Susceptibility of Meropenem-Resistant and/or Carbapenemase-Producing Clinical Isolates of Enterobacterales (Enterobacteriaceae) and Pseudomonas aeruginosa to Ceftazidime-Avibactam and Ceftolozane-Tazobactam as Assessed by In Vitro Testing Methods

This study aimed to assess the comparability of in vitro susceptibility testing methods to ceftazidime-avibactam (CZA) and ceftolozane-tazobactam (C/T). Meropenem-resistant and/or carbapenemase-producing clinical isolates of Enterobacterales (Enterobacteriaceae) and Pseudomonas aeruginosa were tested by both bioMérieux ETEST and VITEK-2 AST-N397 card and compared with a Micronaut AST-system broth microdilution (BMD) method. CZA and C/T MICs were interpreted using EUCAST breakpoints. Of the 153 Enterobacteriaceae isolates, 55.6% and 0.0% (VITEK 2) and 56.9% and 0.0% (ETEST and BMD) were susceptible to CZA and C/T, respectively. Of 52 P. aeruginosa isolates, 50.0% and 40.4% (VITEK 2, ETEST, and BMD) were susceptible to CZA and C/T, respectively. The essential agreement (EA) was 96.1% (197/205; VITEK 2 versus BMD) and 95.6% (196/205; ETEST versus BMD) for CZA testing, whereas EA was 98.0% (201/205; VITEK 2 versus BMD) and 96.6% (198/205; ETEST versus BMD) for C/T testing. The categorical agreement (CA) was 98.0% (201/205; VITEK 2 versus BMD) and 100% (ETEST versus BMD) for CZA testing, whereas CA was 100% (VITEK 2 versus BMD) and 100% (ETEST versus BMD) for C/T testing. Categorical errors regarded four Enterobacteriaceae isolates. VITEK 2 and ETEST yielded equivalent CZA and C/T susceptibility testing results, compared to the BMD method, in such a clinical context.

In keeping with an ever-increasing demand for clinical microbiology laboratories to perform in vitro antimicrobial susceptibility testing (AST) for CZA and C/T in clinically relevant Gram-negative bacteria [5,6], it has become essential to assess the comparability of available AST methods for these antibiotics. To our knowledge, only three European studies (two from Germany and one from France) have recently reported on this issue [7][8][9], with one of them including the VITEK 2 AST-XN12 card (bioMérieux, Marcy l'Étoile, France; this card includes CZA and C/T in the panel of routinely tested antimicrobial agents) among assessed AST methods [8]. Consistently, studies used EUCAST breakpoints to categorize isolates from their collections of Enterobacterales (Enterobacteriaceae) and/or P. aeruginosa against CZA alone [8,9] or those of P. aeruginosa against CZA and C/T concomitantly [7]. Here, our isolates' collection was tailored to include meropenem-resistant and/or carbapenemase-producing Enterobacteriaceae and P. aeruginosa organisms. With this collection, we assessed the abilities of VITEK 2 AST-N397 card (the version currently marketed by bioMérieux in Italy) and ETEST (bioMérieux) to yield equivalent results for CZA or C/T susceptible/resistant isolates. Using the Micronaut AST system (Merlin Diagnostika, Bornheim, Germany) based broth microdilution (BMD) method as the reference method, agreement, and error rates of VITEK and ETEST results, respectively, were calculated.
Overall, there were 98.0% (201/205) of categorical agreement (CA) results for CZA by the VITEK 2 method as well as 100% (205/205) of CA results for CZA by the ETEST method or for C/T by both the methods ( Table 2). As detailed in Figure 1, VITEK 2 or ETEST methods generated MICs that differed from Micronaut AST MICs, respectively, up to 3-or 5-fold for CZA and up to 2-or 6-fold for C/T. However, differences went unnoticed by CA in 24 of 28 instances because the isolates' MICs to CZA or C/T were far below the EUCAST susceptible breakpoint (6 isolates) or far above the EUCAST resistant breakpoint (18 isolates). Only involving K. pneumoniae isolates when tested against CZA with the VITEK 2 (AST-N397 card) method, very major error (VME) and major error (ME) rates in our study were 1.1% (1/92) and 2.7% (3/113), respectively. Table 2. Agreement and error rates in testing 205 Gram-negative bacterial isolates against ceftazidime/avibactam and ceftolozane/tazobactam.

