Abstract
BACKGROUND: Ceftazidime-avibactam (CZA), ceftolozane-tazobactam (CT), and imipenem-relebactam (IMR) are newer β-lactam/β-lactamase inhibitor (BL/BLI) combinations used for treatment of multidrug resistant (MDR) Pseudomonas aeruginosa, although resistance has already emerged. Optimal use of these agents relies on timely, accurate susceptibility testing results and understanding of local resistance mechanisms. METHODS: 183 MDR P. aeruginosa clinical isolates were used to evaluate the performance of commercial Sensititre and Phoenix panels for CZA, CT, and IMR susceptibility testing compared to broth microdilution (BMD). Genomic resistance determinants were also predicted for each isolate with AMRFinderPlus. RESULTS: Categorical agreement (CA) of the Sensititre panel compared to BMD was 95.8%, 90.1%, and 95.8% for CZA, CT, and IMR, respectively. CA of the Phoenix panel was 83.0% for CZA and 85.7% for CT. The Phoenix panel was biased toward under-calling CZA resistance while error rates were acceptable for CT by the error-rate-bound method. AMRFinderPlus identified acquired β-lactamases in 4.6% of first-time patient isolates. CA of genotype with BMD was 74.9% for CZA, 91.9% for CT, and 90.7% for IMR. However, all agents had unacceptably high VME rates using genotypic testing because many phenotypically resistant isolates had no identifiable genotypic resistance determinants. CONCLUSIONS: The Sensititre panel met standard acceptance criteria while the Phoenix panel had low CA for all tested BL/BLIs compared to BMD. Acquired β-lactamases were a rare cause of BL/BLI resistance. Further understanding of resistance mechanisms is required before phenotypic BL/BLI resistance can be reliably predicted from genotype, especially in settings where prevalence of acquired β-lactamases is low.