Comparison of BD Phoenix and disk diffusion to broth microdilution for determining cefepime susceptibility among carbapenem-resistant Enterobacterales.

Abstract

There are increasing reports of carbapenem-resistant Enterobacterales (CRE) that test as cefepime-susceptible (S) or susceptible-dose dependent (SDD). However, there are no data to compare the cefepime testing performance of BD Phoenix automated susceptibility system (BD Phoenix) and disk diffusion (DD) relative to reference broth microdilution (BMD) against carbapenemase-producing (CPblaKPC-CRE) and non-producing (non-CP CRE) isolates. Cefepime susceptibility results were interpreted according to CLSI M100Ed32. Essential agreement (EA), categorical agreement (CA), minor errors (miEs), major errors (MEs), and very major errors (VMEs) were calculated for BD Phoenix (NMIC-306 Gram-negative panel) and DD relative to BMD. Correlates were also analyzed by the error rate-bounded method. EA and CA for CPblaKPC-CRE isolates (n = 64) were <90% with BD Phoenix while among non-CP CRE isolates (n = 58), EA and CA were 96.6%, and 79.3%, respectively. CA was <90% with DD for both cohorts. No ME or VME was observed for either isolate cohort; however, miEs were >10% for CPblaKPC-CRE and non-CP CRE with BD Phoenix and DD tests. For error rate-bounded method, miEs were <40% for IHigh + 1 to ILow - 1 ranges for CPblaKPC-CRE and non-CP CRE with BD Phoenix. Regarding disk diffusion, miEs were unacceptable for all MIC ranges among CPblaKPC-CRE. For non-CP CRE isolates, only IHigh + 1 to ILow - 1 range was acceptable at 37.2%. Using this challenge set of genotypic-phenotypic discordant CRE, the BD Phoenix MICs and DD susceptibility results trended higher (toward SDD and resistant phenotypes) relative to reference BMD results yielding lower CA. These results were more prominent among CPblaKPC-CRE than non-CP CRE.