Species of Isolates
Tested (n)

Method Evaluated in Comparison with BMD
Ceftazidime/Avibactam (CZA) a Ceftolozane/Tazobactam (C/T) a % EA (n) % CA (n) % ME (n) b % VME (n) c % EA (n) % CA (n) % ME (n) b % VME (n) c  (expressed as µg/mL) obtained by testing Enterobacteriaceae (n = 153) and P. aeruginosa (n = 52) isolates with the BMD (reference), VITEK 2, or ETEST methods, respectively. Darker and lighter grey squares indicate, respectively, the number of isolates with a VITEK 2 or ETEST MIC corresponding to the BMD MIC and 1-log2 concentration for each of two antimicrobial agents. Numbers outside the squares indicate the isolates with VITEK 2 or ETEST MICs that differed by more than 1-log2 concentration from the BMD MIC, which determined essential agreement rates below 100% for ceftazidime/avibactam or ceftolozane/tazobactam, respectively. Dashed lines delimitate VME or ME areas to represent, respectively, the isolates falsely categorized as susceptible or resistant according to EUCAST clinical breakpoints. Table 3 summarizes the results from similar studies (i.e., European studies that used interpretive EUCAST breakpoints) conducted so far [7][8][9]. Like us, one study tested both CZA and C/T combination antibiotics using VITEK 2 and ETEST methods [7], whereas two other studies tested only CZA using the ETEST method [8,9]. The correlation between ceftazidime/avibactam (a) and ceftolozane/tazobactam (b) MICs (expressed as µg/mL) obtained by testing Enterobacteriaceae (n = 153) and P. aeruginosa (n = 52) isolates with the BMD (reference), VITEK 2, or ETEST methods, respectively. Darker and lighter grey squares indicate, respectively, the number of isolates with a VITEK 2 or ETEST MIC corresponding to the BMD MIC and 1-log 2 concentration for each of two antimicrobial agents. Numbers outside the squares indicate the isolates with VITEK 2 or ETEST MICs that differed by more than 1-log 2 concentration from the BMD MIC, which determined essential agreement rates below 100% for ceftazidime/avibactam or ceftolozane/tazobactam, respectively. Dashed lines delimitate VME or ME areas to represent, respectively, the isolates falsely categorized as susceptible or resistant according to EUCAST clinical breakpoints. Table 3 summarizes the results from similar studies (i.e., European studies that used interpretive EUCAST breakpoints) conducted so far [7][8][9]. Like us, one study tested both CZA and C/T combination antibiotics using VITEK 2 and ETEST methods [7], whereas two other studies tested only CZA using the ETEST method [8,9]. BMD, broth microdilution; EA, essential agreement; CA, categorical agreement; ME, major error; VME, very major error; CZA, ceftazidime/avibactam; C/T, ceftolozane/tazobactam. a Except for the present study, three other studies here listed are referred to as first author, year, and reference number. b Except for one study that used self-prepared BMD panels [9], three other studies used commercial BMD panels and, consequently, concentration ranges in these panels were less broaden. c The XN12 or AST-N397 cards used with the VITEK 2 method in two studies [7 and this study] are equivalent. d Currently available EUCAST breakpoints were used to interpret isolates' testing results in each study. e For VITEK 2 or ETEST methods, performance was assessed in comparison with the indicated BMD method, which was used as the reference method in each study.

