Multicenter Evaluation of an MIC-Based Aztreonam and Ceftazidime-Avibactam Broth Disk Elution Test

ABSTRACT

Due to limited therapeutic options, there is a clinical need to assess the in vitro activity of the combination of aztreonam (ATM) and ceftazidime-avibactam (CZA) to guide the therapeutic management of multidrug-resistant (MDR) Gram-negative organism infections. We set out to develop a practical MIC-based broth disk elution (BDE) method to determine the in vitro activity of the combination ATM-CZA using readily available supplies and compare it to reference broth microdilution (BMD). For the BDE method, a 30-μg ATM disk, a 30/20-μg CZA disk, both disks in combination, and no disks were added to 4 separate 5-mL cation-adjusted Mueller-Hinton broth (CA-MHB) tubes, using various manufacturers. Three testing sites performed both BDE and reference BMD testing of bacterial isolates in parallel from a single 0.5 McFarland standard inoculum and after overnight incubation, assessed them for growth (not susceptible) or no growth (susceptible) at a final concentration of 6/6/4 μg/mL ATM-CZA. During the first phase, the precision and accuracy of the BDE were analyzed by testing 61 Enterobacterales isolates at all sites. This testing yielded 98.3% precision between sites, with 98.3% categorical agreement and 1.8% major errors (ME). During the second phase, at each site, we evaluated unique, clinical isolates of metallo-β-lactamase (MBL)-producing Enterobacterales (n = 75), carbapenem-resistant Pseudomonas aeruginosa (n = 25), Stenotrophomonas maltophilia (n = 46), and Myroides sp. (n = 1). This testing resulted in 97.9% categorical agreement, with 2.4% ME. Different results were observed for different disk and CA-MHB manufacturers, requiring a supplemental ATM-CZA-not-susceptible quality control organism to ensure the accuracy of results. The BDE is a precise and effective methodology for determining susceptibility to the combination ATM-CZA.

INTRODUCTION

Antimicrobial resistance to β-lactams was first noted near the discovery of penicillin in 1940 (1) and has rapidly expanded alongside the introduction of new antimicrobial agents, generating a significant global health problem (2). Resistance to carbapenems, the broadest class of β-lactams, is caused broadly by two categories of resistance mechanisms: carbapenemases, enzymes able to directly hydrolyze the carbapenems, and noncarbapenemase mechanisms like AmpC or extended-spectrum β-lactamases in conjunction with cell wall permeability defects (3, 4). Metallo-β-lactamases (MBLs), a type of carbapenemase requiring zinc for activity, have been the most common of the carbapenemase family in Eastern Europe, the Indian subcontinent, and parts of Southeast Asia, and their prevalence worldwide is increasing (3, 5, 6). MBLs are adept at hydrolyzing almost all traditional β-lactams and are not inhibited by FDA-approved β-lactamase inhibitor (BL-BLI) combination regimens, such as ceftazidime-avibactam (CZA), imipenem-relebactam, and meropenem-vaborbactam. However, MBLs are not able to hydrolyze aztreonam (ATM). Many MBL-producing isolates coexpress one or more other β-lactamases with activity against ATM but that are inhibited by avibactam (AVI). With a dwindling list of available treatment options for MBL-producing Gram-negative organisms, aztreonam in combination with avibactam is a new therapeutic approach to target these bacterial infections. However, this combination is not yet commercially available. Combination treatment with aztreonam and ceftazidime-avibactam (ATM-CZA) has shown success in vitro and in treatment against multidrug-resistant and extensively drug-resistant Enterobacterales isolates (7, 8). ATM-CZA has been recommended by the Infectious Diseases Society of America (IDSA) as an alternative treatment for select multidrug-resistant Gram-negative infections, like MBL-producing carbapenem-resistant Enterobacterales (CRE) and Stenotrophomonas maltophilia, with limited therapeutic options (9). Combination aztreonam-avibactam (ATM-AVI) is in clinical trials (10–13), but it will be several years before this combination is available clinically. Although cefiderocol demonstrates in vitro activity, there are concerns regarding the reliability of susceptibility testing methods, clinical efficacy, and development of resistance during therapy (14–18). Clinicians need an immediate solution to target MBL infections, and utilizing ATM and CZA together can provide an additional option for treatment in these cases.

