Multicenter Evaluation of Ceftazidime-Avibactam Susceptibility Testing of Enterobacterales and Pseudomonas aeruginosa on the Vitek 2 System

In this multisite study, Vitek 2 AST-Gram-Negative Ceftazidime/Avibactam test results for 1,073 isolates (866 Enterobacterales and 207 Pseudomonas aeruginosa) were compared to the Clinical and Laboratory Standards Institute (CLSI) broth microdilution (BMD) reference method. The results were analyzed for essential agreement (EA), category agreement (CA), major error rates, and very major error rates following FDA/ISO performance criteria using the FDA-recognized CLSI/EUCAST breakpoints (sensitive [S], ≤8/4 μg/ml; resistant [R], ≥16/4 μg/ml).

ABSTRACT

In this multisite study, Vitek 2 AST-Gram-Negative Ceftazidime/Avibactam test results for 1,073 isolates (866 Enterobacterales and 207 Pseudomonas aeruginosa) were compared to the Clinical and Laboratory Standards Institute (CLSI) broth microdilution (BMD) reference method. The results were analyzed for essential agreement (EA), category agreement (CA), major error rates, and very major error rates following FDA/ISO performance criteria using the FDA-recognized CLSI/EUCAST breakpoints (sensitive [S], ≤8/4 μg/ml; resistant [R], ≥16/4 μg/ml). The overall EA was 94.5% (1,014/1,073) and CA was 98.7% (1,059/1,073). No very major errors were reported. The major error rate was 1.4% (14/998). Out of 14 major errors, 9 were within EA. Based on the EA and lack of an intermediate category for ceftazidime-avibactam (CZA), the adjusted major error rate for FDA criteria was 0.5% (5/998). The performance for ISO criteria after error resolutions included EA of 94.5% (1,014/1,073), CA of 98.9% (1,061/1,073), major error of 1.2% (12/998), and no very major error. Vitek 2 met the ISO and FDA criteria of ≥95% reproducibility and ≥95% quality control (QC) results within acceptable ranges for QC organisms. Vitek 2 overall performance for Enterobacterales and P. aeruginosa met or exceeded the FDA and ISO performance criteria; thus, it is a reliable alternative to the BMD reference method for routine CZA susceptibility testing.

INTRODUCTION

Multidrug-resistant (MDR) Gram-negative bacterial infections, including carbapenem-resistant Enterobacterales (CRE), continue to pose a major threat to public health (1). CRE have been listed as an urgent threat in the 2019 CDC report, and MDR Pseudomonas aeruginosa has been listed as a serious threat (2). Global spread and high mortality associated with both organism groups emphasize the immediate need for effective antimicrobial therapies and rapid antimicrobial susceptibility testing methods to aid in treatment decisions. Ceftazidime-avibactam (CZA) is a combination of the third-generation cephalosporin ceftazidime and the novel, non-β-lactam β-lactamase inhibitor avibactam (3). The addition of avibactam improves in vitro activity of ceftazidime against Enterobacterales and Pseudomonas aeruginosa (4). CZA received FDA approval for treatment of complicated intra-abdominal infections, complicated urinary tract infections, and hospital-acquired and ventilator-associated bacterial pneumonia caused by Gram-negative bacteria (5). Recent studies have revealed that CZA also exhibits a good outcome in the treatment of patients infected with Klebsiella pneumoniae carbapenemase (KPC)-producing strains (6). Susceptibility testing on automated systems has been unavailable for this agent, although there are studies showing the efficacy of commercial susceptibility testing methods compared to standard broth microdilution techniques (7, 8). Otherwise, very few studies on the multicenter evaluation and use of automated systems for antimicrobial susceptibility testing of CZA have been published at this time (9).

The Vitek 2 AST-GN Ceftazidime/Avibactam test (bioMérieux, Inc., Durham, NC) was designed for antimicrobial susceptibility testing of Enterobacterales and P. aeruginosa against CZA using the Vitek 2 and Vitek 2 COMPACT systems. Here, we report a multicenter systematic evaluation of the Vitek 2 CZA test for antimicrobial susceptibility testing of Enterobacterales and P. aeruginosa.

