Abstract

Ceftolozane/tazbactam (C/T) is a potent anti-pseudomonal agent that has clinical utility against infections caused by non-carbapenemase, producing-carbapenemresistant Pseudomonas aeruginosa (non-CP-CR-PA). Accurate, precise, and reliable antimicrobial susceptibility testing (AST) is crucial to guide clinical decisions. However, studies assessing the performance of different AST methods against non-CP-CR-PA (the main clinical niche for C/T), are lacking. Here, we evaluated performance of gradient strips (Etest and MIC test strip [MTS], and disk diffusion [DD]) using CLSI breakpoints. Additionally, we assessed the performance of DD using EUCAST breakpoints. For all susceptibility tests, we used a collection of 97 non-CP-CR-PA clinical isolates recovered from 11 Chilean hospitals. Both gradient strips and DD had acceptable performance when using CLSI breakpoints, yielding a categorical agreement (CA) of .90% and 92%, respectively. In contrast, DD using EUCAST breakpoints performed suboptimally (CA 81%). MTS yielded a higher essential agreement (EA, .90%) than Etest (84%). Importantly, the performance of all methods varied significantly when the isolates were stratified by their degree of susceptibility to other anti-pseudomonal b-lactams. All methods had 100% CA when testing isolates that were pan-susceptible to all b-lactams (Pan-β-S). However, the CA markedly decreased when testing isolates resistant to all b-lactams (Pan-β-R). Indeed, the CA was 81% for Etest (six errors), 78% for MTS (seven errors), and 78% and 56% for DD when using CLSI (seven errors) or EUCAST breakpoints (14 errors), respectively. Our results suggest that all manual AST methods have strikingly decreased performance in the context of Pan-β-R P. aeruginosa with potentially major clinical implications.