To administer ATM and CZA, clinicians will need results from a reliable methodology to determine the in vitro susceptibility of isolates to the combination to guide treatment. In the treatment of MBLs, ATM is the active agent, with AVI having no independent antimicrobial activity and ceftazidime not contributing significantly to the combination for Enterobacterales infections (19). As shown in Table 1, the proposed breakpoints for ATM-AVI (and by extension ATM-CZA) are ≤4/4 μg/mL for Enterobacterales (20) and ≤8/4 μg/mL for Pseudomonas aeruginosa and S. maltophilia (21–24), consistent with the Clinical and Laboratory Standards Institute (CLSI) ATM breakpoints for Enterobacterales and P. aeruginosa, respectively. Currently, there is no validated, practical, and readily available antimicrobial susceptibility testing (AST) method for assessing the activity of the combination in clinical microbiology laboratories. Combination disk and gradient diffusion methods are subjective and poorly standardized, with results varying between the methods used, the individual readers, and the reagent manufacturers (25). Other options like checkerboard, time-kill, and reference broth microdilution (BMD) assays are not practical in a clinical setting, as they are too resource intensive to be utilized in a patient care setting. Previous research investigated the use of a broth disk elution (BDE) synergy test for ATM and CZA against 10 CRE and difficult-to-treat resistant P. aeruginosa isolates (25). However, the assay was not MIC based and only focused on the detection of synergy, making it difficult to translate the results to the patient care setting. The Antibiotic Resistance Laboratory Network (ARLN) of the Centers for Disease Control and Prevention (CDC) does currently provide expanded AST for MBL-producing CRE isolates (26, 27), which can give clinicians MIC data for use with ATM-CZA combination treatment. However, this requires samples to be sent out for testing, which has a minimum of a 3-day turnaround time to results and may lead to a delay in treatment administration that could be critical for patient care. Based on the clinical need and the limitations of the current methods, we set out to develop a practical MIC-based BDE method to determine the in vitro activity of the combination ATM-CZA using supplies readily available in clinical microbiology laboratories.

TABLE 1.

Clinical breakpoints for aztreonam, ceftazidime-avibactam, and aztreonam-avibactam^a

Organism	Breakpoint (μg/mL) for:
	Aztreonam			Ceftazidime-avibactam		Aztreonam-avibactam
	S	I	R	S	R	S
Enterobacterales	≤4	8	≥16	≤8/4	≥16/4	≤4/4b
Pseudomonas aeruginosa	≤8	16	≥32	≤8/4	≥16/4	≤8/4c
Stenotrophomonas maltophilia						≤8/4c

^aCLSI breakpoints (21) were used for aztreonam and ceftazidime-avibactam. The proposed breakpoints are shown for aztreonam-avibactam. S, susceptible; I, intermediate; R, resistant.

^bReference 20.

^cThe cutoff of ≤8 μg/mL indicates the aztreonam-susceptible breakpoint published by CLSI for P. aeruginosa and the tentative aztreonam-avibactam PK/PD susceptible breakpoint (20, 22).

MATERIALS AND METHODS

Study design.

Testing was performed across three laboratories, as part of a CLSI multicenter study: Johns Hopkins University School of Medicine (JHU), Vanderbilt University Medical Center (VUMC), and University of Illinois Chicago (UIC). The study proceeded in two phases: in phase 1, the same collection of isolates was evaluated across all three sites; in phase 2, each site selected a unique collection of challenge isolates, as detailed below. Each testing site performed both BDE testing and reference BMD from a single 0.5 McFarland standard sample of bacterial isolates. Each testing method is described in further detail below.

Bacterial isolates.