(Partial study data were presented at ASM Microbe in Atlanta, GA, on 8 June 2018 [10].)

MATERIALS AND METHODS

Ethics statement.

Study sites received approval or waiver from each institution’s review board before study initiation.

Trial product.

The Vitek 2 AST-GN Ceftazidime/Avibactam test has concentrations of CZA in the card of 0.06/4, 0.25/4, 1/4, 4/4, and 8/4 μg/ml. The CZA MIC reporting ranges for the card are ≤0.12/4 to ≥16/4 μg/ml for Enterobacterales and ≤0.25/4 to ≥16/4 μg/ml for P. aeruginosa. The Vitek 2 AST cards are used in conjunction with the Vitek 2 systems. The broth microdilution (BMD) test was performed in accordance with Clinical and Laboratory Standards Institute (CLSI) standards (11). BMD were prepared using ceftazidime concentrations of 0.015 to 64 μg/ml with avibactam concentration fixed at 4 μg/ml. The Vitek 2 AST-GN Ceftazidime/Avibactam assay received FDA 510(k) clearance in December 2017.

Settings.

Four testing sites participated in this study: (i) NewYork-Presbyterian Queens (NYPQ), Flushing, New York; (ii) Indiana University School of Medicine (IUSM), Indianapolis, Indiana; (iii) University of California—Los Angeles (UCLA), Los Angeles, California; and (iv) bioMérieux, Inc. (STLCA), Hazelwood, Missouri. The detailed study scheme at each site is presented in Table 1.

TABLE 1.

Study testing overview^a

Test	Clinical and QC	Challenge and QC	Reproducibility
1st card	Vitek 2 automatic dilution	Vitek 2 automatic dilution	Vitek 2 automatic dilution
2nd card		Vitek 2 manual dilution	Vitek 2 manual dilution
3rd card		Vitek 2 COMPACT manual dilution	Vitek 2 COMPACT manual dilution
BMD	Yes	Yes	No
Clinical trial sites	IUSM, NYPQ, UCLA, STLCA	STLCA	IUSM, NYPQ, UCLA

^aBMD, broth microdilution; IUSM, Indiana University School of Medicine; NYPQ, NewYork-Presbyterian Queens; QC, quality control; STLCA, bioMérieux, Inc.; UCLA, University of California—Los Angeles.

Reproducibility.

The reproducibility set consisted of 10 Gram-negative organisms, including P. aeruginosa (n = 2), Serratia marcescens (n = 2), Citrobacter freundii (n = 1), Escherichia coli (n = 1), Klebsiella (Enterobacter) aerogenes (n = 2), Citrobacter braakii (n = 1), and K. pneumoniae subsp. pneumoniae (n = 1). Reproducibility testing was performed for Vitek 2 systems at 3 participating clinical trial sites. Each reproducibility isolate was tested in triplicate for 3 days by the following methods: Vitek 2 autodilution and manual dilution methods and Vitek 2 COMPACT manual dilution. At the end of the study, each reproducibility isolate had 27 card results for each method. Best-case calculation for reproducibility assumed that the off-scale result is within one well from the mode. Worst-case calculation for reproducibility assumed that the off-scale result is greater than one well from the mode.

QC.

Quality control (QC) was performed for the Vitek 2 and BMD each day of comparative testing at each site using the FDA/CLSI-recommended QC set, including E. coli ATCC 25922, E. coli ATCC 35218, P. aeruginosa ATCC 27853, and K. pneumoniae subsp. pneumoniae ATCC 700603. The QC result for each QC organism was within acceptable limits for 99% of the QC tests performed. QC was within the normal limits on every day that comparative analyses (clinical and challenge studies) were performed. QC testing was performed a minimum of 20 times at each site.

Sample types.

Overall, 1,083 isolates (combined clinical and challenge organisms) of Enterobacterales and P. aeruginosa isolates were tested. Isolates that failed to grow were excluded from the analysis. Growth failure rates for clinical and challenge isolates were calculated.

Clinical isolates.

