B and colistin) is associated with several methodological issues. Currently, broth microdilution (BMD) is recommended for colistin and polymyxin B. BMD is laborious and the utility of alternative methods needs to be evaluated for polymyxin B susceptibility testing. In this study, using BMD as a reference method, the performance of agar dilution (AD) and MIC test strips (MTS) were evaluated in polymyxin B susceptibility testing. BMD, AD and MTS were used to determine MICs of 193 clinical isolates of Escherichia coli. Seventy-nine were positive for the polymyxin resistance gene mcr-1. Method performances were evaluated based on pair-wise agreements with the reference method (BMD) and statistical testing. AD and MTS showed an unacceptable number of very major errors (VMEs) compared with BMD, 9·3 and 10·7%, respectively. The essential agreement (EA) was low for AD (49·7%), but high for MTS (97·8%). However, statistical testing showed that MTS tended to yield a one-step lower MIC (P < 0·01) compared with BMD. The discordances observed with MTS and AD in comparison with BMD for polymyxin B susceptibility testing for E. coli suggest their inapplicability in routine testing. A large number of isolates clustered around the susceptibility breakpoint (2-4 mg l ) and several mcr-1 positive isolates (17%) were determined as susceptible with BMD. A screening breakpoint for mcr-1 of 2 mg l should also be considered.