Comparison of Disk Diffusion, E-Test, and Broth Microdilution Methods for Testing In Vitro Activity of Cefiderocol in Acinetobacter baumannii

The reference method for cefiderocol antimicrobial susceptibility testing is broth microdilution (BMD) with iron-depleted-Mueller-Hinton (ID-MH) medium, whereas breakpoints recommended for disk diffusion (DD) are based on MH-agar plates. We aimed to compare the performance of the commercial BMD tests ComASP (Liofilchem) and UMIC (Bruker), and DD and E-test using MH- and ID-MH-agar plates with the reference BMD method using 100 carbapenem-resistant-A. baumannii isolates. Standard BMD was performed according to the EUCAST guidelines; DD and E-test were carried out using two commercial MH-agar plates (BioMérieux and Liofilchem) and an in-house ID-MH-agar plate, while ComASP and UMIC were performed according to the manufacturer’s guidelines. DD performed with the ID-MH-agar plates led to a higher categorical agreement (CA, 95.1%) with standard BMD and fewer categorization errors compared to the commercial MH-agar plates (CA BioMérieux 91.1%, Liofilchem 89.2%). E-test on ID-MH-agar plates exhibited a significantly higher essential agreement (EA, 75%) with standard BMD compared to the two MH-agar plates (EA BioMérieux 57%, Liofilchem 44%), and showed a higher performance in detecting high-level resistance than ComASP and UMIC (mean log2 difference with standard BMD for resistant isolates of 0.5, 2.83, and 2.08, respectively). In conclusion, DD and E-test on ID-MH-agar plates exhibit a higher diagnostic performance than on MH-agar plates and the commercial BMD methods. Therefore, we recommend using ID-MH-agar plates for cefiderocol susceptibility testing of A. baumannii.


Introduction
Cefiderocol is often the last active agent in gram-negative bacteria before pan-resistance ensues, in particular for carbapenem-resistant Acinetobacter baumannii (CRAB) [1], which represents a major cause of hospital-acquired infections worldwide [2]. Cefiderocol exploits its siderophore moiety to gain access into bacterial cells through active iron transporters. However, this uniqueness also poses a great challenge for antibiotic susceptibility testing (AST), which is reflected by differences in the interpretative criteria established by the EUCAST, CLSI [3,4], and FDA [5]. In fact, accurate in vitro AST requires iron-depleted conditions (ID-MH) to induce siderophore-mediated entry. Despite the complexity of the method and the difficulties that sometimes occur with reading the minimal inhibitory concentrations (MICs) due to the emergence of trailing points, there is a consensus that broth microdilution (BMD) represents the gold standard for MIC testing. Recently, two commercial kits (Compact Antimicrobial Susceptibility Panel, ComASP, Liofilchem, Italy and UMIC, Bruker, Germany) to determine cefiderocol MICs through BMD have been (ciprofloxacin and levofloxacin, see Figure S1). Antibiotic disks were from Oxoid Limited (Basingstoke, United Kingdom). Sirweb/Sirscan system (i2a) measured the inhibition zone diameters, which were visually controlled [20]. The gradient strip test (E-test) was used to determine MICs of ceftazidime-avibactam, ceftolozane-tazobactam, ampicillin-sulbactam, tigecycline, and eravacycline ( Figure S2). All antibiotic gradient strips were from Liofilchem. Colistin MICs were determined by BMD using the UMIC Colistin kit (Biocentric, Bandol, France). MIC values were rounded and adjusted to a binary log scale (i.e., 0.002, (. . .), 128, 256). Pseudomonas aeruginosa ATCC 27853 and A. baumannii NCTC13304 were used as internal quality control strains to assess the accuracy and reproducibility of different AST methods ( Figure S3).

MIC Test Strip for Cefiderocol AST
The MIC test strip method was performed on regular CAMH-agar plates from BioMérieux (Marcy L'Etoile, France). Cefiderocol MIC test strip was also performed on CAMH-agar plates from Liofilchem and on in-house produced ID-CAMH-agar plates. For synergy testing of ceftazidime-avibactam with cefiderocol, the ceftazidime-avibactam strip was first applied on an in-house produced ID-CAMH-agar plate previously inoculated with A. baumannii (McFarland 0.5) using a swab. After 1 h incubation at room temperature, the ceftazidime-avibactam strip was carefully removed and the cefiderocol strip was placed on the same location. Cefiderocol MICs were read after 18 h incubation at 35 • C.