The sites collected 980 clinical isolates (NYPQ, 325; IUSM, 292; UCLA, 335; STLCA, 28), which were comprised of 960 CZA-susceptible isolates and 20 CZA-resistant isolates. Out of 980 clinical isolates, 105 were retrospectively (stock) and 875 were prospectively collected. Clinical isolates were C. freundii (n = 52), Citrobacter koseri (n = 44), K. (E.) aerogenes (n = 46), Enterobacter cloacae (n = 34), E. cloacae subsp. cloacae (n = 24), E. cloacae complex (n = 19), E. coli (n = 159), Klebsiella oxytoca (n = 44), K. pneumoniae subsp. pneumoniae (n = 97), K. pneumoniae (n = 55), Morganella morganii subsp. morganii (n = 18), Morganella morganii (n = 12), Proteus mirabilis (n = 59), Proteus vulgaris (n = 29), Providencia stuartii (n = 24), P. aeruginosa (n = 161), Salmonella enterica serotype Enteritidis (n = 7), Salmonella enterica serotype Typhi (n = 4), S. marcescens (n = 75), Edwardsiella tarda (n = 1), Shigella flexneri (n = 1), and Shigella sonnei (n = 15). Each isolate was tested by both the Vitek 2 and BMD reference methods. Delineation of K. pneumoniae subsp. pneumoniae, M. morganii subsp. morganii, and E. cloacae subsp. cloacae was noted when possible to present more robust data through additional grouping.

Challenge isolates.

A total of 93 challenge isolates were tested, consisting of 38 CZA-susceptible isolates and 55 CZA-resistant isolates, as confirmed with the BMD reference method and category interpretation per FDA-recognized CLSI/EUCAST breakpoints. They included E. cloacae (n = 9), E. coli (n = 10), P. aeruginosa (n = 46), S. marcescens (n = 1), P. stuartii (n = 1), K. pneumoniae (n = 10), K. pneumoniae subsp. pneumoniae (n = 6), K. oxytoca (n = 4), K. (E.) aerogenes (n = 1), and C. freundii (n = 5). The challenge isolates were tested by Vitek 2 and BMD at one site (Table 1). The challenge isolates included representative isolates with various β-lactamase and carbapenemase genes, including E. cloacae (AmpC, KPC), E. coli (AmpC, CTX-M, TEM), K. pneumoniae (CTX-M, TEM, SHV, KPC), and P. aeruginosa (AmpC variants PDC, KPC, GES, VEB, and outer membrane porin loss OprD).

Testing.

Each isolate was first subcultured on tryptic soy agar with 5% sheep blood. After an 18- to 24-h culture at 35 ± 2°C, suspension with a 0.5 to 0.63 McFarland range was prepared using the DensiCHEK Plus instrument (bioMérieux, Inc., Durham, NC). Clinical, challenge, reproducibility, and QC isolates were all tested on the Vitek 2 autodilution method. The Vitek 2 manual dilution method was evaluated for QC, reproducibility, and challenge testing, as was the Vitek 2 COMPACT with manual dilution. The summary of testing types and testing sites for different methods is provided in Table 1. For each isolate, a single initial McFarland suspension was prepared for inoculation of Vitek 2 cards and BMD method for comparative testing. Both methods were performed in singlet unless mentioned otherwise.

Interpretation of results and data analysis.