In phase 1, the testing sites used 61 Enterobacterales isolates from the CDC and FDA Antibiotic Resistance (AR) Isolate Bank (26). Of these isolates, 57 were susceptible (≤6/6/4 μg/mL), and 4 were not susceptible (NS; intermediate and resistant, >6/6/4 μg/mL) to ATM-CZA. In phase 2, each site tested additional clinical isolates of MBL-producing Enterobacterales (total tested across sites, 75), carbapenem-resistant Pseudomonas aeruginosa (n = 25), S. maltophilia (n = 46), and Myroides sp. (n = 1). The species breakdown of the isolates used in phases 1 and 2 are shown in Table 2 and in Tables S1 and S2 in the supplemental material. Additionally, Escherichia coli ATCC 25922 (susceptible to all agents), Klebsiella pneumoniae BAA-1705 (not susceptible to ATM), K. pneumoniae BAA-2146 (New Delhi metallo-β-lactamase [NDM] producer; not susceptible to ATM or CZA individually), and E. coli AR348 (not susceptible to ATM, CZA, or ATM-CZA) were included as quality control (QC) strains to assess the expected growth/no-growth results in the BDE tubes.

TABLE 2.

Isolates included in each phase of the study

Organism	Phase 1	Phase 2	No. of isolates not susceptible to ATM-CZAa
Citrobacter amalonaticus	0	1	0
Citrobacter freundii	2	1	0
Citrobacter koseri	1	0	0
Enterobacter aerogenes	5	0	0
Enterobacter cloacae complex	7	8	1
Escherichia coli	16b	23	4
Klebsiella oxytoca	2	1	0
Klebsiella ozaenae	1	0	0
Klebsiella pneumoniae	14b	35	1
Myroides sp.	0	1	1
Morganella morganii	1	0	0
Proteus mirabilis	1	2	0
Providencia rettgeri	1	2	1
Providencia sp.	1	1	0
Pseudomonas aeruginosa	0	25	14
Salmonella spp.	4	0	0
Serratia spp.	4	1	0
Shigella sonnei	1	0	0
Stenotrophomonas maltophilia	0	46	2
Total	61	147	24

^aNot susceptible (NS; intermediate and resistant, >6/6/4 μg/mL) to ATM-CZA. ATM, aztreonam; CZA, ceftazidime-avibactam.

^bOne E. coli isolate was excluded due to missing BMD data, and one K. pneumoniae isolate was excluded due to no reference result as the comparator (i.e., disagreement between test sites and repeats).

Broth disk elution method.

For the BDE method, a 30-μg ATM disk (JHU/VUMC: Becton, Dickinson [Sparks, MD]; UIC: Oxoid Ltd., Thermo Fisher Scientific [Waltham, MA]), a 30/20-μg CZA disk (JHU/UIC: Hardy Diagnostics [Santa Maria, CA]; VUMC: Becton, Dickinson), both disks in combination, and no disks (growth control [GC]) were added to 4 separate 5-mL cation-adjusted Mueller-Hinton broth (CA-MHB) tubes (Becton, Dickinson prepared tubes, Hardy Diagnostics prepared tubes, Thermo Fisher Scientific prepared tubes, and Oxoid base prepared in-house [using Thermo Fisher Scientific tubes]) (Fig. 1). The final concentrations in the tubes were 6 μg/mL ATM and 6/4 μg/mL CZA, individually or in combination. The disk content/CA-MHB ratio was purposefully chosen to result in a final ATM-CZA concentration of 6/6/4 μg/mL in tube 4 to provide an ATM concentration around the CLSI susceptible breakpoint for both the Enterobacterales (4 μg/mL) and P. aeruginosa (8 μg/mL) strains and to match the fixed AVI concentration of 4 μg/mL used in BMD. Routine weekly QC of the disks was performed at each site prior to setting up the BDE method. The methodology and inocula applied were similar to those of the colistin BDE method (28). Briefly, the disks were added to each tube accordingly and incubated for 30 to 60 min at room temperature to allow diffusion of the antimicrobials from the disk(s). A 0.5 McFarland standard was prepared from a second serial overnight subculture from freezer stock on blood agar plates of the bacterial isolate in 5-mL saline tubes, and a 10-μL inoculating loop was used to streak the cultures onto purity plates. Aliquots (25 μL) of the McFarland culture were added to each tube (final bacterial concentration in each tube, 7.5 × 10⁵ CFU/mL), vortexed vigorously to ensure even distribution, and incubated at 35 ± 2°C under ambient air for 16 to 20 h (Enterobacterales, P. aeruginosa, and Myroides sp. strains) or 20 to 24 h (S. maltophilia strains). After overnight incubation, the tubes were assessed for no growth (susceptible) or growth (not susceptible) at 6/6/4 μg/mL of ATM-CZA, based on the breakpoints outlined in Table 1.

