Does the presence of multiple β-lactamases in Gram-negative bacilli impact the results of antimicrobial susceptibility tests and extended-spectrum β-lactamase and carbapenemase confirmation methods?

•Disk diffusion testing remains accurate even in the presence of multiple β-lactamases.•mCIM/eCIM and Xpert Carba-R accurately differentiate between serine and metallo-β-lactamases.•ESBL screening tests produced inconsistent results that may confound therapy.•Predicting carbapenem resistance from WGS data produces very major errors. Many multidrug-resistant Gram-negative bacilli (MDR-GNB) harbour multiple β-lactamases. The aim of this study was to assess the impact of multiple β-lactamase carriage on the accuracy of susceptibility tests and extended-spectrum β-lactamase (ESBL) and carbapenemase confirmation methods. A total of 50 MDR-GNB, of which 29 carried multiple β-lactamases, underwent broth microdilution (BMD) and disk diffusion (DD) testing as well as confirmation tests for ESBLs and carbapenemases. Whole-genome sequencing (WGS) was used for β-lactamase gene identification. Categorical agreement of BMD and DD testing results ranged from 86.5 to 97.7% for 10 β-lactam agents. BMD and DD algorithms for ESBL detection were highly variable; 6 of 8 positive strains carried an ESBL plus a carbapenemase or an AmpC enzyme, which may confound antimicrobial selection. The sensitivity and specificity of the modified carbapenem inactivation method (mCIM) were both 100%, whilst mCIM and EDTA-modified carbapenem inactivation method (eCIM) when used together to differentiate serine from metallo-β-lactamase carriage were both 96%. Xpert® Carba-R results (in vitro diagnostic test) were consistent with WGS results. Predicting phenotypic carbapenem resistance from WGS data overall showed 100% specificity but only 66.7% sensitivity for Enterobacterales isolates that were non-susceptible to imipenem and meropenem. Multiple β-lactamases in MDR-GNB does not impact DD results, the utility of mCIM/eCIM tests, or Xpert Carba-R results. However, ESBL algorithms produced inconsistent results and predicting carbapenem resistance from WGS data was problematic in such strains.