Abstract
Susceptibility results with low reproducibility by the same or different methods have been observed for metallo-beta-lactamase (MBL)-producing Enterobacteriaceae. Eighteen VIM-1-producing Klebsiella pneumoniae isolates (one per patient) belonging to a single epidemic clone in our hospital (2005 to 2008) but with different susceptibilities to carbapenems were studied. Imipenem MICs ranged from 8 to >128 mg/liter by standard CLSI microdilution, from ≤1 to >8 mg/liter by the semiautomatic Wider system, and from 0.75 to >32 mg/liter by Etest. Meropenem MICs ranged from 0.5 to 128, ≤1 to >8, and 0.38 to >32 mg/liter, respectively. Ertapenem MICs by CLSI microdilution and Etest ranged from 1 to 64 and 0.75 to >32 mg/liter, respectively. The rates of essential agreement (±1 log(2) dilution) for imipenem and meropenem MICs between the Wider system and the reference microdilution method were 45% and 49%, respectively. Those between Etest and the reference microdilution method for imipenem, meropenem, and ertapenem MICs were 33%, 67%, and 84%. The rates of very major errors for the Wider system and Etest were 33% and 28% for imipenem and 25% and 75% for meropenem, respectively. Low MIC reproducibility was observed even when the same inoculum was used (differences up to 4-fold dilutions). Heteroresistance was suspected due to the presence of colonies in the Etest inhibition zone. It was confirmed by population analysis profiles of 4 isolates displaying different imipenem MICs, with the exception of an OmpK36-porin-deficient isolate that homogeneously expressed carbapenem resistance (MIC, >128 mg/liter). Low carbapenem MIC reproducibility could be due to the presence of resistant subpopulations and variable expression of the resistance mechanisms. Since carbapenem MICs are not good markers of MBL production, reliable and reproducible phenotypic methods are needed to detect the presence of this mechanism.