FIG 1.

Aztreonam (ATM) and ceftazidime-avibactam (CZA) BDE results for (left) a Pseudomonas aeruginosa strain that was not susceptible to ATM and CZA individually or in combination (growth) and (right) a Klebsiella pneumoniae strain that was not susceptible to ATM and CZA individually but susceptible to the combination ATM-CZA (no growth).

Reference broth microdilution method.

Reference BMD was performed according to CLSI standard M07 (29). BMD panels were manufactured in-house at VUMC to include ATM (Sigma; 0.5 to 32 μg/mL), CZA (CZA from Sigma, AVI from MedChemExpress; 0.5/4 to 32/4 μg/mL), and ATM-CZA (0.5/8/4 to 32/8/4 μg/mL), prepared in CA-MHB (Becton, Dickinson), frozen, shipped on dry ice, and stored at −80°C at each site. In addition, a single well containing 6 μg/mL ATM, 6/4 μg/mL CZA, and 6/6/4 μg/mL ATM-CZA was included to mimic the concentrations in the BDE tubes. These single-well results were used as the primary comparator to the BDE method. The same 0.5 McFarland standard sample (1 mL) as used for the BDE was further diluted into a 29-mL tube of demineralized water. A 95-pin inoculator was used to transfer 10 μL/well of diluted inoculum to a thawed BMD panel (100 μL/well) for a final concentration of 3 × 10⁴ CFU/well of bacteria. The BMD plates were sealed, incubated at 35 ± 2°C under ambient air for 16 to 20 h (Enterobacterales, P. aeruginosa, and Myroides sp. strains) or 20 to 24 h (S. maltophilia strain) and assessed for growth following the M07 standard (29).

Manufacturer and lot-to-lot comparison study.

To determine if there were any differences in BDE results dependent on the manufacturer, various manufacturers of disks and broths were evaluated in multiple permutations. The disks and broths tested were sourced as follows: CA-MHB from Hardy Diagnostics and Becton, Dickinson and Oxoid dehydrated culture medium that was cation adjusted (Thermo Fisher Scientific); ATM disks from Becton, Dickinson and Oxoid Ltd.; and CZA disks from Hardy Diagnostics and Becton, Dickinson. Bacterial strains used for the comparison study are listed in Table S3. BDE was performed as described above. When available, lot-to-lot comparisons of the various disks and MHB broths were conducted within and between testing sites.

Analysis of results.

As a comparator value, the references MICs of each strain were determined for each combination of ATM, CZA, and ATM-CZA (referred to below as the BMD MICs), including the result of a single well in the BMD panel containing 6 μg/mL ATM, 6/4 μg/mL CZA, and 6/6/4 μg/mL ATM-CZA to replicate the concentration in the BDE tubes (referred to below as the BMD reference well). The results from the single BMD reference well were used as the primary comparator to the BDE method and were interpreted as susceptible (no growth) or not susceptible (growth). The BMD MICs for the organism of each agent/combination were evaluated if a discordant result occurred, to assess whether the MIC bracketed the concentration in the BDE tube. The BMD results (both MIC and reference well) from VUMC and UIC for phase 1 were used due to a limited number of panels available at JHU. Thus, the JHU BMD panels were reserved for the clinical isolate testing in phase 2. If the two BMD results from phase 1 were not in categorical agreement, all methods of testing were repeated. If the discordance did not resolve on repeat, the strain was excluded due to the inability to determine a reference result as the comparator. For phase 2, the BMD results at each testing site were used as the comparator. Categorical agreement (CA) was determined if the BDE result corresponded to the BMD reference well interpretation (e.g., no growth in the 6/6/4 μg/mL BMD reference well, and the ATM-CZA BDE tube interpreted both as susceptible). A major error was recorded if there was growth in the BDE tube but not in the BMD reference well, whereas a very major error was recorded if there was no growth in the BDE tube but growth in the BMD reference well. If on repeat the results confirmed the initial results, this was deemed to be a confirmed CA error (a major error or very major error, accordingly). If on repeat the results did not confirm the initial results, the initial results were removed from the final analysis. The precision of the BDE methodology was also determined in phase 1. Precision was determined by the percent categorical agreement of susceptible/not susceptible classification across the 3 testing sites.

