Abstract
Clinical isolates of Escherichia coli (n = 554) were tested against fosfomycin using agar dilution, disk diffusion, and Etest. Agar dilution (reference method) identified few isolates with fosfomycin MICs of 64 (n = 3), 128 (n = 4), and ≥256 μg/ml (n = 2). Applying CLSI (M100, 2020) and EUCAST (v. 10.0, 2020) breakpoints, 98.9% and 98.4% (agar dilution), 99.3% and 99.1% (disk diffusion), and 99.1% and 98.9% (Etest) of isolates were fosfomycin susceptible, respectively. Essential agreement (agar dilution versus Etest) was low (40.8%); 59.3% (131/221) of isolates with agar dilution MICs of 2 to 128 μg/ml tested 2 to 4 doubling dilutions lower by Etest. Applying CLSI breakpoints, categorical agreement was >99% for both disk diffusion and Etest; no major errors (MEs) or very major errors (VMEs) were identified, and rates of minor errors (mEs) were <1%. EUCAST breakpoints yielded categorical agreements of >99% and no MEs for both disk diffusion and Etest; however, VMEs occurred at unacceptable rates of 44.4% (disk diffusion) and 33.3% (Etest). All isolates with agar dilution MICs of ≥32 μg/ml (n = 12) and a subset of isolates with MICs of ≤16 μg/ml (n = 49) were also tested using the Vitek 2 AST-N391 card and generated fosfomycin MICs 1 to ≥3 doubling dilutions lower than agar dilution for 11/12 isolates with agar dilution MICs of ≥32 μg/ml. We conclude that performing fosfomycin disk diffusion or Etest on urinary isolates of E. coli and interpreting results using CLSI breakpoints reliably identified fosfomycin-susceptible isolates regardless of differences in endpoint reading criteria. EUCAST breakpoints generated excessive rates of VMEs for our isolate collection of high fosfomycin susceptibility.