Data Analysis
Essential agreement (EA) was defined as MIC ± one twofold dilution of the reference MIC (determined with the reference BMD method). Categorical agreement (CA) and clinical errors (major error, ME; very major error, vME) were determined according to the 'CLSI Methods Development and Standardization Working Group Best Practices for Evaluation of Antimicrobial Susceptibility Tests (2018)' on the basis of the CLSI and EUCAST breakpoints [3,4]. Expected congruent performances were: EA/CA ≥ 90%, ME ≤ 5%, vME ≤ 1.5% [24].

Whole Genome Sequencing and Typing
Whole genome sequencing determined the presence of beta-lactamase genes. Bacterial genomic DNA was extracted using the DNeasy ® Ultraclean ® Microbial kit (Qiagen, Hilden, Germany). Library preparation was performed with the QIAGEN QIASeq FX kit (Qiagen, Hilden, Germany). Library quality and fragment size distribution were analyzed on an automated CE system (Advanced Analytical Technologies Inc., Heidelberg, Germany). Paired-end sequencing (2 × 150 bp) of DNA libraries was performed using an Illumina MiSeq platform (Illumina ® , San Diego, CA, USA). Trimmomatic (version 0.39) was used to filter and trim raw sequencing data [25]. Reads were assembled using Unicycler v0.4.8 [26]. All genome assemblies were typed in Ridom Seqsphere+ v8.5.1 by multilocus sequence typing (MLST) according to the Pasteur (sequence type, ST) scheme and by core genome multi-locus sequence typing (MLST) [27].
The A. baumannii isolates belonged to 34 different sequence types (STs), one of which was novel and four were not determined ( Figure S4). ST2 was the most prevalent ST (49/100, 49%). The phylogeny falls into two major clusters, one dominated by ST2. This phylogenetic organization has previously been described [30]. The dominance of ST2 and, in particular, one sublineage of ST2, reflects what is received by our laboratory in Zürich, Switzerland. No association between cefiderocol resistance and phylogenetic clusters was detected, as resistant isolates are often adjacent to sensitive isolates in the phylogeny, suggesting independent emergence of cefiderocol resistance through gene acquisition and de novo mutations ( Figure S4).
(49/100, 49%). The phylogeny falls into two major clusters, one dominated by ST2. This phylogenetic organization has previously been described [30]. The dominance of ST2 and, in particular, one sublineage of ST2, reflects what is received by our laboratory in Zürich, Switzerland. No association between cefiderocol resistance and phylogenetic clusters was detected, as resistant isolates are often adjacent to sensitive isolates in the phylogeny, suggesting independent emergence of cefiderocol resistance through gene acquisition and de novo mutations ( Figure S4). The P. aeruginosa ATCC27853 quality control (QC) strain was tested throughout the experiments (eight times) and exhibited MIC values within the EUCAST range (0.06-0.5 µg/mL, MIC mean 0.22 ± 0.1 µg/mL), while the A. baumannii NCTC13304 QC strain, for which there are neither EUCAST nor CLSI DD QC MIC range values, exhibited a MIC mean of 0.62 ± 0.2 µg/mL ( Figure S3).
Based on the EUCAST and CLSI CBPs for A. baumannii, and when not available for P. aeruginosa, nearly all the isolates were resistant towards piperacillin-tazobactam, cephalosporins (ceftazidime and cefepime), carbapenems (imipenem and meropenem), and quinolones (ciprofloxacin and levofloxacin, see Table S3 and Figure S1). Also, the great majority displayed resistance against all classic aminoglycosides (amikacin, gentamicin, and tobramycin). As expected, all isolates showed high MICs of ceftazidime-avibactam, ceftolozane-tazobactam, and ampicillin-sulbactam (for all MIC90 > 256 µg/mL), irrespective of their β-lactamase content (Table S2, Figure S2). For tigecycline and eravacycline, there are no EUCAST nor CLSI CBPs on A. baumannii or P. aeruginosa (EUCAST has so far published a CBP for E. coli, which is 0.5 µg/mL for both tigecycline and eravacycline). The The P. aeruginosa ATCC27853 quality control (QC) strain was tested throughout the experiments (eight times) and exhibited MIC values within the EUCAST range (0.06-0.5 µg/mL, MIC mean 0.22 ± 0.1 µg/mL), while the A. baumannii NCTC13304 QC strain, for which there are neither EUCAST nor CLSI DD QC MIC range values, exhibited a MIC mean of 0.62 ± 0.2 µg/mL ( Figure S3).