RESULTS

Phase 1: AR Bank isolate testing.

In phase 1, 61 isolates from the CDC AR Bank were tested to compare the BDE method results to the BMD results (Table 2). Fifty-nine strains were included in the final analysis. Two strains were excluded: one E. coli strain due to missing BMD data and one K. pneumoniae strain due to no reference result as the comparator (i.e., disagreement between the test sites and repeats). Furthermore, data were missing for 2 isolates at one site, resulting in a total of 175 comparisons for phase 1.

We evaluated each BDE tube independently from the set compared to the BMD reference well appropriate for each agent. For ATM, 15 (25.4%) isolates were susceptible and 44 (74.6%) were not susceptible by BMD reference well. Compared to the BMD reference well, 97.1% categorical agreement (170/175) was observed for BDE, with 3 (7.0%) major errors and 2 (1.5%) very major errors. For CZA, 35 (59.3%) isolates were susceptible and 24 (40.7%) were not susceptible by BMD reference well. The BDE method achieved 97.1% categorical agreement (170/175), with 2 (1.9%) major errors and 3 (4.2%) very major errors, compared to the BMD reference well.

Fifty-seven of the isolates were susceptible and 2 were not susceptible to the combination ATM-CZA based on the BMD reference well results. The ATM-CZA reference BMD MICs ranged from <0.5/8/4 to 8/8/4 μg/mL. Our initial BDE testing yielded 9 major errors (ME) and 2 very major errors, of which 8 resolved upon repeat testing, including both very major errors. These 8 errors were due to manufacturer-related issues observed at a single site that resolved upon repeat (see “Manufacturer comparison study,” below). The remaining three major errors occurred with the same isolate across all three testing sites. Escherichia coli AR0137 had no growth in the BMD reference well at the 2 testing sites (6/6/4 μg/mL) and was interpreted as susceptible to ATM-CZA. However, all three testing sites’ BDE results determined it to be not susceptible (growth in the BDE tube; 6/6/4 μg/mL, whereas the reference MIC for the isolate was 8/8/4 μg/mL ATM-CZA). The final analysis after the repeats yielded 172 of 175 (98.3%) categorical agreements (CA), a major error rate of 1.8% (3/169) across the 3 centers, and no very major errors (Table 3). We also used this phase to analyze the precision between BDE and the BMD reference well, since identical isolates were tested at each site. After repeat testing, the precision was 98.3% (58/59), with the one disagreement occurring at all three sites with E. coli AR0137 (as discussed above).

TABLE 3.

Summary of BDE study results after repeat testing

Phase of study	Antimicrobial agent	No. of isolates		% of isolates (no. of isolates/total no. of isolates) with:
		Susceptible	Not susceptible	Categorical agreement	Major errors	Very major errors
Phase 1a	ATM	43	132	97.1 (170/175)	7.0 (3/43)	1.5 (2/132)
	CZA	103	72	97.1 (170/175)	1.9 (2/103)	4.2 (3/72)
	ATM-CZA	169	6	98.3 (172/175)	1.8 (3/169)	0
Phase 2b	ATM	11	136	100 (147/147)	0	0
	CZA	23	124	98.0 (144/147)	4.3 (1/23)	1.6 (2/124)
	ATM-CZA	125	22	97.9 (144/147)	2.4 (3/125)	0
Phases 1 and 2 combined	ATM	54	268	98.4 (317/322)	5.6 (3/54)	0.7 (2/268)
	CZA	126	196	97.5 (314/322)	2.4 (3/126)	2.6 (5/196)
	ATM-CZA	294	28	98.1 (316/322)	2.0 (6/294)	0

^aPhase 1 included testing 59 Enterobacterales isolates from the CDC AR Bank at 3 sites. Data were missing for two isolates at one site, resulting in a total of 175 comparisons.