Results from the comparative testing of Vitek 2 and BMD methods were included in the analysis. FDA (12) and ISO 20776-2 (13) susceptibility performance criteria and methods of data analysis were followed to evaluate the performance using the FDA-recognized CLSI and EUCAST breakpoints (Enterobacterales and P. aeruginosa, sensitive [S], ≤8/4 μg/ml; resistant [R], ≥16/4 μg/ml), respectively. Comparative clinical performance, challenge testing, and reproducibility analyses were performed for the Vitek 2 autodilution, Vitek 2 manual dilution, and Vitek 2 COMPACT methods. Comparative performance was evaluated using essential agreement (EA), categorical agreement (CA), very major error, and major error rates (see Table S1 in the supplemental material). To fulfill FDA criteria, alternate analysis was performed in accordance with the Susceptibility Testing Manufacturers Association (STMA) letter and FDA favorable disposition in 2016 (14). In brief, the lack of an intermediate breakpoint does not necessarily affect the CA between the test device and the reference method. However, the lack of an intermediate breakpoint may affect the number and percentage of isolates with major and very major errors. Isolates that, by definition, were very major or major errors but were within the EA of the reference method were noted as errors but were considered acceptable when taking the EA into consideration. Adjusted errors were calculated in these circumstances and presented along with original errors. For ISO criteria, errors were resolved per ISO standard 20776-2, section 5.2.7 (13), and overall performance was recorded after error resolution. In brief, the discrepancy was resolved by a onetime triplicate testing of the reference method using separate bacterial inoculum suspensions. The CA mode of the triplicate results for the reference method replaced the original result for the purpose of determining the error rate.

RESULTS

Reproducibility.

The line-listing of reproducibility isolate performance, including the mode MIC value for Vitek 2 automatic, manual, and COMPACT methods, is included in the supplemental material (Table S2a, S2b, and S2c). The reproducibility was calculated based on MIC values that fell within one doubling dilution of the mode MIC value. The Vitek 2 best-case overall reproducibility for autodilution was 99.6%, and worst-case overall reproducibility was 99.3% (Table S2a). Best-case overall reproducibility for manual dilution was 99.6%, and worst-case overall reproducibility was 99.3% (Table S2b). The Vitek 2 COMPACT best-case overall reproducibility was 100%, and worst-case overall reproducibility was 98.9% (Table S2c). All reproducibility results met the ≥95% reproducibility requirement for ISO and FDA.

Quality control.

QC testing was performed for Vitek 2 autodilution, Vitek 2 manual dilution, Vitek 2 COMPACT dilution, and the reference method at each site. QC results were within acceptable limits of ≥95% of QC tests performed on each quality control organism for all Vitek 2 methods (Table S3). QC was verified to be within normal limits on every day that comparative testing (challenge and clinical isolates) was performed.

Clinical performance.

We examined the clinical performance of Vitek 2 at four previously noted sites (Table 1). Complete test results were available for 980 isolates from a total of 990 clinical isolates tested in this study. The growth failure rate was 1.0% (10/980), well within acceptable FDA criteria (<10%). Ten clinical isolates that did not grow in Vitek 2 cards included P. aeruginosa (3), Shigella (6), and E. coli (1). A total of 980 clinical isolates, including 98.0% (960/980) of susceptible and 2.0% (20/980) of resistant isolates, were evaluated across the four sites using Vitek 2 autodilution and BMD. The EA for clinical isolates was 94.1% (922/980), and the CA was 98.7% (967/980). No very major errors were found, and the major error rate was 1.4% (13/960) (Table 2).

TABLE 2.

Performance of clinical and challenge isolates, Vitek 2 automatic dilution method^d

Ceftazidime-avibactam treatment	Total tested, n	EA, n	EA, %	Eval			CA, n	CA, %	R, n	Major, n	Very major, n
				EA total, n	EA, n	EA, %
Enterobacterales, ≤8 susceptible, none intermediate, ≥16 resistant
Clinical	819	770	94.0	265	216	81.5	813	99.3	10	6	0
Challenge	47	46	97.9	20	19	95.0	46	97.9	27	1	0
Combined	866	816	94.2	285	235	82.5	859	99.2	37	7a (0.8%)	0 (0%)
P. aeruginosa, ≤8 susceptible, none intermediate, ≥16 resistant
Clinical	161	152	94.4	145	136	93.8	154	95.7	10	7	0
Challenge	46	46	100	18	18	100	46	100	28	0	0
Combined	207	198	95.7	163	154	94.5	200	96.6	38	7b (4.1%)	0 (0%)
Enterobacterales plus P. aeruginosa
Clinical	980	922	94.1	410	352	85.9	967	98.7	20	13	0
Challenge	93	92	98.9	38	37	97.4	92	98.9	55	1	0
Combined	1,073	1,014	94.5	448	389	86.8	1,059	98.7	75	14c (1.4%)	0 (0%)

^aOut of seven major errors, four major errors were one dilution apart from the reference method and, as such, fall within essential agreement.