Performances of Disk Diffusion to Assess Cefiderocol Susceptibility
Disk diffusion was performed on two commercially available CAMH-agar plates (BioMérieux and Liofilchem) and on an in-house produced ID-CAMH-agar plate. Results were compared with MICs determined with the reference BMD method (Figure 2). Based on CLSI guidelines, whereby cefiderocol susceptibility of A. baumannii should be reported only when the inhibition zone is bigger than 14 mm, with both commercial CAMH-plates more than 90% (BioMérieux 90/100, Liofilchem 93/100) of the isolates were classified as susceptible (inhibition zone > 14 mm). While using the in-house produced ID-CAMH-agar plate, only 81% (81/100, Table 1) were susceptible. However, considering only the interpretable results, the CA with the reference BMD method was higher with the ID-CAMHagar plate (77/81, 95.1%) than with the commercial CAMH-plates (BioMérieux 82/90, 91.1%; Liofilchem 83/93, 89.2%). Furthermore, the ID-CAMH-agar plate caused significantly less categorization errors (3/81 mE, 1/81 vME) as compared to CAMH-agar plates (BioMérieux 4/90 mE, 4/90 vME; Liofilchem 4/93 mE, 6/93 vME). Based on the EUCAST PK-PD breakpoint (S ≥ 17 mm, R< 17 mm), the CA was 87%, 84%, and 86% with the Bio-Mérieux-, Liofilchem-and ID-CAMH-agar plates, respectively. Again, the ID-CAMH plates generated less vME (7/100) compared to the plain CAMH-agar plates (BioMérieux 11/100; Liofilchem 16/100).   The P. aeruginosa ATCC27853 QC strain exhibited DD values within the EUCAST range (23-29 mm) when using the CAMH-agar plates (both showing a mean growth inhibition zone of 28 ± 1 mm). While the mean inhibition zone on the homemade ID-CAMH-agar plates was slightly bigger (29.8 ± 1.5 mm) and was in three cases above the higher range value ( Figure S3). The growth inhibition zones of the A. baumannii NCTC13304 QC strain were stable throughout the experiments. The inhibition zone varied in size based on the media used: with the Liofilchem-CAMH-agar plate of 25 ± 0.7 mm, with the BioMérieux-CAMH-agar plate of 23.6 ± 0.9 mm, and with the ID-CAMH-agar plate of 21 mm.
The P. aeruginosa ATCC27853 QC strain exhibited comparable MIC values when using the three different CAMH-agar plates, and all the growth inhibition values were within the QC range ( Figure S3). The MICs of the A. baumannii NCTC13304 QC strain were stable throughout the experiments and were identical when using the commercial CAMH-agar plates (mean 0.12 µg/mL), while being 1-2 log 2 higher when using the ID-CAMH-agar plate (mean 0.41 µg/mL).

Performances of ComASP to Assess Cefiderocol Susceptibility
ComASP showed 76% EA with the standard BMD method (Table 1, Figures 3 and S5). According to CLSI and EUCAST guidelines, ComASP exhibited 86% and 88% CA with the reference method and produced vME with 6/100 and 7/100 isolates, respectively.