^bPhase 2 included 147 clinical isolates of metallo-β-lactamase-producing Enterobacterales (n = 75), carbapenem-resistant Pseudomonas aeruginosa (n = 25), Stenotrophomonas maltophilia (n = 46), and Myroides sp. (n = 1).

Phase 2: clinical isolate testing.

In phase 2, 147 clinical isolates were tested. We evaluated each BDE tube independently from the set compared to the BMD reference well appropriate for each agent. For ATM, 11 (7.5%) isolates were susceptible and 136 (93.2%) were not susceptible by BMD reference well (or considered intrinsically resistant, for S. maltophilia). One hundred percent categorical agreement (147/147) was observed for ATM BDE compared to the BMD reference well. For CZA, 23 (15.6%) isolates were susceptible and 124 (84.4%) were not susceptible by BMD reference well (or considered intrinsically resistant, for S. maltophilia, despite it sometimes demonstrating in vitro activity to ceftazidime). CZA BDE compared to the BMD reference well achieved 98.0% categorical agreement (144/147) with 1 major error (4.3%) and 2 very major errors (1.6%).

The BMD reference well results revealed that 125 of the isolates were susceptible and 22 were not susceptible to ATM-CZA. Initial testing resulted in 93.2% (137/147) categorical agreement, with a major error rate of 7.2% (9/125) and a very major error (VME) rate of 4.5% (1/22) by the BDE method compared to the BMD reference well. The VME occurred in a S. maltophilia isolate at VUMC. Seven errors were later resolved as manufacturer related, as described below. Upon repeat, seven errors were resolved, including the one VME. The three major errors that did not resolve on repeat occurred in one E. coli and two P. aeruginosa isolates. All three errors occurred at a single site and had BMD MIC results of ≤0.5/8/4 μg/mL but nonsusceptible BDE results (growth in the tube). The final analysis of the BDE testing for phase 2 resulted in 97.9% (144/147) categorical agreement, with 2.4% (3/125) major errors and no very major errors across the 3 centers (Table 3).

Manufacturer comparison study.

Upon review of the initial results, it was noted prior to repeat testing that a single site was observing higher errors than the other 2 sites. Thus, to further analyze the discrepancies between the testing centers, predominantly with the QC strain E. coli AR348, which is not susceptible to ATM, CZA, or ATM-CZA, various permutations of manufacturers of the CA-MHB, ATM disks, and CZA disks were tested to look for discordant results being linked to specific manufacturers. The results of all experiments are presented in Table S3. Focusing on E. coli AR348, as the QC strain that was expected to be NS to all test conditions for BDE, the discordant results suggested manufacturer-related issues (Table 4). We discovered that the CZA disks manufactured by Becton, Dickinson produced no growth and thus a susceptible interpretation for AR348, whereas the Hardy Diagnostic CZA disks produced positive growth, leading to a nonsusceptible interpretation, unless used in combination with Hardy Diagnostic CA-MHB. This same pattern was seen in two other nonsusceptible E. coli strains: AR0434 and AR0450 (Table S2). Interestingly, purchased premade CA-MHB manufactured by Thermo Fisher Scientific produced positive growth and the expected nonsusceptible interpretation with all disk manufacturers for E. coli AR348. The manufacturer of the ATM disk did not alter the results of the test. For the ATM disks (BD, Oxoid), CZA disks (Hardy), and CA-MHB (Hardy, BD), no lot-to-lot variability was observed within sites and/or between sites.

TABLE 4.

Manufacturer comparison study results for Escherichia coli AR348^a

Broth manufacturer	Disk manufacturer for:		Result for:			Interpretation
	ATM	CZA	ATM	CZA	ATM-CZA
Hardy Diagnostics	O	H	+	−	−	S
	B	B	+	−	−	S
	O	B	+	−	−	S
	B	H	+	−	−	S
Becton-Dickinson	O	H	+	+	+	NS
	B	B	+	−	−	S
	O	B	+	−	−	S
	B	H	+	+	+	NS
Thermo Fisher Scientific	O	H	+	+	+	NS
	B	B	+	+	+	NS
	O	B	+	+	+	NS
	B	H	+	+	+	NS

^aH, Hardy Diagnostics; B, Becton-Dickinson; O, Oxoid Ltd.; +, growth; −, no growth; S, susceptible; NS, not susceptible (intermediate and resistant); ATM, aztreonam; CZA, ceftazidime-avibactam.