^bAdjusted major error rate for FDA criteria for P. aeruginosa was calculated due to lack of an intermediate category. Before adjustment, major error rate as presented in the table was 4.1% (7/169). After adjustment, major error rate was 1.2% (2/169). Five P. aeruginosa isolates were within essential agreement compared to the reference method.

^cThe overall categorical major error rate for Enterobacterales and P. aeruginosa combined was 1.4% (14/998). Nine major errors were one dilution apart from the reference method and, as such, fall within essential agreement. Based on the essential agreement and the lack of an intermediate breakpoint for ceftazidime-avibactam, the adjusted categorical major error rate for FDA criteria is 0.5% (5/998). For resolved errors per ISO standards, refer to Table S4.

^dEssential agreement (EA; ±1 doubling dilution) occurs when there is agreement between the result of the reference method and that of the Vitek 2 test card within one serial twofold dilution of the antibiotic. Evaluable results are those that are on scale for both the Vitek 2 test card and the reference method. Category agreement (CA) occurs when the interpretation of the result of the reference method agrees exactly with the interpretation of the Vitek 2 test card. Eval, evaluable isolates; major, major discrepancies; n, number of isolates; R, resistant isolates; very major, very major discrepancies.

Out of 980 clinical isolates, Enterobacterales comprised 83.6% (819/980) and P. aeruginosa comprised 16.4% (161/980). Per FDA/CLSI/EUCAST breakpoints for Enterobacterales, the study included 98.8% (809/819) CZA-susceptible and 1.2% (10/819) CZA-resistant isolates by BMD. Enterobacterales clinical isolate performance included 94.0% (770/819) EA and 99.3% (813/819) CA. No very major errors were found, and the major error rate was 0.7% (6/809) (Table 2). P. aeruginosa isolates were 93.8% (151/161) of susceptible and 6.2% (10/161) of resistant isolates, as confirmed with BMD per FDA/CLSI/EUCAST breakpoints for P. aeruginosa. EA and CA for P. aeruginosa were 94.4% (152/161) and 95.7% (154/161), respectively. No very major errors were found, and the major error rate was 4.6% (7/151) (Table 2).

Challenge performance.

To examine test accuracy, MICs derived by Vitek 2 and BMD methods were compared to those of the challenge set. All 93 challenge organisms grew in the Vitek 2 cards. They included 40.9% (38/93) CZA-susceptible and 59.1% (55/93) CZA-resistant isolates. The overall EA for challenge testing was 98.9% (92/93), and the CA was 98.9% (92/93). No very major errors were reported, and the major error rate was 2.6% (1). The challenge performance was similar for all the Vitek 2 methods (Table 3).

TABLE 3.

Performance of challenge isolates^a

Test	Total isolates, n	EA, n	EA, %	Eval			CA, n	CA, %	R, n	Major, n	Major, %	Very major, n	Very major, %
				EA total, n	EA, n	EA, %
Vitek 2
Automatic dilution	93	92	98.9	38	37	97.4	92	98.9	55	1	2.6	0	0
Manual dilution	93	92	98.9	38	37	97.4	92	98.9	55	1	2.6	0	0
Vitek 2 COMPACT
Manual dilution	93	92	98.9	38	37	97.4	92	98.9	55	1	2.6	0	0

^aCA, category agreement; EA, essential agreement (±1 doubling dilution); Eval, evaluable isolates; major, major discrepancies; n, number of isolates; R, resistant isolates; very major, very major discrepancies.

The 93 isolates in the challenge set included 47 Enterobacterales and 46 P. aeruginosa isolates. The Enterobacterales challenge set included 20 susceptible and 27 resistant isolates. Performance of Enterobacterales challenge testing included 97.9% (46/47) EA and 97.9% (46/47) CA. No very major error was recorded; one major error was recorded (Table 2). The P. aeruginosa challenge set included 18 susceptible and 28 resistant isolates. Performance of P. aeruginosa challenge testing included 100% (46/46) EA and 100% (46/46) CA. No very major or major errors were reported (Table 2).

