Susceptibility testing accuracy of a CTX-M–type extended-spectrum β-lactamase organism-producing population of Enterobacteriaceae: intermethod analysis for 9 β-lactams

Abstract

To assess the wide geographical applicability of the current and proposed susceptibility breakpoint criteria for 9 β-lactam antimicrobials, the performance characteristics of 2 standardized methods were analyzed by testing a contemporary collection of 354 isolates of Enterobacteriaceae, enriched (76; 21.5%) for extended-spectrum β-lactamase (ESBL)–producing strains. Molecular characterization of 57 ESBL strains revealed that majority of the strains (94.7%) were CTX-M type, with a predominance (85.2%) of CTX-M-14 and -3 types, those types prevalent in China. Bloodstream isolates constituted 68.6% of the entire collection. The 9 β-lactam antimicrobials analyzed were aztreonam, cefepime, cefotaxime, cefotetan, cefoxitin, ceftazidime, ceftriaxone, ceftizoxime, and cefuroxime. Reference broth microdilution minimal inhibitory concentration (MIC) results were compared to those zone diameters obtained by disk diffusion testing. The regression coefficient was acceptable for most antimicrobials, ranging from r = 0.84 (cefotetan) to r = 0.98 (cefotaxime, cefuroxime, and ceftriaxone). Using the current breakpoint criteria, the absolute intermethod categorical agreement was acceptable for 8 of the 9 antimicrobials (not cefoxitin, 85.1%) ranging from 92.6% (cefotaxime) to 97.8% (ceftazidime). Very major (false-susceptible) and major (false-resistant) errors were nil (0.0%) for 5 of the β-lactams and minor errors ranged from 0.8% (cefotetan) to 14.1% (cefoxitin). The proposed (generally lower) MIC breakpoint criteria also had acceptable intermethod concordance ranging from 91.6% (cefoxitin) to 99.2% (cefotaxime and ceftriaxone). Furthermore, an improvement in the intermethod absolute categorical agreement ranging from +0.8% (ceftazidime) to +6.6% (cefotaxime) was observed for 7 of 9 antimicrobials tested, including the ESBL screening test compounds (aztreonam, ceftazidime, cefotaxime, and ceftriaxone). No modification of the susceptible breakpoint criteria, with removal of the intermediate category, was proposed for cefoxitin and cefuroxime, resulted in a shift of error type from minor to major or very major, but the categorical agreement improved (+ ≥1.2%) for both cephems. In conclusion, our results confirm that both the current and the proposed MIC breakpoint criteria with appropriately selected zone diameter correlates have acceptable intermethod error rates even when tested against an Enterobacteriaceae collection enriched with CTX-M–type ESBL-producing strains that are endemic in locations (China) outside the United States.