Final analysis.

After the discovery of a manufacturer-based effect on the results, any discordant results from the initial analysis were repeated with an adjustment to the manufacturer of the CZA disks (from BD to Hardy Diagnostics at one site) and CA-MHB (from made in-house to purchased premade Thermo Fisher Scientific at one site). Overall, the ATM-CZA categorical agreement was 98.1% (316/322), with 2.0% (6/294) major errors and no very major errors. The final results of the study are summarized in Table 3.

DISCUSSION

Due to limited therapeutic options, there is a clinical need to assess the in vitro activity of the combination of ATM and CZA to guide therapeutic management of MDR Gram-negative bacilli, especially metallo-β-lactamase (MBL) producers. Currently, there is no validated, practical and readily available antimicrobial susceptibility testing method for assesing the activity of the combination in clinical microbiology laboratories. The current CLSI multicenter study evaluated the ATM-CZA BDE method to provide an MIC-based assessment of ATM-CZA compared to reference broth microdilution for MDR Gram-negative bacilli. Overall, we found the BDE method to be a practical, reproducible, and accurate method of determining ATM-CZA susceptibility, with an overall categorical agreement of 98.1% and 2.0% major errors.

Other methods evaluating the combined in vitro activity of ATM-CZA have focused on labor-intensive methods such as reference BMD, checkerboard, or time-kill assays that are normally reserved for research settings or simpler, more subjective methods, such as disk and gradient diffusion proximity or stacking methods (30). Recently, a BDE synergy method was described for ATM-CZA, applying a similar concept to the colistin BDE method that was recently endorsed by CLSI (28, 31). The BDE method overcomes many limitations of other ATM-CZA synergy methods by providing an objective (i.e., growth versus no growth) and practical method applying supplies that are readily available in most clinical laboratories, including those in resource-limited settings. We set out to further develop the BDE to provide an MIC-based result that can be interpreted by applying the proposed clinical breakpoints, which can be translated to patient-facing clinicians as susceptible/not susceptible to guide patient care.

Currently, there are no established interpretive criteria by standards development organizations to interpret ATM-AVI or ATM-CZA MICs. However, other studies have defined a putative susceptible breakpoint for ATM-AVI (and ATM-CZA) at ≤4/4 μg/mL for Enterobacterales (20) and ≤8/4 μg/mL for P. aeruginosa and S. maltophilia (21–24), aligning with the aztreonam breakpoint for Enterobacterales and P. aeruginosa (see Table 1). The BDE method developed and evaluated in this study was established to bracket the breakpoints of ≤4/4 μg/mL and ≤8/4 μg/mL with a final concentration of 6/6/4 μg/mL to provide a susceptible (no growth) or not susceptible (growth) result. The most commonly utilized ATM-CZA dosing regimen of 2 g/2.5 g given every 8 h (q8h) has demonstrated rapid, sustained bactericidal activity in vitro, along with favorable clinical response rates and reduced mortality versus comparators in clinical studies against MBL-producing, carbapenem-resistant Gram-negative isolates (7, 32). Using pharmacokinetics (PK) data generated from 41 critically ill patients treated with ATM-CZA 2 g/2.5 g q8h for MBL-producing CRE, Monte Carlo simulation demonstrated a probability of ≥90% of target attainment (PTA) at ATM-AVI MICs of ≤8/4 mg/L and CrCl values of ≤90 mL/min (33). The inclusion of the concentration 6/6/4 μg/mL in the BMD wells and BDE tubes in the current study allowed for an evaluation of the performance at both purported ATM-AVI breakpoints (4/4 and 8/4 μg/mL), while remaining within the systemically achievable and adequate PK/pharmacodynamics (PD) exposure (≤8/4 μg/mL). This is advantageous as it eases the translation of the results herein to clinical use and/or breakpoint determination, compared to previous studies which have utilized ATM-AVI concentrations in the BDE method of >8/>4 μg/mL (25). Thus, the current BDE method described provides an MIC-based assessment of the combination to guide therapy. However, if an expanded range of doubling dilutions is required for a more precise MIC for management, then BMD or another validated MIC method would be required, as the BDE only provides assessment at a single dilution bracketing the susceptible/intermediate breakpoints.