Performances of UMIC to Assess Cefiderocol Susceptibility
As for ComASP, the UMIC test showed 76% EA with the standard BMD method ( Table 1, Figures 3 and S5). According to CLSI and EUCAST guidelines, UMIC exhibited 86% and 89% CA with the reference method, and produced vME with 3/100 and 9/100 isolates, respectively.
MIC values of the P. aeruginosa ATCC27853 QC strain were all within the EUCAST QC range and the MIC mean (0.23 ± 0.2 µg/mL) was comparable to that of the standard BMD method (0.22 ± 0.1 µg/mL). Instead, the MICs of the A. baumannii NCTC13304 QC strain as determined with UMIC (MIC mean 0.31 ± 0.1 µg/mL) were on average one log 2 lower than those assessed with the BMD reference method (MIC mean 0.62 ± 0.2 µg/mL).

Overall Performances of the Various Methods to Assess Cefiderocol Susceptibility
Overall, based on the CLSI CBPs and considering the CA, DD performed better than the MIC-based methods (DD CA = 89.2-95.1%, E-test, ComASP and UMIC CA = 85-87%) and DD with ID-MH-plates exhibited the highest congruence with the BMD method (CA = 95.1%, see Table 1). Importantly, using ID-MH-plates led to significantly less vME both by DD (1.2%) and by E-test (2%) as compared with the other commercially available MH-plates (≥4.4%), mostly due to detection of cefiderocol highly resistant isolates. UMIC also produced few vME (3%) as compared to the other MIC-based methods (≥6%). Among the MIC-based methods, the E-test with ID-MH-plates, ComASP and UMIC ex-hibited the highest EA with the standard BMD method (75-76%). However, ComASP produced significantly more vMEs (6% vs. 2-3%) as a result of the failure to detect highly resistant isolates.

Synergy between Cefiderocol and Avibactam
We found that addition of avibactam decreased the cefiderocol MICs by three-or morefold dilutions (synergistic activity) and restored in vitro susceptibility in 3/5 intermediate and all 9 resistant A. baumannii strains non-producing MBL-carbapenemases (i.e., of type NDM) and exhibiting cefiderocol MICs ≥ 8 mg/L (Table 2). Interestingly, in one cefiderocol intermediate (isolate 30, OXA-58-producer) and one resistant (isolate 92, OXA-23/-72producer) A. baumannii strain, the addition of avibactam did not affect cefiderocol MICs as determined by standard BMD. Synergy tests using the MIC gradient strip method exhibited concordant data with the BMD method in all but one A. baumannii strains (see an explanatory example on Figure 4). The one discordant A. baumannii isolate (isolate 73, OXA-23-producer) tested cefiderocol resistant with the BMD method (MIC = 16 µg/mL) but resulted susceptible with the MIC gradient strip method (MIC = 0.75 µg/mL). Also, cefiderocol susceptibility was not affected by avibactam. The two A. baumannii strains for which with the standard BMD method avibactam did not show synergistic activity with cefiderocol (isolates 30 and 92), neither a synergistic effect nor restoration of cefiderocol susceptibility was observed with the MIC gradient strip test. Finally, growth inhibitory effects (halos) between cefiderocol and avibactam disks (either ceftazidime/avibactam 10/4 µg and/or ceftazidime/avibactam 40/10 µg, see an explanatory example on Figure 4) were detected by DD with 8/10 A. baumannii isolates showing synergistic activity with the BMD method, while it was not detected in the remaining two A. baumannii strains (isolates 57 and 90).