Overall performance combining clinical and challenge isolates.

A combined total of 1,073 clinical and challenge isolates of Enterobacterales and P. aeruginosa were evaluated, with exclusion of only one organism, as noted in Discussion. The overall population included 7.0% (75/1073) resistant isolates and 93.0% (998/1073) susceptible isolates. The overall EA was 94.5% (1014/1073) and CA was 98.7% (1059/1073). No very major error (0/75) was reported, and the major error rate was 1.4% (14/998) (Table 4). As such, overall performance met the FDA and ISO criteria.

TABLE 4.

Overall performance of Enterobacterales and P. aeruginosa

Organism	Total isolates, n	EA, n	EA, %	CA, n	CA, %	R, n	Major,a n	Major, %	Very major, n	Very major, %
C. freundii	57	49	86.0	57	100.0	5	0	0	0	0
C. koseri	44	43	97.7	44	100.0	0	0	0	0	0
E. tarda	1	1	100.0	1	100.0	0	0	0	0	0
K. (E.) aerogenes	47	43	91.5	46	97.9	1	1	2.2	0	0
E. cloacae	43	43	100.0	43	100.0	9	0	0	0	0
E. cloacae subsp. cloacae	24	20	83.3	24	100.0	0	0	0	0	0
E. cloacae complex	19	19	100.0	19	100.0	0	0	0	0	0
E. coli	169	163	96.4	168	99.4	7	1	0.6	0	0
K. oxytoca	48	46	95.8	48	100.0	3	0	0	0	0
K. pneumoniae subsp. pneumoniae	103	97	94.2	103	100.0	1	0	0	0	0
K. pneumoniae	65	63	96.9	64	98.5	7	1	1.7	0	0
Morganella morganii subsp. morganii	18	18	100.0	18	100.0	0	0	0	0	0
M. morganii	12	11	91.7	12	100.0	0	0	0	0	0
P. mirabilis	59	57	96.6	59	100.0	0	0	0	0	0
P. vulgaris	29	25	86.2	29	100.0	0	0	0	0	0
P. stuartii	25	23	92.0	23	92.0	3	2	9.1	0	0
Salmonella Enteritidis	7	7	100.0	7	100.0	0	0	0	0	0
Salmonella Typhi	4	3	75.0	4	100.0	0	0	0	0	0
S. marcescens	76	69	90.8	74	97.4	1	2	2.7	0	0
S. flexneri	1	1	100.0	1	100.0	0	0	0	0	0
S. sonnei	15	15	100.0	15	100.0	0	0	0	0	0
P. aeruginosa	207	198	95.7	200	96.6	38	7	4.1	0	0
All species combined	1,073	1,014	94.5	1,059	98.7	75	14b	1.4	0	0

^aThe overall categorical major error rate for Enterobacterales and P. aeruginosa combined was 1.4% (14/998). Nine major errors were one dilution apart from the reference method and, as such, fall within essential agreement. Based on the essential agreement and the lack of an intermediate breakpoint for ceftazidime-avibactam, the adjusted categorical major error rate for FDA criteria was 0.5% (5/998).

^bAdjusted major error rate for P. aeruginosa was calculated due to lack of an intermediate category. Before adjustment, major error rate as presented in the table was 4.1% (7/169). After adjustment, the major error rate for FDA criteria was 1.2% (2/169). Five P. aeruginosa isolates were within essential agreement compared to the reference method. Out of seven major errors for Enterobacterales, four major errors were one dilution apart from the reference method and, as such, fall within essential agreement. For resolved errors per ISO standards, refer to Table S4.

Out of 14 major errors, nine were one dilution apart from the reference method and fell within the EA. Based on the EA and the lack of an intermediate breakpoint for CZA, the adjusted major error rate for FDA criteria was 0.5% (5/998) (Table 2).