Laboratories considering validating the ATM-CZA BDE method for patient care should devise test and report strategies for ATM-CZA as appropriate for their institution and in conjunction with their antimicrobial stewardship team. For example, Enterobacterales or P. aeruginosa isolates could be tested when an organism is identified as an MBL producer by molecular (e.g., bla_NDM on a molecular panel) or phenotypic (e.g., EDTA carbapenem inactivation method positive) approaches. For S. maltophilia, testing could be considered when an MDR strain is encountered (e.g., resistant to trimethoprim-sulfamethoxazole, levofloxacin, and minocycline). Alternatively, the BDE method can be set up by request when the combination is being considered for therapy. Validation of the ATM-CZA method could be performed by comparing results to the expected CDC FDA AR Bank isolates outlined in phase 1 of this study or to results obtained by reference BMD or another validated method (e.g., the ARLN Lab Network results [26, 27]).

Although we evaluated the BDE method by assessing ATM and CZA individually and in combination, the method can be further simplified for clinical use by only assessing the combination of agents in the single tube (i.e., an ATM-CZA tube only). However, including the set of tubes provides a nice visual effect demonstrating the activity of the combination of ATM-CZA (e.g., no growth when in combination) when individually, they may not be active (e.g., growth present in the individual ATM and CZA tubes), and this visualization could help with potentially identifying any unusual AST profiles/discrepant results (e.g., no growth in ATM and growth in both CZA and ATM-CZA tubes). Testing the four tubes also provides additional granularity to QC results.

Quality control strains are important to assess disk and CA-MHB performance during the initial validation of AST methods and for routine daily and/or weekly QC. In this study, we did identify manufacturer-related differences that influenced the accuracy of the results, highlighting the importance of including a supplemental ATM-CZA-nonsusceptible isolate control, such as E. coli AR348. We found that the CZA disks produced by different manufacturers yielded variable results depending on the CA-MHB broth that each disk was combined with. The ATM disks did not alter the results. The differences between manufacturers could have various sources: the quality of the CZA disks or differences in the CA-MHB ion concentrations and the rate and/or efficiency of drug elution from the disks, resulting in various final drug concentrations in the tubes. Previous studies have demonstrated the variability of zinc concentrations in CA-MHB produced by various manufacturers, demonstrating up to an 8-fold difference in the meropenem MIC for MBL-producing Enterobacterales isolates (34). Although zinc concentrations are a possibility for the manufacturer-related issues we observed, we observed variation in our results mostly with non-MBL-producing ATM-CZA-nonsusceptible isolates and found the manufacturer of CZA disk to be a more critical component. Due to this variability, it is critical that laboratories perform QC testing, particularly with an ATM-CZA-nonsusceptible control strain (E. coli AR348), with each new lot of reagents, new shipments, and any changes to reagent manufacturers to ensure the ability to detect ATM-CZA resistance, or there is a risk of very major errors.

The limitations of our study include the lack of a complete lot-to-lot comparison study. However, when available for comparison, no lot-to-lot differences in the disks of CA-MHB were observed. Furthermore, we were not able to elucidate the specific mechanism affecting the manufacturer-related differences in the study. However, we did identify a method for laboratories to appropriately QC reagents to ensure accurate results by including a supplemental ATM-CZA-nonsusceptible control. Last, repeats due to manufacturer-related issues at one of the sites were only conducted on discordant results.

Overall, our findings show that BDE is a precise and effective methodology for determining the susceptibility of the combination ATM-CZA for the treatment of multidrug-resistant Gram-negative bacillus infections. The results from this multicenter study are currently under consideration by CLSI as an acceptable method for in vitro assessment of the combination.