Discussion
Using a large collection of CRAB with a wide range of cefiderocol susceptibilities, we showed that DD and E-test performed on CAMH-agar plates exhibit a poor correlation with the standard BMD. Importantly, both methods tend to underestimate MICs, especially with highly resistant strains. Likewise, we found that the recently commercialized ComASP, and to a lesser extent, UMIC, also failed to detect high-level resistance, mostly because of underestimation of high MICs, even though UMIC exhibited a higher congruence with the standard BMD method. The congruence with the standard BMD values significantly increased when both DD and E-test were performed with the same medium, namely ID-CAMHB. Like for the exemplary isolate depicted on Figure S6, tiny yet visible colonies within the growth inhibition zones of resistant isolates, which may result from the emergence of hetero-resistant subpopulations, appeared more consistently on ID-CAMH agar plates. This improved the correlation with the MIC values obtained with the standard BMD method. Consistent with these findings, the ID-CAMH-agar plates exhibited a significantly lower mean log 2 difference of MICs with the standard BMD of the intermediate and resistant populations as compared to the other methods ( Figure S5). To our knowledge, this is the first study evaluating the performance of DD and E-test for cefiderocol and A. baumannii using the ID-CAMH-agar plates.
To improve the reliability of the cefiderocol ASTs, two tests instead of one may be performed and interpreted. For example, considering the DD and E-test values obtained on ID-CAMH agar plates (interpreted according to the CLSI CBPs and based on the rule that by discrepant categorization between the methods resistant overtake susceptible results), CA with the BMD method was observed in in 88/100 of the cases, mE in 9/100, ME in 1/100 and vME in 2/100 ( Figure S7). Applying the same rules and considering the DD values obtained on ID-CAMH agar plates and the UMIC MIC values, CA with the BMD method was observed in in 89/100 of the cases, mE in 8/100, and vME in 3/100 cases. Finally, considering the DD values obtained on ID-CAMH agar plates and the ComASP MIC values, CA with the BMD method was observed in 88/100 of the cases, and mE and vME both in 6/100 cases.
Repeated testing showed that the medium had no impact on the cefiderocol E-test values and had only a small impact on the DD values of the P. aeruginosa ATCC27853 QC strain. Conversely, the A. baumannii NCTC13304 QC strain exhibited on average significantly higher MICs and smaller inhibition zones when using the ID-CAMH-agar plates, suggesting a bigger impact of the medium on the cefiderocol AST of susceptible A. baumannii strains. Moreover, the ID-CAMH-agar plates improves the detection of putative resistant subpopulations. A set of two A. baumannii QC strains, one susceptible and one resistant to cefiderocol, may also be considered for internal QC to ensure the quality of the CAMH-agar-plates.
Previous studies have shown that PER-like β-lactamases, and to a lesser degree, NDM β-lactamases, are associated with elevated MICs in A. baumannii, although production of these enzymes alone does not lead to MICs above the EUCAST PK-PD breakpoint (≤2 µg/mL) [9]. Likewise, in our study all PER-producing strains exhibited high MIC values (≥32 µg/mL), while NDM-producing isolates showed either reduced susceptibility or resistance with MIC values closely above to the CBPs. Cefiderocol reduced susceptibility and/or resistance was not associated with specific markers. Thus, cefiderocol susceptibility cannot be inferred by the presence specific acquired resistance mechanisms and should always be determined in vitro. The only exception to this rule is for PERtype β-lactamases, whose prompt detection may help guide decision making in therapy against MDR A. baumannii infections. In this regard, we showed that the addition of avibactam restored the susceptibility in all but one cefiderocol resistant A. baumannii isolate producing OXA-type and/or PER-type β-lactamases, as also reported in previous studies [8][9][10]14,17,18]. We also found that, using ID-CAMH-agar plates, the synergistic activity of avibactam with cefiderocol can be tested quantitatively and qualitatively by E-test and DD, respectively.
In conclusion, we showed that DD and E-test on ID-CAMH-agar plates can produce more consistent results with the standard BMD method than on CAMH-agar plates, which are currently recommended by EUCAST and CLSI. Synergy between cefiderocol and avibactam can also be detected both by E-test and DD on ID-CAMH-agar plates. Based on the findings of this study, ID-CAMH-agar plates may be considered for in vitro susceptibility testing of cefiderocol and A. baumannii.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antibiotics12071212/s1, Table S1. Features of the isolates used in this study; Table S2. MIC-based susceptibility rates of the A. bau-mannii isolates towards secondline and last-resort antibiotics; Table S3. DD-based susceptibility rates of the A. baumannii isolates towards first-line antibiotics; Figure S1. Distribution of growth inhibition zones of first-line antibiotics; Figure S2. Distribution of MICs of second-line and last-resort antibiotics; Figure S3. Quality control; Figure S4. Phylogenetic neighbor joining tree generated in Ridom Seqspere+ based on core genes with associated metadata; Figure S5. Performance of E-test on different MH-agars and ComASP and UMIC BMD assays; Figure S6