The overall performance of Vitek 2 for ISO criteria (applicable to EUCAST breakpoints) after error resolution included EA of 94.5% (1,014/1,073), CA of 98.9% (1,061/1,073), major error of 1.2% (12/998), and no very major error (Table S4).

A combined 866 clinical and challenge isolates of Enterobacterales were evaluated. The overall population included 4.3% (37/866) resistant isolates and 95.7% (829/866) susceptible isolates, as determined by BMD per FDA/CLSI/EUCAST breakpoints. The overall EA was 94.2% (816/866), and the CA was 99.2% (859/866). No very major errors, and a 0.8% (7/829) major error rate, were reported for the overall performance. For P. stuartii, which showed an overall major error of 2/22 (9.1%), both errors were one dilution apart from the reference method and fell within the EA. Based on the EA and the lack of an intermediate breakpoint for CZA, the adjusted major error rate was 0% (0/22). Overall, for Enterobacterales, out of 7 major errors, 4 were one dilution apart from the reference method and fell within the EA. The adjusted major error rate was 0.4% (3/829) (Table 2). Overall performance by organism groups within Enterobacterales included ≥90% EA and ≥90% CA for all organism group except for C. freundii (EA of 86.0%, CA of 100%) and E. cloacae subsp. cloacae (EA of 83.3%, CA of 100%) (Table 4). There was 100% CA; therefore, no major or very major errors for either organism group were noted, demonstrating an acceptable overall performance. Further, species-level EA was 94% for E. cloacae when calculated after combining performance of E. cloacae isolates with E. cloacae subsp. cloacae isolates (data not shown).

A combined total of 207 clinical and challenge isolates of P. aeruginosa were evaluated. The Overall population included 18.4% (38/207) resistant isolates and 81.6% (169/207) susceptible isolates, as determined by BMD per FDA/CLSI/EUCAST breakpoints. The overall EA was 95.7% (198/207), and overall CA was 96.6% (200/207). No very major errors, and a 4.1% (7/169) major error rate, were reported for the overall performance. Of the 7 major errors, five major errors were one dilution apart from the reference method and fell within the EA. Based on the EA and the lack of an intermediate breakpoint for CZA for P. aeruginosa, the adjusted categorical major error rate was 1.2% (2/169) (Table 4).

The challenge isolates used in the study included representative Enterobacterales isolates with β-lactamases including KPC (n = 6), CTX-M (n = 4), TEM (n = 11), SHV (n = 12), and plasmid-mediated AmpC (n = 4). The Enterobacterales with beta-lactamase enzyme characterization included K. pneumoniae (KPC [n = 5], SHV [n = 12], TEM [n = 9], and CTX-M [n = 2]), E. coli (AmpC [n = 2], CTX-M [n = 2], and TEM [n = 2]), and E. cloacae (KPC [n = 1], and AmpC [n = 2]). The study showed susceptibility to CZA with Vitek 2 MIC values ranging from 0.5/4 to 2/4 μg/ml (data not shown). Additionally, the challenge set included seven P. aeruginosa isolates with β-lactamase AmpC variants PDC (n = 7), KPC (n = 1), GES (n = 1), VEB (n = 1), and outer membrane porin loss OprD (n = 7). Two isolates demonstrated resistance with MICs of ≥16/4 μg/ml, and five isolates showed susceptibility ranging from 2/4 to 8/4 μg/ml (data not shown).

A trending analysis of evaluable clinical and challenge results for Enterobacterales and P. aeruginosa was also performed. A trend toward higher MIC was observed with Enterobacterales and P. aeruginosa compared to the CLSI broth microdilution reference method (data not shown).

DISCUSSION

CZA is a novel combination of the third-generation cephalosporin ceftazidime and the non-β-lactam β-lactamase inhibitor avibactam. CZA has excellent in vitro activity against important Gram-negative pathogens, including extended-spectrum β-lactamase-, AmpC-, KPC-, and OXA-48-producing Enterobacterales and drug-resistant P. aeruginosa isolates. CZA does not demonstrate activity against metallo-β-lactamase-producing strains (3, 5). As CZA was recently approved for the treatment of multidrug-resistant bacteria, susceptibility testing methods have only recently become available. Systematic assessments of alternative methods to BMD have not been widely published for CZA susceptibility testing, with only a few such studies on alternative methods having been published (8, 9, 15). For example, Shields et al. compared the disk diffusion and Etest methods to the BMD reference method for testing CZA against a diverse collection of CRE isolates (8). Among the 74 mixed CZA-susceptible and -resistant CRE isolates, BMD CA was higher with Etest (96%) than with disk diffusion (72%; P = 0.0003), demonstrating that Etest measurements of CZA susceptibility correlated closely with the standard BMD method.