Discussion
Comparing our study with that by Daragon et al. [7] revealed an almost identical number of Gram-negative isolates (200 (albeit restricted to P. aeruginosa) and 205, respectively) studied in total. However, numbers/proportions of P. aeruginosa isolates susceptible to CZA (34 isolates, 17.0%) or C/T (40 isolates, 20.0%) based on BMD MICs were relatively lower in that study [7]. Furthermore, a consistent number of isolates had CZA (and C/T) MICs near to the EUCAST breakpoints (±1-log 2 ); thus, in part, accounting for 18.1% (30/166) of false-resistant isolates (MEs) or 8.8% (3/34) of false-susceptible isolates (VMEs) when tested for CZA with the VITEK 2 (AST-XN12 card) method [7]. Otherwise, CZA testing with the ETEST method on the same isolates revealed an inverse situation, being the ME rate as low as 3.0% (5/166 isolates) and the VME rate as high as 14.7% (5/34 isolates). These error rates were reminiscent of those reported by Schaumburg et al. [8] for a clinical collection of MDR/XDR P. aeruginosa isolates tested against CZA (Table 3) but were dissimilar from the error rates found by Kresken et al. [9], which were equal to (ME) or a very little greater than (VME) zero. If looking at the Kresken et al. [9] study's findings for Enterobacterales (Enterobacteriaceae) and P. aeruginosa isolates at a whole (200 isolates, of which 34 were CZAresistant), we noticed them to be consistent with our study's findings (205 isolates, of which 92 were CZA-resistant). Since 2018 (the year the Kresken et al.'s study was published), until the present study (Table 3), the number of Gram-negative bacterial isolates with CZA or C/T MICs higher than the EUCAST susceptible breakpoint (if those studied by the authors [9] were from a clinical collection) appears to be substantially increased. However, the possibility raised by Daragon et al. [7] to overestimate the CZA resistance when testing clinical P. aeruginosa isolates with the VITEK 2 AST-XN12 card cannot be excluded based on the findings from our study.
Confirming independent U.S. multicenter evaluation findings [17,18], in a much more challenging set of isolates, we have shown that the performance of the VITEK 2 AST-N397 card to determine the susceptibility of clinically relevant Gram-negative bacterial species to CZA and C/T combination antibiotics was comparable to that of the ETEST method. However, performance assessment of both VITEK 2 and ETEST methods in our study might be biased because we used a comparison method (i.e., the Micronaut AST system) that is really a surrogate of the widely accepted, albeit less practicable, BMD reference method [19]. Otherwise, final assessment of VITEK 2/ETEST MIC results implied repeat testing of isolates with initial CA discrepancies. Knowing the reasons behind these discrepancies could make the occurrence of categorical errors with both VITEK 2 and ETEST methods very negligible in the future. Finally, our study was intrinsically limited by the relatively small number of Gram-negative bacterial isolates tested.

Agreement Analysis
Using the Micronaut AST system-based BMD as the reference method, essential agreement (EA; defined as MIC results differing by a maximum of 1-log 2 concentration), CA (defined as MIC results within the same susceptibility category), ME (defined as false resistance), and VME (defined as false susceptibility) assessments were performed according to standard criteria. Because the (BMD) reference method used in this study does not contain ceftazidime and ceftolozane concentrations lower/higher than 1/64 and 0.5/64 µg/mL, respectively, EA calculation for the VITEK 2 or the ETEST method included both isolates with and without on-scale results (i.e., those that fall within the test range of the reference method). This was in accordance with the AST systems guidance document published by the U.S. FDA in 2009 (http://www.fda.gov/MedicalDevices/ DeviceRegulationandGuidance/GuidanceDocuments/ucm080564.htm; accessed on 1 July 2022). When MIC results fell between conventional two-fold dilutions (half dilutions), ETEST results were rounded up to the next upper doubling dilution that was used with the BMD method. The above-mentioned EUCAST breakpoints allowed assessing the CA of the VITEK 2 or the ETEST method with BMD as well as the error rates for both VITEK 2 and ETEST methods. Isolates with VME or ME were retested with the VITEK 2 or the ETEST method, as appropriate, and repeat testing results were used for analysis (Table S2).

Conclusions
Our analysis of CZA and C/T MIC results of clinically relevant Enterobacteriaceae or P. aeruginosa organisms allowed us to show that CA was excellent (98%) for the VITEK 2 AST-N397 card and complete (100%) for the ETEST method. While both ETEST and VITEK 2 allow testing the susceptibility to CZA and C/T with less hands-on time compared to reference method, the automation brought by VITEK 2 also allows its use in a routine workflow. ETEST remains an accurate alternative when these drugs are not yet available in VITEK 2 AST cards.

Informed Consent Statement: Not applicable.
Data Availability Statement: Data may be available upon reasonable request.