Automated antimicrobial susceptibility testing methods can standardize the reading of endpoints and often produce susceptibility test results in a shorter time period than the BMD reference method (16). In this study, a shorter mean time of call for Vitek 2 results (9.80 ± 2.80 and 10.19 ± 2.77 h, respectively, for clinical and challenge isolates using Vitek 2 autodilution) was recorded.

We evaluated Vitek 2 as an alternative automated method to BMD per FDA and ISO performance criteria for validation of antimicrobial susceptibility tests (12, 13). The evaluation of Vitek 2 was conducted on a large number of Gram-negative bacterial isolates in a multisite study with geographically diverse locations across the continental United States. The MIC values derived by the Vitek 2 correlated well with MIC values obtained by the reference BMD for both Enterobacterales spp. and P. aeruginosa, exceeding the ≥90% threshold of EA required by the FDA and ISO. MIC trending analysis showed that Vitek 2 MIC values tended to be in exact agreement or at least one doubling dilution higher when testing Enterobacterales or P. aeruginosa than the reference BMD. However, the high EA (>90%) reaffirms the desired performance.

The CA for Enterobacterales spp. and P. aeruginosa using the FDA/CLSI/EUCAST breakpoint exceeded the ≥90% threshold of CA required by the FDA and ISO. Overall, no very major errors (false susceptibility) were reported, and the major errors (false resistance) were well within the FDA and ISO threshold limit (≤3.0%) and often occurred due to the lack of an intermediate breakpoint category. Testing should be repeated using an alternative testing/reference method prior to reporting results for P. aeruginosa when the Vitek 2 MIC is ≥16 μg/ml due to observed major errors (4.1%), with an adjusted major error rate of 1.2% in this study.

P. rettgeri data were excluded from all analyses, as it did not meet the FDA and ISO performance criteria for Vitek 2 testing in this study. Therefore, a performance limitation was noted, and P. rettgeri data were excluded from the analysis. It is advised that an alternative method of testing for CZA susceptibility or resistance in P. rettgeri should be performed prior to the reporting of Vitek 2 AST results.

Best- and worst-case assessments of combined reproducibility for all three sites were well above the acceptance criteria for the FDA and ISO (≥95%). Consistently high reproducibility above the acceptance criteria was reported for all Vitek 2 methods. Reproducibility is one of the key indicators to establish the performance of the automated method as an alternative to BMD (17).

Vitek 2 performance was evaluated using the challenge set isolates of Enterobacterales with β-lactamases, including KPC, CTX-M, TEM, SHV, and plasmid-mediated AmpC, along with P. aeruginosa isolates with β-lactamases AmpC variants PDC, KPC, GES, VEB, and outer membrane porin loss OprD. Enzyme group characterization was not available for Enterobacterales having OXA and metallo-β-lactamases and for P. aeruginosa having metallo-β-lactamases given the lack of either clinical or challenge organisms with the presence of OXA-like enzymes. Thus, the performance of Vitek 2 is unknown for these enzyme classes. However, it must be noted that CZA is not active against bacteria that produce metallo-β-lactamase enzymes and may not have activity against Gram-negative bacteria that overexpress efflux pumps or have porin mutations (1, 5).

Overall, this multicenter evaluation of Vitek 2 AST-GN Ceftazidime/Avibactam on a large number of clinical strains from different geographic regions within the United States, along with quality control, reproducibility, and challenge studies, establishes that Vitek 2 is a reproducible and accurate method for CZA susceptibility testing of Enterobacterales spp. and P. aeruginosa.